THE DIFFERENTIAL DISTRIBUTION OF MARKET "NEWS FUNCTIONS: AN INTERPRETATIVE STUDY OF SELECTED STRUCTURAL DIMENSIONS 7 W Fl! I! m 5 HI. I. MICHIGAN 51m uumasm ROBERT C. mum . I 962 IIIIIMLIITIMIWW This is to certify that the thesis entitled THE DIFFERENTIAL DISTRIBUTION OF MARKET NEWS FUNCTIONS: AN INTERPRETIVE STUDY OF SELECTED STRUCTURAL DIMENSIONS presented bg ROBERT C . BEALER has been accepted towards fulfillment of the requirements for Ph.D Soc. 8 Anthro. ' degree in Date 11.03/39] b §LJ.LITJ_~_‘._9.62_ 0-169 LIBRARY Michigan State University ,b.._h-‘. , _,..., ”J W Ac. -Awww MSU LIBRARIES “ RETURNING MATERIALS: PTace Thgbodk drop to remove this checkout from your record. FINES wil] be charged if book is returned after the date stamped beIow. THE DIFFERENTIAL DISTRIBUTION OF MAdKET NEWS FUNCTIONS: AN INTJRPRBTATIVE STUDY OF SELECTED STRUCTURAL DIMSNSIONS ‘1) L by ‘0‘ Robert C. Bealer A THESIS Submitted to the College of Science and Arts Michigan State University of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY _ Department of Sociology and Anthropology 1962 Approved: Mtge [{WM O! I: A" I Abstract In this study a limited type of functional analysis was used to explain certain uses of market news by a representative sample of 356 lower peninsula Michigan farmers receiving such information. This analytical model was selected to compensate for certain deficiencies both in previous research and theory about market news, and in , “effects” studies of mass commication. The organizing orientation was that of structural constraint; i.e_., the function of an item is conditioned by the context of its appearance. The item to which functions are imputed is the market news infor- mation received. The significant structural context was taken to be the individual's occupational role and was originally conceived as having five important dimensions: social-psychological involvement; W from non-preference economic factors; characteristic diffuse or specific type social relations with farm nei hbors, and with major product deg-lg; and degree of ggtiongitz in the managerial practices used. The possible uses for market news were categorized as latent or manifest on the basis of the intent of the United States Department of Agriculture Market News Service program. They included nine latent and seven manifest functions. The total number of each occurring, considered separately and then jointly, were the three dependent vari- ables for the first phase analysis. A factorial analysis of variance ledel (after Keyfits) assessed the ability of the structural dimensions in accounting for differentiating functions scores. Mode of market news receipt and type of information was controlled in this analysis. .. 2 .. The factorial analysis showed that rationality and both types of social relations significantly differentiated total and latent scores, but-that none of these significantly differentiated manifest scores. High rationality levels and diffuse relations were associated with greater occurrence of these functions. Neither commitment nor involve- ment showed any consistent significant differentiating ability. Structure was redefined on the basis of this evidence. Only those variables that were consistently significant differentiators were used in this phase of the analysis. Social relations were classed as diffuse, mixed, or specific on the basis of the neighbor and dealer Guttman-type scales. Rationality, measured by Guttman technique, was dichotomized. Cross classification of these two dimensions yielded six homogeneous types which were examined with respect—to their rela- tionship with each of the nine specific 'types of behavior construed as possible latent functions. in "Opportunity-motivation" proposition was evolved, which con- sistently differentiated the occurrences of the specific latent functions. Because this proposition could be derived as a working hypothesis from the notion of functional constraint, and because it could tie the disparate behavior together in a logical and meaningful fashion, it was taken as evidence for the validity of the study guide of functional constraint. It could not be taken as a formal test, however, since the analysis is 1933 mm. Further research on functional constraint in communication research is urged. Acknowledgements Like every human product, this dissertation is a social endeavor with a history of indebtedness too long to be listed, too complex and subtle even to be known by the writer. However, for their special prominence I would like to thank: Dr. Joel Smith, for assuming leadership of the research project upon which this dissertation is based; for numerous perceptive insights into the problem; for advising the dissertation to its completion and contributing editorial work well beyond normal expectations. Dre. Charles P. Loomis and John Useem who, as successive Heads of the Department of Sociology and Anthropology, made available the research facilities of that Imit for the study as well as providing the means for following out graduate study. Dean Glenn Taggart, for providing initial interest in the research project and original chairmanship both of the project and the thesis and for friendly encouragement in that endeavor. Dr. Macklin E. John, Department of Rural Sociology of the Pennsyl- vania State University, for obtaining arrangements facilitating the writing completion of this dissertation during the author's mployment at that institution. Mr. Francis M. Sim who, as joint field supervisor and colleague, shared the experiences of the research from start to finish and provided, apart from intellectually rewarding comradery, the enjoyment of peremal friendship. Mrs. Barbara Frantz and Mrs. Jo anith for typing the final draft Of the manuscript. My wife, Barbara Ann, and the children, Anita Carol and Andy Glenn, for being themselves and providing the vital support of the inseparable "whole” nan. l Bible of Content! Page on?” I file H031“ eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 1 m 3‘3de 0: m 8m: eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 1 ”0109136 “I. P13913193! eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 1‘" Bone Relevant Generalizations from Pest Research ......... 4 the Relevance of .Beeeerch Generalizations for the Present SW eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 11 lunctiensl Analysis: A homework for Studying “ht I." H193“ eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 19 he Special Significance of “sectional Alternatives“ m 'WthOIi" « m 3m eeeeeeeeeeeeeeeeeeeeeeeeeeee 21 Chapter II he Scheme for Analysis: General Prepertiee .......... 25 M1”. 0: Strmm. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 25 M01381 Definition eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 25 Conlideratienl from the ROM DB's-g eeeeeeeeeeeeeeeeee 27 m MMt Variablee eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 32 Definition 0: function eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 38 8030 General P136510” 0: W1. eeeeeeeeeeeeeeeeeeeeeeeeeeee “1 ”$013181 “.1“ eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 53 General ”mtio' eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 53 Adaptation $0 the Ram}! 130.1511 eeeeeeeeeeeeeeeeeeeeeeee 58 chapter 111 he Sch. for Analysis: Operationalizatien ......... 63 kt.» 001106‘1“. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 63 Study Development, Schedule Construction. and. h.“.t1n‘ eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 63 m kNOM eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 6“ ‘hMS- eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee ‘47 J\ 3W0 B‘Pr..ut1'm" eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 69 WOO: Independent Variables eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 7“ Mill 31.3103. and. ”$103.11” eeeeeeeeeeeeeeeeeeeeeeeee 8° Infilmt and. CMtMt eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 90 M10": Wt Verieblel eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 1-02 MOUt 8001'. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 103 but 3001'. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 1“ 136.100.} centre]. "Pi‘blO. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 11“ Mm 1' mention Of “1. Structural MCI.” eeeeeeeeeeeeee 117 lum 0: th. Independent ”1.310. eeeeeeeeeeeeeeeeeeeeeee 117 Relation of the Structural Dimensions to reaction Scores ..... 125 ’1'. Variables, N0 Oentrele eeeeeeeeeeeeeeeeeeeeeeeeeeeeee 125 m 78318310., 3113510 GOBtI'Ol. eeeeeeeeeeeeeeeeeeeeeeeeee m” vui‘blo'. M 003ml. eeeeeeeeeeeeeeeeeeeeeeeeeeee 151 Nation of the Structural Dineneione' Relations to t1“. SOOI‘QI eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 155 h structural Dimension! .0 8"“. eeeeeeeyeeeeeeeeeeeeeee 157 5‘ Pres-em criterion and Structural Dimensions ........ 160 I" l' u fable of Oentents chapter 7 he Structural Dimensions and Structural Constraint ... Intrehotion eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeseeeee structural ”PO. and Junction-l Constraint eeeeeeeeeeeeeeeeeee Defining the Structural m0. eeeeeeeeeeeeeeeeeeeeeeeeeeee Use of lbs-hot News as a Convenient Conversation Item ..... Getting General Educational Material from Msrloetlews .... VIC 01 Surrogates #0 Obtain Mei-lost N6“ eeeeeeeeeeeeeeeeee )lsrlnet News as a criterion for Esteem Evaluations ........ xerket news as e rector in Evaluating USDA Performs ... “200‘ Inward. ”10111:”. ud- lhrket NO“ eeeeeeeeeeeeeeee Market News 88 3 Manta]- ”31th “661183.138 eeeeeeeeeeeeeeeee Ion-solicited Intent Functions and Mai-lost News ........... Over-ell Evaluation of the Evidence for Structural Consul-int eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee chapter 11 M end conclusions eeeeeeeeeeeeeeeeeeeeeeeeeeeeee W ‘ eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee W 3 eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee AM o eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Bibliomw eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 167 167 170 17 172 183 195 198 201 206 210 212 215 226 279 310 'nl' fable # CD ~§ Ch kn 10 12 13 1h» 15 16 List of fables Anew-is of Variance for fetal, Latent and mnifest 8601’. Classified. by 8.6mm sapling me. eeeeeeeeeeeee Scale of Intimacy in Social Relations with Dealer or MOI-MCQrt. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Scale ef Intimacy in Social Relations with Neighbors ..... Scale of Rationality in the Use of Supplementary h'WOO' in th. ”10‘1“ Operations eeeeeeeeeeeeeeeee Criteria for Evaluating the “It“ 834310. eeeeeeeeeeeeeee m 0! Constituent to farming eeeeeeeeeeeeeeeeeeeeeeeeeee Index Of Involvement in m eeeeeeeeeeeeeeeeeeeeeeeeee lean i'otal, intent and Manifest Score for Each Level Of the ’17. Structural M1011. eeeeeeeeeeeeeeeeeeeeeeee Kean Donation Scores by Factorial Cell Classified by m ’17. 8mm Dimension! eeeeeeeeeeeeeeeeeeeeeeeeeee Computational. Material for Assessing the Siaificance of Differences in Kean Total function Scores for nationality: Basis of Classification, rive Structural Dmaion. ......OOOOOOOOCOOOO0.0.0.0.........OOOOOOOOOOOO 'Interections' Matrix for the live Structural Dinensions.. Differences in Mean Scores, Standard. Errors, and t Values for the live Structural Dimensions Factorial “I1“, I'D-“110d EO't. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Analysis of Variance for total, Latent, and Manifest Score Classified by Levels of Commitment and of 137.17“; eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Differences in Kean Scores, Standard lrrors, and 2 Values of the live Structural Dimensions with He mmm Controls eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Differences in lean Scores, Standard More, and t Values of the four Significant Structural Variables Isotorisl Design vith Controls Separately Added for Source 0: Market I.“ and Definition 0: Market I." eeeeee Contingency of the Control Variables and the Structural ”W103. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee 81 83 87 96 101 120 127 133 135 1&2 1% 1b? 150 1.! “I “- (u _'.\’ fable 1? 18 19 20 23 25 27 29 30 31 32 List of Tables Differences in lean Scores, Standard Errors, and t Values for fhree Structural Dimensions and Two Control "rmJ-O. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeseeeeeeeeeeeeeeeeee lieu Differences in leans. Standard Rrrers, and t Values for the Structural and Control Variables “HCGtOd 0V0: th‘ 1170 “tori-‘1‘ eeeeeeeeeeeeeeeeeeeeeee Contingencies Along the Structural Dimensions Sigifioant in Differentiating lune tion Scores . . . . . . . . . . . Equation for Predicting Total and Instant functions Scores and the Per Cent of Variance Accounted for by mh .....OOOOOOIQOOOOOOOOOO....O.......OOOOOCIOOCOOOOOOOO Use of Market Rows as a Convenient Conversation Piece “0113 the Structural TYPO. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Persons Attending Para Bureau and Coop Meetings and Participating in the SOS Program Among the Structural we. .0.......O.........OOO......OOOOOOOOOOOOOOO0.0.0.... Persons the Discuss the Market News Thaw Get with Persons Bot Well Known Anon; the Structural i‘ypes ........ Use of Market News as a Conversation Primer Among m award Ema eeeeeeeeeeeeeeeeeeeeeeeseeeeeeeeeeeee Distribution of Persons Who Use Market laws as a Conversation Primer Iith Persons Rot Well Known Among the Structural m0. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee fears of Residence in the Comnity Among the Structural m. 0.0000............OOCOOOOOOOOOOOOOOO......OOOOOOOOOO Proportien of Close Friends in the Comunity of 30.1“”. “On; the Structural ”PO. eeeeeeeeeeeeeeeeeeeee Off-farm Deployment and Proportion of Total Income his York Provides Among the Structural w. eeeeeeeeeeee Obtaining General Knowledge from Continued Receipt of “ht I“. “038 “1. Structural w. eeeeeeeeeeeeeeeeeee 11.701. offered. 8611001135 m m Strum“ w. eeee ”IMtI' ‘30 Among m Structural m. eeeeeeeeeeeeee Use of Surrogates to Obtain thrket Rows Among the 8mm m. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Page 153 156 159 16% 172 175 1?? 178 179 180 180 181 181!- 187 190 \4.‘ Cable 33 35 37 b0 ‘bl #2 List of Tables Peeling of Deprivation with Less of Market News and Why Surroptes Are Rot Used Among the Structural Types ... Available family Surrogates Among the Structural iypes ... Use of Market News as a Criterion for Rsteen valuations Anne the Structural me. eeeeeeeeeeeeeeeeeee Use of Market flows as a hater in Evaluating USDA 203103”. Along the 8mm“ m0. eeeeeeeeeeeeeeeeeee Sources of Market News Among the Structural i‘ypes ........ Use of Marat lows in Effecting Rulings Toward mimm. mu m Structural me. eeeeeeeeeeeeeeoeeee Use of Market News as Succor for Mistakes Among the 8mm W. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Use of Hornet News as a Scapegoat for Mistakes “a; the Structural MOI eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Ron-solicited Latent Functions of Market News Among the 8mm m0. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Sal-.17 of the Support for Functional Constraint in the Relationship of the Structural Dimensions and the Latent Functions eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Page 192 193 197 199 200 2015 2% 209 211 213 List of Iiguree Figure Page 1 Distribution of. the Primary Sample Units and ‘ Rasher of Interviews Usable IronEach County ............. 65 2 Distribution of Kean total Score by Levels of. the ’1'. 8mm“- Yum}... ...OOOOOOOOOOOCOOOO00.00.00.000. 122 3 Distribution of Mean Latent Scores by Levels of the 31's swmu Yeti-Que“ oasesseeeeeeeeeeaessaoasosooeeoe 123 11- Distribution of Mean West Score by Levels of. the ’1'. 8mm“ Nab]... eeeseooesseeeseeeeeeeeeeeeoeoeso 12” List of Charts Chart Page 1 A Mothetical mket HOV. Situation eseeeeeeseeeoeeaoeaee 15 Chapter I The Problem 2.1.1.9 W 91 21.9 £39111 In the United States prior to the Civil War agriculture was largely either of a subsistence type or oriented to local markets where both producer and buyer could be reasonably informed of relevant price set- ting conditions. After the War the urban-industrial upsurge resulted in a dramatic shift of papulation from the farms to the cities. Con- comitantly, a large segment of the farms shifted from a subsistence to a commercial basis.1 Generally this move forced sales to markets removed both in space and time from the producers. Early in this re- organisation, dealers in farm products develoPed their own private intonation sources for supply and demand data necessary for rationally pricing those products they handled.2 However inadequate these sources might have been, they tended to place the dealers in a relatively more advantaged position than the farm producer in regard to market and price information.3 1. Mnmy Ben-dict, rm lanai” 9.: as man was, 12242.04 0. New York: Twentieth Century Fund, 1953, chapter five, particularly pp. 85-93. This characterization is less true of the cotton and to- bacco eeonoxv of certain parts of the South. Also, it should be noted that the "supporting" factors for this move were not all "ur- ban" in location. Technological advances in agricultural production increasing per operator output complemented the labor demands of the new manufacturing industries. 2. Bot infrequently the form of such "information“ lay in the quasi- nonopoliatie organisation of the buyer. See 1.154., pp. 513-514. 3- R. E. Bette, 91. 51., "the What News Services," in _1:_h_e. i icul a man: as Mum .....nServic 2; as lei—ted ......sStato 22m 3231 93, W: Washington, D.C.: Bureau of Agricultural Economics and Production and Lhrketing Administration, United States DQMQt Of Agriculture, NO. 703, DOM”, 1949, p0 1720 . 2 .. Growing out of the particular, unique history of Homestead settlement and the ideological valuation of the ”family farm" in ag- ricultural policy, America's commercial farms were generally, and still are, at once sufficiently large enough to make sale for distant markets a mndate but not of such scale to allow development by each producer of his own private information sources. "This situation placed the producer at a disadvantage in bargaining with buyers better infoned than himself regarding actual market conditionle‘ Farmer agitation for public rectification of this information imbalance bore fruit at approximately the time of World War I. 5 Ever since 1915 the United States Department of Agriculture (here- after to be referred to as the USDA) has operated a prog-am to provide farmers with information about daily market prices and conditions. A supplementary program to obtain long range market outlook information is also maintained. The program's intended purpose is ”to aid in the effective distribution and fair pricing of farm products."6 The Department collects relevant information at the market place and trans- site it to either the press wire services or its own leased wire. Eventually, the information becomes public through the mass media of radio, television, newspapers, magazines and commercial mrket letters. These mass media, through the decisions of those charged with editorial —-__ 1.. gm. 5. For a more complete history of the Market News Service and the net-- ure of its Operation see mg" pp. 172-193. 6‘ Ma, Pa 172a .. 3 . responsibility, may use or ignore the material as it is received or edit and supplement it with information collected by their own or other pri- vate sources. As a consequence, the majority of farmers look upon the mass media as originators of market news even though the primary source is the USDA. In 1955 the USDA through provisions of the Research and Marketing Act of 1946 extended a contract? to the Department of Sociology and Anthropology and the Social Research Service of Michigan State Univer- sity to conduct investigations designed to ascertain where and how farmers obtain information regarding current and prospective market conditions, how they use it, and how adequate such information is for helping them make marketing and related production decisions. Accord- ingly, data were collected in the summer of 1957 from 375 farm oper- ators drawn by area sampling methods to represent the farm population of lower peninsula Michigan. A survey schedule administered by interviewers elicited the information necessary for serving these ends. This dissertation is a partial analysis of these data. The focus of the study is upon the "uses" made of market news by farmers. Com- bining somewhat unique methodology with substantive exploration, it will 'interpret' the functions of a particular type of information within Varying structural contexts. The problems which the dissertation “dresses are substantive, theoretic, and methodological. Necessarily these are intimately intertwined. The present chapter will aphasise the first two through a somewhat historic perspective. The next chapter ——-A 7. Contract No. 12-25-010-94. .. 4 - will consider the more specifically methodological problems. magnum MWWMMW Initially the substantive interests stem from the two broad cate- gories, l'effects" studies in mass conununications research and substantive analysis of market news in rural BOClOIOgical research. However, rather than representing a full or simple linear developnent from each, the specific points of interest and the manner of attack must be understood as an attempt to meet certain analytic inadequacies in both areas. Since, in substance, the dissertation draws its greatest inpetus from rural sociology, the pertinent investigations there will be considered first. Although government-supplied market news information has existed a 8 Among the first studies, Green in 1926, O. V. Wells in sociologists. 10 1930 for the USDA, and Youngstrom in 1932 evaluated the accuracy of long term outlook reports for both prices and supply. The substance of relatively long time, it has not received much attention from rural R. M. Green, "Batting Averages in Agricultural Forecasting," 8. install. 91 W. v01. 8 (1926). pp. 174-193. 9. 0. v. None .4 W 9; Meat figtflentg With Sgbgeguefit , Bureau of Agricultural Economics, ted m, Washington, 13.0.8 States Department of Agriculture, 1930, (mimeo). 10. C. 0. Youngstrom, A R the Ac an T e S n , Moscow, Idaho: Idaho Agricultural Experiment Station, Circular 62, 1932. .. 5 .. that research, however, is not relevant here. Market news as studied here refers only to short term and daily price and supply information. A second type of study has been concerned with simply ascertaining the sources from which market news is obtained by farm people. For example, we know that 79 per cent of the farmers in an (1110 sample ”obtained market information from the radio before deciding to sell whereas ”radio served 91. per cent of the hog sellers”13 their hogs"12 Similarly, in the former study, 65 per cent of the in an Iowa study. sample read daily newspapers for market information before selling hogs but only 1.5 per cent in the latter study engaged in this behavior. In these and all other such studies, a theory to account for these findings is not considered. ~ In addition to ascertaining market news sources, some studies have gone on to elicit farmers' evaluations of ways to ”improve" market news reporting and/or have obtained criticisms of specific mass media cover- 11. It can be noted that evaluation studies in terms of economic accur- acy of outlook information has continued through the years. See, @tlook Work: Ihg first :9 193;, Washington, D.C.: United States Department of Agriculture, 1942; John D. Mar, Jr., and Don Paarlberg, "How Accurate is Outlook," Journal 2;: {arm Economics, V01. 34 (1952 ), pp. 509-519; Q Appraisal 9;. New Egrk City Live Poultgz Market Re orts, gag-29, Ithaca, New York: Cornell Univer- sity Agricultural Experiment Station, Bulletin 884, 1952; John F. Beer, "Accuracy of Iowa Farm Outlook Infomation," Joggm_ Farm Economics, Vol. 36 (1954), PP. 143-147. 12. Francis E. McCormick, An Analysis__ of the Market News Service in gido Wooster, Ohio: die Agricultural ”Experiment Station, Research Bulletin 71.4, May, 1954, p. 23. J. Parry Dodds and K. R. Marvin, flog Q9 Iowa Farmers Obtain 3951. Egg 13. Market News?, Anaes, Iowa: Iowa State College Agricultural Experi- ment Station, Research Bulletin 417, Novanber, 1954, p. 129. .. 6 .. age. The level of analysis has paralleled that of simple source enumeration. Throughout these inquiries it is assumed implicitly that market news information is used only for the purposes for which the USDA intends it. The investigations have ignored the possibility that any social arrangement can and often will take on a variety of unintended or unanticipated functions in given social structures. Such possible functions, beyond those intended by the USDA have gone completely uninvesti gated . With but a single exception, all the work has remained at the level of empirical generalizations,” enumerated without any attempts at con- scious and explicit causal explanations for the behavioral differences and/or similarities found.16 Furthermore, there have been no attempts 11.. See for further examples, Market New; Starvices _i_n_ the Midwest, Washington, D.C.; United States Department of Agriculture, 1949; Noah D. Holmes, Commmications Media Through which Iowa Farm er- gtors thain Aggié'ultural Outlpgk Information, Ames, Iowa: Unpublished M.S. thesis, 1951; Listening Survey 9;; Livestock and Other Market Reportp, Chicago: Chicago Producers Commission, December, 191.9, Glimeo). After Merton we take this to mean, "an isolated proposition summer-- 15. ising observed uniformities of relationships between two or more Robert K. Merton, Social Theory and iocial Structure, Fran Dewey we take variables." _. Glencoe: The Free Press, 1957 (rem ed.), p. 95. that, ”empirical means that the subject-matter of a given prOpos- ition which has existential inferences, represents merely a set of uniform conjunctions of traits ... without any understanding of HE! the conjunction occurs.” John Dewey, IONA: 229. ......EZTheO 9.1: Maui 9 New York: Henry Holt and Company, 1938, p. 305. 16. See, Dodds and Marvin, 9p. £159.; mCormick, pp. git-,4 Dodds and Marvin, What Qpes gap 19143 Farmer Want £30m Radio Egrket gaps-'2, Ames, Iowa: Iowa State College Agricultural Experiment Station, Research Bulletin 4.13, August, 1954; Dodds and Marvin, Egg Does 331.2 Mi m Want From Newspaper Market News?, Ames, Iowa: Iowa State College, Special Report, 1951., amineo); Ways 3.9. Improve Market News 52g Information, report of the Phrket News and Inform- ation Conference, Kansas City. Missouri, “33'. 19559 compiled by Gene Futrell, published by Iowa State College, Ames, Iowa, 1955. - 7 - to develOp a coherent theory to codify the disparate empirical general- If any thing, there is a disclamation to phrase research in izations. Such research, therefore, tends to be unorganized extant theoretic terms. in theory and non-cumulative in substance; a distinctly inefficient situation.17 Only one study to data has faced some of the deficiencies thus far indicated. Smithlg sought not only the sources from which farmers obtained market news and the criticisms of these sources which were 19 made but also asked, "How may these variations in use be explained?" One possible explanatory factor was explored. This was "occupational commitment" or the degree to which a person is tied to farming as a It was defined operationally by dichotomizing source of livelihood. (he Guttman scale types for two unidimensional behavioral universes. was the use of supplementary production resources. The relative use or non-use of accounting bookkeeping methods on the farm, of extension specialists, and of Experiment Station publications were the specific behaviors utilized. The second "universe was the nature of the inter- personal relationships maintained with the dealer to whom the farmer 17. This argument follows the exposition of Merton, 92. 333., pp. 95- 101 . Joel Smith, "Michigan Farmers' Use of Radio and Newspaper Market Bulletin Michigan Agricultural Experiment 1. News, " erl Station?3%gil“§§,“2hey,'1956), pp. 612-627; Joel Smith, crggg mg}; g; the Farm egg Mass Commmicgtion, Evanston, Illinois; Northwestern University, unpublished PhD. dissertation, May 1951,, 18. 190 Mth, “Michigan M3138," 22. 21-320, P0 6120 sold his main product. . 8 .. That relationship we dichotomized as "intimate Three levels of and non-specialized" or "distant and specialised." 21 "eminent" resulted when the two scales were cross-classified. Those classes were shown to be highly related to obtaining market news from mass media sources. In every case, the use of mass media as sources of market news is positively and significantly related to the degree of occupational commitment. The greater the commitment, the greater the extent of use.‘< In turning to W in mass communications and its effects, a parallel to market news studies may be observed. Generally, analysis is restricted only to the level of "empirical generalization;' although 23 the meaning of this term needs to be extended to include, after Parsons, repeated observation. The major organizing principle in mass commun- ication research has been the now classic question 'who says what to when in what context with what effect?"24 Following Lazarsfeld, communication research can be categorized then into five general areas: 20. 21. 22. 23. 24. mt!» Matias. 22o silo. po 65. The hm types of cases of high use of supplementary information but 'spocific' dealer relations or low use of supplementary infom- ation and "diffuse" dealer relations were combined as a I'mixed" commitment category. Smith, “Michigan Farmers“, 93. 213., pp. 624-625. Talcott Parsons, The, M m, Glencos: The Free Hess, 1951, p. ‘87. It is interesting to note that while this question is widely quoted, M13 credit for its origination has eluded the present MtOro - 9 .. control, content, audience, functional analyses, and effects. shall not attempt to review systematically the vast literature in each We of these areas. Only the last is of direct substantive interest and only selected cements there are relevant to the study in hand. Most research in mass comunications tends to follow out or em- phasize a single category of the Lazarsfeld schema. There is one notable exception. The notion of effects has been placed in an inex- tricable relationship with message transfer. The chief reason we study this process (of message transfer or communication] is to learn something about how it achieves effects. We want to know what a given kind of communication does to people.2 In no small part, this goal is fostered by the "applied", commer- cial nature of much research in mass coxmnunications' effects. The research goals are limited very often simply to evaluating the rela- tive effectiveness of different situational conditions in obtaining desired goals of the communicators. In fact, the measure used to assess the ”success" of a particular communication attempt is frequently the degee to which an implied or directly stated directive of a message is realised in communicates behavior. For instance, as one of innumerable Onnples, the degree of communication of a "documentary” anti-Fascist broadcast has been measured by the degree to which attitude questions showed that listeners had increased their negative evaluations of -‘ I'tlonlnunication Research and the Social Psycholo- 25. P. F. lasersfeld, silt.'1n U Donni- (edJ. Meat Trends in £4 W. Pittsburgh: University of Pittsburgh Press, 1943. PP. 218-273. 26. Wilbur Schranm, "How Communication Works," in Wilbur Schrsn- (06.), s _. .f .- . .; _ , Urban: University 1”". ~- 0‘ -10... Fascism after hearing the program. This evaluation procedure seems reasonable because 'connnunication" is inplicitly or explicitly defined in terms of action taken upon receipt of symbols not merely as the degree to which literal understanding of symbols has occurred. Given this orientation, an interest in the "effects" of communi- cation generally results in an examination of the degree to which the goals intended by a particular message are achieved in a given popu- 28 lation. Ehnpirical studies of "effects" attempt to explain why a message was or was not effective in obtaining the given goal the message initiator had in mind. Under these circumstances, empirical studies of "effect" in the mass communication literature can be char- acterized perhaps more accurately as studies of effectiveness. A second line of work concerning "effects" has been aptly summar- ized and commented on by Smith as follows: A 27. E. C. Vilson, "The Effectiveness of Documentary Broadcasts,” 29212 See also the gum W, Vol. 12 (191.8), pp. 19-29. emery article by H. F. Lionberger, "The Diffusion of Fhrm and Home Information as an Area of Sociological Research," M W, Vol. 17 (1952), pp. 132-140. Lionberger considers a number of studies attempting to evaluate comunication effective- ness and concludes, "As in previous studies, changes in farm practices attributed to bulletins, and acknowledgement of inform- ation from them, were taken as the measure of usefulness (4-2.: extent of comunicationl.” p. 131.. 28. 9:. the sumary articles of Carl I. Hovland, "Effects of the Mass Media of Communication " in Gardner Lindsey (ed.), Muggézgg m W, Cambridge: Addison-Wesley, 1954, pp. 10 4103; Joseph T. Klap er, “The Comparative Effects of the Various Media," 11”- Schrm add. flafrasassandhfihatsafmmumm, as £13.09 ppe 91’105. . Kate and Lazarsfeld in reviewing comunications studies write, "We 29. are suggesting that the over-riding interest in mass media research is in the study of effectiveness of mass media attempts to influence inions in the very short -- y to change - attitudes and cg run.” See Elihu Katz and P. r. unarsfel . 29312931 Mam c . Glencoe: The Free Press, 1955, PP- 1349- -11.. For some reason, there are few empirical studies available on which to build reliable propositions about the functions of mass communication. This does not mean that sociologists have not concerned themselves with the social functions of connnunications in general, or mass communications Specific- ally. Small, Cooley, Park, Burgess, Simmel and other earlier sociologists concerned themselves with the matter. in However, their writings on the subject are largely 8ression- istic, and serve, at best, as unverified hypothesis.3 It can be added that the gommentggy on the effects of mass commun- ication is suply enormous and in comparison to the inductive research on l'effects" has been extremely far reaching into the latent aspects.31 However, these speculations have not led to many attempts at empirical 32 verification. While highly stimulating, few have atteznpted to bridge the gap between these broad generalizations and empirical verification. For mass communication in general we may summarize that when "effects" are studied empirically, interest remains with manifest function, and when latent functions are focal, empirical inquiry is neglected. The Enlargers a: haaearsh __s.2;______i__Gen film”: one £9.12 .2119 m e t .....rs‘tud In the present study, communication is conceived as being independ- ent of its effects. "Comunication' and the "effects" of communication are viewed as separate matters of inquiry. By communication is meant "the transfer of a set of meanings embodied in a message form in a M 30- Smith. Missiles. 22o 9112-. so 2. 31. See Bernhard Rosenberg and D. M. White (eds.), .1183: Elm: .T_h_e, 1’ gm jig m, Glencoe: The Free Press and Falcon's Wing Press Joint publication rights) 1957.. 32s WBfeld, no me, Po 219e .. 12 . manner that permits it to be received in a preferred way by a specific person or persons." 33 In other words, communication is a process of interaction involving the exchange of significant symbols. A signifi-r cant symbol is defined, in the sense of George Herbert Mead, as any symbol for which both actor and alter share the same mental image. "Effect" refers to any change in the behavior and/or mental condition (including "no change”) of message recipients, as well as to any other condition viewed as possibly being dependent on a given communication situation. This separation of communication and effects has both theoretical and practical support. Perhaps the former-34 can be understood better by considering the following table which summarizes the possible rela- tionships between given communication situations and any possible previously defined consequence. Existence of Specific Communication Situation Yes No Occurrence of Yes Cell 1 Cell 2 Specific Effect No Cell 3 Cell 4 “- 33. Joel Smith, Robert C. Bealer, and Francis M. Sim, 'Commmication and the 'Consequences'. of Communication," paper read at the American Sociological Society Meetings, Washington, D. 0., August 27-29, 1957, p. 3. Contained as Appendix C. 340 Adapted from Igido, pp. 7’8e .. 13 .. If given communication situations relate to their potential effects so as to occur empirically only in cells 1 and 4 then effects can be accounted for by a communication theory and no differentiation is necess- In predicting effects, only those things need be known which are ary. specified in the theory as necessary in accounting for communication. The occurrence of empirical cases in cells 2 and 3, however, suggests Knowledge of the ccmnuni- the independence of these two phenomena. cation situation, by itself, will be inadequate for predicting effects. Information indicated by a theory of communication as necessary in accounting for a communicative situation would be insufficient for pre- dicting the occurrence of the consequence (effect). Undetermined additional information concerning factors and/or relationships super- To simply try fluous for the communication theory would be necessary. and apply a comunication theory to predict effects probably would require both the inclusion of certain variables necessary for explain- ing communication but not effects and the omission of some variables necessary for effects but not necessary to explain communication. The principle of parsimony in theory building would be doubly violated. 35. M. L. DeFle‘ur, "A Mass Communication Model of Stimulus Response An Experiment in Leaflet Message Diffusion," Relationships: W. Vol. 19 (Web, 1956) pp. 12-25; and Kate and Iazarsfeld, 93. m” offer mmples of this observation from among recent researches. 36. N. R. Cohen and Ernest Nasal. in W is Lexis and W W, New York: Harcourt, Brace, 1934, p. 395. -14.. The "practical" reason for maintaining this separation is tied in A simple, hypothetical with the theoretical and is depicted in Chart 1. market news information example is used to bring out the salient features. The depicted situation is one in which there is a fairly wide price differential between two markets. is not as unambiguous as an advertisement for cars, cigarettes, or soap, Although intent of the message it does fairly clearly imply that the intent is “sell at I", all other Of course, in any real situation the “other considerations being equal. Still, within a given “almost equal” factors" are hardly ever equal. strata, some will be found among those exposed to the message who 7 and some will be pre- would explicitly deny the intent of the message, vented from complying by such extenuating circumstances as are illus- trated in the chart. 37. The United States Department of Agriculture maintains only that it wishes to aid in "efficient" pricing. A priori there is no reason then to believe that the "denied" category should be construed as In point of fact, however, an impairment of "efficient pricing." most employees of the Market News Service have their training in economics and tend to accept "efficient" as orthodox economic This position implies man's supreme valuation is upon efficiency. pecuniary reward and accepts the model of man as homg rat10ggggg, if not in feet, then in llto be striven for” goal. This latter assumption validates the educational aspects of the economist's If the farmer is not rational (i.e., desirous of obtaining In role. mximum price) he is acting either in ignorance or in error. either case he is in need of education. See below Chapter III on methodology for evidence supporting this view. For a clear delineation of efficiency types relevant to economics see L. W. Witt, "Economic Efficiency and Social Welfare,“ (mines, undated) Michigan State University and for the broadest philosOphical exploration and mandate of the implied positivistic position demonstrated in this view see Talcott Parsons, The; 9‘ m £93123: Glencoe: The Free Press, second ed.,) 1949, pp. 51-82. , i . Chart 1 - A Hypothetical Market News Situation Mammalian y wag-E23295 Desired action Equivalent meaning / of message (literal /’ (Sold at 1) understanding. / Person exposed to / Prevented (wanted to message) \ sell at x. \ No trans- \\ portation.) \No desired action Comication attempt. Messa e FPrice: at x— l at y" 3080‘ (Implied intent: Denied (sold at y. sell at 1) Does not like dealer at x.) No equivalent esired meaning of message 1/ Action (sold at 1) (Message garbled // because of lack \ of shared meaning \ among those exposed. \ No desired \Action (sold at y) Or, no exposure.) L Time Axis In either case, using "effect" as the measure of communication can lead evaluation programs astray. Lack of desired actions can result either from the immediate communication situation or not. To attribute all failure of desired action to communication mechanisms -- choice of Billbols, syntax, delivery order, nature of appeal, etc. -- inplies that the denial of a message is impossible, and that 'successful' communi- cation and intended consequences can not vary independently of each another. -16... The "practical" aspects of the communication - effects disjunction There are a number of problems currently in- has further significance. volved with communication as defined here without adding the compli- cations to analysis presented by the "denied" and ”prevented" categories. For instance, analysis of market news reports by the Flesch Readibility 8 Nearly Scale show that they tend to be of above average difficulty. one-third of the farmers in an Iowa study 9 reported they did not understand radio reports suffici mtly to feel that they could translate personally such information meaningfully to their own situation, Lg” [.0 Whether they @ng could not Judge what grade their animals were. have used the information in the intended way ,to set grades and thereby allow reasonable price expectations or evaluations of dealer honesty in the process of efficient pricing, if they could have made this trans- lation is a'question apart. It is best answered separately. It should be understood that this argument does not mean that communication and effects are unrelated. They most certainly are, but .... 38. Unpublished United States Department of Agriculture materials. 39. Dodd and Marvin, £93 23 Log, 22. git. 40. Actually whether this (isra problem of communication (as here de- If the farmer who fined) or of effects is somewhat indeterminate. says he cannot translate the information ,say by grade,can under- stand that choice grade hogs are bringing a certain price, commun- ication in our sense has occurred. This assumes that the program message was to convey simply the information that hogs of a certain grade at a given location were selling at a given price range. Being able to reproduce the qualities of a given grade hog could be a communication problem if the program's intent was to convey specifications on what constitutes a given class of hogs. . 17 .. a careful analysis would look on the relationship as problematic rather than necessary. Certain implications of the evaluation that many investimtiam of “effects" can be understood more readily as studies of effectiveness have direct import for the study. In large measure, the study of ”effectiveness” commits the investigator to accepting as his evaluative criterion only the intended or manifest consequences desired by the communicator. Merton's observations on this point are particularly cogent. What is his [the sociologistb] task if he confines himself to the study of manifest function? He is than c oncerned very largely with determining whether a practice instituted for a particular purpose does, in fact, achieve this purpose ... He Will ask whether a prepaganda campaign has indeed gained its objective of increasing "willingness to fight" ... so long as sociologists ggpfigg themselves to the study of manifest functions, their inquiry is set for them by practical men of affairs, rather than by the theoretical problems which are at the core of the discipline ... with the concept of latent function, the sociologist extends his inquiry in those very directions which promise most iir the theoretical development of the discipline. In this light it is to be eacpected that sociological investigations should be concerned with more than the determination of the effective- ness of market news programs in reaching farmers, or of ascertaining ways in which to implement the intended effects of rational economic decision linking relevant to the farm enterprise. A sociological anal- ysis of market news should be concerned with both the manifest m the latent effects of cammunication.acts, systems, or structures. -__. a. “non, 99- fig, ppe 65’660 fr! '1} . 13 - If "effects" and "effectiveness" were coterminous in fact, much of mass communication effects research would still be inadequate. Fearing indicates the crux of the difficulty when he notes, "many widely held notions about the communicator—content-effects relation- ships, expecially in the mass media, grossly oversimplify the problem)“; Much research in this area is posed in essentially unanswerable forms. In part, Fearing recognizes this when he asks rhetorically: How are the effects of mass media influenced by the curacter and content of the comunication? In answering this question one must consider both the types of appeals employed and the arrangement of the elements sequentially.43 A stronger recommendation than this, however, needs to be made. A sharp delineation of the "conditions under which" the "character and contentu are to be observed would help to resolve much of the ambiguous evidence on mass communication ”effects.“ For instance, there is uncertainty on the relative merit of "type" of appeals to win support for a measure.“ Hovland indicates that while two studies show "superiority of emotional 45 propaganda over logical argumentation" the remaining available have I'failed to find clear-cut superiority of one type of appeal over the 42. Franklin Fearing,) “Social Impact of the Mass Media of Communication," ed inN.B. Henry .),.Ma§§__LMed m______nEducatio. Jmm assisting them... of Educating Chicago: University of Chicago Press, 1954, p. 171. 430 c. Ie H071“, no we, pe 1075e U» Mo 45. mg. hi Ii. 5! d l ~91 .I 'u ‘5 .. 19 .. other)“. He cogently comments, ”The lack of consistency in outcome suggests the need of further research _o_n 3.13.9. conditions which affect the relative advantage of the two types of appeals."47 In the present context, this comment should have the effect of directing inquiry into the effects or consequences of market news in- formation into tightly delimited bounds rather than in the broad terms often found in effects research. Hovland's general directive to ex— plicate the ”if-then" context of all "effects" studies can and should be extended. In its germ is an outline of a general method for studying communication effects which contains correctives for the shortcomings of the investigations thus far examined. The reference is to functional analysis. After briefly examining its major tenets, it will be possible to delimit specifically the problan for this dissertation. W Mg: 5 fiamework £9; Stugflng Market News Effects. Perhaps the most programatic statement on the functionalist frame- work for sociology is in Merton's workf8 He indicates that "the central orientation of functionalism - (is) expressed in the practice of interpreting data by establishing their consequences for larger structures in which they are implicated."49 Specifically then, "the theoretical framework of functional analysis must expressly require that there be M of the m for which a given social or cultural it. is functional."50 1.6. 221.51., p. 1076. 47. m. (Italics mine) 48. “Itch, 22. Mo, pp. 19‘&. 49. 1mg” pp. 46-47. 50. Iii-£09 Po 3°. -_ .l k. .. 20 .. In contrast to the more delimited notion of functionalism in anthropology, Merton argues that the unit to which functional importance is to be attributed M £93, be society in its totality. It is necessary 'to consider a mggg of units for which the [any] item has designated consequences ... Terminologically, this implies the concepts of psy- chological function, group function, societal function, cultural function, etc.'51 In delimiting item(s) to which functions are inputed, Merton in- dicates that ”the basic requirement is that the object of analysis represent a W (1.3., patterned and repetitive) item."52 And then, in assessing the functions of the unit, we are required "to intro- duce a conceptual distinction between the cases in which subjective aim-in-view (of actors) coincides with the objective consequence, and the cases in which they diverge."53 In other words, a distinction needs to be drawn between manifest and latent functions. The former "are those objective consequences contributing to the adjustment or adaption of the system which are intended and recognized by participants in the 5 systom.‘ 4 Conversely, latent functions are ”those which are neither 55 intended nor recognised.“ 51' m" P0 520 52° may P0 50- 53. m.) p. 510 54. Mo 55. mg. .3 oil .. 21 .. 56 This abbreviated general outline will be the framework used for analyzing the effects (functions) of market news systems of information in this dissertation. Functional analysis as described here has no necessary substantive bounds. "The central orientation ... (of) in- terpreting data by establishing their consequences for larger structures in which they are implicated - has been found in virtually all the sciences of man.“ Functionalism as this study uses it ‘8 represents a general analytical procedure.’ ct ona Alternative ”8 2113 "Hmthggig' m2 Mia; W a: " 91 its aim: - One of the basic tenets of functional analysis is that "the range of variation in the items which can fulfill designated functions in a social structure is not unlimited.’I There is structural constraint. In turn, this inplies the notion of a system of various "parts" in some determinate complex of interconnections. To use a simple mechanical 56. A number of other details of functional analyses are taken up by Merton but are not germane at this point. 57. Merton, 92. m" pp. 46—47. 58. This is not to represent the field historically. Barber has correctl observed that functionalism (which he calls structural in: in part a body of substantive sociological concepts analysis and theories in part a method . . . of analyzing the relations B. Barber, ”Structural-Functional among structural parts.‘l Analysis: Some Problems and Misunderstandings,‘ Amegicag Sggigu 1.22121 series. Vol. 21 (April, 1956) p. 130. 59. Merton. 22- 21.2., p. 52. 1‘7 .. 22 .. analogy as an example, gasoline injected into a diesel engine would stop the Operation. The some fuel injected into a gasoline engine would start or continue operation. The difference in function could be "ex- plained" in terms of structural context, 1.2., the articulation of the given set of relationships (the,engine) with the "new" element. Simi- larly, within a growing season it is manifestly impossible for a sugar beet farmer who has contracted his crop at the beginning of the year to have market news function in the making of a decision as to when or where to sell at harvest. However this may be the crucial role of market news for a wheat farmer. equal, variability in structure should result in different patterns of This argument It seems reasonable that, other things effect when the same or similar message is received. rests on an extension of the postulate of functional alternatives to The particular function of a given item is the its more general form. This postu- result of the structural context in which it is entered. late in its general form provides the major orientation of this analysis. At the outset it should be emphasized that the general "hypothesis" of the study is not a hypothesis in the strict sense, but rather a ”sensitizing” orientation in the manner which Blumer distinguishes de- finitive and sensitizing concepts. Whereas definitive concepts provide perscription of what to see, sensitizing concepts merely suggest directions along which to look . .66 they rest on a general sense of that is relevant. Here the former meaning applicable, the study could have proceeded with “Plicit and precise W operational or working hypotheses M 60. Herbert Blumer, What is Wrong With Social Theory," M m... 12am lanes. Vol. 19, (February, 195/.) p. 7. A: I'. .. 23 .- translating the general directive to specific substantive references. This was not the case both because of the paucity of previous research in market news (particularly with reference to latent functions) and 61 the lack of adequate theory explicating communication as such. Primarily this study is meant to be exPloratory and demonstrative. First, it intends to ascertain the efficiacy of certain selected struc- tural dimensions for differentiating market news functions. Second, given positive results in this endeavor, the study will examine in a functional framework certain of the specific functions with respect to delimited structural contexts. In both, the study can only demonstrate 62 A definitive sub- what is possible methodolOgically and analytically. stantive setting out of the relationships between structure (as will be defined below) and market news functions is impossible. The small sample size as well as the necessary lack of depth inherent in survey research (the research model used for the more general study) preclude this latter goal. This limited goal is not without significance. Merton has cogently assessed, "all interpretative schemes . . . depend upon a triple alliance 61. Snith, Sim, Bealer, 29. git. ”Demonstration” will take on broader meaning in the next chapter There an attempt when the model. for data analysis is examined. Will be made to reconcile the "tests of significance" school and the recently revocalized "cross-tabulation, causal analysis" See, Hanan G. Selvin, "A Critique of Tests of Signifi- school. cance in Survey Research," Amegican Sociologigl Rgflew, Vol. 22 (October, 1957) pp. 519-527. 62. - 24 .. between theory, method and data. Of the three allies, method is by all odds the weakest.“63 This is particularly true of rural sociology both in the specific area of market news and in general.64 The present study is oriented specifically toward workers in this field. 01’ course, the study is equally relevant for mass communication effects analysis. One further word about the hypothesis of the study is necessary. The argument that a "difference" in social structure prior to the intro- duction of an item will result in a functional difference does not mean that an item will function in the same manner in one and only one type structure or that function is treated as a dichotomous all-or-nothing phenomena. Rather, one of the intended by-results of the dissertation is to gain some insight into the range of structures for which a single item can serve the same function. The line of attack preposed will look at a we; 22 different structures simultaneously at the same point in time. The functionalist generally is concerned with the other aspect 0f the time-space axis, 5 m 9;: W through time or in articu- lation with different type of structures in the institutional sense. Md 63. Merton, pp. 9333., p. 19. 64- 569 W. H. Sewell, "Some Observations on Theory Testing," Rural W, Vol. 21 (Wrch, 1956) pp. 1-12. The appraisal article by Taves and Gross, "A Critique of Rural Sociological Research, 1950" 133:; Sogiology, Vol. 17 (June, 1952) pp. 109-118 gives incisive evidence on the point. Chapter II The Schem for Analysis: General Preperties Dimensions 9; Stacture finals]. Definition In functional analysis data must be interpreted by establishing their consequences for the structure in which they are implicated. Levy has defined structure as "a pattern, 1.g., an observable uniformity, of action and operation"65 and, notes that, IIthe general form of this con- cept is deliberately left in to cover a wide range of possibilities from highly stable uniformities to highly fleeting ones."66 He further indi- cates those aspects of the definition which have significant methodolog- ical implications: Structure ... refers to an aspect of empirical phenomena divorced from time. The patterns of action, gyg patterns, do not exist as concrete objects in the same sense that sticks and stones do. The patterns of action in this sense are abstractions from concrete empirical phenomena, and they “exist" and are "empirically verifiable" in the same sense that the sq eness of a box “exists" and is “empirically verifiable." 7 Thus. "ear elem no.1 contain an element indicative 9; g gtructure ingofar 93 £3 .13 considered with reard 22 its nonunigue aspects 9}; character- istics."68 —__ 65. bhfion J. levy, Jr., gigs smctpge g: Society, Princeton: Princeton University Hess, 1952, p. 57. 66' M" P- 58. Persons puts it this way, 'Structure does not refer to any ontological stability in phenomena but only to a relative stabil- ity - to sufficiently stable uniformities in the results of under- lying Processes so that their constancy within certain limits is a workable pragmatic assumption." Talcott Parsons, ”The Present Positim “‘1 PrOBPecta of Systematic Theory in Sociology,“ in m 93 again-- Lhm (revised edition), Glencoe: The Free Hess, 1954, 9217- 67: My, 220 21:10: P0 57' 68. L113. (Italics added.) .u an. - 26 - Functional analysts have usually defined structure in terms of institutions or similar specific concepts reflecting order, for functional analysis has generally used society as the point of refer- 69 Howaver, there is nothing in the logic of functional analysis This has, ence. to prevent the consideration of structures of lesser scepe. in fact, been done. Merton, in discussing middle range theory, uses as favorable examples of functional analysis studies taking as their points of reference empirical phenomena of less global content than “society. "70 Recognizing that they are addressing society as the unit of analysis, Bennett and Tumin's comments on structure are important as guides in the current study. The sociological name for position in society is “status". This is to be distinguished from the common-sense use of the word where status is held to mean "prestige." ... In all human societies we find statuses based on differences in sex, skill, economic productivity, power, dependency, education, age, strength, attractiveness, marital status, parent- hood, and native birth among others....If we can think of these various possibilities as building blocks, then we may note that the social structure of any society consists of a selected number of these build- ing blocks arranged in such a way that together they form a total body of relationshi s to which... are attached different rules [modesjpof behavior...71 *— 69. See, and T. Parsons, £133, Social 5y__s_tem, Glencoe: The Free Press, 1951; D. F. Aberle, gt. $1., "The Functional Prerequisites of a Society,“ E32123, vol. 60 (January, 1950) pp. 100-111. Note also the cments of Merton 92. 91.3., pp. 25-30, 52. 7°. Merton. an. sue im- slis, pp. 55-82. 71-. M. Tunis and John Bennett, $2.15!: M: W gag Funct New York: Knopf, 1949, pp. 87-88. 'e e a. 0 :0 h. ’1 I. k or. - 27 - It is further emphasized that: The status is not the same as the individual who occupies it...Status is not something tangible or perceivahle by the senses; it is a way of thinking about and describing social phenomena which helps us make sense out of social life. It thus becomes possible to distinguish between the position and its occupant.72 In the present case, the dissertation is concerned with the "status" factors (called "structural dimensions") in which a given individual is implicated and which constitute his "structure" at a point in time.73 This interest is in the pattern and/or ”status", (1.§., the ”position“ in Bennett and Tumin's terms) however, not the individual. The study will focus on certain of those "status" factors simultaneously and examine them for their relationship to market news, holding, of course, that the nature of the relationship (function) will be conditioned hy the structural arrangement of which it is a part. angidgzatiogg‘frgm Egg Rgfiegggh Desigg Just what empirical events the notion of structure should refer to even in the limited sense in which "functional" analysis is used here However, a number of guides are present. In the first is problematic. It is intended place, market news information is highly specialized. 72' M. Ppe 89’900 73. tfllternative terminology which may be somewhat less troublesome is available in.A. H. Barton, "The Concept of Preperty-Space in Social Research,“ in Lazarsfeld and Rosenberg, (eds.) Thglggnggggg 2; m Eggegzgh, Glencoe: The Free Press, 1955, pp. 40-53. While in essential agreement with this, that exposition fails to convey the sense of variable interrelationships so lacking in market news studies and so greatly implicated by functional analysis. .\ po - .- vi .5 Ga Vb -28... to serve an occupationally specific audience, that is, farmers and business personnel involved in processing and marketing agricultural Restriction of "structure“ to the indiVidual farmer was commodities. Furthermore, the contract directed that such an imediate possibility. Accordingly, the range of functions that could be a group be studied. considered was limited by this restriction in the “structural" dimensiomu Within these limits further considerations prevailed. The aspects of structure had to be universally applicable in the sense that all study participants could be categorized in their terms frcn indices Such univerwl obtainable under survey conditions of data collection. categories would promote continuity in research if, with apprOpriate changes in operationalization, they defined variables important for describing and mediating behavior in occupational enclaves beyond ag— riculturefl'5 This was attempted, as will be taken up later. or somewhat intermediate concern were over all research purposes. The pertinence of separating "connnunication" and "effects" was discussed previously. Here it can be noted that the dissertation is meant to a, ’ m 74. For a fuller examination of these exact research limits steming from the joint, "applied", contract financed aspects see Joel Smith, F. M. Sim and R. C. Bealer, "Research Design in Structural- Functional Analysis: A Case Study from Applied Research on a Comunication System," paper read at the Ohio Valley Sociological Society Meetings, Columbus, Chic, April 26-27, 1957 and contained in Appendix C. . The effects of trade Journals on the structure and functioning of industrial concerns is one such area. Given both the magnitude of these information sources and the importance of knowledge in cm- petitive organization such research would seem substantively 75. significant. e" e" a I”. '3' ’0 J stand in a complimentary relationship to a separate analysis of commun- ication behavior, 1.5., eXposure to and understanding of market news information. Hence, the dimensions of structure also were selected for their possible cogency in differentiating eXposure to media for market news. By utilizing the same sets of independent variables in each analysis, insight can be gained as to the extent to which the variables of a communication model can account for the consequences of communi- cation . Examination of the independence of these two types of questions is not, however, a part of this dissertation. Finally, by way of theory, a functional approach to the question of I'effects" implies commitment to the tenet that behavior is systemic It is manifestly impossible with present knowledge in its determinants. As Parsons to control for or examine all possible facets of a system. indicates, it is this imperfect knowledge which forces us to what he calls structural-functional analysis. In line with Merton’s usage we call it simply functional analysis. 76. For fuller explication of this problem see Smith, Sim and Bealer, "Communication", 92. 23;. 77. Parsons, “The Positions and Prospects", with recognition that his referent is orthodox functionalism, pp. Li}. pp. 215-218. He puts it this way: “The ideal of scientific theory must be to extend the dynamic scope of complex systems as a whole as far as possible... [and] the essential feature of dynamic analysis in the fullest sense is the treatment of a body of igterdepengent phenomena sim- ultaneously, in the mathematical sense .. The ideal solution ... state ] all the elements of reciprocal interdependence betwoen all The ideal has ... been attained only the variables of the system. All other sciences are limited to a in ... analytical mechanics. more primitive level or systematic theoretical analysis. . .. [For sociology it is] the functional reference of all particular con- ditions and process to the gtate of the gystem 35 5 going concgzg which provides the logical equivalent of simultaneous equations in a fully developed system of analytical theory.‘I See also Par- 80118, The .....islSoc m. pp. 483-484- l 0. '- Given these limitations and considerations it was necessary to delineate and define Operationally as structure those aspects of systems which were both relatively independent and significant differentiators of behavior.78 ‘While relative independence is necessary if knowledge and prediction of behavior is to be maximized, the "independent" var- iables could not be completely unrelated or the assumption of a system 79 that is basic in functional analysis would be violated. In this light, two broad research Options were immediately apparent. One involved the omnibus task of searching out empirically all the var- iables that might best explain the variance in behavior resulting from the receipt of market news. In this case one would use data on as many factors as possible to obtain the most efficient predicting equation. This approach is endless in its scope because it lacks a theoretical orientation. It would be no improvement over the research previously 80 reviewed. The second alternative, while not ignoring substance, would focus less explicitly and primarily on the particulars of the sample in select- ing“variables to explain any differential distribution of market news 73. This is the same kind of thinking which basically underlies Parsons' use of primacy in defining and interrelating the modes of orien- tation and the pattern variables. See T. Parsons, Th3 Social m, 2H- siio, pp. 58 ff. 79. tMerton,‘gp.lgit. 30. Merton, gp. p_i_§., pp. 96-101. See Joel Smith and Francis M. Sim, hfichiggg's Lower Peninsula Farmers' Cons tion and Use 2; Market Newg, East Lansing, Michigan, (mimeol, the research report to the Uhited States Department of Agriculture for a fuller probing of variables differentiating functions of market news. 1.8 “ u I”. - 31 - effects. The central analysis would shift from merely explaining the particular sample variance to using the sample as a means for also examining certain theoretic and methodological propositions. ‘As in- dicated in the previous chapter, the dissertation follows this second approach. The specific substantive variables selected were obtained from the research legacy of mass communication studies. Katz and Lazarsfeld have succinctly observed that : ...The intellectual history of mass media research may, perhaps, be seen best in terms of the successive intro- duction of research concerns - such as audience, content and the like - which are basically attempts to ippptg effects by means of an analysis of some more readily accessible intermediate factor with which effects are associated.81 These authors cite four such general sets: 1) exposure, access, or’attention - variables of audience analysis; 2) medium, here it is assumed that effectiveness is modified by the channel which delivers the message; 3) content — in the sense of form, presentation, language, order, etc.; 4) predispositions - attitude structure of the communicatees involved. Each has been found important in mediating the effectiveness Of communication.82 'While these have been derived largely from studies concerned only with manifest consequences, they are relevant to the present study because it is concerned with manifest as well as latent functions. Furthermore, there are no comparable guides for the study of latent effects. An effort was made to Operationalize structure by taking into consideration each of these four general categories. 31. Katz and Lazarsfeld, pp. g;§., p. 20. 32- 170.1.qu pp. 21-25; Hovland an. git" pp. 1071-1099. 1.1.. ‘V i". In ...w B}. - 32 .. The mm Earlene: At the study's outset it was assumed that few if any persons who qualified as "farmers" by the study definition, and who were thereby eligible to be included in the sample83 would not receive some market news. It turned out that there were only 9 of 375 interviewees for whom no evidence of the receipt of market news could be uncovered. It was possible, therefore, to control for "attention" or "eXposure" without losing many cases by restricting the analysis to persons who were exposed to and received market news information. (he consequence of the argument that separates communication from its effects, is that the Katz-Lazarsfeld category of “content" could not be considered as an aspect of structure in the analysis. Rather, the item to which functions are attributed is market news information. For methodological rigor, the messages of which market news consists should be the same for all individuals if the variability in the dependent variable (uses of market news) is to be attributable to structural difference. If not, variability in functions may be attributable to my; differences in content or in structure. Therefore, content should be controlled and cannot be an independent variable. Like content, the medium through which market news information is received has an important bearing on effectiveness. Nevertheless, it could not be considered as an independent variable in this design. The logic behind the definition of structure is consistent with the argument for a separation of communication and effects. In the present study the M 83. See chapter three for the specifications on sampling. fact of communication is taken as given and questions refer to phenomena that may follow from.that fact. Therefore, to the extent that the med- ium is a significant part of communication, it should be controlled and not considered an independent variable. This leaves only the general "predisposition" category of the Kate and Lazarsfeld schema as a source from which to draw the independent variables. Hewever, in the study cited,Katz and Lazarsfeld also suggest a fifth general category: s..it appears that communications studies have greatly underestimated the extent to which an individual's social attachments to other people, and the character of the Opinions and activities which he shares with them, will influence his response to the mass media. We are suggest- ing, in other words, that the response of an individual to a campaign cannot be accounted for without reference... to the character of his interpersonal relations.34 The dimensions of structure were framed with the guidance of these two general categories. As previously indicated, market news is occupationally specific and specially meant for the occupational or work world. In general terms, there are at least two dimensions to gay niche in the economic divisions of labor. 0n the one hand there is the actor's conscious awareness and evaluation of his own psychological relationship to the Job. The work role is defined by ego as satisfying or noxious; as a source of pres- tige cumulation, affective indifference or disgrace; as something to be defended or rejected. Since men act in terms of their definition of the M 84. Kat! and IazaI‘Sfeld, 23. me, P. 250 .1 I1 - 34 - action situation,85 ego's relational definition to his job is of theo- retical and empirical importance. "The social world of the average adult is primarily patterned about work activity ... The satisfactions of daily life are largely bounded by the rewards of work. And so, too, the dissatisfactions.86" This social psychological orientation to one's job will be referred to as 'occupational involvement." It is manifested in the degree to which a person has a preference for and identifies with an occupation. Some form of this relationship must exist for all occupations. A complementary feature of the job, always present is the degree of “occupational commitment.“ By this is meant the comparative degree to which factors other than those of psychological preference tie a person to farming (or any other occupation) at a given point in time. In this sense the variable is Specifically intended as a residual category as defined by Parsons.87 The use of occupational commitment 85. For the clearest statement of this position in social psychological terms see Walter Coutu, Emeggent Humag Naturg, New York: Knopf, l9“; 86. D. Miller and w. Form, ggdggtggel geiglegz, New York: Harpera, 1951, p. 7. Haire observes that, "There is probably no other field in the general area of social psychological problems in industry in‘which there are so many publications as there are under the general heading of morale.“ M. Hairs, “Industrial Social Ps chol- ogy,‘ in Hangpggk g; Socig; szghologz, Gardner Lindsey (ed.§, Cambridge: AddisonAWesley, 1954, p. 1104. ‘Morale and involvement tap quite similar phenomena. 87. Persons, Stggctgg, gp. git... pp. 16-20. - 35 - in this study pivots about the idea of economic cost. At least two highly related facets of the phenomena can be identified. These are the actor's position relative to present and future facilities88 within and outside the given occupation and the articulation of the job with such other non-occupational factors as, for instance, the point in the life cycle. Occupational commitment as described here has sharp over tones of involuntary or non-recognized coercion. Its meaning will be explored further in the next chapter. or course, involvement and commitment can vary independently. For example, a young, part-time farmer, factory employed for the major part of his income may be highly identified with agriculture but not partic- ularly committed to it presently (though he may desire to be in the future.) A full time farmer highly committed to agriculture because of advancing age, lack of education, a growing family, and an inherited dairy farm.may or may not identify with farming. There are both happy and disgruntled dairy farmers. Furthermore, it seems reasonable to expect that variation in either involvement or commitment will produce significant effects on job performance and, hence, on the utility and functions of the occupationally specific market news information. A third type of “predisposition" used in this study is the ration- ality of the individual in his orientation to the job. For presenta- tional purposes, it will be called simply “rationality," and will refer to the degree to which the individual farmer behaves in ways which, M‘ 88. C. P. Ioomie and J. A. Beagle, M W, Edgewood Cliffs, New Jersey: Prentice-Hall, 1957, p. 7. ...... - 36 - from economists' perspectives, should maximize income from the farm. For instance, some farmers go out of their way to use the latest improved farm technologies and practices. Others make no efforts in these directions and farm by the "seat" of their cultural heritage.89 Since new technologies often are more efficient, they allow a greater return per unit of input. Other things equal, the person who adOpts a new technology should accrue greater income than the person who does not. In this terminology, adoption would be construed as "rational" while non-adoption would be I'non-rational.“ The last two structural dimensions used in the study were drawn from the area of interpersonal relations. At least two broad aspects of this area are variable; first, the character or quality of the relationship, and second, the types of statuses to which the status in question must be related. Both may vary widely. However, Smith has demonstrated the relevance of a diffuse-specificity definition of in- terpersonal relationship to dealers and hired labor for explaining variations in some patterns of exposure to market news.90 While it would be desirable to use the same variables in the present study, such direct replication was not possible. Since over half of the respondents (54.4%) used no hired labor, this status could not be used because it did not meet the criterion of applying to all sample respondents. In contrast, the residence pattern imposes some form of relationship with m 89. See for example J. H. Cepp, ”Toward Generalization in Farm.Practice Research," Burg; M21231. vol. 23 (June, 1958) pp. 103-111. 90. Smith, ”Michigan," 93. 93.; J. Smith. mm. as- sif- H II! - 37 - neighbors on all respondents. Mereover, a number of studies have demon— strated the important part played by neighbors and peers in communi— cation.91 In addition, "no relationship" with neighbors can be ascribed a meaningful polar position on a diffuse - specificity con- tinuum, while ”no relationship" with hired labor is.more ambiguous because it may be either a response to the specificity desired of the relationship or, an economic inability to support the conditions to have the contact. Therefore, relationship with the status "neighbors" was substituted for that with the status "hired labor.“ It was decided to include two statuses because it had been found that the tone of interpersonal relations could vary independently between dealers and hired labor.92 The following list summarizes the disposition in this research of the "sets" of variables for effects. 91. See for instance, Katz and Lazarsfeld, gp. gi§.;.A. Lee Coleman and C. Paul Marsh, nDifferential Communication.Among Farmers in a Kentucky County," Rpggl Sociology, vol. 20 (June, 1955) pp. 93-101; Herbert F. Lionberger, "The Relation of Informal Social Groups to the Diffusion of Farm Information in a Northeast Missouri Farming Community," Rgzal Sgciology, v01. 19 (September, 1954) pp. 233-243. 92. Smith, gzggnization, pp. 213. ‘e In 'A- ‘A‘ '1 2. 3. 4. 5. Dameggion of Effect Audience or exposure 1. Predispositions 2. Intervening diffuse groups 3. Media (sources) 4. Content 5. Dgginitiog g; ction Place in Design Control. Study includes only those eXposed. Independent variables - rationality, involvement, commitment Independent variables - dealer and neighbor relations Control Control Function is defined as a “condition, or state of affairs, resultant from the Operation (including in the term Operation mere persistence)of a structure through.time." 93 In less formal terms, nFunctions refer to what is done and structure to how (including in the meaning of "how" the concept "by what") what is done is done.“’4 0 The empirical phenomena seen as dependent (i.§., functions) were a variety of possible uses that market news might be put to by farmers receiving it. utility or Operate toward several ends. ion in: 1.) 2.) 3.) From the USDA viewpoint, market news is intended to have These include use as a criter- making economic decisions about such aspects of marketing as when, where, how much, and in what form to sell, and what price to expect or charge. making decisions about changes in production plans. evaluating the honesty of dealers and the efficiency of local marketing arrangements 93. Levy, 93. git" p. 56. 940 we, pp. 60-61. - 39 - In addition there are a number of uses not specifically intended by the USDA to which market news, nevertheless, could be put. A sel— ected number Of these conclude the dependent variables. These include the possibilities that it may be to: 1.) act as a resource in interpersonal relations i.§., as a conversational gambit. 2.) reinforce occupational involvement. 3. provide new ideas and concepts - to change rationality through education. A.) serve a mental health function by Offering a target for hostilities derived from dissatisfaction in other areas. 5.) provide a criterion (among others) for ordering persons in a social prestige order for the community. 6.) permit surrogates to perform the task of Obtaining market information. 7.) evaluate partially the performance of the USDA. The range of both types of functions considered was not intended to be exhaustive for the respondents. Functions were chosen only to possibly articulate the independent variables and to demonstrate their breadth and, thus, were not meant to be exhaustive. The distinction between intended and unintended uses corresponds in part to Merton's manifest-latent distinction.95 Categorization was determined by the intent of market news as seen by the ultimate commun- icator (and client) - the federal government agencies. The functions, however, were defined in their relevance for the recipients — farm peqpkt Thus, the manifest—latent distinction may not yield the same result when 96 the point Of reference is changed from communicator to communicates. ____ 95. krton, 2p. 9_:l._‘l;., pp. 60-82. 96- Though no direct evidence was available bearing on the point it was the writer's impression that, following Merton's definition of la- tent es unrecognized by the participants 1.5., farmers, the dis- tinction as made was not broad enough. For many respondents the latent functions included not only those here defined as latent but manFof what are here termed manifest functions. While this is not crucial to the design, it does limit the research, 0 particularly in terms of Merton's functional paradigm.’ A final aspect of structural-functional analysis relevant for con- sideration here is that: .... patterns of action are themselves the results of the Operation This of other patterns, and in this sense they are functions. consideration points to a special characteristic of the referents The same empirical of the concepts of function and structure. phenomena may be an example of either a function 05 a structure depending upon the point from which it is viewed.9 What is a "function" at one point in time may at a later point he "structure.” In the present study, at the time of sampling, persons are All persons in a structural categorized as having a certain structure. type show certain distinctive behavorial and social characteristics. Each type differs, at least in these selected respects, from the others. However, each category contains a mixture of peOple, some who would always have been so classified and always will be, and others who may be moving into or out of the category in which they happen to fit at the Of particular interest, in this respect, is the fact time of the study. that the occurrence of some of the functions under study here can result in.a change in the future structural placement of an individual.99 Uh- fbrtunately, the extent to which these functions may have this effect 97. Merton, 99- EU... p. 51 gives as a "basic query,""What are the effects of the transformation of a previously latent function into a manifest function?“ 93° “'7, .22. 24.3» p. 61- 990 366 P..F. Lazarsfeld and R. K. Merton, ”Friendship as Social Pro- cess:.A Substantive and Methodological Analysis,” in.ML Berger, 21‘:- alo. means as M tr in ways £29223. New York: Van Nostrand, 1951., pp. 18-66 esp. pp. 37-55 for discussionct the Point in a somewhat different substantive context. .‘J- tn *1 '1 Is: ‘ 1‘? - 41 - cannot be ascertained from the present study. However, the fact that they can have this effect is abundantly illustrated by the fact that two of the possible functions are changes in the level of commitment and/hr involvement. It should be made clear that, given the technique of a cross-sectional sample survey, the only question the study can address is that of some possible roles of a single facet of structure (market The answer can be news information) in varying structural systems. Causal analysis only in terms of correlation rather than causation. requires a necessary time dimension which the study did not obtain. The inability to bring conclusive data to bear on possible alterations on structure of a given "function" is one of the many limits of the research. .....eSom _srslGen Bream 2.11 inalzsis To this point the discussion has indicated certain analytical short- comings in "effects" studies generally and in market news studies It has been argued that a limited type of functional specifically. Within this framework a rationale analysis will meet those inadequacies. 1.) the dimensions of has been given for selecting and defining: structure as independent variables; 2.) the unit to which functions are to be imputed -.market news information; and 3.) the functions of in- terest as dependent variables. In bringing these three items to bear— om the study hypothesis, a number of related considerations have to be taken into account. As has already been indicated, functional analysis is based on the tenet of interpreting data by establishing their consequences for larger - 42 - structures in which they are implicated. Buckley has commented on what this has meant substantively in sociology. ... integrated social wholes are analyzed into structures and functions; the term "function" is used primarily in its nonemathematical sense; and the focus is primarily on the consequences of the operation of structures for the "sur- vival" or "maintenance'I of some specified or unSpecified state of the whole.100 In this light Merton is led to the distinction between “function" and "dysfunction"; the former, "those observed consequences which make for the adaptation or adjustment of a given system'; the latter, "those ... which lessen the adaptation.“ This distinction supposes some reasonably crystallized set of be- havioral phenomena which may be designated as the system. Generally this means professionally conventional and not in any sense "given" in the existential world and directly perceived. This is vividly recog- nized when it is noted that: ... persons are no more concrete entities than are groups... we cannot "see" persons any more than we can see groups: both are realities which extend beyond the range of human perception. Both 923 abstractions from and summaries of10 our observati“ons"'or r222 was arrests 2.1: he. Lire-a1 t . 2 103 y and In this very important aspect of consensual assignment to realit 100. ‘Walter Buckley, "Structural-Functional Analysis in Medern Sociol— ogy,“ in.Alvin Boskoff and Edward Becker (eds.) Modern Sociolog- igéiylhggzz, New York: Dryden, 1957, p. 249. 101. Merton, 93. 91,5" p. 51. 102. C. K; warriner, "Groups Are Real: .A Reaffirmation,"lgmg;igan Eggiglggiggl‘figzigg, vol. 21 (October, 1956) p. 552, italics 103. See Coutu,lgp.‘g1§., particularly pp. 200-209. . I A -‘43 - percept, that which is considered to be a'btructure" in sociology is an empirically "observable" whole - for example, a ward political club, a city machine, a national political party. In short, functional analysis 104 conventionally presupposes boundary maintaining social systems. Then, in the example, one can be in a favorable position to assess the 'net balance of an aggregate of consequences"105 of a phenomenon like patronage. That is so because there is a backlog of interpretation and research cast in terms of reality perceived in this certain way. The present study does not deal with a structure as an empirically ”observable" whole in this same sense. The empirical referent of structure (i.§., "system”) is a-conventional for functional analysis. This is so in two senses. First, as indicated above, the study defines as structure phenomena less general than society or other organized Second, in focusing on combinations of selected status— social systems. as, the customary ”sort factors" or "social correlates" as variables of "social structure" are utilized in the study not as significant or ob- jects of interest in and of themselves. Rather, they are interpreted 104- Parsons, The £29121 M. 9.2. 21.2.. pp. 482-433. A150: loom-’18 and Beegle,lgp. 913., chapter 1. 105..Merton,‘gp.‘git., p. 51, offers this process to counteract the inp cisive criticism that functionalism (if not functional analysis) tends to view all items in a structure as functioning positively 1.3., toward maintaining an equilibrium) and, hence, tends not to egg., Buckley, 92. £13.; A. W. handle the problem of change. Gouldner, "Some Observations on Systematic Theory, 1945-1955,“ in Ho L- Zetterbers (ed.) ____J._aSocio o in the 1121.10.22 gains 22 America. Paris: United Nations Educational, Social and Cultural Organi- zations, 1956, pp. 34~42; Carl C. Hempel, "The Logic of Functional Analysis," in L. Gross (.d.) W as W T_h__xeor . Evanston: Row, Peterson, 1959, pp. 271-307. rn !"l I] - 44 - in terms of their meaning or implication for more synthesized variables.106 Fbr instance, age, amount of education, and size of farm are interpreted as significant only as they'contribute or do not contribute to commit- ment or to involvement. In rural sociological research the variables of age, education, and size of farm are usually treated independently, often with little regard to their mutual relations. Sometimes these variables are treated simultaneously through a scaling procedure or Rarely are these variables i.e., for This is composite index construction technique. aggregated into a series or battery of synthesized variables, example commitment and involvement, and considered jointly. Hence, there is very little particularly true in market news studies. substantive basis for ascertaining possible “functional" or "dysfunc- tional" roles for market news information. In the present context, since the study projects a flat picture timewise, and because only small numbers are available for any one of the structural types to be defined, the specification of a "net balance of consequences“ is both impossible and premature. Rather, it seems first necessary to defend the designation as a structure of the complex being studied to demon- strate the programatic utility of this decision. The analysis proced- ures must facilitate this task. 106. George Baal in a different context has urged the need for more attention to conventional "structural" variables in their “dynamic“ See his, nAdditional Hypotheses rather than their “static" sense. in Participation Research," Burg; Sociology, vol. 21 (September - December, 1956) pp. 249-256. This dissertation does not accept Beal's definition of dynamic as "subject to control“ nor does it accept his strong orientation to directed social change. This StudYdoes agree with the plea to deemphasize the traditional social structural variables in their static usage. -45... While the above considerations hold, it is also true that func- tional analysis supposes systemic determinants to behavior. This seems to suggest a holistic analysis; the need for a methodology for compre- hending total phenomena sets simultaneously in time. However, if a sociological study is to be scientific, methodology with all that this implies. " are lacking in it must employ an objective, intersubj active 107 Obj ective Operational techniques for comprehending such “wholes 108 the social sciences. Therefore, whatever else its merit, this 107. For a concise statement see Roy G. Frances, "The Nature of Scien— tific Research,“ in J. '1'. Baby (ed.) 133 Igtgodgction 3,9, Socigl 3-20. For a fuller W, Harrisburg: Stackpole, 1954, pp. treatment see Morris R. Cohen and Ernest Nagel, .A_n_ Introduction 19, ngic 9E §§ientific Method, New York: Harcourt, Brace, 1934, especially pp. 197 ff. And, in the most severe and at times overdrawn terms, George Lundeberg, ngdationg g; Sociology, New York: MacMilJan, 1939. 108. Cultural anthropology in particular has been committed to this type of analysis through its use of “culture" as its most central 93., A. L. Kroeber and C. Kluckhohn, "The Concept of concept. Culture: A Critical Review of Definitions,‘I m 2; he Peabod , Cambridge: Harvard University Hess, vol. (.1, 19505. At the cane time there is a startling dearth of methodology texts in cultural anthr0pology. Noteg and Querieg 93 Anthromlogy (issued by A Camittee of the Royal Anthropological Institute of Great Britain and Ireland, London: Routledge and Keegan, Paul; 6th edition, 1951.) is the only general methodology volume known to this writer. Its focus is on idiographic data collection (i.e., "Theory and fact should not be merged. The observer who his material should wishes to give a theoretical construction to consider this separately after recording his facts.“ p. 27) Radin's New York : McGraw-Hill, volume (mmmmizW. 1933) is even more distressing, 2.5., "Most good investi tors are hardly aware of the ecise manner in which they gather or analyze their date.“ (p. ix . A o o Tgaz (A. L. Kroeber, («1.) Chicago: University of Chicago Press, 1953) pp. 401-487, likewise contains certain data collection discussion. In a more specialized area, "culture-at-a-distance," M. Mead and R- “emu (as m at 9911?: at 5 mm. (eds.). Chicago: University of Chicago Press, 1953 specifically attempt to set out and codify methodological canons. That the methodology here, and procedure must be rejected as a modus opgrandi. What is required issmm form of what Blumer has called “variable analysis" - "the scheme of sociological analysis which seeks to reduce human group life to 109 variables and their relations." 109. in the studies done in its name, and in the closely allied endeavor of "culture and personality" is grossly deficient bears abundant testimony. See among others: A. R. Lindesmith and Anslem Straus, nA Critique of Culture - Personality Writings,“ Americag Sociglog- .195; Meg, vol. 15 (1950) pp. 587—600; A. Inkeles and D. J. Levinson, "National Character: The Study of Model Personality and Sociocultural Systems" in Gardner Lindzey (ed.) gp. gi£., pp. 977- 1020; M. L. Farber, "The Problem of National Character: A Math- odological Analysis," Jourggl Q; P cholo , vol. 30 (1950) pp. 307-316. Mead's counterargument to the criticism is particu- larly instructive and damning. See Margaret Mead, "National Character," in Anthropology Today, 2p. gi§., pp. 642-667. Redfield's candid admission goes to the heart of the matter: "The validity of a characterization of a culture by any of the models employed ... is not today established by eXperimental or any other precise proof. ... Rather it may be said that the reader of an account of a culture or system of social institutions is satisfied as to the truth of what he reads only in part by the correspondence between the more comprehensive prOpositions and the documentation offered. In part the proof, if proof it he, seems to issue from ... an act of apprehension of the totality ... and such an act of apprehension is characteristic of the under- "Relations of Anthropology to the standing of a work of art.“ Social Sciences and to the Humanities“ in Anthrgpology Today,gp. 2132-: Pp. 735-736. Herbert Blumer, "Sociological Analysis and the 'Variable' ”, Amgriggn §ggiglogical Revigy, vol. 21 (December, 1956) p. 683. This article is highly critical of variable analysis and correctly points to certain deficiencies. His plea to a holistic inter— pretation (p. 689), characteristically, is unsupported by'a working methodology -” This is not the occasion to spell out the For the most succinct and sOphisticated nature of the scheme." statement of variable analysis see Lazarsfeld and Rosenberg, 220 £50 - 47 - A crucial problem is inherent in this analysis schema. Variable analysis requires a procedure or rule fbr judging the significance of variate correlations. Because statistical correlation between the Operationally defined indices of the independent variables and the de- pendent variables offers the evidence to evaluate the study hypothesis in variable analysis, the question arises as to whether a given corre- lation is important and significant or not. There is no clearcut answer to this question. Among those who subscribe to this methodology a significant split has developed on this issue. Selvin writes: With only slight exaggeration it is possible to divide empir- ical researchers into two groups: (1) those who test each conclusion for significance but seldom cross tabulate exten- sively to discern causal or eXplanatory factors; and (2) those whole substantive analyses are based on extensive cross-tabulations, with no tests of significance. Although the members of the first group are by no means of one mind, the few critics within it have generally concluded that the tests do perform a valid function in sociolo ical research. Exactly the Opposite point of view has been Eand is here] argued. — A 110. Selvin, 93. 913., p. 519. Fbr generally concurring appraisal see J. S. Coleman, "Statistical Problems," in S. M; Lipset, Martin Trow and J. S. Coleman, Union Democracy, Glencoe: The Free Press, 1956, PPo 427r432; Patricia L. Kendall, "Note on Si nificance Tests“ in R. K. Merton, G. G. Reader and P. L. Kendall eds.), Th2 Student Phygician, Cambridge: Harvard University Press, 1957, pp. 301- 305; Herman Wold, "Causal Inference from Observational Data,n Jogznal‘gf the Royal §tati§ticg1 Society, vol. 119 (Series.A, part 1, 1956)_pp. 28-50. Without direct statement though infer- ly‘With rentially supporting this position, is H. Zeisel, Sa Harper's (4th ed.), 1957, especially pp. 131- Eigyygg, New York: 214 and nearly all of Paul F. Lazarsfeld's work. See particularly, "Interpretation of Statmdical Relations as a Research Operation," in Lazarsfeld and Rosenberg, 92.,§i§.; with P. L. Kendall, "Prob- lems of Survey Analysis," in R. K. Merton and P; F. Lazarsfeld figfiggzgh, Glencoe: The Free Press, (eds.), ngtingities in _Sg_gial 1950, pp. 131-196; with Elihu Katz, Peysonal Influengg,_gp.‘git. ..- -43.. Selvin's critique is trenchant. Above the technical difficulties following from the logic of the null hypothesis to which significance 111 tests are inextricably wed, one criticism stands out above all his others. It is that users of tests of significance tend to confuse statistical significance with substantive significance and, therefore, make no attempt to distinguish between "true" and "spurious" corre- lations. The former, "reflects a true causal connection.u "A causal relationship betVeen two variables Exists] if _t_h_e_ p§_rtial pg- lgtionships [between them] pevgz gigappegr, even when every ggnceinble A relationship is spurious antecedent test factor ig’introduced." if at first "X" and "Y“ are found to be related but this relation is due to their concurrent correlation with a third variable "Z". Selvin argues: ... most users of significance tests do not even attempt to deal with the correlated biases; instead, they move directly from the observed difference to a test of significance. These tests mu_t be the last step in statistical analysis, not the first. 14 Selvin and his former colleagues of the Bureau of Applied Social 111. See W. A. wallis and H. V. Roberts, Statigticg A Egy.Approach, Glencoe: The Free Press, 1956, chapter 12; R. A. Fisher, 2%2 De n Qi‘fingzimggtg, London: Oliver and Boyd, 1951 (6th edition pp. 13-17. We shall reconsider the technical difficulties in a some- what different context below. 112. Zeisel, gp.‘gi§., p. 205. 1130 13231.8de and Kendall, 220 9%., p. 1580 114. Selvin, 22.,9i§., p. 522. ’i‘! - 1,9 - 115 propose other guide lines for assessing data drawing inferences. Coleman116 points out two: test by partials and tests by implications. The former is achieved simply by introducing control variables as part of the n tables rather than single variable Research at Columbia and usual contingency analysis. The re— sult is manifold classificatio tables. It is particularly exemplified by the work of Zeisel.117 The second schema states merely that, ”if there is a causal relationship between two variables, then this should inply other relationships as well."118 This criterion is argued by Merton, 1t... 51. , as “internal consistency" - "a finding with regard to one question is held to be valid only if it also holds true in connection with a closely related question. '119 A second criterion offered by Merton and his co—workers is labeled "replicgtive consistency." This means that ”a finding in one group must also hold true in a second independent group, if the same general conditions prevail in both."120 __._‘ 115. See James A. Davis' review of lhe ment, is , pp. cit., in the Amoric leases; 21: W. ”1.63 Jammy, 19583—59- 445- . 116. James 3. Coleman, letter to the editor, W m 9; w, vol. 64 (July, 1958) pp. 59-60. 117. “18.1, 90 m. 118. Coleman, letter, 22. 919., p. 59. 119. Merton, Reader and Kendall, 93. 9.110, Po 303- Of course, neither of these criteria are new. They herally understood as l'scien‘lziflc knowledge.” out for special emphasis underscores proof. 120' M0, Po 30’» are basic to what is ge The fact that it is singled the confusion over the criteria of hwo. as.» m.. a he. I. .. 5o .. Davis has apposed this position. In his review of The £3292!!! M2153 he points to mmerous instances in which similar or identical data differences lead now to positive findings; s response to that review is cogent. It goes now to negative ones. His rejoinder to Coleman' to the heart of the general methodological problem of concern here. Celeman says: "If the relationship is truly a chance one, then it will tend to show inconsistencies under different values of the third variable.' Probably so, but how can you demonstrate ”inconsistencies” without some measure of in- consistencies and some allowance for chance fluctuation? Does he propose to reject every relationship which does not show identical degrees of relationship in each value of the third variable? If not, which ones does he propose to keep and which does he prepose to reject? ... Coleman says that, if the implications of an interpretation pan out, the interpretation "is strongly reinforced.” ... What bothers ‘me is how you tell whether the implications pan.out or not, if you do not use some sort of criterion.for measur- ing the ”pan-out ability" of a given.re1ationship and evaluate the net “pan-out ability“ of a series of relation» ships. Again, something akin to significance tests is sneaking in the back door. Simply, the problem is how to minimize spurious correlation and, yet, to present formalised, quantifiable techniques so that two research analysts must come to the same conclusion when faced with the same set of data. Davis suggests: ... The BASR methods and traditional significance tests are @otJmtually exclusive, but each provides data on a differ- ent type of potential error. Full error control requires the application of both types of tests.122 ny detail in that direction, however. lem.more technically and explicitly. Davis does not specify a Others have followed the prob h‘ 121. Jules A. Davis, rejoinder to Coleman's letter, w M 2; m, vol. 61. (July, 1958) p. 61. 122‘ mo -51.. 12 A separate and pointed rejoinder to Selvin has appeared. 3 It will be useful to consider it. MbGinnis argues that the general inability in sociology to randomize correlated biases is quite correct. However, this fact only vitiates the use of statistical tests of significance in survey data if one accepts the meta-sociological tenet of complete and absolute cause - effect interpretation. MCGinnis denies that this is necessary by asserting that there are three types of hypotheses: Type III hypotheses impose the condition that ell correlated biases be controlled, which can only be accomplished by randomization. Type II hypotheses require that some finite number of related effects be eliminated, presumably in the statistical model itself. Type I hypotheses make no demands whatsoever of this sort.12 He notes further that, I'No test of significance requires 9f itgelf that all correlated biases be removed."125 This is required only for Type III hypotheses. McGinnis observes that Selvin slips to the hidden belief, "that social scientists should be concerned immediately and exclusively with Type III hypotheses and their tests. This conviction leads Selvin to the conclusion that all hypotheses of sociology are automatically of Type III."126 This being the case, all of Selvin's critique rides on 123. See Robert McGinnis, "Randomization and Inference in Sociological Research,“.Americ§n Soclgloglcal‘fieglgg, vol. 23 (August, 1958) pp. 408-414. . 1240 1.13—1.90: P0 412- 125. Ibid., p. 413. 126e Me, Po “2e _ 52 - false grounds and "the innocent test of significance becomes a false 127 Villa-inc I Tests of significance have a legitimate function to perform in survey research. If correlated biases are present in a sample or pop— ulation and operating, this does not influence the outcome of a test for a null hypothesis of Type I or Type II. It is crucial for a Type III hypothesis but, such types are not attainable presently.129 Tests of significance offer a "prearranged decision procedure" for conforming to what Sewell argues is the law of parsimony in theory testing: If a scientist's purpose is to test a hypothesis concerning the relationship between variables, he begins ... with the simplest possible hypothesis ... The hypothesis of no re- lationship, or the null hypothesis ... only if the hypothesis of no relationship can be rejected is there any point in entertaining more complex hypotheses. 127. Ibid., p. 413. 128. Ibig. Such hypotheses are necessarily time, space, and data bound. 129. Whether the future offers any hope is a moot point. MCGinnis argues the extreme that there is no "completely general relation- ship which is independent of pepulation, time and space." (lb;g,, p. 412) If one takes the "every conceivable antecedent factor" part of the definition of "true" correlation literally, it would seem that Selvin is not far from McGinnis. Selvin's own enumer- ation of why all correlated biases cannot be removed lfl‘pgggilgg from sociological research argues against his own demand for definitive cause-effect schemes. 130. William.Sewell, "Some Observations on Theory'Testing," Rural So 0 , vol. 21 (March, 1956) p. 7. This analysis will proceed from the general position of Davis and McGinnis. In the present stage of knowledge, Type III hypotheses are working impossibilities. All correlated biases are neither known nor controlable. NOnetheless, Type III hypotheses are an ideal worth achieving. It is intended in thisstudy to demonstrate a form of analysis which it is believed is a maximum.effort in this direction under current conditions. Since it is a mode not generally used in sociology, the objective of "demonstrating" will be fulfilled. Factorial Degigg Gegezgl Prepositions The model for analysis used in the study is a factorial arrange— ment of variables evaluated by the analysis of variance. "Analysis of variance is, as its name implies, the ... breaking up of variance into portions arising from specified sources and the testing of these portions to discover if they are significantly different."131 In gen- eral it is a method "fbr analyzing and describing association between one quantitative and one or more nonquantitative characteristics."132 In a factorial design the effects of a number of different factors on a dependent factor are investigated simultaneously. The treatments (of standard experimental terminology ) consist of all combinations 131. Margaret J. Hagood and D. 0. Price, Statistic f9; Sociologlsts, New York: Henry Holt, 1952 (revised edition p. 379. 132. Ibig., p. 381. 133. Ostle correctly indicates that a large measure of semantic error inheres in the use of "factorial" which generally implies a type But, he notes "factorial really refers to of egperhmental design. how the 'treatments' were formed and not to the basic design." See Bernard Ostle, Statistlcg lg Research, Ames: Iowa State College Press, 1954, p. 341. O—fi - 54 - that can be formed from the different factors. For example, in a 2 X 2 factorial, years of education as the necessary quantitative dependent variable could be investigated in terms of two independent qualitative factors, say race and religion each operationalized at two leve13134 of white and Negro: Protestant and Catholic, respectively. Then there would be four possible “treatments" - Negro Protestant, Negro Catholic, white hotestant, and white Catholic. The effect of any one treatment on education is considered as the sum of three effects: race, religion, and the interaction of race and religion. In this instance interaction measures the failure of race effects to be the same for each religion and, conversely, the failure of religion effects to be the same for each Interaction is the measure with which "two factors combine to 135 produce an added effect not due to one of them alone.“ 1'8“. Now, it is true as Hagood and Price point out that, ”there has been no thorough treatment of the utility of the analysis of variance ... in the analysis of data from purely ghgenatlpgl situations as differen- tiated from W.'136 The analysis of variance was devised for use where relatively strict experimentation is possible. However, as Ostle points out, there is no necessary reason why factorials could not 134. Hence, the desigmtion 2 X 2. If 3 levels of one factor were used and 2 of the second the designation would be 2 I 3 and so on. There is no limit to the number of independent variables or the number of “levels” for each, aside fran practical considerations. 1350 Oatle, 2D. 23., pe 345s 136s $80“ and R100, me Me, pe 380e be applied to survey data.137 Yates concurs cautioning only, as with the results of other tests of significance, that: O deductions as to the magnitudes E: angof the effects of given factors can never be made with ce inty from survey data. ... In order to determine with certainty the magnitude in the causal sense of the effect of any given actor, eXperi- ments ”.1813? undertaken ... Nevertheless they urveys] are Of “1116 e Nmnerous sociological surveys have used the analysis of variance, generally without careful or explicit concern with whether the necessary assumptions involved are satisfied by their data.13 9 Only one published work in sociolog which has used a factorial arrangement has come to the writer's attention. That is a study by Keyfitz investigating the sig- nificance of distance from city on family size. 140 Using a 26 factorial arrangement, the effect of any one of 6 inde- pendent variables was examined while holding "constant" the level of the other five. Only two classes for each independent variable were used; this for "vastly simpler Land efficient—l calculation purposes,"m and 137. 0stle, 99. git” pp. 341; 374—375. 138. 139. Frank Yates, W Me h {9; Gen e and m, London: Charles Griffin, 1953 revised edition p. 131. See particularly Churchill Eisenhart, "The Assumptions Underlying the Analysis of Variance,“ W, vol. 3, (March, 1947) pp. 1-21. Fortunately, this probably has not caused overly great @9- 11291 repercussions. See, William G. Cochran, "Some Consequences When The Assumptions for The Analysis of Variance Are Not Satis- fid," W, 701. 3, (M311, 1947) PP. 22-38. Nathan Keyfitz, ”A Factorial Arrangement of Comparisons of Family 5120." WM. vol. 58 (larch. 1953) pp. 470-480. In the nature of the case unequal numbers arose in the “treatment" types for the analysis of variance. Such a situation presents spec'hl calculation problems for the analysis of variance. The dichotomous case short circuits these. See G. W. Snedecore, agengng; Methgg Ames: Iowa State College Press, 1956 (5th edition , pp. 268-270. ’ c. t”) .. 56 .. because I’the problem was conceived essentially as one of finding out whether or not there was an effect ... rather than how much effect.u2" Sixty-four cells resulted. These were all the possible combinations of the six dichotomized factors. Each case in the sample of 1056 families was assigned to one and only one cell of the master table. The average number of children for those cases in any one cell constituted the data for analysis. On this basis it was found that certain of the results "contradict other work on differential fertility.” That is, a direct relationship of fertility with both income and years of schooling were found in con- trast to the usual merge relationships.143 Thus, in the BASR term- inology, a spurious correlation was ferreted out. Keyfitz‘ concludes “holding relevant variables constant may reverse familiar relation- ships." The significance of this for exploratory studies is that if "rele-- vant" factors are controlled at the outset, the likelihood of “spurious" correlations will probably be less.]'45 The likelihood of reversing "familiar relationships" will be less because Type II hypotheses replace __l 1420 Keyfitz, 22. we, p. 4720 1430 Me, p. 477. 144. M. 145. Smith, 29. £11., achieved control on age and education by sampling within a single age-education level and thus avoided spurious correlation with respect to these factors. .. 57 - Type I from the beginning. Research effort need not be spent in costly duplicative, though necessary, studies whose only purpose is to inves- tigate the relationship of certain factors when the effects of others are removed.l4'6 Furthermore, a factorial arrangement of variables facilitates in- vestigation of the impact of systemic relations, by use of estimates of interaction effects. Cochran and Cox observe that: day. By this . for example} we mean the response of nitrogen is the same whe her ploughing is shallow or deep, and that the difference between the effects of deep and shallow plough- ing is the same whether nitrogen is present or not ... The question arises: How do we know whether factors are inde- pendent? ... A factorial experimefi. itself provides a test of the assumption of independence. 7 It sometimes minens that the effects of factors are indepen— The existence of significant interactions effects is evidence for complex systemic action of the factors involved. Interaction is measured as the variance in a factorial cell not accounted for by the components taken singly when the level of the remaining components is held constant. Significant interactions ”indicate" that the simple effects of a factor vary according to the particular combination of other factors with which 8 they are produced,“14 the complex form of “system." Since "interaction 146. See for instance, William Sewell, A. O. Haller and M. A. Straus, "Social Status and Education and Occupational Aspiration,“ Amegigg W m, vol. 22 (February, 1957) pp. 67-73. 147. William G. Cochran and Gertrude M. Cox, W 295;“, New York: John Wiley, 1950, pp. 123-121.. g 148. MO, p. 1250 It'— - 53 - is the important effect about which the factorial design can give infor- mation,‘u9 factorial analyses are particularly fruitful for studies of systemic relations.”O message m the Rama ch ma The last consideration in this chapter is the adaptations of the study data to the demands of a factorial design. Two major problems were involved. In the first place, given the sample size, it was appar- out that the probable maximum limit for a dichotomized factorial arrangement was a 25 model. Using more than five qualitative factors meant that there would be no cases for some cells of the factorial, an ambiguous situation for tests of significance.151 In addition, a model beyond 25 would also yield a very small number of cases in many other cells. Estimates of error based on such small numbers of cases would be so wide as to be meaningless. Since the study pr0posed five struc- tural dimensions and two control variables, some compromise was required. It will be recalled that, for methodological rigor, communication should be I'constant" to assess whether differences in functions could 149. R. L. Anderson and T. A. Bancroft, WT he or: i3 Reggargh, New York: McCraw-Hill, 1952, p. 267. 150. Yates, 93. 311., p. 313, cogently observes with specific regard for Keyfitz's problem, but with wider import, that, 'The procedure of grouping and working with factors at two levels provides an alter- native to multiple regression analysis. In data of this complexity 1.3., Keyfitz's example] regression analysis would be exceedingly borlous, requiring the evaluation of 28 sums of squares and pro- ducts and the inversion of a 6 X 6 untrix. Moreover, the regressicn technique does not readily lend itself to the investigation of the existence of interactions.“ 151- Soc Oscar Kempthorne. m m and AM s: W New York: John Wiley, 1952, pp. 287-288. .. 59 .. be attributed to the independent variables in question.152 At the same time the study was aimed at an investigation of the structural dimensions as a system. Because of the shortage of cases, both goals could not be attained. One had to be chosen. The initial compromise was mde in favor of a study of system. First, evaluations will be made without the use of control variables. Subsequently controls will be added to: (l) the various subsets of independent variables significantly related to the dependent variables; and (2) to the various subsets of independent variables not significant- ly related. In no case will it be possible to combine “system" and 'control' in complete patterns. The next best alternative seems to be to ask whether significant relationships continue to prevail when con- trol variables are substituted, or whether significant relations emerge when none existed before if control variables are substituted. In case all five items comprising a “system“ show significant relations (or, the converse, of all five showing no relationship) each of the com- binations of two controls and three independent variables will be examined. The second problem concerns the dependent variable. In the analysis of variance the dependent variable must be quantitative and , hence, con- tinuously distributed. In this respect Guttnan's observation is cogent: confusion may arise as to what is a quantitative variable because when one is dealing with a statistical variable, _* rather than positive (causal relationship to which survey data 152. Within, of course, the more general recognition of the "weak" are currently subject. . 3‘? (D -60-- one is elgezg dealing with not gee variable but with gee variables simultaneously. One variable is the content of interest ... the other variable is the frequency function. The fizeguencz fleection ie alwaye quantitative as it arises fromva counting process. The variate itself may be either quantitative or qualitative; the fact that the frequency function is quantitative has nothing to do with the nature of the variate.153 The dependent variables of the study singly considered do not meet the qnantitative requirement directly. on the other hand, by Jointly con- sidering all functions simultaneously a new variable quantitative in content can be defined. If we assume that the functions are reason- ably independent, and if we entertain the hypothesis of no relationship betweenuany structural type and any function, then it would be possible to obtain a measure of 'functionality" through simply counting and summing the number of functions which can be identified as being served for a given individual irrespective of their contribution (functional or dysfunctional) to a structure. If there is no relationship between the structural dimensions and the dependent variables, then the number of functions associated with varying levels for any structural dimen- sion should be a matter of chance and consequently should be distributed randomly; The distribution of the total number of functions to any cell of the factorial also should be distributed randomly. This new variable of ”functionality” has the same claim to uquan- titativeness' as do many other borderline cases, for instance, number 154 of children in the family or income. This point is not to be overly ___ 153. In S. Stouffer, e3. 91., W 29g. Men, Princeton: Princeton University Press, 1950, p. 193. 15;. Econcmists have for years operated as if each dollar of income were equal “in content", using Guttman's terminology. This gave the- -61.. stressed, for in the last analysis the assignment as quantitative or qualitative is not ccmpletelyunequivocal.155 In the present case this assumption is of particular heuristic value. It allows the development of a procedure which, under given conditions, maximizes simultaneous controls (the plea of the BASR school) while yet giving objective rules for rejecting or accepting a relationship as significant. At the same time it helps in advancing a necessary substitute to a Mertonian "net balance of functions" analysis precluded by certain conditions of the research.”6 The factorial analysis allows us to cut through a large number of possible structural variables and winnow them down to man- agible prOportions for the substantive functional analysis of chapter five. Here the dependent behaviors, considered singly, are considered as “qualitative" variables and here, the tests crucial for the study "hypothesis” are addressed. Before turning to an examination of Operational procedures, we may summarize what the factorial analysis will provide. If the analysis shows that the independent variables ("structure") are significantly sufficient grounds for the parametric models characteristically employed in the field. Without necessarily refuting the parametric models some economists have recently explicitly recognized the artificiality of complete faith in this assumption. Hence, they have recognized the truth implicit in the old law of "diminishing utility.” See for example, James S. Duesenberry, Income, Saving grad its Them sf Emmet Bahama. Cambridge: Harvard University see, 1949. 155. For.an opposed view see Hagood and Price,‘ep. e13. It is implied also in Guttman, 39. £15. 156. See the discussion pages 42 e 44. .- be: '1' r1 (1 -62.. related to "functionality“ this M e21. constitute direct proof for the study "hypothesis." The "hypothesis” of the study is only a sen- sitizing construct which cannot be verified by these data. However, such an outcome would be evidence for the associated objective of differentiating significant structural dimensions, i.e., verifying the utility of defining structure as has been done. Depending on the out- come of this first analysis, exploratory inquiries into the specific functional roles of variant communication receiving structure (i.e., complex of statuses) may also be made. This second analysis bears more directly on the “hypothesis" of structural constraint. -63- Chapter III The Scheme for Analysis: Operationalization 313W Siegy'DeveloEgent, Schedule Construction. and Protesting Initial steps toward operationalizing the research project were taken early in January, 1956. For eight weeks the members of the research committee of the Department of Sociology and Anthropologyls? met with various USDA personnel and Michigan State specialists in agricultural economics and communication skills. Conferences were held on the average of every four days for three to four hours per session. These meetings were intended to serve as a background for understanding market news information and the part played by it in American agriculture. A research design that emerged from these meetings was submitted to the research.committee on March 7th. It was revised in light of that response and resubmitted march 15th. The design was again re- vised and suhnitted formally on April 23rd. This time presentation was to the entire Department of Sociology and Anthropology faculty'and graduate students at the Department's colloquium series. The session showed the need for further slight revisions. Approval of the spon- soring agency was obtained only after this exhaustive examination of the research design. 157. The Research Committee was composed of Dr. Glen Taggart, chair- man, Joel Smith, Edward Mbe and Duane Gibson. Dr. Smith assumed ‘working chairmanship upon Dr. Taggart's appointment as Dean of the International Studies Program.and was in full charge of the research execution. Nb. Francis Sim, hr. Harold Israel and the present writer were affiliated as graduate research assistants. -64- Work was begun immediately on both the sampling plan and schedule construction. The former will be treated in the next section. The latter received considerable attention before its adoption in final form (contained in‘Appendix B). The information that had to be obtained in interviews with farmers was of four types: (1) latent consequences; (2) manifest consequences; (3) sources of market news; and (4) demographic and descriptive data to serve as indicators of the independent variables. For each of these areas it was possible to construct a section of the interview schedule somewhat independent of the other three. Between July 16th and December 26th four revisions of the latent consequences section, three of the sources section, and two each of the manifest and demo- graphic sections were tested on farm Operators in Ingham and Idvingston counties. In light of these reconnaisances, by the research assis- tants, revisions were drafted under Dr. Smith's direction and retested with new respondents. The sections were then assembled into the total schedule and this was pretested, revised, and given further testing. Protesting was included as part of the instruction for interviewers. In total, 24 full schedule interviews were completed in the pretest phase. Sampling Procedure The sample was drawn for the Department by the Agricultural Parketing Service of the USDA. It was a probability area sample of - 65 - 158 almost all cpen country areas of lower peninsula Michigan. The procedures followed were those outlined in.Agricultural Handbook No. 67 by E. E. Heusaman and T. J. Reed.159 The senior author of that handbook was responsible personally for the sample drawn. Thirtybone sample areas'1.§., townships, which were drawn were ' located in the counties indicated by name in Figure 1. Within each of the 31 townships further areal segment sampling was applied. Thus, the sample was a two stage design with Minor Civil Divisions as the primary sampling unit and smaller areal segments within the township as the secondary sampling unit. By using the latest available aerial maps, attempts were made at the USDA to define the secondary units with probably equal numbers of farms.160 The primary units were defined to make them as homogeneous by type of farming area and avail- ability to mass media as possible while still requiring: (I) that each primary unit be a contiguous area and (2) that each area follow political boundaries for Minor Civil Divisions. 158. The population excluded certain areas of Ingham and Livingston counties because they had been used for protesting. Townships with ‘ special township extension agent programs were withdrawn as werev townships immediately contiguous to heavily populated urban areas which had substantial suburban populations. Finally, certain town- ships were excluded because interviewing for a study of livestock farmers' use of market news had been conducted recently in them by the Agricultural Economics Department of Michigan State University. All these areas constituted but a minor part of the total state farm population based on the 1954.Agricultural Census. 159. See Houseman and Reed, Anyhow": 0 2f Mohahilit 9......1‘93 m5 11 £2 EEIE §gzz§1§, Washington, D.C.: United States Department of Agriculture, Agricultural Marketing Service, May, 1954. 160. These maps were generally feur to eight years outdated. —66- Figure l - Distribution of the Primary Sample Units and Number of Interviews Usable From Each County. 33 Hugs /6 ’6 mad 9% Gra‘hol' 53"“6 l0 /4 6 Lac-ev- C 0 x4 ’6 O \n C V“! _ v v—f ' Two sample areas in county. 8 .h4 /0 ‘ M. Joseph Bra-w. ”BM 7 .. 67 .. On advice from the Sociology and Anthropology Department as to the time-cost factor for interviewing, revealed in pretesting, a sample of 330 farmers was judged to be the largest possible with the given resources. on this basis a sampling rate of 1/200 of all secondary units was deemed appropriate. Within each of the 31 primary sampling units three secondary areas were drawn, constituting the sample. Each secondary sampling unit constituted an area for complete enumeration by the interviewers. Schedules were completed only for those persons who resided in the area and were farmer Operators, 1.g., who had marketed at least 150 dollars gross value of farm.commodities in the calendar year preceding contact. Furthermore, such a person actually had to be responsible for the decisions made as to how much, where, when, and how the commodities were actually sold and had to be residing in the sample segment. Thus, a farmer tenant who had the marketing decisions dic- tated by a landlord was excluded from the sample by virtue of the definition of the universe. Similarly, persons farming land in an enumeration segment, but living outside the segment were excluded from the sample. Conversely, persons living in an enumeration segment and farming in another not included in the sample were included. No minimum acreage control requirement was used in defining a farm Operator. Iflierviewingg The data were collected by trained interviewers, all of whom were doctoral candidates in the Sociology'and Anthropology Department at -53.. Nfichigan State University'and experienced in interviewing. ‘While five interviewers161 were engaged, four-fifths of the completed in- terviews were taken by three,162 including the writer and Mr. Sim. The latter two jointly supervised the field work. Interviewing was begun the last week of June, 1957, and completed approximately the first week in October. A total of 375 completed schedules were ob— tained;\ For purposes of this study, however, only 356 were usable. Nine of the 375 respondents claimed to receive no market news. .Another ten schedules were missing information on one or more of the independ- ent variables and could not be assigned a place in the basic factorial. Generally, interviewing was done on the farm at the respondent's convenience. The schedule was rather lengthy by normal survey stand- ards. Interviews averaged between two and two and one-half hours. Since a large number of the respondents - 47.7% - worked Off the farm, and nearly feur in five of these worked at least half the year for a full day, a large number of very early morning and late evening appointments were required. A further factor contributing to inter- Views at such odd hours was the coincidence of schedule length and harvest schedules in the field bean and fruit areas. In spite of these conditions, cooperation was extremely high. Only five persons refused to be interviewed. This constituted a refusal rate of just over one ‘ 161. Misters David Lewis, Robert Hicks, Ellwyn Stoddard, Franci: Sim and the writer completed the interviewing. The writer wis es 1:0 acknowledge his appreciation to these persons. due to 162- The small ercenta e for the remaining two interviewers was their earl; withdriwal to accept permanent professional appointments. Fri per cent based on the total number of persons eligible for interviewb ing, i.e., discounting those excluded by the fact of their unavailability. A minimum of three calls,163 including two call-backs, were necessary before excluding a residence from the sample. The day was construed as falling into three parts - midnight to noon, noon to 6 p.m., and 6 p.m. to midnight. Call backs were to be made during the two time periods which would round out a full day; At least one of the two had to be made on a day other than that of the original contact. All "first" refusals were revisitedat least once by'a field supervisor inuan.attempt to obtain cooperation. Persons who were not home but eliminated from.any segment as non-farmers were so classified only when this status was verified by cross-checks with at least two neigh- bors. The final disposition of the original sample drawn is summarized in the working tabulation, Table 143, Appendixrfi. The cases used in this study are distributed as in Figure 1. §gmplg Reprgsentativeness The present study, in general, is not claimed to represent any parameters other than those for the sample itself. There will be no _ 163. In an effort to obtain complete coverage in the sample areas up to eight call-backs were actually made. Maximum limits were flexible depending primarily on the time-cost feasibility involved. Gener- ally each of the 31 primary sampling units was worked by the interviewers as a group. Under these circumstances the first persons encountered in a segment generally would receive more call-backs than those visited last. Since there was no set order for beginning the enumeration of a segment, any biases in- herent in the procedure should randomize out over the entire sample 0 - 7o - attempt to ascertain the probable distribution of structural types in the Michigan farming papulation. Only the relationships among struc- tural elements with respect to their possible bearing on the uses made of’market news will be investigated. It is assumed that the variables under consideration go together in "determining" the dependent be- 164 havior (1.9., functions) in finite and reasonably limited ways. For this purpose it is only necessary to have sufficient numbers in each cell of the factorial to obtain reasonable stability in the inter- actions among the independent variables. Thus, complete representa- tiveness of the sample for the Michigan farm pepulation, while imper— ative for estimating the distribution of structural types, is less important for the goal of this analysis. It is true that a sample which accurately reflects the distri- bution of lower peninsula Michigan farmers in the factorial would enhance confidence that theoases also represent reliably the true relationships of the study variables in this population. Certainly, with complete representativeness the typicality of the variable re- lationships would be maximized for the given sample size. This matter is significant in light of the sampling design utilized. Kish has observed that with cluster sampling, "The individuals in these sampling units tend to resemble each other - there is usually some homogeneity of characteristics, of attitudes, of be- 165 havior." On this suggestion, an analysis of the variance in the 164. See the discussion of structural constraint, chapter one. 165. Leslie Kish, "Confidence Intervals for Clustered Samples," Amer- igag Sociological figzigg, vol. 22 (April, 1957), p. 155. 1‘ - 71 - dependent scores was conducted, in which the secondary sampling seg- ments were used as the criterion for classification, Table 1. As can be seen, the F tests are statistically significant. From this it can be concluded that there is greater variability among than within the Table 1 -.Analysis of Variance for Total, Latent and Manifest Score Classified by Secondary Sampling Units. Score and Analysis Sum of Degrees of Mean F ratio+ of Variance Item Squares Freedom Square Total Score Total 2,155.90 345 Among 674. 03 79 8. 53 Within 1,481.37 266 5057 1053** latent Score Total 939.37 345 Among 266.04 79 3.37 within 673.33 266 2.53 1.33* anifest Score Total 767.93 345 ” Among 230.64 79 2.92 Within 537.29 266 2.02 1.45* + F ratios are read from tabular values of 75 and 200 degrees of‘ freedom respectively for the greater and lesser mean squares - the closest tabular values available and the most conservative under the circumstances. The F values for the latent score are on the border» line of statistical significance. With 200 degrees of freedom the F value associated with an .05 probability level is 1.35; with 400 degrees of freedom, the next closest tabular value, the same F Value is 1032. l 166 . c asses. In other words, there is a tendency, significantly above chance, for persons within sampling clusters to be more similar in score than peeple in different clusters. 156. In the analysis only those segments were used that had two or more schedules completed. Necessarily there would have to be complete agreement on score for only one case. Ten secondary sampling units contained but one eligible respondent. Three units yielded no inter- views. - 72 - The significance of this is not clear. It is conceivable that the"homogeneity" in the dependent variable is attributable to similar- ity in structural types. The data available were not satisfactory to test this possibility. There are 80 secondary sampling units with two or more respondents but, almost 40 per cent (31) contain only two or three cases. Without being able to justify any ordering for the sampling units besides nominal distribution, only lower power statis- l tical measures could be applied. A chi-square goodness of fit test was not possible since the data did not meet the criterion that no expected frequency be smaller than one or that no more than 20 per 168 cent be less than five. Use of the binominal test for each secondary sampling unit with an assessment of the probability of the occurrence of the number of significant tests actually obtained was not warranted. .A binominal test, under a hypothesis of p=q=% and with no direction predicted,canln§zer show statistically significant diff- erences for samples of three.169 Extrapolating from the remaining cases is not valid because of the large amount of data necessarily excluded. 167. See Sidney Siegel, Nonpgrgmetrig Statistics £9; Egg Behggigra; Sciences, New York: McGraw-Hill, 1956, especially chpter 3, PP. 18-340 163. W. G. Cochran, "Some Methods for Strengthening the Common X2 Tests," Biometrigs, vol. 10 (1954), pp. 417—451. 169. See Siegel, gp. gi§., pp. 38-39. - 73 - It was the writer's impression that some structural dimensions, particularly those of interpersonal relations and rationality, occasionally showed striking homogeneity within sampling segments but 170 that, on the whole, homogeneity was only moderate. It may be surmised that homogeneity within segments with respect to the number of functions served is probably not due to homogeneity in structural assignment. Territoriality as such, or as a correlated index to other variables, is another variable whose relationship to market news might be investigated in future research. It could receive no further consideration in the present study. Of course, and finally, it should be noted that the similar sums occuring within segments might have been based on sharply different structural components. It is this interpretation which is made for the remainder of the analysis. If the dissertation was concerned with making estimates of the distribution of structural types in Michigan , the possible biasing effect of cluster sampling would be highly important and limiting. Even with the more modest, exploratory goal of the present analysis it retains some of its significance, for it is assumed that the cases in any cell of the basic factorial give a reasonable representation of both the range and the interaction of the independent variables. With M 170. See working Table 2-W,.Appendix.A, for support of this observation. This conclusion is given some further weight by the fact that the F values obtained were not extremely large. - 74 - significant homogeneity in the secondary sampling segments, the actual number of cases which are viewed effectively in the study is possibly less than it would be if a simple random sample design had been used for selecting individual farmers. There is no way of assessing adequately the biasing effect of the cluster techniques used here. While it is possible that this is minimal, following Keyfitz's suggestion,171 an altered within-class error expression will be used as a "safety" feature in computing the significance of differences in mean scores. Indicesz Independent Variables It was assumed that each of the independent variables would yield scales by Guttman scaling techniques.172 This preference was reflected in the study design. It was indicated previously that, ideally, the structural dimensions should be independent to maximize predictive knowledge. Scale analysis is particularly geared to seek out and test the assumption of unidimensionality in a composite set of qualitative data. Scale analysis, “affords a procedure for ordering individuals or groups along a single dimension, at the same time testing the assumption that the several acts or items 'hang together' to represent 1'71. Keyfitz, 92. $3., pp. 1.76-4.77. 172. For a detailed discussion of these procedures see S. Stouffer, . 21¢ 21,, Measurement and Prediction, Princeton: Princeton Univer31ty ' = terms Press 1950 rticularly pp. 3—19, 60-90. Hereafter the "scalhgram.dn§:ysis", "Guttman techniques" and "scale analysis" will be used interchangeably. - 75 - a unitary concept."l73 Of course, unidimensionality is apt equal to independence. It is vitally interrelated, however. Unidimensionality, in the sense that a number of different types of behavior go together throughout a sample in a given "step" pattern does not indicate whether the behavioral universe i.e., a structural dimension, is correlated with a second set defined as structure. Eat, it seems reasonable to suspect that if each of the independent var- iables is shown to be unidimensional, and if the behavioral areas so defined are different substantively, a minimum of spurious correlations will result and a maximum of prediction under the given conditions 174 will be achieved. Each of the structural dimensions should yield 173. J. W. Riley, Jr. "The Sociological Variable,“ in Riley, Riley and Toby, Sociological Studies in Scale Analzsis, New Brunswick: Rutgers University Press, 1954, p. 18. 174. The problem of "independence" needs to be carefully considered. Statistical independence refers to the simple fact that "the .222EEE222§ 2; 222 [aventj does not affect the chance of the currenc 9; 329 other.“ ZAnderson and Bancroft, 22. g;§., Po 125. Under the conditions of present limited knowledge, efforts are usually made to obtain independent variables that are independent in the statistical sense but which are correlated highly with the "dependent" behavior. However, the notion of system demands statistical dependence for the so called "inde- Pendent" variables of usual variable analysis models. The level of one factor must, in a systemic model, set the level of all others. This is true for any and all delineatable aspects of the System. The current assumption, which this study does not vitally contradict, is that maximum knowledge is attained 11‘ variables are selected which are gelgtively uncorrelated so that while they'may indeed be part of a systemic model, the rate of a change in one of the variables means that the other independent variables change so little as to be, for all practical purposes, considered as statistically independent. The loss of knowledge for true systemic behavior inherent in this process needs to be constantly recalled. .. 76 .. correlative material not contributed to by the others because each is a different phenomenon. ‘Without knowing that an "independent” vari- able is unidimensional one is less sure that the correlations of any independent variables x and y with any dependent variable s are not due to the fact that x and y are essentially measuring the same thing. This condition is minimized when the independent variables are uni- dimensional rather than multidimensional. Scale analysis is not the only procedure for assessing the unitary nature of a concept and its derivable indices. However, as Guttman says, ”one of the contributions of scaling [i.g., scalogram theoryjis to do away with untested and unnecessary hypotheses about normal distributions,” 175 which adhere in such alternatives; proce- dures intended specifically for quantitative variables.176 The variables of the research are conceived of as primarily qualitative. It was apparent that all the independent and control variables could be included in the analysis only if each was in some way dichotomised. The use of Guttman scales to define the variables of interest provides a reasonable basis for dichotomizing. Each of the scale types produced by valid Guttman scaling is separable from types either higher or lower than it in the rank terms of the dimension 175° W3 we t sati. males. 22- 23-. p. 71. 176. See ibid., pp. 172-212 for a discussion of factor analysis, regression and other product-moment series statistics in relation to scale analysis. Note also J. W. Riley, Jr., "The Sociological Variable,” 22. git. - 77 - from which the scale is built. "The scalogramthypothesis is that the items have an order such that, ideally, persons who answer a given question favorably all have higher ranks on the scale than persons who answer the same question unfavorablyi‘177 In other words, "people in higher scale positions have all of the characteristics of people in lower scale positions plus at least one additional positive response.“ 178 The points for dichotomization must follow the lines of the data, i.g., must occur at a "natural' break in the variables. The separation for contingency analysis by Guttman technique is less capricious than that involved in most other forms. The problem of cutting points is, of course, broader than the design of the current research. It is at the heart of attribute analysis generally as well as with tests of significance specifically. Selvin correctly notes that: ... an attribute or a scale with several values, such as a nine-point scale of ideological sensitivity, can be dichotomised or trichotomized by using several different cutting points, some of which may increase the correlation between the scale and another variable, while others may reduce it. In fact, one occasionally finds a table where the direction of the association can be changed by shifting the cutting points. 179 He argues then that, M 177. Ihg’AQprican Soldier, 22, gi§., p. 9 (italics removed). 178. E. A. Suchman and Roy G. Francis, "Scaling Techniques in Social Research,” in John T. Doby (ed.), ‘2, troductio t9 §2§l§l Eegggggh, Harrisburg: The Stackpole Company, 1954, p. 129. 1790 8013111, 22. me, p. 5270 - 78 .- In such cases, where the analyst can drastically alter the apparent relationship between the variables by manipulating the cutting points ... . Why test for statistical significance when even the very direction of the results is in question? 180 There can be no question that the determination of cutting points is a serious problem. Selvin suggests that, "Mg 353.12 £2!!! 9;; mnemmnmmmmmn isotrgpig 93; 232 23 m isotropic by re-arragging £933 9; 901mm." 181 This is only of value after the relationship between factors is obtained. How does one go about maximizing the chances that cats- gories of variables will be collapsed legitimately? Or, should one never collapse categories and simply abandon data when relationships are not isotropic? It is this study's contention that the need for separation of variables for contingency analysis and their review by tests of significance is: (l) legitimate, on the basis of previous ergunentl82 and (2) enhanced by the Guttman technique. It is true, of course, that little can be done before the fact to assure obtaining isotrOpic tables. Isotropy is a matter of the Joint distribution of two variables in a population.183 Unidimcnsion- ality refers to the nature of the content of a single variable. Even ~— 190. 1‘21.- 181. Ibid. (italics added). See G. Udny Yule and M. G. Kendall, in Milan is the, 111.2221 2% Miss. London: Charles Griffin Company, 1948 (13th ed. , pp. 71-71. for a discussion of isotropic tables . 182. See pages 47-53. 183. Yule and Kendall, pp. 941., pp. 71-72. .2'. [2-3- '~;. 3'! ' "I - 79 - with two unidimensional attributes, their joint or correlative occur- rence may take any form -- linear, curvilinear or random. However, under the conditions of a true Guttman scale, the variable defined unidimensionally must be cumulative in its distribution function. If a second variable in a contingency table is perfectly correlated with the first, defined by Guttman technique, then an isotrOpic table must result. If the correlation is not perfect, then, for certain purposes, cutting point selection can be drastically affected and the possibil- ities for collapsing data confounded. However, it is still true that the Operational definition of an "independent" variable by scalogrmm procedures maximizes the possibility of isotrOpy because it defines half the table in cumulative terms. Isotropy is essentially a cumula- tive property in a joint distribution. While defining the second variable in a contingency analysis independently in cumulative terms (1.2., a Gmttman scale) does not necessarily help obtain isotropy, it does not hinder it either. In any case, the previously cited advan- tages of Guttman scaling would justify the use of it for all the inde- pendent variables apart from this argument for isotropy. With respect to that phenomenon, the small advantage offered toward objectivity is not to be minimized in social science research. There is still another advantage in the use of Guttman scaling. Knowing that questions of similar content scale together shows them to be measuring a single attitude or Opinion [or other] variable and warns one against attempting to interpret them as measuring different (albeit related) variables on the basis of their manifest content. The -80- warning is particularly appropriate when one is attempting some "causal" connection from a cross tabulation. 131+ A simple ”correlates" analysis can include items as "independent” variables which are merely aspects of the same phenomena, or in scalo- gram terms are items in the same behavioral universe. Working with these canponent items increases the possibilities of spurious correla- tion, as previously understood. Social flglations an}; Rationality Although attempts were made to scale all five variables, only three met the criteria of acceptable scale. These included rational- ity, relations with dealers, and relations with neighbors. The items included in each scale, the scale types, and their distribution are given in Tables 2, 3 and 1.. Both scales for intimacy are read downward. Scale type I repre- sents a completely diffuse relationship while its linear Opposite represents the maximum in Specific and socially distant relations as measured in the study. In terms of "intimate“ response frequencies, the relations with both neighbors and dealers for the sample as a whole tended toward the distant and specific rather than the diffuse. Almost one-third of the sample did not consider their dealer even an ”acquaintance,“ the most frequent type of diffuse behavior. Similarly, Just under one in five did not ever visit with their neighbors. The Specific-distant tenor of relations was particularly true for the M 184. m ficldieg, 22. git” pp. 154-155. Table 2 - Scale of Intimacy in Social Relations with Dealer or Dealer-Surrogate.” Items Response Number (Descending order of —**. degree of intimacy) Intimate (+) Distant (-) 1 Engages in recreational activities Yes NO with dealer 2 Sees dealer socially outside place Yes No of business 3 Personally knew dealer before Yes NO doing business 4 Knows location of dealer's home Yes No 5 Manner of addressing dealer Nickname or Other first name 6 Considers dealer Friend or Comparative acquaintance stranger Ideal 1"“ hub” and Nmnber Per cent of Total Scale Pattern of Res onse of cases in the Type Types 1 2 3 4 5 3 ‘ I + + + + + + 23 6.5 II - + + + + 1.6 12-9 III — - + + + + 35 9-8 V - - - - + + 32 9.0 VII - - - - - _ 106 29.8 Total 356 100.0 ’ Dealer-surrogate was defined as any employee of a dealer with whom a respondent had regular contacts if no relationship at all was main- tained with the owner or other managerially responsible person. An Ollmple Of a surrogate frequently encountered with dairy farmers was the milk hauler or route driver. Suchtatperzfin, howezer, reprS-gf sentin the buyin or ization const u es s pg;§2_i£1§3_12_ the castracting OE bufiizg organisation to the respondent and in this Dense is interchangeable with the dealer status. Thble 3 - Scale of Intimacy in Social Relations with Neighbors. ‘w u Items Res onse Number (Descending order of - degree of int! y) Intimate (+ Distant (v) 1 Lend farm equipment or supplies Yes . No 2 Plan and execute joint trips Yes No to town 3 Do small errands for others in Yes No town 4 Visit with one another Yes NO Ideal Item number and Number Per cent of Total Scale ‘22£££££_2£.§2§222§2. of cases in the Type Type 1 2 3 4 I + + + + 41 11.5 II - + + + 80 22.5 III - - + + 53 14.9 IV - - - + 118 33.1 V - - - - 61. 18.0 Total 356 100.0 3' . *m-mm- ’ -33.. Teble 4 - Scale of Rationality in the Use of Supplementary Resources in the Agricultural Operations. saEaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa=========================== Items Response ___ Number (Descending order of degree of rationality) Rational (+) Non rational () 1 Computes net return per acre Yes Nb or animal 2 Attends demonstrations and Yes No field days 3 Reads Extension or Emperiment Yes No Station publications 4 Participates in county agent Yes NO program, visits with county agent or seeks direct advice at M.S.U.* 5 Nature of bookkeeping Book of some No books or sort kept retains only receipts Ideal Item Number and Number Per cent of Total Scale flattern of Regponse of cases in the Type Type 1 2 3 4 5 I + + + + + 38 10.7 II - + + + + 32 9.0 III - - + + 42 1108 Iv .. .. - + + 38 10.6 V - - - - + 90 2503 -‘ *- ’ A positive response to m 923 of these behaviors was scored as a rational response. -34... dealer. The distribution of cases in "intimate" categories was con- sistently below that for the neighbors. This is so in spite Of the fact that the dealer scale was taken with regard to route drivers or milk haulers for any respondent who claimed to know absolutely nothing about his dealer. These characterizations for relationships with both hbors and dealers accord with notions Of the increasing segmentali- on of rural areas and with the subsequent tendency 185 neig zation and urbanizati for social relations to move from diffuse to specific types. The scale for rationality is also read downward. Those persons availing themselves of all supplementary resources in the pattern defined were considered the most rational while those using none of the resources were considered the least rational. On the basis of the distribution of cases, the sample members can be characterized as some- what non-rational in their orientation to the vocational aspects of agriculture. For instance, fully a third of the sample members did not so much as keep a set of accounting books, irrespective of its accuracy. In contrast, only slightly more than one in ten computed This non—rational eturns on a work unit of acres or animals. 186 not r characterization in general accords with other recent studies. 185. See Lowry Nelson, "Rural Life in a Mass-Industrial Society," Rural Sociology, vol. 22 (March, 1957), PP- 20'303 00 D. Duncan, , vol. 19 (March, "Rural Sociology Coming of Age," £2331 c 0 John L. Haer, "Conservatism -— 1954), sapecially pp. 8-12; Radicalism and the Rural-Urban Continuum," 335;; §g§iologz, vol. 17 (December, 1952), pp. 343-347. and R. F. Bittner, Earn Prgctige aggptigg in an, East Lansing: Michigan State Agricultural meri- ment Station, Technical Bulletin 263 (January, 1958); A. Dean, H. A. Aurbach and C. Paul Marsh, "Some Factors Related to Rationality in Decision Making Among Farm Operators,“ Raga; segiglgzz. vol. 23 (June, 1958), pp. 121-135. 186. See James H. Nielson - 35 - However, armors careful evaluation needs to be made of this and the other two scales. The evaluation of Guttman type scales involves both substantive content and formal methodological criteria. ”Scale analysis as such gives no judgment on content; ii presumes ihgi_ihg universe 9; content i§_gi;§§gy defined, and merely tests whether or not the area is repre- sentable by a single variable." 187 Content has few guides for evalu- ating except for the nebulous, and hence controversial, "face validity." Indeed, Suchman, in evaluating the utility of scale analysis, remarks, "there is no way to decide whether or not an item belongs to the uni- verse, except through a decision on the part of the investigator or a group of Judges." 188 While the three scales are not composed of all the possible items in the universe of attributes,189 it does seem reasonable to include each item in the field for which it is defined. All the behaviors included in the rationality scale are, in intent, if not effect,meant as rational means of increasing or main- taining earnings from the farm. All the items in the social relation- ship scales are clues to the scope of the actor's interest in the relevant other. This is the pith Of the diffuse-specificity'dimen- sion.190 187. Mines __l___S° dier. 22- 9.13.. p- 85. 188. 1pm., p. 187. 189. Fer a discussion of this point see ibid., pp. 80-82. 190~ lbs we}. mm. 22- 91?.» pp- 65-66. - 86 - It should be recognized further that the study was not intended as an investigation of the limits of the attribute universes defined as structure. No unequivocal conclusion as to the unidimensionality of the universe within the time-space population represented by the sample is intended. Scale analysis was adopted only because it offered an efficient, objective means for building structural typologies. It is possible that an item, especially one not investigated, may be scalable in.more than one attribute universe.191 This should not de- tract from the present types, except in the possibly constructive sense of allowing more useful and predictive types to emerge if the structural dimensions are poorly conceived. The methodological criteria are more objective but far from unequivocal. In Table 5 the results of a variety of tests of the existence of valid scalogram.patterns in the three scales are summa- rized. Each scale has a coefficient of reproducibility well above the .900 minimum proposed by Guttman.192 While this measure of the amount 191. In the literature on scale analysis, particularly in Guttman's work, it is implied that an item.must be unique in behavioral content so that it is assignable to one and only one universe of attributes. See pp. 83-85, The eric Soldier. This makes more tenable the basic assumption and claimed advantage) of scale analysis that the rank order of individuals in a sample, "exists not only for the given series of questions, but is the same as the rank order that would be obtained with any other series of questions in the same area.” (p. 154, The 53erican §gldier). Such an assumption of invariant order, while the ideal, is not crucial for a pragmatic methodology. 192. The American Soldier, 99,. £15. , pp. 77-78 offers a discussion for the measure. It is determined by the following formula: number of scale errors 1 ' number of questions i number of respondents -37... Table 5 - Criteria for Evaluating the Guttman Scales. sale _ Grit” 1°” Dealer Neighbors Rationality Coefficient of reproducibility .968 .945 .929 Coefficient of reproducibility in a chance pattern .880 .890 .867 Range of marginals (% "favorable” 6e? "" 6809 16e5 "' 7606 15e4 "’ 644/; No. of items between 40%.and 60% l l 1 No. of items less than 90% reproducibility’ 0 O 1 no. of items failing en'improvementn 0 0 0 No. of non-scale groupings with N 18 0 l 1 Per cent of total cases non-scale 16.8 20.7 31.9 - .‘g of deviation from an ideal scale pattern is the "principal test," it supposes that other conditions have also been met.193 NOt all items can have extreme distributions of sample cases among their categories. The reproducibility of any individual item can never be less than the percentage of respondents falling into a single answer category of that item, regardless of whether or not a scale exists ... . attempts should be made to include in the sample as wide a range of marginal distributions as possible, and specifically to attempt to include items with marginals around 50-50. 194 Rows 3 and A of the table show that these conditions have been met. 193. Ibid., p. 78. Generally, the characteristics of the scale with respect to the remaining criteria for scalibility are not reported 0 194. 121.. -88- It has been suggested also that each separate item in the scale should have a reproducibility not much below .90. 195 Only one item in the three scales, row 5, failed this criterion and its reproduci- bility "as .8880 Two further, related criteria are important. Guttman notes: ... for any item whatsoever, whether it belongs to a scalable universe or not, the reproducibility cannot be less than its highest category frequency ... . Hence, it is important to guard against spuriously high sample reproducibility for items which have modal categories which contain a vast majority of the population ... . Not only much reproducibility of each item be high from the trial scale score, but the scale error must be at most half of that which would be obtained without knowledge of the scale pattern, that is, from the modal fre- quencies alone. 1 Row 6 shows that all the items passed this test. Similarly, a chance reproducibility coefficient can be defined. It involves comparing, "the actual amount of total error with the amount of error to be expected by chance, given the marginals of the items in the scale under consideration." 197 Row 7 shows the values obtained. In every instance they are lower than the reproducibility coefficients actually obtained. Hence, there is less deviation from the pure types ($.g., 'error") in the scales as constituted than one would expect on a chance basis. All of these test results support the methodological adequacy of the scales. 195. Ibid., p. 287. See also, 8. A. Stouffer, E. F. Borgatta, D. G. Hays and A. F. Henry, "A Technique For Improving Cumulative Scales," in Scale M, 9-20 22-39, pe ”Be 196. American Soldier, 22. 933., pp. 287-288. 197. Riley, Riley and Toby, Sociological Studies i3 §ga1e Agglysis, 22. cit., pp. 317-318. For a full explication of the procedures .00 PP. 317-320e _ i as at. :.. -89- The pattern of error is another criterion for evaluating scales, though its application is less clearly delineated. It is contended only that, "the pattern of error should be inspected to see that there are no substantial non-scale types of persons." 198 Suchman does indicate, in discussing an example using a sample of 100, that five or more respondents with a non-perfect pattern consti- tute a non-scale type.199 By linear extrapolation, 18 or more cases in the present sample would indicate a ”significant" non-scale type. Row 7 indicates that one such grouping emerged in both the Neighbor and Rationality scales. Should these scales be accepted? The Rutgers' studies, perhaps the most comprehensive in scope and method, recognize this five per cent non-scale criterion explicity but in practice repeatedly ignore it so long as not more than one or two such groupings appear.200 That procedure will be followed here. The number of response categories and the number of items used is the last criterion to be used and the most ambiguous. It is unanimously agreed that the more items used and "the more categories that can remain uncombined, the more credible is the inference that the universe is scalable." 201 Any minimum level is unstated. However, Guttman has noted that: 198. gaggiggg Soldier, 22. gi§., p. 119. 199. mm. 200. at. m is p. 316 and pp. 88 89 91 125 134 135 179,180,2fiandé37. ’ ’ " ’ , ’ 201- mm ma 1'. 212- mo. 9- 117- -90.. At present it sews quite clear that in general the proba- bility of finding a sample of items to form a scale by chance for a sample of individuals is quite negligible, even if there are as few as six dichotomous items in the sample and as few as one hundred individuals. 202 In practice, most research studies use about the number of items used in the present scales and often use much smaller sample sizes. Further- more, it should be remembered that the intent of the scales was not to determine unidimensionality for the universe definitely but to provide a convenient and reasonably objective method for constructing structural typologies. Thus, while more items in each scale would be desirable, the scales actually obtained do appear as more than chance phenomena.203 Igvolvement Egg Commitment Both these variables were originally conceived to be unidimensional in character with the possibility that commitment might have two related axes. As the analysis progressed it became clear that this assumption was not valid. Neither the involvement nor the commitment items could meet the formal demands of scale analysis. This conclusion was based on extensive investigation. A thorough examination was made of the items chosen to represent each attribute universe. Through a special * 2020 Me, p. 82o 203. The scales in each instance incorporated information from at least 50 per cent of all the questions in the schedule which were in- tended to solicit material for the scales. The majority of those questions were asked as simple dichotomies. Thus, under the practical research limits a fairly efficient use of available data was achieved. The validity of both the dealer and rationality scales are given sane further credence by the fact that they followed the lines set out in an earlier study of a Michigan farm population. See anith, 0rg_a_gization 2; the Farm, pp. 913;. M1 comparability is not obtainable and should be interpreted cau- tiously given the warnings on the relativity of Guttman type scales. See 53% §gldier, 22. cit., pp. 82-83; 168-170. - 91 - “count“ board wired for the IBM 3101' statistical counterzOA it was possible to assess very rapidly the scalability for the items combined Even under these circumstances it was not in a wide variety of ways. possible to construct the desired scales. Further attempts to sub. divide each variable into subuniverses yielded no satisfactory results. It seems reasonable to conclude either that the items chosen to repre- sent both involvement and commitment are not good indicators or that both variables are not susceptible to unidimensional representation. The latter is a more tenable interpretation. These factors can be con- sidered only as "focal" rather than unidimensional in character. The sense of this can be seen perhaps best with commitment. Commitment was defined as the degree to which a person is 204. The wiring arrangement was devised by Mr. Sim and to the best of our knowledge is unique to him. It has not appeared in print. The method begins by punching all items construed as being in the universe onto IBM cards in the full array of categories of answers as originally coded. In the nature of the case, all the items were enterable on single columns. The count board was then wired so that any combination of answer categories for any column could be quickly made to be either a positive or negative response. However, each column had to be counted as a dichotomy in obtaining a total score for a respondent. The maximum number of items (dichotomized) that could be handled for any one scoring was 20. The number of columns used could be quickly changed by simply removing from the count board the lead wire of all the columns to be removed. The codes construed as "positive" or "negative" were almost as quickly changeable through altering the codes wired in series to either count (positive) or not count (negative). The 101 counts and distributes the respondents by total scores. Any scoring that provided reasonably distributed totals was run off by a second, specially wired board on the IBM type 107 tabulating printer. This produced a normally appearing scalogram type array which could be inspected for leads in recombining codes in columns; dropping or adding items. With.the Operator having an understanding of the criteria for valid Guttman scales, the pro- cedure allowed extremely rapid evaluation of a large number of combinations and permutations of items scored in many different W8. II II - 92 - inextricably bound to agriculture as a vocation by any or all of a variety of factors other than those of preference and a sense of its worth and importance. The specific sources for high or low commitment might be varied. For example, age could be a factor making for commit- ment. The older a person is, the less likely will he be able to find employment outside agriculture even if he has skills necessary to such an alternative, or the desire to obtain such employment. Similarly, the number of jobs open to a person with more formal education is higher than for persons with less education, for the former has avail- able all the Jobs Open to the person with less education plus some specifically denied to persons of little education. Similarly, a man with liabilities against the famm, other things equal, would tend to be less occupationally mobile than the person free of debt. These factors are to some extent related to each other. For example, older peeple, in general, have less education and fewer debts than do younger peeple. If these three factors constituted items from a unidimensional attribute universe, knowledge of a person's age would also allow one to know, within a certain probability of error, the amount of education and his liabilities position through the usual conditions of scalogram prediction. Empirically, it is required that the relations between all variables be reasonably constant, jointly considered. However, if the relations depart from constancy then set patterns in scale terms cannot be obtained. Approximate ”homogeneity" can be attained through following a line of'reasoning somewhat similar to that used in categorizing the depend- .. 93 .. ent variables.205 A number of behavioral items that apply to all respondents are selected and weights are assigned to each level of these factors. The position of each respondent with respect to each factor is then ascertained. Finally a count can be made of the number of items weighting toward commitment or not weighting toward continu- ance in agriculture. The "levels" of commitment derived in this way do not represent a single unidimensional phenomena. They represent inter- sects for a large number of attributes which might or might not be uni- dimensional themselves. Commitment, in this process, has become a focus, or intersect, of constituent elements that may or may not be completely similar. The counting procedure provides an approximate differentiation into "high" or "low" commitment which is all that is required for the analysis design. It does not make the dimension quantitative. The criterion for selecting items indicating commihnent was that of social "costs," specifically those that are economic. Leaving agri- culture extracts a price from the individual who has no other skills. Transition to outside employment normally requires the expenditure of time, possible expenses for education, often less than maximal return for the econanic means at one‘s disposal, and diminished possibilities to earn income by obtaining employment at all. Economic risks are involved in changing j obs. Certain related status features of any job —__ 205. The similarity is in the use of counting as a first step in a hem‘istic process. It was assumed that the number of functions itself is a meaningful, indicative construct. Here we count as a convenient way of differentiating the underlying qualitative variable, 'cmmitment.“ - 94 - tend to either lessen or enlarge the risks involved in Job changes. For instance, the return for one's managerial skill obtained through farming is probably at its maximum in farming. While this skill could be used in certain areas outside agriculture, its greatest return is in farm employment. Furthermore, it can be assumed that there is a reasonably positive relationship between amount of eXperience and level of managerial skill, other things equal. Then, increasing years of managing farms for oneself means a greater loss if agriculture is left as a vocation. Hence, greater years of managerial experience is more committing than fewer years of experience. The choice of the specific items used in the commitment index stemmed from.an interesting professional develOpment. Speaking about research in organizational participation, but with broader import, Beal has observed, I'there has develOped almost a traditional set of factors that are analysed ... age, formal education, stage of family cycle [etc.] ' 206 Within the limited analysis characteristic of market news studies, these "traditional" variables have been the main sort factors used. They form the basis for use here, also, because all are subsumable under the economic cost idea.207 The present study used that set of nine such variables which: (a) could be interpreted most directly in economic costs terms; (b) were relatively independent and universal in applicability; and 206. George Beal, ”Additional Hypotheses in Participation Research,” 3255; §ggiglggy, vol. 21 (September-December, 1956), p. 251. 207. This does not mean that they lack relevance for other possible analytical variables. See the discussion pp. 43-44. - 95 - (c) had differentiating potential. Therefore, certain factors which might at first appear relevant were excluded. For example, acres owned or controlled was not used because its relation to commitment depends primarily on the type of farm involved. The same features for commit- ment are covered more adequately through joint consideration of gross farm income and type of farm. In another case, tenure status and specific vocational training outside of agriculture were not used be- cause they differentiated very little of the sample. Very few renters and very few persons with.vocational agricultural training were included in the sample. Length of residence also was excluded. It was probably an important social cost factor in lateral, intra-agricultural moves because part of "successful" farming is in the building of a function- ing network of personal relationships with neighbors, dealers, and other parts of the social organization of agriculture. Physical or intra-agricultural movements often sever such relationships. However, this network of “personal" relationships, built up through time and specific in its functions, is often of little utility or relevance for interoccupational transfers. Furthermore, a transfer out of agricul- ture does not necessarily imply as great a geographic movement as does intra-agricultural movement. The items used in the commitment index are given in Table 6. Aside from item selection, weighting was also a crucial problem in constructing the commitment index. The procedure adopted was simple. Each variable was dichotomized. A weight of one was assigned to that side of the dichotomy that should be most heavily committing, and zero to the least heavily'committing side. The commitment "level" was Table 6 - Index of Commitment to Farming. Item Point of Dichotomization (Committing Category) 1. Age Over 40 years 2. Formal education Less than 12 grades completed 3. Number of dependents* Four or more 4. Type of farm+ Dairy, part dairy, fruit, fat stock 5. Amount of gross farm income Over $7,500 6. Amount of total assets All invested in farm 7. Farm.has unpaid financial Yes liabilities 8. Preportion of total income Less than 1/2 from off-farm work 9. Years managing farm for self 20 or more Index Grouping Number Per cent 1 11 3.1 2 37 10.4 3 43 12.1 4 67 18.8 5 87 24.5 6 66 18.5 7 37 10.4 8 8 2.2 Total 356 100.0 M— h ’ Includes children and aged parents. *’ See Appendix A, Table 34W, for the basis of this classification. - 97 - obtained by simply counting the number of "committing" factors each respondent possessed. The items were each dichotomized on the basis of the marginal distributions used in the preliminary Guttman scaling attempts. Cutting points were taken from the "best" plots which also made reasonable sense substantively. The points of dichotomization and the distribution by groupings are given in Table 6. subjective judgments played an important role in the construction of the commitment index since, as previously noted, none of the trial scale plots met minimal scale criteria. Choosing among "failures” in this sense is precarious and the more so for its recognition. That is, arbitrary weighting and selection of index items has a long and contin- uing record in sociological research. Of course, this is not to condone such ”error," but only to indicate a further limit on the validity of the present analysis. Hewever, the commitment index accepted does seem to have face validity. It can be used for the gross classification for which it is intended. The desirability of further work and refinement is obvious. In this respect, a word may be said about alternative procedures of index construction. If each of the variables in the commitment index were basically quantitative, exact weights for each could have been ascertained by multiple regression techniques. However, this technique was not applied for a number of reasons: (1) The labor and time involved in handling the number of variables under consideration made the work unmanageable for the resources at hand. (2) Some of the variables were not recorded precisely even though they were quantita- tive, while others were not quantitative at all. (3) This procedure - 98 .. would have interfered with the overall study design. Regression weights are set on the basis of predictive ability £9; §_gig§g’dgpggd¢ ggt vgrigblg. A new weighting problem arises each time the dependent variable is changed unless all dependent variables have a perfect or near perfect joint distribution. Thus, while weighting through multi- ple regression is much more accurate than any other procedure, it can be totally inaccurate for a second dependent variable. NMltiple regres- sion weighting is highly specific. The overall research design required reasonably high generality for the Operational definition of commit- ment. Hence, the decision was made to weight arbitrarily and count. The last independent variable to be Operationally specified is involvement. It refers to the actor's conscious awareness and evalua- tion of his own psychological relationship to farming as a work role. It is manifested in the degree to which an individual shows a prefer- ence for and identifies with his occupational role in agriculture. It seems reasonable to expect that those persons who are highly involved in their work role would show a preference for staying in agriculture and would generally be willing to name agriculture as their preferred choice if they could relive their careers. Similarly, persons highly involved in their work would tend to describe that activity in compli- mentary terms and avoid, ridicule or devaluation of the role. The need for favorable response to one's self by others is well known. Since such responses are in terms of shared norms and symbolic mean- ings, persons asked to describe farming should select these phrases - 99 - which reflect the most positive evaluation.208 Both of these ideas were centrally involved in operationalizing involvement. The respondents' job preferences were sought in direct and indirect questions. A check list of Opposed adjectives was used to elicit respondents' descriptions of farming (Question 17, Appendix B). After the check list was completed, respondents were asked which single adjective most aptly described farming as a job and which was the most misleading. "Complimentary" adjectives included: creative, healthy, interesting, own boss, takes brains, takes special skills, clean and good paying. ”Derogatory" adjectives were: routine, unhealthy, boring, tied down, don't have to be smart, anyone can farm, dirty and poor paying. It was assumed that positively involved respondents would select complimentary phrases as best describing the job and negative adjectives as most misleading.209 Positive involvement also was assumed to be a condition of long standing even though its intensity might change through time. Persons who are highly involved with their work role should not only prefer farming in a hypothetical situation but also should have demonstrated this by their own behavior. Given a choice of entering or not entering agriculture they should have shown a preference for the agricultural 208. Coutu, 92, 923?: has built an entire social psychological system about the core of selective perception symbolically mediated in light of group norms. See his discussion for the full develop- ment of this position and its implications. See Robin Williams, Jr., AmggiggnLSociet , New York: Knapf, 1951, for an account of the value premises in American society. 209. The concept of saliency underlies this assumption. See Theodore W. Newcomb, Socigl szghology, New Ibrk: Dryden, 1950, PP e 151‘153 e - 100 - role and, when given a chance to reconstruct their entrance into farming, they should express positive feelings about the situation. Information regarding respondents' entry into agriculture was elicited. The respondents were asked if they had preferred any jObs to farming when they began farming for themselves. They were asked also to reconstruct the train of events by which they came to farm. The responses were examined for indications of positive evaluations of agriculture. Positive responses included such remarks as: ”the farm is a good place to raise a family;" "healthy work;" "secure, dependable job;' "I like it, enjoy it.” Negative responses included such things as: ”wife wanted it;" "hated the city;" "just grew up on a farm, stayed on." 210 The procedure for obtaining the involvement index followed closely that used for commitment and shares the weaknesses and strengths of that procedure. The items of the index, the points of dichotomization, and the distribution by classes are given in Table 7. 210. The fact that slightly over 50 per cent of the sample claimed the last as the manner by which they entered farming in part negates the assumed long standing qualities of involvement. In order to discount this, as well as to more heavily accen- tuate the emotive aspects, the selection of the most descrip- tive adjective was weighted double the other items. - 101 - Table 7 - Index of Involvement in Farmdng. l. 2. Item New respondent got into farming When began farming would have Point of Dichotomization (Involvement Category) Passive acceptance or coerced” Yes preferred other type job 3. WOuld now prefer another type Yes of job more than farming 4. If had chance to relive past No would choose farming again 5. Most descriptive phrase for "Derogatory" farming++ 6. Least descriptive phrase for "Complimentary" farming Index.Grouping Number Per cent 0 11 3.1 1 34 9.6 2 68 19.1 3 83 23.3 4 77 21.6 5 53 14.9 6 30 8.4 Total 356 100.0 ++ Multiple response possibility. A single positive response classified person as positive on the item. Weight is double all other items. - 102 - Lgdices: Depgndent Vhriableg The functions of market news investigated were classified according to a manifest-latent distinction defined from.the USDA view- point. “Scores" were independently computed for each and then combined for a joint "functionality" score. 211 The rationale for counting and adding the functions has been discussed previously. The exact nature of the items entering each score and their limitations shall be con- sidered now. As operationalized, the structural "item" from.which the functions were assumed to follow was the information characteristically dissemi- nated by the USDA Market News Service and transmitted through the mass communication.media. 212 However, in the overall research design that "item" was not easily or always available for examination. One of the questions that the overall research design intended to answer was the degree of congruence between the USDA.definition of market news and farmers' definitions of this data. Fammers' definitions might or might not coincide with.market news as it is construed in the formal programs of the USDA. ‘When definitions were not congruent, the inter- view situation was more meaningful when the questioning as to the 211. In line with the exploratory nature of the study, this was done in order to increase the scope of possible significant rela- tionships between the structural variables and the “functions." The product-moment correlation for umber of manifest and mm- ber of extent items for respondents was only +.243. 212. The contract nature of the research strongly dictated this. The evidence of the sources for market news as found in the study supported this interpretation. - 103 - functions of market news was based on the respondent's definition. If the definitions were congruent, then the "item" for functional analysis had to be "directly" functional if it was scored for function- ality. In cases where market news was defined as broader than the formal pregram of the USDA, the formal program of the Market News Service might contribute the specific information which was serving a function or it might come from those aspects outside the formal USDA program. In the latter case, where any function was served, the fermal Market News Service program had the possibilities of being totally'and solely responsible for it, of contributing to it, or finally, of being irrelevant. The available data did not permit the separation of these possibilities. However, this does not impair the research findings. It merely means that ”market news“ should not be equated with the Market News Service. Mgnifest §gore An eight item manifest function score was used. The largest part of these items was taken from a series of questions which attempted an exhaustive reconstruction of the reapondent's last sale of what he considered to be his main product.213 The'main product might or might not be the product which.provided the largest amount of the gross farm income. Primary importance was given to what the respondent defined as his main product. This was done in order to: (1) increase his involvement in a rather lengthy interview; (2) increase the chances 213. See questions 23-31 of the schedule, Appendix.B. For detailed information on scoring see Smith and Sin, 22. 21.3., pp. 93-107. - 104 - that more details about the sale would be remembered; and (3) maxi- mize the possibility of finding market news' utility.214 When the main product did not lend itself to questions about economic decision-making functions, careful directions215 indicated the product to be used. For instance, a farmer who indicated milk as his main product was not asked about his last "sale" of milk. Milk is marketed on a continuous, daliy basis and, hence, "last sale" would be both ambiguous and meaningless. Furthermore, no formal USDA market news exists for fluid milk or other continuously marketed items. In this situation a cash "crap" such as grain or cull cows was selected. Questions about economic decision-making were never referred to contin- uously marketed products. The last sale was probed to see whether the market news indicated as generally received was used in deciding: (l) the time of sale, (2) the place of the sale, and (3) the form in which the product was marketed. Any acknowledgement of market news as an aid, irrespective of its value, was scored separately as serving a function. 214. It was assumed that market news probably would serve more func- tions and be utilized more in relation to those products which were of greatest interest to the respondent. Standardization on the last sale assumed that the typicality of sales procedure and the role of market news therein would be high, though, of course, in certain particular cases 'atypical sales would be elicited. The decision to seek information about a specific sale in con- trast to sales in general was predicated upon the facts that (1) the second procedure would compound any “articulate" bias in the schedule and (2) asking for the respondent to talk about speci- fics would probably elicit more valid information because people, in general, can recall specific information more accurately. 215. See pp. 1-2 of the schedule, Appendix B. - 105 - Time of sale was interpreted broadly. Originally it was assumed that timing of a sale could be distinguished from a decision about whether to break up a cr0p or lot or sell it all at once. However, the respondents indicated that, in general, timing and amount of prod- uct to sell at once were inextricably intertwined. Therefore, any affirmation that market news was considered in any way in deciding about either aspect of the sale was scored positively. Other items (questions 29 through 31, Appendix B) allowed infer- ences as to whether respondents used market news information to check on the honesty and fairness of dealers in pricing. They also elicited respondents‘ dissatisfactions or satisfactions with the specific prices received. This last information is a possible index of respondents' feelings with regard to the price setting and marketing system in general. Both these functions are uses intended for market news by the USDA. A function was credited if the respondent indicated that he contrasted the price offered by his dealer with the price information contained in the market news he was receiving. Similarly, in ascer- taining satisfaction with price, irrespective of its level, respondents were questioned as to why they felt as they did. Indication that market news helped create these feelings were scored as functions. Information on both planned changes and changes actually made in the farm operation (size or type of enterprise, not production prac- tices) in the past three years was obtained. The role of market news as a factor in such planning was probed and scored separately. It was assumed originally that the effect of market news on commitment could be separated from its other roles in the planning process, but this - 106 - was not possible. The detailed information needed for this separation was not recorded in many interviews, so the two related ideas were merged and scored as one function if the respondent indicated that market news information was considered in changes made in the past or considered for the future. Finally, an attempt was made to ascertain whether it would make a difference to the person if he could no longer get any information about markets. A positive response followed by one or more reasons which indicated a manifest function was being served was scored as functional for the manifest total. Islamic £29.22 Latent scores were develOped from information about nine such possible functions. Obtaining information about them was, perhaps, the most difficult aspect of the entire schedule, for in many instances the original definition of "latent“ from the USDA perspective was also "latent" from respondents' standpoints. Respondent unawareness of certain functions for market news was brought out early in the testing of schedule sections. As a result, a variety of projective and indi- rect questions approaches was tested. None of these approaches showed marked superiority over direct questions skillfully probed in obtaining valid information.216 In the final field schedule, then, most of the latent functions were ascertained through direct questions and probing, using examples of the types of behavior of interest. This possibly h; 216. See Smith and Sim, pp, 913., pp. 90-93 for further discussion. - 107 - introduced cognitive bias as a cost. In any case holistic, depth study is needed to extend knowledge beyond the level that survey pro- cedures permit. The items used for "latent" scores were diverse and necessarily neither exhaustive of possible latent functions nor completely unique as to the form which might serve in a particular role. The latent functions were selected only for their range and probable relationships to the independent variables of the study. For instance, a behavioral item concerning the facilitation of interpersonal relations was devel- Oped. Respondents were asked whether they ever discussed market news information in the sense of a general topic much as the weather might be used. It was assumed that market news represents an impersonal tepic around which sociable conversation can turn.217 Not only is market news occupationally specific, but it bears directly on the financial aspects of work. under these circumstances, its universality as a conversation piece should be virtually assured. Whatever the customary mode(s) of interpersonal relations, it seemed reasonable to expect that farm people would sometimes utilize market news in this way. When respOndents indicated that they had used market news in this way they were asked to furnish details of a recent example in order to verify this claim, Since well over half of the reSpondents did furnish 217. See Thg Sociology 9f Georg Simmel, (trans. Kurt H. Wolff), Glencoe: The Free Press, 1950, Chapter 3, particularly pp. 45-46; 51-53. - 108 - such examples, the claim was accepted even if no example was given.218 The possible bearing of market news on occupational involvement was a second latent function examined. Respondents were asked whether how'much they knew about markets and marketing affected how they felt about farming. Only persons who explicitly claimed in answering this question that information that they specifically defined as ”market news" was involved in their current feelings were credited with having As is true for all the this involvement reinforcement function served. The "score" items, the direction of the impact was not considered. basis for including this function, aside from previous considerations, is the well substantiated relationship between knowledge, affect, and 219 use for supplementary resources in agriculture. Two somewhat similar fimental health" functions were hypothesized as possible consequences of market news exposure. Farmers operate in a business organization vested with rather high uncertainty. This is attributable to both the vagaries of the crucial weather factor and the generally small size of farms relative to processors plus the farm's 218. Questions also sought to ascertain whether market news, within this general sense of sociability, might be utilized specifically to stimulate a lagging discussion. The respondent was asked first whether he had ever been witness to such an occurrence or whether, secondly, he himself had used market news in this way. The correlations (tetrachloric) between this Specific function and the item used was + .66 and that between witnessing and using For this market news as a conversation stimulant was + .84. reason only the one more general item was included in the latent score. See George M. Beal, The Boots 91 Earticipation _i_n_ Earmer Cooyerg- The College Bookstore, 1954; R. C. Bealer, tiveg, Ames, IOwa: "value Orientations and Behavioral Correlates of Producer—Patrons in Purchasing Cooperatives," unpublished Master's thesis, The annsylvania State university, 1953, especially pp. 14917. 219. - 109 .. situation at the least subtle end of the processing chain.220 Under these circumstances it seems reasonable to expect a fairly high level of personal anxiety about the marketing situation would exist. Here the final measure of the quality of the year's performance will be reflected in the price received for the product. Market information is meant specifically to allay some of this uncertainty about price. A failure of market news information to coincide with the price offered to the individual could elicit two broad reactions, _i_._e_., catharsis or succor. If a discrepancy is viewed as major, it could trigger a release of hostility toward market news information because it is supposed to be authoritative and accurate. Thus, market news reports can be a convenient scapegoat for hostilities arising from other sources. Like weather forecasts, market news is a general description, hence highly prone to be in error as a description of the individual's experience.221 Thus, its potentials for scapegoating are heightened. It is an easy thing to "pick on." In either case, the "failures" of market news can facilitate some cathartic effects. Instead of stimulating a release of hostility, the inaccuracy of market news information might serve 220. For a clear, concise exposition on this point substantiated as a global phenomenon, see L. W. Witt and Hordecai Ezekiel, Th3 {gym gpg_§hg Qitz, Rome: Food.and Agriculture Organization of the united Nations, 1953, especially pp. 9-16. 221. The reasons for this are many, ranging from.message failure due primarily to the respondent's misinterpretations, to such inappropriate data as insufficient detail, covering all grades, and so on. - 110 _ as a succor.222 The individual can find relief and strength in the fact that the assumed "expertise" of the market reports can also be "wrong." Admittedly, these possible functions for market news are difficult to assess in a survey interview. After considerable pretest, a direct series of channeled probes were devised (Question 22, Appendix B). A score for the "succor" function was given if the individual indicated that he was bothered by making mistakes, that he sometimes excused his mistakes by recalling that experts also made them, and that he had specific reference to market experts when he had such thoughts. Examples were sought to verify these claims. As might be suspected, only a small minority, 15 per cent, gave sufficient evidence of this function being served. A score was given for the “catharsis" function only if the indi- vidual indicated that he "blow off steam" when things went wrong, that on some such occasions he picked on things that were not really bothering him, and that he had specifically used market information as such an unjust target. Examples were then sought to validate such claims. Less than eight per cent of the sample was scored as having this function served. The fifth item in the latent score was market news' possible educational consequences. It was considered that over time the reception of market news information might serve as a.means for 222. Though related conceptually, they were empirically independent. The tetrachloric correlation between the two items was + .12. - 111 - learning about such things as supplybdemand-price relationships, price cycles and movements, and quality-price relationships. A direct question inquired as to whether the respondent believed he had learned anything of a general nature through the receipt of market news over time. An affirmative answer with an adequate example was scored as a latent functional, though this really only meant might have been responsible for the knowledge. Definite etiology was impossible to trace. A somewhat clearer understanding was afforded for another aspect of the latent score. It is well known that,either at the level of the community or the small group, differentiation occurs and that social relations between the differentiated statuses is thereby either en- hanced or discouraged.223 It is conceivable that knowledgeability with regard to market news might be a correlate of esteem. The criteria for assigning esteem tend to be those things most valued by or most func- tionally important to the group. Given the centrality of material well being in American culture and the relation of market news to efficient role performance for the farmer in his work, knowledge about markets might be a correlate of esteem in a farm population. One question bore directly on this possibility (Question 13.3, Appendix B). It was scored as a function Only if the level of market news information 223. See John Useem, Pierre Tangent and Ruth Useem, "Stratification in a Prairie Town," émerican §g§iglggig§lpfigzigg, vol. 7 (1943). pp. 331-342 for evidence of this at the community level. For research.bearing on the point in small groups, see Henry‘u. Riecken and George C. Hemans, "Psychological Aspects of Social Structure,“ in Gardner Lindsey (ed.), gggdbook g; Social PBzghOlng, Cambridge: Addison Wesley, 1954, pp. 786-829. - 112 - was an active part of the respondent's evaluative processes. The question first inquired as to whether high esteem and knowledge of markets were correlated. Irrespective of this answer, the respondents were then pressed for the criteria by which they personally accorded esteem to other farmers. If knowledge about markets and marketing was included among the respondent's criteria, a score was given for this function. The seventh item making up the latent score arose from the usual manner of obtaining market news. The widespread dissemination of market news through mass communication media enables the farm Operator to be released from directly obtaining such information himself. A surrogate can perform the act. This "surrogate" function should be understood in a time perspective, where it is particularly striking. Prior to the government sponsored program.fbr market news and its mass distribution, knowledge about markets, if it existed from reason- able "first hand" sources, generally depended upon "knowing" persons who engaged in selling at central markets; "catching" the "string butcher" on his rounds and getting by his double—edged and often eva- sive answers on market conditions224 or, on similar exacting particg~ Lgristig considerations. At present such particularistic relationships are less necessary in the information system. The point is not that the need for a surrogate might have changed through time, but only that 224. Part of the "string butcher's" profit depended on his ability to keep farmers in ignorance about going prices and, thereby, extract the rewards of a monopoly market. In this case it was a monopoly of information. This situation was also true of other such middlemen in the older agricultural market. _ 113‘. the reliability of the surrogate is increased by the fact that most of the market news information originates with the mass media and, hence, governmental sources. Score was given for this function on the basis of whether or not other persons in the house did regularly ob- tain market news as a matter of course or were specifically asked to obtain such information for the reapondent. A second function was also related to the formal structure of the market news dissemination program. Respondents were asked whether the USDA provided some or all of the information to the sources from which they received their market news. If they indicated some knowl- edge of the USDA's role in the system, an evaluation of the USDA's performance in that role was elicited. Respondents who indicated a belief that the USDA did operate in this capacity, irrespective of the nature of their evaluations, were credited with having this partial evaluation function of the USDA served by market news. Finally, the omnibus question previously discussed was tallied for a latent function score if one or more responses to it indicated latent functions were being served by market news. Before discussing the control variables, the limitations of the dependent variables may be summarized. The "functions" scores refer only to the types of behavior included and the representativeness of these is unknown. Furthermore, respondents were ggt excluded if their behavior with respect to a function was unascertainable. The score for any individ- ual is his maximum score ascertainable from the data available. The number of not ascertainable answers for the functions as a whole ran - 114 - slightly over three per cent and such answers did not appear to cluster in specific schedules. It seems reasonable to suspect, therefore, that information on the not ascertained functions would not greatly alter any conclusions drawn. Indices: Control Variables The content of market news information and the medium through which it was received were the factors controlled. The interview was designed to provide a richness of detail about both the sources of market news and the respondents' definitions of market news.225 The control indices cut across this detail rather simply. Since all of the variables in the analysis design could only be eXpressed as dichot- omies, it was necessary to maximize the substantive significance of any dichotomization. The variability in what farmers specifically considered as market news was great. This disparity was apparent irrespective of the level of specificity of the categorical system applied. Therefore, it was decided to dichotomize on the amount of detail rather than the type of content as such. The number of types of information included in the reapondent's definition of market news was contrasted with the types 0f information contained in market news releases of the USDA. This USDA standard was interpreted as including price, supply, demand, and short-run changes, trends, and conditions (including market volumes) 225. See questions A, 5 and 10, Appendix B; less directly note also questions 21 and 23-31. - 115 - for daily or weekly periods.226 Information falling outside this defi- nition included: (1) price outlodk, 1.2., predictions and projections beyond a week in advance; (2) demand outlook; (3) supply outlook; (A) outlook of a mixed or unspecified nature; (5) futures markets; (6) long term reviews and trends; (7) marketing techniques and proce- dures; (8) production techniques and procedures; (9) costs of inputs; (10) government policy in relation to agriculture; (11) outlook for economic activity for the national economy; (12) current weather; (13) long-range weather forecasts; (14) farm group activities and meetings; and (15) other miscellaneous information. The two cats- gories for control on content are those definitions which include only USDA items, and those which contain items not included in the USDA definition.227 The division ._ into two groupings according to the media used as market news sources recOgnized the crucial role of interpersonal rela- tions as a link between the mass media and individual recipients.228 226. No consensus on the meaning for market news could be obtained in discussion with the USDA representatives. The definition was taken from empirical evidence of the characteristic information disseminated by the USDA Market News Service and carried specifi- cally by'mass media as ”market reports" or ”market news." 227. Respondents were put into the category of "only USDA items" if they indicated no content categories outside of the USDA.defini- tion or one such item. The inclusion of the latter groups was necessary in order to obtain a somewhat even split between this grouping and its apposite and thus assure sufficient cases for all.¢H£Lls ill'the iburborial. 228. See Katz and Lazarsfeld, QB. git.; Matilda White Riley and Jchn W. Riley, Jr., "A Sociological Approach to Communications Research,” in Schramm, 22. cit., pp. 389-401. - 116 - Those persons who obtained their market news information solely through.mass media sources were separated from those who received it from either mass media sources and more informal, personal sources or such personal sources only. The media defined as "mass" included radio, daily newspapers, television, magazines and weekly newspapers. Informal or personal sources included friends and neighbors, dealers, truckers, route drivers, and extension or other government personnel. The basis of the distinction was not the type of information likely to be received, its form or accuracy but, rather, the existence of the possibility for an immediate reaction to any information in human interaction and group process. - 117 - Chapter IV Evaluation of the Structural Dimensions Prior to examining any functional prOpositions about structural constraint, the Operational definitions for structure need validation. The procedure for testing That is the concern of the present chapter. utilizes the factorial design applied by Keyfitz. 229 turning to the factorial analysis it is necessary to first inquire However, before briefly into the problem of isotropy. Isotropy in thg indepgndent Variables The most important limitation of the Ksyfitz factorial design is The general prob- the requirement of dichotomous independent variables. 1am of dichotomizing continuous variables was considered in chapter two. Here the adequacy of the cutting points used in the factorials will be assessed. A complete analysis as to whether the structural variables are isotropic with respect to functions scores would require setting out each structural dimension against the complete array of scores and assessing the association of each "elementary tetrad." 230 This would 229. Keflitz’ ‘22. 9-4-2. See Yule and Kendall, 22. gi§., pp. 71-74. Isotropic tables are those in which the ratio of all (AmBn)/(Am+1Bn) terms in compari- son to the (Am +ll/(AmflBn) terms have the same association - positive, negat ve, or zero. The terms An and Bn have the following meaning, as an example: instant; e] of' cmmmimnent = II High Medium low . Sccuwaz=.n O ‘AmBn '%m+1Bn F 1 I ‘maml Am+13n+1 L 2 l V—' 230. -113- be extremely tedious and would only here the visible fact that the structural variables are not isotropic.231 That is, even one inverted tetrad, for example, in the 84 possible elementary tetrads of the commitment-total score table destroys the possibility of isotropism. Furthermore, the tables could not be made isotropic by rearranging the levels of the independent variables for this would violate the sense of the Guttman scale types. What conclusion can be drawn then? To dismiss the data as useless seems unrealistic. In fact, non- isotropic relationships are quite frequently used in sociology. They seem to form the general norm.232 Certainly isotrOpic relations between the structural variables and the dependent variables would make conclus sions drawn from this study less risky, particularly from the factorial analysis where dichotomization is necessary. But, perhaps high risk is inherent in social science research and the more so when refinement of variables is attempted.233 The least that can be done, in the absence of isotropic relationships is to indicate in.mcre detail the form of the relationships between the structural dimensions and function scores. To this end, mean scores for total, latent, and manifest functions for each The AmBn symbol associated with "high commitment", latent score zero moves to "medium.commitment", latent score zero and the Am.an to "low commitment", latent score zero and in similar fashion for the AmBn-1, Am-an-l symbolism to pick up the second "elemen- tary tetrad." It can be seen then that the number of elementary tetrads equals the degrees of freedom for any contingency table. 231. See Tables 44W through 84W, Appendix A. 232. SGlVin, 9.2. 9—1.2" p0 5270 233. Paradoxically, the cutting point problem can be completely allayed by defining all variables as dichotomies in initial operationaliza- tion. Since a two by two table has only one degree of freedom it also has only one elementary tetrad and, ipso f to, must be iso- trop 10 Q .119- of the five independent variables were computed for all levels of each. These are given in Table 8. As can be seen, in only few instances is the ordering of the dependent variable perfect with respect to the structural dimensions. Conversely, except for commitment, there is in no instance of a sharp departure from either a decreasing or increasing function indicating an approximately linear relationship. Thus, with respect to total scores, involvement and relations with neighbors show no inversions, rationality one, dealer two, and commitment three inversions from a linear function. In none did the inversion take place in mean scores across the dichotomy line that had been set in advance of this analysis purely on the basis of equating halves in order to maximize the sc0pe of the factorial analysis. Somewhat similar rates of inversion hold for latent and manifest scores. The complete reversal on latent scores for both anchors of the commitment index is striking. The very small number of cases in each of these two classes is also apparent. The remaining variables, except for involvement, are regular as far as any inversions across the dichot- omy. Within the variables small inversions do occur. The inversion of involvement on latent scores across the dichotomy points up a peculiar feature of this variable. Throughout all the functions scores, the differences among the middle four classes of involvement are extremely small. This is particularly true for the two classes between which the dichotomies are drawn. In view of this pattern, any dichotomy will distort the data somewhat. A trichotomy of high, medium, and low perhaps would be best for simplest representation. Such a procedure, however, could not be followed for it would make the Table 8 - Mean Total, Latent and Manifest Score for Each Level of the -120- Five Structural Dimensions. L L Variable and Number Mean Total Mean Latent Mean Manifest Levels for Each of Cases Score Score Score Commitment 1 (high) 8 6.00 4.00 2.00 2 37 5.81 3.49 2.32 3 66 5.61 3.51 2.10 4 87 §,84 3,61 2,12 5 67 6.24 3.85 2.39 6 43 6.67 4.18 2.49 7 37 6.03 3.70 2.33 8 (low) 11 6.91 3.55 3.36 Involvement 1 (high) 30 7.07 4.27 2.80 2 53 6.28 4.00 2.28 3 77 6,1g 3.74 2.38 4 83 6.10 3.78 2.31 5 68 5.85 3.66 2.19 6 34 5.05 2.91 2.15 7 (low) 11 4.82 3.18 1.64 Dealer Relations 1 (diffuse) 23 7.00 4.39 2.61 2 46 6.98 4.54 2.44 3 35 7.06 4.34 2.72 4 102 5,87 3.65 2,22 5 32 5.75 3.53 2.22 6 12 5.82 3.58 2.23 7 (specific) 106 5.31 3.16 2.15 Neighbor Relations 1 (diffuse) 41 6.71 4.20 2.51 2 80 6.70 4.16 2.54 3 53 6,12 2.22 ;.g; 4 118 5.91 3.57 2.34 5 (Specific) 64 4.88 2.98 1.89 Rationality 2 32 6.81 4.53 2.28 3 42 6.45 3.95 2.50 4 38 6,63 4,18 5 90 5.72 3.49 2.23 6 (low) 116 5.36 3.22 2.14 Denotes point of dichotomization. “— ~121- factorial unworkable under the present design. This limitation on the analysis will be considered more Specifically later. Manifest scores show two inversions across the dichotomy, one in- volving a level of dealer relations maintained by only twelve respond- ents. The second occurs in the neighbor relations scale. Since it involves classes with larger numbers of cases, it must be considered more serious than any previous inversion. Similarly, the greater number of inversions for all the variables on this respect are more serious, since manifest score means have a more restricted range than the means of the other two types of scores. Despite these irregularities, it appears reasonable to assume linearity in the relationship of the structural variables and functions scores for heuristic purposes. To literally portray the reasonableness of this assumption somewhat better, Figures 2, 3, and 4 were constructed. The joint distribution of mean function score and level of the independent variables were plotted and the line of relationship approximated by sight. The supporting evidence is clear. Of course, it should also be clear that this brief inquiry into the isotrOphy of the independent variables does not allow full assessment of the significance of changing the cutting points for the independent variables. Such a procedure under the research design used requires the calculation of factorials for each and every level of each structural dimension, a procedure that was simply impossible. The probability of finding enough cases for all the cells of the factorial if the cutting points were to be moved to the extremes approaches zero. It is clear, however, that the assumption of linear relationships between the structural dimensions and the function scores is reasonable, and will be - --.1 A... JI‘II'KV H'III.-annl:ib\~lll '>.dh Ills... 4‘5! t I Flt! en's“ “A. ”1.0.1!” I '.§. (II- allll‘l -1 I -l 22- .4» u—d .190 I EEBEEJ .3353; 7.59855 2.2 one «o 32.3 .3 988 H38 can: no Sensors-3 oom m.m 0.0 m5 04. canon -l 23- ... mom L 0.: .8355» is. «8.56 o» E .5 no #23 .3 8.88 833 in: no 8.333 as .- M 0.55 L o.n -124- nu a 3 Gabon Jada-«.8» 139386 Bah 05 no snob-a hp 0.88 0835: use: He 83sn§3 .. a shaman ~125- accepted without further direct evidence. In a similar situation Yates has observed cogently that the gains in refined analysis on this point, "would not Justify the additional computation labour, which is better devoted to extending the scope of the investigation in other direc- tions.” 234 It should also be clear that if the relationships are linear, the results of the factorial analysis would not change greatly with different dichotomization points. Relation of the Structural Dimensions to Function Scores *_WW~ Eizg variables, 33 Controls As previously indicated, the research design could not analyze simultaneously the variables composing the structural system and those of the system which were to be controlled. The former were chosen as the point of departure. A first part of the data necessary to evaluate structure then is given in Table 9. The criteria for classification into the factorial cells are the five dimensions of structure. Later tables will set out somewhat different bases for classification as explicated in chapter two. In order to orient the reader to the manner of extracting and assessing the information contained in the tables of this chapter, a step by step analysis of the first table will be made. Later tables of this chapter will present only results. The first and most important information necessary is the mean number of functions served which are associated with a factor when the remaining four are held constant. To do this the data were rearranged as shown in Table 10. For purposes of illustration, only one variable F.— 2340 Yates, 22. $0, p0 3130 -126— Table 9 - Mean Function Scores by Factorial Cell Classified by the Five Structural Dimensions. Number Mean Moan ‘Mean Factorial Cell of Total Latent Manifest Cases Score Score Score Commit H, Invol H, Deal Diff, Neigh Diff, Rat H 14 8.00 4.93 3.07 ' ' ” Rat L 10 6.00 4.30 2.30 a u Neigh Spec, Rat H 20 7.15 4.50 2.65 " " " Rat L 16 5.25 3.25 2.00 ' ” Deal Spec, Neigh Diff, Rat H 10 6.10 3.90 2.20 " " ” Rat L 9 5.00 3.22 1.78 I n Neigh Spec, Rat H 5 5.60 3.20 2.40 n V ” Hat L 10 5.20 3.20 2.00 a Invol L, Deal Diff, Neigh Diff, Rat H 17 6.77 4.06 2.71 “ " " Rat L 14 5.00 2.93 2.07 n u Neigh Spec, Rat H 13 5.92 3.77 2.15 " " " Rat L 19 4.53 2.84 1.69 n a Deal Spec, Neigh Diff, Rat H 6 6.17 4.17 2.00 n R 0 “ Rat L 13 4.15 2.86 1.29 a n Neigh Spec, Rat H 6 6.50 4.17 2.33 " " " Rat L 16 4.25 2.38 1.87 Commit L, Invol a, Deal Diff, Neigh Diff, Hat 3 11 6.72 4.54 2.18 I I! I II R817 L 10 7090 4.60 3030 n u Neigh Spec, Rat H 4 6.75 4.25 2.50 ll H II Rat L 7 5.29 3086 1043 " " Deal Spec, Neigh Diff, Rat H 5 6.20 4.40 1.80 " fl “ Rat L 15 6.27 3.47 2.80 I . " Neigh Spec, Rat H 6 5.83 3.67 2017 a u I R81} L 8 4.75 2.00 2075 - Invol L, Deal Diff, Neigh Diff, Rat H 12 7.59 4.92 2.67 " a " Rat L 17 7.47 4.76 2.71 n ' Neigh Spec, Rat H 10 6.90 4.30 2.60 " " ” Rat L 12 5.92 3.75 2.17 " ” Deal Spec, Neigh Diff, Rat H 4 5.75 4.00 1.75 “ a I ' Rat L 7 7.71 4.57 3.14 " " Neigh Spec, Rat H 7 6.71 3.71 3.00 ' “ ' Rat L 23 4.65 2.61 2.04 The abbreviations used here carry the same meaning throughout the remainder of the dissertation. Commit - Commitment Invol ,Deal Neigh Rat They are as follows: - Involvement - Dealer Relations - Neighbor Relations - Rationality H - High L - Low Diff - Diffuse Spec - Specific h. -127- Table 10 — Computational Material for Assessing the Significance of Differences in Mean Total Function Scores for Rationality; Basis of Classification, Five Structural Dimensions. “—“— Differ-'Weighted Square of No. Mean ence in No. of Differ— Factorial Call of Score Mean Cases ence in Cases Score Mean Score Commit H,Invol HtDeal Diff,Heigh Diff,Rat H 14 8.000 +1.400 5.833 1.960 I n I ' Rat L 10 6.600 0 7‘ II Neigh Spec,Rat H 20 7.150 +10900 8.889 30610 n u H ” Rat L 16 5.250 " " Deal Spec,Neigh Diff,Rat H 10 6.100 +1.100 4.737 1.120 I N I I R31; L 9 50000 n n v Neigh Spec,Hat H 5 5.600 + .400 3.333 .160 u u u U Rat L 10 5.200 " Invol L,Deal Diff,Neigh Diff,Hat H 17 6.764 +1.764 7.677 3.112 I! II II It Rate L 14 50000 n u n Neigh Spec,Rat H 13 5.923 +1.397 7.719 1.952 n u R Rat L 19 4.526 " Deal Spec,Neigh Diff,Rat H 6 6.167 +2.024 4.200 4.097 I! 1‘ ll kt L 14 40143 n u " Neigh Spec,33t H 6 6.500 +20250 4.364 50062 n n n " Rat L 16 4.250 Commit L,Invol H,Deal Diff,Neigh Diff,Rat H 11 6.727 -1.173 5.238 1.376 n n n " Rat L 10 7.900 n n n Neigh spec,Rat H 4 6.750 +1.461. 2.545 2.143 II C! N I Eat L 7 50286 n a Deal Spec,Neigh Diff,Hat H 5 6.200 - .067 3.750 .004 n u n ” Rat L 15 6.267 n u n Neigh spec,Rat H 6 5.833 +1.083 3.429 1.173 u N N I Rat L 8 40750 " Invol L,Deal Diff,Neigh Diff,Rat H 12 7.583 + .121 7.034 .015 n n n a Hat L 17 7.471 " 3 " Neigh Spec,Rat H'10 6.900 + .983 5.454 .966 a u n ' Rat L 12 5.917 " I Deal Spec,Neigh Diff,Rat H 4 5.750 4.964 2.545 3.857 I! I! II II RAt L 7 7cm '3 I I Neigh SPGC,Rat H 7 6.714 +2.062 50367 40252 I I I ' Rat L 23 4.652 ======:: -127- Table 10 - Computational Material for Assessing the Significance of Differences in Mean Total Function Scores for Rationality; Basis of Classification, Five Structural Dimensions. Square Differ- Weighted of No. Mean ence in No. of Differ— Factorial Cell of Score Mean Cases once in Cases Score Eben Score Commit H,Invol H,Deal Diff,Neigh Diff,Rat H 14 8.000 +1.400 5.833 1.960 " ' ' ' Rat L 10 6.600 " " " Neigh Spec,Rat H 20 7.150 +1.900 8.889 3.610 " " ' Rat L 16 5.250 " " Deal Spec,Neigh Diff,Rat H 10 6.100 +1.100 4.737 1.120 " " " ' Rat L 9 5.000 '3 II I! Neigh SpeC,R8t H 5 5.600 + 0400 30333 0160 " " I Rat L 10 5.200 I Invol L,Deal Diff,Neigh Diff,Rat H 17 6.764 +1.764 7.677 3.112 " " " ” Rat L 14 5.000 " ” " Neigh Spec,Rat H 13 5.923 +1.397 7.719 1.952 “ I I Rat L 19 4.526 " " Deal Spec,Neigh Diff,Rat H 6 6.167 +2.024 4.200 4.097 ” I I “ Rat L 14 4.143 n u " Neigh Spec,Rat H 6 6.500 +2.250 4.364 5.062 " I I I Rat L 16 4.250 Commit L,Invol H,Deal Diff,Neigh Diff,Rat H 11 6.727 -1.173 5.238 1.376 n n n I Rat L 10 7.900 " " Neigh Spec,Rat H 4 6.750 +1.464 2.545 2.143 .. n n a Rat L 7 5.286 " " N i h Diff Rat H 5 6.200 - .067 3.750 .004 [I n Deal”Spec, 3 g” ,Rat L 15 6.267 I u n Neigh Spec,Rat H 6 5.833 +1.083 3.429 1.173 n n n I Rat L 8 4.750 " Invol L,Dea1 Diff,Neigh Diff,Rat H 12 7.583 + .121 7.034 .015 u n a I Rat L 17 7.471 I n « Neigh Spec,Rat H‘lO 6.900 + .983 5.454 .966 II II N ' Rat L 12 50917 “ I Deal Spec,Neigh Diff,Rat H 4 3.3;: -1.964 2.545 3.857 n u I! " Rat L o n u u Neigh Spec,Rat H 7 6.714 +2.062 5.367 4.252 I I I I Rat L 23 4.652 I! -128- on one score is used, "rationality" and "total," respectively. For any one set of variables three such tables were required. Hence, with five variables to a factorial, fifteen such tables in all were required. For the example, a simple way of making comparisons between the means for all functions for rationality is to use the average number of functions in the 16 paired cells combined without weights. This procedure is recommended by Keyfitz as Ithe best answer that can be secured if interactions are large.” 235 Hewever, "it is subject to unnecessarily large sampling error, for the cells containing only two or three families,[i.g., observations]have as big a weight as those in which forty are averaged." 236 Immediately two problems are given then: first, assessing the magnitude of interactions, and second, weighting sub-groups of unequal size. Keyfitz argues that the amount of information contributed by each and every one of the sixteen differences contained in the facto- rial Iis exactly that which would be given for an average, in material which varies as our individual families do within cells, of nlnz / (n1 + n2) cases." 237 Then, using the substance of his study, he con- tinues that: In general, if we have averages fog numbers of children born in families far from cities, X11 (where i ranges over the 32 ways of holding the 5 other variables constant [in our case 16 ways and 4 variablng), and corresponding averageg fog families near cities, 121, the unit compari- son is X11-X21. Being based on D11 and n21 observations, 235. Keyfitz, 22.,git., p. 473. 236. 121.- 237. Ibid., pp. 474-475. -129- it contains a quantity of information proportional to hllhzl .. N -—-----* ~ 1 , and the efficient combination of n1:1 * 1“21 all such differences is 2 N1 (in - E21) (1) 238 iiNi The error of this term, to which the actual differences need be compared in significance tests, is stated as follows: if the within-cell variance is g72, then the variance of X11421 is $3.3... - .1.) - .gL and the variance of (1) D11 H21 1 is 21712 (r2 / N1 , which reduces to (1" 2 / EN; 2 However, cluster sampling effects of the type occurring in this study may invalidate the use of this within-class error estimate. An alter- native expression for error in this situation is offered by Keyfits. Consider the 32 blocks, each of which gives a difference between the average size of family in near and distant parts and is matched on the other 5 variables. The calculation of the weighted mean difference due to distance has removed 1 degree of freedom; the consistency with which the several blocks resemble their average in respect of excess of children for distant places furnishes 31 degrees of freedom for the error of the weighted mean. Consider that in each of the 32 pairs of cells‘there are N1 cases ... and a difference of means equal to X1. This is as though we had_N1, N2, ..., N32 observations, giving means X1, X2, ..., 132, respectively, of a new variable X which has the same variance as family size,(T'2. 240 238. ago, p. 4750 239. 191 . 2400 19111., p. 476. -130- Thus the formula for calculating the significance of mean differences becomes: 2‘. N131 52111021)? - [(2111551)2 / zNflr (K-l) SM (3) 241 where all symbols are as defined before and k stands for the total number of paired cells. Referring again to Table 10, it can be seen that the numerator of this last expression is the difference in mean scores taken over all 16 "observations", i.g., cells in which only rationality is left to vary while the other four variables are held constant. It is the weighted mean difference in total scores attributable to variability in rationality. It is obtained by multiplying the difference in mean scores between each two adjacent cells by the weighted number of cases, 1.2., column 3 multiplied by column 4 and summing these algebraically over the 16 differences. The weighted number of cases for each pair of cells was obtained, as shown by Keyfitz above, as nlnz /'n1 + n2. Thus, for the first two cells 14 x-10 = 140 divided by 10 plus 14 yielded the quotient 5.833. For Table 10 the sum of the differences multiplied by the weighted number of cases is 87.570. That figure divided by the total number of weighted cases (82.114) yields a mean difference of 1.006 in favor of high rationality versus low rationality levels. To test whether that difference; to decide if the variability might be due simply to chance fluctuations in sampling, the standard 241. 121 O -131- error term stated above is required. In Table 10, the first term of the numerator within the radical is simply the sum of squared differ- ences in the sixteen compared cells weighted by the number of observa- tions in each (i.e., the sum of column 4 multiplied by column 5 or, in our example, 182.753). All the remaining terms are readily seen. The sum of differences weighted by number of observations, the quantity squared, ( £1N1Xi)2, was seen above as (87.57)2 or 7668.5049. This, divided by the total weighted number of cases, (safifi), yields the quotient 93.389. The denominator of the term under the radical is simply the prod- uct of the total number of paired cells minus one (k-l), 15, times the weighted number of cases, 82.114, or 1231.71. Performing the arith— metic calculations yields an error term of .269. Dividing the mean difference of 1.066 by this error yields a t value of 3.96, which, with 15 degrees of freedom, is significant beyond the .01 probability level. However, before this conclusion can be accepted, the role of interactions in producing this result had to be assessed; for the error term used in the example applies only if interactions are negligible. The interaction of rationality with the four remaining variables is assessed in a manner similar to that used above. For example, the interaction of rationality and neighbor relations is obtained by noting the eight differences of differences between cases of high.and low rationality when type of neighbor relations is also allowed to vary. The first such difference, from Table 10, column 3, is -l.400 minus -1.900 or +.500. This is the difference in scores for high and low rationality cases when, in the first instance, neighbor relations are diffuse, and in the second case they are specific. ~132- A weighted average must then be computed for these eight "observa- tions" in the manner of the test for main effects. The weights for number of cases suggested by Keyfitz in this case is, "the reciprocal of the sum of the reciprocals of the number of observations in the four cells." 242 In the present example the weight for the first inter- action would be 1/ 1/14 1.1/10 +-1/20 + 1/16, or 3.522. The error to which this mean difference of differences is subject is the square root of the estimated within-cell variance of the 32 factorial cells divided by the sum of the weights for the eight differ- ences of differences.243 In the example this value is .529 and the weighted mean difference is -.790. The resultant t value of 1.49 is clearly not significth with seven degrees of freedom. The remaining interaction values for rationality as well as those for the other variables of the five factor factorial, for all three mean function scores are given in Table 11. It can be seen that only the interaction of commitment with involvement for total and latent scores shows a statistically signifi- cant difference under a two tailed test. Significance of that relation is vitiated when a strict definition of interaction is invoked. Yates notes, "it is customary to define the interaction as gag-hglf the difference of effect" 245 in which case the difference elicited is no 24-2. Igid.’ p0 A780 243. Ibid. 244. A "significant" difference here, and throughout the remainder of the dissertation means a probability level of .05 or less associated with the results of the appropriate test of statis- tical significance used. 2450 Yates, £22. 22:20, p. 31.2. -133+ Table 11 - "Interactions" Matrix for the Five Structural Dimensions.+ Interaction Tgtgl Score of Commitment Involvement Dealer Neighbor Rationality Commitment -- +2.38 - .19 -1.70 +2.26 Involvement -- +1.08 + .32 - .66 Dealer - + .96 + .04 Neighbor -— -1.49 Rationality - _Latent Score Commitment Involvement Dealer Neighbor Rationality Commitment -- +2.51 -l.24 -1.78 +1.44 Involvement -— +1.61 + .24 - .35 Dealer "" "’ 050 " 063 Neighbor .- .1008 Rationality' - Manifest Score Commitment Involvement Dealer Neighbor Rationality Commitment -- +1.19 + .97 - .86 +2.09 Involvement -- - .06 + .22 - .65 Dealer -- +2.03 + .71 Neighbor - -1.26 Rationality - + Cell values are values of t to be read with seven degrees of freedcxn. 246 longer significant. The obvious conclusion is that the weighting procedure used was justified. 246. Higher order interactions were not computed since the first order interactions were not statistically significant. With the error term used, the first order interactions would have to Show reasonably consistent levels of significance before the higher order interactions could Show other than chance Significant rela- I wish to thank K. R. Bennett, statistician for the tionships. Agricultural Experiment Station, Fennsylvania State university, fer calling this point to my attention. -134- With the help of this example the results of the factorial It will be analysis can be presented without further digression. remembered that the main datum sought from the analysis by factorials is the mean difference in score for each structural dimension when Each is evaluated the levels of all others are held constant. The relevant infor- against its standard error, i.g., t values.24 mation from the first factorial is contained in Table 12. It can be seen that all of the structural dimensions except involvement show statistically significant differences in mean scores The differences associated with for both total and latent functions. variations in commitment exceed the .05 level in both instances while the other three sets exceed the .01 level.248 NOne Of the five vari- Whether ables significantly differentiates manifest function scores. this reflects uniform use on all structural levels or whether it reflects a deficiency in assessing and measuring’manifest functions is not clear. Some inferences can be made, however. It is apparent from the data in Table 9 that there is a general low level of reported use of market.news with respect to all of its manifest purposes. This low level of use is particularly noticeable 247. The values obtained by analysis of variance calculations with two means compared, as in the present model, yield results identical The comparing of with t values since t is the square root of F. more than two means is, of course, more efficient with the F test, see the discussion in F. A. Pearson and K. R. Bennett, Statistical Methods Applied pg Aggicultural Economics, New York: John Wiley & Sons, 1942, p. 354 ff. The use of t values is predicated primar- ily upon the greater explication by statisticians of the particu- lar factorial model in this form, see Keyfitz, 22. git.; Yates, 220 £22- The t values are interpreted with 15 degrees of freedom here and See 248. in all the factorials analysis unless otherwise noted. Keyfitz,{22. gi§., p. 476, for the basis of this selection. -l35- Table 12 — Differences in Mean Scores, Standard Errors, and t values for the Five Structural Dimensions Factorial Design, Two- tailed Tests. Manifest Scores t Gil-32 61-1-12 t -0311 0171 1.82 Total Scopes Latent Scores v b — _ — — - _ - _ aria 1° 611-312“ 511-12++ ti 611-12 le-xz .300 2.58** -.463 .195 2.37* .197 .170 1.16 .150 .110 1.36 1.21#* .185 .139 1.33 Commitment -.774 Involvement .347 .228 1.52 .804 .188 4.28** .619 .147 Dealer .647 .134 4.83““ .219 .161 1.36 Neighbor .866 .245 3.53”” Rationality 1.066 .269 3.96** .799 .145 5.51** .267 .171 1.56 + A difference in mean score between the two parts of the dichotomized variable. Unless otherwise noted, X refers reSpectively to high commitment, high involvement, diffuse dealer relations, diffuse neighbor relations, and high rationality. Unless otherwise indi- cated, mean score differences are read as positive. This termi- nology applies to all tables reporting results of the factorial analysis 0 ++ Standard error of the difference indicated. i. Ratio of the actual difference to its error term. The five and one per cent probability levels for the test of significance are designated, respectively, by * and **. They refer to two-tailed tests unless otherwise noted. when the extent of occurrences of manifest and latent functions is compared. Slightly over one-third of the sample indicated none or only one manifest function served. In contrast, only one in twelve indi- cated no more than one latent function served. A somewhat similar picture holds at the other extreme of use where the figures are reversed. Nearly one in three of the respondents claimed more than half the latent functions served whereas only slightly over eight per cent claimed more than four manifest functions served by market news. -136- In part, these differences are probably due to the difference in time span characteristically involved in the two types of functions as they were obtained in the schedule. Manifest functions had a tightly limited time span. Information was obtained for the last sale, not the typical or the usual sale. Similarly, changes in the last three years or for the upcoming year were probed. Only one question relating to manifest functions was Open-ended on time. That was the question, ”would it make a difference to you if you could no longer get market news; why?" In contrast, the questions with reapect to latent functions either specifically defined or implied no limit to the time in which they applied. While this difference in time reference may contribute to the failure of the structural variables to relate to differences in the occurrence of manifest functions, contrary evidence is also available. It seems reasonable to eXpect that the use of market information for its manifest purposes is part of a more general factor of business acumen or managerial ability.249 Any such factor might also be expected to manifest itself in such allied behavior as the adaption of new farm practices, technOIOgies and business procedures. If this argument is valid, this behavior should be strongly correlated with the use of market news for manifest purposes. In general, however, an overall rate of adeption for new practices is low. A 1954 study of 471‘Michigan farmers is revealing in this respect since the sample was drawn from much the same geographic area as that utilized in the present M 249. McCormick, o . 913., pp. 45-46 deveIOps this argument on the basis of his Ohio data. ~137- study.250 Fifty-four recommended farm practices were investigated. "The median percentage of adoption was 34.” 251 For specific manage- ment practices the percentage of adaption ran from 81 per cent keeping some sort of books to 38 per cent buying fertilizer ahead of the time needed, 22 per cent using written leases on rented land, 12 per cent buying supplement in quantities to get discounts, and 6 per cent buying whatever protein feed supplement is cheapest on a per pound of protein basis.252 Evidence is also available in other market news studies supporting a generally low level of use for manifest market news functions. It is true that the evidence appears contradictory without careful examina- tion. Thus, McCormick in Ohio found that between 82 and 92 per cent of the farmers in his sample of 656 who sold hogs, sheep, wheat, corn or soybeans "used various sources of farm market information before selling," and 71 per cent of the cattle sellers did likewise.253 On similar commodities, an Iowa study254 found somewhat comparable percent- ages of people who either listened to radio or read daily market news reports in newspapers prior to their last sale. However, when asked, 250. Nielsen and Bittner, 22. 913., compare particularly Figure l with their page seven. 251. 191a., p. 1. 252. IQ;Q., pp. 21-26. The low rate of adoption for the last two practices probably is not due to capital inabilities or "rationing" since 61 per cent of the farmers report using short term credit to buy fertilizer and feed. See page 22. 253. McCormick, pp. 313., pp. 22 and 37. 254. Dodds and Marvin, Egg 29 Iowa Fgrmers Obtain gag Egg Market News, 99. cit. .138... "Where did you get the information that helped you decide on the buyer of the hogs,[or corn, or cattle, etc.,]the weight at which to sell and the time to sell?" ... large numbers of farmers named none of the usual market news media sources.255 Specifically, 68 per cent of the hog sellers said market news was not used in helping to decide selling weight; 67 per cent said it was of no aid in fixing the selling time; and 63 per cent considered it of no value in deciding on sales outlet/'256 Similar figures applied both to cattle sales and grain sales.257 The Ohio study allows no insight into the discrepancy between possible communication or, at least, exposure, on the one hand, and effect, on the other. It is only stated that "those farmers who sold hogs were asked from what they obtained market news before they decided to sell." 258 Presumably, the same approach was used with other commod- ities. The loaded aspect of the question is obvious since it assumes, first, that the farmer did have information and, second, that it was used, if available. The Iowa study does not force the second part of this assumption. The current study found that slightly over 81 per cent of the members of the total sample were getting market information at the time they were getting ready to sell the product to which the manifest functions referred. However, only 52 per cent indicated that such 2550 M‘, p. [9. 256. :91 . 2570 1915., _c_f_:, pp. 137 and 1420 2580 MBCormick, 22. 91.3., p. 180 -139- market news was actually utilized in deciding when or how much to sell. The discrepancy between possible communication and its intended effects is again apparent. At the same time it is somewhat paradoxical for the data show a relatively high rate of market news use in comparison with the Iowa study. At least with respect to other market news studies, this evidence means either that the Michigan farm p0pulation has a different rate of use of market news or some improvement in measuring use has been achieved. An unequivocal answer is not available. It is also somewhat aside from the point. The problem is less one of evalua- ting the current study against the results of relevant studies done elsewhere than it is to assess the validity with which the various indices are measured and, hence, are able to differentiate in the present sample. Here, also, the case is not clear. We can begin by considering involvement and commitment. It will be recalled that in evaluating the indices for the structural dimensions both involvement and commitment were deemed the least satisfactory. Yet, g priori, these two variables appear as the most relevant for differentiating manifest functions. This is particularly true for in- volvement. There is a.more developed basis in theory.for eXpecting relations with this variable than with any of the five structural vari- ables, for both reference group and role theory suggest its pertinence.”9 _—..__ 259. See particularly Merton, Sggigl §trugture, gp. gi§., pp. 225-386; Persons, 1;; Social System, 22. gi§., particularly his discussions of "Identification"; Coutu, 22, gi§.; Ralph Turner, uRole-Taking, Role Standpoint and Reference Group Behavior," American Journgl g; Sociology, vol. XLI, (January, 1956), pp. 316-328; Theodore Serbin, "Role Theory" in Gardner Lindsey (ed), Handbook, 0 . §i§., pp 0 223.258 0 -140- There is same evidence in the research findings that the indices for involvement and commitment may not be completely adequate. For one thing, the sample showed a low level of involvement in agriculture. Approximately two-thirds of the respondents in reconstructing the situa- tion by which they became farmers indicated that either coercive or passive elements were fundamentally involved. In this light, the fact that 80 per cent of the sample indicated no preference for a job other than farming when they started farming, and the fact that more than nine-tenths of these currently felt that they would not "enjoy" any other job more than farming can take on a new'meaning. Rather than indicating involvement in agriculture, these facts may simply reflect a realistic appraisal of highly limited alternatives that leads to a resignation to rather than an involvement in agriculture. Other data support this reinterpretation. Given the hypothetical chance to relive their lives, 55 per cent of the sample members chose to remain in agriculture. But, of those choosing agriculture, one in four gave no positive reasons for their choice. Rather they exPressed such deep-seated resignation that they were unable to entertain seriously the hypothetical alternative posed. They could only see that they had no possibilities for employment outside of agriculture. Under these circumstances, the involvement questions could have tapped simultane- ousdy involvement (as defined) and forced resignation, its near Opposite, with regard to the dependent variable. If so, each level of involvement would contain two very different classes of people. This would account for the failure of involvement either to conform to the Guttman scale pattern or to relate to the rate at which functions were served by market news. .141- Going along to confound this possible substantive ambiguity of involvement were certain requirements of the factorial analysis schema. In the discussion of isotropy, the need to dichotomize involvement, even though a trichotomy seemed more apprOpriate, has been reviewed.260 The point of dichotomization of the involvement variable was between the two single largest groupings. In mean scores these two groupings never varied beyond .07 points on any of the three dependent variables. The possibility of the inapprOpriateness of the dichotomization is underscored by noting Table 13 where the results of an analysis of variance using all levels of involvement are contained. This test showed statistically significant differences at the .01 probability level for both latent and total scores. If the involvement index were not substantively ambiguous, this result would cast serious doubt on the dichotomous procedure. With the substantive ambiguity, its meaning is less definite. What is clear is that, within the limits of this analysis, involve- ment showed no differentiating power. Similarly, a better index for involvement is needed. This study will not attempt to "back track" and create it. The data are not present. The failure of the involvement index is still another limit of the research and concurrently a challenge to further, future research. The situation with respect to commitment is also somewhat hazy. An analysis of variance using all levels of commitment showed no signi- ficant differences on all three dependent variables, Table 13. This __ 260. See page 119. .142- Table 13 - Analysis of variance for Total, Latent, and Manifest Score Classified by Levels of Commitment and of Involvement. r l. “u“ f I Degrees Basis of Classification, Score, Sum of of Mean F and Analysis of Variance Item Squares Freedom Square Ratio Involvement - Total Score Total 2,242.77 355 Among 112.21 6 18.70 Within 2,130.56 349 6.10 3.07”” Involvement - Latent Score Among 49.21 6 8.20 Within 922.26 31.9 2.61. 3.3.1H Involvement - Manifest Score Total 803.34 355 Among 16 011,9 6 2 o 75 Within 786.85 349 2.25 1.22 Commitment - Total Score Total 2,242.77 355 Within 2,195.83 348 6.31 1.06 Commitment - Latent Score Total 971.47 355 Among 16.93 7 2.42 Within 954.54 348 2.74 .88 Commitment - Manifest Score Total 803.34 355 Among 19.54 7 2.79 Within 783.80 348 2.25 1.24 stands in contrast to the factorial result. these two, it could be argued that the significant outcome of the In attempting to reconcile factorial analysis was simply the result of dichotomizing a non- linear variable and not reflection of a ”real“ relationship. However, this interpretation is tenable only if it is also assumed that the levels of commitment used in the analysis of variance are meaningful -143- and distinct strata of the phenomena. In many respects accepting this assumption seems riskier than accepting the dichotomy. Commitment was constructed from an unweighted count of the occurrence of a number of disparate phenomena. In this case,the more cutting points used, the more errors in the placement of individuals would occur. Errors in placement would be at a minimum, everything else equal,with only dichot- omization used. Errors in placement would have no effect except as they involve error of placement across the dichotomy line. These would have to be the same for both forms of analysis. In addition, the analysis of variance would have additional chance for error in placement. Further- more, it is much easier to see why commitment should be related to functions score. Hence, the results from the dichotomy analysis were accepted. Quite naturally, the kind of reasoning engaged in here reflects on the validity of the conclusions. A further problem arises in interpreting the test results based on the dichotomized index of commitment. Should direction be used in interpreting the results of the factorial? If the analysis specifies direction, a one-tailed statement of probability applies and it becomes apparent that commitment is inversely related to manifest scores at the .05 probability level. High commitment levels have lower average manifest scores than do low commitment levels. This is a problem because common sense leads one to expect the reverse. The higher the commitment, the more important would it seem to be to use market news to make the most of one's situation. The inverse relationship also seems to contradict past research findings. Smith found a direct relationship between his index.of - 144 - commitment and exposure to market news.261 ‘While it has been argued that exposure to market news and use of market news are not identical, they are not unrelated. In the current sample, better than 60 per cent of those who were exposed to market news information at the time of sale cited market news as functional in deciding where and when to sell. Semantic differences are largely behind this apparent incongru- ity. Smith built his commitment index from the combined position on two Guttman scales, one measuring diffuseness of relations with dealer and the other measuring level of use for supplementary information sources. Essentially, these scales were reproduced in the current study. They are used as the structural variables “dealer" and “ration- ality,” respectively. Both of these variables, although not signifi- cantly related to manifest score, are positive in direction and, hence, do not contradict Smith's findings. Further consideration of the current commitment index and the manner of weighting makes the inverse relation.more understandable. In general, as commitment is measured in this study, the actor is not necessarily nor generally aware of his relative level of commitment. Advancing age and low educational attainment, in combination with finan- cial position and job experience, in particular, underlie "high" commitment. Then if normal correlative behaviors could be expected to hold, older persons would tend both to receive less information and use it less than would younger persons; farmers with larger Operations and more income would tend to receive and use more information than would 261. Smith, ggganization of thg_Farm, o . git., p. 64 ff. - 145 - smaller farmers, and so on. 262 At the same time certain of these associations signal conflicting tendencies. Fer instance, the larger farm Operations and higher incomes in farming, other things equal, tend to go to older Operators.' With other things not equal, the empirical case, there would then be a cross-checking tendency among, say, age, education, size of farm, and jOb experience. In part, this cross- checking may be responsible for the campgratively low differentiating ability Of commitment relative to the three remaining structural vari- ables. With respect to the remaining three structural variables, there is no evidence of inadequate measurement. It must be concluded that these do not significantly differentiate manifest functions. 263 The results to this point also demonstrate certain advantages of the factorial model in simultaneously forcing controlled comparisons and suggesting Objective standards for hypothesis acceptance or rejec- tion. What this can mean is demonstrated when the relationships of the five dichotomized structural dimensions to total, latent, and manifest scores without regard to the factorial controls are examined (See Table 14). While no directions Of difference change, the level of 262. See, for instance, Alfred Dean, H. A. Aurbach, and C. Paul Marsh, "Some Factors Related to Rationality,” o . git.; C. Paul Marsh and A. Lee Coleman, "The Relation of Farmer Characteristics to the Adoption of Recommended Farm Practices," Rural Sociology, 761. 20, (September-December, 1955) Pp. 289-296, and "Differential Communi- cation Among Farmers in a Kentucky County,” Rural Sociology, Vol. 20, (June 1955) pp. 93-101; McCormick, 22. .c_1_t,., pp. 45-46. 263. An analysis of variance run using all levels of these three vari- ables yielded results identical with those by the factorial analysis, hence, support this conclusion. See Table 94W, Appendix A. -146- Table 14 - Differences in Mean Scores, Standard Errors, and Z values of the Five Structural Dimensions with No Factorial Control. Tota1_§core§ Latent Scores Manifest Scores Variable G§;:§é 63143; Z 631;}? (iiiiz z ‘til’i2 031432 Z Commitment -.592 .535 1.11 -.295 .226 1.31 -.297 .221 1.34 Involvement .476 .543 .88 .301 .227 1.33 .186 .213 .87 Dealer 1.000 .537 1.86 .773 .228 3.39** .213 .217 .98 Neighbor .986 .540 1.83 .735 .226 3.25** .261 .217 1.20 Rationality 1.327 .533 2.4% 1.022 .225 4.54.. .305 .225 1.36 assessed significance does. Only four of the differences are statisti- cally significant in contrast to the eight in the factorial design. The beta, or type II, error is avoided by drastically reducing the sources of sampling error to which the means are subject by attaining homogeneity within the cells of the factorial. may; Vgiables, sgpglp Controls The second factorials necessary to the analysis schema involved adding controls to the set of structural variables shown to be signifi- cantly related to functions scores. Since only involvement failed to differentiate at the chosen significance level, it was temporarily drOpped from the analysis and the two "controls", source of market news and definition of market news, were added to the remaining four Structural variables. The results are shown in Table 15.264 261.. Tables lO-W and ll-W, Appendix A, contain the data from which these results are Obtainable. Tables 149W and 154W demonstrate the legitimacy of utilizing the Keyfitz model of analysis since the interactions are shown to be not significant. - 147 - Table 15 - Differences in Mean Scores, Standard Errors, and t Values Of the Four Significant Structural Variables Factorial Design with Controls Separately Added for Source of Market News and Definition of Market News. Total Scores Latent Scores Manifest Scores Eéridble Xi-XQ Xl-Xz t Xi-XQ Xi-XQ t Xl-XZ Xi-XQ t Commitment ‘0564 0300 1088 ‘0349 0183 1091 “0215 .160 1034 Dealer 0691 .217 3088*‘ .574 .158 3063** .117 0139 .84 Neighbor .810 o 321 2 o 52. o 633 .172 3 068*‘ o 177 .176 1. 01 Rationality .855 .293 2.92* .684 .194 3055** .171 .149 1015 Definition+ 1.027 .321 3.20** .515 .208 2.431' .512 .150 3.41** Commitment .590 .299 1.97 -.352 .158 2.23* -.238 .179 1.33 Dealer .762 .184 4.14" .614 .150 4.09** .148 .141 1.05 Neighbor .887 .275 3.23“ .611. .148 4033‘! 02.46 0163 1.51 Rationality 1.046 0312 3035** e780 0161 4084*’ .266 .170 1.56 Source++ .549 .223 2.45” .286 .165 1.73 .263 .123 2.14 -.- 121 refers to a definition of market news broader than the USDA's. ++Xi refers to the use of mass media plus informal sources or informal sources only. Two findings are particularly noteworthy. First, the insertion of "controls" does not alter the relationships of dealer relations, neigh- bor relations, or rationality to the three scores. Total score and latent score remain significant but, in all instances for all scores, the t value is reduced when the market news definition is used as a control. Thus, the prObability level for neighbor relations and rationality are reduced to the .05 level. To retain significance for commitment, a one-tailed interpretation is necessary. There is no change in the inability of these variables to differentiate manifest scores. -148- The picture does not change greatly when sources of market news is used as a control. The relationship of the structural variables remain as they were in the factorial without controls except for changes in the level of significance. All four structural variables are signifi- cantly related to total and latent scores at the .01 level with a two- tailed test except for commitment which is significant for latent score at the .05 level with a two-tailed interpretation and significant at the .05 level for total score with a one-tailed interpretation. The reduc- tion in t values is far less even when sources is the control. In fact, the t value increases for commitment. Since the original relationships are maintained there is no need to think of either source of market news or definition of market news as being by themselves necessary parts of the analysis of possible spurious correlation between the structural dimensions and the dependent behavior. The second fact standing out is that the "control" variables, while superfluous as controls, significantly differentiate the dependent be- havior (except for sources on latent score). Those persons who define market news more broadly than does the USDA have higher functions scores than those whose definitions conform more closely to that of the USDA. Similarly, those persons who use market news sources not limited to mass media tend to have higher scores than those who use only mass media sources. Of particular interest is the differentiating ability of the "control" variables with regard to manifest score. They are much more effective than the structural variables in this respect. In the pre- vious tests, only commitment had a significant relationship with manifest -149- score. However, even with a one-tailed interpretation, the importance of commitment in this respect was not supported with the insertion of controls. What can we conclude? It will be remembered that the t value in the factorial with no controls was only barely significant at the .05 level with a one-tailed interpretation. But, a one-tailed interpretation for the commitment variable is hard to Justify as pre- viously shown. Furthermore, mean differences and standard errors are both subject to sampling fluctuation and hence any obtained t values would fluctuate about the "real" value. Therefore, it seems reasonable and necessary under the rules of analysis to conclude that commitment does not significantly differentiate manifest scores. Before going on to the final step in the factorial analysis, a brief comment on certain further advantages of the factorial analysis can be demonstrated. This pertains to the problem of correlated bias in analysis which was discussed above.265 The factorial offers a distinct aid to that problem. One of the conclusions drawn above was that the structural dimensions exerted independent influence apart from the control variables. This conclusion of independent effect would be less evident from a simple contingency analysis. In fact, it could not be obtained except by some form of cross-tabulated analysis - a feature essentially forced by the sheer fact of the factorial. Short of a cross-classification of all five variables simple contingency analysis could estimate the possible independent effects by noting the joint distribution of the independent variables themselves. For example, it can be seen from Table 16 that those persons with high 265. See pages 47-53. -150- Table 16 - Contingency of the Control Variables and the Structural Dimensions+. Structural Market News Definition_ Market News Sources Dimension same or’less More than ‘Rass Mass and in- and Level than USDA's USDA's only formal or in- formal only Commit H 97 (+)” 101 101 (+) 97 Commit L 61 97 57 101 12 = .10 .8 > p > .7 f = 7.91 .02 7 p >.31 Deal Diff 73 H 133 85 H 121 Deal Spec 68 82 73 77 12 = 3.37 .1 7 p 7 .05 x2 = 1.92 .2 7 p 7 .1 Neigh Diff 67 (-) 107 76 (-) 98 Neigh Spec 74 108 82 100 12:01.6 o77p7o5 127-006 09>p7o8 Rat H 41 H 109 58 H 92 Rat L 100 106 100 106 X2 = 16048 P < 0001 12 = 3046 01 7 P 7005 * Involvement is not included for reasons given on pp. 140-141. ++ The sign indicates the direction of departure from the theoretically expected number. Only one sign is shown. All the other signs are fixed by the fact that the contingency has only one degree of freedom. Thus, the 101 must be negative, the 61 negative and 97 positive. rationality levels tend also to have broader definitions of market news than those persons with low rationality levels. The probability is significant at beyond the .001 level measured by chi-square. The dis- tribution of market news definition with regard to neighbor relations, dealer relations, and commitment is no different from that which.might be expected from chance. What would be the conclusion with respect to independent effect of each of the variables fron.a simple contingency ~151- analysis? Since definitions of market news are evenly distributed with respect to the two levels of each of neighbor relations and commit- ment, one would normally expect that any effect of definition differ- ences would cancel out across the two levels.266 This conclusion would be less certain for the dealer variable and denied for the rationality dimension. In the latter case, one could not be certain whether higher functions scores correlated with more inclusive market news definitions and with high rationality were independent or due to the fact that these two variables were themselves correlated, as shown by the simple contingency analysis. An even.more cloudy picture would hold with regard to the sources variables. While only one of the four relationships is statistically significant under a null hypothesis, it is also true that there is a tendency for diSproportionate numbers of persons with low commitment to use other than strict mass media sources; for persons who maintain diffuse relations to use such sources; and, for persons with high rationality levels to use formal and informal sources in contrast to solely mass media sources. Since it was shown that the use of other than strict mass media sources is associated with higher functions scores, the problem of correlated bias would again intrude. Ihree variableg, 23g Controls It was indicated earlier that it would not be possible to consider simultaneously the "system" and "control" variables in full. However, 266. Such an assumption neglects the possibility that interactions occur, in which case, this sort of analysis is completely in- validated. The factorial analysis gives direct evidence on interactions. -152- the results of the factorials analyzed to this point indicate that the need for this sort of assessment is less than might have been supposed. Nonetheless, it seemed safer to consider both control variables in relation to the structural dimensions. Involvement, for example, has not yet been examined with respect to the controls. Two further factorials were produced. The first included the three structural variables with the greatest mean differences in average effect; neighbor relations, dealer relations, and rationality. The second included involvement, commitment, and rationality. Commit- ment was added because it had to be excluded from the other factorial. Further, the results of the interaction analysis in the factorial that included the five structural dimensions indicated the unique joint dis- tributions of these two, and logically, of course, commitment was defined residually to involvement. Rationality was added because, in one sense, nothing could be lost by its inclusion while, on the other hand, some insight might be gained into the relations of involvement, commitment, and rationality. As will be shown these three tend to be fairly highly correlated in occurrence. The results of the two factorials are given in Table 17.267 Two features are worth noting. The first is the general stability of all the relationships. Involvement remains unrelated to the functions scores whereas all the remaining structural variables are significantly related to total and latent scores. A one-tailed interpretation is required to show significance for the relationships of commitment and 267. Tables 124W and 134W, Appendix.A, contain the data on which these results are based. Tables 164W and 174W show that the interactions are not significant. ~153- Table 17 - Differences in Mean Scores, Standard Errors, and t Values for Three Structural Dimensions and Two Control Variables. Total Scores Latent Scores Manifest Scores variable “i132 @1352 t @132 Git-72 t mm t Dealer .661 .223 2.96** .530 .129 4.11** .131 .173 .76 Neighbor .935 .294 3.18** .676 .142 4.76Mr .259 .181 1.43 Rationality .740 .208 3.568% .579 .120 4.82** .161 .130 1.24 Definition 1.008 .224 4.50** .448 .146 3.o7** .560 .180 3.11** Source .544 .167 3.26** .335 .118 2.84* .209 .130 1.61 Commitment -.598 .323 1.85 -.348 .184 1.89 -.250 .176 1.42 Involvement .364 .297 1.23 .230 .194 1.19 .134 .145 .92 Rationality 1.007 .243 4.14** .806 .177 4.55** .201 .122 1.65 Definition .911 .342 2.66* .447 .216 2.07 .464 .161 2.88* Source .447 .260 1.72 .240 .169 1.42 .207 .120 1.72 of market news definition to latent scores in the commitment-involvement- rationality factorial. The differentiating ability of market news definition for manifest score is maintained while the structural vari- ables continue no differential. The second point worth noting is that market news source fails to show a significant relationship with manifest score in the first facto- rial even with a one-tailed interpretation. In the second factorial it fails to differentiate any of the three sets of scores. This instability of sources as a control is more apparent than real. The mean differ- ences and their standard errors are both subject to sampling variability. ~154- Even if the actual value of mean difference for the dichotomized vari- able is such that it would be measured as statistically significant, sampling variability would still lead to some fluctuations in the results of particular "t" tests. This may be the case here. In effect, the five factorials correspond closely to the statisticians‘ classic "balls in an urn." Each reSpondent represents one ball with ten values printed on it; three for the functions scores, five for the structural dimensions, and two for the control variables. Instead of reaching into the urn and drawing out balls, the spheroids are preassigned on the basis of whether they are high or low rationality, high or low commitment, and so on. But, for each factorial something occurs that corresponds to the drawing of a new sample - by returning the balls to the urn and redrawing. While in no instance will any of the inscribed values change, the group of persons who constitute a cell change from factorial to factorial. For instance, take the cell containing high commitment, high involvement, diffuse dealer and neighbor relations, and high rationality cases. In the next factorial where market news definition is substituted for involvement some reSpondents will drop out because they have a USDA type definition and others will be added in who, in the first factorial, were summed for functions score in a different cell. Each factorial, then, represents a new "sample" from the universe of samples possible under the fixed assignments of the independent and dependent variables. The sources variable then might sometimes show "non-significant" test results even when its true value is one which, under the rules of analysis set out earlier, would make the variable a "significant" differentiator. -155- It is true, of course, that this line of reasoning is two-edged. One could also argue away significant test results with the same rationale. However, one is hard put in this second cutting position when the test results: (1) show differences whose probability of occurrence through sampling variability is low and (2) the tests show stability at these levels through replication. In order to make this situation clearer the mean differences in means over the dichotomies, the standard errors, and t values for all independent and control vari- ables in the five factorials are collected in Table 18. The high stability particularly for the first two columns is readily apparent and needs no comment. Quite naturally, the t value shows somewhat greater variability but they too are markedly limited in fluctuation. Evaluation 2; thg Structural Dimensions' Relations 39 Functions Sggggg In the preceding section the relations of the structural variables to functions scores were presented. It was found that diffuse dealer relations, diffuse neighbor relations, high rationality, and low commitment were significantly related to higher total and latent functions scores than were their dichotomized opposites. Involvement showed no such significant relationships. None of the structural vari- ables were significantly related to manifest score. Mereover, the control variables did not alter the effects of these structural vari- ables but did themselves constitute sources of significant differentia- tion in their own right. Heepondents with market news definitions which included more types of information than are contained in the USDA program had higher functions scores for all categories; total, latent, and manifest. -156- Table 18 - Mean Differences in.Nbans, Standard Errors, and t values for the Structural and Control variables Collected Over the Five Factorials.+ “ * Variable and Function Score Mean Difference Standard Error t Value Commitment” Total 77, 56, 59, 60 30, 30, 30, 32 258, 188, 197, 185 Latent 46, 35, 35, 35 20, 18, 16, 18 237, 191, 223, 189 Manifest 31, 21, 24, 25 17, 16, l8, 18 182, 134, 133, 142 Involvement Total 35, 36 23, 30 152, 123 Latent 20, 23 17, 19 116, 119 Manifest 15, 13 ll, 14 136, 92 Dealer Total 80, 69, 76, 66 19, 22, 18, 22 428, 318, 414, 296 Latent 62, 57, 61, 53 15, 16, 15, 13 421, 363, 409, 411 Manifest 18, 12, l5, 13 14, l4, 14, 17 133, 84, 105, 76 Neighbor Total 87, 81, 89, 94 25, 32, 28, 29 353, 252, 323, 318 Latent 65, 63, 64, 68 13, 17, 15, 14 483, 368, 433, 476 Manifest 22, 18, 25, 26 16, 18, 16, 18 136, 101, 151, 143 Rationality Total 107, 85, 105 27, 29, 31 396, 292, 335 74, 101 21, 24 356, 414 Latent 80, 68, 78 14, 19, 16 551, 353, 484 58, 81 12, 18 482, 455 hhnifest 27, 17, 27 17, 15, 17 156, 115, 156 16, 20 13, 12 124, 165 Definition Total 103, 101, 91 32, 22, 34 320, 450, 266 Latent 52, 45, 45 21, 15, 22 248, 307, 207 Manifest 51, 56, 46 15, 18, 16 341, 311, 288 Source Total 55. 54, 45 22, 17, 26 246, 317, 172 Latent 29, 33, 24 16, 12, 17 173’ 284, 142 Manifest 26, 21, 21 12, 13, 12 214, 161, 172 + All values in the table have been multiplied by 100 and the thousands digit rounded. ++ Negative signs on mean differences have been omitted. ~157- The factorial design utilized in this analysis allows an assess- ment of both the independence of effect for the various structural variables and the possible interactive effect between and among these variables. This section will examine the results with respect to the more general question of whether the defined "structure" constitutes a system and whether such a definition of structure has pragmatic value. The Structurgl Dimensions 5 g Sygtgg System can be investigated in two different senses. The first is the degree to which the structural variables ointl gffggt the depend- ent variable. Here the question is one of systemic determination of effects. System in this sense will result in the variables assuming a "complex" form where the change in value of any one of the variables alters the effect of all on the dependent variable. The second sense of system is what we call a "simple" form. It is the degree to which the structural variables have joint occurrence irreSpective of whether they have "complex" systemic determination. Positive evidence for "simple" system is the non—uniform.distribution of the independent variables with respect to one another. The factorial is intrinsically geared to answer questions about "complex" systems. By controlling out the effects of the remaining structural variables, the independent effect of a single variable could be isolated. The results reported showed that the significantly related variables did have independent effects, irrespective of their impact considered simultaneously. At the same time, and without exception, the relations of the interactions to the various dependent function score means were not significant. In the less rigorous form, simply ~158- as the difference of differences, only 2 of the possible 84 first order "interactions" for the three function scores in the five factorials were statistically significant. A significant difference of difference between commitment and involvement for total and latent scores was found. Interpreted strictly as an interaction, a significant probabil- ity level was not attained. Furthermore, the independent effects of involvement were not significant and, in repeated trials, the differ- ence in differences did not continue to be significant. It seems reasonable to conclude, therefore, that the two interactions occurred by chance. This failure to show significant interactions offers evi- dence of the non-existence of what has been called previously a "complex" system. Without significant interactions there is no evidence that the effect of any of the variables on the function scores changes when the levels of the other variables are changed. This is what inter- action means in a factorial analysis. The evidence indicates that the variables in the study which define structure gag pgt svstemic ig determining consequences. The answer to whether the variables form a "simple" system is almost unequivocal. Since all the cells of the first factorial con- tained four or more cases, there cannot be complete correlation among any of the structural variables. The existence of associations among the structural dimensions was also more carefully considered. The independent variables significant in differentiating function scores were investigated by chi-square tests of association, Table 19. Only ~159- Table 19 - Contingencies Among the Structural Dimensions Significant in Differentiating Function Scores. Structural Structural Dimension and Level Dimension Rationality Dealer g_mNeighbor and Level High Low Diff Spec Diff Spec Commit H 91 (+)* 107 123 (+) 75 93 M 105 Commit L 59 99 83 75 81 77 12 = 2.77 .1>p7.05 12 = 3.30 .lrp'.05 12 == .63 .5713 7.3 Ration H’ 101 (+) 49 79 (+) 71 Ration L 105 101 95 111 12 =14.27 p < .001 X2 =1./.9 .3’p » .2 Dealer Diff 105 (+) 101 69 81 1273085 05713703 * The sign indicates the direction of departure from the theoretically expected number. Only one sign is shown. All the other signs are fixed by the fact that the contingency has only one degree of freedom. one of the six relations was statistically significant.268 Persons with high rationality levels also tended to have diffuse dealer relations; the magnitude of the chi-square was significant beyond the .001 level of probability. Furthermore, the results of an analysis of all five structural dimensions, utilizing as many levels of each as possible 268. Two of the remaining five test results showed probabilities between .1 and .05. However, in both instances, the meanings of the possible relationships are somewhat anomolous. Persons with high commitment tended also to have both high rationality levels and diffuse dealer relations. But, these levels of structure have contradictory effects on function scores. High commitment is associated with lower functions scores than is low commitment, whereas high rationality and diffuse dealer relations have higher function scores. - 160 - without violating the suggested minimum expected frequencies, showed only small differences from those shown by the analysis of the 2 x 2 tables. Involvement showed no significant relations under the null hypothesis with the distributions of any of the remaining four structural variables. In this more extended analysis the degrees of freedom varied between 16 and 25 and, hence, allowed finer variations in the joint distributions to be expressed. The relations of the structural variables to the control variables discussed earlier (see Table 16) are also relevant here. Without reviewing the details, it can be stated that even expanding the definition of system to include the control dimensions does not greatly alter the general unrelatedness among the study variables in terms of joint occurrence. The conclusion seems justified that the definition of "system“ used in the study does not correspond in any simple fashion to units of the existential world. The variables of structure do not cluster markedly to reveal clearly recognizable types. However, this does not critically damage the significance of the analysis. It is part of the sociolOgist's job to seek out significant typologies where none was generally believed to exist. Current methodology seeks aspects of true, existential systems which are relatively independent both in effect and distribution so as to maximize pragmatic predictive goals under limited knowledge and ways of knowing. The paramount criterion of prediction is our next point of attention. Thg Pragggtic Criterion gag §tructurgl Dimensions One of the more important features of factorial analysis stems -161- from its basis in the analysis of variance. It is possible to parti- tion the total variance observed in the dependent variable among the variables used in classification. It is then possible to assess the amount of total variance accounted for by all of the independent vari- ables together. The greater the amount of variability which can be explained, the higher is the probability of predicting correctly the dependent behavior when given knowledge of the independent variables. Predictability is inextricably part of any pragmatic criterion measuring research. Following Keyfit2269, we can look upon each observation of the 356 as consisting of a sum of effects independently arising from the particular relationships of the structural dimensions and control dimensions to the given dependent variables, ;.g., total, latent, or manifest score. Thus, for example, on latent score, in the case of the factorial including only the five structural dimensions, any single score algebraically is wixyz =‘m + av +~bw + cx +-dy + evwxyz, where m is the grand mean for all 356 observations, av is the level of commit- ment, bu the type of dealer relations, ox the type of neighbor relations, dy the level of rationality, and evwxyz is the unexplained variability. Any score to be predicted then is predicated on first assuming that the grand mean is the best estimate, plus or minus one-half the observed mean differences in score for the remaining four classification vari- ables. These mean differences represent the effect of the given inde- pendent variables on the dependent behavior. In the present example, 269. Keyfitz, 22. 923., p. 478. -162- the independent variables have the following values: a1 = high commitment =-.232 a2 = low commitment = +.232 bl = diffuse dealer relation 2 +.310 b2 = specific dealer relation = -.3lO c1 = diffuse neighbor relation 2 +.324 c2 = specific neighbor relation = -.324 d1 = high rationality = +.400 d2 = low rationality = -.LOO m = grand mean = 3.722 To predict for the cell, high commitment, diffuse dealer relations, diffuse neighbor relations, and high rationality, we take 3.722 (the grand mean); subtract .232 for commitment since high commitment has lower scores than low commitment; add .310 for dealer because diffuse dealer relations have higher scores than specific dealer relations; and, for similar reasons, add .324 for neighbor relations and .400 for rationality. No estimate for involvement is used since the analysis did not show this variable as significantly differentiating. Then the expected value for this cell is 4.524. In similar fashion, the expected value of each.cell can be esti- mated from the prediction equation. The difference between the sum of squared deviations about this hyperplane and the sum of squared devia- tions about the grand mean gives an estimate of the amount of total variability explained by the differentiating variables. The total sum of squares about the grand mean is 971.4691 while that about the hyper- plane is 813.0307. Hence, the total sum of squares accounted for is 158.4384. The squared standard deviations, or variances, for the mean and the hyperplane respectively are 2.7288 and 2.2838. The variances -163- about the arithmetic mean and the "regression" hyperplane27o measure the degree to which the mean and the hyperplane fail to characterize the data completely. The difference between the two variances measures the amount of original variability about the arithmetic mean eliminated by knowledge of the structural variables. The proportion of this vari- ability eliminated is given by the ratio of this difference to the original variability (i.g., 2.7288 - 2.2838 / 2.7288 = .1631) and is the coefficient of determination, r2. 271 Just over 16 per cent of the variance is explained by the four variables used here. Similar results were obtained when latent scores were predicted from the remaining factorials, Table 20. The lowest per- centage was derived from the dealer-neighbor-rationality factorial with two controls, 16.2 per cent, and the highest was with the four struc- tural dimensions - 'market news definition factorial utilizing the information available on the definition's relation to latent score. No predicting equation was written for the second, two-control factorial because of the absence of differentiating variables. The percentage of explained variance was less in every case for total scores than for latent scores. This is probably a reflection of wider range of variability, the low correlation between manifest and latent scores, and the non-differentiating quality of the structural dimensions for manifest function scores. For this latter reason, also, “...—...“ 270. Regression is not literally involved here as previously indicated, "the procedure of grouping and working with factors at two levels do not yield direct estimates of regression coefficients." Yates, 22° 9.13.09 P0 3130 271. See the discussion in Pearson and Bennett, 92, git., p. 146 ff., especially p. 149. -164- Table 20 - Equation for Predicting Total and Latent Functions Scores and the Per cent of Variance Accounted for by Bach. Per cent magi: and Predicting Equation“ Variance Explained 5 structural dimensions total score 6.025‘i.387a 3.402b 1.433c 1.533d 12.54 5 structural dimensions latent score 3.722 1.232a 3931013 1.3246 i.4.00d 16.31 4 structural dimensions and definition control total score A structural dimensions and definition control latent score 4 structural dimensions and sources control total score 4 structural dimensions and sources control latent score 3 structural dimensions and both controls total score 3 structural dimensions and both controls latent score 6.025 1.28221 1.346b 3.4.056 35.41.21 1.51m 30722 iel728 i0287b 133160 i0342d 1.2586 6.025 ¢.295a 1.3811: 1.4.1.12: 1.523d 1.27M 3.722 i.l76a 1.30% i.320c 3.39% 6.025 1.3301) 1.468c 35.370d 1.504e 1.2721“ 3.722 1.2651: 1.338c 1.29% $22.49 5168: 18.73 19.12 14.29 17.73 16.02 16.22 * Where a = commitment level; b = type of neighbor relations; c = type of dealer relations; d = rationality level; 6 = market news defini- tion; f = sources of market news. -165- no predictions were made for scores on manifest functions.272 All of these prediction equations constitute "explanations" of variance far above chance under the null hypothesis. For the lowest coefficient of determination, five structural variables on total score, the corresponding r273 is .354. With 356 observations such a value is highly significant. The tabular value for the .01 level on a four vari- able multiple coefficient with 200 degrees of freedom is .235 and for 400 degrees of freedom is .167. 274 While this had to be taken with caution because multiple correlations tend to rise rapidly due to associations among the error variances as more variables are added, there is still a basis for concluding that the predicting equation is Operating above chance. Of course, the crucial question is whether the predicting equations derived from the variables of the study are substantively meaningful. Certainly the largest part of the variance is left unexplained. How- ever, in view of both the eXplanatory nature of the study and profes- 272. In a.more extended investigation of patterns of manifest functions, particularly the role of market news in the sales situation, no satisfactory account could be made for differentials in usage. See Smith and Sim, gp. £13., p. 83 ff. 273. Pearson and Bennett, 91),. 931., p. 150. 274. Ibid., p. 412. -l66— sional precedents,275 the level of prediction would seem of sufficient magnitude for at least suggesting the need to probe some of the leads turned up thus far in the analysis. This will be the task of the next chapter where data bearing on the functional hypothesis of constraint will be examined. 275. Keyfitz, upon whose work the analysis to date is largely built, could explain but seven per cent of the total variance in family size. Q§,, also, E. M. Rogers, "A Conceptual variable Analysis of Technological Change," Rural Sociology, vol. 23, (June, 1958) pp. 136-145. F. c. Fliegel, "A Multiple Correlation Analysis of Factors Associated with AdOption of Farm Practices," Rural Sociolo , vol. 21, (September-December, 1956) pp. 284-292. J. R. Christiansen, "The Behavioral Correlation of Membership in Rural Neighborhoods," Rural Sociology, vol. 22, (March, 1957) pp. 12—19. C. F. Westoff, P. C. Sagi and E. L. Kelly, "Fertility Through Twenty Years of marriage: A Study of Predictive Possi- bilities," American Sociological Review, vol. 23, (October, 1958) PP. 549‘5560 The per cent of variance explained and the number of inde- pendent variables used in the prediction for these studies respectively are: 16.7 per cent with six; 32 per cent with six; 12 per cent and 33 per cent with six; and, 27 per cent with nine. ~167- Chapter V The Structural Diminsions and Functional Constraint Iptroduction The present chapter examines the data more closely with respect to the functional hypothesis of constraint. Of the five structural dimensions originally included in the study, dealer relations, neighbor relations, and rationality have been accepted as relevant and valid on all considerations. All three conform to the expected scale pattern and have been shown to be appropriate by the criteria adopted. All three significantly differentiate both total and latent functions scores, though none of them significantly differentiate manifest score. Lastly, all three continue to differentiate when controls for differ- ences in market news sources and market news definitions are instituted. Propositions concerning the relevance of commitment and involve- ment were not supported. In the case of involvement, there was some doubt as to whether the index used measured a unidimensional area or a compound of involvement and resignation. Mbreover, the dichotomy required for the factorial probably did not represent the data most satisfactorily. ‘Whether the rejection of the involvement variable as significant for differentiating function scores is attributable to these features or to a lack of true relationship could not be made. Under these circumstances it seems fruitless to continue to probe into the role of this factor and, therefore, it will be omitted from further consideration in this chapter. The appropriate disposition of commitment is considerably more uncertain than that of involvement. Of all the structural.variables -l68- originally proposed, it appeared least likely to be unidimensional. Experience in operationalising the factor supported this expectation. The first working assumption about commitment, however, was that it was linearly related to the dependent variables. As the analysis proceeded, evidence contradicting this assumption appeared. In the factorial , commitment generally was shown to be related inversely to both total and latent scores at statistically significant levels. However, a one- tsiled interpretation had to be used for this conclusion. The nature of the cmitnent index made such an interpretation tenuous. Further- more, an analysis of variance of function scores using all the levels of eminent showed no significance. The implication of this was that the result obtained in the factorial might have been an artifact of the dichotomisation procedure .276 The ambiguity connected with this vari- able has led to the conclusion that further analysis ought also to be deferred on commitment. Therefore, it will not be considered in this chapter. The evidence about the "control" variable, market news comes, closely paralleled that concerning commitment. It did not maintain a constant relationship with the dependent variables. Of the six assess- nents of its relationship to latent and total scores, half were not statistically significant and two of the remaining three yielded test results barely significant at the .05 probability level. While these results merit further investigation for the variable, it caused best 276. It is also true that assmptions of risk were involved in the analysis of variance of co-itnent using all levels. Sea PP- “2’1430 -169- here to eliminate it from the analysis.277 Variations in market news definitions were clear cut in their effects. This factor showed an independent effect on all functions scores in all factorials. A broader than the USDA type of definition was significantly related to wider use of information for various functions. Because market news information is the item to which functions were to be imputed under the original design "hypothesis," its incorporation in the analysis was also imperative apart fran this research result. The relation of the function scores to the structural dimensions was also clear cut. Total and latent scores always showed a statisti- cally significant relationship to the structural variables cited above. Manifest scores were only occasionally related. Hence, it did not seen worth while to probe the specific substantive behaviors composing this dependent variable. Since the total score was simply an arithmetic on of latent and manifest scores, the total score contained no specific behaviors not subsumed in its two components. Therefore, this chapter will be concerned with probing how certain of the structural dimensions are related only to the various specific behaviors constituting the latent functions score. 277. The reasoning behind this unlusion applies to those considered prior. Under conditions of low substantive knowledge about what news attempting to w or]: both with relationships between variables not completely consistent (i.g., commitment and market news source) or with variables ambiguously indexed (1.3. , involvement) confounds interpretation. Uith these conditions , any incongruity in the substance of the data leaves uncertain the source of the 'error.‘ Is it in the methodology or is it in the substance? By diminishing the level of possible methodological ambiguity more definite insight into the substance of market news might occur. -l70- A two-dimensional structure will fonn the backbone for the analysis. On one axis will be the nature of social relationships, and, on the other, the level of rationality. The type of definition of’market news will be controlled on both axes in this respect. The distributions of the specific latent functions will be considered with respect to the structure in which they appear and possible or plausible 'explanation' given for the distribution as it occurs. In this way it will be possible to gain some insight into the hypothesis of structural constraint. This investigation will also serve as grounds for specific functional hypoth- eses concerning market news to be verified in future research. This procedure is in accord with the sentiment that, most hypotheses in survey research are formulated ggtg; examining the data. There is nothing intrinsically wrong with this procedure, as long as the hypotheses are subse- quently tested on gthgg data. But when the hypotheses are tested on the same data that suggested than 2&3 thon a spurious inpression of validity may result. WMMWW WQWM It was shown in Chapter IV that both dealer and neighbor relations had independent effects in the same direction on latent scores when levels of all other variables were held constant. Therefore, a new and rather simple social relations typology can be constructed to simplify the structural types. There are three broad possible types of persons when both of the social relations scales are considered Jointly'with respect to their effects on latent functions scores. There are persons 278. 8.1.7111, ”e we, p. 5&0 -171... who are diffuse in both relationships as well as those who are specific in both. For either of these types the net effect of the component variables on latent scores is unambiguous. Both diffuse dealer and diffuse neighbor relations are correlated with higher function scores, while specific relationships are associated with lower scores. A third, nixed grouping is also possible. A person could maintain specific relations with his dealer and diffuse relations with his neighbors or vice versa. In either case, the effects of each variable on latent scores should cancel each other. Both variables produced mean differences in latent scores of similar magnitude when they were dichotomised.279 In terms of mean functional scores this mixed grouping falls between the completely diffuse and the canpletely specific types. There will be no alteration in the way that rationality is treated. Crosseclassification of the new social relations typolog with rational- ity yields six types of structure: diffuse relations, high rationality; diffuse relations, low rationality; mined relations, high rationality; mixed relations, low rationality; specific relations, high rationality; specific relations, low rationality. The relation of these structural types to each of the various latent functions will be the basis for evaluating the "hypothesis" of structural constraint. These functions will be considered in the descending order of frequency of occurrence in the sample. 279. See Table 18. -l72- mnnnnmnmemnmmwrs mm The most common latent function of market news is its use as an easy tepic of conversation with others in a variety of informal situa- tions. Data with respect to this function are contained in Table 21. Table 21 - Use of Market News as a Convenient Conversation Piece Among _ the Structural Types. Definition of Market News Structural more tgg USDA as or less Total.— Type Total 1 Total % Num- S Ies _ Yes ber Yes Specific-Lo Rationality 28 64.3 29 51.7 57 57.9 Specific-Hi Rationality 15 86.7 9 100.0 2!. 91.7 Mixed-Lo Rationality 50 92.0 1.8 83.3 98 87.7 Mixed-Hi Rationality 54 88.9 18 72.2 72 84.7 Diffuse-Lo Rationality“ 27 96.3 23 100.0 50 98.0 Diffuse-Hi Rationality 1.0 90.0 11. 86.7 54 88.9 Chi-square test d.f. 12 Social relations alone 2 23.21 p < .001 Rationality alone 1 1.92 .2 > p > .1 Market news definition alone 1 5.39 .05 7 p 7 .02 Relations and rationality, no control 5 39.69 p < .001 Relations and rationality, USDA definition 4 22.77 p < .001 Relations and rationality, ' One case, no infomation. The distribution of cases by structural types for the sample is as shown. In the remaining tables of this chapter showing the distribution of latent functions, unless other- wise noted, deviations from this distribution are attributable to cases of no information. ~173- The most striking aspect of the data is the sharp departure of the specific-low rationality grouping from the remaining five classes. In the comparison without the control for market news definition the latter groupings report frequencies of such use from 84.7 per cent up- ward. In contrast, only 57.9 per cent of the manbers of the specific low rationality grouping report a conversational use for market news. This picture does not change greatly with the definition of market news controlled. There is a slight tendency for greater use for all struc- tural classes with definitions of market news broader than the USDA. The distribution of this function by the structural classes is statis- tically significant both with and without the market news definition control. Most of the variability in the distributions is due to the overappearance of the function in the diffuse groupings and the under- appearance in the specific-low rationality class. Since market news is occupationally specific and has limited interest outside of agricultural contexts, this distribution is perhaps understandable. while farmers of the specific-low rationality type operate in a structure which does not facilitate conversation with others with like occupational interests, all five of the remaining types of structure have at least one facilitative condition for easy and appropriate use of market news in conversation. People who have diffuse relations with either dealer and/ or neighbors have many occa- sions for sociability with others interested in agriculture. Similarly, persons with high rationality scores visit more with occupational specialists than do those with low rationality scores. Thus, the social relations variable is significantly related to the function and and in the correct direction at beyond the .001 probability level. ~174- The rationality variable, although in the correct direction, has a probability of occurrence between .2 and .1, Table 21. These five structural classes all have more Opportunity to use market news for a convenient conversational item. The increased opportunity for use hypothesis is sharply focused by asking how the members of the specific-low rationality type can behave in this manner at all, since it is implied that there is so little opportunity for the function to occur. The question can be answered in part by recalling the operational definition of ”specific” neighbor relations. Dropping by to see a neighbor without engaging in such more personal social intercourse as picnicking, hunting, sharing trips, and so on was accepted as the second level of intimate neighbor relations. In diohotomising, this second level was placed in the 'specific' relations category. Thirty-one of the 57 specific-low ratimality cases were on this second level and only 33 cases of the conversational function for the specific-low rationality grouping were reported. Further probing showed that 17 of the 31 cases from the second Guttman type reported using market news in this conversational manner. If we can assume that most of those persons used market news in the conversational function with neighbors, and evidence to this effect will be given below, than less than half of the occurrences of this function need to be explained possibly by reference to other types of occasions. Other data also lend support to the interpretation of limited Opportunity. When the six typologies are ranked on the basis of the percentage of constituents attending one or more meetings of the Farm Bureau and local cooperatives, or participating in the Soil Conservation .175... Service (SOS) progran, the specific-low rationality grouping was always last, Table 22. The actual percentages for this grouping were low in Table 22 - Persons Attending Earn Bureau and Coop matings and Participating in the 808 Program Among the Structural Types. * _~ M Structural Total Ear cent Attgnding mitigating m. arm can oops in 805 Specific-Lo Rationality 57 10.5 1.7 15.8 Specific-Hi Rationality 21. 29.2 20.8 58.3 Mixed-Lo Rationality 98 11.2 13.3 16.3 Mixed-Hi Rationality 72 1.4.1. 40.3 56.9 Diffuse-Lo Rationality 51 15.6 9.8 29.4 Diffuse-Hi Rationality 54 44.4 31.5 48.1 W *- all three areas, being under two per cent for attendance at coop meetings, Just over ten per cent attending Farm Bureau meetings and Just under sixteen per cent participating in the 508 program. The remaining five structural types ranged frost a low of Just under ten per cent for coop attendance to a high of Just over 1.0 per cent; frm Just over sixteen per cent SCS participation to almost 60 per cent; and, finally, fro- Just over eleven per cent to almost 1.5 per cent on nil-a Bm-eau attendance. In all three cases, high rationality is distinctly related to higher participation. with type of social relations con- trolled, high rationality types always exneed the low rationality types.280 The nagniimde of difference between the specific-high * 280. This is true partly by definition but it does not alter the bearing on the hypothesis of opportunity. ~176- rationality and specific-low rationality types is particularly impor- tant since here opportunity for both grotmings is minimal with respect to neighbors and dealers. If "opportunity“ is to be made up, its source has to be the features associated with the rationality dimension. The specific-high rationality class has participation rates in the Farm Bureau and 808 three times greater than those of its opposite and ten times as great attendance at coOp meetings. In addition to being asked whether market news was used for sociable conversation, respondents were asked with when this was normally done. If the information was not included in the free response, answers were probed to ascertain whether such behavior was carried on with persons not well known. In general, persons with high rationality levels were more given to using market news in this sense with strangers. In the high rationality classes 60 of 129, or 1.5.8 per cent, were interacting with strangers when using narket news as a means of engaging in sociable conversation. In contrast, only 58 of 168, or 34,5 per cent, with low rationality levels included strangers when using larket news in this way, Table 23. This difference was significant statistically at a .05 probability level. This general tendency' was also reflected in the specific-low rationality class. 01' the 33 who did use market news in this way, two in every three indicated they did not talk with persons not well known. In this respect persons with high rationality levels, irrespective of the type social relations maintained have naxinun opportunities for using narket news in this way. These opportunities are particularly important for people when their relations are specific. When farmers of this type use narket news in this way, they tend to do so with -177. Table 23 - Persons Who Discuss the Market News They Get With Persons Not Well Known Among the Structural Types. 4“ J 4" t J g Structural er cent Who so Type ““1 with strangers Friends only Specific-Lo Rationality 33 33.3 66.7 Specific-Hi Rationality 22 1.0.9 59.1 Mixed-Lo Rationality 86 31.4 68.6 Mixed-Hi Rationality 61 41.0 59.0 Diffuse-Lo Rationality 1.9 1.0.8 59.2 Diffuse-Hi Rationality 1.8 54.2 45.8 peOple they know well even though the maintenance of specific relations with neighbors and dealer impairs the development of relationships with other persons with whom market news can be conversed appropriately. This feature would bear hardest on the specific-low rationality class. In addition to having more opportunities, persons with high ration- ality levels might see two functions in the use of market news in con- versation and consequently be given greater motivation to use . (hi the one hand, market news may be used to promote and facilitate sociability, while, on the other hand, it may be preliminary to collecting and organising information to help in the rational planning of marketing. Those with high rationality levels would be expected to see these impli- cations more than those with low rationality levels. Some of the data collected were revealing in this respect. Respondents were asked in the interview whether they ever used market news information they received to perk up or prime a lagging -178- Table 24 - use of Mkrket laws as a Conversation Primer Among the Structural Types. # e t on ket e Structura1 More than USDA Same or less _To§gl Type Total $ Total 1 Num— 5 Tea Tea ber Ies Specific-Lo Rationality 27 48.0 29 27.6 56 37.5 Specific-Hi Rationality' 12 41.7 9 55.5 21 47.6 Mixed-Lo Rationality 48 50.0 46 1.3.5 94 46.8 Mixed-Hi Rationality 51 60.8 18 44.4 69 56.5 Diffuse-Lo Rationality 28 64.3 23 60.9 51 62.7 A conversation and with‘whan'this might be done. A.pattern similar to the less conscious manipulative use of market news in sociable conversation emerged, Table 24. Fifty-eight and three-tenths per cent of persons with high rationality levels used market news in this 'priming' sense while only 48.3 per cent for those with low rationality levels reported such.use. This difference had a probability of occurrence under the null hypothesis of between .1 and .05. But, not only did the high rationality cases tend to use market news more often in this way, they also showed greater propensity to do so with.pe0ple whom.they did not know well, Table 25. The difference of 13.7 per cent is statistically significant when evaluated by chi-square. The explanation of limited opportunity supports the hypothesis of structural congruence. In advancing “limited opportunity“ as an.explanae tion, it can be asked whether or not opportunity for usingpmarket news in a conversational sense is a.result of forced circumstances rather -l79- Table 25 - Distribution of Persons Who Use Market News as a Conversation Primer With Persons Rot Well Known Among the Structural Types. V Structural Total or c t Who 3 us e Type With strangers friends only Specific-Lo Rationality 21 38.1 61.9 Specific-Hi Rationality 10 50.0 50.0 Mined-Lo Rationality 44 40.9 59.1 Mixed-Hi Rationality 39 53.8 46.2 Diffuse-Lo Rationality 32 46.9 53.1 Diffuse-Hi Rationality 35 60.0 40.0 fi‘ fi than choice, as is implied. It is conceivable that persons in the low rationality-specific class might be new to the community and, hence, more likely to be denied diffuse type relations. Time is often a factor in deve10ping this mode of social intercourse. Again, persons in the low rationality-spec ific relations class might work off the farm more and have less time to socialize with neighbors and dealer. Neither of these possibilities are supported by the data. Loss than six per cent of the entire sample reported ”living around here. less than five years. When the percentages of each type residing in the area eight years or less were compared, three of the five groupings had larger proportions of such newcomers than did the spec ific-low ration- ality class, Table 26. Paralleling length of residence was the distri- bution of 'close' friends living in the community of residence. Three of the five groupings had higher proportions of persons who claimed to have “none” or only “a few friends” in the calmunity, Table 27. Mabel-s -180- table 26 - Years of Residence in the Community Among the Structural TIP“- Years of Residence Structural Total Less 8 15 30 Entire Type than to to or life _ 8 15 30 more ------ per cent - -1- - - - - Specific-Lo Rationality 57 14.0 12.3 19.2 31.6 22.8 Diffuse-Hi Rationality 5!. 5.6 11.1 27.8 35.2 20.3 Table 27 - Proportion of Close Friends in the Community of Residence Among the Structural Types. __ x Proportion of 91039 Friend: Strg;;:ral Total. :11 gzggt Roggszr nest a few - - - - - per cent - - - - - - SpecificaLo Rationality 57 35.1 31.6 33.3 Specific-Hi Rationality 24 29.2 33.3 37.5 Mixed-Lo Rationality 97 40.2 22.7 37.1 Mixed-m Rationality ‘72 1.7 .2 26 .I. 26.1. Diffuse-Lo Rationality 51 39.2 23.5 37.3 sumo-31 mummy 51. 61.1 24.1 14.8 .181. Table 28 - Off-farm Employment and Proportion of Total Income This Work Provides Among the Structural Types. A S Pr0p0rtion of Total Income Structural Working 131‘ Off ~f work? Type ”“1 _ 03- 172 or 172 to More ' ram less 3/1. than 3/4 ----p;rcent~-—-- Specific-Lo Rationality 57 1.5 .6 16.0 28.0 56.0 Specific-Hi Rationality 21. 54.2 76.9 7.7 15.1. Mixed-Lo Rationality 98 1.8.0 31.9 31.9 36.2 Mixed-R1 Rationality 71 36.6 68.0 12.0 20.0 Diffuse-Lo Rationality 51 37.3 11.1 1.4.4 44.1. 48.1 58.3 29.2 12.5 Diffuse-Hi Rationality 51. C“ V g A r * Percentages are computed only on cases where income information is available. Specific-lo rationality had two cases of data missing. All the other types had one case of data missing except the specific- hi rationality and mined-lo rationality, where canplete data was available. of three of the structural types also reported higher rates of working off the farm, Table 28. All of this evidence points to choice rather than force of circumstances behind linited opportunity. Opportunity to this point has been considered primarily as physical potentiality to interact in the appropriate fashion. However, opportun- ity also can be thought of as mental or attitudinal since it is patently true that people in physical proximity need not interact. The social relations variable offers a good inference to attitude in this respect. Persons in diffuse type relations, by the very nature of then, show a acre favorable attitude toward sociable interaction than do people in specific type relations. This inclination bears favorably on using ~182- market news in a sociability sense. The distribution of this function by type of social relationship is statistically significant at beyond the .001 probability level and in the correct direction, Table 21. If the line of reasoning to this point is valid, the data exhibit certain incongruities. Inboth the diffuse relations and mixed rela- tions classes the degreeof use of market news for sociability is higher anong those exhibiting low rationality levels than among those With high rationality levels, Table 21. While it might strengthen the explanation if the reverse were true, the existence of sufficient cannon opportunities for all of the five structural types that are absent for the specific-low rationality type is the most crucial point. This is the major point at which opportunity matters, whereas differences among the other five types of structure with respect to the opportunities they offer for this use of narket news are likely to be random. Thus, differences facilitating higher use on the low rationality levels of the same relations types are not statistically significant. (h the evidence presented ,the role of differences in Opportunity resulting in differences in use continues to 10011 as a fruitful hypoth- esis to be tested with new data. The explanation will also be examined where relevant in terms of certain data of this stutiy. There is another respect in which the ability of market news to function as a conversation topic seas to vary systuatically. In the specific-low rationality class, 64.3 per cent of those who define narket news more broadly than does the USDA report use of this function whereas only 51.7 per cent whose definition conforms to that of the USDA do so. Shilnrly, within this structural type, those with USDA type definitions are slightly ncre prone to confine these conversations to persons they _ --183- now well, 14., 69.2 per cent in contrast to 61.1 per cent. Perhaps respondents with broader definitions of market news can report legiti- nately for a wider variety of things that they are using as “market news' and thus pruote or facilitate sociability since many more of the things that farm peeple find easy and interesting to discuss will fall within these definitions. This possibility is given some support by examining differences in the degree of sociability use in the remaining five structural types when the market news definition varies. Three of the five show less of this use with restricted market news definition, while only one shows a slight increase, and the other shows a thirteen per cent increase. The last, however, is based on one cell containing fewer than ten cases. It is also reinforced by comparable data concerning conversation priming, Table 24. Of those whose definitions are broader than that of the USDA, 56.1. per cent report this function. In contrast, only 1.2.9 per cent of those with definitions more congruent with that of the USDA do so. Five of the six structural types reflect this general pattern.“ The one exception again involves the snallest class containing but nine cases. Gama: mesa}. maximal M £29 mat Em The second most frequently occurring latent function in the sample was the role of market news in furnishing ideas of a general educational nature. Often this consisted of calling to awareness or reenfcrcing an understanding of the simple broad mechanics of the effects of supply- demand relationships on price. it other times, more subtle awarenesses developed. One man said that by getting market news regularly “you appreciate how much all prices are tied together. Long term trends in .184... agriculture determine ultimate price and profit. " Another figured he could make more money be going against majority marketing patterns. A paraphrase of his oo-ent runs, 'I watched wheat prices, found out the highest price comes just before the new harvest. Discounting on the farm storage as use of otherwise waste space, it makes sense to hold on to the wheat as long as not too many famers do it." Irrespective of the level of sephistication of what was learned, this function was dis- tributed among the structural types as shown in Table 29. Table 29 - mitaining General Knowledge from Continued Receipt of Market News Among the Structural Types. Jain-W Structural re SD e o as Type Total Total Nua- __ 7 Tee Tee Aber Yes Specific-Lo Rationality 24 95.8 25 40.0 49 67.3 Specific-Hi Rationality 15 86.7 9 77.8 24 83.3 Mixed-Lo Rationality 45 73.3 42 61.9 87 67.8 Mined-Hi Rationality 51 86.3 16 87.5 67 86.6 Diffuse-Lo Rationality 26 80.8 21 76.2 47 78.7 Diffuse-Hi Rationality 39 94.9 13 84.6 52 92.3 Chi-square test d.f. 12 Social relations alone 2 5.01 .1 > p 7 .05 Rationality alone 1 14.71 p < .001 Market news definition alone 1 16.20 p < .001 Relations and rationality, no control 5 20.63 p < .001 Relations and rationality, USDA definition 5 10.44 .1 7 p 7 .05 Relations and rationality, non-USDA d.fiflitifin 5 18 e0]. e01 7 p 7 .001 W -185- The nest obvious feature of this table is the difference among the high and low rationality classes. Whether the type of relations is held constant or allowed to vary the difference is statistically sipificant. Given the meaning of rationality, this pattern is certainly to be expected. The ideas reported as having emerged frm following narket news were overwhelmingly things that constitute rational behavior with respect to far: marketing. This congruence of the function with the stmcture in which it some can be extended. If it can be seemed that learning has at least a dual experience referent -- variety and number - then sale of the data regarding the conversational function are also relevant here. It nay be remembered that persons maintaining diffuse relations were nore likely to report such behavior than were persons in mixed relations, and these, in turn, were more likely to do so than were those maintaining specific relationships. This pattern was interpreted as reflecting differences in the umber of opportunities to talk with other persons interested in agriculture for when market news would be a relevant topic. It was also shown that high rationality cases tended not only to have more opportunities for such social intercourse, but also to experience more diversified contacts within their social contexts in the sense that they nore often used narket news in conversations with strangers. All data concerning this conversational function were intended to be limited to these instances where market news was used in a strict sociability sense. However, the exaaples the respondents supplied often verified the can sense idea that the use of narket news for sociability could blend with nore purposive actiaa. That is, the individual would find hinself trying to pick up information for its own value either to help ~186- in an mediate sale situation or, we would hypothesise, as possibly relevant for its educational value. In view of these findings, it is plausible to hypothesize that the most learning fran market news would take place among respondents of the diffuse-high rationality types and that the least would occur in the specific-low rationality types, for, in those types of structure both Opportunity and motivation are, respectively, at their highest and their lowest. bong the remaining types one might expect motivation to exert more influence than Opportunity would. If, in addition, it is seemed that rationality primarily exemplifies motivation toward learning and that type of social relations exwplifies the relative Opportunities for learning contexts, then the rank order for proportions using market news in the general educational sense among the remaining four types would run: high rationality-nixed relations, high rationality-specific relations, low rationality-diffuse relations, and low rationality-nixed relations. This is the order shown in the data where definition of market news is not controlled. It was also argued with respect to the conversational function that persons who define market news more broadly than does the USDA can refer to a wider range of bits of infomation which they may utilize as conversational ganbite of market news. Similarly, such persons should also have wider possibilities for learning.281 This is borne out by the data. 0f persons with definitions broader than the USDA's, 85.5 per cent gave evidence of this function in contrast to 66.7 per cent for 281. The definitions of market news tended to resuble closely the types of information actually obtained by the respondent. ~187- those persons with definitions of the USDA type. Considered individ- ually, five of the six structural types showed this relation. The sixth showed virtually no difference in this use. Moreover, the rank order predicted above on the basis of I'motivation-ropportunities" generally held up for both types of market news definitions, though one very noticeable and inexplicable inversion appeared. Among those with definitions more inclusive than the USDA's, the specific-low rationality type had the highest level of use! As was the case with the conversation function, certain questions can be raised to elaborate the structural typologies in relation to the educational function and, at the same time, underline the rationale of the analysis in this chapter. Thus, questions about differential ability to extract educational material from diversified and amorous contacts could be made. (be possible index of this ability is years of schooling coupleted. The distribution of years of schooling among the six structural types is not uniform, Table 30. Measured by chi-square, Table w - Levels of Fornal Schooling Among the Structural Types. St’fyg‘r‘l Total Grammar Somthigh High school or less school grad or more 7 -------percent------- Specific-Io Rationality 57 73.7 12.3 14.0 Specific-Hi Rationality 2!. 37.5 20.8 41.6 Mind-Lo Rationality 98 65 .3 15.3 19.6 Mixed-Bi Rationality 72 47.2 18.1 34.7 Diffuse-Lo Rationality 51 68 .6 11.8 19.6 Diffuse-Bi htimslity 54 42.6 24.1 33.3 W ~188- Table 31 - Respondents' Age Among the Structural Types. A‘ Age in gear; fl SW“ see if. 25’ 1:3 22 .33 ‘ less 44 54 64 over 1 o-----percant ------ - Specific-Lo Rationality 57 12.3 19.3 21.1 21.1 26.3 Specific-Hi Rationality 21. 8.3 25.0 16.7 37.5 12.5 Mixed-Lo Rationality 98 10.2 19.4 26.5 26.5 17.3 nixed-Hi Rationality 71 12.7 23.9 35.2 16.9 11.3 Diffuse-Lo Rationality 51 9.8 29.4 27.4 17.6 15.7 Diffuse-Hi Rationality 51. 16.7 27.8 27.8 22.2 5.6 - .— the distribution varies significantly at the .01 probability level. The high rationality types tend to have comparative overabundance of persons with higher levels of formal education. This "ability“ criterion does not appear as a result of age differences in the individuals included in the types. The distributions of respondents' ages in each structural type do not differ significantly, Table 31. 282 The typologies in this respect scan to reflect some active choice aspects on the part of the component individuals rather than simply an Certainly, also, a w W for occurrence of an educational function for market now should include imposed categorisation. A large part of the variability that is shown is due to the sharp differences in the diffuse-high rationality and specific-low rationality types for persons over 65. The former has almost five times as many persons in the age grouping. This difference does not detract, however, from the sale conclusion. 282. ~189- differential ability to learn. However, the center of attention in the present analysis is testing constraint with the dimensions of structure utilised by the study-283 rather than a full explanation of differential distributions for the particular functions. A first requirement for that hypothesis is that the data can be understood plausibly by using the definition of structure adopted. A11 correlated biases and, hence, other possible explanations, can be neither controlled nor considered. Correlated variables, such as age and education, are considered only when data with respect to them are salient and available. Has 2: Was in Elsie Mat lain Another possible function of the USDA's program of making market news available to the public through the mass media is the freeing of farmers from the restrictive condition of having to get market news for themselves by permitting than to rely on surrogates for this purpose without increasing the risks of inaccuracy. The distribution of what will be here referred to as the “surrogate” function is given in Table 32. The only apparent pattern is in the significance of the typeof social relations maintained. Seventy-one and four-tenths per cent of the diffuse relations, 63.5 per cent of the mixed relations, and 54.3 per cent of the specific relations cases use surrogates to obtain information about markets and marketing. This distribution when evalu- ated by chi-square has a probability of occurrence between .1 and .05. This is, of course, the order to be expected from the point of view of the opportunity hypothesis develOped with regard to the previous two 283. See the discussion of this point, pp. 30-31. .190- table 32 - use of Surrogates to Obtain Market news Among the Structural Types. gggitiog of Mket yea Structural re than SD or as To Type Total Total x N‘lm- x Yes Yes ber Yes Specific-Le Rationality 28 67.8 29 44.8 32 56.1 Specific-Hi Rationality 15 33.3 9 77.8 12 50.0 MixedéLo Rationality 50 64.0 48 58.3 60 61.2 Mixed-Hi Rationality 5!. 50 .o 18 61.1 1.8 66.7 Diffuse-Lo Rationality 28 64.3 23 69.6 34 66.7 Diffuse-Hi Rationality* 40 75.0 _ 14 78.6 41 75.9 Chi—square test d.f. 12 Social relations alone 2 5.7L .1 > p 7'.05 Rationality alone 1 1.45 .3 > p >’.2 Market news definition alone 1 .77 .5 7 p 7'.3 Relations and rationality, no control 5 7.59 .2 7 p 7'e1 Relations and rationality, USDA admition 5 8e73 e2 7 p 7 e1 Relations and rationality, non-USDA definition 5 8.80 e2 > p 7 e1 ii .191- functions. In diffuse relations the scope of ego's interest in alter is both broad and personal enough to make specific designation for obtaining market news a legitimate expectation. The same is not true for specific relations except under unusual circumstances. Even if there is no specific designation of a surrogate,28” a type of surro- gating "by accident" could obtain more readily with diffuse relations. It was shown that persons in diffuse relations tend to discuss market news more than do those in specific relations. In any such instance, where market news information was used not solely in its sociability sense, surrogating ”by accident” would result if and when the individual , did not himself receive the relevant information but got it in the con- versational exchange. While the hypothesis of Opportunity is consistent with the data on the use of sm'rogates, a question as to the need for such persons can be raised. Some data for assessing possible need for surrogates are available. Examination of the proportions of respondents of each structural type working off the farm indicates that the specific rela- tions types and the mixed-low rationality type have then of the four highest rates of off-farm uployment (See Table 28). All other factors being equal, famers uployed off the farm, should have probably a greater need for surrogates since they would be generally gone from the farm during the tines of the day when market news infomtion is nonally available. let, the three structural types which should have the greatest need for mrogates show the lowest use rates. 284. See the discussion of scoring for this faction, p. 113. -192- This evidence, however, overlooks a fundamental distincticn‘between objective need and subjectively';gltlneed. It has been accepted as axiomatic that action is mediated in tonne of the gglf,perggived situap tion. Reed has to be defined from the actor's reference. Some data are available on this point. Respondents were asked if it would.make any difference to them if they could no longer get market news information. The distribution of answers among the structural types is shown in Table 33. It can be seen that the specific relations types and the mixed low rationality groupings which have significantly less use of surrogates also have lower felt need for them. Table 33<- Feeling of Deprivation with Less of Market News and Why . SurrOgates Are Not Used Among the Structural Types. Per cent Per cent Structural Who would Who Can Get Type Total Net Miss Total Enough news Market News ‘ By Themselves Specific-Lo Rationality 57 28.1 25 52.0 Specific-Hi Rationality 24 16.7 12 83.3 Mixed-Lo Rationality 98 26.5 38 68.4 Mixed-Bi Rationality' 72 9.7 24 79.2 Diffuse-Lo Rationality 51 15.7 17 64.7 W The factor of preference in defining need is supported inferen- tially through other data. It was shown above that there are no signi- ficant differences on the distribution of number of friends in the community of residence among the structural types and no significant differences on age distributions, Tables 27 and 31. Similarly, there --193- Table 3A - Available Family Surrogates Among the Structural Types. J —-:..___ -::__. j_ Relative Availability of String 8.]. Total Surro to Best‘ Intermediate“ Worst“. Specific-Lo Rationality 56 8.9 57 .1. 27.8 Specific-Hi Rationality 21. 12.5 58.8 27 .4 Hind-Lo Rationality 97 16.5 65.3 22.2 nixed-Hi Rationality 72 12.5 58.8 24.7 Diffuse-Lo Rationality 51 13.7 66.7 20.8 Diffuse-Hi Rationality 51. 14.8 66.7 30.4 W * Consists of unmarried persons living at home or with others present and married persons with grown (i.g., over 18 years old) children at home. ** Consists of married persons with wife present only or with grow ing children (1.9. , all between 5 and 18) present along with mCe "5* Consists of persons who are widowed and living alone, unmarried living alone or married with all children under 5 years of age. are virtually no differences among the structural types when these are compared on the basis of household composition ranked by the degree of probably opportunity for' obtaining surrogates from the family, Table 31.. The explanation of need is also relevant to the non-significant distribution of this function between the two levsls of the ration- ality variable. One might expect that higher rationality levels would be associated with higher rates of use of surrogates, given the motiva- tion connotations ascribed to the variable. Although the distribution was in the correct direction for this proposition, the probability of ~194- ocourrence was only between .3 and .2, Table 32. However, it could be argued that those persons with higher rationality levels would also be more willing and likely to get their own intonation and/or use surro- gates. This interpretation is supported by the data. There is virtu- ally no difference between high and low rationality levels on the proportion of members working off the farm (43.6 per cent versus 44.7 per cent). But, 63.1 per cent of the high rationality cases receive less than half their total incane from that work in contrast to 22.8 per cent of the low rationality cases. Objective need here expressed is matched by subjectively felt need. 01' the low rationality cases, 24.3 per cent would not miss market news if it were unavailable but only 12.0 per cent of the high rationality cases answer this way. At the same time, the high rationality cases were more likely to get the market news they needed themselves if they could not assign a surrogate. Among the high rationality cases, 81.6 per cent of those who did not use surrogates indicated they got all the market news they needed then-- selves whereas only 62.5 per cent of the low rationality cases indicated this. With respect to the role of market news definition in the occur- rence of this function, it could be argued that definitions more inclusive than the USDA's create greater needs for surrogates since more types of intonation are required and, other things equal, the more information the individual feels he needs the less likely is he to be able to get it all by himself. However, as was shown in the 'priming" function, those who define market news more broadly are also more willing to go out of their way to have contacts and presullabhtr obtain necessary information. The data, Table 32, are inconclusive on --195-- this matter but tend in the direction of the first point. Without controls on relations and rationality the difference is slightly in the direction of greater surrogate use by persons with non-USDA type definitions. With controls on relations and rationality three of the six comparisons involving differences in definition show higher use of surrogates in the groupings with more inclusive definitions, two show slightly less use, and one shows markedly less use in this class. The tendency of those persons with high rationality levels to be more self reliant for market news would seem to apply also to the joint effects of relations and rationality. Table 32 shows that the same probability of occurrence for the distribution of the function among the structural types exists with or without the market news definition control. In all three, the underrepresentation of the function in the specific relations classes and the overrepresentation in the diffuse- high rationality class accounts for the largest part of the variability. (n the opportunity hypothesis as it has been extended above, this is to be expected. However, none of the distributions are statistically significant. mummunmi magnum; Differential esteaa exists in every group. Hanans in summarising a number of empirical stmies has noted, "the more nearly a member in his activity realises the nouns and values of the group, the higher [is] his rank [1.9, sateen] .- 235 The centrality of material well being in hericsn culture and the relation of market news to efficient role 285. Humans and Riechen, pp. 913., p. 789. 4%- performance for the farmer in the economic sphere would imply that persons with high rationality levels should have greater use for this function. This assumes that high rationality denotes greater concern with the efficient managment aspects of farming. The possible relationship of market news definition or social relations to this function are not so sharply specifiable. Since differential esteem applies in all types of social relationships and since the function was one simply of use and not relative importance of the criterion, which would tend to dispose higher use in specific rela- tions, no expectation was held for the effect of social relations on the function. Similarly, no expectation was held for the market news definition variable. Differences here might effect esteem Judgments in situations where a person with a broader than the USDA type definition might interact with a person defining market news less inclusively. Because of that definition, the latter person could appear as deficient in knowledge to the other and receive, thereby, less esteem. This would be particularly evidenced in structural types with high rational- ity levels. The necessary information to evaluate this possibility was not part of the schedule. From Table 35 it is clear that use of knowledge about markets and marketing as a criterion in esteem Judgments of others is evenly dis- tributed through all of the structural types . The chi-square test applied to the six structural types , without regard to market news definition, yields a value with a probability between .8 and .7. Repetition of the tests for each definitional class taken separately makes no great difference. lior are there any significant differences when the structural types are grouped so as to reveal differences with .197.- Table 35 - Use of thrket News as a Criterion for Esteem Evaluations Among the Structural Types. .___._-. I Definition of Market New; Structural e t USD e or less Type Total $ Total % Num- . Yes Tee ber Tes Specific-Le Rationality 24 54.2 26 30.8 50 42.0 Specific-Hi Rationality 14 57.1 9 1.4.4 23 52.2 Mixed-Lo Rationality 49 1.4.9 45 1.8.9 9!. 1.6.8 Mixed-Hi Rationality 53 50.9 20 55.0 71 53.5 Diffuse—Lo Rationality 27 55.6 22 54.0 1.9 55.1 Diffuse-Hi Rationality 40 52.5 14 57.1 54 53.7 Chi-square test d.f. 12 Social relations alone 2 1.44 .5 > p > .3 Rationality alone 1 1.10 .3 > p > .2 Market news definition alone 1 .24 .7 7 p 7 .5 Relatims and rationality, no control 5 2.86 .8 > p 7 .7 Relations and rationality, USDA. definition 5 1029 095 7 P ’ e9 Relations and rationality, non-USDA definition 5 5.23 .5 7 p 7 .3 W respect to either type of social relations or rationality. The proba- bility of occurrence of the chi-square value for the latter is between .3 and .2. However, five of the six comparisons in which only ration- ality varies are in accord with expectation. is an equally likely chance for differences to go in either direction, If we seems that there the probability of five in six going in the same direction is .09. This would contribute sue evidence for the expectation about rationality. .193. m [m 5; 3 m in, Evaluating USDA Performance The perceived quality of market news can evoke evaluative responses from farmers and, in the event that the USDA is known to be the origin of the service, these evaluations are directly transferable in sane degree to the USDA as a whole. Thus, the existence of the market news system may function to provide acme farmers with a basis for evaluating the USDA. Formulation of evaluations of the USDA through market news involves learning, inasmuch as respondents had to be aware that the USDA was involved in the chain of information dissemination in order to be eligible to have this function attributed to then. This awareness almost necessarily requires learning because the USDA does not greatly publicise its role in market news dissanination. By analogy with the reasoning presented previously with respect to the general educational function, high rationality, the maintenance of diffuse relationships, and the holding of broad definitions of market news would all be most facilitative for this function. The distribution of this function among the structural types is given in Table 36. Respondents with high rationality levels show 55.9 per cent use in this way in contrast to 27.4 per cent use by those with low rationality levels . Similarly, the proportion of respondents indi- cating occurrence of this function when they are classified according to the type of social relationships maintained ranges frus 51.9 per cent for diffuse relations through 38.2 per cent for mixed relations to 24.7 per cent for specific type relations. The extent of use snong those persons with broader than the USDA definitions is 47 .7 per cent in contrast to 27.5 per cent for those with USDA type definitions. All of these differences are highly significant statistically ~199- Table 36 - Use of lurket News as a Factor in Evaluating USDA Performance Among the Structural Types . ..._.— —-_ ‘4'... D tion of ket N Structural r t D e r is Type Total 1 Total Hum- Yes Yes ber Ies Mixed-Io Rationality 49 32.7 46 17.4 95 25.3 Mixed-Hi Rationality 56 62.5 17 41.2 70 55.7 Diffuse—Lo Rationality 28 50.0 22 31.8 50 42.0 Diffuse-Hi Rationality 40 60.0 14 64.3 54 61.1 Chi-square test d.f. 12 Social relations alone 2 13.50 .01 > p > .001 Rationality alone 1 28.39 p < .001 Mket news definition alone 1 13.79 p < .001 Relatims and rationality, no control 5 36.07 p < .001 Relations and rationality, USDA definition 5 18.98 .01 > p > .001 Relations and rationality, non-USDA definition 5 16.89 .01 7 p 7 .001 interpreted by chi-square . The ordering of the six structural types can also be predicted for this function 1: the rationale developed for the relative importance of the relations and rationality variables in respect to the general educa- tion mu“ 1. uc.ptod.286 The rank order of the structural types should be : diffuse-high rationality, mixed-high rationality, specific- 286. See pages 185-.186 this material. .200... high rationality, diffuse-low rationality, mixed-low rationality, and specific-low rationality. This order is shown generally both with the market news definition controlled and uncontrolled. One small break in the rank order occurs when market news is uncontrolled and when control is added in the cases with definitions at variance with the USDA. This, as well as the previous tests on the evaluative function support the preposition of structural constraint. The results obtained also do not appear to be attributable to differences in the potentiality for this function but in certain respects appear in the face of adverse possibilities. The USDA indi- cates itself as the source of market news releases made to the mass media. To the extent that a respondent uses only the mass media for market news he would be expected to have no less and possibly more con- centrated exposure to the fact of the USDA's role than those who use other than only mass media sources. Yet as shown in Table 37, the Table 37 - Sources of Market Hews Among the Structural Types. W Sourcg Structural Total Mass media Mass media and informal W x only or informal only Specific-Lo Rationality 52 50.0 50.0 Specific-Hi Rationality 21 38.1 62.9 Mixed-Lo Rationality 95 48 .4 51.6 Mixed-Hi Rationality 70 47.1 53.0 Diffuse-Lo Rationality so 50.0 50.0 Diffuse-El Rationality 54 27.8 72.2 .201- distribution of market news sources among the structural types shows that the high rationality cases tend not to concentrate their market news sources with the mass media only and therefore have possibly less exposure to the USDA's role. Affect ngard Aggicultge 9291 Market News Ehotional affect toward farming was an aspect of structure on which it was seemed that the effects of market news would be directly felt. There are major uncertainties inherent in agriculture with regard to income flow, and these are crystallized in the prices that the farm operator receives for his product. Information about current prices is the heart of the USDA program and is included in nearly all respondents' definitions of market news. While it seems reasonable to expect that concern with price uncertainty is differentially distri- buted in the sample, it also seems reasonable to expect that market information functions implement affective feelings for the occupation more for those persons who are most concerned with uncertainty than for those persons who are less so. News about an area of interest influences the attitudes of more of those persons interested in the field than for those not interested in the area. Previously it has been argued that rationality indicates the mode of a respondent's adjustment to his circumstances. Whether a high rationality level is associated with an outlook of satisfaction with agriculture or dissatisfaction and is simply a way of making the best of a bad situation is irrelevant to our purpose. Exhibiting behavior which normally promotes maximm monetary incmne is direct evidence of -202. a concern withprice.287 Moreover, the direction.of the affect pro- moted by'market information is not important. What is important is that these individuals probably differ from.those with low levels of rationality. The behavior of the latter suggests less interest in price and, relatedly, monetagz income from farming. Whether such lower interest is a result of choice or ignorance is also unimportant. In either case, market news should play a less significant role in determining the individual's feelings toward agriculture. The bearing of social relations on this function is less clear. It could be argued that, without regard for any other conditions, the wider latitude for expression of affect offered by diffuse type rela- tions would induce an association between this type of social relation- ship and a higher rate of occurrence of this function than for specific type relations with their restricted emotive contexts. The assumption behind this preposition is that affect is contagious. Here a person- ality variable differentiating what Riesman has called the "inner- directed" and ”ether-directed" personality types might be sigificant.288 Affect contagion would be particularly true for the other-directed personality. Riesman has implied that the other- directed character is the current modal type. The inner-directed type, whatever preference for type of social.relations, could draw as 287. Further evidence for this contention.cames from.chapter four. While rationality did not differentiate manifest functions at statistically significant levels, under conventional under- standings, all:mean differences in the.five factorials were in the direction of greater use by persons with higher levels of rationality; 288. See David flies-an, lathsn Glaser and Renal Denney, Ihg,fggsly 9223;, low Haven, Connecticut: Tale university Press, . ~203- much affect from specific relations as from diffuse. The inner- directed character is the typical characterization of the farmer. These cross-checking considerations render an expectation difficult. We do know from previous analysis that persons in diffuse relations both discuss market news more and have wider contacts in their dis- cussions. Similarly, they use market news more both for sociability and ”priming" purposes more often. On this basis we can resolve the expectation in favor of greater use for diffuse relation cases. It would seem reasonable to expect, if the above be true, that the type of social relations maintained would be particularly signifi- cant for high-rationality cases, but less significant for low- rationality cases. Since prices are of less concern to the latter type of people, the possibility of more opportunities for discussion should have no great effect. Similarly, for market news definition, the broader the definition the more should market news influence occupational feelings. Again, this influence should be particularly clear in the high rationality cases, but not specially important amen rationality is low. Table 38 contains the relevant data for evaluating these preposi- tions. The distributions of occurrence among the social relations classes, the rationality levels, and the types of market news defini- tions each taken singly with others left uncontrolled are all statis- tically significant when evaluated by chi-square. The rates of mun-mes for this function are, respectively: diffuse relations, 39.4 per cent; mixed relations, 31.8 per cent; and specific relations, 22.2 per cent; high rationality levels, 1.1.6 per cent; low rationality .204. Table 38 - Use of Market News in Effecting Feelings Toward Agriculture Among the Structural Types. M De inition of ket Ne Structural or t U D one or e 0 Type Total $ Total 5 Num- S Yes Yes ber Yes Specific-Lo Rationality 28 32.1 29 6.9 57 19.3 Specific-Hi Rationality 15 33.3 9 22.2 24 41.7 Mixed-Lo Rationality 50 32.0 1.8 1.4.6 98 23.5 Mixed-Hi Rationality 54 46.3 18 33.3 72 43 .0 Diffuse-L0 Rationality 28 35.7 23 43.5 51 39.2 Diffuse-Hi Rationality 39 33.3 14 57.1 53 39.6 Chi-square test d.f. 12 Social relations alone 2 6.22 .05 ’ p 7 .02 Rationality alone 1 7e17 002 7 p 7 e01 thrket news definition alone 1 5.31 .05 7 p 7 .02 Relations and rationality, no control 5 14.28 .02 7 p 7 .01 Relations and rationality, USDA definition 5 3.12 .7 > p 7 .5 Rolations and rationality, non-USDA definition 5 20.40 .01 > p 7 .001 m 5 levels, 26.6 per cent: and, definitions of market news more inclusive than that of the USDA, 36.4 per cent; definitions of the USDA type 24.8 per cent. The expected pattern of decreasing occurrence through the social relations types of the high rationality cases is clearly apparent in the group whose definitions of market news approximate that of the USDA, the percentages being 57.1 for diffuse relations, 33.3 for mixed, and 22.2 for specific relations. Surprisingly, the .205- same decreasing pattern holds for the low rationality cases. lo comparable effects can be discerned in the grouping whose definitions are broader than that of the USDA. The uniform occurrence of the function in the low rationality cases is the enacted one. Just why the decreasing pattern fails to exist for the high rationality cases is an interesting but inexplicable problem. One possible lead might be the content of the definitions of market news most at vari- ance with that of the USDA. Respondents in this grouping and who claimed that their feelings about agriculture were being influenced by the market news they received, could have been selectively referring to that information which is not directly or primarily price oriented. This would tend to be the case particularly among the low rationality cases given their lower price orientation. Concepts of market news which included more than price would tend to obscure the effects of rationality. In contrast, this is less likely to happen with respond- ents with more limited, USDAI-like definitions, since these must involve price data and not much more. Those persons whose definitions were broader than that of the USDA had either none or only one “type“ of information falling outside the USDA program. Although this explanation would account for the failure of rationality to discriminate in the groupings with broader than the USDA type market news definitions and, parallelly, would account for the failure of the enacted pattern of social relations in the high rationality cases, it does not resolve, the failure of social rela- tions by itself to be differentially associated with the function for this type definition cases. It must stand as non-confirming evidsnu for the 'opportunity" hypothesis and, in turn, the preposition of structural constraint. -206- use: has .2 . some sans mans-l The two mental health functions will be considered together, even though they are uncorrelated, since the arguments that apply to both are identical. Both functions involve complex psychological mechanisms of personal adjustment to social stimuli. To set out the relation of the social situation to these is extremely difficult, partly because of the welter of ambiguous, often conflicting data in the area289 and, on the other hand, because of the dearth of relevant material in the study. Quite contradictory arguments and propositions of cross- checking effect tendencies could be raised. For example, it has been argued that high rationality indicates a greater concern with farm prices than does low rationality. It was also suggested, in purposing the possibility of mental health functions for market news, that the inherent high uncertainty of agriculture is focused in the prices for farm products. If a fairly high personal anxiety level about the marketing situation exists, one might expect it to be concentrated among those who are more interested in price, i.g., the high rationality cases. At the same time, it is reasonable to argue that persons with higher rationality levels also have more non-emotive and logical resources to cope with their anxiety and, hence, that they have less need to rely on these non-rational mechanisms. In similar fashion a case for contradictory tendencies for low rationality cases could be made which, in brief, would argue less resources but also less need. The dearth of relevant data in the study focuses primarily on the 289' 2!: WM M- Rm. sebum mam mam 121mm: a W m. n... tom Norton. 1955. -207- need for a measure of basic personality structure.290 It seems entirely clear that the incidence of any mental health mechanism.would involve an interactive equation based on: (1) the nature of the social situa- tion and (2) the basic personality of the participants. Of course, the latter has been at best tentatively inferred to the present without as serious difficulty as is involved for the mental health functions. In the present context, however, to suppose that high rationality cases have greater anxiety would not allow one to Specify the simple fact of whether release from anxiety (should it come) would be in succoring or scapegoating. It seemed best, therefore, not to raise propositions about the possible distribution of these two functions but, rather, to simply note the distributions and obtain what insights are possible. Tables 39 and 40«contain the distributions of the two mental health functions among the structural types. It can be seen that the type of social relations is significantly associated with the succor function, whether the market news definition is controlled or uncontrolled. At the same time, neither rationality nor market news definition classes are significantly related to the function, with or without the other two variables controlled. Rationality and social relations considered without market news definition controlled was highly statistically significant. This was also true for the caseSivith market news defini- tion at variance with the USDA.291 In these last two comparisons the 290. The discussion, pages 202-203, is also relevant here. 291. No statistical test was run for cases with.USDA-like market news definitions because the collapsing of cells required to meet the test criteria destroyed substantive meaning. This was also true for both tests of relations and rationality for the scapegoat function with.market news definition controlled. -208- Table 39 - Use of Market News as Succor for Mistakes Among the Structural Types. Definition of ‘ Market News Structural re t USDA e or les 0 Types Total Total 1 Num- Ies Yes ber Tea Specific-Lo Rationality 28 0.0 25 4.0 53 1.9 Specific-Hi Rationality 11. 0.0 9 22.0 23 8.7 Mixed-Lo Rationality 49 18.1. 44 9.1 93 14.0 Mixed-Hi Rationality 48 14.6 16 6.2 61. 12.5 Diffuse-Lo Rationality 26 23.1 22 22.7 1.8 22.9 Diffuse-Hi Rationality 39 33.3 M 21.4 53 30.2 Chi-square test d.f. 12 Social relations alone 2 18.28 p < .001 Rationality alone 1 2.01 .2 7 p 7 .l )hrket news definition alone 1 1.1.1 .3 7 p 7 .2 Relations and rationality, no control 5 20.70 p 4 .001 Relations and rationality, non-USDA definition I. 14.62 .01 > p 7 .001 ~209- Table 40 - use of Market News as a Scapegoat for Mistakes Among the Structural Types. Definition of Market News Structural "33‘ESEE‘EEB“""‘S‘3§‘$3§E 0 Type Total S Total % Num- , Yes Tes ber Ies Specific-Lo Rationality 27 0.0 27 3.7 54 1.9 Specific-Hi Rationality 15 0.0 9 22.2 24 8.3 Mixed-Lo Rationality 50 8.0 29 8.2 99 8.1 Mixed-Hi Rationality 53 9.4 17 5.9 70 8.6 Diffuse-LO Rationality 27 11.1 23 4.3 50 8.0 Diffuse-Hi Rationality' 40 7.5 14 7.1 54 7.4 Chi-square test d.f. 1? Social relations alone 2 1.71 .5 P p >'.3 Rationality alone 1 .51 .5 > p 7'e3 Market news definition alone 1 .001 .98 7 p > .95 Relations and rationality, no control 3 2.66 .5 7 p 7 .3 largest part of the variability was attributable to overrepresentation of the function in the diffuse-high rationality type and underrepre- sentation in the specific-low rationality class. Whatever else may be involved in the dynamics of this mechanism, it would appear that diffuse relations facilitate this function. The scapegoating nechanism.is distinguished by its extremely low incidence in the total sample as well as its rather even distribution among the six structural types, particularly without the definition of market news controlled. none of the distributions among any of the structural variables is statistically significant. -210... Non-solicited gtent Functions ad Eket News The final contribution to the latent function score was derived from an Open ended question which sought to discover if missing market news information would make a difference to the respondent. Affirma- tive answers were probed to ascertain why it would be missed and these responses were categorised as indicating manifest and/or latent functions being served by market news in that case. The crucial point of this function is saliency of latent functions. However, respondents were not made aware of the manifest-latent distinction along which their answers were categorized. As a result of this non-awareness, the most reasonable expectation of pattern would be one of equal dis- tribution among the structural types since chance alone should be involved in determining the occurrence of the function. Similarly, since none of the structural dimensions implies latency as such, chance alone should determine the distributions by the various levels of these variables. Table 41 shows the relevant distributions. Irrespective of the variable examined, an extremely uniform distribu- tion exists with none deviating from a chance pattern. These results, while bearing out the expectation, also testify favorable for the methodological procedures of the study in an interesting manner. A number of the questions for obtaining infome- tien on the functions on their face might appear to have an articulate or cognitive bias. However, the distribution of answers for the saliency function closely matches the distribution of education in the sample. Fifty-eight and (me-tenth per cent of the sample members had no more than a grenar school education, 16.7 per cent had sue high school training, and 25.2 per cent were at least high school graduates. .211- Table 41 - Hon-solicited Latent Functions of thrket News Among the Structural Variables . Definition of mket News Structural More USD Same or ess 0 Type Total 1 Total 3 Num- % Ice Ice ber Ies Specific-Lo Rationality 28 10.7 29 13.8 57 12.3 Specific-Hi Rationality 15 6.7 9 22.2 24 12.5 Mixed-Lo Rationality 50 18.0 1.8 8.3 98 13.3 Mixed-Hi Rationality 54 14.8 18 22.2 72 16.7 Diffuse-Lo Rationality 28 14.3 23 13.0 51 13.7 Diffuse-Hi Rationality 40 12.5 14 14.3 54 13.0 Chi-square test d.f. 12 Social relations alone 2 .29 .9 > p ‘7 .8 Rationality alone 1 .16 .7 7 p 7 .5 Market news definition alone 1 .02 .9 7 p 7 .8 Relations and rationality, no control 5 .69 .99 7 p 7 .98 0f sample nembers who indicated latent functions for market news in responding to this question, 66.7 per cent had no more than gramar school, 12.5 per cent had sme high school training, and 20.8 per cent were at least high school graduates. There is, thus, a slight tendency for persons indicating latent functions for market news in spontaneous exchange to have lower education levels than those who suggest only manifest functions. If there was a cognitive bias in the schedule, it might be expected that persons indicating latent functions for market news would have more formal education. mar mummmw tur Writ This chapter has attempted to evaluate the evidence relevant to the hypothesis of structural constraint, though this evaluation could not constitute a formal test. As a theoretical guide in analysis, the hypothesis of structural constraint was expressed as an "opportunityb motivationfi mechanism which, in turn, as it was deve10ped and extended suggested some relations between the limited structural dimensions used and the specific behavior construed as possible latent functions for'market news. Although this explanation could not be extended equally well to all the specific functions and although necessary data was not always available to adequately investigate relationships, it can be seen from Table 42, which summarizes the various relevant tests, that there is considerable evidence supporting the opportunity- netivation proposition. 0f the 54 possible relationships set out by the table, 23 could not be interpreted meaningfully by the data avail- able.292 Twentyhfive, or slightly over fourbfifths, of the remaining tests supported the explanatory proposition. The remaining six rela- tionships were inconclusive, i.g,, certain evidence for and against the "expected" distribution patterns was found which denied clear evaluation. Thus, the opportunityhmotivation interpretation with reasonable plausibility related the structural types to a rather wide and disparate range of behavior. Inasmuch as the data at least hang together on the skeleton of this proposition, there is, than, some evidence for the validity of structural constraint. Definitive tests 292. These are designated as H.A. or ”not applicable" in Table 42. -213- Ascended n.2— uné. I at ebaasaosoocu u oaH a mm m .2 4 .2 m m .mfifioasu. 333. no heal»: o 4.2 4.2 4.2 4.2 4.2 4.2 2833 o 4.2 4.2 4.2 4.2 4.2 4.2 «82.2.8 0 .<.z 32.2 .4.z .<.z 4.2 3.4.2 .3095 m 3H 0.30295 #20295 #20296 9.3293... 0.322225 mad-rash V3300 noon: 0 tong V3295 0.3.29.6 Ema 0.32326 9.32926 «am: no . 83315 o 4.2 4.2 44.2 4.2 23H 4.2 .83.“ n 278225 0.32% 0.322% 3H 0.3295 #2235 aeoeuosm e £8226 4.322% 2.322% £2826 £0225 «822.5 838.62 n 8H a can .3253 «.3225 tong 333.358 .3282 2032528 5325.282 2958» on 83 :25. 428.28 4222 521838 .253. bad. 28338 82 822 33a 223.3833 220338392 one when 19035 38m swung no es. aaoepflsmfl Ea 2833.2 33312 2.32.32 » glen. so. . 5. a 2. 1.22.122. ....»iIQfism.-. . 283822 p.233 3.. e5 833.52 1528.5 23 no 22363302 .5. a passage 2303252 .82 «20232 on» no 2.3.5.2 .. 3. .35 -214. must await the further research, which is the appropriate next step from post-rectum interpretations. Furthermore, such research vitally needs a strong time dimension to assess whether any specific latent function is contributing to the maintenance or’change of the structure constituted by the relevant variables. In the current study the implicit assumption has been that the functions are consonant with past structure and, hence, contribute solely to maintenance. This assumption needs checking and elaboration. - 215 2 Chapter VI Summary and Conclusions This study has invoked a limited type of functional analysis to explore substantively and explain certain uses of market news informa- tion(by;a sample of 356 Michigan farmers who get such information. This analytical model was chosen because it seemed capable of compen- sating for a number of deficiencies in previous research. There are few substantive studies of market news by rural sociologists. Those that have been done characteristically have been confined to eliciting empirical generalizations. Little or no attempt has been.made to interpret the results obtained in any single study; to codify the data from different studies in any theoretic system or even to utilize analytical concepts which can be readily related to current theory. Furthermore, what investigations have been made into market news have confined attention to only the manifest consequences of the formal USDA programs of market news dissemination. All of these features offend the commonly assumed relations between theory and research. l'i Studies in "effects" of mass communication by which this study might also have been guided were also found to be inadequate for this purpose. Mass communication studies tend to injudiciously lump commu- nication and effects of communication as a single phenomenon with attendant analytical (and practical) difficulties. 293 In addition, most such research has been "practically" oriented with.a "problem' alleviating motivation. 0n the one hand, such research problems tend 2931 See—the discussion pp. 11-16. -216- to be posed in essentially unanswerable foms —— "what type of appeal is more effective" studies - and on the other hand, the definition of the major problem as that of studying the effectiveness of a given communication system, structure or situation results in a concern with only manifest consequences. When the effects of mass communication are studied empirically, interest has been limited to manifest function. When latent functions are focal, empirical inquiry has been ignored. It was argued that studies of effects had to be separated from studies of communication. In this way the limits of a middle range communication theory and its congruence with models explicating effects of communication could be expedited as well as forcing interest in the particularly sociologically relevant phenomena of latent functions. Furthermore, while interpretation of data is vitally necessary, the framework for that analysis should be delimited rather than global so that research can.be posed in answerable terms from research. Functional analysis rectifies the above shortcomings. Its central tenet is simply that of ”interpreting data by establishing their conse- quences for larger structures in which they are implicated." A supplementary requirement is that in assessing consequences a concep- tual distinction be made between the cases in which subjective aim-in- view coincides with the objective consequence and those in which they diverge. In these circumstances, bgth latent and manifest consequences will be investigated. It was argued that functional analysis most fruitfully is understood as a general analytical.model applicable throughout science rather than in its substantive and historical ties to certain dubious assumptions in sociology. -217.. The item to which function was imputed was the fact of a given type of communication, namely, market news information received by farmers. The structure in which the item Operated was defined at the individual status level and originally was five dimensional, including: involvement, the social psychological preference for and identification with agriculture as a means of employment; commitment, the degree to which non-preference factors of economic cost tie the individual to agriculture as an income source; characteristic relations with farm neighbors and with one's dealer for the major product sold, both separately measured on a diffuse-specificity axis; rationality, the degree to which supplementary agricultural information sources are used and managerial practices followed which would normally tend to maximize farm income. The potential functions of’market news within this structure were separated as latent or manifest on the basis of the intent of the USDA program. The fact that the research was financed by the USDA for edification of its program.prompted this definition. The manifest functions investigated were: 1. To guide economic decisions about such aspects of marketing as when, where, how'much, in what foam and what price to expect and/or charge. 2. To serve as a criterion for making'changes in production plans. 3. To provide a partial basis for evaluating dealer honesty and the efficiency of local marketing arrangements. The latent functions included: 1. To act as a resource in interpersonal relations either as a convenient conversational topic in a general sense or as a :neans for priming lagging conversation. 2. To influence affect toward agriculture. -218- 3. To be a source of new ideas or concepts -— to provide a general educational mechanism. 4. To serve a mental health function by offering a convenient and simple target for hostility derived from dissatisfaction in other areas - scapegoating - or to offer "succor" in the knowledge that expert information is not always correct. 5. To provide a criterion (among others) for according esteem Judgments. 6. To permit surrogates to perform the task of obtaining data necessary to farm management. 7. To partially evaluate the USDA's performance of service to farmers. Controls on the farmers' definition of market news and the sources of that information were deemed necessary. In this way communication and effects could be analytically distinguished and the empirical sig- nificance of the former on effects neutralized.294 This distinction was also necessary to maintain.similarity on the item to which function is imputed. The organizing ”hypothesis" for the study was a general extension of the idea that the range of variation in the items which.can serve designated functions in a social structure is not unlimited. There is a structural constraint. The particular function of a given item is the result of the structural context in which it is entered. It was recognized as a pg§3,fagtgm interpretive schema and, hence, definitive evidence for it and evaluation of it could not be made with the data in hand. At the same time, however, this orientation focused attention on interpretations based on the interrelations of data. The dimensions of ....— 294. Separating communication and effects of communication does not mean communication is an unimportant variable determining any given effect. The two phenomena are highly related. -219- structure defined were meant as systemic elements of low enough inter- relations to maximize independent effects (and hence boost prediction) and high enough to justify the assumption of systemic determination implied in the functional hypothesis of constraint. Certain general problems of analysis were involved in bringing the data to bear on this guiding hypothesis. In particular certain deci- sions had to be made in light of the controversy regarding the place of tests of significance in survey research. It was precipitated in this fashion. The study eschewed the substantive historical referent (i,g., society) for assigning the larger structure to which the item for functional analysis is implicated. Because of the survey methodology, the time dimension to utilize Merton's conception of net balance of effects for more delimited functional analysis was not possible. At the same time, structure was not defined in terms of a boundary- maintaining social system. It seemed vital, therefore, to justify the "structure" defined. How then justify? Given acquiescence to sociology as science, empirical methods were deemed necessary. Objective criteria for accepting or rejecting evi- dence then are required. Tests of significance afford such a principle. Hannah Selvin, speaking for those who would abandon all such tests in survey research, argues that all correlated biases are not removable under these research conditions. Therefore, tests of significance are ridiculous and useless because they identify only sampling variability as IDS source of differences in obtained results not produced by the experimental variable. until all correlated biases can.be removed, tests of significance have no place in survey research. In contrast, -220- it was argued, with McGuinnis, that such tests are legitimate to survey research so long as it is recognized that a result from them does not mean a definitiVe causal relation has been established. All correlated biases have not and cannot be removed but tests of signifi- cance do provide objective criteria for Type II hypotheses, the current best approximation to full causal understanding.295 At the same time, Selvin's emphasis on the obvious interrelations of social phenomena could not be overlodked even though it often is when simple correlation studies utilize tests of significance. There is some need to control for intercorrelations. However, the procedures for statistically controlling for correlated biases are often based upon assumptions not generally met by sociological data. Extensive cross-tabulating with appropriately met techniques is a procedure alien enough to be conveniently forgotten in most studies.296 In order to achieve some control over these intercorrelations without making grossly dubious assumptions about the nature of the data, a factorial analysis of variance model (after Keyfitz) was chosen as the judicious model for analysis. It allows for an assessment of the differentiating ability of the structural dimensions while it permits controlling a maximum number of variables under limited sample size.297 This assess- 295. See the discussion pp. 47—53 for a fuller review of this highly hmportant question. 296. See the comprehensive summary of rural sociological research in Edmmdde s.Br\mner.ThsMhs.f.a§sims= AW QMWWQQMWNWYWH Harpers, 1957, sap. p. 148. 297. See the discussion pp. 53-58 and 125-135. .221- ment is a pre-requisite for defining the structure to which the functional hypothesis of constraint would be assessed. The five structural dimensions were to be operationalized by Guttman scaling techniques in order to minimise the cutting point problem since the factorial design required dichotomies for all inde~ pendent variables.298 The variables of social relations and rational- ity produced such scales and were shown as valid both substantively and with the more carefully worked out conventional.methodological standards. Commitment and involvement did not.meet the criteria of acceptable scales and, therefore, were Operationalized by arbitrary indexes.299 Information for all of these scales, as well as for the occurrence of the functions, was elicited through lengthy personal interviews con- ducted with schedules by trained interviewers using directed probes. Interviewing was conducted during the summer of 1957. The sample, selected through a cluster procedure, was meant to be representative of lower peninsula.Nflchigan agriculture. It was shown that this procedure on the basis of functions scores obtained greater homogeneity in the sampling units than expected. This fact did not bear on the present study in ways that could not be corrected but it does raise cautions for error estimates on the occurrence of structure and of functions. The number of functions scored as being served for any respond- ent,300 separated by the manifest-latent distinction, and then summed 298. See the discussion pp. 77-79 on the prdblems of cutting points in sociology. 299s 588 We 74-101. 300. See pp. 102-114 for the detail of the scoring procedure. .222- over these for a ”total" functions score, constituted the three dependent variables for the factorial analysis. The five structural dimensions, dichotomized at points which would yield to the extent possible an even number of cases in each class, were used as the inde- pendent variables. After assessing this procedure as a reasonable working assumption, five different factorial arrangements of the inde- pendent variables and the control variables were established to assess the differentiating power and the independence of effects of the inde- pendent variables and to cross-check possible interactive effects between differing levels of each. It was found that rationality and both types of social relations significantly differentiated total and latent scores and that none of these significantly differentiated manifest scores. High rationality levels and diffuse relations were associated with higher levels of functions usage. J Involvement shogedno differentiating ability on any of the scores in any of the factorials. Commitment was ambiguous, sometimes showing statistically significant differencesfdnd at other times not. When significant, high commitment was inversely related to functions scores. In this respect, market news programming faces a paradox. The USDA . Market News Service program is meant to help the farmer adjust his occupational role to maximize income, but it is those who are most in need of this adjustment process who least use market news in this way. The picture on relations of the structural dimensions to the functions scores was not changed when the control dimensions were added. But, it was found that market news definition itself significantly -223- differentiated functions for all three scores. Those definitions of market news which were more inclusive than the USDA definitions, as manifested in the content of its formal program, were associated with higher occurrence of the functions. The sources of market news dichoto- mized as those which used only mass media for market news and those which used mass media plus personal sources or personal sources only, like commitment, were ambiguous. Sometimes statistically significant differences occurred, sometimes not. It was concluded on evaluation of the structural variables that both involvement and commitment could not be given detailed attention in the further analysis but, at the same time, they should not be dismissed as unimportant structural dimensions affecting market news functions. Their inability to consistently differentiate functions scores could easily have been the result of inadequate measuring tech- niques and procedures. Some evidence for this was presented. The need for more research here was pointed out and urged since both these vari- ables are more highly rationalized in theoretic terms than the remaining three and because it seems as though they should be particularly relevant to manifest functions. The overall inability to significantly differentiate manifest scores was suggested as probably not due to measurement deficiencies but a result of possible low level relations of the structural variables. ‘With regard to the defined structure as an empirical system, it was found that all of the structural dimensions were independent in ' their effects when they were significant and control had been exercised over the remaining structural variables. The lack of significant -224, interactions in the factorials further emphasized the statistically independent effect of the variables and implied the non-existence of what was called ”complex" system.301 Since only high rationality level and diffuse dealer relations of all the structural variables considered in joint occurrences required rejection of the null hypothesis, it was concluded that the structure empirically defined does not denote con- spicuously occurring emergent types. However, this condition was not assessed as being of overbearing importance. It is part of the sociolo- gists's job to detect significant typologies where none were previously conceived. Current methodOIOgy seeks aspects of true, existential systems which are relatively independent both in effect and distribup tion so as to maximize pragmatic predictive goals. This pressed fore- ward the criterion of prediction. 0n the basis of the factorial evidence, multiple regression type predicting equations were set up. The percentage of total variance explained by the variables significant in differentiating functions scores ranged for total score between 12.54 and 18.73 per cent and between 16.22 and 19.12 per cent for latent scores over the various factorials. The largest part of variability was left unaccounted for. However, it was deemed justified, on the basis of professional prece- dent and the objective tests of significance criterion, to conclude that at least certain aspects of the defined structure were important factors in.mediating functions for market news. All of the multiple ”regression" coefficients corresponding to the coefficients of 301. See pp. 57-58. ~225- determination (i.g., the percentage of total variance accounted for) were statistically significant. Structure then was redefined on the above evidence. Only those variables that were completely ubiquitous and significant were used. A two dimensional "structure" then emerged. On one axis was rationality and on the other the nature of social relations. Social relations were classed as diffuse, mixed, or specific on the basis of the cross-classed position of the neighbor and dealer scales. Placement in the diffuse dichotomy cell on both scales positioned a person as diffuse. Mixed types involved diffuse relations on one scale and specific on the other. Six structural types then were obtained and these examined relevant to the nine specific behaviors construed as the latent functions. A rationale was evolved which posited high rationality as indi- cating relativelz greater concern with price (or maximizing income) and, hence, implicating certain motivations, and diffuseness as offering the greatest opportunity for discharging certain behaviors stemming from these motivations on a continuum through least opportunity for specific relations for discharging and enacting the functions. It was shown to be consistent in explicating differences in the specific latent function behaviors. The fact that this rationale could be derived as a working hypothesis from the notion of functional constraint and could plausibly tie the disparate behaviors together in a logical fashion was taken as evidence for the validity of the study guide of functional constraint though it could not be taken as a formal test. Ebro definitive evi- dence for both the working hypothesis of ”motivation-opportunity“ and, in turn, functional constraint requires Egg gggtgg|hypothesis testing on new data; a procedure recommended on the basis of this study. -226- APPENDIX A -- Working Tables Table 14W - Disposition of the Original Sample by Sample Segments. -227- Not a county Mm 1:225:32. $333;ng 12:92.: Allegan Chesire 8 12 0 Martin 11 13 0 Barry Prairieville l4 19 O Berrien weesaw 14 12 0 Branch Nbble 16 10 0 Charlevoix South Arm 20 11 0 Clinton Riley 7 12 1 Emmet Readmond 9 ll 0 Gladwin Bourret 3 16 O Gratiot Pine River 18 29 O Hillsdale Reading 13 10 O Huron Sigel 18 3 0 Windsor 16 16 O Ingham Aurelius 17 16 0 Ionia Berlin 14 16 0 Lapeer Attica 10 12 O ~228- Table law - (Continued) Disposition of the Original Sample by Sample Segments. County Township 1:351:23, mining new 3;; Decision-maker Livingston Conway 17 12 O O Macomb Ray 14 17 O 0 Mason Custer 15 14 O O Mccosta Colfax 11 ll 0 0 Midland Hope 9 23 O O Menroe London 7 23 O O IMuskegon Sullivan 11 34 O 0 Oakland Oakland 9 90 O l Oceans Claybanks 20 9 O O Osceola Highland 6 13 O O Presque Isle Case 6 8 0 0 Saint Joseph Nottawa 8 9 0 0 Sanilac Buel l6 9 O O Tuscola Arbela 10 22 O O washtenaw Saline 11 9 1 0 Total 375 521 5 3 .229.. Table 24” - Distribution of the Structural variables by Sample Segments. Nei hbor RationauTz Hi Lo ...E._.. Diff Spec Dealer Diff Spec Structural Dimension and Leve Involvement Hi Lo Hi Lo ommitment Segment County 16021 10133 141/43 12011 22111 04043 12131 114023 22023 nu14111/1. 23456 Allegan 23A. 2142 02/4 Barry Berrien 212 [423 42/4 211.. 123 Branch I411.» [412 2 3 1 Charlevoix 1 Clinton 23 Emmet Gladwin 421 324 23 1 Gratiot 142211 031 141111 23 1 Hillsdale 31.1. 122 [462 [4141 353 231 122 I462 Huron -230- Table 24W - (Continued) Distribution of the Structural Variables by Sample Segments. County A .u m u am am 7 him at nr 08 an an .m Dt am mw O w v «to ..u m .m .1 m t m m. sea Lo Diff Spec Diff Spec Hi Lo H1 H1 L0 Huron [431 22]» Ingham Ionia Lapeer 312 2 3 Livingston 1 212 Macomb 411 351 Mason Mecosta l 23 Midland Monroe 353 000 1141 212 213 140 251 102 211.. 1142 Muskegon -231- he Structural Variables by Sample Segments. Table 26W - (Continued) Distribution of t Neighbor Rationality Hi Lo Structural Dimension and Level Degler Diff Spec Diff Spec lgvolvement Hi Lo Segment Commitment Hi Lo County 301 [410 310 Oakland 633 quaav [412 124 nlnc.4 Oceana Osceola Presque Isle 23 St. Joseph 1 11]». 226 23 1 Sanilac 152 Tuscola 23 1 washtenaw 150 206 150 174 182 206 198 158 160 196 Total .232.. Table 34W - Classification Basis for Type of Farm. ======================================================§—' 2===mr Classification Basis if Type Farm. Percentage of Income from Type Is Mixed General 15-40% from each of any combination of three single types Cash crap and dairy 30-40% from each or 40-50% from each Cash crOp and fat stock 30-40% from each or 40-50% from each .§iflzls 1129 Dairy 40% Fat stock (total of hogs, steers, 40% lambs, etc.) Cash crOp (total of wheat, corn, 40% field beans, sugar beets, pickles, soy beans, potatoes, etc.) Fruits (tree) and vegetables 40% (total of tree fruits and string beans, onions, green peas, asparagus, red beets, etc.) Poultry and eggs 40% Truck (small fruits and/or 40% vegetables) total of berries, melons, grapes, carrots, radishes, tomatoes, celery, etc.) Fat stock and dairy Biafgngggzhegzzhor Dairy and poultry Biafgngggzhezzghor .233- Table 44” - Distribution of Total, Latent, Level of Commitment. and Manifest Scores by Cgit- TQM” N Mean Level 0 1 2 3 4 5 6 7 3 9 10 11 12 x Score 1 1 1 3 2 1 2 1 ll 76 6.91 2 1 3 1 3 4 11 6 3 2 2 1 37 223 6.03 3 1 2 3 3 3 6 7 9 4 4 43 287 6.67 4 l 9 7 10 14 5 7 7 4 2 l 67 418 6.24 5 3 8 9 6 12 13 10 13 7 4 l 1 87 508 5.84 6 3 6 7 9 s 7 9 7 z. 5 1 66 370 5.61 7 2 5 6 5 4 6 3 3 3 37 215 5.81 8 1 l 2 1 1 l l 8 48 6.00 Latent Score 1 3 3 2 2 1 11 39 3.55 2 3 6 6 11 7 3 l 37 137 3.70 3 6 4 5 9 6 7 5 1 43 180 4.18 4 3 10 16 17 12 5 3 1 67 258 3.85 5 1 6 17 2O 15 15 8 4 1 87 319 3.67 6 2 6 8 18 17 5 8 2 66 231 3.51 7 2 ll 6 6 9 3 37 129 3.4 8 1 1 3 2 1 8 32 4.00 manifest Score 1 2 2 2 1 3 1 11 37 3.36 2 4 9 7 9 5 2 1 37 86 2.33 3 3 7 13 10 7 2 1 43 107 2.49 4 7 15 16 12 11 4 1 1 67 160 2.39 5 12 21 16 24 7 6 1 87 189 2.17 6 12 14 15 9 12 4 66 139 2.10 7 3 6 12 9 6 1 37 86 2.32 8 2 2 1 1 l l 8 16 2.00 * Commitment level 3 had one case with a score of 13. -234- Table Saw - Distribution of Total, Latent, and Manifest Scores by Level of Involvement. Involve- ___ Total Score* Mean L232; 0 1 2 3 4 5 6 7 8 9 10 11 12 N x Score 1 1 1 1 2 4 7 3 5 3 1 1 29 205 7.07 2 l 3 4 10 10 10 9 2 4 53 333 6.28 3 1 6 6 11 11 11 8 7 6 6 4 1 78 478 6.13 4 1 3 4 7 7 7’ 18 9 11 7 6 1 82 501 6.10 5 3 1 15 5 5 9 9 9 7 4 1 68 398 5.85 6 5 3 5 6 9 3 3 1 35 177 5.05 7 1 3 2 1 2 1 1 11 53 4.82 Latent Score 1 2 3 4 6 7 4 3 29 124 4.28 2 l 2 6 11 14 8 9 1 l 53 212 4.00 3 5 17 12 17 17 6 4 78 292 3.74 4 2 7 10 16 21 9 13 3 1 82 311 3.79 5 1 3 13 19 11 13 3 4 1 63 249 3.66 6 6 8 9 9 2 1 35 102 2.92 7 l 2 4 2 2 11 335 3.18 More 1 6 4 7 9 2 29 81 2.79 2 2 12 17 16 3 3 53 121 2-28 3 11 14 20 11 15 4 2 l 78 186 2.39 4 13 14 16 23 8 5 3 82 190 2.32 5 11 17 10 15 8 7 68 149 2.19 6 3 8 14 3 5 2 35 75 2.14 7 2 5 1 1 2 11 18 1.64 * Involvement level 4 had one case with a score of 13. ~235- Table 6-W - Distribution of Total, Latent,eand Manifest Scores by Type of Dealer Relations. Dealer Rela— _‘ Total Score* Mban tion80123456789101112N xScore Level l 2 1 1 1 4 3 5 2 2 2 23 161 7.00 2 1 4 4 9 11 6 5 3 2 1 46 321 6.98 3 2 2 4 4 2 3 7 4 4 1. l 35 247 7.06 4 4 7 6 15 12 17 13 12 7 8 1 102 599 5.87 5 l 6 4 5 7 1 2 3 2 l 32 184 5.75 6 1 1 4 2 1 2 1 12 71 5.92 7 1 4 8 16 11 14 18 14 9 8 3 106 562 5.30 ¥_L§tent Score 1 l 2 4 3 7 5 1 23 101 4.39 2 2 3 2 15 13 7 4 46 209 4.54 3 9 6 3 5 6 5 1 35 152 4.34 4 2 6 13 24 3O 18 5 3 1 102 373 3.65 5 4 4 11 4 4 4 1 32 113 3.53 6 2 4 4 1 1 12 43 3.58 7 2 13 26 24 21 10 7 3 106 334 3.15 16.1mm 1 4 4 5 2 1 23 60 2.61 2 5 7 11. 9 8 1 1 1 46 112 2.44 3 3 6 7 8 5 5 1 35 95 2.72 4 11 25 23 25 12 4 2 102 226 2.22 5 4 7 7 9 2 3 32 '71 2-22 6 1 3 3 2 2 1 12 28 2.33 7 16 24 21. 19 16 7 106 228 2.15 ‘ Dealer relations level 3 had one case with a score of 13. Table 74W - Distribution of Total, Latent, and Manifest Scores by Type of Neighbor Relations. Neighbor Rela— Total Score‘ N Eben tions 0 1 2 3 4 5 6 7 8 9 1o 11 12 xScore Level 1 1 l l 5 5 7 6 3 5 5 2 41 275 6.71 2 1 3 7 4 10 10 14 12 6 10 2 l 80 536 6.70 3 5 8 7 15 4 6 4 3 1 53 325 6.13 4 3 10 ll 9 16 20 18 16 7 4 2 1 118 697 5.91 5 1 4 7 11 10 6 7 4 6 7 1 64 312 4.88 Latent_§gore 1 2 3 4 15 12 3 2 41 172 4.20 2 l 4 7 l6 17 18 10 7 80 333 4.16 3 2 8 15 10 8 6 3 1 53 208 3.92 4 1 8 23 29 27 15 10 3 2 118 421 3.57 5 2 10 18 11 ll 5 6 l 64 191 2.98 Manifest Score 1 5 10 7 8 2 7 2 41 103 2.51 2 6 17 17 19 13 6 1 l 80 203 2.54 3 6 12 11 14 9 1 53 117 2.21 4 15 20 31 24 20 6 2 118 276 2.34 5 11 17 16 ll 6 3 64 121 1.89 * Neighbor relations level 4 had one case with a score of 13. Table 84W - Distribution of Total, Latent, and -237- Level of Rationality. w Manifest Scores by Ration- Totgl Score" Mean fility 0 1 2 3 4 5 7 8 9 10 ll 12 N szcore evel 1 1 2 2 6 4 5 7 4 6 l 38 268 7.05 2 1 3 4 5 9 3 4 3 32 218 6.81 3 1 2 3 2 4 9 9 5 2 2 2 1 42 271 6.45 4 2 2 6 l 7 4 5 9 2 38 252 6.63 5 2 7 9 9 16 16 7 13 5 5 1 90 514 5.71 6 l 6 9 18 14 13 18 12 10 5 7 1 1 116 622 5. Latent Score 1 4 7 9 9 7 1 1 38 167 4.39 2 l 7 8 8 6 2 32 145 4.53 3 l 1 4 11 9 9 6 1 42 166 3.95 4 3 11 10 8 3 2 1 38 159 4.18 5 10 16 21 23 7 9 4 90 314 3.49 6 3 15 31 18 21 17 4 6 1 116 374 3.22 Manifest Score 1 4 7 4 10 9 4 38 101 2.66 2 2 8 9 6 6 1 32 73 2.28 3 5 7 ll 7 7 4 1 42 105 2.50 4 5 7 6 11 5 3 1 38 93 2.45 5 12 18 20 23 12 5 90 200 2.22 6 15 29 32 19 11 6 4 116 248 2.14 * Rationality level 6 had one case with a score of 13. Table 94W - Analysis of Variance for Total, Latent, and Manifest Score Classified by the Structural Dimensions of Rationality, Dealer Relations, and Neighbor Relations. Structural Dimension, Score, Sum of Deggees Mean F and Analysis of variance Item Squares Freedom Square ratio Rationality - Total Score Total 2,242.77 355 Among 120.99 5 24.20 Within 2,121.78 350 6.06 3.99** Rationality - Latent Score Among 82.37 5 16.47 Within 889.10 350 2.54 6.48** Rationality - Manifest Score Total 803.34 355 Among 10.80 5 2.6 Within 792.54 350 2.26 .96 Dealer Relations - Total Score 2 242 77 355 Total 0 Among 362.37. 6 27.14 Within 2,079 .93 349 5096 4055** Dealer Relations - Latent Score 971 47 355 Total e . Among 91.27 6 15.21 Within 880.20 349 2.52 6.04** Deal r Relations - Manifest Score 8Total 803.34 355 Among 13.39 6 2.23 Within 789.95 349 2.26 .98 N i hbor Relations - Total Score 0 g Total 2,242.77 355 Among 142.77 4 35.69 Within 2,100.00 351 5.98 5.97** N b Rel ti 8 - Latent Score eigthzal a on 971.47 355 Among 64 .08 4 16 .02 Within 907.39 351 2.59 6.19** ti - Manifest Score “$32315 ”a 303.31. 355 Among 18.61 4 4.65 Within 784.73 2.08 -239- Table 104w - Mean Function Scores by Factorial Cell; Basis of Classifi- cation, Four Structural Variables and Definition Control. Num- Mean Mean Mean Factorial Cell b” Total Latent ””11" Of Score Score fest Cases Score Commit H, MN more, Deal Diff, Neigh Diff, Rat 25 7.16 4.34 2.80 I I I I Rat 15 5.93 3.67 2.27 a u w Neigh Spec, Rat 24 7.29 4.66 2.83 I I I. I Rat 15 5.93 3.60 2.33 " " Deal Spec, Neigh Diff, Rat 12 6.42 4.00 2.42 I I I I Hat 9 5.67 3.78 1.89 n u a Neigh spec, Rat 6 5.33 2.83 2.50 I I I I Rat 12 6.25 3.67 2.58 I MN less, Deal Diff, Neigh Diff, Rat 6 8.00 4.83 3.17 n a I I Rat 9 5.22 3.22 2.00 a u . Neigh Spec, Rat 9 5.00 3.56 1.1.4 I I I I Bat 20 4.05 2.60 1045 I I Deal Spec, Neigh Diff, Rat 4 5.23 4.00 1.25 n u u a Rat 13 4.00 2.69 1.31 I I I Neigh Spec, Rat 5 7.00 4.80 2-20 n I I I Rat 3.21 1.79 1.43 Commit L, MN more, Deal Diff, Neigh Diff, Rat 15 7.27 4.87 2.40 I I I I flit‘fll t‘flit‘fil f‘flit‘fli r'nzrln: rln:r*na E‘DSE‘HI r‘nit‘UI t*n:r4n: Rat 13 7.54 4.77 2.77 I I I Neigh SPOC, Rat 12 7.17 4.50 2.67 I I I I Bat 14 5.93 4.00 1.93 n n D 1 S N61 h Diff Rat 6 6.00 4.00 2.00 I I ea I pee, g I ’ Rat 12 6.42 3.3; 2.3; I I I N81 h S SC Hat 9 6.78 3. 2. I I I g ' p ’ Rat 16 5.56 2.88 2.69 Diff N i h Diff Hat 8 7.00 4.50 2.50 : MN £838, D881“ , 9 S n , Rat 14 7.71 4.64 3.33 I I N i h S ec Rat 2 5.00 3.00 . : I I e g I p , Rat 5 5.00 3.20 1.80 N 1 h Diff Rat 3 6.00 4.67 1.33 z : DealIspec’ 9 g ' , Rat 10 7.10 4.22 3.88 I I N 1 h S 60 Rat 4 5.25 3. . : I I e g ' p , RBt L 15 3.73 2.00 1.73 .240- Table 114W - Mean Function Scores by Factorial Cell; Basis of Classifi- cation, Fbur Structural Dimensions and Sources Control. 532- Mean Mean 3:2:- Factorial Cell f Total Latent f t C 0 Score Score es ases Score Commit H,Mass Plus,Dealer Diff,Reigh Diff,Rat H 21 7.00 4.14 2.86 a u u. a Rat L 9 6.33 3.67 2.67 I I I Neigh Spec,Rat H 17 6.76 4.18 2.59 I I I Rat L 17 5.00 2.94 2.06 I‘ I Dealer Spec,Neigh Diff,Rat H 6 6.67 4.33 2.33 n n n n Rat L 9 5.11 3.67 1.44 " I. I Neigh SPGC,R&t H 6 7.00 4.33 2.67 I I I Rat L 12 4.75 3.00 1.75 I Mass 0nly,Dealer Diff,Neigh Diff,Rat H 10 8.00 5.10 2.90 I I I .' Rat L 15 5.27 3.40 1.87 I I I Neigh Spec,Rat H 16 6.56 [9.25 2.31 I I I Rat L 18 4.72 3.11 1.61 I I Dealer Spec,Neigh Diff,Rat H 10 5.80 3.80 2.00 I I I I Rat L 13 4.38 2.77 1.61 I I Neigh Spec,Rat H 5 5.00 3.00 2.00 I I I I Rat L 14 4.50 2.43 2.07 Commit L,Mass P1us,Dealer Diff,Neigh Diff,Rat H 18 7.72 5.00 2.72 n I I I Rat L 17 7.76 4.76 3.00 I I I Neigh SpBC,R8t H 8 6.25 3.75 2.50 n u I I Rat L 14 6.00 3.93 2.07 I I Dealer Spec,Neigh Diff,Rat H 7 5.86 4.29 1.57 I I I I REt L 11 7.00 4.27 2.73 a a n Neigh Spec,P.at H 9 6.41. 3.89 2.56 n n u a Rat L 17 5.35 2.76 2.59 I 8 er Diff N61 h Diff Rat H 5 5.20 3.80 1.40 a Mass.0nly,De 1 I ’ g I ,Rat L 10 7.40 4.60 2.80 I I I Neigh SpBC,R&t H 6 7.50 5.00 2.50 I I I ' Rat L 5 4.80 3.40 1.40 n a Diff Nei h s ec Rat H 2 6.50 4.00 2.50 I I Dealer" , g I p ’Rat L 11 6.45 3036 3009 u n u Neigh Diff,Rat H 4 6.00 3.25 2.75 n a w a Rat L 14 3.86 2.07 1.79 -241_. Table 124W - Mban Function Scores by Factorial Cell; Basis of Classifi- cation, Three Structural Variablea and Two Control Variables. ET‘M was: Factorial Cell ; Total Latent f t- 0 Score Score es Cases Score MN more,Masa plua,Dealer Diff ,Neigh Diff, Bat 31 7.32 4.55 2.77 H ' Rat L 15 7.40 4.40 3.00 ” ” " Neigh Spec, Rat Hi 19 7.05 4.26 2.79 " ” ” Hat L 19 5.95 3.74 2.21 I! ll Dealer “Spec Neigh. Diff, Rat H 9 60410 4011 2033 7' II Rat L 11 6036 4009 2.27 “ " ” NeighI Spec, Rat H’ 8 7.25 4.00 3.25 " " " ” Rat L 18 6.12 3.47 2.65 " Mass only,Dealern Diff ,Neigh Diff, Rat H 9 6.78 4.56 2.22 " ' ' Rat L 13 5.85 3.92 1.92 ” * Neigh Spec, Rat H 17 7.47 4.47 3.00 n n N ' Rat L 10 5.90 3.90 2.00 I! 3' Dealer "Spec, Neigh” Diff, Rat H 9 60].]. 3.89 2.22 n " Rat L 10 5.80 3.30 2. 50 u n n Neigh Spec, Rat H 7 5.00 2.86 2.11. ” " " “ Rat L 11 5.45 2.82 2. 64 1 D a1 Diff Nei h Diff Rat H 8 7.33 4.50 2.88 MN less,Massnp us, 6 er" , g a ,Rat L 11 7.09 4.36 2.73 fl 0 " Neigh Spec,Rat H 6 5.17 3.33 1.83 n n fl " Rat L 12 4.67 2.75 1.92 ' " Dealer S ec Neigh Diff Rat H A 5.75 4.75 1.00 a u I p ’ I ,Rat L 9 5.89 4.00 1.89 0 fl * Neigh Spec,Rat H 7 6.00 4.14 1.86 8 fl " a Rat L 12 3.83 2042 1042 " Diff Nei h Diff Rat H 6 7.50 4.83 2.67 " Massuonly,Dealeru . g ” ’Rat L 12 6.42 3 23 $.33 u u " Nei h Spec Rat H 5 4.80 3 . I II N g 01 ’Rat L 13 3.85 2069 1015 N i h Diff Rat H 3 5.33 3.67 1.67 : : Dealer-Spec’ e g a ’Rat L 11. 5.00 2.86 2.115 a a 9 Nei h Spec Rat H 2 7.00 4.00 3.0 I u I 8 ' ,Rat L 17 3035 1088 1047 ..242- Table 134W - Mean Function Scores by Factorial Cell; Basis of Classifi- cation, Commitment, Involvement, Rationality and 'No Control variables. Num- Mean Mean Wan Factorial Cell be? Total Latent Vani- Of Score Score fGSt Cases Score Commit H, Invol H, MN more, Mass plus, Rat H 20 7.30 4.40 2.90 " ' ' ' Rat L 12 7.25 4.58 2.67 ' ” ' Mass only, Rat H 17 6.65 4.12 2.53 " " " " Rat L 13 5.69 3.69 2.00 “ " MN less, Mass plus, Rat H 6 6.00 4.00 2.00 n n I I Rat L 5 3.80 2.40 1.40 " " " Mass only, Rat H 6 8.17 5.33 2.83 N n N I. Rat L 15 4047 2073 1073 " Invol L, MN more, Mass plus, Hat H 20 7.15 4.15 3.00 n I ' " Rat L 13 5038 3008 2031 " fl " Mass only, Rat H 11 6.45 4.18 2.27 " ” " a Rat L 12 5.25 3.17 2.08 " " MN less, Mhss plus, Rat H 5 5.80 4.20 1.60 " fl ” "' Rat L 16 3.69 2.33 1.31 " fl " Mass only, Rat H 7 5.00 3.43 1.57 I! II N I Eat In 20 4.00 2055 1045 MN e Mass lus Rat H 10 6.50 4.20 2.30 Commit L, Invol H, mor , "p , Rat L 10 6.30 4.00 2.30 n n 1‘ Mass only, Rat H 8 6012 4000 2012 u u n I Rat L 8 5.12 2.75 2.37 " " MN 1688 Mass plus Rat H 7 6.71 4.57 2.14 n n a ’ a ' Rat L 11 2.73 3.82 2.91 0 fl " Mass only Rat H 1 .00 5.00 1.00 n n n n ' Hat L 11 6.36 3.36 3.00 I Mass lus Rat H 18 7.39 4.67 2.72 " Invol L, MN more, I.P 9 Rat L 26 6.46 3.35 2.3; n I " Mass only Rat H 6 7.50 4. 7 2. N 0 fl ” ’ Rat L 11 6.82 4.18 2.64 ' Mass lus Rat H 7 6.14 3.86 2.29 2 I MN £888, Hp ’ Rat L 12 6.58 4.00 2.33 n a Mass onl Rat H 2 4.50 2.50 - z n I N y, Rat L 10 3.70 2.40 1.30 M -243- Table 144W - "Interactions" Matrix for the Four Structural Variables and Market News Control Factoria1.* Interaction Total Score of Commitment Dealer Neighbor Rationality Definition 1 Mitment .— - .16 -2 o 02 +1 0 92 +1 .0 5 Dealer "" +1.01 ‘1' 036 " 027 I Neighbor "" .1003 -2084 Rationality -— - .99 Latent Score Commitment Dealer Neighbor Rationality Definition Commitment -- -l.16 -1.88 +1.24 + .54 Dealer ." "’ 045 " 043 + .56 Neighbor - - .55 -2.19 Rationality - -l.85 Definition - Manifest Score Commitment Dealer Neighbor Rationality Definition cmtment - + .92 -1020 +1.72 +1.07 Dealer ."' +2.01 + 098 " 099 Neighbor -— -1.03 -2.15 Rationality -- + .11. Definition - M ’ Cell values are values of t read with seven degrees of freedom. -244- Table 154W - "Interactions" Matrix for the Four Structural Variables and Sources Control Factorial.* Interaction Total Score of Commitment Dealer Neighbor Rationality Sources 1 Commitment - - .20 -1.47 +2.33 + .55 Dealer '“' + 051 + 031 " 074 ‘ Neighbor -— -1.58 - .13 Rationality -— - .57 7 Sources -— 1 Latenttgcore Commitment Dealer Neighbor Rationality Sources Commitment - -1.60 -1.37 +1.70 - .49 Dealer —- - .35 ’ - .19 -2.34 NBigthr .. " 099 + 028 Rationality - - .64 Sources .— Manifest Score Commitment Dealer Neighbor Rationality Sources Commitment - +1.31 - .91 +1.92 + .36 Dealer - +1.93 + .67 +1.23 Neighbor - -1.46 - .49 Rationality - - .23 Sources " ========================================================================= * Cell values are values of t read with seven degrees of freedom. -245- Table 164w - "Interactions" Matrix for the Three Structural Variables and Two Controls Factorial.’ Interaction Total Score of Dealer Neighbor Rationality Definition Sources Dealer ." +1.34 4’ 014» - 054 - 050 Rationality - - .59 - .72 Definition - + .34 Sources - Lgtent Score Dealer Neighbor Rationality Definition Sources Dealer -— - .12 - .42 + .62 -1.83 Neighbor -- - .80 ~2.15 + .76 Rationality -- ~1.35 - .75 Definition -— - .06 Sources - Manifest Score Dealer Neighbor Rationality Definition Sources Dealer -- +2.23 + .66 -l.54 +1.24 Neighbor - -l.65 -2.53 + .07 Rationality -— + .52 - .22 Definition -- + . 0 Sources “ * Cell values are values of t read with seven degrees of freedom. -246- Table 174W - "Interactions" Matrix for the Commitment, Involvement, Rationality and Two Controls Factorial.* Interaction Total Score of Commitment Involvement Rationality Definition Sources 1 Commitment -- +1.33 +1.69 +1.70 --1.01 1 Involvement - - .24 -l.91 - .31 I Definition —— + .24 Sources -- Latent Score Commitment Involvement Rationality Definition Sources Commitment -.. +1.45 + .95 + .96 -1.33 Invalvement "’" " 014 -1017 + .22 Rationality - -l.42 - .61 Definition - - .03 Sources *- Manifest Score Commitment Involvement Rationality Definition Sources COMiment "" + .58 +1067 +1066 " 019 Involvement -- - .23 -1.77 - .71 Rationality -- + J+4 + .21 Definition -- + . 7 Sources '“ __._‘-_ * Cell values are values of t read with seven degrees of freedom. .247-' Appendix B - Survey Schedule" * The schedule contained in the.Appendix has been condensed to reduce it fron.its original length of 45 pages. These changes effect only the appearance of the schedule. All questions appear in their original wording and sequences. ~248- hrn whet News: June at, 1957 Budget Doreen lo. Ibo-5711 Social Research Service (Agricultural hperinont Station) mix-ation new 10/15/57 Elohim State University How do you do. I'm from the Sociology Section of the Agionltnrol hperinont Station st Hichigon State thiversity. Io are conducting a study of market ness end farming practices. I'd appreciate your answering some questions for no. 1. sir-t .: .11. how many acres, on together; I. are you renting this year! 1:. do you on? Tfim)m.ymmm.r Tfim)snm.:th.o. those are you actually using so: are you actually using so: ____orop lend a rotation pasta-e ___orop lend a rotation pester. Mt pasture MM posture “reminder __rent out or put out on shores Jeanine» 2. low I'd like to find out some things about shot you produce here. first of ell, mt do you consider use you ‘15 product lost your “‘— Ihst did you do with your {gyms} lost your? that other things did you have lost year? that else! Did you ..11 all these things in this fern for cash? (11‘ “HO. 111D (I)! WHICH use roe SOLD AND um comm no son ormm man was 801.13. mmormroaulsnrmrusr.) . Approximately what percentage of your gross income from forming did. 11 (START fling! PRODUCT} amount to! RODEO! SOLD at $ 01' 03088 mm mm (on 75%) . thin ___!es __ lo Other “Yes __ lo _ (PRODUCT 20 mos (PRODUCT TO HIGH mun “103331? 0113821013 32 or 35 momma-mm III-I. mm. .11 THIS PRODUCT 18 m Bull WHOSE 23-31 AS 1'03 m MORTON-m QDISEIGB, mm mm) moursnmazubcmr 35. rumour 41.. -249. mums QMIOIZTORWIOHWWEWNBMII W910” 23-31 D! MEETING DNISIWG AND “@1018 32 or 35 03 mmxmmmnmnmummmsrmumm 01 m m.IIG PAH. ran as: momma-lime 003910180341). Dan momma! mum moor mos nouns moms: mpdamor or cacao rm mom. amonmnsmsrnonccrsmrm 091.1me mmmmnmmmnmmlmcsosh. , (1) 11' on 18 "(411", can 11!; (z nonlsncoruamroimcrm W, mu: (3 11' m 011 mm m 00117111310118.0me on or gmmnmmr; (t) 11' non 13 A comxos, 011008: on 1113111113111. 1011 m we MIWSHIP cannons (32 or 351.03: mm menace mos mom-z mum! momma or ences rm noon. 1: 1110 03mm 1m,m1mmcmomnmomc1mmtonwn snow. (1) 11' on IS Ilium, can 11; (2)1ron18100mmrommm Inn". 13111: (3) Irmcnmmoommzons,monronormosnwn mnmmmmmcmmm, Daemon. 11m 611 as can) 103 Daemon-111nm 9111511013, amass on mm- nmu’. nmonuoncnsncmson mamma— , more conscious. use an out: on can 111 mm 11101310:- umrc cannons. , _ , (b) nmmcommon.monrcnamcummm momma-11mm commons, as m on. 11 son on as can son memos-unite consume, 03008] on 123111112111. 11 no on mm on as am 1011 momma-mo cannons, on m on can 102 Daemon—11mm QUISTIOIB. . 1! 80381011“! WINS-AID 133130021035 m ms 'MLII' AID 'lflJ’Ol' mmmmmmmmmmonmmmms P10902110! 01' me I'm mm. . 3. Doyonordinerilydoownon-tsrsvorktorinoosohrmtnoyoofl W _____lo ____r.o: untamed 1.1931 c . (1:!. :03 18 “6310051024”, Sn? so QUESTIOIIM») am a. you work: 1 - > 1 W m for is that from here! silos . Sow lmqm you worked there? yosrs Doyonlnvorogdsryoer-roondsork,ordoyon:utvorkoflthotors ports orthoyoort myesr: Isitstnudsy'ssorkt “hum __1'ertlsy :Pertofthoyeorsmtpertoftheyocdoyonmkt (mam) DoyonsorkefcndsyorJnstpsrtotthelsyr ____1'u114oy ____Psrtw -250- Ihst proportion of your total gross income from all sources came from your non-fun work last year? (mmnnm PRES!!! cm) __ Less than l/lb __ About 1/2 __ More than 3/11» __ About l/lt __ Botwun 1/2 8. 3/lb __ D.K.. not all __ Between 1]“ d 1/2 __ About 3/4 __ No answer ll». low I'd like to set you about some specific ways nos-hot news can be gotten. 4.1 Do you read s dsily newspaper fairly regularly? __ 11. (am so QUISTIOI 4.3 or mm) __ Yes: Which ones! w ... 10.2 Do you look at any form nsrhet news in any of the dsily newspapers you read regularly! __ No __ Yes: lhich ones? __ _ _ Now how about radio! “.3 Do you have s radio in working condition? __ n. (sup ro QUISTION 4.5 on own sensors) __ Yes '0.“ Are there any radio progress that you listen to fairly regularly for foru‘ market news? __ Yes: which ones! (as moons um, swim. m nus) .___ 15.5 low 1m lino you to 1»): st this 11st and 1.11 so if you get uni-ht news from any of those sources. (PRISM clan) lone (our in cinemas 5.2) : Yes: (SPICIH) ___Magesinos: mob ones? _ ”weekly newspapers: men ones? ___1'e1ovisionx which stations! (m 0m 1.1m M... cm moms) A.— —— vv ors ”fl-ushers end route drivers “County agents and other government personnel ___pth.ru (8mm) (11' conscious 4.1 to 4.5 1111110113 ran an mmmr D038 m noun mar nus, no»: 1111 or men sensors. 3119 low 10 common 9, cm! 10. an AS! 11) 5. Now I'd like to find out more specifically what whet news you got. 5.1 5.2 5.3 -251- (‘8! m HOLLOW 1'03 W! 80030! mom 1! WHO! 1|». 5) lint kinds of things are carried 11y (MIL, (roams-comm. ASK: 'Ihstkinds of thingsdoyougotfrsn (somzr' ) (roam soonest ”W08”, mmnmmm. USING BUCK QJIISTIOHB AS What about PRODUCT 7', “Myth”; else" 11‘ P310], SUPPLY, OUTLOOK, 1150. 'for whst products"l ) E11351 saunas: ) __ sworn sown: ) 11 (DO 30! A8! 11' ms 10 QUIBTIOI 4.1 or h.2 us '30“) I" I'd 11b to till: o. little bit about newspapers. You said you read the W for nurlnet news. mt nrkot nous does it curry! 1! mm 18 our some moon. ASK: “ “at kind! of com MW? Is there snything besides in the nrket news from 11 mm 13 on! mowers, ASK: that things about __ (11g! PRODUCT) does it cover? mthing else! How about _ W that things a... it cover! that market news a... the on mama ...-m (WNPMABKBOVI) (DO IO! 18! 11' mm 20 0113521038 I01.3 or 1t.“ WAS '10“) low I'd like to not: s few more questions about radio. You said you listen to I ! lhet nsrket newl does it cerer 11' mm 18 on! com CATEGORY , 18K: 3 out: n 3 ‘ O that kinds of ) a... that include? 1. thereonythlubesidos w . : 03141 ) in the mket news from 11' mm IS ONLY PROMISES. ASK: Ihst things about cover? Anything else? How about lhst things does it cover! Int mm am too- W ...-m does it curry or (MNWEL‘NW) .252- W: mmnmmmr mom: 6. 6.1 6.2 5.3 _.__8eens to have s clear definition of vhnt nsrket news is. __!ries to figure out s clear definition but no indication that it one thought out in advance. ____8eons very vague: definition of no apparent lasting mooningfulness. (18! 1'03 RAGE mum m RADIO PROM MENTION A8 A 80030] 01' HARD! W) I'd liloe to find out some thing- obout the w you would rote (such of) the newspaper“) and radio program“) so to how well they hondlo nrket ness. (mm 80m cm) In terms of the kind of Job _ (800mg) does in handling nsrket ness, vhere would. you place it on this scale (TE! 80”.! 18 A! m BOTTOM 01' an NT PM. 33003!) EACH OHOIGI rm. AIDJ'OR m 80118.3! 30! mm momma AS! mmnmn) What are your reseone for not rsting {80%) higher? (an rm mmm m m amour Prom RENEE As soon) (mom 1'03 mans non ms LIB!) n. It doesn't cover all the products I'm interested in. b. It doesn't cover all the nerlnet locations 1'): interested. in. o. he reports of local whats ore inaccurate. d. be information is not up-to-dete. _ WPAPIBS ONLY: A RADIO P200308 011qu f; Reports are not easy to loonto. p. Frog-en is on st o bed. tins. Q. Printing or set-up is bed. In. Poper is hard to get. (1. Station reception is bad. J. Paper cones too late. r. Announcer does a poor Job. some: M0118: (mono 30m 0301613 ma non mum on mm mm A! m man or ms Pm. mxcmztm sown. rm rm cmxcns m m Hummus or m melanoma Em: m m, m 11 rm 1: m m on soon sooner, 8m m 80m TO m>mronnm man) Is this ebout the soy you would rsto the newspaper-(s) end rsdio mono-(I) you got nurhet news front __ Yes ___1Io: (A8! momma: no annoy me mares. 11mm 18 to m 3 am an m. mom) mm.) , 453- 6.» (sm no 80112018 in um um: cum-m, m cmnoxm 10 m m nun, Ant) N do 7'! 1‘“. w m hid!!! than ~m7 Emu.“ “ om m:- Poor 7. is for us you know, does the United Btutes Deportnent of Agriculture originally provide u considersblo usount of the nsrhot nous infern- uticn to the sources you get it from __ Io __ lost Are you pretty sure of the“ ___l‘o: mob sources do you think probsbly get informtion from the USDA? h ___Ios: Is that true of all the sources mget nurhet ness from! _.__ Ios __ 3o: mick does it include! (£83103 EROS! somsmmcnmm 611'“ m OBIGIILEOR) . 4 You-mentioned that (sons or ell) of the nterisl you get from (QUE!) cones fron the USDA. how do you feel about the Job that the vent does in supplying such infern- ution to (M)? (sooner: . ) _ k fl 8‘. (mxmnmpommmmzmrmnmmnmon smmmm) You've mentioned thet you get sorbet ness in s nunbor of ways. Do you prefer sons of then nore then others or ere they ubout equnlt .. mu preference: If in the future you could continue to get nut-bet ness fron Just one of those sources that you're using, which would you prefer to continue vitht in A __ Reference: thich one source, of all those you new use do you like nostt __ u m that one? 9. Do you use any of tho follouing kinds of infornution in figuring out :hetigg problsllt (mom m m 0! QHISHM 5.1. 5.2, 5.3 '0 * Grading. quality, M 39011180 mind Weather Consumer preferences Different we of nurhoting Other (Volunteerod) 10; 10.1 10.2 10.3 11. 12. 45»- (Asxnmormmnmmmmxmmmmmmn ABOIPMWWIIIB) We've been tulhing shout for: sorbet nous for s. for minutes not end you've described .- u '« ' ; : 5.1” 5.2 , and 5,3] us nsrket nous. Are theroenyotherthingsthatyuuthinkofus-rlcetnevst ____lo fires: that are they! Is there any other inforntion about fern nerketing thet you'd like to get but can't or don't get now! __ lo __ res: mt! Is there mthing thut you now get in whet nous thst you'd Just us soon not get! ___llo ___Ies: whet! lould it ushe ony difference to you if you could no longer get on informtion about nsrhaetsl __ to: W not? (II 'MBMGWBW’MWAEM I! cum, A813) Suppose there wasn't any in shioh you could get this kind of infornution. lould it nob ow difference to you! __ No: my not? ____Ies: In whet says! __ __ test In what ways? (11%me 01mm 01m, mmm HP! mm) (H W110“ tin-10 INDIGLEI mu m mm: D03 '0! mm m mm! NOMIOI DIEIGBI, MI! NWO MAD-II summm 'J'irst ofull, ' IIQHEIIOIIZ. MAM W 01' 91131101 12) tell. new, you seid'thst you get fern nsrhet new on u fuirly reg- ular basis. I'd like now to not ubout sons things mch don't hove too uch to do directly with nsrhoting. but which other for:- ers hove ssid ere isportent. first of oil, do you sonetines hove swone else. libs fuily non- bors or e friend, got nerhet infers-Men for you? __ Yes: ihoeould thst be! shut kinds of informtion do they get for you! lhent um __ to: m not? (11' momxon 18 nor VOW, 181:) Doesonyonoolseinthefenilyaetnsrhetnevsinforn» tion? __ Yes to 13. 13.1 13.2 13.3 -255- (IT QUISTIOE “-10 mxunmmmmmnorsmmnm m WIOIDIENELI. AID mm 10 QIIIS'I'IOI’ 12 13 '10.. own ALL 03' “531018 13 and 20-32, LED SKIP NOV 50 WTIOI 1'4) Bonetiaes when you are in general conversation do you discuss the whet news you are getting? We aean Just talking about narkots the weyyounight talkabout the weather-not particularlytoget any new information. __ lo: m not! ___Ies: Oanyourecall the last tine, or any tine. that this happened! mat is, can you tell no a little about it? (P3031: When was it? there was it? Iho was it with! lhat type of informtion was it") Ingenoral insuch situations.whatkindsofpeopledo youtalkwithinthiswayl (Ir 'reominos'rnowroom' Ismmmmm. sermmmonm) How about with people you don't know too well? __ Yes __ lo Have you ever been in a situation where conversation was logging and then sonebody started talking about asrkets and the conversation picked up again! __ res __ lo haveyouever donethisyourselft Iaean, haveyoueverbroushtup marketing when the talk soeaed to be stalled! __ lo: thy not? __ res: with when? (11 mom I DON'T ”W100 ml.“ 18 30! mm, Adm) . How about with people you don't know too well? __ Yes __ lo Most people think me hiyaly of sore persons than they do of others. l'or ennple, there are probahly sons far-ere around here that you regard acre highly than others. they may or say not be the biggest and wealthiest farners. low I'd like to have you think about these farmers that you regard highly. I don't want to know their names, but I'd Just like to have you think about then to yourself. how, would you on that those farnors when you regard aost highly know more about whts and marketing than other formers around here do! No: Doyouthinkthatwhetherornotafaraerknowsnch about markets and asrketing has anything to do with how hidily you regard hint __ lo : shy not? that things are iaportautt _ __ Yes: In what ways? um __ 1%. 11151 11152 14.3 up.» 11?. 5 1“. 6 15. .255- ___Yes: Do you think that the fact that these far-ere nowhere about minute and urinating tends to asks you regard than lore highly! ' __ lo: W not! h __ Yes: Ihyt _ how I'd like to find out soae things about your experience in fan-ing. Didyoug'owuponafarn‘l ___mofchildhoodspentonfarn ____1'ert ____lone that are the mass of how long did Did they give you the schools you attended! you go there! any training in . agriculture? ____Yes _____lo hat was the last pads of school you completed? Have you had any additional training, such as short courses or vocational training! ____lo ____Yosx Ihatkindofcourse Bowlong did. it run? didyou taker Did you ever belong to: a LE Club! __ Yes __ lo the Future Turners of Anericar __ Yes __ lo Ioreyouever outofforningfor awhile? __ Io __ Yes: for how long? * Ihat kinds of work did you do during this time? Bveyou over livodinacityd . _ D‘o V __ Yes: Ihat kinds of work did you do during that period! Inthepast, haveyoueverhadanon-farajobwhileyouworefsraiag‘r (11mm: ID! .103 W! m. nus mum END Bill mm 1' 01133131013.) ..'._ 3° __ too: low how long! lhat kinds of work did you do baring that tine? Could you tell no how you happened to get into faraing? .1 -257... 15.1 At the tine you first started to fern, were there sone other kinds 16. 16.1 17. 19. M of Jobs you would have preferred! ___lo :Yes: mydidyougo into faraingrather thaninto one of these other Jobs! Howdoyoufoelaboutyourworknow? Doyouthinktheresreeny other kinds of Jobs that you would enjoy acre than feraingt __ Yes: lhat are they! - that wouldyouwant to do 1:wa ofan occupation ifyouwere younpr and Just starting out on your first Job? (uncouth-cur SITOJ‘HIASOEB ISGIm): Doyouevor thinkof getting out of faraing and starting over min! __ Do: may not? ____ Yes: lhy? that would you rather do? Why that? he more are a number of different things that farmers have said to us about farming. I'm going to read you a list of these and I'd like you to tell as which side cones closest to your own point of view. easy work - hard work creative work - routine, repetitive work healtlw work - unhealtlw work takes special skills not everybody has - anybody can fare: interesting work - boring work aasnis his ownboss ~ananis tieddownby his far- freo tine whenever you want it - long hours endocrine-postman: Is there anything elseyouwouldliks toaddabouthowyouwould describe fuming! A H I H Ill Ifyouhadto describe howyoufeel about farninginiust one of the above phrases, which one would you choose! (01361.!) Iron your point of view. which one of these descriptions gives the host misleading inpression‘l (80pm!) Does how much you know about nsrkets and narketing have sow effect on how you feel about forming! _____lo: m not! :1» : could you tell no a little about it! ’ In nuabering the original schedule, the washer 18 was inadvertinently Iupped. 20. 21. 21.1 21.2 .253. (11’ m1” lib-12 IIDIOAEID m2 YE! escrow: DQIS 10! mm in MM! 130M103, rum DIRECTLY OR rumour. mun “£1088 20—32, Ind. SKIP m 10 QUISIIOI 33.) You've been getting market inforaation on a fairly regular basis. Asidefrcadirectusos, deyoufeel that ithashadaweducationsl value? that is, has it helped you develop any generally useful ideas about farming or marketing! __ No: W not! __ Yes: Ihat are acne of those ideas? __ Don't know how I'd. like to ask some questions about the relation between market informtion and your production practices and plans. Yirst. areyoufiguringonanychanges intheanounts orkinds of products or enterprises for m year? ‘ ___-le: Is the fact thstyouaren'tplanningtcnskeanychanges next year a result of any whet news you've been getting? : ____lo (SKIP 1'0 21.5) __ Yes: lhat informtion is that? where did you get it! (sup so 21.3) __ Yes: that changes- (I'r SPIDIIIO 1 that in- fill it in- fill it in- mm 01' DASH ease or de- crease or crease or am , IHGL'UD- reuse your decrease decrease ING mm investnent? flexibility? costs? Finn" , DROP, or SWINE.) mm mums! 0081's CRAIG]! '. g a e I g a e g a a a. e a a e o e 8 ... a .. a .. Are you considering any of these changes on the basis of an mrket information you've been getting? ____lo :Yes: lhich ones? that infornation there did you (US! D08.) is it! get it: -259- 21.3 Is this the way you usually figure out whether to nuke changes in enterprises? Yes __ no: In what ways were these cases different? 2.1.1:» Have you aade any changes in amounts or kinds of products or enterprises in the a; three years? who: Is the fact that you haven't node aw changes a result of any aarket news you’ve been getting? __ no (SKIP ‘1'0 22) __ Yes: What information was that? Where did you get it? (SKIP !O 21.6) ____ Yes: that changes. (ASK ’03 mg) were these? (can smalnc id that 111- Did it in- Did it in- mm 01' EACH reuse or de- crease or crease or cum. , INCLUD- crease your decrease decrease 1116 mm investment? flexibility? costs? 'ADD','DBOP', 0311018 more same less more same less more same less 21.5 Did you asks an of these changes on the basis. of on market information you had been getting? ' .... 1° __ Yes: - uich ones? What infornation there did you no]: nos.) was it? t it? 21.6 Is this the way you usually figure out whether to rake changes in enterprises? __ Yes __ Do: In what ways were these cases different? _ 22. Does it bother you to nuke nistahes? .. lo . ___Yes: Howdoyougot out offeelingbadlyabout then? (I! nor murmur. ssxz) 22.1 22.2 23. «260- Do you ever tell yourself thst experts nuke sistokes toe? __ lo __ I'es: Vonld you sq that st such times so you've .de o. mistake and remembered that the experts ere often wrong too, that you hsd marketing experts in mind? __ Yes: Would you give me u instance? 3o: ibywss that? __ When things go wrong do you ever "blow off stesm' in order to feel better? __ Ho __ Yes: Do you ever find when "blowing off stone" that you pick on things which sre not really those which are bothering you? ____]l’o (m ALSO IN QUESTION 22: mm SKIP TO QUESTION 23) Yes Hsve you ever conpluined shout msrhst infornstion because of other things going wrong? ~ o __ You Could you give me on sample? _ low I'd like to find out some things shout whet news and how you urket. here are e number of msJor problems thst s former hes to settle in marketing. We would like to find out «nothing about how you handled them the lust tine you sold (mermmmmcr. HIT 1811131031608, OBI]I I! 18 mmmmc um 03 AOOMINUING BASIS, smear m P309110! IHIOH 18 nor W common: 21m PROVIDE m HIGHS! 8m 01' 1100)!) Did you sell your lest (crop or lot) of (@5200!) all st once to one dealerer didyouhreokitnpforflg? anon 1201’ urns 10 non: mm on mo: mm) __ m st once: use use that? (on? common 28) __ Broke it up: Uhst wus the period during which you sold it? (ASK qmsrxons at, 26, end 28: am qwrzon 25 end 27) it the tine you were getting resh to sell, were you getting on (rm. 1]? mm: mommy is cum I! QWI'IOHS 5.1, W Immunos on . ($921101er lo: (new! (3an m on mm mm in mm momma!) lhy not? Is this ususlly the ones when you hsve (mg to sell? __ lo: (d8! nsmmm 1'0 8mm MOS! mom TYPICAL an: an US] “18 II FOLLOWING qmsnons. QUINCY 23 HUS! Bl M.) __ Yes: Vould you like to he able to get infomtion to help you with such sales? .. so __ Yes: that would it be? In what particular ways would such in- formation help you? __ (DWPRODOU! IS USED, BHPIOIYOQUISYIOI , 301' 0M1! 30.1 and 31:13 MAJOR P301300! I! M USED, SKIP YO QUESTIOI 33.) __ Yes: What infornstion were you getting? here were you getting it? ‘Do you usually get such information when you have (mum) to sell? .... Y“ lo (1! quarter 23 “moms rm one: o; m m m pg, an? M QUISTION 25. ) 25. Could you have sold your (m1!) either eulier or later than you did? __ Yes: hen why did you pick Just that time to sell? Did your whet informtion help you in coming to this decision? __ lo: Why not? __ Yes: In what way? __ Is itusuallyuseful in this w? .. ro- ___1lo: Bowwas it more useful this tine than it usually is? ___lo: thywas it thatyeucouldonly sell then? Didyoufoel this wsyat all as aresult of the market informtion you were getting? lo : Yes: not was that infornation? _ there did you get it? Is itusuallyuseful inthis way? __ Yes ___lo: nowwas it norouseful this tine tun it usually is? (man ammnm 11‘ QMTIOI’ 25 w 8W) 26. 2?. couldyouhavo soldyour D Y inanyother fora? (mmumm __ Yes: my did you decide not to? Did the market information you were getting help you in this decision? __ lo: Ily not? __ Yes: hat was that information? lhero did you get it? Is this the way you usually handle this kind of problem? __ Do: that do you usually do? he: by couldn't you? ~ (”0an nmmmxcinsucxormmor WI?” 03 my” PHYSIGAL IAGIDIYIE roe ammo) Did you feel this n at all as a result of the aarkst informtion you were getting? __ No: Why not? __ Yes: hat was that infer-tion? Ihore did you get it? Dey;uusuallydeoideaboutwhattodewithit thisww? es ____so: Ihatdoyouusuallyde? (mqugsrlosmmrnonoronmussmmnmroon new You mentioned that you sold your (mg; to one dealer inmsalg, Didyouhave to sell it this W? __ Yes: by was that? Didyoufeel thiswayatsllasaresultof thenarket inforntion you were getting? __ lo: lhy not? __ You use information was that? _ lhere did you get it? Do you usually have this kind of inforntion? Yes : Io: that position are you usually in with regard to infomtion? lo: How else night you have sold it? Ihydidyoudeoide to sell itthewaythstyoudid? Did the market informtion you were getting help you in deciding to handle it this way? __ no: Ihy not? __ Yes: that was the inforsation? then did you got it? Do you usually have this kind of inforastisu? __ Yes __ lo: that poition are you usually in with regard to informtien? 29. 29.1 29.2 29.3 .263. (USIQUISHOIZBOEYHGBDPORM'ASSPLIYUPNRW) You mentioned that you didn't «11 all of your 1." crop (or lot) ofwallst one time. Howdidyouget ridef it? mydidyoudecide todeit thisway? Did the market information you were getting help you in deciding to handle it this my? __ lo: W not? __ Yes: mo inforntion was that? How did you get it? Is this thowsyyouusuallyhandle thoproblenofhewmch ofs crop (or lot) to sell at any one tine? __ Yes __ Is: How is it different free usual? (IISOLDmuOMEO (IYSPLIYUPJ'ORBALI) 0!! mm) Iherodidyousellyour Iheredidyousellthelargest‘part AMY) ? of your last crop (or lot) of M Y Dealer's name hown in which the business is located lumber of niles from fare: to dealer's business Do you usually sell your (MI) there? __ Yes: that are your reasons for selling there? (II‘GOOD PRIGD'IS a MOD) Howdidyouknowtbtyou'ddoaswell thereasswwhere else? . lo: M did you choose this place? (I! |'CiOOD mos" 18 A reason) Bowdidyounow that you'ddesswoll therossonywhere else? Is thisthowyounsuauydecidewheretosollyour M1 1 ___Yes ____lo: Bewdoyouusuallydeoidewheretosell?___ Bowmnyyecsagodid youfirst sell Monm- dealer? Years Bowmanyyearshashohesainthis area? Years -26'l- 29.” How long have you been producing (PRODUCE } in this area? Years 30. How was the price set for the sale to (menu AID Locum 1'30?! QIIISIIOI’ 29)?_ lss the price set when you delivered your {mm or was it set earlier? __ 8st at delivery __ Bet earlier: Did you have a written contract or did you use a verbal ages-ent? __ Iritten __ Verbal 30.1 mu you feel that the price was right when you speed to it? ___Yes: wudynfed that way? _ lo: is any of the aarket inforaation you were getting responsible for how you felt? * 'o __ Yes: What was the inforsation? Ibere did you get it? (soar am Iron 'Set at delivery') Did you feel that the ice was right when you delivered DUO! ? ___Yes: Vhydidyoufoel that way? Io: Use any of the market information you were getting responsible for how you felt? lo __ Yes: lhat was the information? here did you get it? 31. Did you follow the prices of (M!) for while after you sold it? __ lo: Do you usually check prices after you sell? ___lo: shy not? Yes: thy didn't you check prices this tine? __ Yes: How? Did it go up. down, or stay the same? __ Vent up __Stoyed the ease _lent down How did you feel about the sale later? __ Satisfied __Dissatisfied: Ihst did you feel like doing? Doyouthinkyoumightlike to sellyour PRO 1' somewhere else next time? Do you usually check prices after you sell? ___Yos: my? - :ls: W not? thy did you check this tine? __ 32. 32. (IL 32.02 32.11 -265- (n consumes 23.31 am: 193, mm: co one run: moor, our no 91335103 33) Now getting back to the dealer, Do you ever stop in at this dealer's place of business Just to chat with him even if you have no business to do there? No no, but no regular dealer Yes, but no regular dealer Yes: About how often would you say that you drop in? that-kinds of things deyouusually talksbout during these visits? a low, is there another nan connected with this dealer -not the seas onewithwhonyouarrongs the sale—who cones outhere frequently (once a south, once a week, or more often) to pick up or check with you on your (PRODUCT 130M Expo! 23) ? __ lo (11' "ushmnsumrmrnmsmnnmummmm 1.01am commons mm to an own. mama, on SOMEONE 11: ion- uomrr In on PLACE or names. monomers in no as M meomnnnmm. H 'm'. mm com, mom conscious 32.21—32.28) Do you and your dealer,or any aenhers of each of your foailios. ever visit one another at hone? Yes - regularly Yes - soaetines lever Doesn't new dealer at all (our no consume 32.21 m 11 mm 20 qunsnon 32.02 mm 16 Ino', 531? no qmsuon 33) 32.12 Do you know where your dealer's hone is? 32.13 32.11» __ lo __ Yes: there is it located? How many silos from here is it? "w Are there any other things like bowling, card playing, or hunting and fishing that you ever do with your defler? __ Io ___ Yes: Ihst are sons of these things? Do you and your dealer, or usabers of your immediate faailios, attend the sane church? __ Sane church Different churches One does not attend church leither attends church Doesn't knew about the dealer 32.15 32.16 32.17 32.18 32.21 32.22 -266- ' Did you know your dealer personally before you started doing bus- iness with his? __ Io __ Yes: How many years age did you first meet him? mt were the circumstances under which you first met ? Are there any tines when you see your dealer socially outside of his place of business? __ No __ Yes: Would you say that you see him this way fairly regularly or only occasionally? .... mum, ______ Occasionally On what kinds of occasions do you get together socially with him? how long have you known his socially? Years Do youusuallycellyour dealerhyhis first sans, last name, or a nichans? Iirst name is compared with how you feel about your closer friends, would you call this an a close friend, friend, acquaintance, or comparative stranger? __ Close friend __ J'riend __ Acquaintance __ Comparative stranger __ Don' t know (I, ANSWER 10 32.02 IS “'0', SKIP 1‘0 33) low Just a few questions about the man who cones out about your (PRODUCT). Do you and he, or on members of each of your families, over visit one another at home? __ Yos - regularly Yes - soaetimes lever "" ___Doesn't know hit: at all (am so qansnos 33) Do you know where his heme is? __lo :Yes: there is it located? Bow new miles from here is it? .257. 32.23 Are there any other things like howling. card playing. or Imnting 32.21? 32.25 32.26 32.27 32.28 and fishing that you ever do with his? .... to ___Yes: What are some of these things? A Do you and he, or members of your imediste families, attend the same church? Same church Different crunches Neither attends church One of the two does not attend church Doesn't know about the dealer you know his personally before you started doing business with IEEIIIII __ Yes: How saw years ago did you first meet his? Years Are there any times when you see him socially outside of his visits here? __ Yes: Would you any that you see him this woy fairly regularly, or only occasionally? __ Regularly __ Occasionally On what kinds of occasions do you got together socially with him? How long have you known him socially? Yeara Do you usually address hinhy his first name. last name. or s e aixture of these Doesn't know his name is «spared with how you feel about your closer friends. would you call this man a close friend. friend, acquaintance. or a compara- tive stranger? __ Olose friend Iriend __ Acquaintaneo __ Comparative stranger __ Don't know 33. Did you use an hired labor in running your form last year? ___lo (SKIP i'O QUETIOY 3’!) :Yes: Did they work for you year round or part time? __ Year round: now mm were there? __ Part time: Bow many were there? On the averagehowmsndes didtheavo- rage pert-tine worker work for you? on that was the total number of days worked for you by part-time workers? Do you ever visit with. or socialise in any other fly with people who do farm work for you? __ Yes __ lo __ Depends Do youusuallyhsve the semehirodholpyear after year, or are there usually different ones? __ Same ___Different __Somo same, some different Do you —-usually hire local people or migrants? ___Usually local __ About half and half ____ Usually migrants ___Usually migrants , but prefer local people when I can get them Other (SPIOIJI'Y) Do you ...—usually address your regular hired help by their first or lest names? __ Iirst name __ last names ___licknames :Some of each When you have hired help working here. are they usually allowedtouse thehousefor thingslikewashiagand eating? __ Allowed to use the house ___Depends on who it is and the circumstances __ Don't use the house _k 3". Did you attend meetings of Do you belong to new “210! 1 lo Yes 50mm no You? during the past year? . 12mm ? l‘erm hrem the Mn Daren. the Orange the Orange the Yarmer's Union the Dormer's Onion local oo-ops an local co-gps; I -269- 3151 :14 you participate 1! F05?” res lo. (ommwmr) lost your? nigihlo let mum. Soil Conservation Service County Agent‘s programs 10.3. 1.5. 3t.2 Did you do any of these things lost year? no Yes Hove home or office visits with the county agent Attend demonstrations and field days Attend county agent's Eating- Make use of hp. Sta. or ht. Sort. publiootions Gone in to MSU for information or advice Corry crop and] or livestock insuronce Figure out costs and returns for each sore and/or Balm 310.3 Do you beep s set of books on your expenses and sales? No __ lo, but keeps hills, receipts, checks __ Yes: Do you keep your records on everything that you prohce ell together, or for each product or enterprise sepnrute- ly, ordoyouonlykeop thenfor theone that'syour njor source of income? __ m records together __ Separate set of records for each product __ Records only for the mJor source of income __ Eotel records and o seporste record for main product (11' manor 23.31 mo g9; mm 20 um mm mm in motion 32 us ogrm, moprsnon 35) (n qursrxous 23.31 mm co m & moor m qunsuor 32 m ASKED, on! qmsnon 35 up 31:12 no quarto: 36) _ Now we'd lilne to get some information on how you nerket (ms PRODUCT non @229! g} . these questions that follow ash: ehout a dealer. and whet we mean is the sun with when you erronge to sell your product. .270- 35.01 could you pleue give no the nuns and the place of business of the dealer to whom you sell this product? Dealer's none Town in which the business is located No. of miles from farm to desler's business 35.02 Do you usually sell this product to the some deoler? __ Yes \ __ to 35.03 o) How mum yours ego did you first sell this product to the dealer? yrs. b) How nsnyyeers ho he been in business in this ores? __yrs. c) How long have you been producing this product in this tree? ’1"- 35.0¢t Do you usually hove e. written contract with this dealer or do you use o. verbal agreement? __ Written contract __ Verbal agreement 35.05 Do you ever stop in at your dealer's place of business Just to chat with him even if you have no business to do there? __ No A __ No, but no regular dealer __ Yes, but no regular dealer __ too: About how often would you say that you drop in? that kinds of things do you usually talk shout during these visits? 35.06 How, is there another nun connected with this dealer who cones out here at least once u nonth to pick up or check: with you on your __ much manor mom QUESTION 2) ? ...!“ (11' 'ns', mm: sun mu m msrom mums __ No rm m FOLLOVIEG QUESTIONS mun no in owns, Nil- m, 03. 80mm 11! murmur! m on rust or 3081-- NESS. Ammomms ARE To BE TREATED LIKE m Dim mum. 1mm, mu no comm, m 0141'! QUISIIOBB 35.21 - 35.2“) - _ 35.11 Do you and you: dealer. or any nembers of eooh of your families, over visit one another st home? Yes - regularly res - sonetines lever : Doesn't know dealer et en. (3m to qussnou 35.21 % 11 35.06 18 “150'. 8m 10 3 35.12 35.13 35.15 35.15 35.16 35.17 35.18 .271... Do you know where your dealer's home is? Io __ Yes: We is it located? _ How many miles from here is it? __. Are there any other things like bowling, card playing, or hunting and fishing that you ever do with your dealer? __ Do __ Yes: What are some of these things? m Do you and your dealer, or members of your immediate families, attend the cane church? __ The same church __ Different churches __ One of the two does not attend church __ Neither attends church __ Doesn't know about the dealer Did you know your dealer personally before you started doing business with him? __ Io . __ Yes: How many years ago did you first meet him? lhat were the circumstances under which you first met him? i Are there any times now when you see your dealer socially outside of his place of business? __ Io __ fee: would you say that you see him this way fairly regularly or only occasionally? ...... Regularly __ Occasionally On what kinds of occasions do you get together socially with him? How long have you known him socially? Do you usually call your dealer by his first name. lost none or a. nickname? __ l‘irst name ___Last name ___Iickname ___A mixture of those ‘ is compared with how you feel about your closer friends, would you call this man a. close friend, friend, acquaintance. or comparative stranger? __ Close friend __ Friend __ Acquaintance __ comparative stranger F's 35.21 35.22 35.23 asgzu 35.25 35.26 35.27 .272. (I! 35.06 18 'no', our no 36) Now Just a few questions about the man who comes out about your 1mm: PRODUCT Do you and he. or any members of each of your families, over vis- it one another at hone? Yes - regularly Yes - sometimes lover Doesn't know him at all (SKIP Io qrntsnon 36) 5‘ you know where his home is? No __ fee: fibers is it located? How many miles from here is it? A Are there any other things like bowling, card playing, or hunting and fishing that you ever do with him? ... x. __ Yes: What are some of these things? Do you and he, or members of your immediate families, attend the same church? __ the same church __ Different churches ___Oneofthe twodoes notattendchurch __ Neither attends church __ Doesn't know about the dealer Did you know him personally before you started doing business with him? __ lo __ Yes: How many years ago did you first meet him? __ Are there any times now when you see him socially outside of his visits here? , lo : Yes: Would you say that you see him this way fairly regularly or only occasionally? .... Bosnian: __ Occasionally On what kinds of occasions do you get together socially with his? ‘ ' how long have you known him socially? ___yrs. Do youusually address hinby his first none, last name. or a __ nickname __ A mixture of those -273- 35.28 is compared with how you feel about your closer friends, would you 36. 36.1 36.2 35.3 36.“ 35.5 36.6 36.? call this an a close friend, friend, acquaintance, or a compart- tive stranger? __ Olose friend __ Friend . __ Acquaintanco __ Oompuetive strnger I'd like to ask you .... questions about the things you yourself do with your neighbors. I mean, the three or four families who live closest to you. Do most of their farm? __ Yes __ Do: (ID 'NO', SKIP TO QUESTION 36.11.) Do you lend farm equipment or supplies to each other? __ No __ test how often? __ Often __ Sometimes Do you help each other out with getting work done? __ No __ Yes: Do you Just do this whenever an of you needs some help, or is it something you keep tabs on so that everyone gets about as much help as he gives? __ Work exchange informal __ Work exchange equalised mmumhcnleuhotherbyfmtnaustchms, orby last names? __ lirst names or nicknames Last names __ Doesn't know names Is there borrowing of household supplies and items among those families? __ lo __ Yes Do you sometimes make arrangements to go in to town together rather, than always go in alone? __ Ho . __ Yes . Do youask each other todo errands'in townwhenyouneedtohave something done and can't go in yourself? ...—3° 36.8 36.9 37. 37.1 37.2 37.3 471!» Do you or members of your families ever visit in each others' homes for a whole afternoon or evening together? lo __ Yes Do you or members of your families ever go on outings or picnics or go out hunting or fishing together? lo __ Isa low I'd like to find out a few things about the local area. Ihat do you call this place around here? (114118er qmsrlos37.1 Ismrarmwaisrmorm 01.0325! rows ma) Do you sometimes call it (M or 010%! Tom m): __ Isa __ lo: m is that? _ A (11' A3013 mm 18 m0! mom mm 18 rqumsmr no m on MORE ram) ‘ . Do you have a preference for one of these? __ lo __ Yes: Which? (as: on! 11* A mm am: Is GIVEN IN 37.1) How much of an area does (m GIVEN 13 411852101? 37.1) take in? (use mus Am I! mrmommun. moo museum Ame, IBM, m mmsmxonm'mrmmrom, 11' m, 1(1an rim. mmm mam non rm ro row.) , 37.“ (AB! on! 11' A FLAG]: m Dmnmr 130M 37d 13 cum I! 37.2) How much of an area does (1mm GIVEN I! QUISTIOI 37.2) take in? (us: mic imp In councilman. moo sures Insom- nns, mm mu. m on memos or m rm m eons, n m, 111ml; mu. mxcm msnnos mow rm so nous.) 37.5 38. 38.1 38.11 38.12 38.13 -275- (11 mm 1'0 37.2 18 'ES', 231?! QUISTIOHB 37.5 to 33.3 mm 1'0 mm: m III 37.2) . (11' mm 20 37.2 18 "no" no A sum 18 cum I! 37.1. rm qussrzons 37.5 to 38.1 ms no mm mm mm In 37.1 m cannons 38.2 and 38.3 m scum s! m nous morons 38.2) How long have you lived around here? l'ears __ Lived here entire life (RECORD AGE, M '1') Now we 'd like to find out a little about how you feel about the commnity. a. What things do you like about b. [hat things do you dislike living around here? about living around here? (SKIP r0 QUESTION 38.12 IF mam r0 QUESTION 37.5 Immune mm mm: mm 1.11!) How does this place around here compare with other places you've lived in? , __ This place evaluated better __ Ms place evaluated as same __ this place evaluated as worse m. place evaluated as better in some WI. worse in others (spasm) # How does it compare with other places you've seen or visited? __ It's the best It's better than most It's about the same as most It's poorer than most It's the worst It's better in some ways, worse in others (SPICE!) Do you take pride in living around here? (11' WEAR!) l'or example, do you have a feeling of pride when you tell a stranger where you're from? Yes __ D‘s (usm Locum m non unusual 37.2 ms muons 38.2 and 38.3. 1r mums 37.2 18 Inc! BECAUSE mm 18 mm-msmr m m canon rows, em: on macsomnmmwcm 38.3rormmmmm. nmmrommmsnormmos in All nmmm 1mm 18 cam 13 ensues 37.1, can mm mm. nxomncmnmumm Isms ms, mosses: um. moms.) -276- 38.2 Here is a list of a number of things which you may or may not do. Ifyoudothem, I'dclike toknowifyoumdothesethingsin (10W mun.) usuallzdo them there. only onetsig dothoafirmdocthomin W (I! 'don't do it at all“ one: no mo: 81m m! m noun! 18 MI]! W) Use banking services Don't n- Usu- sm- Four Go to the movie! 3w fern machinery Tarn machinery repair done 31w farm supplies Vi sit relatives Buy hardware supplies av clothing Buy gasoline Oar repairs done Attend sports events Attend church (ll' WWW um.) Could you attend church in Ihat is, is there a church there that would be of your choice? __ Yes __ No __ Don't how 1: you wanted to? __ All nest About half Just a few ... 10M 38.3 that proportion of your close friends live in the community? 40. lbl. “2.1 142.2 42.3 #2.“ .277. Is this the only fern you've operated for yourself? __ Yes: How many years have you run this place? Years _ No: How many years have you operated ferns for yourself? Years ' ' How many years have you run this place? Years We'd appreciate knowing who else lives with you and what their approxiaate ages and education are. Relation to Age Last grade Sex Itarital respondent completed status in school Respondent (DON'T ASK) __ M __ I __ __ It __ I M 1' About how much of all your assets are invested in the farm? ____ Less than l/lb ____ Between 1/2 and 3/4 __ About l/o __ About 3/1' __ Between 1/14 and 1/2 ____ More than 3/1. __ About l/z __ All About what proportion of your total assets invested in the farm are tied up in things that you couldn't sell for close to their value in ten days? __ Less than l/lt __ Betveen 1/2 and 3”" __ About l/t __ About 3/" __ Betwaen 1/4 and 1/2 __ More than BI“ __ About 1/2 __ All would you say your liabilities compare with your assets? __ Ho liabilities Liabilities less than 1/u of assets Liabilities about l/llv of assets Liabilities between l/h and 1/2 of assets Liabilities about 1/2 of assets Liabilities between 1/2 and 3/lr of assets Liabilities about 3/11» of assets Liabilities between 3]“ and all of assets Liabilities equal to assets __ Liabilities greater than assets 5' that would you say you- average annual gross farm income has been for the past three years? .278— Vell, that's all of the questions. Thank you very much for your time and cooperation. Date fl County w w Earner A m Address Seg. No. Interviewer l'arm No. M type MA score -2 79.. ~280- RESEARCH DESIGN IN STRUCTURAL—FUNCTIONAL.AKALYSIS A CASE STUDY FEOM.APPLIED RESEARCH OH.A COMMUEIGATICH SYSTEM* Joel Smith, Robert C. Bealer, and Francis M. Sim Michigan State University Introduction The writers of this paper have been concerned for some time with developing and testing theories of communication, in general, and mass communication, in particular. They were, therefore, favorably disposed to accept an opportunity that arose to engage in an applied research project involving a particular communication system being operated for program purposes. Specifically, the problem concerned the uses made of information about agricultural marketing being disseminated by the Agricultural Marketing Service of the United States Department of Ag- riculture. This seemed to provide an opportunity for a structural- functionel analysis of a specific communication system since the programs of large-scale formal bureaucracies like government agencies are generally embodied in structures Specifically designed for these purposes and since the question of use could easily be seen as a question of function. We are concerned here with considering certain difficulties in- volved in doing any structural—functional analysis and in doing such an analysis in the kind of applied research situation in which we are * Journal paper No. _ of the Michigan Agricultural Experiment Station, East Lansing, Enchigan, Project No. 69. The project is also supported by the United States Department of Agriculture through con- tract Number 12-25—010-94. Read at the annual Chic Valley Sociological Society Meetings, Columbus, Ohio, April, 1957. .281... .... I‘ 1 involved. We intend to indicate what those problems are and what sorts of adjustments had to be made in our research design in response to these problems. When we talk of problems in research, it should be understood that reference is to the very mundane problems of empirical research operations and not the more general methodological problems of structural-functional analysis that have concerned such people as Par- sons, Levy, and Bredemeier.2 The kinds of materials and eXperiences to be reported here should help to shed light on the general question raised by Merton as to the implications of research conditions for the outcomes of structural-functional analyses. The research design will not be discussed in detail but will be referred to only to illustrate the sorts of adjustments between the ideal and the reality that had to be made in a response to the sorts of problems experienced. Qperational Problems Arising in.Any Structural-Functional Analvsig In accord with Levy's usage of the terms, three types of struc- tural-functional analysis may be distinguished: (1) functional analysis in which the concern is with consequences attributable to the existence and Operation of a unit under analysis; (2) structural anal— ysis which is concerned with establishing propositions about the types of structures capable of serving specified functions; and (3) structur- al-functional analysis in which there is simultaneous concern with both types of questions. Our interest is in establishing the consequences of a particular social arrangement for some larger social arrangement of which it is a part and/or for other social arrangements, also includ- ed within this larger social arrangement, which can stand in the required causal relation with the unit under analysis. As we have defined our terms, this type of question calls for functional analysis. However, for purposes of improved communication it shall be referred to as structural-functional analysis in order to be consistent with current usage among sociologists. Structural-functional analysis, as defined here, raises a number of difficult problems in the designing of research to deal with a spe- cific empirical case. Some of these problems revolve around definition. some around models 92 proof which may be utilized, and some around practical considerations concerning the social environment of eXperi— mental research. These problems will be discussed in this order. There are at least three definitional problems: defining the unit which serves functions; defining units for which these functions may be served; and defining the units to be referred to in distinguishing latent and manifest functions in a given empirical case. The adopted definition refers to a unit with identifiable structure which has con- sequences for other units of identifiable structure. Accordingly, the question arises as to the kinds of units whose relationships may be in- vestigated with a structural-functional approach. Are they to be de- fined as units by the groups acting within them? Are there such criteria as complexity of the structure of the unit, number of personnel participating in the unit, or degree of independence between the pOpu- lations involved in each unit that need to be considered in deciding whether a given empirical problem is an appropriate vehicle for struc- tural-functional analysis? There seems to be little consideration of this problem of defining units in the sociological literature on the .233- structural-functional approach. However, even if an analyst may look at any units that derive from his theoretical framework, the problem I; I of defining the limits of these units remains. Formal criteria for distinguishing between manifest and latent functions have never been specified. While the twin criteria of in- tention and ree0gnition have been suggested for this purpose, an important definitional problem remains in specific empirical research.6 Whose intentions and whose recognitions are crucial in assessing the status of a specific function? How is time to be handled? Granting that the analyst knows to whom he should look in determining intention and recognition, he must still decide the point in time at which these should be determined. This is an especially difficult matter in a structural-functional analysis conducted under conditions of applied research where the client, with a program, wanting to know what he is actually doing, is in rather constant contact with the investigator. Thus, the client is informing himself of the investigator's ideas and findings at every step of the way. Under these conditions, the client's recognition of the functions of his program are subject to change through time. Less likely, but quite possible, is a related change in intent. An investigator must select a time at which he will attempt to determine the intentions and recognitions of the group or person which he defines as crucial for this purpose. As regards models of proof, the research design of the controlled projected experiment is usually deemed to be the most powerful investi- gatory tool available for establishing causal relationships with high degrees of probability among empirical units. If one grants the possibility of functional alternatives, equivalents, or substitutes- and the preponents of the structural-functional approach seem com- pletely in accord on this point-it is extremely difficult, if not ime possible, to determine the functions of existing units experimentally. The extreme degree of urgency attached to most functions suggests that, in most cases, the removal of a structure hypothesized as serving a particular function would result either in compensatory adjustments in other structures so that the function might still be served, or in the direct taking-over of the function by other units with no changes in their structures for this purpose. Attempting to overcome this diffi- culty by removing all units with structures capable of serving the function simply confounds the experiment, since it would be impossible to know which of the many units removed was reSponsible for the function if it appeared that the function was no longer being served. Of course, it is extremely unlikely that our knowledge of social structure is ade- quate for indicating what all of the possible "alternatives" are. While it is conceivable that a very complex experimental design might be evolved to get around the difficulties posed by the possibility of functional alternatives, it is very unlikely that such a design might be translated into an effective set of research operations. At the present time a direct experimental test of structural-functional hy- potheses is not feasible and less powerful methods of hypothesis evaluation need to be substituted. If the difficulties in using experimental model of proof did not exist, there would still be few opportunities for testing structural- functional hypotheses eXperimentally because of the intrusion of other -285- practical considerations. The concept of manifest function often re- flects positive evaluation for at least some of the participants. Therefore, an experiment necessitating the removal of a structure so as to evaluate its functions would probably meet strong resistance on the part of persons involved in the units that might possibly be affected. The experimenter is planning to destroy a "good", not do away with an "evil." This would be a source of social resistance to social eXperi- ments that goes beyond objections based on the general notion that sociological experiments are immoral and unethical because an experi- menter (one person) manipulates other persons. Difficulties in Executing Structural—Functional Analyses in Applied Reseggch Settingg’ The above discussion was limited to those problems which arise in doing any empirical research from a structural-functional approach, and of which we are most acutely aware by reason of our experiences. Now we should like to consider some other difficulties which we experienced and which we attribute to the fact that our attempt at structural- functional analysis has an applied research setting. These difficul- ties are not merely a by-product of the particular client or his specxfl. problem, but will occur frequently when an applied research problem is used as a vehicle for executing a structural-functional analysis. How- ever, as these problems are stated, they shall be illustrated by examples drawn from this project. The client for applied research frequently tends to see his prob- lems as specific and limited in scope. One of the strictures about—- and claimed advantages of—-structural-functional analysis is that the -286- analysis should be complete. While we would hesitate to argue that an investigator must pursue his analyses of the functions of a particular unit to the point at which he can give a definitive enumeration of all the functions of a unit, we do interpret this to mean that an investi- gator should be able to pursue the identification of both latent and manifest functions as far as he wishes. Such broad research pursuits are usually discouraged by clients interested only in the extent to which certain chosen functions are being served by the structures of the units with which they are concerned. If the research does not produce evidence that the specified functions are being served ade- quately, the client is usually uninterested in pursuing the question of whether any other functions are being served by the unit. In our particular case, the client's representatives are concerned largely with whether or not there is what may be considered an economic gain for persons who consume market information. Attempts to place this type of consequence in the perspective of other types of consequences have usually received no support. The possibilities of executing a "com, plete" structural-functional analysis are very severely limited by what Read and Bain has noted as a frequent inability of peOple with "prac- tical" concerns to see research—worthy problems.9 The applied problem is often couched in such a way that the inves- tigator has no choice as to his units of observation. These client- determined units are often not the ones with which a structural—func- tional analysis might most apprOpriately be concerned. Uhile the investigator may be interested in individuals as the units in which consequences of a given unit's structure might appear, he is almost -237- certainly interested in such units as institutions, organizations, in- formal groups and patterned systems. In our research most of the interests of the client forces the use of individual farmers as units of observation. By aggregating data collected from individuals, it might be possible to get some approximate ideas of the functions of market information for other units of interest-~but these other units can only be those in which farmers may be involved. Moreover, for those units in which farmers are involved, we can only get a notion of those functions in which farmers are involved. Thus, data collected from farmers yields no information about the functions of market in- formation for the exchange relationships between dealers in agricul- tural products and the people to whom they sell. However, something can be learned about the exchange relationship for agricultural commodities as it exists between farmers and dealers, though even for this relationship the unit of observation is too limited to yield a run. picture of functions. The function of market information for the farm- er‘s formulation of his price expectation can be explored but the function of the same information in the dealer's formulation of his offered price cannot be determined. Therefore, in as much as an applied investigation forces certain limited units of observation on the investigator, an incomplete and systematically biased picture of the functions of the structural unit being investigated will result. The client of applied research does more than define the problem and force units of observation on the structural-functional analyst. He also forces the investigator to adapt certain techniques of data -288- collection. After all, he is, more than likely, a man who sees himself beset with critical problems and, accordingly, he is eXperiencing un- certainty as something unpleasant. Therefore, he may well insist that the investigator utilize techniques which will bring him the kind of answer about which he can feel certain. Our client is concerned that we apply techniques which yield "quantifiable" data. We place quanti— fiable in quotation marks because it is probably not the apprOpriate concept for describing the data he desires. Terminology aside, he prefers interview questions which may be answered by counting or meas- uring to the kind of question that yields a detailed response which must be interpreted by thematic analysis. Regardless of the client's reasons for preferring particular "quantitative" techniques (though it appears to be a reliance on what he spuriously conceives as quantitative as a reliable way to increase his feelings of certainty in a situation of uncertainty), the net effect is often to make the data unrevealing for purposes of structural-functional analysis, even if the units of observation are appropriate for this purpose. Finally, the research client rarely sees the desirability of supporting an inquiry into latent functions. These lack significance and meaning for him and this is to be expected. If such functions were significant and meaningful, they would very likely have been manifest functions for the unit. We have engaged in long controversies with our client in efforts to retain items in the interview schedule that might indicate some of the latent functions of the market information system. The controversies are usually lost when the issue comes to a head. If one eventually wins such a controversy after a series of such losses, he begins to wonder whether the function at issue is not, in fact, - manifest. Thus, the situation of the structural-functional analyst engaged in applied research, biases the outcome of his work in the direction of certain functions, which are most likely manifest, and which are limited both by the kinds of units of observation and by the kinds of techniques of observation that will be allowed to him. Apparent Advantages of Applied Research Settings for Structural- Functional Analysis We should like to assess now some of the apparent advantages of the applied research setting for structural-functional analysis in the light of our experiences. It will be assumed, of course, for purposes of this discussion that the difficulties just considered either do not exist or can be solved.10 Our evaluation of these advantages is based completely on our own case at this point. We do not know whether our impressions would be borne out by other experiences. Anticipating the discussion, to clarify this point, we will indicate what appear to be certain advantages of the applied research situation for conducting a structural—functional analysis, but conclude that these advantages may be illusory. This conclusion will be based on problems particular to our own study which may be so idiosyncratic that they have led to an unwarranted conclusion about the apparent advantages of the situation. What are the apparent advantages of an applied research setting for a structural-functional analysis? Briefly it would appear that it is an attractive situation both for minimizing the problems of defining the limits and the structure of the unit whose functions are being an- alyzed and for determining the intentions and recognition necessary for -290- distinguishing latent and manifest functions. The typical client for applied research is a formal organization with a program and set of goals. They know what they are and what they want and resort to re- search because they sense some problems regarding their effectiveness. Thus, the unit whose functions are to be analyzed is predefined and its structure readily accessible for examination. Moreover, because a structure had to be designed to implement the ends of the organi- zation, the intentions of the unit are likely to be clear and recog- nizable. For these reasons it might be concluded that the important definitional problems just mentioned would be minimized. While this still seems to be an eminently reasonable conclusion, our experiences with this study force us to question it. At the time that the contract was being negotiated we were per- fectly willing to talk to the clients in terms of research concerning farmers' use of market information. Both our common sense and their informal examples of things they were concerned with convinced us that we could easily formulate meanings for such terms as "use" and "market information" once the project was underway. To our surprise and dis- comfort, this was not so. We managed to get some consensus that the information was to be of use in the sense that it would help a farmer ‘ to improve his status. Beyond this we could get no consensus. The pui- gram is Operated to help farmers improve their status through more effective marketing but the concept of marketing is completely unclear. Improvements may be reflected in such diverse activities as formulating a precise estimate of what price to require, on the one hand, and .291- decisions about commodities to be produced, on the other. Moreover, no agreement as to the referents of status could be elicited from reaponsi- ble program officials. ‘Uhile individuals were willing to give their own interpretations, they were not motivated to resolve their differ- ences and achieve consensus. The Operators of the program do not share precisely defined intentions. Perhaps the more amazing experience came from our attempts to identify the unit whose structure is presumably geared to execute these ill-defined functions. There is, as far as we are able to determine, no clearly distinguishable unit structured to implement this program. There are units charged with collecting and disseminating such inform- ation as current prices of specific commodities at specific markets, supply information on specific commodities, price outlooks, and so on. But these supply only a part of what is considered market information. There are many other categories of information that have shifting status and are considered to be market information only if they are brought to bear on the marketing process; gggg, weather reports, pro- duction technique information. In interviewing various representatives of the client we could find no informational program of the entire Department that might not under some circumstances be defined as market information. Moreover, since the personnel involved consider it their task to encourage non—governmental units to collect and disseminate such information, the structure of their prOgram knows no bounds. We have been unable to benefit from the apparent advantages of the applied research situation for structural-functional analysis. Superior hindsight brought to beer as a consequence of these ex- periences enables us to see some general reasons for the illusory character of the apparent advantages of the applied problem setting. Most studies of formal organization show that extremely significant informal structures develop within the formal structure. The identi- fication of the structure of the formal organization (iggg, the unit whose functions are being investigated) with a problem is probably newmr a simple process. In addition, studies of bureaucratic organizations repeatedly reveal that the intentions of the personnel often diverge from those formally stated for the organization. One very effective technique for making it possible for personnel to maintain their varied intentions without feeling themselves to be unfaithful to or in con— flict with the aims of the organization is to keep the program and aims of the organization vague so that anyone may apply his own interpre- tation. Moreover, vagueness in aims also makes evaluation of perform- ance in terms of effectiveness impossible. Some of these considerationa, all of which are untested, may make even more attractive the hypothesis of the inherent disadvantages of the applied setting for structural- functional analysis. Consequences_for the Research Design While the discussion of structural-functional analysis in general and in the context of applied research indicated a number of difficulthxs in executing research we do not conclude that such analyses should not be undertaken. All executed research involves a series of compromises with some ideal model. Structural-functional analysis in an applied .293- setting is not unique in this respect. Our aim is to profit from our recognition of these difficulties by using our experience to indicate both the kinds of limitations of research findings deriving from such research situations and the sorts of steps that may be taken to cir— cumvent these difficulties. The problems involved in carrying out any structural-functional analysis are, of course, not uniquely experienced by us. Some are definitional and can only be handled by arbitrary definition. We can- not indicate how definitions were established or what they are because of space limitations.11 However, the solution is not as easy as the term "arbitrary" might suggest and can hardly be resolved satisfactor- ily without having some previously developed theoretical models for the unit under analysis. The general difficulties blocking the application of a controlled experimental model do not appear resolvable at this time. Such models cannot be used, at present, for a structural-func- tional analysis of any large-scale social unit. Approximations to this model by correlational approaches seem to offer the best compromise and .that is to be our analytic approach. ‘We have noted a number of difficulties in conducting a "complete" structural-functional analysis in an applied setting. All of these seem to work in the direction of limiting the functions that will be established for the unit under analysis to a small number of manifest functions. To the extent that the pressures in this direction cannot be avoided, the analyst can only recognize these qualifications on his data and try to appreciate them for what they do tell him. The limiting effects of our client's preference for certain units of observation and -2914. observational techniques could not be avoided. we shall try to convert these restrictions to advantages by Specifying our areas of omission in more detail than is usual in the reporting of sociological research. However, some of the restrictions on functions to be investigated with- in those limits set by the units and techniques of observation were circumvented. We were fortunate in having some independent support for our research and so it will be possible to put additional questions to the respondents which may shed light on other functions with which the client is not concerned. This additional effort will be directed main- ly to some possible latent functions of participation in the market information system by farmers. These additional resources could not have been applied to an expansion of our analysis, of course, without first solving the prob- . lems caused by the client's inability to identify either the structure involved in executing his program or the intentions of the program. The problem of structure was resolved by taking advantage of the fact that all of the client's representatives agreed that some identifiable units were a part of the program. Therefore, even though the client's interests will force us to catalogue all information brought to bear in determining a given action, we should be able to ascertain whether the materials disseminated by one of these units-~Market News Services --had a function in the situation. We will, therefore, attempt a structural-functional analysis of a smaller, more discrete unit than that with which the client is concerned. The problem of establishing which functions might be considered latent and which manifest was handled somewhat differently. On the -295- rather simple assumption that a pregram designed to disseminate inform— ation about farm markets could have been meant to affect what a farmer did as he sold his products, it was decided that consequences of the system for all aspects of how a farmer behaved in the various seg- ments of the marketing process would be considered manifest. This decision was somewhat validated by the relative willingness of the client's representatives to suggest some points in the marketing processes of farmers at which they thought effects of market inform— ation could appear. Although they were not willing to assert that the creation of such effects is part of the program's £2332; goals, this degree of consensus was accepted as sufficient. Given a selection of manifest functions, we were then in a position to ask ourselves what other functions we as sociologists could see as possibly emenating from the Operation of a structure like that of Market News Services. The selection was based on the range of possibilities afforded by im- agination and intuition. These still seem to be the major means by which latent functions are apprehended. When this research is concluded we hOpe to know something about some of the functions of Market News Services and their relative impor- tance. If this communication system can be related to a pypg of com- munication system, we may know somewhat more than this. Failing either or both these aims, the experience may at least shed a little light both on the actual experience of undertaking and carrying through such an analysis and on some of the ways in which ideological factors affect the outcomes of structural-functional analyses. ~296- Footnotes This would appear to be very relevant to many sociologists since, although we are unable to document this quantitatively, a considerable portion of the research energy of sociologists is devoted to applied research problems under conditions like those of a government contract. Talcott Parsons, "The Present Position and Prospects of Systematic Theory in Sociology," in Georges Gurvitch and Wilbert Moore (eds.), Twentieth Century SocioloEX, New York: Philos0phical Library, 1945, pp. 42-69; Marion J. Levy, Jr., Egg Structure 9f Societ‘, Princeton, New Jersey: Princeton University Press, 1952; Harry C. Bredemeier, "The Methodology of Functionalism," American Sociological Review, 20 (April, 1955), pp. 172-180. 3 In the eleventh category Of his paradigm for structural-functional analysis Merton specifically raises the question of the impact of client interests on the outcome of such analyses. Robert K. Merton, §Q§i§l Ihgggy gpd Social Structure, Glencoe, Illinois: The Free Press, 1949, p. 54. 4 or. Marion J. Levy, Jr., 92. 213., p. 62. 5 _I_1_a_i_q., pp. 34—38. 6 lpig., p. 83. Also Robert K. Merton, 93. g;3., p. 51. 7 None of the discussion which follows is intended as criticism of our particular client. 8 Talcott Parsons, 22. £i§., p. 48. Read Bain, "Action Research and Group Dynamics," unpublished paper delivered to the Sociological Research Association on September 7, 1950, p. 5. .297. There is some evidence from our experience that these problems are solvable. Research support from other sources, for example, can be used to do those things necessary for filling in the gaps created by the conditions imposed by the client for applied research. 11 This matter will be discussed in considerable detail in a paper now in preparation. -293- Communication and the "Consequences" of Communication* Joel Smith, Robert C. Bealer, and Francis M. Sim Michigan State University In his recent assessment of the current state of sociological theory for our British Colleagues, Robert Merton reasserts the need for and desirability of theories of the middle range.l Such theories apply only to restricted ranges of phenomena and may be illustrated by com- munication theory.2 Unfortunately, Merton does not go on to consider the "restrictions" characteristic of such theories. NO procedures are suggested which may be used for setting the limits to the range of phenomena to be considered by such a theory. This, however, is pre- cisely the problem that must be addressed and resolved if progress is to be made toward the develOpment of an adequate middle range socio- logical theory of communication.3 Unless major advances are being reported in some of the papers now being delivered, it seems safe to say that there has been little progress since the need for such a theory was noted some years ago. In our own work at develOping such a theory, we have been led to the conclusion that a major breakthrough can be achieved only if the problem of the range of phenomena or types of questions to be covered by a middle range theory is attacked directly. Essentially this is a problem in definition, but eXperience suggests that formally acceptable definitions alone do not solve such problems. Although there is a *Journal Paper No. of the Michigan Agricultural Experiment Station, East Lansing, Michigan, Project No. 69. The project is also supported by the United States Department of Agriculture through con- tract Number 12—25-010-94. Read at the annual American Sociological Society Meetings, Washington, D. 0., August 27—29, 1957. -299. generally adequate definition of communication in common use, many peeple claiming interests in the subject translate these interests into research on questions which fall outside the range indicated as appropri- ate by this definition. Since these investigations are conducted, at least in part, to test theoretical prOpositions concerning "communi- cation," it seems important to ask what kinds of questions do provide apprOpriate data for testing "communication theories." Most investigations commonly considered to be "communications" re- search have either or both of two types of phenomena as foci. One of these types is the transfer of a set of meanings embodied in a message form in a manner that permits it to be received in a preferred way by a Specified person or persons. Questions regarding phenomena of this type concern all aspects of communicative situations - communication acts, processes, structures, systems, or combinations of these - and we would refer to research on such questions as communication research. The central core of interest is the conditions of meaning transfer. The other type of phenomenon investigated in "communications" research are events or circumstances which.may be considered as dependent on prior communication activities and/or conditions. This we term the "consequences“ of communication and distinguish from the other re- search focus by the type of factor seen as dependent and problematic. In communication research we must determine whether potential recip— ients of messages have information, represented by the form of these messages, as a result of having been eXposed to these messages. The change 212213 the recipient is the only change of interest. Questions regarding consequences refer not only to any and all other changes in ~300- both the behavior and mental condition of the message recipients, but also to any other condition viewed as possibly being dependent on a given communicative situation. Should an adequate theory of communication apply to questions in- volving either or both of these phenomena? Apparently, Opinions diverge on this matter. Communication theories currently being formulated differ in the range of phenomena with which they deal. Some major efforts at theoretical formulations, largely those of information theo- rists and cyberneticists, tend to focus only on message transfer. Others, in contrast, and here are found most sociologists, journalists, psychologists, and educators, seem to be concerned with both types of phenomena.8 The implications of this breadth of interest for the de- velopment of an adequate sociological theory of communication might be seen most easily if we consider first the meaning usually assigned to the term communication. Communication has been used here with the following definition im- plicit: the process through which a set of meanings embodied in a message is conveyed to a person or persons in such a way that the mean- ings received are equivalent to those which the initiator (s) of the message intended. While this definition is not a formal statement to which we might adhere in the systematic elaboration of a theory, it seems to embody the meaning to which most sociOIOgists subscribe. It suggests that the empirical referents of an adequate sociological theory of communication would be limited to only the first of the phenomena we have described, i.e., to message transfer. In this sense of the defin- ition, most current efforts at theoretical formulations by sociologists -3 0],- are misdirected and inefficient. The breadth of coverage intended will result in a less than "par- simonious" theory of communication. In order to see the grounds for this assertion, consider the following table which summarizes all poss- ible relationships between given communicative situations and any given phenomena which may be considered potential consequences of these situations. Existence of Specifig Communicative Situation in Les chggrence of Yes Specified Conseguence No If given communicative situations relate to their potential consequences so as to fall only in the lower left and/or upper right hand cells of this table, then consequences are distinguished from communication only semantically and can be accounted for completely by a communication theory. In predicting consequences, only those things need be known which are specified in the theory as necessary for accounting for the communicative situation. The distinction in empirical phenomena that has been suggested may be ignored and communication should be redefined to include the possibility of all kinds of consequences in the communi- cative situation. The occurrence of cases in the upper left and/or lower right hand cells, however, suggests the independence of the phenomena we have dis- tinguished. Knowledge of the communicative situation, by itself, is 0 inadequate for predicting a consequence.’ Relatedly, information -302- ~- . indicated by a theory of communication as necessary in accounting for a communicative situation would be insufficient for predicting the occurrence of the consequence. Undetermined additional information con- cerning factors and/or relationships superfluous for the communication theory would be necessary, and the canon of parsimony in theory build- ing would be violated. Indeed in directly addressing the problem of consequences, one would probably find that some of the information necessary for accounting for the communicative situation was inapplicable.lo While no formal proof has been attempted, we would assert that many cases occur which can apprOpriately be classified in the upper left and lower right hand cells. Whenever a directive message is received but its intentions not implemented because the receivers lack the means of implementation, we have the conditions of the lower right hand cell. The upper left hand cell is represented by all cases in which intended consequences occur among members of intended audiences even though these persons never receive the message. It is because both these situations do occur that the success of a communication attempt cannot be assessed by observing voluntary behavior. In both types of cases, the explana- tions of the communication situation and of the consequence arrangement require reference to different sets of factors. If more detailed consideration could be given to the distinction between the upper left and lower right hand cells, it would become quite clear that the viewbpoint being expressed here is largely that of functionalism. ‘While we are unable to discuss the relevance of the problems raised for functional analysis in the brief space available -303... here, it seems advisable to indicate how the problems of communication theory relate to the functionalist framework. Very briefly, both com- municative situations and consequences may be analyzed as structure and function. Either phenomenon may be assigned either position depending on the point in the empirical process that is chosen as the starting point for analysis.11 In analyzing the communicative situation, the existence or non-existence of acts, systems, or processes which are viewed as communicative is treated as a function of a configuration of other factors which constitute the structure for structural—func- tional analysis. In analyzing consequences, the communicative situation provides the structure, and its relation to other specified conditions viewed as possible consequences becomes problematic. The possibility of both structural and functional alternatives or equiv- alents here is well-known, i.e., any given structure may have a variety of different consequences and any given consequence may be produced by a variety of different structures.12 This structural-functional view of communication analysis points up the need for analytic disjunction of communicative situations and possible consequences, since their relation is contingent. but it also emphasizes that we are not suggesting that communicative situations and their possible consequences are unrelated. Perhaps the problems we have been discussing in this paper have gone unrecognized because com- munication analysts have not made their commitments to structural- functional analysis explicit and, accordingly, have been unable to see its implications. The efficacy of divorcing the two questions of communicative situation and their possible consequences might be demonstrated by formulating and explicating models necessary for considering each ques- tion as independent. Comparison of these models would indicate the extent of commonality between the two questions. In a current inves- tigation of specific communication systems this is being attempted. We are delimiting and supplying a general structural model of communication (appropriate to research concerning communication pg; pg) and, then, executing a functional analysis of the consequences of communication. The extent to which the variables of the communication model account for the consequences can then be examined as data concerning the in- dependenceof these two types of questions. We are led to conclude from the considerations reviewed that de- velOpment of an adequate sociological theory of communication will be delayed as long as we fail to distinguish communication and its conse- quences. Communication theory should not be designed with any concern for its ability to account for the consequences of communication. The relationship of communicative situations to their consequences is appropriately treated as a separate matter. While this does not help directly in developing theory, it may lead to a more efficient concen- tration of effort in this direction by providing a criterion for judging the relevance of particular empirical situations under consideration. We do not, however, wish to argue that questions concerning the consequences of communication are not sociologically relevant. On the contrary, these are often as or more interesting than communication pg; 88.14 We feel, though, that such matters will be understood more ads- quately if treated in their own right. Perhaps a more general concern -305- with the limits of theories is necessary for a more rapid develOpment of adequate sociological theories of less than global scope. -306- Footnotes 1 R. K. Merton, "The Role Set: Problems in Sociological Theory," Bgitigh Journal 21; Sociology, VII, 2 (June, 1957), 106-120. 2 lb;d., 108-109. Communication theories, of course, provide only one example of middle range theories. Stouffer's theory of intervening opportunities, Burgess' theory of urban spatial organization, Winch's theory of homogamy in mate selection, and Sutherland's theory of differential association are just a few examples of the wide variety of such theories with which sociology abounds. When we speak, throughout this paper, of a sociological theory we use the term sociology in the sense that colleagueship within the discipline implies shared orientations and points of view which are reflected in common lines of conceptual development and use. The sense of orien- tation is used in the same way as used by Robert Merton who suggests that the provision of such an orientation is one of the prime functions of theory. “Sociological Theory," American gournal pf Sociology, L, 6 (May, 1945), 464-465. 4 M. W. Riley and J. W. Riley, Jr., "A Sociological Approach to Com- munications Research,“ m @inion gu arterly, xv, 3 (June, 1951), 445-460. The absence of such a theory has resulted in little progress in codifying (R. K. Merton. ibid., 472-473) the vast number of investi- gations of mass communication. The major effort of Paul Lazarsfeld prior to the Riley's statement ("Communications Research and the Social Psychologist," in ngrent Trends in Social Psychology, (ed. W. Dennis) Pittsburgh: University of Pittsburgh Press, 1948, 218-273.) Handel's work on the movies (L. Handel, Hollygood nggg gt lgg Audience, Urbana: ~30?- University of Illinois Press, 1950), and Bogart's on television (L. Bogart, Egg égg 9: Television, New York: Ungar, 1957) are not effective efforts in this direction. The work of Katz and Lazarsfeld is a more meaningful step in this direction, but happens to suffer certain major flaws if the arguments to be presented here have merit (E. Katz and P. F. Lazarsfeld, Personal Igfluence, Glencoe: The Free Press, 1955). An important step in the direction of a useful logically consistent set of definitions for these various concepts is contained in a paper delivered to the Seminar on the Sociology of Mass Communication at Michigan State University by Francis M. Sim, James Harkness, and James Flynn on March 22, 1956. 6 These two categories cross-cut the common "control, content, audi- ence, effects" classification of communications research. Depending upon what is seen as problematic, studies would fall into either or both of these categories irrespective of how they might be classified into the more traditional scheme. W. R. Ashby, fig Introductjon 3g Cybernetics, (New York: John Wiley, 1956); E. Shannon and w. Weaver, The Mathematical Theory g_f_ Communi- ggtigp (Urbana: University of Illinois Press, 1949); Scientific.Amerigag. ed.,.Automatic Control (New York: Simon and Schuster, 1955), 83-121; exemplify theories of this type. ‘Wilbur Schramm also proposes a theory which, on its manifest level, appears to exclude consequences from its province ("Procedures and Effects on Mass Communication," in Mggg Eggig gag EduCgtion, Ihg Eifgyeghigg Yearbook 9f the National Society fig; thg §tudy Qf Education, 2gp; II, ed. N. B. Henry (Chicago: University of Chicago Press, 1954), 11-138. However, an analysis of the conceptual distinctions preposed suggests that Schramm's interest in communication is seriously colored by a concern with the consequences of communication. E. Katz and P. F. Lazarsfeld, pp. 323., E. Katz, "The Two Step Flow of Communication: An Up-to-Date Report on an Hypothesis," £32139 Opinion Qparterly, XXI, 1 (Spring, 1957), 61-78; B. Berelson, "Com- munications and Public Opinion," in Communications ip gpdpgp SocietV, ed. N. Schramm (Urbana: University of Illinois Press, 1948), 168-185; J. T. Klapper, The Effects pf Mass Media (New York: Bureau of Applied Social Research, Columbia University, 1949), mimeo: B. H. Westley and M. S. MacLean, Jr., "A Conceptual Model for Communications Research," Audio-Visual Communication Review III, (Winter, 1955), 3-12; M. S. MacLean, Jr., and B. H. Westley, "Research on Fortuitous Communi- cation: A Review," Audfio-ngual Communication Re‘.riew_:I_I_]_f, (Spring, 1955, 119-137); C. Gerbner, "Toward a General Model of Communication," épdip— Visual Communication Review, Iv (Sumner, 1956), 171-199; F. Fearing, "Social Impact of the Mass Media of Communication" in Mass Media and Education, pp. pip., 165-191; and T. M. Newcomb, "An Approach to the Study of Communicative Acts," Psychological Review, LX, 6 (November, 1953), 393-404. Perhaps this tendency to see both types of phenomena as subsumable under the same theory, so clearly eXpressed by Newcomb when he says with regard to his own efforts that "It seems likely that the dynamics of such a system are such that from an adequate understand- ing of its properties at a given moment there can be predicted both the likelihood of a given act of communication and the nature of changes in those preperties which.will result from that act," (gggg., p. 403), results from an uncritical acceptance of the now classic statement that .309- A.— the concern of communications research is "who says what to whom under what circumstances with what effect." 9 M. L. DeFleur, "A Mass Communication MOdel of Stimulus Responses Relationships: An Experiment in Leaflet Message Diffusion," Sggigm- 2353, XIX, (March, 1956); and E. Katz and P. F. Lazarsfeld, pp. 213.. 137-334, offer two cases of important recent research illustrating this point. 10 M. L. DeFleur, lgig., 12-25. M. J. Levy, Jr., The Structugg‘gg Society (Princeton: Princeton University Press, 1952), 60-62. 12 R. K. Merton, Social Theory and Social Structure, (Glencoe, The Free Press, 1949) 35-36. 1 3 Ibig., p. 36. 14 Nor do we agree with HacLean and Westley ("Research on Fortuitous Communication: A Review," 22- git.) that, "Pure message studies would probably bear little fruit." (p. 126). -310- Bibliopnphy Books Anderson. R. L. and '13. L. Bancroft, Statistics; m in Rooms!» New York: Dicker-Hill. 1952. M. George 14.. .12-..th- ammumm an I hoe. Iowa: The College Book-tore. 1951+. Benedict, Murray, m Policies 9; 3.29. m States, 1220-1250. How York: Twentieth Century land, 1953. Cochran, William G. and Gertrude H. Cox, Laminate]. Denim. In York: John Wiley. 1950. Cohen, Merrie n. and lrneot Easel. g introduction to; Lo c gn_d_ Scientific Method. new York: momma-toe. 19%. Coutu, Walter, argent m Nature. New York: Knopf, 19149. 408 manner. Edmund. mmgam One ‘ immune; Sociologcd Eugen in Q Unite; States, lee York: Ens-pore, 1957. Dewey. John. m: 21 Mg; m. New York: Henry Holt. 1938. Duoenberry. June 5., Income, gem g._n__d m noon 2;; Con 3271». Cambridge: Harvard. Univeroity Prone, 1%. Iilher, B. 1., am gflrmntl. Sixth edition. London: Oliver and Boyd, 1951. Eagood, Margaret J. and Daniel 0. Price. Statistic- Lo; Sociologiotg. Revised edition. New York: Henry Holt. 1952. 3mm. x. 3.. and I. «7- Reed. Macaw m 33 M m. Vuhington, D. 0.: United State: Dep;tnent of Agriculture. Agricultural Hunting Service, 19 . rots, mum and Paul I. mar-fold. mound m. Glencoe: in. FIG. P139... 1955. number». Oscar. m m 5% 52:12.1; 2: W lav York: John racy. 1952. hooker. A. I... g. Anthropglog 3393. Chicago: University or Ghioege no“, 1953. -311- Bibliog'eplw Boon Levy, union J..J‘:.. m 8mm 21 mm. Princeton. new Jerky: Pr1nceton Univereity Prue, 1952. Leonie, Cherlee P. and J. Allen Beegle. Rural Sociolog: 3:; 8mm 9!. m. Rammed 01121:. new Jersey: Prentice-Hell. 1957. Lnndeberg. George; Ieundntione _e_£_ Sociolog. New York: Macmillan, 1939. Need. Hergeret end B. Metrm, g... 11:; m g; @522; at 9; meme. Chicago: University of Chicago Preee, 1953. Morton, Robert L, gocig; @9922 E4. Geog Strncmg, Revised. edition. Glencoe: he Tree Prue, 1957. Miller. Delbert C..end. William Iorn, [mule]. Sociolog, lee York: Eat-pore, 1951. Newconb. Model-e. Social W. New York: Dryden. 1950. Oetle, Bernard, totie ce gm hoe. love: love State College ha.., 19 e Pereone, Mcott, E23221 9g Sociolo c Yheggz. noticed. edition. Glencoe: no Tree Pro”, 19 . . a; Social £232. Glencoe: no Zl'ree Preee, 1951. _ -MW£MA‘M¢ 51m008m1'ru Preee, 191:9. Pearson, J'. A. and I. 3. Bennett. S..i!.§é____‘t1°'~1 LN“ 52214.9; 2. gamma mm. lev York: John Wiley, 19%. m. .2. . me am an m 2: mm. new term mam-mu. 1933. niece-n, Devin, lethen Glen: end m1 Denney. .13. m 933$. new Keven, Connecticut: Me wivereity Preee. 1950. Riley, J. V.. Jr., )letilde Unite Riley and Jeclmon new, W Studiee in m my, New Brunswick: htgere univereity Prue. 19?!“ 3m. Arnold. Lie. mg MMWM! AW mach. New York: V. V. lorton. 1955. ..5‘ “3V -312. Bibliography Books Decanters. Bernhard end B. H. White. 39,. ELSE-3221‘ glam Art; in mg. menace: Yhe l‘reo Prue end moon'o Yin; ”9” o 1957o 510501. 314307. W 3“““1” £2.11 ....“9. Lil-....“ “"3- £22.22.- let York: Moan-Hill, 1953. sin—.1. Georg. mg m g; Georg my (Kurt a. tent. tune.) Glencoe: The Free Fun, 1950. Won. 0. n. gtetiotioel 59%. 111th edition. hoe, Ion: Iowa State College Preee, 19 e . ‘ Stonffer. Camel. _e_§_. 51... moment 2; Prediction, Princeton: Princeton University Preeo, 1950. The Bond Anthropological lnetitnte oi' Greet Britain end Ireland, 1. Committee Report. 123.9. 222 mm 22 W m: Rutledge end hogan, m1. 1951. Min. M. and John Bennett, Coca age: Structure 94m Nev York: Knopf, 191:9. Vellie, W. 1. end H. Y. Roberto, §tetiotige: A 12! Month. Glencoe: the tree Prose, 1956. Villim. Robin. Jr., M1939. 82213131, New York: Knopf, 1951. Iitt. Lawrence W. and Mordecei hekiel. 95:51 a mom Home: Food and Agiculturel Orgenisetion of the thited lotione, 1953. mm. frank. WELD“ mimosa. £22m. Mud edition. London: Chnrlee Griffin. 19 53. Yale, G. Udny end )1. Gt Xendell. g introduction 2&Mgg Stetiotice. thirteenth edition. London: Cherlee Griffin , 191:8. Zeieel. hone. in; it lit; m. l'onrth edition. Yew York: Barrel. 1957. ,Articloe in Booke Berton. A. 8.. '1!» Concept of Property-Space in Sociel Reeoerch,' a 213829151 Quack. (Poul usereteld end It. Boeen‘oerg. egg. Glencoe: the Tree Pro". 1955. PP. l1:0-53. ~313- Diblioyephy drtioloe in Booke Buckley, Welter, 'Btructurel-Ponctionel Anelyeie in Modern Sociology,‘ llodern gociological Theory, (Retard Decker Ind Alvin Beckett, _e_de_.s New York: Dryden. 1957. PP. 236-259. Coleman, J. 8.. 'Stetioticel Problem,‘ Mm. (s. :4. Lipeet, )lertin Brow end J. 8. Cblenen) Glencoe: The Preo Proeo, 1956, pp. 427-432. Peering, Franklin, 'Bociel Inpect of the nose Medic of. Comicetion,’ mmemm communism at mammogram (n. 13. 301117. 99.. Chicago: University of Chicago Preee, 195”, pp.165-l91. Prencee, Roy 0.. "the Nature of Scientific neutron," a W 1:; Social ee 1:. (J. 1'. new, 99..) mrioburg. Penneylvenie: Stockpelo. 19 . pp. 3—20. Gouldner, A. v.. 'Sone Oheervetione on Syotentic rheory, 1945-19553 °1 ngmmnm (H. 1.. zetterbere. ed, Paris: United Netione Educational, Social end Cultural Craniutiono, 1956, pp. 3942. 3‘11". 14., 'Induetriel 30619-1 thflhge' 32432.22! 21 mm- 92- (Gardner Lindzey, A.) Cambridge, Mum: Addieon-Voeley, 195“. PP. nah-1123. Helpel, Carl 0.. 'he Logic of Junctionel Analylilfl m _og Sgiolo‘icg noon, (1.. (iron, ad.) Iveneton, Illinoie: Dov. Peterson, 1959. pp. 271-307. Home, George and Henry V. Biochen, 'Peychologicel Aepocte of Social Structure.“ Mlmh 2:. will. 222.211.11.25:- (Gardnu- Linda-r. . 5.) Cambridge: Addieon-Ieelq. 1954. pp. 786-829. novlend, Carl 1., "Brecte o: the lieu Medic or Commotion} p; Sociel mm. (Gardner Lindsey. &) Cambridge: Addison-Volley. 1959. pp. 1062-1103. Inhlee. A. end D. J. Levinoen, “national Character: on. Study of node]. Personality end Sociocultural Breton-J Handbook 91 Social Peggholog, (Goirdner Lindsey, 3d.) Contridge: dddileon' -Veeley. 195». pp. 977-1020. Kendall, Petricie I... _"lote on Significance leete,‘ m M Minion. (B. x. llorton, o. a. lender endP. 1.. Kendall, .2.) Cambridge, the»: Renard University Proeo, 1957, pp. 301-305. -311+- Bibliogreplw Articlee in Dooke mapper, Joeeph 1., "The Comparative Effects of the Various Media} no.4.» 8- 29.13.24: 1: 2:. ......ms W ('1. Schema. pd.) ‘ Ill-bone: mivereity of Illinoie Preoe, 1951:, pp. 91-105. Musteld, Pull 1‘.,~ 'Comunicotion Beeeerch end the Social Plycholo- giot,‘ Currgt bends _ip_ Social Penholog, (V. Donnie, g.) Pitteburgh: University of Pittsburgh Prue, 19%. PP. 218-273. . 'Interpretetion of Statistical Beletione ee o. Research Operationfl Inn m: 2:. 8 ma- (m1 1'. Lucreteld end ll. Rooenberg. .991. menace: m Proe Proee, 1955. PP- 1154-25. N . end Patricia L. Kendall, 'Problene of Survey Anelnie,‘ Cogtinudtige a Social Egeeuch. (n. 1. Norton and real I. Legerefeld, 293.5 Glencoe: The Iroe Prose, 1950. pp. 131- 19 . , . and Robert K. Norton, “hiendehip as Social Proceeo: A Substantive and Methodological Lennie," Freedo g Control in; Medea Sociegz, (H. Berger. gt. 3!," 53,5 New York: Venlootrend. 1954, pp 18-66. Mead, Margaret, “Rational Character,’ thro lo Long. (A. L. noeber, 99,.) Chicago: (hivereity of Chicago Preeo, 1953, me 642-667. Rodfield, Robert, ' Relations of Anthropolog to the Social Science: and to the Humanities,' ppm-mug pm. (A. L. Iroober, , $> Chicago: university of Chicago Preee, 1953, pp. 728-738. Riley, John W., Jr., "The Sociological Variable) Soc ole col Sadiee in; Scalp Analgio, (J. I. Riley, Jr., lutilde U. Riley, and Jackson Toby, 93;.) New Brunnick, New Joreey: mtgero University Preee, 195b, pp. 5-20. Riley. Metildnflhdte and John W. Riley, Jr., ' A Sociologcdl Approach to Communication Research,“ the Eoceee g Egan 9_f_ £9; Commotion. (I. Schrenn, 93:5 urbane: university of Illinoie Prone, 1951:, pp. 389-1401. Sal-bin, Theodore, 'Role Mry,‘ m 9;. Social Meg. - (mm Lindsey. :4.) “bridge: Addioon-Veeley, 19 , pp. 223-258. Schren, Wilbur, “How Commotion Vorh,‘ p.13 Eocene 3d 112%“ 2; Neon Wig. (I. Schren, _e_d.) arm: University or: Illinois Preee, 1951:. pp. 3-26. . O -315- Bibliography Articles in Books Stouflor, Sensual A..H. 1‘. Dorptta, D. G. Hays, end A. 1'. Henry, IA Technique Per Improving Cumlative Scales," Sociologcal m 1.5 Scale My, (J. 11. Riley. Jr., Matilda V. Riley and Jackson Toby, 1g.) New h-unswick, new Jersey: Rutgers University Prue, 1951:, pp. 372—389. _ Suchman, n. A., and Roy G. Frances, "Scaling Techniques in Social Research," _A_n; gtroduction pg Social Research. (John 1'. Doby, pg.) Harrisburg, Pennsylvania: Stackpole, 1954, pp. 123-137. Periodical Articles Aberle, p. 1., a, el., ”rho; Functional Prerequisites ed! a Society,“ Ithics, vol.—60 (January, 1950) pp. loo-111. Baker. John D., Jr., and Don Paarlberg, 'Hov Accurate is Outlook?“ m p; Q legal-nice, vol. 3” (1952) pp. 509-519. Barber, Bernard, ”Structural-Motional Analysis: Cone Problems and Misunderetandings," _Aner_;_ican Sociological 91.111: vol. 21 (April, 1956) We 129‘135o Beal. George. 'Additicnal Hypotheses in Participation Research,“ Rural Sociolo vol. 21 (September-Decenber, 1956) pp, W55 . Blumer. Herbert, 'Socielogical Analysis and tha‘Variable" Anericm Sociological m vol. 21 (Mentor, 1956) pp. 683390. ___, I"that is wrong With Social Iheory,‘ Asericg Sgciolggical More vol. 19 (February. 1951+) pp. 3-10. Christiansen, J. R. , “no Behavioral Correlation of Membership in mired. Neighborhoods,“ Rural Sociolog, vol. 22 (dial-ch, 1957) PP. 1249. Cochran. William 6., 'Sone Consequences When Elle Aesunptions tor the Analysis of. Variance Are Not Satisfied,“ Diogtriog, vol. 3 (“011, 1914‘?) PPo 22’380 , 'Sone Methods for Strengthening the Coleen x2 route,- We '01- 10 (1954) PP. 417-451, Colonn, A. Lee. and C. Paul Marsh, “Differential Communication Anon; Parners in e Kentucn County." M Sgiolog, vol. 20 (June, 1955) We 93‘101o * «316- Dibl ical-aw Periodical Articles Cola-an, Janes 8.. letter to the editor, 419111.932 my, 9; m, 7910 “(Me 1958) PP. 59.600 Cepp, Jones 3.. 'i'ovard Generalization in Porn hactice Beuareh,‘ m Sociolog, vol. 23 (June, 1958) pp. 103-111. Davis, Janos A., rejoinder to Janos 8. Coleman letter. M m 2;. Sociolog, vol. 61: (July, 1958) p. 61. 1: ins! t ici ric J Soc 1 ---' .3363 chasm, 1958pr. nae-see --’-—‘-" ' Dean. 1.. H. A. Aurbach, and a. Paul Harsh. 'Sone rectors Deleted to nationality in Decision Making Along Para Operators, Egg; Sociolog, vol. 23 (June. 1958) pp. 121-135. Deneur, M. I.., 'A Kass Commotion Model of Stimlus Desponso Relationships: An Experimental Study in Leaflet lounge Diffu- sion.“ Socione vol. 19 (lurch, 1956) pp. 12-25. Duncan, D. D., “Rural Sociolog Coning of Age) M Sociolog, vol. 19 (larch. 1955) pp. 1-12. lisonhart, Churchill, 'Ihe Assmsptions Underlying the Analysis of Variance," Biometrics vol. 3 (March, 1947) pp. 1-21. Farber, H. I... “he Problen of rational Character: A lethodological Amy-1'." m 21.. W '01. 30 (1950) pp. 307-316. Illegal, 1'. C., 'A Multiple Correlation Analysis of. Doctors Associated with Adoption of Porn Practices,“ and m vol. 21 (September-Deceaber, 1956) pp. 284-292. Green. B. 11.. 'Batting hex-egos in Agricultural Forecasting,‘l gm 3; Farm Eoncnics, vol. 8 (1926) pp. Nit-193. Hear, John I... I'Conservetisn-hdicalism and the Mal-when Continue,’ m sociolog, vol. 17 (December, 1952) pp. aha-3&7. Hoer, John 1., 'Accuraocy of loss rare Outlook Intornetiu,‘ m 2; 1g; Eunice, vol. 36 (1954) pp. 143-147. Wits, Hathan, 'A hoterial Arrangement of Coaparisons of really Shadow W Sooiolog, vol. 58 (lurch, 1953) PPO ‘97 00 .317- Bibliograplw Periodical Articles Rich. Leslie, 'Coniidence Intervals for Clustered Samples,‘ M Sociglogcal m vol. 22 (April, 1957) pp. lit-165. Kroeber. A. 13., and Clyde Kluckhohn, “me Concept of Culture: A Critical Rerieu oi Definitionsfl l’apggs 9; g; m m vol. #1 (1950) entire issue. Lionberger. R. 1.. “The Diifusion of Farm and Rome Iniormtion as an Area of Sociological Research,‘I M Sociolog, vol. 17 (June, 1952) pp. 132-1110. . 'The Relation of Internal Boo ial Groups to the Diffusion of Darn Information in a Northeast Missouri Farming kanmunity,‘I Rural Sociolog, vol. 19 (September, 195%) pp. 233-2146. Lindesnith, A. D., and Anslen Straus, 'A Critique of Culture- Personality Writings} American Sociologicg grieve, vol. 15 (1950) pp. 587-600. Harsh, C. Pavel, and A. Loo Coleman, “Differential Consummation Anon; Iarners in a Kentucky County," 1mm Sociolgg, vol. 20 (June, 1955) pp. 93-101. . . "the Relation of Iarner Characteristics to the Adoption of Recommended l‘arn I’ractices,‘I £935; Sociolo-g, vol. 20 (September-December. 1955) pp. 289-296. McGinnis, Robert , 'Randonisation and Inference in Sociological Research,“ Aneriggg Mioloflcg; Bevin, vol. 23 (August, 1958) pp. “OB-MA». Nelson, Lorry, “Enrol Life in a use-Industrial Society) mg; Sociolog, vol. 22 (March, 1957) pp. 20-30. Rogers. 1:. 14., "A Conceptual Variable Analysis of technological Change,u £111. 3.2112122. 1'01. 23 (M9. 1958) pp. 1345-1115. Selvin. Hanan G., 'A Critique of 'rests o: Sigincance in Survey Research,“ Mica Sociologigol 3313, vol. 22 (October .1957) pp. 519.5270 Sewell, Iillian 3., 'Some Observations on Theory iestiu.‘ m gociolofl, vol. 21 (March, 1956) pp. 1412. __ . :L’go. Ealler and M. A. Straus, “Social Status and m tion and Occupational Aspiration.' w W £21.19. vol. 22 (rat-nary. 1957) pp. 67-73. -318- Bibliography Periodical Articles Smith. Joel, “uichigan Par-ers' Use of hdio and levspaper Jarret 1'0".“ w Bullet Michigan Agricultural hperisent Station, vol. 38 Mar, 1956) pp. 612-627. Eaves. lbrvin, and Real Cross, “A Critique of Rural Sociological Research, 1950,‘I Rural Socioloa, vol. 17 (June, 1952) pp. 109-118. . Turner, Ralph, “Role-taking. Role Standpoint and Reference Group Behavior,“ Anericon Journal 9; Sociolgg, vol. 61 (January, 1955) PP. 31:328. Useen, John, Pierre Tangent, and Ruth Useen, “Stratification in a Prairie bun,“ American Sociolggical Review, vol. 7 (19113) Pp. 33196sz Isrriner, c. 1., “Groups Are 11.1.1: 1 Reattirmtion,“ m Sociological m, n1. 21 (October. 1956) pp. 919-5511. Westoff, 'C. 1., P. C. Sagi and R. 1.. Kelly, “Pertility brough Eventy Years of Marriage: A Study of Predictive Possibilities,“ W Bociolgical m, vol. 23 (October, 1958) pp. 5 e , Wilson, I. C., “The Effectiveness of Docunentary Broadcasts,“ m gainion m. vol. 12 (1948) pp. 1.9-29. Wold, 11.1-m, “Causal Inference from Observational Data,“ m 21 the no Statistical Societz, vol. 119, series A, part 1 .63 We 28-50e Bulletins, Circulars and Reports Anonymous, Listenggg 5mg; 9; Livestocg fl 0331: abet W, Chicago: Chicago Producers Conission, Decanter, 19119 (nines). - 252.1191 10:; £111.92 in the. 314.2911. mmgton. 13.0.: United States Department of Ayiculture, 19119. __ A A gtlook lork: garirst pp, Years Washington, D.C.: United States Departmt of Agiculture, l 2, .319... Bibliograplw Bulletins, Circulars and Reports Betta, r. 3.,” g” MEN M!“ thg and Heart” Merit of 31.19. .3191. M Roms. 21 W Washington, D. C. : Mean of Agricultural Economics and Produc- tion and Marketing Administration, United States Depart-ent of Agriculture. No. 703, December, 1949. Dodd-I. J. Perry. and x. 1:. Mann. magmmmmm Market Revel, Ames: Iowa State College Agicultural hperinent Station, Research Bulletin l+17, November, 195‘}. _MMEQLMW Want From 1322291 kit 1. Area: Iowa State College, Special Report, 195”. (mineo ...... 951132221.»an mmgl‘rom gamma 1...: love State College Agicultural Experiment Station, Research Bulletin 1513, August, 195141. I‘m-ell. Geno. (compiler) Here 12 1221:9129. m 3.9.." 29. 111022521. report of the Market Revs and Information Conference, Kansas City, Missouri, Hay, 1955, published by Iowa State College, “‘0 19550 ment. (2.1.. mammmmmm Reports, 1 - 0. Ithaca, New York: Cornell University Agicultural Experiment Station, Bulletin 8811, 1952. moom.1,n.m1. 3.. 11 1111211 21 111 meet Em San—ice 1.1 Ohi__9_, Wooster: Ohio Agricultural Experiment Station, Research Bulletin 77+, May. 195“- Rielson, James 3., and R. l'. Bittner, M Pracgicg mg Michi , not Lansing: Michigan State Agricultural Experiment Station, Technical Bulletin 263, January, 1958. Wells, 0. 17.. AC camison 9__£0 Outlook W Vith Rants, Washington, D.C. . Bureau of Agricultural Economics, United States Department of Agriculture, 1930, (line). ““38““"3. C. on A__.!1_£l'._ °f fimmIflWfim... Statements, Moscow, Idaho: Idaho Agricultural Experiment Station, cm» 62,1932. ~320- Bibliography tbpuhlished Interials Bealer, Robert C., Value Qflgntatiens pp; RM Cogelatge g; Boduceg-gtrons 29. W Coomativgs, Master's thesis, . Ehe Pennsylvania State University, 1955. Holmes, loah D., Communication yEdia mm Which Iowa FEE. tors Obtain giggltural Outlook Information, water's thesis, Iesa State College, 1951. smith. Joel. W21 magmas Co 122 . PhD. dissertation, Northwestern University, 19 . , and hancis )1. Sin, fichim's Doug; 32an m Conmtion and Egg at Market gm, Department of Sociolog and AnthroPology, Michigan State mirersity, East Lansing. undated, (mimeo ). Witt, Lawrence 11.. “Economic Efficiency and Social Welfare,“ Department of Agricultural Economics, Michigan State Univer- sity, East Lansing, undated, (mince). if I U, - . . \ ' c e o v . \1/ A 4. a t. «2}; H r f 0 \‘ S o I . fi . . p. r 9 AU R0035 USE GEE-LY. "Ic'{flifliluflll‘wfl‘l‘mfllfliflmlfifs