‘KQAW’ “7‘ """“°‘ .tlr-O'. . o 1 a “‘9‘ A3 ,\ . " ’3 t ‘.‘. I“;"‘.'\:r . ‘ , ‘ . :4; m4;- . . It‘ :' .3. I. "k'..lll‘t‘:‘. ' V . l I. .o‘ ‘l. .'_.A H ‘ -" I '1‘?“ ‘1}: ‘. ‘_ ‘ I+bl§ "r.’ ~ _v'§:l“ t. ' 'l‘ ‘ - " A‘ if) #7? " , ' - V 3.\ . _“_‘TV 1"} ”'3" T I} 712% "-- t ‘ ' ' I . ' "‘.:..‘. o ‘ r"-' .. v ~67:- L'v' 3‘." I . '- -."'.\ > ':J"| . ', - -">$ 4' “ -.~|“|v 1". . Or 3 . . 'n . J ‘ u .'o‘(.a..,-l ‘. ‘ v ‘ ‘ .> . x .‘I711:"vf.- J‘u‘flu‘ 'I q; l . a“. .n,|.‘l- IIIVIIII‘ . 93:5 ~ I" '5‘“! 44 fig“. \ ' w". .‘ -" .. ~ , 'nl . N ‘1.» '.' . I, ‘ .IVI..'_I o ‘- "EMIJl ‘n'. o '4'... 'll‘lr h y :7“ "7 ' '1": IL . - ~.'- HI)... lI-I.I"\\V_ \L. "' ‘ 'u-b '. ‘ _- llfllilll'flml llllllllflilfllllll WWW L/ “we 3 1293 01060 0637 0 ‘¢ ‘1 t , . , , r This is to certify that the thesis entitled AN EXAMINATION OF THE RELATIONSHIP BETWEEN SOCIAL STRUCTURAL FACTORS, COMPLEXITY OF THE INFORMATION ENVIRONMENT, INDIVIDUALS' PROCESSING STYLE, AND COGNITIVE STRUCTURE presented by N. J. STOYANOFF has been accepted towards fulfillment of the requirements for Ph.D. degree in COMMUNICATION £40.40 @AA Major professor 4f/17/f/ Date 0-7639 MSU RETURNING MATERIALS: Place in book drop to remove this checkout from LIBRARIES --_. your record. FINES will be charged if book is returned after the date stamped below. 3.134 1 1 15:; AN EXAMINATION OF THE RELATIONSHIP BETWEEN SOCIAL STRUCTURAL FACTORS, COMPLEXITY OF THE INFORMATION ENVIRONMENT, INDIVIDUALS' PROCESSING STYLE, AND COGNITIVE STRUCTURE BY N.J. Stoyanoff A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Communication 1981 taich struc 39 u: iata “ ‘ . r 35. ABSTRACT AN EXAMINATION OF THE RELATIONSHIP BETWEEN SOCIAL STRUCTURAL FACTORS, COMPLEXITY OF THE INFORMATION ENVIRONMENT, INDIVIDUALS' PROCESSING STYLE, AND COGNITIVE STRUCTURE BY N. J. Stoyanoff A theory of information processing is presented which explains how individual differences in cognitive structure occur. Data were collected by questionnaire from 99 undergraduate students, and the theory was tested using a linear structural equation model. The results of the data analysis indicate that while there were specification problems with the model, the relationships prOposed by the theory were supported. Accepted by the faculty of the Department of Communication, College of Communication Arts and Sciences, Michigan State University, in partial ful- fillment of the requirements for the Doctor of 6%W, / £3024 Philosophy degree. Director of Dfésertation {99% / Guidance Committee: , Chairman :&~%~ /\ 24244 f (//z Aw / W ”Kfléf/é/ZV/ .1) / To Pando Stoyanoff and Traico Kostoff, whose lives inspired me to pursue my own dreams. iii v Reaa‘ u u , O A..Ar V'v. n2}- . .125 b. ‘u‘ . .VA“ va‘u' . ‘a‘s 3 -; l ACKNOWLEDGMENTS This dissertation is concerned with a topic I first began investigating in September of 1975. Needless to say, over the past 5 1/2 years my thinking patterns, behavioral patterns and general well-being have been influenced by a number of persons whose contributions I would now like to recognize. In keeping with tradition, I would first like to thank my advisor Dr. Edward L. Pink for his many contribu- tions to my overall development. In spite of his genius, patience, and kindness, I still managed to make this quite a trying experience for him. Also, I'd like to thank the other members of my doctoral committee--Dr. Gerald R. Miller, Dr. Thomas Muth and Dr. Joseph Woelfel, whose advise and guidance have allowed me to accomplish the many formid- able tasks required for the completion of this dissertation. There are a number of people I have worked with over the past 5 1/2 years who contributed something of themselves to help me along, and I'd like to say thanks to: Drs. Charles K. Atkin, David C. Ralph, Everett M. Rogers, Gerald Goldhaber, Charles Petrie, Allen Lichtenstein, George A. Barnett, Felipe Korzenny, D. Lawrence Kincaid, June Ock Yum, iv Dari: Jane: n, l... Bare: ' u V'fln "*th H. 5.ch 23c e>13: (‘1 '4 David R. Brandt, James A. Danowski, Thomas O. Mwanika, James Gillham, Edmund P. Kaminski, Steven T. McDermott, Barbara Ann Walker, and Martin P. Block. During the course of this project, there were a number of people who volunteered their time and energy to help me. Without their efforts this endeavor would have been impossible. Consequently, I humbly pay my respects to the following people who performed outstanding work for a tyrant who demanded perfection: Joe Pierce, Tamie Crespo, Michelle Kantor, Marcie Wolfe, Tom Faes, Mike Panzegrau and Carole Galloway. These people helped make this project an enjoyable experience. While I think they believe I often forgot them, I would have never been able to keep my sanity were it not for my many good friends. At the sake of making an unfor- giveable error and leaving out someone's name, I'd like to thank all those people who didn't know what the hell was going on with this dissertation, or why it was taking so long to finish, but who stood by me nonetheless--Chris Costantino, Doug McDuff, Jody Hale, Ron Parkinson, Pablo Logan, James Ferguson, Kathleen Muglia, Larry Slade, Melody Lees, Ken and Teresa Quartermus, Peter and Marcie Hill, Ken Miller, Bob Cliza, Rollie Legg, Bob Duff, and Bill Sundstrom. There are four people I'd like to single out and express my deepest gratitude to. They are Mitzi Jarrett, Bill Donohue, Carolyn Ruth Fox, and Jamie DinKelacker. v TC 3*! .‘fiRA {Luv to; ”in ‘ H... my.” D” 9‘1; To Mitzi I credit saving my life and providing me with encouragement to engage on this project, and for this I will always be grateful. Bill Donohue has been a great friend and colleague, whose ability to see the lighter side of life was an inspiration during the more exasperating periods of my graduate career. CRF is by far the most com— passionate person I have ever met, and her comfort and support I will always cherish. To Jamie, what can I say? He is the person I most often turned to when I needed to think something out. He is brilliant. He is honest. He is generous. He is a true friend with whom I look forward to interacting with for a long time to come. Finally, I'd like to express my sincere love for the members of my family--Maria, Jimmy, Pete, Judy, Diana and Bill, whose support never waned, whose love only grows stronger, and without whom I'd surely be lost. And last, but certainly not least, there's Andrea. For the past year she has weathered all my trials and tribulations, and in the process captured my heart. To all these people, my heartfelt thanks. vi Chapter TABLE OF CONTENTS A THEORY OF INFORMATION PROCESSING . . Introduction. . . . . . A Review of the Work on Cognitive Structure . . . . . . . Theoretic roots . . . . The static- -structural models. . The dynamic- processing models . Summary of the dynamic—processing models . . . . . A summary of the work on cognitive structure . . . . . The Distributional Patterns of Individuals' Responses . . . . Psychophysics and human responses Personality theory and human responses . . . . . . Cognitive structure and human responses . . . . . . Summary: the distribution pattern of individuals' responses . . The Problems with Prior Research. . A Theory of Information Processing . Information processing: the detection of change. . . . Information processing as a measurement procedure . . . Environmental change: adaptation to new environments. . . . Personal change: stages of development. . . . . . Social structural change: adap- tation to meet role aspirations and/or expectations. . . . Chapter Summary . . . . . . FOOTNOTES . . . . . . . . vii Page 22 29 3O 32 36 36 39 48 50 53 56 58 " 7' Chigte. III Chapter Page II METHOD . . . . . . . . . . 59 Overview. . . . . . . . 59 Operationalization . . . . . . 60 Primary measures. . . . . . 60 Secondary measures . . . . . 63 Background items. . . . . . 65 The Pre-Test. . . . . . . . 6S Objectives . . . . . . . 65 Procedure . . . . . . . 67 Pre-test results. . . . . . 67 Discussion of the pre-test results . 81 The Final Instrument. . . . . 84 The social— structural items . . . 85 The paired comparison items . . 86 The Scott "listing and Comparing" task . . . . . . . 87 The Administration of the Final Instrument . . . . . . . 87 Selection of subjects . . . . 87 Administration of the questionnaire . 89 Data Handling Procedures. . . . . 92 FOOTNOTES . . . . . . . . . 94 III RESULTS AND DISCUSSION . . . . . . 96 A Basic Discussion of Linear Regression Analysis . . . . . . . 98 Analysis of the Correlations Among Indicators . . . . . . 108 Analysis of the correlations among the exogenous indicators . . 108 Analysis of the Correlations Among the Indicators of Predictor Variables with the Indicators of Predicted Variables . 117 Analysis of the correlations among the indicators of E and n . . . 118 Analysis of the correlatiohs among the indicators of n and n . . 120 Analysis of the corrélationg among the indicators of n and n . . 120 Analysis of the corrélationg among the indicators of n and n3. . . A Summary and Discussion of the Cor- relation Analyses Performed on the Indicators of the Model. . . . . 125 viii 122 Cu .O. b FUL :3 h a . Y. Y\' o .A \xx “bu . u: Q My in. H Chapter III (cont'd.) FOOTNOTES . . . . . . . . IV CONCLUSION . . . . . . . . Theoretic summary . . . . Methodological summary . . . Summary of findings . . . . Problems and Limitations. . . . Theoretic considerations. . . Methodological considerations . Directions for Future Research . . Dissertation Summary. . . . . FOOTNOTES . . . . . . . . APPENDICES . . . . . . . . . . A. The Pre-Test Instrument . . . . . B. Concept Categories and Terms Extracted From the Pre-Test Instrument which were Concerned with the United States - Iranian Hostage Situation (N=99). . . C. The Final Instrument . . . . D. Concept Categories and Terms Extracted From the Re-Administration of the Request For Information Regarding the United States - Iranian Hostage Situation (N=38). . . . . . E. A Brief Description of the Data Assessed From Student Records. . . . . F. Instructions for Calling Participants. . G. Procedures that were Followed to Greet Participants and Introduce Them to the Study . . . . . . . . . H. Consent Forms A and B. . . . . . I. Model Indicators, Lables, and Corresponding Questionnaire Items . . . . . . J. Codebook for x - x . An Examination of the Fit of the Indicators . . . . . Test of the Full Model . . . A Discussion of the Overall Results 2 5 . . . . . BIBLIOGRAPHY. . . . . . . . . ix Page 126 138 144 150 152 152 155 156 158 158 159 161 165 166 167 167 175 183 211 219 228 230 233 235 249 250 LIST OF TABLES Table Page 1. Participants' Listing of the Two Most Important News Events in the Past Year (N=29) o o o o o o o o o o 68 2. Concept-Categories Extracted from the 459 Concepts Generated by the 28 Participants who Listed the Taking of the American Hostages in Tehran, Iran as an Important News Event . . . . . . . . . 70 3. Mean (X), Standard Deviation (S.D.), Coefficient of Variation (C.V.), Minimum Score (Min), Maximum Score (Max) and Range for All Indicators of Cognitive Structure (N=29). . . . . . . . 71 4. Correlations Among the Indicators of Cognitive Structure (N=29) . . . . . 72 5. Means (X) and Standard Deviations (s.d.) for the Indicators of Information Process- ing Style Derived From the MMDS Instrument (N=29) . . . . . . . . . . 75 6. Variances, Covariances and Correlations for the Five Indicators Used to Determine Nominal-Dominant and Ratio-Dominant Information Processors (N=29) . . . . 76 7. Unstandardized and Standardized Factor Score Coefficients for the Five Indicators Used to Determine Nominal-Dominant and Ratio-Dominant Information Processors (N=29) . . . . . . . . . . 79 8. Mean Scores for the Indicators of Cognitive Structure: Upper vs. Lower Half Means and Upper vs. Lower Quartile Means (N=29) . 80 X Table 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. F-Ratios for the Indicators of Cognitive Structure by Information Processing Style: Comparison of the Upper vs. Lower Half and Upper vs. Lower Quartiles (N=29) . . . Descriptive Statistics for All Indicators in the Full Model (N=99) . . . . . Correlations Among the Exogenous Indicators (N=99). . . . . . . Correlations Among the Endogenous Indicators (N=99). . . . . . Correlations Among the Exogenous and Endogenous Indicators (N=99) . . . . Correlations Among the Indicators of ml and n2 (N=99) . . . . . . . Correlations Among the Indicators of ml and n3 (N=99) . . . . . . . Correlations Among the Indicators of n2 and n3 (N=99) . . . . . . . A Confirmatory Factor Analysis On the Indicators of E (N=99) . . . . . A Confirmatory Factor Analysis On the Indicators of mi (N=99) . . . . . A Confirmatory Factor Analysis On the Indicators of n2 (N=99) . . . . . A confirmatory Factor Analysis On the Indicators of n3 (N=99) . . . . . Canonical Correlations Among the Indicators of E with the Indicators of ”1 (N=64) O O O O O O O O O Canonical Correlations Among the Indicators of n with the Indicators of _ l n (N-64) o o o o o o o o o 2 Canonical Correlations Among the Indicators of n with the Indicators of _ 1 n3 (N—64) o o o o o o o o 0 xi Page 82 103 109 112 119 121 123 124 127 128 129 130 133 134 135 Table Page 24. Canonical Correlations Among the Indicators of n2 with the Indicators of n3 (N=64). . . . . . . . . . 136 25. Estimates of Full Model Solution and A Test (N=99) . . . . . . . . . 139 I-l. Summary Statistics for the Indicators of Printime (N=99) . . . . . . . . 243 I-2. A Confirmatory Factor Analysis On the Indicators of Printime (n=99). . . . . 245 I-3. Summary Statistics for the Indicators of Newstime (N=99) . . . . . . . . 246 I—4. A Confirmatory Factor Analysis On the Indicators of Newstime (N=99). . . . . 248 xii . Thaw- o‘UH‘ J LIST OF FIGURES Figure 1. The Relationship Between Environmental Complexity and Level of Differentiation, Discrimination and Integration. . . 2. Theoretic Model . . . . . . 3. Full Model. . . . . . . . xiii Page . 15 . 99 . 102 CHAPTER I A THEORY OF INFORMATION PROCESSING Introduction The purpose of this dissertation is to present a theory of information processing which details the rela- tionship between symbolic environments and an individual's responses to symbolic environments. The fundamental pre- mise of this theory is that information processing is a measurement process individuals engage in to reduce envi- ronmental uncertainty. The theory will argue that there are only two basic processing styles which an individual can adopt to reduce uncertainty--nominal-dominant and ratio- dominant. Further, the theory will explain how the infor- mation processing style an individual adopts is largely a function of the environmental factors to which the individ- ual has been exposed. Moreover, the theory will show how one can infer which style dominates an individual's reper- toire from the individual's responses to a specific infor- mation processing task. Finally, the theory will illustrate the relationship between dominant processing style and cog- nitive structure, and argue that individuals who employ a ratio-dominant processing style have relatively more complex cognitive structures than individuals who employ a nominal- l dominant style. To accomplish the objectives specified above, the remainder of this chapter will be devoted to the presenta- tion of four topics. First, a review of the work concerned with the topic of cognitive structure will be presented. A chronological approach will be taken to explicate the development of this variable, from its theoretic under- pinnings to the state-of—the-art models which now influence social-psychological thought. Second, a review of the work concerned with the distributional patterns of individuals' responses to information processing tasks will be presented. In particular, this section will closely examine the manner in which individuals respond to questionnaire items. These first two sections are structured so as to acquaint the reader with the fundamental ideas concerned with informa- tion processing style and cognitive structure. Next, the third section will detail the problems of prior research in the area of information processing and cognitive structure. These ideas will then provide the framework for a theoretic reformulation of human information processing to be pre- sented in the fourth section of this chapter. This reform- ulation will focus on how information processing is accomplished, and the effect of each dominant style of in- formation processing on the individual's cognitive struc- ture. Key theoretic constructs will be cast into relationships that will be used to generate a causal model. 1” “din From this model, specific hypotheses concerning information processing style and cognitive structure will be presented. Finally, a summary of this chapter will be provided. A Review of the Work on Cognitive Structure Theoretic roots. Much of the contemporary research on the topic of cognitive structure is based upon the social- psychological work of Lewin (1935, 1936, 1948, 1951), Heider (1946, 1958) and Kelly (1955). Lewin's Principles of Topo- logical Psychology introduced the concept of cognitive com- plexity as the differentiation and integration of the individual's life space. For Lewin, the individual's "life space” referred to the internal organization of the environ- mental stimuli to which the individual had been exposed. This "life space" was conceived by Lewin to be comprised of regions which become articulated (i.e., differentiated) into sub-regions as a result of diverse experiences. As Lewin (1936) writes: It is typical of the process of orientation into a new environment that the regions which are at first unclear gradually become clearer. The degree of clearness is an essen- tial determinant of the cognitive structure of the life space. It is closely related to the degree to which one can differentiate the life space into different regions. (p. 39) In addition to the process of differentiation, Lewin notes that the individual is also capable of integrating regions ()1- VA, 53913 :11 I! 51935) Cc 4 (>15 1:11e life space in order to gain understanding of the environment. Lewin (1951) comments: The increasing differentiation of the life space into relatively separated subparts is somehow counteracted by the increasing organization of the life space. . . . It refers to the increasing scope of coexisting parts of the life space which can be organ" ized as a unit and the increasingly larger sequence of actions which are unitedly governed. (p. 108) Levvj.r1's major contributions to the development of a theory of c:c>gnitive structure are these concepts of differentia- tirarx and integration (Yum, 1979), and his conceptualization of g>ssychological processes using topological principles. This; latter point is important because: (a) it represents an attempt to apply a spatial metaphor to the study of a sociaall science problem, and (b) as we shall see, most of the C:C>ntemporary theories of cognitive structure are based on SE>E3fltia1 (i.e., tOpological) principles. As it turns out, this same approach will be adopted in the theoretic and Operational reformulations to be presented. (1936) As Lewin comments: The nature of things whose system constitutes a mathematical space is entirely irrelevant for modern mathe- matics. It does not matter whether one thinks of them as physical objects, temperatures, numbers, colors, events, or anything else. Only certain rela- tionships and the possibility of certain operations are relevant. It is these which finally define space. As far as mathematics is concerned there is 5 therefore no fundamental objection to applying the mathematical concept of space to psychological facts. The crucial point is whether the relation— ships that characterize space in mathe- matics can be applied adequately to psychological facts, and whether one can coordinate psychological processes uni- vocally to mathematical operations- (pp. 52-53) 2X11 .janortant extension of Lewin's Gestalt perspective of psychological processes is the theoretic work of Fritz 13€2:i.cier'(1946, 1958) and George A. Kelly (1955). Among Heider's many contributions, probably the most relevant to (3111? concerns here is his work on attribution theory. In an attempt to understand how attitudes developed, Heider PCS ited that individuals processed information by defining their experiences according to the properties (or attributes) ‘tljualit they perceived the phenomena of their experience to p":>S>sess. Further, Heider contended that an individual's understanding of the environment, and the people in it, Could be determined by examining the attributions the indi- Vi dLlal made with respect to these stimuli. Utilizing this 1(635?’ :notion, Kelly developed a theory of personal constructs 1111’ “Vlnich he contended that individuals engaged in a process of construing (or interpreting) events, by observing the Sinai larities and/or differences among them, and developing a set of "constructs" (bi-polar adjective continua) for Ci - e fining these events. As Kelly (1955) writes: 6 In construing, the person notes the features in a series of elements which characterize some of the elements and are particularly characteristic of others. Thus, he erects constructs of similarity or contrast. Both the simi- larity and contrast are inherent in the same construct. (p. 50) Instruction system which organized a finite number of cii.c:rlotomous constructs into a meaningful hierarchy, consist- ‘itifig' (of many levels of ordinal relationships between con- 55tl1TIJCtS, which can vary as the individual further construes ‘3‘7éerats. From these initial studies have come a multitude of trl€3C>retic models concerning how individuals organize informa- t41<2>ra (for a good review of these models see: Goldstein and E31-éi<:kma.n, 1978; Streufert and Streufert, 1978; Scott, Osgood ‘aJ‘<51 Peterson, 1979). A convenient framework for examining these models is to distinguish the static-structural models from the dynamic-processing models. The static-structural models. The term "static- sst135‘lactural models" refers to that class of models which Silt13E>lyattempt to specify the components of the individual's co grlitive structure, and which do not include "time" as a variable or index. The work of Bieri (1955, 1961, 1966, JIEDIT711.), Crockett (1965) and Zajonc (1960, 1968) is represent- ative of this class of models. The common thread among the V'ca. :I:]1?<23kett's work is in his conceptualization of the changes that take place in the individual's cognitive system when hie rarchical integration takes place. Here Crockett pro- Poses that individuals form impressions of phenomena by fi rs t defining the properties of the phenomena in terms of th e and then by extant constructs in the cognitive system, de - filling the properties of the phenomena using a new set of co lustructs, which are a result of the individual's 9 ,irutzeagration of extant constructs. Further, Crockett strong- .1)! crontends that the frequency of contact with non-redundant 51::eruili is the major predictor of cognitive complexity. Work by Zajonc (1960, 1968) represents the fj.1::3 t attempt to: (a) develOp a distinction between a "co gnitive universe" and ”cognitive structure" (i.e., a dj.£3‘tLiDCtiOD between the individual's overall structure and triee ssub-structures that pertain to various domains of ex- Peazrzitance), and (b) suggest the use of a dimensional model CCDrisscisting of independent constructs (attributes) treated EH5 ‘xrectors in a multidimensional space. Zajonc (1968) Stléifties: A psychological dimension is one's capacity to map consistently a set of responses onto a collection of stimuli that is itself ordered. A specific act of "perceiving" or "cognizing" a given stimulus object or event is regarded as involving the projection of the stimulus onto a set of psychological dimensions, and thereby attributing to it one value from each of these dimensions. These projected values, attributes, are the elements of the cognitive structure under analysis. They are what is commonly understood by the traits, characteristics, qualities, etc., of the object, event, or con- cept as the person perceives them. (p. 328) T . . . bee ‘Jse of a dimensional model, and the distinction between t . . . . . tie: :Lndiv1dual's overall structure and particular domains are :important because they allow: (1) the possibility of an 10 individual having a relatively complex structure with re- spe ct to one domain, and a relatively simple structure with respect to another, and (2) the use of factor analytic techniques to develop dimensional models that can represent certain organizational features of the individual's cogni— ti ve structure. This latter point was an important influ- ence on the work of Schroder, Driver and Streufert (1967) , We xler and Romney (1972), and Scott, Osgood and Peterson (1 9 '79) , all of whom developed models of cognitive structure baSed upon factor analytic principles. Summary of the static-structural models. While there has been a great deal of variety in the approaches discussed thus far, there have been three commonalities in the concep- tUal explications which we can review. First, the basic Premise of each of these models is that as individuals en- CO unter phenomena, they define these stimuli according to the properties (attributes) they perceive the stimuli to p03$.ess. In general, cognitive theorists use the number of independent attributes an individual utilizes to define stimuli as a primary indicator of the complexity of the ind:ividual's cognitive structure. That is, the more inde- perudent attributes the individual employs, the more rela- ti\’ely complex the cognitive structure. In passing, it s1‘CPIIlld be noted that two variables have been discussed thus far = (l) the total number of attributes that the individual utilizes to define a set of stimuli, and (2) the total 11 nanber of independent attributes that are used to define the set of stimuli (i.e., the dimensionality of an attri- bute space). The process of defining stimuli according to the properties (attributes) they are perceived to possess will be referred to as differentiation. Second, static- structural models emphasize the importance of examining the degree to which an individual makes distinctions among stimuli defined along the same attribute as an indicant of complexity. The capability to make distinctions along any attribute, sometimes called articulation, will be referred to as discrimination. In general, the more distinctions an individual makes, the greater the discrimination. Third, the static-structural models examine the kind of relation- ships an individual establishes among stimuli. The cap- abi lity to combine information and establish relationships among stimuli, such that the individual can generate several perspectives as to how the stimuli and attributes inter— relate, will be referred to as integration. In general, the more different ways an individual perceives certain s‘b‘JIJ-T‘uili in relationship with other stimuli, the greater the level of integration. From the static-structuralists' perspective an indi- vicilzlal's cognitive structure can be considered relatively complex when: (a) the individual uses a large number of att-J’l‘ibutes to define a particular set of stimuli (differen- Nl. (b) the individual makes fine discriminations 12 among the stimuli, and (c) the individual exhibits a high level of integration. Further, these variables are expect- ed to be correlated with each other, such that an individual who utilizes a large number of attributes to define a set of stimuli is also likely to exhibit a high level of dis- crimination and integration. As it turns out, the theoret- ical reformulation that will be presented will contend that these three variables (differentiation, discrimination, and integration) are the fundamental aspects (indicators) of an individual's cognitive structure. The dynamicgrocessing models. The term "dynamic- pro cessing models" refers to that class of models which at- te mpt to examine the manner in which the components of the CO 9nitive structure change over time. Based upon the work of Murphy (1947), Piaget (1926, 1932, 1952, 1954) and We rner (1957) , these developmental models focus on the interaction between environmental complexity and the com- plentity of the individual's cognitive structure. These mode ls are considered here to be dynamic because of their emphasis on interaction processes and responses to forces. Representative of this class of models is the work of Harvey, Hunt and Schroder (1961) , Schroder, Driver and Streufert (1967) , and Streufert and Streufert (1978) . Harvey, Hunt and Schroder (1961) consider cognitive complexity the result of a process which entails the acqui- s' . . . - ltlon of increaSingly abstract constructs, produc1ng a l3 cchggriitive system which is highly differentiated and which j.£; (:haracterized by a high level of discrimination and integration. Commenting upon this process of abstraction, Harvey, et a1. write: We assume that the most important structural characteristic is the degree of concreteness or abstract- ness. The more concrete, the struc— ture is assumed to be restricted to, or dependent upon, physical attri- butes of the activating stimulus. (p. 3) They specifically define abstraction as: . . . how the ambiguous or undiffer- entiated is broken down or differen- tiated into parts and then integrated or interrelated into a conceptual pattern. (p. 22) We assume that learning occurs through a process of differentiation and inte- gration, during which time the person breaks down the environment into parts relevant to his current conceptual structure and then integrates these parts in ways compatible with his current organization. (p. 4) U1-1:imately, Harvey, et al. posit that each individual pass- es; izhrough stages of development that are characterized by ce31‘tain (hierarchical) levels of complexity. However, they nc’tlee that development may be arrested at any level due to St1<211 factors as changes in training, or environmental com- pleEiatity. As these authors comment: A given level or stage of concreteness- abstractness is attained through dif- ferentiation-integration, but this same 14 level or stage of abstractness affects both what kind of differentiations or discriminations are made and how these are subsequently organized or integrated by their being related to existing internal standards- (p. 23) Schroder, Driver and Streufert (1967) consider cog- rii_1:;ive complexity to be "the complexity of the schemata that determines the organization of several dimensions in a complex cognitive structure" (p. 65). Schroder, et al. f1:>cziis their theory around the three variables of differen- tli.£astion, discrimination, and integration, and how these \VEIJC2iables vary with respect to different levels of environ- nflezrirtal complexity. Interestingly, these authors posit that <35L:f?ferences between relatively simple and complex individ- ‘16351.s can only be uncovered when they are presented with an in formation processing task which presents an "optimally" <3C>Iruolex set of stimuli (i.e., when the task presents stimu- :L5L inhich is neither "underloading" nor "overloading"). tr}1€e:ir rationale is that a too extreme level of complexity plrcDduces stressful situations in which no meaningful dif- fe rentiation, discrimination, or integration should be ex- pected to occur. However, when an "Optimal" environment is; I>resent, individuals should exhibit differences in pro- c6253using style, particularly with respect to their level of differentiation and integration. More specifically, Sel"Iroder, et a1. expect relatively complex individuals to '3t3‘1ain a higher peak point on an inverted U—shaped curve 15 which relates environmental complexity (on the ordinate) 13c> (Bifferentiation, discrimination, and integration (on 1:riea abscissa). Further, they expect relatively complex individuals to attain this peak at a higher level of en- xhihzrc1nmental complexity than relatively simple individuals (see Figure 1). 1» JL£exmfl.of Di fferentiatior fl, ,_ '_ Discrimination ’ ~ ‘ 1! Environmental Complexity ---- relatively complex individuals relatively simple individuals Figure l. The Relationship Between Environmental Complexity and Level of Differentia- tion, Discrimination and Integration 16 Recent work by Streufert and Streufert (1978) has expanded upon the work of Schroder, et al. to include a c:c>11:sideration of how individuals adapt their information p>1r<><=essing behaviors to meet varying levels of environ- rneexnrtal complexity. However, rather than use the term "e211dvironmental complexity," Streufert and Streufert discuss ezrithironmental conditions in terms of the relative amount of 54r1<2<3ngruity that is present at any one time. In discussing this notion of incongruity these authors write: One could hypothesize that organisms form expectations concerning the probable amount of incongruity they will encounter. . . . By "general incongruity" is meant the total amount of novelty, imbalance, disso- nance, disagreement, failure, con- flict, etc. which an individual typically encounters. . . . Organisms could average their prior general in- congruity experience over time and thus develop general expectations con- cerning the ”normal" (consistent!) amount of general incongruity to ex- pect in their environment. The expec— tation concerning general incongruity can be termed the General Incongruity Adaptation Level (GIAL). Organisms with past experiences rich in general incongruity would develop high GIALs, and those with relatively constant pasts would evolve with low GIALs. Both would define points or ranges where the incongruity to which they may be exposed would be experienced as consistent. (pp. 172-173) Stll=£eufert and Streufert go on to posit that: (a) during eaI‘ly cognitive stages, the GIAL will be subject to radical $115—fts as general incongruity varies, due to the fact that 17 t;11ngruity to offset the level established on the basis of F>Ireaxrious experience. Finally, Streufert and Streufert argue that, in general, individuals with a moderate GIAL V95.Ilfil tend to be relatively more cognitively complex. This 3153 Ibecause an individual with a moderate GIAL will typical- :L§?' encounter environmental conditions that allow the ulti- li zation of complex search activities. As these authors conunent: A very low GIAL would not allow much incongruity to occur before cloze actions predominate.1. . . Similarly, a person with a very high GIAL may be so continuously overloaded that his ability to develop or utilize multi- dimensional thought processes could be seriously reduced. . . . A moderate GIAL . . . with its better balance be- tween load and complex search-producing incongruity levels may aid in the development of multidimensionality. . . . This activity of the moderate GIAL person may, in the long run, result in the development of higher complexities. (pp. 276-277) ‘HEPIltze, these authors contend that a balance between environ- mel'ltal conditions and the individual's cognitive structure 1ea-ds to the development to a relatively complex structure. 18 Summary of the dynamic-processing models. As we have seen, the dynamic—processing models focus on the de- velopment of an individual's cognitive structure. Basic- ally , the emphasis of these models is on the interaction be tween the individual's cognitive structure and the envi- romnental conditions that the individual experiences. In general, the more diverse experiences the individual has, and the greater the complexity of the environment character— izing those experiences, the more likely the individual is to develop a relatively complex cognitive structure, and exhibit behavioral responses which expose him/her to envi- ronmental conditions which correspond (or complement) the COmplexity of his/her cognitive structure. A summary of the work on cognitive structure. The p‘15.:‘1305e of the preceding sections was to provide an intro- duction to the literature on cognitive structure, the major cri terion variable in the theory to be presented. This introduction first discussed the theoretic roots found in tlQ work of Lewin, Kelly and Heider. Next, two classes of Qcratic) behavior. More specifically, these authors PMDES€3d the following research questions: If a potentially Fascist individual exists, what, precisely, is he like? What goes to make up anti-democratic thought? What are the organizing forces within the person? If such a person exists, how commonly does he exist in our society? And if such a person exists, what have been the de- terminants and what the course of his development? (Adorno, et al., 1950, p. 2) IPIij.nnarily as a result of Nazism, anti-democratic style was iiiaxrsst Operationalized as prejudice toward Jews in the Anti- SGEITIj.tism scale developed by Levinson and Sanford (1944). CDtifleer scales studied at about this time by these investi- gators were the Ethnocentrism Scale (Levinson, 1949) , the pc>31.:"Ltico--E:conomic Conservatism Scale, and the Fascism Scale (reviewed by Adorno, et al., 1950). In general, these sczEiIlLes were Likert-type scales in which the respondent is asJfied to indicate the degree to which he/she agrees (or di Sagrees) with a particular statement (e.g., "Jews are J:'..“‘*":11'11ess"). Further, statements were posed such that a rle‘SEflu score on these items would reflect extreme prejudice alljl‘fi high authoritarianism. Finally, investigators have t:§?>3§>ically found high positive correlations among these seales (e.g., Adorno, et al., 1950; Christie, 1954; Rule 7‘51 Hewitt, 1970), providing some evidence for the valid- 51 t3? of these measures. 24 While many interesting findings have been generated lxy' t:he research on authoritarianism (for a crisp review see 13c>].615tein and Blackman, 1978), our discussion here will be rtessizricted to those results which are concerned with the reasszponse patterns exhibited by authoritarians. Further, trizreee particular response patterns examined by the above reasseearch seem pertinent to our concerns: (1) intolerance fc>1r ambiguity, (2) rigidity, and (3) response-set acqui- esiczeence. Research clearly indicates that highly authori- tai:::ian individuals tend to be less tolerant of ambiguity (aridi more rigid than less authoritarian individuals. Com- menting upon these two findings, Goldstein and Blackman (l 9 '78) write: A person who is intolerant of ambiguity is likely to make infrequent use of limiting and qualifying language. The concept of rigidity refers to thought and behavior that is exceptionally re- sistant to modification. Rigidity was evident when the authoritarian individu- al refused to relinquish ethnic stereotypes when faced with information contradicting the stereotype. Another characteristic of rigidity is that the individual's cognitions are compartmentalized and walled-off from each other, resulting in an apparent lack of consistency. (p. 19) w , es: JLnterpret these findings to indicate two important t: - lulu«rugs. First, high authoritarians tend to make extreme j udgments. That is, they do not use qualifying language a; ‘hd when provided with a bounded Likert-type scale, they t: Qrid to use the end points of these scales. Second, high 25 authoritarians tend to have relatively simple cognitive structures. That is, we interpret the authoritarian tend- encies to employ stereotypes as an organizing scheme for others, to adhere to stereotypic patterns even when pre- sented with conflicting information, and to have difficulty inter-relating cognitions, as reflective of a simple cog- nitive structure. Finally, with respect to authoritarianism, we come to the issue of response-set acquiescence. As was discussed earlier, the scales that were developed to tap authoritari- anism were Likert-type scales wherein agreement with each item contributed to a higher authoritarian score. One problem with this type of scale is the tendency for individ- uals to respond to the questionnaire item in an agreeing (or acquiescent) manner, regardless of the content of the prompt. Consequently, some researchers argue that these (authoritarianism) scales measure authoritarianism as a result of the authoritarian's tendency to acquiesce (e.g., Cohn, 1953, 1956; Chapman and Bock, 1958; Chapman and Campbell, 1959; Zuckerman, Norton and Sprague, 1958), while others dispute the degree to which authoritarian scores can be attributed to acquiescence (e.g., Brown and Datta, 1959; Couch and Keniston, 1960; Clayton and Jackson, 1961). Nonetheless, it appears that at least some amount of vari- ance in authoritarian scores is due to acquiescence, and that individuals who acquiesce are also likely to be authori then, '1 points extreme acquie: author anism ser'v'at 1Y0 a! 26 authoritarian (Messick and Frederiksen, 1958). In sum then, it appears that the research on authoritarianism points out that high authoritarians tend to make: (a) extreme judgments, (b) categorical judgments, and (c) more acquiescent responses than low authoritarians. Dogmatism. One of the main criticisms of the authorianism scales was that they only measured authoritari- anism of the "right." That is, that they focused on "con- servative" authoritarian beliefs (Shils, 1954). Consequent- ly, an effort was undertaken, primarily by Rokeach (1951, 1954, 1956, 1960), to develop a measure of cognitive style which would reflect close-minded (i.e., dogmatic) beliefs regardless of their content. Stated somewhat differently, Rokeach attempted to develop a measure of cognitive style which would reflect authoritarianism of the "left" as well as the ”right.” According to Rokeach, individuals vary along an open-minded-close-minded continuum. He posited that a person who was extremely close-minded (i.e., highly dogmatic): (a) polarized his/her beliefs into a belief- disbelief system, totally accepting all ideas (and peOple who hold those ideas) in the belief region, and totally rejecting all ideas (and people who hold those ideas) in the disbelief region, (b) maintained an authoritarian per- spective, glorifying authorities who supported his/her be- lief region and supportive of an elite class, and (c) was likely to be dogmatic across all domains of experience. 27 To assess this form of cognitive style, Rokeach developed the Dogmatism Scale (see Rokeach, 1960). Similar to the authoritarianism scales discussed in the previous section, the Dogmatism Scale was a Likert-type instrument which provided a prompt (statement), and asked the respond- ent to place a checkmark along a six-point scale to express the degree to which he/she agreed or disagreed with the prompt. Because of the wording of the prompts, agreement was scored as a reflection of close-mindedness, while dis— agreement was scored as Open-mindedness. Unlike the author- itarianism scales with their concern for prejudice beliefs, the Dogmatism Scale focused on measuring the individual's belief—disbelief system; perceived adequacy of self; un— certainty concerning the future; perceived friendliness of the environment; perception of authority; intolerance of conflicting ideas; and time orientation (i.e., a concern for the past, present, or future). A high score across these items reflected a highly dogmatic cognitive style. To determine whether or not Rokeach was successful in his attempt to measure dogmatism, investigators experi- mented with different research strategies. For example, one strategy was to examine the relationship between dogma— tism and authoritarianism (e.g., Schroder and Streufert, 1962; Zippel and Norman, 1966; Sheikh, 1968). Another strategy was to examine the relationship between dogmatism and political ideology (e.g., Di Renzo, 1968; Parrot and 28 Brown, 1972; Thompson and Michel, 1972; Steininger and Lesser, 1974; Stimpson and D'Alo, 1974). In a review of the different research strategies that were employed, Gold- stein and Blackman (1978) report that these studies show: (a) a "consistent positive relationship" between dogmatism and authoritarianism, and (b) that high dogmatism is asso- ciated with the political orientations and attitudes of the extreme "right wing." As a result, Goldstein and Blackman conclude that "Rokeach's attempt to develop an instrument that would be equally sensitive to dogmatism of the right and left was not successful" (p. 70). Nevertheless, even though the Dogmatism Scale does not completely satisfy its objective to be content free, it does provide us with some useful information about response patterns and cognitive style. First, because the Dogmatism Scale is structured like the authoritarianism instruments, the same observation concerning response set acquiescence has been made for these scales (e.g., Couch and Keniston, 1960; Peabody, 1961; Roberts, 1962). That is to say, highly dogmatic individuals tend to be more acquiescent. Second, while relatively few studies have been conducted with re- Spect to examining the relationship between dogmatism and intolerance for ambiguity, the research evidence indicates that highly dogmatic individuals are less tolerant of ambi- guity than less dogmatic individuals (e.g., see Barker, 1963; Feather, 1969; MacDonald, 1970). Third, parallel to the a indiv and ( 29 the authoritarianism research findings, highly dogmatic individuals tend to be more cognitively rigid (e.g., Korn and Giddan, 1964; Schroder and Streufert, 1962; Hession and McCarthy, 1975; White and Atler, 1965). In sum then, it appears that dogmatism research findings closely paral- lel the research findings cited for authoritarianism, and points out that highly dogmatic individuals tend to: (a) make more acquiescent responses, (b) make extremely polar- ized judgments, (c) be less tolerant of ambiguity, and (d) be more cognitively rigid. Cognitive structure and human responses. Since the theoretic work with respect to cognitive structure has already been reviewed,2 this section will simply discuss two key findings that have been generated with respect to response patterns. Probably the most important finding with respect to response pattern is that cognitively simple persons tend to use extreme scores on semantic-differential items more frequently than cognitively complex persons (White and Harvey, 1965; Sawatzsky and Zingle, 1971; cf., Nidorf and Argabrite, 1970). This result is interpreted here as reflective of the individual's discrimination abilities. More specifically, cognitively simple persons make fewer discriminations among stimuli and hence, operate at a more "concrete" level of information processing in which stimuli are defined terms of dichotomies. More- (over, this style of cognitive functioning is interpreted as the ma Meme indivi Furthe foils Suppo Harve towar toler inte tent Com; 30 the major factor determining dichotomized (i.e., extreme) judgments on questionnaire items. A more complex individual makes finer discriminations among stimuli. Further, such an individual uses more alternative response foils when presented with a series of questionnaire items. Support for this contention is provided by Miller and Harvey (1973) who report that "concrete" individuals tend toward extreme and polarized judgments, and are less tolerant of ambiguity. The second finding to be discussed here is the observation that relatively complex individuals tend to use more "constructs” in describing stimuli. This observation, supported by the research of Campbell (1960), Supnick (1964) and the several studies reviewed by Crockett (1965), is interpreted here to indicate strong support for the con- tention that differentiation is a prime indicant of the complexity of an individual's cognitive structure. Further, this observation supports the contention that an individual's differentiation capability can be empirically assessed by the individual's performance on particular information processing tasks. Summary: the distribution pattern of individuals' responses. The purpose of this section was to review the relevant research concerned with the patterns of individual's responses to information processing tasks as they relate to upor ity lite are an 31 cognitive style. Three areas of investigation were focused upon: (1) the psychophysical literature, (2) the personal- ity theory literature, and (3) the cognitive structure literature. The review of the research in these three areas yielded several interesting findings. First, there are clear differences in the manner in which individuals process information (e.g., some individuals tend to use certain numerical responses on a questionnaire task more frequently than other responses). Second, there exist in- dividual differences in the use of polarized, stereotypic judgments and extreme scores on questionnaire tasks. These tendencies are consistently associated with authoritarian- ism, dogmatism, conservatism, intolerance for uncertainty, and rigidity. As White and Harvey (1965) comment: . . . the tendency for the individual to dichotomize his psycholoqical scale and pile up his judgments of the issue at the ends of the scale instead of distributing them more evenly over the entire range of the scale . . . would be predicted from at least three different but related personality theories, that of Adorno, Frenkel-Brunswik, Levinson, and Sanford (1950), of Rokeach (1960) and of Harvey, Hunt, and Schroder (1961). Greater author- itarianism (Frenkel-Brunswik, 1949; Adorno, et al., 1950), higher dogmatism (Rokeach, 1951, 1960), or greater concreteness (Harvey, et al., 1961) each of which is presumed to be underlaid by poorly differ- entiated and integrated cognitive struc- tures, should dispose the individual toward the usage of more absolute, more undiffer- entiated, and more discontinuous internal standards. (pp. 334-335) expre: r1356 « Push . s 32 Third, given the psychophysical research findings on the manner in which individuals perceive the magnitudes of certain stimuli, serious consideration must be given to re- expressing raw data using some transformation if meaningful patterns are to be uncovered. Moreover, it is not unlikely that several different transformations will have to be experimented with before meaningful patterns are extracted from the raw data of questionnaire responses. That is to say, the transformations that have been of high utility to psychophysicists may not necessarily be the same ones which best re-express questionnaire data. These issues will be dealt with in the next few sections. The Problems with Prior Research In the next section a formal theory of human infor- mation processing will be presented. Essentially, this work will be a theoretical and operational reformulation of the work described in the previous two sections. Conse- quently, this section will first review the problematic aspects of prior conceptualizations and operationalizations. While there have been numerous attempts to examine the relationship between information processing and cog- nitive structure, these previous efforts have suffered from several shortcomings. First, there has been a general weakness in conceptualizing the indicators of cognitive structure. Commenting upon this situation, Streufert and Streufert (1978) write: .‘IJ def tha 1.)) E va: Ta wi 33 The term cognitive complexity has appeared with increasing frequency and increasing confusion in the psy- chological literature. In reviewing the literature, both theory and research, the reason for the con- fusion becomes apparent. . . . The ways in which the theories differ are many, including the underlying assumptions upon which they are based and the definitions of terms. (p. 12) For example, with respect to the problem of the conflicting definition of terms, Scott, Osgood and Peterson (1979) note that several different labels have been used to refer to the concept of "attribute” including: personal construct (Kelly, 1955), dimension of judgment (Anderson, 1971), cue variable (Hammond, 1972), and trait (Bruner, Shapiro and Taiguri, 1958). Sometimes however, the problem worsens with definitions simply lacking. For example, Scott, et a1. note the work of Asch (1952), and Kretch and Crutchfield (1948) who ". . . used the term cognitive structure quite frequently; nevertheless, it was left undefined, and was used for a variety of purposes. Nowhere did these authors offer a specific designation of just what was structured" (p. 34). So then, without delving further into detail, the first major problem noted with previous research efforts has been the weak theoretical development, as indicated by the weaknesses in the conceptualization of indicators and the specification of relationships. Second, there has been a general weakness in opera- tionalizing the indicators that have been developed. That is, t) scale and S 1966; lack manne Golds bloci theo nota ers dome Stre lit St mo 34 is, the connection between the theoretic indicator and the scale item(s) is not always clear (e.g., see Harvey, Hunt and Schroder, 1961; Schroder and Streufert, 1962; Harvey, 1966; Tuckman, 1966). Further, there has been a general lack of consistency across investigators with regard to the manner in which constructs have been operationalized. As Goldstein and Blackman (1978) write: "The major stumbling block to widespread research on integrative complexity theory is the problem of measurement" (p. 172). Most notably, these authors are critical of the lack for research- ers to develop an instrument that can be used across domains, and individuals. Nevertheless, as Streufert and Streufert (1978) point out in their review of the literature: The most striking aspect of complexity as a phenomenon in psychology is the inconsistency among the various measure- ment techniques, yet the similarity of experimental results. While different researchers have made similar predictions, they have used diverse (and uncorrelated) measures of complexity as a personality characteristic. Nonetheless, they have typically obtained identical results. (pp. 68-69) Streufert and Streufert conclude that this observation is most likely a result of the different measures being . . . geared toward separate and differ- ent components of complexity, and that each of these components is likely to produce the expected differences in per- ceptions or behavior. (p. 69) Hence effor hensi ture Whil Sele tors and SeeIr POSS that 35 Hence, the second major problem with previous research efforts has been with the clarity, consistency and compre- hensiveness with which the indicators of cognitive struc- ture have been operationalized. Third, there has been a predominant tendency to employ imprecise scaling procedures in the operationaliza- tion of indicators. The major problem here is that the nominal and ordinal scales that have been most popular among cognitive theorists have severe limitations with re- spect to the amount and type of information which they can provide. As Scott, et a1. (1979) remark: Psychologists have developed rather arbitrary measuring procedures, such as Likert scales and Thurstone scales, relying on particular sets of questions that may be apprOpriate to one group, not another, applicable today, but not tomorrow. The scores yielded by [such] attitude scales typically depend on the number of items with which the subject agrees, and this in turn depends more than anything else on the number and type of items in the scale. (p. 80) While all measurement is, to a degree, arbitrary, the selection of a scaling procedure to operationalize indica- tors is important in that it largely determines the quality and quantity of the information gathered. Consequently, it seems important to try and create the most precise scales possible, and the problem with prior research efforts is that this has not been carefully attempted. inv me 321 I "21 PT I! 36 Finally, and largely a consequence of the above, investigators have been limited in the methods that they have employed to seek out meaningful patterns in the data (e.g., see Cronbach, 1955; Wyer, 1964; Crockett, 1965; Schroder, Driver and Streufert, 1967; Wexler and Romney, 1972). What is important to realize is that a variety of transformations can be used to find the correct functional form of the relationship between two (or more) variables when the scales which are employed yield data rich enough to apply them. However, when imprecise scales are used, only a small number of transformations may be "appropriate- ly" applied. This, then, implies the need to use scales which allow the greatest number of transformations. A Theory of Information Processing Information processing: the detection of change. From birth, individuals are part of an environment in which a multiplicity of processes are taking place. Each human is equipped with a set of sensory organs which are sensi- tive to certain ranges of fluctuation in the environment. Not all processes which take place in the environment are capable of being directly monitored by the human sensory mechanisms. That is, while some processes generate changes in the environment which stimulate one or more of the indi- vidual's sensory organs, others are beyond human perception. Among the set of sensory mechanisms which an individual p05: ity sou sur tic r1 1:! 37 possesses, five are generally distinguished: (l) sensitiv- ity to light wave fluctuations (sight), (2) sensitivity to sound wave fluctuations (hearing), (3) sensitivity to pres- sure fluctuations (touch), (4) sensitivity to fluctua— tions in the air's composition (smell), and (5) sensitivity to chemical reactions which take place when stimuli inter— act with the enzymes of the mouth (taste). Using this per- spective, human experience may be conceptualized as the processing of changes. The interpretation of our experience is facilitated by the brain's capacity to store information (signals), and to utilize this stored information in the future to make comparison with other signals. As reported by Guyton (1971), it is believed that patterns of sensation are stored as a result of neural signals providing similar (or redundant) spatial and temporal stimulation to the cortex of the brain. That is, these neural signals stimulate certain areas of the brain in a particular sequence. Over time, an individual is likely to encounter various environ- mental conditions which stimulate his/her sensory organs to generate similar and/or redundant neural signals. Thought of in this way, memory may be conceptualized as a process in which neural signals stimulate the cortex of the brain, such that cognitions of previous experiences are evoked. Further, the previous experience(s) serve as the basis for identifying similarities and/or differences in the present Sig: Sig: sen: re ft sen as ext tic tic St: St: the SE! pr- du ni be ts it 01 38 signal. The phrase "redundant neural signals" refers to signals which stimulate the cortex to provide equivalent sensations, while the phrase "similar neural signals" refers to signals which stimulate the cortex to provide sensations of greater and/or lesser magnitude (or intensity) as a previous signal. An individual's ability to survive is, to a large extent, dependent upon his/her ability to monitor varia- tions in the environment, interpret the resultant sensa- tions, and respond with behaviors. It appears that the primary manner in which an individual accomplishes these tasks:n3by directing his/her attention toward particular stimuli, and forego the processing of other stimuli. Stimuli which exhibit large fluctuations tend to dominate the individual's attention. Beyond these phenomena, it seems that individuals selectively attend to stimuli which provide similar and/or redundant sensations. In a funda- mental sense, we may say that information (signals) can re- duce environmental uncertainty when the individual is cog- nizant that redundant and/or similar neural signals are being processed. Stated somewhat differently, environmental uncertainty is reduced when an association can be made be- tween two (or more) experiences. Moreover, as this capabil- ity to make associations develop, the individual begins to notice that particular sensations either precede, co-occur, or follow other sensations. This is important, for once a tine ence ual no tab exp ces ent me: nm ti l9 fu Wh 39 time-order can be established between two (or more) experi- ences, the individual can begin to learn about the process— ual nature of the environment. Information processing as a measurement procedure. In the preceding section, it was argued that environmental uncertainty could be reduced when the individual could es- tablish a correspondence (association) between two (or more) experiences. Conceived of in this manner, information pro- cessing is essentially a measurement procedure individuals engage in to reduce environmental uncertainty. At the mention of the phrase "measurement procedure," typically one thinks in the abstract, and considers it a process wherein numbers are assigned to represent the attributes or proper- ties of the phenomena of our experience (e.g., see Campbell, 1938; Russell, 1938; Stevens, 1951). However, at a more fundamental level, numbers are simply one set of symbols which can be utilized to represent cognitions. From our perspective, any time a correspondence can be established between the elements of two sets, such that the elements of the first set are used to refer to (i.e., designate) the elements of the second set, a measurement procedure has been accomplished. Further, it is argued that there are only two fundamental measurement procedures which can be employed to reduce environmental uncertainty: nominal procedures and ratio procedures. The next few paragraphs will discuss each of these procedures, and the manner in which they facilitate UDC( t0 nit pie 0 D-‘h he an t5 li E's} 40 uncertainty reduction. With human memory, the elements which are utilized to represent stimuli (or attributes of stimuli) are cog- nitions. While the processes of human memory are too com- plex to detail here, in general, it can be described as a process in which the individual establishes a one-to—one correspondence between a referent and a cognition (for a fuller discussion, see Lawson, 1967). That is, the human brain can preserve cognitions of various referents, and can evoke them by similar and/or redundant neural stimulation of the cortex. Further, there are two important aSpects of human memory which are relevant to our discussion. First, while there is sometimes a high iconicity between cognitions and their referents, this is not always the case. That is, the human brain does not store exact replications of stimu- li, but rather, stores either: (a) neural signals that evoke cognitions of particular features of the stimuli, or (b) neural signals that evoke cognitions that are not iconic with the referent, but which are utilized to desig- nate the referent (e.g., the symbols of a person's name). Second, the individual's capability to evoke cognitions through similar and/or redundant stimulation of the cortex indicates that stimuli which differ somewhat from the re- ferent may be considered as part of the same gia§§_of phe- nomena, to the extent that they evoke the same cognition(s). These two aspects of human memory, when considered together, seem t the te lation to the CIitE] group: indiv. an in or re lar c reduc ident prese Vidua (or . Cedu Ence Such Hate meas Cede 41 seem to indicate a general principle that individuals have the tendency to group referents which provide similar stimu- lation, and utilize a minimum number of cognitions to refer to these referents. While referents may be grouped by any criterion, what is important to realize is that it is this grouping (or categorization) which provides meaning for the individual concerning the referents. That is to say, when an individual encounters stimuli which provide similar and/ or redundant stimulation to the cortex to evoke a particu- lar cognition (or set of cognitions), he/she is capable of reducing uncertainty in the sense that he/she can identify the presence of the referent (or at least the presence of some feature of the referent). When the indi- vidual is capable of cognizing the presence of a referent (or an attribute of a referent), a nominal measurement pro- cedure has been accomplished. In general, when a correspond- ence can be established between the elements of two sets, such that the elements of the first set are used to desig- nate the presence of elements in the second set, a nominal measurement procedure has been accomplished. Nominal pro- cedures allow the individual to group the stimuli of experi- ence, and establish correspondences between such groups. Typically, symbols are assigned to each group and utilized to designate the elements of the group. Nevertheless, the uncertainty reduction capabilities afforded by nominal pro- cedures are somewhat limited, since these procedures only allc aha red a 1 H .J( b inl 50 th 42 allow the individual to detect the presence or absence of an attribute or referent. As an individual's experiences with similar but non- redundant phenomena increase, the individual becomes in- creasingly capable of differentiating among stimuli within the same class. That is, the individual learns to consist- ently detect differences in the magnitude of certain attri- butes among elements of the same class. When the discrepancy between any two phenomena can be judged along some attribute, a £3339 measurement procedure has been accomplished. More specifically, ratio procedures are accomplished when the individual can judge the discrepancy among stimuli (along some attribute) with respect to some standard. Further, there are three types of ratio level procedures which may be employed to reduce environmental uncertainty--direct ordinal estimation, direct magnitude estimation, and direct interval estimation.3 With direct ordinal estimation the individual uses a previously experienced stimulus as the standard, and simply classifies novel stimuli on the basis of whether the stimuli possess more or less of a particular attribute than the standard.4 Hence, the individual establishes an "order relationship" among the elements of a class. So, for example, with direct‘ordinal procedures an individual can not only distinguish between hot and cold, but also among ranges (or degrees) of hot and cold. The capability to employ direct ordi1 vide proc phen exis dirl sti di ir me 43 ordinal estimation procedures is important, for they pro- vide the individual with more information than nominal procedures, and the capability to develOp expectations of phenomena not yet experienced, but which may potentially exist. Similar to direct ordinal scaling, direct magnitude measurement procedures are accomplished when the individual judges the discrepancy between two stimuli in terms of the relative magnitude (or intensity) of some attribute the stimuli possess. The subtle difference here is that with direct magnitude estimation, the discrepancy between the stimuli is judged in proportion to the amount of the attri- bute some arbitrarily defined standard possesses. Hence, the general form of direct magnitude scaling is as follows: if x possesses u units of attribute a, how many units of attribute a does y possess? With direct interval estimation, the discrepancy between any twp stimuli are judged in pro- portion to an arbitrarily selected interval of discrepancy. Hence, the general form of direct interval estimation is as follows: if there are u units of difference between a and b, how different are x and y? Consequently, while both direct magnitude estimation and direct interval estimation both use an interval of discrepancy as a standard,5 the key distinctions between these two procedures are that a direct interval procedure: (1) does not require the respondent to make judgments of discrepancy along any one attribute (dime the ‘ cans esta fcr only diff 44 (dimension), but rather, allows the respondent to judge the total (multidimensional) discrepancy among stimuli, and (2) requires the individual to make judgments among paired stimuli. Nonetheless, both direct magnitude estimation and direct interval estimation procedures are important, for they provide even more information than direct ordinal esti— mation and nominal procedures, respectively. This is be- cause direct magnitude and direct interval procedures establish proportions of discrepancy among stimuli. So, for example, with these procedures an individual can not only distinquish among ranges (or degrees) of discrepancy, but also among the proportionate relationships of these differences. Thus far, it has been argued that information process- ing is a measurement procedure individuals engage in to reduce environmental uncertainty. Further, it has been argued that there are only two fundamental measurement pro- cedures an individual may employ to reduce uncertainty-- nominal procedures and ratio procedures. Moreover, it is hypothesized that there is a hierarchical relationship among these two procedures, in the sense that ratio pro- cedures are higher order than nominal procedures because: (a) ratio procedures cannot be accomplished until nominal procedures have already been applied to define the set of stimuli, (b) ratio procedures allow for discriminations to be made among similar but non-identical stimuli, and (c) In: 45 ratio procedures generate more information, which presum- ably could be utilized to reduce environmental uncertainty. Finally, it is argued that the measurement procedure which is dominant in an individual can be determined from a re- sponse type the individual should exhibit when responding to an information processing task which would allow such dif- ferences to emerge. More specifically, the type of measure- ment procedure which is dominant can be determined from examining the distributional pattern of responses the indi- vidual exhibits when responding to a questionnaire that em- ploys direct interval estimation techniques (i.e., that uses direct-interval, paired-comparison scales). Further, we would expect the dominant use of nominal procedures to be reflected by the predominant use of extreme scores, with most scores being either zero or integer multiples of the stand- ard. On the other hand, we would expect the dominant use of ratio procedures to be reflected by a greater differentiation among scores, with scores along the entire continuum being utilized, and with a large percentage of scores 22E being multiples or proportions of the standard. The terminology ”nominal-dominant” and "ratio-dominant" will be utilized to refer to these two hypothesized information processing styles. The relationship between information processing style and cognitive structure. In the preceding section, two distinct measurement procedures were delineated, and two information processing styles which result from the denim pothe is on indit disc: are viou whic hati SOCi 46 dominant use of these procedures were discussed. It is hy- pothesized that an individual's information processing style is one of the major determinants of the complexity of the individual's cognitive structure. Because of the increased discriminatory capabilities afforded by the ratio measure- ment procedures, individuals who maintain a ratio-dominant processing style are more likely to develop a relatively complex cognitive structure than individuals who maintain a nominal-dominant style. Moreover, it should be possible to test this hypothesis by examining the relationship between processing style (as indicated by the response type the individual exhibits) and relevant indicators of cognitive structure. The following indices of cognitive structure are proposed because of their: (a) consistency with pre- vious conceptualizations, (b) isomorphism with the tasks which are accomplished when an individual processes infor- mation, and (c) ability to be observed using established social science techniques. First, it would be important to examine how many attributes an individual utilizes to discriminate among stimuli. This process of defining stimuli according to the properties they are perceived to possess was referred to as differentiation. In general, the greater the number of independent attributes an individual utilizes to define a particular set of stimuli, the greater the differentiation, and the greater the relative complexity of the individual's f \- co; the stf DC ir (is ‘ ‘w‘: is Ce v: 47 cognitive structure. Second, it is important to examine the degree to which an individual makes distinctions among stimuli. In general, the greater the discrimination, the greater the relative complexity of the individual's COgni- tive structure. Third, it would be important to examine the kind of relationships an individual establishes among stimuli and attributes. In general, the greater the level of integration, the more relatively complex the individual's cognitive structure. In sum then, an individual's cognitive structure is relatively complex when the individual: (a) uses a large number of independent attributes to define a particular set of stimuli, (b) makes fine discriminations among stimuli, and (c) exhibits a high level of integration. Further, an individual characterized by a ratio-dominant style is expected to differentiate, discriminate and inte- grate in a manner indicative of greater complexity than an individual characterized by a nominal-dominant style. While there is great variation in the way in which nominal-dominant and ratio-dominant individuals process information, it is hypothesized that over time, they will develop distinct patterns for: (a) the rate at which infor- mation is processed, (b) the complexity of the information which is processed, (c) the content of the information which is processed, and (d) the amount of uncertainty they per- ceive in the environment. This hypothesis assumes that en- vironmental conditions determine the relative complexity of the i with to p] we wc point go cc part ized rela C0ur Sent PUrF i266 that 48 the individual's cognitive structure, which in turn will determine the kind of behaviors the individual will respond with in particular environmental conditions.6 If we were to plot the patterns individuals maintained over lifespan, we would note dramatic changes in these curves at particular points in time where information processing behaviors under- go considerable transformation. These fluctuations are: (a) attributable to three types of change--environmental change, personal change and social structural change, and (b) accompanied by changes in the individual's cognitive structure. The next few sections will discuss: (1) each type of change listed above, (2) the expected changes in cognitive structure which accompnay each type of change, and (3) the remaining factors which determining the relative complexity of an individual's cognitive structure. Environmental change: adaptation to new environ- ments. Clearly, the environmental conditions experienced over lifespan vary considerably across individuals. This is particularly true in complex societies which are character— ized by several distinct classes. Nonetheless, even in a relatively simple society, individuals are likely to en- counter a wide range of environmental conditions which pre- sent varying amounts and types of information. For our ,purposes, the term "information environment" will be util- ized to refer to all those processes which produce changes ‘that may be detected by the human sensory organs. Moreover, in' 53: m lec' Ce< 1a} 49 of all components of an information environment, it is hypothesized that the one which has the greatest impact on the individual's cognitive structure are the interperson- al (face-to-face) interactions which the individual engages in with "significant others." More specifically, the factors which have the greatest impact on cognitive struc- ture are the contextual and structural components of "sig— nificant other" interactions. This contention is based upon a large body of literature in sociology which has examined "significant other" influence (for a good review of this literature, see Haller and Woelfel with Fink, 1969).7 Contextual components refer to the content of the information that is exchanged, that is, the tOpics which are discussed, the concepts which are utilized, and the rela— tionships which are established among concepts during con- versations. Contextual components are important because we would expect that the greater the diversity and complexity of the conversations the individual engages in with "sig- nificant others," the more likely the individual is to develop a relatively complex cognitive structure. Further, it is important to realize that it is during conversations with "significant others" that the individual is likely to learn how to apply either nominal or ratio measurement pro- cedures to organize the stimuli of eXperience. In particu- lar, the individual may learn how to apply nominal and/or ratio measurement procedures by either: (a) being expli obser and t other that textt devej C032} Cont the Com: div. Pom ind to th Pe 50 explicitly instructed by a "significant other," and/or (b) observing "significant others" utilize these procedures, and then adopting them (i.e., by having the "significant other" serve as a model). Consequently, it is hypothesized that the greater the diversity and complexity of the con- textual components, the more likely the individual is to develOp a relatively complex cognitive structure. The structural components of the individual's inter- actions refer to the linkages that exist between individuals, the inter-connectedness of their networks, and the rate at which information flows through the network. As these factors vary, individuals should exhibit corresponding changes in their information processing behaviors. It is hypothesized that as the structural components increase in complexity (i.e., as the number of links increase, as inter- connectedness increases, and as information flow increases), the more likely the individual is to develop a relatively complex cognitive structure. In sum then, increases in the diversity and complexity of structural and contextual com- ponents characterize an "enriched environment," and that individuals exposed to enriched environments are more likely to develop: (a) a ratio-dominant processing style, and (b) a relatively complex cognitive structure than individuals who are exposed to less enriched environemnts. Personal change: stages of development. The term "personal change" is utilized here to refer to two kinds of Chan sens and patt for ed a char pro inc it: in: Co! Vi 51 change: (1) changes in the capability of the individual's sense modalities to detect fluctuations in the environment, and (2) changes in the individual's ability to detect patterns in the environment and to store that information for future use. The first type of personal change is treat- ed as "levels of awareness," and the second type of personal change is treated as "learning." However, it is important to realize that the level of awareness changes and learning changes are related to each other, such that learning changes affect the sense modalities' sensitivity to certain fluctuations, and vice versa. While level of awareness changes and learning changes can take place continually over lifespan, probably the most pronounced periods of change occur during early childhood (0-8 years) and late adulthood (over 70). During early childhood, the individual's sensory mechanisms become more fully developed than they were at birth. That is, as the individual grows, the individual's nervous system continues to undergo development, and the individual becomes increas- ingly sensitive U3 fluctuations in the environment. These changes allow the individual to process more information, and are likely to be accompanied by increases in the complex- ity of the individual's cognitive structure. Here, the individual is assumed to be born with a "blank slate." Consequently, any processing of environmental information will necessarily lead to increases in the complexity of the 52 cognitive structure. In addition, there has been consider— able theoretic and empirical work conducted, most notably by Piaget (1926, 1952, 1954), that demonstrates that it is during early childhood that the individual undergoes con- siderable cognitive structural change as a result of learning. During later changes of the individual's lifespan, the individual's sense modalities begin to deteriorate and his/her sensitivity to certain ranges of wave fluctuation diminishes (e.g., see Restak, 1979). It is hypothesized that this type of personal change is likely to be accompa- nied by decreases in learning (relative to learning cap- abilities experienced earlier during lifespan), and by decreases in the relative complexity of the individual's cognitive structure. That is, as the individual's sense modalities diminish in their capacity to detect changes in the environment, less information is processed, less cognitive activity takes place, less previously stored information is reinforced via direct experiences, and it is less likely that the relative complexity of his/her cognitive structure increases. Further, as the individ- ual's memory capabilities diminish, the more likely the individual is to resort to the use of categorical and/or polarized judgments, and seek out less complex environ— ments (i.e., favor high stability, low uncertainty type environments, and become more cognitively rigid and less adapt Whi. fOr WOe S3 adaptable to change). Social structural change: adaptation to meet role aspirations and/or expectations. The final class of change which restructures information processing behaviors and influences c0gnitive structure are the transitions individ- uals make from one stage of the life-cycle to another. That is, there are certain key periods during the life- cycle when individuals restructure their information pro- cessing behaviors to adapt to societal expectations of par- ticular age cohorts. The transition from one period of the life-cycle to another is characterized by an environment in which: (a) the individual is changing due to maturation, (b) the individual's role-position is changing, (c) other's expectations for the individual's behavior are changing due to the changes the individual is undergoing, and (d) the individual is expected to perform coordinating activities which will correspond (or fulfill) the societal expectations for his/her behavior. In describing this environment, Woelfel (1974) writes: Other persons proxemic to the individ- ual . . . observe the individual and form inferences about his or her future potential for attainment based on these observations. At the same time, of course, the person himself or herself makes similar observations and inferences. Out of this information provided by the individual's own observation and the observations and inferences of those surrounding him or her, the individual forms conclusions about his or her likely and desired outcomes (a "self conception"), and acts accordingly, insofar as circumstances and genetic cap- abilitity permits. What is crucial in 54 understanding the complexities of this process, however, is the recognition that no two persons will . . . form pre- cisely the same expectations for the same person. Thus, any individual will be in receipt over time of many sets of expec- tations from many others, all of which differ in minor and major ways from all others and from the individual's own judgments. The question of how the indi- vidual responds to a set of multiple and disparate expectations thus becomes crucial, , , . (p. 3) Further, Woelfel goes on to state that it is the belief of most sociologists who have examined attainment processes that there are certain processes which take place in soci- ety which account for observed differences in aspiration and attainment. This assertion is based upon a large body of literature which has consistently demonstrated that in- dividuals who come from families in the higher socioeconom- ic strata tend to develop higher aspirations for education- al and occupational attainment than do individuals who come from families in the lower socioeconomic strata (e.g., see Sewell and Shah, 1967; Duncan, Featherman and Duncan, 1972). Following Haller and Portes (1973), it is assumed that the influence of social structural factors on the individual's cognitive structure is mediated by the communication patterns the individual maintains with "significant others." That is to say, that social structural factors determine the complexity of the contextual and structural components of the individual's information environment, which in turn determines the relative complexity of the individual's cog. shit foil Fir (p1 pie Sec Vi< dor Ve' CO ex Pr in de ti 55 cognitive structure. A theoretic model and hypotheses. The key relation- ships presented in this theory can be expressed by the following diagram: .- ; ——->nl——->n3 )/ 2 where: F = social structural factors HI = complexity of the individual's information environment n2 = dominant processing style employed by the individual n3 = complexity of the individual's cognitive structure First, this diagram posits that social structural factors (primarily socioeconomic status) determine the relative com- plexity of the individual's information environment. Second, the diagram posits that the complexity of the indi- vidual's information environment determines: (l) the dominant processing style the individual is likely to de- velop, and (2) the relative complexity of the individual's cognitive structure. Third, the final key relationship expressed by the diagram is that the individual's dominant processing style determines the relative complexity of the individual's cognitive structure. Hence, the diagram details how there are two factors which determine the rela- tive complexity of the individual's cognitive structure-- the r envir bivar hm: thi Prc 56 the relative complexity of the individual's information environment, and the individual's dominant processing style. Given the relationships just described, the following bivariate hypotheses are offered: HYPOTHESIS ONE: The more ratio-dominant the individual's process- ing style, the greater the relative complexity of the individual's cognitive structure. HYPOTHESIS TWO: The greater the relative complexity of the individ— ual's information environ- ment, the greater the relative complexity of the individual's cognitive structure. HYPOTHESIS THREE: The greater the relative complexity of the individ- ual's information environ- ment, the more likely the individual is to employ a ratio-dominant style. While there are other hypotheses that could be formulated, these three will serve as our focus for now. Should these relationships be supported, future work would attempt to duplicate these findings, and extend the model. Chapter Summary The purpose of this chapter was to present a theory of human information processing. The major contention of this theory is that information processing is a measurement procedure individuals engage in to reduce environmental un- certainty. Further, the theory argues that there are only 57 two fundamental measurement procedures that an individual can employ to reduce uncertainty--nominal procedures and ratio procedures. The theory discusses the discriminatory and information generation capabilities afforded by each procedure, and contends that the dominant use of one (as opposed to the other) would lead to distinct differences in the development of the individual's cognitive structure. More specifically, it is argued that a ratio—dominant indi- vidual will develop a relatively more complex cognitive structure than a nominal-dominant person. To provide adequate context for this theory, a liter- ature review provided a discussion of cognitive structure, and the response patterns individual's exhibit with respect to certain information processing tasks. Next, a theory of information processing was presented. This theory pro- vided a detailed discussion on the nature of information processing, and then presented a diagram which laid out the major relationships of the theory. Finally, three hypothe- ses are presented which formally express the relationships between the key theoretic constructs. 58 FOOTNOTES l"Cloze actions" is a term which refers to the manner in which individuals completed sentences which were in need of nouns, adjectives, verbs, etc. As used by Streufert and Streufert "cloze actions" refer to the individual's responses made on the basis of stereotype or limited information. That is, when faced with the high level of incongruity in the environment, the individual is likely to respond to initial or familiar cues. 2For a good review of the various instruments that have been developed to assess cognitive structure, see Streufert and Streufert (1978), and Scott, Osgood and Peterson (1979). 3At this point, the discussion on measurement begins to differ from previous conceptualizations in important ways. For a good review of the "traditional" perspective on measurement, see Torgerson (1958). 4Of course, the possibility exists that the novel stimuli will be judged to possess the same amount of the attri- bute as the standard. 5This is because the phrase "if x possesses u units of attribute a" is essentially equivalent to "if there are u units of difference between an object with none of this attribute and x." That is to say, the two types of pro- cedures are very similar when one considers that the direct magnitude scale is always anchored by a zero point. 6For a complete review of interactive complexity theory, see Schroder, Driver and Streufert (1967), Streufert and Streufert (1978) or Scott, Osgood and Peterson (1979). 7While it is clear that "significant other" interactions influence an individual's behavior, it is important to note that the "significant other" literature does not specifically discuss the contextual or structural com- ponents of communication interactions. Hence, this dis- cussion is simply trying to infer what it is about "significant other" interactions that make them so influential in determining an individual's attitudes and behaviors. CHAPTER II METHOD Overview. To adequately test the hypotheses present- ed in Chapter I, a method has to be employed which: (1) demonstrates a logical connection between the conceptual and Operational constructs, (2) overcomes previous inadequacies in measurement, and (3) allows for data transformation tech- niques to detect complex patterns. To accomplish this task requires the implementation of an information processing task which can be utilized to categorize individuals into groups of nominal-dominant and ratio-dominant information processors. Further, the design must include measures of cognitive struc- ture, such that the relationship between dominant processing style and the relative complexity of the individual's cogni- tive structure can be empirically assessed. Finally, the design should incorporate multiple measures of the key con- structs to increase reliability. The purpose of this chapter is to provide a descrip- tion and rationale for the full set of procedures that will be employed. First a set of primary measures that opera- tionalized all of the key theoretic constructs will be dis- cussed. Second, a set of secondary measures for all the 59 60 key dependent variables (constructs) will be presented. Third, a preliminary instrument will be presented, and the pre-test of this instrument will be discussed. Fourth, the final instrument which will be utilized to formally test the hypotheses will be presented. Fifth, the data collection procedures will be discussed. Finally, the data handling procedures will be described. Operationalization Primary measures. The individual's dominant process- ing style can be determined by the manner in which the indi- vidual responds to certain subject-centered scales. More specifically, the individual's dominant processing style will be reflected in the distribution of responses the indi- vidual generates when completing ratio-level paired-compari- son scales. This type of instrument will be utilized because: (a) the completion of a questionnaire item is, in itself, an information processing task, (b) paired-comparison scales are free from the problems of "edge effects" and truncation of response, i.e., these scales have an increased ability (over less precise scales) to allow for differences in response patterns to become evident, and (c) the data provided by these scales is appropriate for metric multidimensional scaling analysis (MMDS), which is the technique which will be utilized to obtain the indices of cognitive structure (for a discussion on MMDS, see Woelfel and Fink, 1980). As was discussed earlier, the dominant use of nominal procedures shou with the proc enti inte star bee: abo* ind pr0' ind dif ext 61 should be reflected by the predominant use of extreme scores, with most scores being either zero or integer multiples of the standard. Alternatively, the dominant use of ratio procedures should be reflected by the use of scores along the entire continuum, with a large percentage of scores pep being integer multiples or integer reciprocal proportions of the standard. Once the data from the paired-comparison items have been analyzed for the distributional patterns discussed above, the response set for both nominal and ratio-dominant individuals will be examined using MMDS, and the information provided by this analysis will be utilized to index the three indicators of cognitive structure discussed in Chapter I -- differentiation, discrimination and integration. First, the extent to which an individual uses different (independent) attributes to define a set of stimuli (differentiation) can be assessed from the number of dimensions which result from the orthogonal decomposition of a transformed distance matrix (i.e., a scalar-products matrix). The greater the number of independent dimensions needed to account for a "significant portion" of the variance in the space, the greater the differ- entiation. Because MMDS procedures recover all n-l dimensions from a n x n symmetric matrix, a scree test will need to be applied to obtained differences in differentiation.1 Since we have no theoretic criteria at this point on which to base our scree test, we shall examine the number of dimensions Whit pro sel the at CC Ca in. CE 62 which account for 70, 80 and 90% of the distance variance in real Space, and inspect the discriminatory capabilities provided by each. From this analysis we should be able to select a level to use for future data analysis. Second, the individual's capability to make distinctions among stimuli (discrimination) can be assessed by examining the trace of the eigenvalue matrix. As the individual makes finer and finer discriminations, distance variance increases and the magnitude of the trace (a measure of variability) increases.2 Finally, the individual's capability to estab- lish relationships among the stimuli of his/her experience (integration) can be assessed by examining the warp factor: (sum of the positive eigenroots ) sum of all (positive and negative) eigenroots ' The warp factor is a summary measure which reflects the relative number and size of triangular inequalities gener- ated by an individual during the evaluation of the paired comparison items. It essentially indicates the individual's capability to maintain a "Euclidean-consistent" relationship among triads of concepts in the set of paired comparisons. A triangular inequality results when the Euclidean rule for planar triangles is violated (i.e., when the individual re- ports distances between a triad of concepts which does not allow the construction of a triangle). This measure can be utilized as an index of the degree to which an individual can establish "consistent" relationships among concepts. 63 However, it is important to remember that the term "con- sistent" simply means, in this case, "Euclidean consistent." The warp factor is a summary measure which expresses the degree to which triangular inequalities are present in the paired comparison data. As the number and size of the tri- angular inequalities increases, the warp factor increases. The smaller the warp factor, and the fewer the number of triangular inequalities, the greater the integration. To assess the relative complexity of the individual's information environment, a series of items will be presented which ask the respondent to: (1) detail his/her communica- tion activity, (2) report his/her rates of attentiveness to mass media sources, and (3) assess the relative sociological and psychological "richness" of his/her present and past experiences. In addition, several measures will be develop- ed to examine how informed the individual is with respect to certain topics. Secondary measures. Some secondary measures of cog— nitive structure will be employed to provide some indication of the validity and reliability of the MMDS measures, insofar as these secondary measures in themselves adequately assess what they purport to assess. These secondary measures will be derived from a modified version of Scott's "Listing and Comparing" instrument (see Scott, Osgood and Peterson, 1979, pp. 86-87, and pp. 104-107 for a complete discussion of this instrument). Briefly, this instrument requires the 64 respondent to examine a list of concepts, and generate a series of sets which group concepts together on the basis of some characteristic. From these data, measures can be derived for each of the indicators of cognitive structure. First, integration may be assessed by examining the number of sets the individual generates, for it reflects the in- dividual's capability to establish relationships among a set of stimuli. Second, discrimination may be assessed by examining the variance in the elements of the grouping patterns generated by the individual. Finally, two measures of differentiation, D and H, can be derived from the "List- ing and Comparing" technique. First the equation: k D 2 where: D = the measure of differentiation k = the number of sets (or grouping patterns) reflects the number of decisions that were made to classify the elements into sets and hence provide an indicator of the number of independent attributes the individual utilizes to define these stimuli. Similarly, the equation: 1 logzn HZnilogzni the degree of differentiation the number of stimuli the number of elements in each grouping pattern where: 5:332 21‘. which is based on the index of dispersion derived from in- formation theory (see Attenave, 1959) provides an additional indicator of the number of independent attributes that were USE COIT In in an SE be ti ti St E) 65 used to define the set of stimuli. As Scott et a1. (1979) comment,this measure: . . . represents the number of dimensions implicit in the respondent's grouping system. . . . Each different combination of group membership [i.e., grouping pattern] (including no group memberships) is considered a distinct way of combining the dichotomous attributes. . . . The more different combinations appear, the greater is the independence of the several attributes and the higher the resulting index of dimensionality. (p. 105) In other words, as the number of distinct grouping patterns increases, the greater the number of independent attributes, and the higher the resulting measure of differentiation.3 Background items. In addition to the primary and secondary measures already discussed, a series of items will be presented to obtain background information on each par- ticipant. The focus of these items will be on the examina- tion of demographics, lifestyle patterns, and other social- structural factors. This information will be gathered to examine the hypothesized relationships between social structural factors, processing style and cognitive structure outlined in Chapter I. The Pre-Test Objectives. The purpose of the pre-test was to: (1) identify the domain the paired-comparison items should be drawn from, (2) generate the actual list of concepts that will be utilized in the final questionnaire,4(3) initially th Da ad CO th Th ma at GE SC 51 66 evaluate the primary and secondary measures, and (4) examine the overall questionnaire as an information processing task. Data analysis of the pre-test instrument would allow for adjustments and refinements to be made before major data collection session was initiated. For pre-test purposes, the instrument which appears in Appendix A was developed. This instrument first asked the participant to identify two major news events which had taken place in the past year, about which most individuals had a firm opinion. Then, for each news event listed, the participant was asked to free- associate and list all major words and phrases that were in some way related to (or described) the news event, and to summarize his/her own opinion of the news event in a para- graph or two. In the second section of the pre-test instru- ment, the participant was presented with a list of 21 emotions, and asked to engage in a version of Scott's "Listing and Comparing" task. Third, participants were presented with 36 paired-comparison items which required them to evalute the discrepancy between 9 concepts associ- ated with the National Aeronautic and Space Administration space shuttle program. Finally, several items were present- ed which asked the participant to provide background infor- mation with respect to socioeconomic status, media patterns, and self-perceptions. In addition, participants were pro- vided with the opportunity to evaluate the relative clarity of the pre-test instrument. '0 "r1 67 Procedure. On June 3-4, 1980, 29 participants were administered the pre-test instrument.5 All of the partici- pants were students enrolled in an introductory communica- tion course at Michigan State University. The participants were administered the pre-test instrument after they had completed their final examination for the course. For their participation, students were informed that they would receive credit toward their final grades; however, no penal- ty was suffered by any student who did not wish to partici- pate. While no verbal instructions were provided, the author was present to answer any questions that arose. Participants required between 18anx130 minutes to complete the pre-test instrument. Since the students were in the midst of exami- nation week at the University, many were anxious to leave without any debriefing. Consequently, the author only pro- vided an informal discussion about the objectives of the instrument to approximately four or five students. Pre-test results. Table 1 shows the participants' responses to the question which asked them to specify the two most important news events of the past year. As can be observed from Table l, 28 of 29 respondents listed the taking of hostages in or around the United States Embassy in Tehran, Iran as one of the two most important news events.6 Of the 28 participants who listed the taking of the hostages, 24 of them listed it first. Further examination of Table 1 shows that no other news event received more than 9 responses. 68 Table 1 Pre-Test Data Participants' Listing of the Two Most Important News Events in the Past Year (N=29) Event Frequency of Response Taking of Hostages in Tehran, Iran 28 Presidential Primary Elections 9 Olympic Boycott 4 Economic Recession in the United States 4 Three Mile Island Nuclear Accident 3 Chrysler Corporation Bankruptcy 2 Russian Invasion of Afghanistan 2 Mt. St. Helens Volcano 2 Oil Crisis 1 The Failed Rescue Attempt of the Hostages in Tehran, Iran 1 The Jonestown Massacre 1 The Death of John Wayne 1 Consequently, only the participants' responses to the Iranian issue were selected for further examination. The 29 participants were asked to free-associate and list as many words or phrases that were in some way related to the news event, and then, in a paragraph or two, to state their own Opinion regarding the news event. A content analysis of these data (conducted by the author) revealed that 459 terms were generated by the participants with respect to the taking 69 of the hostages in Iran (see Appendix B). These concepts were then sorted into categories on the basis of their simi- larity in meaning. Table 2 lists the 15 concept categories that were created from the total set of concepts. A cate- gory was created if at least 5 concepts were grouped together. Table 3 reports the descriptive statistics to all (primary and secondary) indicators of cognitive structure. More specifically, Table 3 reports the mean, standard devia- tion, minimum score, maximum score and range for each indica- tor of cognitive structure examined in the pre-test. Table 4 reports the inter-correlations among all these indicators of cognitive structure. An analysis of these data should pro- vide indication of how well these scales reflect variation in the theoretic constructs, and how strong the association is between the primary and secondary indicators of each con- struct, both "within measure" and "across measures" (i.e., how strong the association is among indicators of the same construct, and among indicators of different constructs). An inspection of the standard deviations in Table 3 indicates that NDN, the number of dimensions needed to ac— count for 90% of the variance of the space, has the highest variance of any of the three MMDS differentiation measures. We interpret this finding to suggest that this measure pro- vides the most discriminatory capabilities, and hence should be used as the criterion level in the scree test which will 70 Table 2 Pre-Test Data Concept-Categories Extracted from the 459 Concepts Generated by the 28 Participants who Listed the Taking of the American Hostages in Tehran, Iran as an Important News Event Concept-Category Iran United States Hostages Khomeini Carter Shah Militants Weak Hostile Media Coverage Diplomacy Military Intervention Religion Irrational Economic Sanctions be used to determine differentiation. As can be observed in Table 4, 25 correlations (out of a total of 91) were significant at the .05 level. Fur- ther, with one exception, all of the correlations which were significant were "within measure." That is to say, ten of the fifteen correlations among the indicators derived from the Scott measure were significant, while fourteen of the 28 correlations among the indicators derived from the MMDS measures were significant; however, only one of the 48 m. 0 p 05:. 71 G 38:. 8s mam» e. -m‘. .a A «if x +m> ><+~> m +6” a,“ H mm A < moon.» o5 o» muoou:o3o o>..3..nmom o5 mo 03:“ u E “SHE: “Spool/53o o5 mo 8:93 on» u gm. RENAME, o5 mo wom M HON £580: on popoo: gag mo Hob-5: o:u n 292 ooqoflug on» «0 wow M Hem 85000: B cocoo: 9:03:26 mo Hon—B: o5 u mdz 85.39 on.» mo mow M MOM £580: 8 pocoo: msoemcwfid no .895: opp u 82 mgflm: on» 3 pououoao msowugo mo msuouumm @5905 m0 .895: on» u 9.2 8534.5 swoon—om @5980 u mwm acmmofifimtmmfl u m sznx mumfic gm u 9 Buzz mm.m 84... 8.? 2.: mm; 8.? m: mica @5888: 8:0. :2. 3.: 8.8 8.0 3.: 8:; 883835 wufimwmuz 84888 88888 8.8 8.8m 34.222 8.838 mucus 8888.883 mes 8.m 8.2. 84 8.8 35 8.~ znz on... 8.m 86 84 8.8 8.0 3.... 82 sot 8.~ 8.m 84 8.8 8.0 8.: 82 8:835:88: 83:8 8.8 8.: 8.... 8.8 8:. 88 .82 88...:ch 888: 2.0 ode 88 3.8 84 mm.~ :8 83285.88: gumboom SA 86 mm; 8.8 88 :..~ m 58: 8.8488 88388 8.8 8.2m 8.883 8:882 o 8338288: @988 omqom xoz 5.: .>.O on x Hons ousmooz 8.30m Ammnzc ousuuauw ofiufifiou mo mnouoofifiH .3: How omfid 0:: Axmzv ouoom 5582 £57: ouoom p555: .A.>.UV :ofloflufla mo ”Eofioflmooogddv couflofiwoo gm .ob Coos Ema ummyHIOHm m manna. '72 m 44 d N H 3 2:3. oomom> 24+; 24+? 24+? 24+42 24 n ma ooonu on» 3 38H:omflo ozuflmom o5 mo 03m: u $23 2.3.4qu “Bogfigo o5 mo 8:93 onu u mug 8:355 on» no woo M new £508: 3 wovoo: “Ho-mg mo Hons: o5 u 752 oogflnmb o5 mo woo M How “Enooom 8 coooo: 903:5 mo Honk—4: on.» n mnz 853.445 o5 mo won M new uSSooo 3 uoooo: 9395 m0 .435: man u 82 mayo-.3 o5 E coon-880 @0395 m0 gouumm 0590.5 mo Month: on» u g 8:385 :uouuom @5953 u Hum w:mmo.nfi:w2IH:.-:~mo.n u m sznx muoné xm n o 6.823 84 8f .8.- 8. 86 .4m. 8. 8f 8. 8- mm «43m 8488on 84 2. :EF :8.- .28.- 44. 8. 8. m4. :5? 848535 84 N4.- 8. N4.- N4. mof m4. 8. BE 832454.484: $282222 84 :8. :8. .5.- 8K 8. mm.- 292 on 84 :8. 8. Sf 84. m4.- 82 so: 84 N4. :8.- mm. 8.- 82 204838.884: 494.48 8.4 mm.- :mm. :8. 82 coflmummufi 888: 84 «3. 4m.- :8 coflmcdfiuoma £qu 84 :3. 2 58: oo . 4 a :oflofléom M40 83qu mm .223 82% 28 82 82 62 mom 2 o 483 $68: 8.38 Ammnzv 9438.446 934.4500 mo muouoofivfi on» 0:05 Eofloaouuoo «boo womb-ohm e Enos 73 correlations between the Scott measures and the MMDS meas- ures were significant. This outcome may be interpreted as the result of the attempt to inter-correlate indicators from two different domains (emotions and space technology). This implies that cognitive complexity appears to be domain specific rather than a general trait across all domains of experience; one's ability to deal with, understand, and/or differentiate among ideas related to space technology is not necessarily related to one's ability to deal with, understand, and/or differentiate among ideas related to human emotions. Consequently, attempts to assess the rela- tive complexity of an individual's cognitive structure using more than one technique (measure) should strive to focus each technique around the same domain of experience (unless the investigator has prior evidence that the com- plexity measures of two or more domains should be correlated). In Chapter I, it was posited that ratio-dominant and nominal dominant individuals could be distinguished by the distribution of their responses to the paired—comparison items. More specifically, it was expected that nominal dominant individuals could exhibit a bi-polar distribution of scores, with most scores being either zero or integer multiples of the standard, while ratio-dominant individuals would use more scores across the entire continuum. To dis- tinguish nominal dominant individuals from ratio dominant individuals using these criteria, five indicators were 74 extracted from the MMDS portion of the pre—test instrument and utilized to create a summary measure of processing style. The five indicators are listed below: (1) (2) (3) (4) (5) The total number of scores the indi- vidual used (NOS). The acronym NOS refers to "number of scores." This indicator reflects the extent to which the respondent used a large variety of scores as opposed to a few scores. The number of scores equal to zero and/or a multiple of 50 (POP). The acronym FOF refers to "frequency of fifty." This indicator reflects the individual's tendency to use integer multiples of the standard, half of the standard and/or zero (i.e., a somewhat limited set of scores). The number of scores greater than or equal to 250 (NEXMAX). The acronym NEXMAX refers to the "number of extreme scores greater than or equal to a selected maximum score of 250." This score of 250 was employed as the cri- terion for determining an extreme score because it was the median of the set of the largest scores provided by each individual. The number of scores less than or equal to 10 (NEXMIN). The acronym NEXMIN refers to the "number of extreme scores less than or equal to a selected minimum score of 10." This score of 10 was em- ployed as the criterion for determining an extreme score because it was the median of the set of the smallest scores provided by each individual. The coefficient of variation for the en- tire set of scores that were used by each individual (C.V.). The acronym "CV" refers to the "coefficient of variation." This coefficient reflects the range of scores that were employed by the respondent rela- tive to the mean. It is derived by divid- ing the standard deviation by the mean. 75 Table 5 reports the mean and standard deviation for each of these indicators of information processing style, while Table 6 reports their variances, covariances and cor- relations. It can be observed from Table 6 that the corre- lations between NOS, FOF and NEXMIN are moderate and negative (-.27 and -.29, respectively), while the correla- tion between FOF and NEXMAX is near zero (.09). These findings indicate that an individual who utilizes a large number of categories tends not to use extreme scores, zero, or integer multiples of the standard. Further, in Table 6 Table 5 Pre-Test Data Means (Y) and Standard Deviations (s.d.) for the Indicators of Information Processing Style Derived From the MMDS Instrument (N=29) Variable i s.d. NOS 11.03 4.14 FOF 21.59 12.57 NBXMAX 6.62 8.72 NEXMIN 6.86 10.73 C.V. 128.97 128.53 where: NOS = number of scores used FOP = number of scores = 0 or an integer multiple of 50 NEXMAX = number of scores 1 250 NBXMIN = number of scores 5 10 C.V. = coefficient of variation 76 Table 6 Pre-Test Data Variances, Covariances and Correlations for the Five Indicators Used to Determine Nominal-Dominant and Ratio-Dominant Information Processors* (N=29) NOS FOF NEXMAX NEXMIN C.V. NOS 17.11 -.27 .09 -.29 -.10 POP -14.20 157.97 .72 -.56 .39 NEXMAX 3.37 78.66 76.10 -.46 .33 NEXMIN -12.89 -75.56 -43.05 115.20 -.11 C.V. -51.63 632.52 374.42 -153.67 16520.24 where: NOS = number of scores used FOF = number of scores = 0 or an integer multiple of 50 NEXMAX = number of scores 1 250 NEXMIN number of scores 1 10 C.V. = coefficient of variation * Covariances are in the lower-left of the matrix, variances are in the diagonal of the matrix, and correlations are in the upper-right of the matrix. 77 it can be observed that the correlation between NEXMAX and FOP is high and positive (.72) while the correlation between NEXMAX and NEXMIN is moderate and negative (-.46). These findings seem to indicate that: (a) there is a tendency for individuals who use extreme scores to use integer multiples of the standard, and (b) that individuals who tend to report extremely large scores have a slight tendency to also report extremely small scores. Similarly, the correla- tion between NEXMIN and FOP is moderate and negative (-.56), indicating that individuals who report small scores are not likely to use integer multiples of 50. Finally, the corre- lations between C.V.,NOS and NEXMIN are small and negative (-.10 and -.11, respectively) while the correlations between C.V., FOFand NEXMAX are moderate and positive (.39 and .33, respectively). We interpret these correlations to indicate a general trend that the coefficient of variation increases as the number of extreme scores increase, and decrease as the number of scores near zero increases. The only para- doxical correlation within this last set of correlations in Table 6 is the small negative correlation between C-V- and NOS. This seems to indicate that as the number of cate- gories employed by the individual increases, the coefficient of variation decreases. This finding is most likely attrib— utable to the sensitivity of the coefficient of variation to extreme scores (see footnote 7). However, it is impor- tant to remember that many of these correlations are 78 essentially zero. Consequently, further data should be examined before any substantive conclusions about this, or any of the other correlations, are drawn. A summary measure of information processing style (PS) was created by first, factor analyzing the set of five indicators discussed above, and then using the unstandard- ized factor score coefficients as the weights in the regression equation: PS = Aylyli+xy2y2i+Ay3y3i+Ay4y4i+xysy5i where: PS = processing style score Ay = the factor score coefficient for NOS Ayl = the factor score coefficient for FOF Ayz = the factor score coefficient for NEXMAX Ay3 = the factor score coefficient for NEXMIN Ay4 = the factor score coeffifiient for CV yli = NOS score for the i:h individual y2i = FOF score for the i individual YBi = NEXMAX score for the i:: individual y4i = NEXMIN score for the i individual Y5: = C.V. score for the ith individual This structural equation describes a "factor model" of one theoretic construct with five indicators. The equation was solved using the LISREL IV (Joreskog and Sérbom, 1978) soft- ware package. This computer program provides unstandardized maximum likelihood estimates for all coefficients in the 8 In addition, the SPSS (Nie, et a1. 1970) pro- equation. cedure FACTOR was employed to obtain standardized coeffi- cients for the parameters of the equation described above. 79 This program uses a principal components technique for obtaining the parameters of the model. Table 7 reports the standardized and unstandardized factor score coefficients for each of the parameters which were derived from each estimation procedure. Table 7 Pre-Test Data Unstandardized and Standardized Factor Score Coefficients for the Five Indicators Used to Determine Nominal-Dominant and Ratio-Dominant Information Processors (N=29) Variable Unstd. Std. NOS -.320 -.004 FOP -.124 .386 NEXMAX .013 .364 NEXMIN -.O43 -.302 C.V. .005 .228 X2 = 12.209 df = 5 probability level = .032 PS = 1 +1 +1 +1 +1' ylyl yzyz y3y3 y4y4 y5y5 where: PS processing style score '< H II II II NOS = number of scores used y2 FOF = number of scores = 0 or an integer multiple of 50 y3 = NEXMAX = number of scores 3 250 y4 = NEXMIN = number of scores 3'10 y5 = C.V. = coefficient of variation Ayi = the unstandardized factor score coefficient for each yi 80 Table 8 Pre-Test Data Mean Scores for the Indicators of Cognitive Structure: Upper vs. Lower Half Means and Upper vs. Lower Quartile Means (N=29) Upper Half Lower Half Upper Quartile lower Quartile Variable N=(14) (N=15) (N=7) (N=7) D 224123.73 39378.28 171053.71 78061.71 H 2.82 2.59 2.89 2.86 SGP 2.94 2.72 3.70 2.58 NGP 10.53 9.07 11.29 10.86 NDS 1.60 1.92 1.42 2.00 NDE 2.07 2.14 2.00 2.14 NDN 2.60 2.79 2.43 2.86 TRACE 1067950.73 201622.86 1878344.00 38752.43 WARP 1.90 1.83 2.14 1.91 where: D = 2k (where k=NGP) H = logzn-ifni—logzni SGP = grouping pattern variance NGP = the number of grouping patterns of emotions generated by the listings NDS = the number of dimensions needed to account for Z 70% of the variance NDE = the number of dimensions needed to account for 1 80% of the variance NDN = the number of dimensions needed to account for 1 90% of the variance TRACE = the trace of the eigenvector matrix WARP = ratio of the positive eigenroots to the trace 81 Following Woelfel et a1. (1977), the unstandardized coefficients were selected for use in the development of the processing style (PS) score. Once a PS score was generated for each individual, the total set of scores was subjected to a median split and a upper vs. lower quartile split. Then, using these groups, a one-way analysis of variance was conducted using every indicator of cognitive structure as a dependent measure.9 Table 8 reports the mean score for each indicator of cognitive structure yield- ed by the median and quartile splits of processing style score. Table 9 reports the analysis of variance.10 An inspection of Table 9 reveals that none of the F statistics was significant at the .05 level. Further, an examination of Table 9 indicates that in many instances the direction of the mean scores was Opposite to that which was expected. What this seems to indicate is that the sample of respond- ents may have been too homogeneous with respect to process- ing style for significant differences in cognitive structure to be uncovered. More will be said about this, as well as other problems, in the next section. Discussion of the pre-test results. The results of this pre-test suggest that several changes should be made in the final instrument to test the relationships proposed in Chapter I. First, the Scott measure and the MMDS measures of cognitive structure should be drawn from the same domain so as to avoid problems with individuals whose complexity varies significantly across domains of experience. 82 Table 9 Pre-Test Data F-Ratios for the Indicators of Cognitive Structure by Information Processing Style: Comparison of the Upper vs. Lower Half and Upper vs. Lower Quartiles (N=29) Upper vs. Lower Upper vs. Lower Dependent Variables Half F Quartile F D 1.330 .317 H .428 .004 SGP .139 1.364 NGP .636 .023 NDS 2.475 3.692 NDE .061 .097 NDN .260 .614 TRACE 2.478 3.262 WARP .072 .246 2k (where k=NGP) .. 1 H - logzn nZnilogzni where: D NGP = the number of grouping patterns of emotions generated by the listings SGP = grouping pattern variance NDS = the number of dimensions needed to account for Z 70% of the variance NDE = the number of dimensions needed to account for Z 80% of the variance the number of dimensions needed to account for Z 90% of the variance NDN TRACE = the trace of the eigenvector matrix WARP = ratio of the positive eigenroots to the trace 83 Second, the MMDS portion of the instrument should: (a) provide an example to demonstrate how the task should be approached, (b) employ more items so that distributional differences can be more readily detected, (c) remind indi- viduals of the standard more frequently so as to increase the likelihood that judgments are made on the basis of the standard, and (d) better inform individuals about the meaning of the distance estimates so that judgments can be made as reliably as possible (see footnote 9). Third, with respect to the Scott measure: (a) an example should be pro- vided to demonstrate how the task should be approached, (b) the items that are being evaluated should appear on every page with boxes to aid in the task at hand, (c) the items should appear at the end of the questionnaire, and (d) individuals should be provided with clear instructions that they should exhaust all possible combinations that come to mind. These recommendations are made because the Scott measure is the most open-ended of all the items, and hence, likely to require the most time to complete. Consequently, if it is placed at the end of the instrument it is less likely to disrupt the individual's processing of the others items and/or cause frustration. Fourth, a greater hetero- geneity in the sample should be sought so that differences in processing style and cognitive structure may be more readily detected. Fifth, a set of oral instructions should accompany at the administration of the instrument. These 84 oral instructions should outline each section of the instru- ment, and detail helpful examples. Finally, the instrument should be timed so that it is known how long it takes each individual to complete the questionnaire. This recommenda- tion is made simply to check an assumption held by the author that a relatively complex person will take more time to complete the instrument than a relatively simple person. The rationale for this assumption is that a relatively com- plex person is likely to require more time to make fine dis- criminations among the set of stimuli being evaluated. The Final Instrument The instrument which was designed for the final data collection appears in Appendix C. There are three major types of items contained in the instrument. First, a set of items appear which measure social-structural factors, i.e., demographic information, communication activity, and life- style patterns. Second, a set of paired comparison items appear which require respondents to make judgments among a set of 15 concepts related to the United States - Iranian Hostage Situation. Third, respondents are asked to engage in the Scott "listing and comparing" task using the same concepts that appeared in the paired comparison portion of the instrument. The next few paragraphs will provide further detail of each of these item-types contained in the final instrument. 85 The social-structural items. The social-structural items are designed to measure various aspects of the indi- vidual's sociological and psychological background. In particular, these items focus on: demographics, family structure, communication activity, and general lifestyle patterns. The demographic and family structure items were designed to gather information about the individual's religion, family composition, and family socioeconomic status. The items dealing with communication activity were designed to provide summary measures of the individual's interpersonal (face-to-face) interactions and exposure patterns to a variety of mass media. The items concerned with lifestyle patterns were designed to gather information related to the kind of social—psychological environment the individual has experienced. More specifically, these items measure the degree to which the individual's experiences were, e.g., nuturing, stimulating, organized, and socially responded to. The total set of social-structural items were presented to obtain a multiple indicator perspective of the individual's background and information processing experi- ences. The specific items which appear were developed by the author after a comprehensive review of literatures con- cerned with the specification of sociological and psycho- logical variables related to cognitive development (e.g., see Bayley, 1940; Werner, 1957; Warner, 1960; Wolf, 1964; Elder, 1968; Levin and Fleischman, 1968; Wachs, Uzgiris and 86 Hunt, 1971; Elrado, Bradley and Caldwell, 1975; Walberg and Majoribanks, 1976) and academic/occupational achievement (e.g., see Duncan, Featherman and Duncan, 1972; Nuttal et al., 1976; Essen, Foggelmark and Head, 1978). The paired comparison items. The second main section of the final instrument consists of a series of 105 paired comparison items, which require the respondents to evaluate the semantic discrepancy among 15 concepts concerned with the United States - Iranian Hostage Situation. These con- cepts were obtained from the pre-test data which were re- ported earlier. Further, since the pre-test was conducted in June, 1980 and the final instrument was not to be admin- istered until October, 1980, the author collected new data to see if any concepts should be deleted or added to the list of concepts obtained from the pre-test. New data were collected from 38 undergraduates en- rolled in an advertising course at Michigan State University during the second week of October, 1980. These 38 students were asked to free associate and list all major words and phrases that were related to the United States - Iranian Hostage Situation. Appendix D shows the list of 555 con- cepts that were generated by the respondents, content analyzed by the author, and sorted into categories. A con- cept category was created if 15 or more terms were grouped together. On the basis of these data, it was decided not to add or delete any concepts from the original list 87 synthesized from the pre-test data. Consequently, the following fifteen concepts appear in the list of paired comparison items: The United States, Iran, Khomeini, Carter, Diplomacy, Military Intervention, Economic Sanction, The Shah, The Hostages, The Militants, Irrationality, Weakness, Hostility, Freedom and Me. The Scott "Listing and Comparing" task. The final section of the instrument consists of the Scott "Listing and Comparing" task. This task requires the respondents to examine a list of concepts and group those concepts which are alike in an important way. Each respondent is to create as many groups as he/she feels are necessary to ex- haust all important groupings. Since the paired comparison data and the data from the "Listing and Comparing" task are to be used to develop the primary and secondary measures of cognitive structure, the same 15 concepts which appeared in the paired comparison portion of the final instrument were used in the "Listing and Comparing" Task. This procedure was employed to make the two measurement techniques as com- parable as possible. The Administration of the Final Instrument Selection of subjects.11 During the analysis of the pre-test data, one conclusion that was reached was that the sample may have been too homogeneous for significant indi- vidual differences to emerge. Consequently, an attempt was made to get a more heterogeneous set of respondents. To 88 accomplish this, the author targeted every junior and senior in the Department of Communication to participate in the study. This particular selection procedure was adopted for two reasons: (1) it was thought that grade point average, since it is a summary measure of the individual's ability to complete information processing tasks (i.e., exams and term papers), would be a good indicator to use to differentiate among the two informa- tion processing types, and (2) it was thought that 80 credits (junior class standing at Michigan State University) would be a good criterion for deciding when an indi- vidual's grade point average had been reliably established. The auther assessed the student records of every student in the Department of Communication, and selected those students who had: (1) completed at least 80 credits at the University, (2) taken at least some of the Michigan State University Entrance Exams, and (3) taken either the American College Test (ACT) or Scholastic Aptitude Test (SAT). Since student records were being assessed, it was thought that the entrance exam information and ACT - SAT informa- tion would be useful indicators for discriminating among individuals' information processing ability. Appendix E contains a brief description of the University Entrance Exams, the Scholastic Aptitude Test, and American College 89 Test, prepared by the University College at Michigan State University. This subject selection process yielded 236 potential participants. Attempts were made to reach all individuals. Of these 236 peOple, 99 either had graduated during the summer or could not be utilized.12 The 137 people who were contacted were informed that a study was currently being conducted in the Department of Communication for the Director of Undergraduate Studies. Each person was also informed that they had been selected to participate in the study, and asked if they could volunteer 1/2 hour of their time to help out in the data collection process. Each student was told that the study was concerned with informa- tion processing, and that some of the results might be utilized by the Director of Undergraduate Studies to aid in curriculuum development. After each student was provided with this brief description of the study, he/she was asked to indicate a time that would be convenient for him/her to come in and participate. Appendix F contains the instuction sheet that was used to guide the request for participation. Of the 137 people who were contacted, 16 refused to partici- pate, and 22 said they would participate but never showed up at the data collection site. Consequently, 99 students in all participated in the study. Administration of the questionnaire. Data collection procedures commenced on Wednesday, October 22, 1980 and were 90 completed on Tuesday, November 4, 1980. Each participant was greeted at the data collection site by the author or an undergraduate research assistant.13 Each participant was briefly informed about the study and asked if he/she were willing to participate. Appendix G provides a brief summary of the procedures that were used to greet participants and introduce them to the study. After each participant orally i1 agreed to participate in the study, he/she was asked to [ sign a consent form (Consent Form A, see Appendix H) which informed him/her of his/her "rights" as a participant, and which outlined the relative costs and benefits of his/her participation. Once the participant had signed the consent form, he/she was placed in an ordinary meeting room which contained six chairs situated around a large table. Each participant was provided with an oral set of instructions about how to complete the questionnaire, and told to: (a) read the printed instructions carefully, (b) consider each item carefully, and (c) ask the author (or research assist- ant) to clarify or interpret any item which was not clear. The author (or research assistant) then noted the time of day, wrote it on the questionnaire, instructed the partici- pant to begin filling out the instrument, and leave when finished. The author (or research assistant) then waited outside the meeting room until the participant left. After noting the time of day that the participant wrote on the questionnaire, the participant was given a comprehensive 91 debriefing on the nature of the study which included: (1) a brief description of the theory and hypotheses generating the inquiry, (2) a discussion on the relationship of the data collection procedures to the theory and hypotheses, (3) a discussion of the null and alternate hypotheses, and (4) a description of how the data were to be processed, and where written reports of this research could be obtained. After all questions from the participants had been answered, the participant was asked for permission to utilize the data contained in his/her student record in the analyses to be conducted. A consent form (Consent Form B, see Appendix H) was provided to those participants who orally agreed to allow the data to be used as part of the investigation. After the response to Consent Form B was obtained, the participant was excused. It should be noted that data were collected between October 22 through November 4, exclusive of and November 1-2, because these were weekends, and November 3, due to scheduling difficulties. Finally, data collection ceased on November 4, due to the election of Ronald Reagan as president of the United States. It was thought that the change of leadership in the United States might serve as an effect that would greatly influence a participant's re- sponse to certain items in the final instrument. One 92 additional note is that not all participants filled out the questionnaire in isolation. While some participants filled out the questionnaire while being seated in the meeting room alone, others completed the questionnaire while seated in the room with other participants who were also filling out the questionnaire. When this latter situation arose, participants were informed to work independently, and to direct all inquiries to the author (or research assistant) outside. Data Handling Procedures The questionnaires were first coupled with a sheet which contained the academic data from the participant's student record, and assigned an arbitrary "participant number." These data were then translated to computer cod- ing sheets. These coding sheets were then verified (and corrected) twice against the questionnaires. After the second verification and correction, the author randomly examined another 20% of the coding sheets and found no errors. The coding sheets were then professionally key- punched by Resource Control Incorporated, a Lansing, Michigan keypunch service. The punched cards were listed and verified (and corrected) against the questionnaires twice. After the second verification and correction, the author randomly examined another 20% of the questionnaires, checked them against the listing of the punched cards, and found no 93 errors. The data were then submitted to the analyses described in Chapter III. 94 FOOTNOTES 1For a good review on procedures for determining the dimensionality of psychological processes, see Barnett and Woelfel (1979) or Woods (1977). 2One problem with this measure is that a high discrimi- nation score may result from two different situations: (1) the individual's tendency to make fine discrimina- tions among a relatively homogeneous set of stimuli, or (2) the individual's tendency to dichotomize the differ- ences among stimuli and use a large number of extreme scores to represent his/her judgments. To overcome this problem one may: (a) transform the distance estimates using some monotonic function, (b) normalize the dis- tance estimates using some arbitrary criterion, or (3) separate the Type 1 individual (described above) from the Type 2 individual, and then examine the differentia- tion scores. If this problem is observed in the data, the latter two procedures will be employed in the data analysis. 3For a more complete discussion of these derived measures see Stoyanoff (1980), and Scott et a1. (1979). 4We would want to select a domain by controlling for past information while maximizing the variance in the concepts that are reported by the participants. That is, we would want to select a domain in which: (1) all participants had about the same prior information, and (2) exhibited a large variance in the number of concepts (across individu- als) that were used to define the domain. STwenty-eight of the participants were administered the pre-test instrument on the evening of June 3, 1980 at a final exam session. The 29th participant was administered the instrument at a "make-up" exam session on the morning of June 4, 1980. 6It should be noted that the one person who did not list this response in the space provided made a note in the margin of the questionnaire that the taking of the hos- tages had temporarily slipped his/her mind, and probably 95 should have been listed as one of the two most important news events. 7One problem with interpreting the coefficient of variation is that some respondents tend to use extreme scores quite liberally. Consequently, an extreme score of 10,000 or 20,000 tends to inflate the coefficient of variation, while a series of extreme scores : tends to deflate the coefficient of variation. 8For a concise discussion which compares maximum likelihood estimation with other estimation techniques, see Stoyanoff (1979), and Hanushek and Jackson (1977). 9It was suspected that many of the extreme scores encoun- tered in the pre-test were given rather hastily, i.e., without much consideration. Consequently, scores larger than 2500 in the MMDS data set were replaced by the extreme score 2500. This "trimming" reduced the influence of these scores to inflate the measures of TRACE and WARP, and deflate the measures of NDS, NDE and NDN. 10The assumption that the errors of prediction are normally distributed and homoskedastic were not checked. 11The procedures which were employed throughout the entire study were submitted to and approved by the Michigan State University Committee on Research Involving Human Subjects, and the University Committee on the Privacy and Release of Student Record Information. 12Three of the 99 were research assistants involved in the project. Consequently, they were not included in the sample. 13The research assistants were three people who were working on an independent study project concerned with learning the research process. Two were female and one was male. All assistants had taken at least one course in research methods prior to being selected to work on the present project. They were trained for a period of four weeks (approximately 16 hours) in the research procedures described here. CHAPTER III RESULTS AND DISCUSSION In Chapter I the following theoretic model was pre- sented to account for individual differences in cognitive structure: )2/ where: E = social structural factors n1 = the relative complexity of the individual's information environment n2 = the individual's information processing style n3 = the relative complexity of the individual's cognitive structure From this model three formal hypotheses were generated: Hypothesis One: the more ratio dominant the individual's processing style, the greater the relative complexity of the individual's cognitive structure. Hypothesis Two: the greater the relative complexity of the individual's information environment, the greater the relative com- plexity of the individual's cognitive structure. 96 97 Hypothesis Three: the greater the relative compleXity of the individual's information environment, the more likely the individual is to employ a ratio-dominant (processing) style. The fundamental purpose of the data collection procedures was to gather data which would allow us to test these hy- potheses and determine the extent to which the model represents the factors and process which determine indi- vidual differences in cognitive structure. The primary means by which this objective will be accomplished is through the utilization of linear estimation techniques. More specifically, the plausibility of the hypotheses and overall model will be examined through correlation analysis and other linear model analyses. This chapter describes how these analytic procedures were used to evaluate the model, and the results that were discovered. The chapter begins with the presentation of a struc- tural equation model and brief discussion of linear regres- sion analysis. This discussion focuses on the theoretic and statistical assumptions of full information maximum likelihood estimation, and examines the extent to which the model and sample data satisfy these assumptions. Second, the correlations among the indicators of each theoretic construct are examined. This analysis describes the magni- tude, direction and significance levels of these correla- tions. Third, the extent to which the indicators of each theoretic construct fit a single factor structure is 98 examined using confirmatory faccor analysis. Essentially, this procedure will test each component of the model inde- pendently of the other components, and provide information on how well the indicators reflect variation in each theoretic construct. Further the extent to which each predictor con- struct is correlated with the each predicted construct is ex- amined using canonical correlation analysis. The results of this analytic procedure indicate how many significant canon- ical correlations there are among the canonical variates of each theoretic construct. Fourth, all model parameters will be estimated using full information maxiumum likelihood, and the overall model will be tested to examine the extent to which the sample data fit the model. Finally, the chapter concludes with a discussion of the major findings. A Basic Discussion of Linear Regression Analysis Since the initial formulation of "path analysis" by the biologist Sewall Wright (1934), regression analysis has been one of the most widely used techniques for examin- ing the direct and indirect effects of variables specified as causes on variables specified as effects (Kerlinger and Pedhazur, 1973, p. 305). To employ regression analysis, a verbal theory must first be cast into a model representing the relationships between the variables expressed by the theory. Figure 2 is such a model. This analytic repre- sentation is essentially a simplified presentation of the theory in Chapter I. 99 C1 €____.nl >6) where: 5 = the theoretic construct "social structural factors" n1 = the theoretic construct "information environment" n = the theoretic construct "processing style" n3 = the theoretic construct "c0gnitive structure" ti = the error of prediction in each ni Figure 2. Theoretic Model 100 Following Kmenta (1971), whenever a structural equation model like the one in Figure 2 is generated, it is assumed that in the model for the pOpulation:l (a) (b) (C) (d) (e) (f) there are no specification errors, i.e., that the correct functional form is specified (with all equations being linear), that no variables pertinent to explaining variance in the endogenous variables are excluded, nor are any extraneous variables included; the model is identified;2 the errors of prediction (C-) are normally distributed about zero and homoskedastic, with the structure of the variance-co- variance matrix among the errors of pre- diction fully specified; the exogenous variables are nonstochastic, with values fixed in repeated samples such that, for any size sample n — 2 2 (xi X) i l n is a finite number greater than zero (Kmenta, 1971, p. 202); the exogenous variables are not characterized by perfect multicollinearity, and exhibit some independent variations; and additionally, the number of observations (i.e., measured variables) is greater than the number of parameters to be estimated. Hence, full specification of the model requires specification of: (l) the functional form of the model, (2) the probabil- ity distribution of the errors of prediction, and (3) how the exogenous variables are determined (Kmenta, 1971, pp. 347- 348). 101 Figure 3 illustrates the fully specified structural equation model which was developed to test the theory in Chapter I. The scale items which correspond to each indi— cator in the model, and the labels which will be used to refer to these indicators appear in Appendix I. An inspec- tion of Figure 3 indicates that the "counting rule" for identification has been met: the number of measured inputs is larger than the number of parameters to be estimated (df=457). Beyond this, however, not much can be said about the extent to which the assumptions of full information maximum likelihood estimation have been met without analyz- ing the sample data. Table 10 (columns 1 through 6) reports the summary statistics for each indicator in the model depicted in Figure 3. As can be seen from Table 10, many of the endo- genous indicators (yi's) have large positive skews. This finding indicates that these measures are not normally dis- tributed about their mean, which in turn implies that the additive errors of prediction associated with these depend- ent measures will not be normally distributed about zero. This means that a key statistical assumption required for linear regression is unlikely to have been met in the population. One procedure that can be employed to alleviate the problem of highly skewed variables calls for the transfor- mation of the raw data using a statistical criterion for: 102 .H xflucommm .000 muoumowncw 023 m0 hummmodm 0 H05 4009‘ :5 .m 030E aw: £000 5 aging m0 g0 03 u Mu Ssmmmfiugmmflnmouonuwgn .u c ”.838 623 88885 B» n :32» .3330 $333500: 30:30:00 3300003 93 u m: c 5 82.80 623 E8850 «5 u W32» gmabm gmmwooumg 30:30:00 0.38093 3 n m.» H c 6038 633 8883: 90 u as». a .3335 5395035.. 305030000 03300.3 93 u .n .xfimwfiugmémoggume w 30058 50.2: 303002 93 u .x =30”.me 3330930 33000.. 30930000 0300093 93 u w "833 103 .000: 00300100 03 000 .0000 003 3530000 + 300333 5 0038503053 03 ._o.~_ A 300.0 m .33» ganmwums 000 mum was on HA .H 5% 000 000300305 03 mo 300003 0 Hon.— .H g 000 “0393.85 0005950 5 000300305 .300 @5000: 03 050 003330 0000 no 3300.» 03 0.8 0030 2 a; mm oo.oooa oo.om m~.v oo.a~a 0H.mmH am flax mm oo.ooo~ oo.oq 04.0 mm.mma 0H.mmH ma aflx mm oo.oom oo.oa oo.~ am.Hm mo.mva mm ax mm oo.oooH oo.mn Nv.m aa.~ma oa.mma mm 0x mm oo.ooo~ oo.o mv.m mm.ma~ am.mma am ax mm oo.oom oo.om k~.~ mo.o~ om.ooH ma 0x . oo.m~ oo.m mm.o mm.m mH.mH ma mx - oo.p~ oo.k Hm.o- mm.m HH.~H am «x . ou.~a oo.mH 00.0 mm.- om.¢m «a mx - o~.~m om.o~ om.o m~.~H om.~m cm «x . oo.ooo.OOH oo.oom~ km.H am.qqmo~ Ho.HaHmm mm ax 0888 880 .5552 380 88688 08:80 88 z 08885 0.0000 55.38: mammnzv 3000: :00 05 5 000300305 :4 000 0030380 93300000 OH wanna. 104 m~.m 2.. 3.0 ma m~.m mx 36 nx 3.0 mx .. 0x I 0* I mx I ~x u ax 8kg 0950000009 80 00 300.0 000000 00600000009 80 00 0030060 00000000 000000 00000ng 09500000.? mo: 00 :00: A0. uCBV OH 0.309 105 . 00.0 00. 00.- 00. 00.0 00 00 00.0 00.0 00.0 00.0 00. 00.0 00 000 00 00.0000000 00.00000 00.0 00.000000 00.000000 00 000 n 00.0 00.0 00.0: 00.0 00.0 00 000 00 00.00 00.0 00.0 00.00 00.0 00 000 . 00.00 00.0 00.0 00.00 00.00 00 000 000 00.000 00.00 00.0 00.00 00.000 00 000 . 00.000 00.0 00.- 00.00 00.00 00 000 . 00.00 00.0 00.0 00.0 00.00 00 000 . 00.0 00.0- 00. 00.0 00.0 00 000 n 00.0 00.0- 00.- 00. 00. 00 000 0 00.000 00.0 00.0 00.00 00.00 00 00 00 00.000 00.00 00.0 00.00 00.000 00 00 00 00.000 00.00 00.0 00.00 00.000 00 00 0 00.0000 00.0 00.0 00.000 00.000 00 00 00- 00.0000 00.00 00.0 00.000 00.000 00 00 00 00.000 00.00 00.0 00.00 00.000 00 00 00 00.0000 00.0 00.0 00.000 00.000 00 00 00 00.000 00.00 00.0 00.00 00.000 00 00 00 00.0000 00.0 00.0 00.000 00.000 00 00 00 00.0000 00.00 00.0 00.000 00.000 00 0 003.008 880 0268: 0.080 0500000: 3000 800.0050 000.0500 :8: 2 0880000 00. 00.080 OH 0.30.0. 106 00.0 00. 00. mm» 00. 00.0 00.00 0 u I u 000 00.0 00. 00.0 000 u u u 000 00.0 00. 00.0 000 - - u 000 n u u 000 - - u 000 u - u 000 00.0 00. 00.0 0M0 00.0- 00. 00.0 00 00.0 00. 00.0 00 ~0.0 00. 00.0 00 00.0 00. 00.0 00 00.0 00. 00.0 00 00.0 00. 00.0 00 00.0 00. 00.0 00 00.0- 00. 00.0 00 00.0 00. 00.0 0 000000 000000 0900000009 80 000000 00000000H 09000000009 0.000 “00 300.0 “00 0000000000 0.0000000 09000000009 0.0000 00 000: IHI‘F 00.0080 OH 0.309 107 (a) a single batch of data, (b) the relationship between two (or more) variables, or (c) a time series of observations. For our purposes here, the method of selecting a transforma- tion which focuses on a single batch of data was deemed as most appropriate given the exploratory nature of the present research. More specifically, the criterion which governed the search for a transformation was to find a single bend transformation to satisfy the equation: x* = ln(x+k) where: x* = the transformed value of the variable x = the raw data value of the variable k = a constant, * such that the skew of the distribution of x was between i 2.00.3 Columns 7-10 of the Table 10 present the constant (k) that satisfied the above equation,4 and the standard deviation, and skew of the transformed scores. As can be seen from the data in columns 2-10, the problems with re- spect to the distribution of the measures have been resolved. Consequently, we are now in a position to estimate (i.e., derive values for) the parameters of the model under inves- tigation. However, prior to solving for the parameters of the model, it would be extremely useful to examine the cor- relations among the indicators of the model. Since a cor- relation reflects the bivariate linear association between two variables (i.e., the extent to which variation in one 108 variable is linearly associated with variation in another variable), a close inspection of the correlations among the indicators of the model should provide us with information on the extent to which: (a) the indicators of each theo- retic construct are linearly related to one another, and (b) the linear relation of the indicators of predictor con- structs to the indicators of predicted constructs. This information will allow us to determine the extent to which the correct functional form of the model has been specified. Analysis of the Correlations Among Indicators Tables ll-13 report the correlations among all indi- cators in the model depicted in Figure 2. More specifically, Table 11 reports the correlations among the indicators of the exogenous variable (5), Table 12 reports the correlations among the indicators for the three endogenous variables (n1, n2, and n3), and Table 13 reports the inter-correlations among the indicators for the exogenous and endogenous vari- ables (i.e., the inter-correlations of the indicators of g with the indicators for n1: n2. and n3). Some of the cor- relations contained in Tables 11-13 will be examined in the next few sections. Analysis of the correlations among the exogenous indicators.5 An examination of Table 11 reveals three dis- tinct groups of inter-correlations. First, the correlations among variables x -x are all positive and low to moderate l 5 in magnitude (.114 to .457), with eight of the ten 109 .H 50:09? 00m .muou0osg 05 mo ammo? 0 Mom .3 3306 80 “:x on wx 3309.6»: HOw wwsoom 20mg :0 0003 ms 033050 mg .uou0oflg 2000 m0 030 39.0w 05 How 3 0.309 080 so..w a I. t mo.vd .1 «. oma. vmo.l HHx ooo.a «taav. «avmm. *«mmm. «emmv. «iqmv. voo.l woo. moa. ooo.s .«vmm. mso. .«mam. ..mm~. smo.- moo.u .mms. oeo.u mmo. osx ooo.s sms. mes. ..0mm. oms.- one. avo.u vos.- mmo. ax ooo.s .«msv. .«mmv. ooo.u mms.u mso.u meo.- .mms.- mx ooo.s ..asm. eso.- smo.- cos. «so.u moo.n N.x ooo.s ~oo.- mmo. mac. mmo. mwo. ox ooo.H .«hmv. ttomv. «cmmH. «emmm. mx ooo.s .«msw. ..-v. .«mmv. ex ooo.s ass. ems. mx coo.s ..smm. Nx coo.s sx ssx osx ox ax ex ex mx ex mx mx sx mamnzc 888:5 Eocmséa 05 992s mcoflmswssoo AH QHQMB llO correlations significant at the .05 level or lower. This suggests a moderate but direct relationship between family income (x1), parents'occupational status (x2 and x3), and 6 parents' educational attainment level (x4 and x5). 6-xll are all positive and low to moderate in magnitude (.018 to .488), Second, the correlations among variables x with twelve of the fifteen correlations significant at the .01 level or lower. These findings suggest a moderate but direct relationship between these six indicators of life- style. Third, the correlations among the variables contain- ed in the x -x set, and the variables in the x -x l 5 6 11 variable set are all near zero or very low in absolute mag- nitude (-.004 to --.193).7 Further, seventeen of the thirty correlations are negative, with only two of the thirty cor- relations significant at the .05 level or lower. These findings suggest that there is little relationship between the socioeconomic status variables (x -x5) and the lifestyle 1 variables (x6-x11) that were used as indicators of the indi- vidual's social structural background (5)- In sum, there is a moderate linear relationship among the five socioeconomic status indicators xl-xs, a moderate linear relationship among the six lifestyle indicators x6- xll’ but very little relationship between these two sets of indicators. From these findings we conclude that while the socioeconomic variables and lifestyle variables correlate moderately and positively within each set (i.e., amongst lll themselves), there is little relationship between socio- economic status and one's lifestyle pattern (at least with respect to which the lifestyle pattern is socially respon- sive, diverse, independent, nurturing, cohesive or intellec- tual). Analysis of the correlations among the endogenous indicators. Table 12 reports the correlations among all three sets of endogenous indicators. For our purposes here, variables y1 will be referred to as the first set of "y12 endogenous variables (since they are indicators of n1): Yl3‘ yl7 will be referred to as the second set of endogenous variables (since they are indicators of n2), and variables le-y21 will be referred to as the third set of endogenous variables (since they are indicators of U3)- For the first set of endogenous variables, an exam— ination of Table 12 reveals three distinct groupings of cor- relations among the twelve indicators of information environ- ment (n1). First, there is a moderate to high positive correlation among variables yl-y9 (.197 to .702), with all thirty-six correlations significant at the .05 level or lower. Second, there is little relationship between the set of variables yl-y9 and the set of variables ylo-y12 (-.001 to -.304). Twenty of these thirty correlations are less than .10 in absolute magnitude. Further, sixteen of these thirty correlations are negative in direction. Finally, only nine of the thirty correlations in this sub-group were 2 l 1 0080305 05 mo E00030 0 sons 00.800 00.58005 00 0003 mm 0003.000 03 .om .OH manna wmm . h 3 ms» 000 5» .3» B a» £30305 0000 m0 00.."0 39000 0:... How ca 0.38. 0000 . H x3009? 000 0030305 MOM so. v m I. «a mo. v d i omo.- «oos.: omo.- oms.u oss.u ooo.- mso.u mso. omo. ooo.- omo.- sos.a mmo.u mmm «.ssm. omo. mos. oms. «mos. sms. ooo.- soo.- ooo. mmo.- omo. smo. mos. as» «mos. omo.- omo. ems. oso. «.ooo. omo. «omm. .«omm. oos. w.omm. mmo. «.mom. ms» .«msm.u moo.u omo.- oms.u ..omm.u oos.- mms. moo. moo.- omo. oos.- moo. omo.: ms» oos. oso.u smo. ooo. omo.- ..oom. moo. .mom. .qmom. «oos. .«vsm. omo.- .«mmm. ms» omo.- oms.- omo.- ooo. mms. mmo.u ..oom.- soo.- ooo.- .«mmm.u smo.n mso.u moo.n ms» .oos. omo.- mso.n oms. .«mom. .«mmm. ooo. .«mmm. «.mom. soo. «.osm. oos. «.omm. es» «.moo.u ooo.- sso. oos. omo.- mms. oms. ..omm. .osm. mos. mms. ooo. .ssm. ms» ooo.s moo. omo. sms. soo.- ooo. omo.- «mos.u mos.- mmo.u moo. omo.: soos.n ms» ooo.s ..mom.- moo.u ooo. omo.- .oms. omo. oso. .mos. «so. .oom. mso.u ssh ooo.s omo.- ooo.- moo. «woom.u mmo.- moo.u .«oom.u mos.- «mmm.n .mos.u os» ooo.s .oom. omo. moo.u omo. soo.- oso. soo. oso. ooo.- o» ooo.H taamv. «emmm. «aaom. «sane. «ammm. «tame. «cmbm. «sawm. m» ooo.s .«oom. ..osm. «.mom. ..mom. .«som. «oos. ..moo. om ooo.a cavvo. «tamv. «emom. «teem. «some. 00550. 0% ooo.H «avmm. «saam. «ammo. «eahm. «05mm. ma ooo.s .«ooo. «.msm. .«omm. «.omo. o» ooo.s .«oom. «.oom. «.mmm. ms ooo.s ..oom. .«mom. ms ooo.s «.omo. s» ooo.s s NH wanmfi aloonzo msoumOsocs msocmooccm any 009:0 000sumswssoo 113 smrs ooo.s .sos.- omo.- mss. ooo. ooo. mms.u moo. om ooo.s .«moo. .«omo. .«mmm.u .osm. .soom. sos. as» ooo.s .omm.u «.ooo. .oos.u .«oom. .«omm. ms» ooo.s moo.- ..mom.u ..mom.u mmo. ms» ooo.s sos.- ..oom. ..mmo. ms» ooo.s «.oom. mos.u ms» ooo.s .oos. as» ooo.s » smm omxs osts ms.s mss msrs msts es.s A0.u000v NH QHDMB 114 significant at the .05 level or lower. Third, there is an extremely small inverse relationship among the set of vari- ables le-yl2 (-.059 to -.293), with only one of the three correlations in this set significant (Eylly12 = -.293, p i .01). In sum, these findings suggest that there is: (a) a moderate to high linear relationship among the variables which tapped the relative complexity of the individual's interpersonal and media experiences (yl-yg), (b) little relationship between variables yl—y9 and variables ylo-y12 (which measure number of hours talking with friends, read- ing print media, and watching TV news, respectively). and (c) a low inverse relationship among the set of variables le—ylz. Hence, while variables which measure the relative complexity of the individual's interpersonal and media en- vironment correlate in a moderate and positive fashion, the remainder of the findings lead us to conclude that the ex— tent to which an individual interacts with friends is in- versely related to the amount of time spent with print and electronic media, and that the relative amount of time spent interacting interpersonally and/or with the mass media is unrelated to the relative complexity of these experiences. The second set of endogenous variables (y13-y17) are the indicators of processing style (n2). An examination of Table 12 indicates that: (a) the correlations of y15 with all other variables in the set are moderate and positive (.189 to .508), with all correlations significant at the .05 115 level or lower, (b) the correlations between y16 and y13, y14, and y17 are small, negative and non-significant (-.058, -.103, and -.l41, respectively}: (C) the correlation be- tween y13 and yl4 is moderate, negative and significant (-.496, p i .05), while the correlation between yl3 and yl7 is small, positive but not significant (.149), and (d) the correlation between yl4 and y17 is moderate, positive and significant (.432, p i .05). These findings suggest that: (l) as the coefficient of variation of the set of MMDS responses increases, the total number of scores which are used, the number of integer multiples of the standard which are used, and the number of extremely small and large scores which are used also increase, (2) as the number of extremely small scores which are used increases, the total number of scores which are used, the number of integer multiples of the standard which are used, and the number of extremely large scores which are used decrease, (3) as the total number of (unique) scores which are used increases, the number of integer multiples of the standard which are used decreases, and the number of extremely large scores which are used increases, and (4) as the number of integer multiples of the standard which are used increases, the number of extremely large scores which are used increases. However, it is important to remember that these conclusions are based on a set of correlations which included a few small and nonsignificant coefficients. 116 The third set of endogenous variables (le-y21) are the indicators of cognitive structure (U ). An examination 3 of Table 12 indicates that: (a) the correlations between )HJ3W1thy19 and y20 are both negative and Significant (-.220 and —.7S4, respectively), while the correlation between le and y21 is small, positive, but not significant (.113), (b) the correlation between yl9 and y20 is positive and signifi- cant (.402, p i .05), while the correlation between y19 and is near zero and nonsignificant, and (c) the correla- y21 tion between y20 and y21 is small, negative and significant (-.191, p i .01). These findings suggest: (I) the correlation between the dimension- ality measure (H) derived from the Scott "Listing and Grouping" task and the MMDS measure of dimensionality is positive but not significant, (2) the dimensionality measure H seems un- related to the MMDS measure of discrimi- nation, and inversely related to the MMDS measure of integration, (3) the MMDS measure of dimensionality is inversely related to the MMDS measure of discrimination and directl related to the MMDS measure of integration, and (4) the MMDS measure of discrimination is inversely related to the MMDS measure of integration. While no Specific predictions were made as to how the indi- cators of cognitive structure would correlate with each other, low to moderate positive (and significant) correla- tions were anticipated. Consequently, the lack of a strong relationship between the Scott measure and the MMDS measures 117 was expected. Further, the inverse relationship between the MMDS measures of dimensionality and integration, while unexpected, indicates that the distance variance in the MMDS space is being accounted for more by imaginary dimen- sions as the individual generates large triangular inequal- ities in the set of paired comparisons. Analysis of the Correlations Among the Indicators of Predictor Variables with the Indicators of Predicted Variables In the preceding section the correlations among the indicators for each theoretic variable were examined to see how well these measures correlated with each other. However, this discussion was limited to examining these measures with- in each set (i.e., theoretic variable by theoretic variable). It would now be useful to examine how the indicators of ppg- dictor theoretic variables correlate with the indicators of predicted theoretic variables. Such an examination would help us determine the extent to which the causal relation- ships among the theoretic variables in the model are plaus- ible given this set of indicators. In the following sections four sets of correlations will be examined: (1) the correlations among the indicators of E and n (2) the 1’ correlations among the indicators of n1 and n2, (3) the cor- relations among the indicators of n1 and n3, and (4) the correlations among the indicators of Hz and n3- Then, a summary of the implications of these relationships will be presented. 118 Analysis of the correlations among the indicators of g and n1. Table 13 reports the correlations among the indi- cators of g(xl-xll) and n1(yl-y12). An inspection of Table 13 reveals that the most distinct grouping of correlations Of the rests in the relationship of yl-y8 with x -x 6 11° forty-eight correlations in the subset, forty-five are low to moderate positive correlations (.035 to .559), with forty-one of these forty-five correlations significant at the .05 level or lower. The three correlations which are negative in direction are all near zero (-.086, -.010 and -.068) and nonsignificant. A second finding among the cor- relations of indicators with n1 indicators is the rela- tionship between x x and x with y1 to y7. Of the twenty- l' 2 4 one correlations in this subset, all are either very low or near zero in absolute magnitude (-.009 to -.l75), with only one significant correlation (:x =-.l75, p i .01). 1Y2 Third, of the sixty-three remaining correlations in the overall set of g and n1 indicators, all are either very low or near zero in absolute magnitude (.000 to .208), with the exception of one correlation (:y9x10=.334, p i .05). Further, there seems to be no distinct trend in the direc- tion of any of these correlations, with the exception of the correlations of x3 with y1 to yll' which are all positive. Finally, only seven of the remaining sixty-three correlations are significant at the .05 level or lower (rx y rx y rx y rx y rx y rx y and rx y ) — 2 12'— 4 ll'— 5 ll'— 5 S'— 5 8'— 8 11' 10 9 ' .H xmucwdm< 000 00000305 05 mo >000on 0 .80 .3 3009 000 waaxnmx 000 ca» 8 Ha 00000009." new 000000 00580000... 00 00003 0a 300an 03 #80300.“ 0000 mo 00mm 3.0000 05 How OH mafia 0000 119 Hmo. omo.! hma. hmo. hmH.I mac. mmo. veo. Nmo.l moo. OOH. mad. mmH. «wha.l «5mm. «aaoa. ONH. atmmm. «tmmm. mvm. «hNN.I mmo. moa. hHo.l «va.l voo.l NHH.I oma. «imam. amaa. mNH. «chow. «tome. «amam. vMH.I mmo. mmo. mmo.l mao.l bmo.l moo.l mqa. cod. cemmm. vac. «can. «evmm. «ewam. «HON.I hmc. toom. «NmH. aha. mHH. «mom. «vhH.I «no. Hmo.l «eomm.l wmo. 00H. hmo.l woo. ovo. moo. hNO.I mao.l mvo. mmo. mmo. «oom.l mmo.l voo.l emom.l hmH.I mmo.l who. mmo.l mva. moo. moa.l had. mmo. mac. boa. «1vmm. OHH. boo. mwd. boa. mmo.l t«omv. «towm. «ammm. «tvww. aammm. «amwv. ¢v>H.I «tomm. «tome. moo.l mmo. «amwm. «wma. one. «aamv. acmmm. aona. «aovm. «tmvv. cammm. mmo. «ammv. «tvmv. aammm. «ammm. atmmm. «ehmm. «moH.I itmov. «tmmv. mad. «mma. «amen. «cmhm. Nmo. «amam. «twee. «mod. «emov. aemvv. «eoav. Nvo.l mma. «000v. OHO.I omo.! mmH. *mNH. moo. «ammv. «aamm. «aamm. «00mm. «*Hhm. aemmm. «HoH.I .3. Iv. 9; mo. v m .1 soo.- mos. Wm» ooo. sms.- as» moo. .sos.- oss omo. oos. ms» ooo. oos.- ms» «*5vm. «swam. mah moo. oso. es» omo.- omo.- ms» omo.- oos.- ms» .oos.u oos.- sss moo. soo. oss moo.u moo. ms mmo. omo. o» sos. moo. o» sos.- .mos.- ms moo.u omo.- mm oms.u ooo.- o» oos.- mms.- ms mso.u oms.u ms oms.u oso. » ooo.- moo.- s» ssx osx ox ox ox ox mx mx sx MA¢®HZV mHOumUflUCH mDOCMVUnuE mde gewg 05 06—05 mCOHUMHwHHOU ma manmh 120 Analysis of the correlations among the indicators of n1 and n2. Table 14 reports the correlations among the indicators of nl(yl -y12) Wlth the indicators of n2(y13-yl7). A close inspection of Table 14 indicates that the correla- tions among the indicators of gland n2 are all moderate to near zero in absolute magnitude (-.015 to .385), with only eighteen of the sixty correlations in this subset signifi- cant at the .05 level or lower. Of these eighteen signifi- cant correlations, six involve y15 and these are all positive, five involve yl7 and these are all positive, and three involve yl4 and these are all positive. The four other significant correlations (£y1y13, £y4y16, £y16y13, ry7yl6) however, are all negative. Consequently, the most noteworthy relationships evident in the set of correlations among the indicators of ml and Hz in general are: (l) the positive linear relationship between n2 indicators yl4, ylS' and y17 with the indicators of n1, (2) the negative linear relationship between n2 indicators y13 and y16 with the indicators of ml, and (3) the trend of fourteen of the eighteen significant correlations to be positive and associated with n2 indicators y14, y15 and y17' Analysis of the correlations among the indicators of nl_§pd_fl3. Table 15 reports the correlations among the indicators of nl(yl ). As can be seen 'yiz) and n3(Y18’Y21 121 ms ms os s .s xsocmddm 000 000003003 05 m0 @000on 0 00.0 .3 0309 000 u 0” 0:0 . w . .m 3 0” 000003003 00w 000000 00530003 00 00000 03 0303000 035. #3003003 0000 m0 030 39:00 05 How 3 magma 0000 S. w 0 I «a mo. v d a. oso.- smo. omo. omo.- .«omm. moo. mom. ..mom. «oos. ..osm. mmo.u «.mom. ms» oms.- omo.- ooo. oms. moo.- ..oom.- sso.- ooo.- .«mmm.u smo.- mso.- mmo.- ms» wmo.l wHO.I mmH. «amvm. 00mmm. ooo. «emwm. «amhm. H0O. camam. OOH. «0mmm. max ooo.- sso. oos. omo.- oms. oms. ..omm. .msm. mos. mms. ooo. .ssm. as» mso. omo. sms. omo.- ooo. omo.- .mos.u mos.- mmo.u omo. omo.- .oos.u ms» 0Amonzv m: 0.0 a: 00 0.800305 05 000.04 0003030008 va 0.309 122 from Table 15, only eight of the forty-eight correlations in this subset are significant at the .05 level or lower, and five of these eight significant correlations involve y19' Further, while there is a generally pgsitive relationship between n3 indicators yl9 and y20 and the indicators of n1. and a generally negative relationship between 03 indicators y18 and y21 and the indicators of 0 only three of the 1' thirty-six correlations including these indicators are sig- nificant. Consequently, the most noteworthy finding yielded by this analysis is that of no linear relationship between the indicators of 01 and n3, with the exception of y19' which seems to be moderately (and positively) related to five of the indicators of 01. Analysis of the correlations among the indicators of 22 and n3. Table 16 reports the correlations between the indicators of 02(yl3-y17) and n3(y18-y21). An examination of Table 16 reveals that twelve of the sixteen correlations between le-yzo are significant at the .05 level or lower. Further, in general, there is a negative relationship be- tween y18 and the indicators of n (for the exception of 2 £y14y18 which is near zero and nonsignificant), and a pOSitive relationship between ylg-y20 and the indicators of 2 significant at the .05 level). Finally, it is quite evident n (with the exception of £yl6y19 which is negative and that y21 has the weakest relationship of any n3 indicator with yl3-yl7, with all correlation coefficients near zero om as os s .s xsoommmo 000 0000003003 0.30 00 #0000030 0 00....— .3 0300.3. 000 u > B h 000 . x B 03 0000003003 00.0 000000 00.0—00000000 00 00000 03 0303000 035. 00003003 0000 no 0030 039000 05 .000 3 0300.3. 0000 123 so..w.o ul- *¥ 00. v a t. «oos.- mmo.- oms.- oss.- ooo.- mso.- mso. mmo. ooo.- omo.- sos.- mmo.- sms mmo. mos. oms. .oos. sos. ooo.- soo.- ooo. mmo.- ooo. smo. oos. om» mmo.- moo. oms. oso. ..ooo. moo. .omm. .«oom. oms. .«omm. mmo. «.mom. ass ooo.- omo.- oms.- «.omm.- oos.- oms. moo. ooo.- omo. oos.- mmo. omo.- ms» mss ssis osKs sos oo o» m.s m.s o» m» m.s s» 0304: mc 000 3: h00 000003003 05 00953 000300300000 03 0340.3. 124 Table 16 Correlations Among the Indicators of n2 and n3 (N=99)a y13 y14 ylS y16 Y17 Y18 -.315** .035 -.506** -.246** -.042 Y19 ~182* -538** .587** -.1a4* .888** yZO -3ll** .101 .669** .217* .255** y21 -.056 .042 -.155 .007 .008 *p g .05 it p 3 .01 aSee Table 10 for the sample size of each indicator. This analysis is based on transformed scores for indicators y For a glossa }§'o Y I Y I lehe igdi and y cator , o I see Table 10. see Appendix I. 125 (-.007 to -.155) and nonsignificant. In review, the most noteworthy findings discovered during the examination of and n correlations were: n2 3 (l) the generally ne ative relationship between y18 and the indicators of n , 2 (2) the generally positive relationship between y and y and the indicators of r 19 20 , , and 2 (3) the lack of a linear relationship between y21 and the indicators of n2. A Summary and Discussion of the Correlation Analyses Performed on the Indicators of the Model In the previous two sections we have examined the correlations among the indicators of the model to see: (a) how indicators of the same theoretic construct correlated with one another, and (b) how the indicators of predictor variables correlated with the indicators of predicted vari- ables. While no specific predictions were presented as to how these indicators would correlate with each other, in an "ideal" situation all correlation coefficients would be moderate to high in absolute magnitude and significant. However, as we have observed in the preceding sections many of the correlation coefficients both within and across theoretic constructs were either near zero and/or nonsignifi- cant. More specifically, only 114 of the 265 inter-correla- tions examined were significant at the .05 level or lower. These findings call into question the validity of some of the indicators. Consequently, further data analysis must be 126 conducted to examine the extent to which the indicators are valid measures of the theoretic constructs. Since the model is essentially comprised of four theoretic factors with multiple indicators, this can be accomplished by conducting a confirmatory factor analysis on the indicators of each theoretic construct to see how well these indica- tors fit a single factor structure. These analyses are reported below. An Examination of the Pit of the Indicators To determine the extent to which the indicators of the measurement model fit the factor structures which are implicit in the model, a confirmatory factor analysis was performed for each theoretic construct. This technique provides information on the extent to which the indicators of the model reflect the underlying theoretic construct. More specifically, the LISREéE)(J6reskog and Sorbom, 1978) program provides maximum likelihood estimates of the factor score coefficients, and a statistic (which is dis- tributed x2 in large samples) which provides information on the "goodness of fit" of the model.8 Tables 17-20 report the results of the confirma— tory factor analyses that were performed on the indicators of 5, n1, n2 and n3, respectively. In these tables the following data are reported: (a) the maximum likelihood estimates of the factor loadings, (b) the standard errors .00000000 00 000000 000 0000000 00000500 Amwwmqu 000 00 0000.000 0000000803 00000 000 00000000 00000 00000000 03090 .000000000 000000000 0 00 0>000 00 00X0m 00 0030> 00090 .3 00000000 000 .0000000000 000 00 >000003m 0 000 .O3 03009 000 “330 00 0x 0000000000 000 000000 00000000000 00 00000 00 000x3000 0009 .000000000 0000 00 0000 030000 000 000 O3 0300B 0000 127 oooo. u 0m>m0 00000000000 oo.om0 u x ow u m0 Ao0o.o oo. ooom. loo ~m.ooo 00x Aomo.o o0. ooo0. loo 0v.00m o0x Aomo.o 00. ooo0. loo om.mo~ ox Ammo.o o0. 00o0. lo. mo.mom ox A0mo.o 00. ooom. loo om.0om ox l~0o.o oo. oomm. loo oo.o0m ox lo0o.o N0. mmmo.- lom.00m0o mm.om- ox Ammo.o m0. oooo.- Amo.o0mo mo.m0- ox Amoo.o 0o. mo0o. Ao0.mmoo oo.m~ mx Amoo.o 0o. oHoo. Amo.000o o0.~ 0x “ooo.o vo. oooo. loo.oo noo.0 0x 000000 00 00000 0000000 000000 0000000 00000000H 00000 .0000 I00000 000000000 I0000000 00000000002 00 00000 0030000 00000000m 00000000000003 0Aomuzv w 00 0000003003 000 00 000%3000 000000 >0000E000000 d 03 0300B .000003003 000000000 0 00 0>000 00 00x30 03 0030> 03000 .3 03000000 000 .0000003003 000 00 00000030 0 000 .o3 03000 000 0°30 00 3% 0000003003 000 000000 00000000000 00 00000 03 03003000 0300 .000003003 0000 00 0030 030000 000 000 03 03008 00m0 128 oooo. n Hm>m0 00000000000 om.m0.~ u x om u m0 Am0.o o~.0 no. Ao0.o m0. 00» loo.o mo. mo.u loo.o oo.- 00» loo.o om. mo. lo0.o oo. o0» 00o.o oo. 00. loo.o om. o0 00o.o oo. 00. loo.o oo. o» Amo.o 0N. ma. loo.oo noo.0 00 Amo.o o0. mm. Ao0.o ~m.0 00 00o.o mo. mm. loo.o om. m0 00o.o o0. mm. A00.o mo. «0 00o.o oo. 00. loo.o 0o. m0 Amo.o o0. om. 000.o «o. 0» 00o.o oo. 0m. loo.o on. 30 000000 00 00000 0003000 000000 0003000 000003003 00000 .0000 I00000 000000000 10000000 00000000002 00 00000 0030000 00000300m 0003000000000: 00mmnzv 3: 00 0000003003 000 00 03003000 000000 000000030000 < 03 0300B 129 .000003003 000000000 0 00 0>000 00 00x30 03 0030> 03090 .H x300000€ 000 .0000003003 000 00 00000030 0 000 .03 03009 000 “030 000 m30 0000003003 000 000000 0000000000» 00 00000 03 03003000 0309 .000003003 0000 00 0030 030000 000 000 03 03009 0000 oooo. u 00>00 00300000000 mm.mh u x m u we Rom.mv «v.0- m0.0 030.00 0o.~ h00 Am0.0~v mo.mm0 0m.- 030.0 00.0- 000 00o.v we. we. 00o.v mo. m00 000.0m 00.000 mm.m Amm.0v om.m 000 000.00 00.00 mm. Aoo.ov noo.0 300 0003000 000000 0003000 00000300H 000000 00 00000 I00000 000000000 I0000000 00000 .000V 00000050002 00 00000 0030000 000003000 0003000000000: N: 00 0000003000 000 00 0300300< 000000 000000030000 0 03 0300B mAmmuzv .uoumuwocfi mocmuwwmu m an m>umm on cwam ma msam> manan .H chcmmm< wmm .muoumofiocfi may no aummmon m mom .OH magma mum “omh on ma» muoumowvcw Mom mmuoom cwfiuommcmuu no.6mmma ma mwmxamcm macs .uoumoficcw comm mo muwm meEmm wnu How ca manna mmmm 130 chmo. u Hm>mH xuflaflnmnoum mm.o u x N"? AHH.V vb. mo. Amo.v NH. AN» Am~.mav mo.H m~.H Aem.~mv Hm.m om» AHm.V HH.~ mo.a AmH.V mv.n ma» Ahm.v H~.H ma. Aoo.o. noo.H ma» .uouuw Mo uouum omNflcum Auouum amnwcum uoumoflccH Houum .numv Iwcmum oumvcmumv Incmumc: acmsmusmmmz mo wouum mcwcmoq nmumeumm cwuwoumccmumca mammuzv m: mo muoumoflccH may no mamaamc¢ uouomm auoum5uwucou < om manna 131 of these coefficients, (c) the measurement error associated with each indicator, and (d) the chi-square statistic, degrees of freedom, and probability level yielded by the "likelihood ratio" technique used to test the overall "goodness of fit" of the model.9 As can be seen in Tables 17-20, one of the indicators in each factor stucture (x1, Y7: yl3 and le for 5, n1, n2 and n3, respectively) was fixed at 1.00 to serve as a reference indicator (i.e., to fix the scales of measurement so that the magnitudes of these are interpreted relative to one another; this is necessary for model identification). However, a detailed discussion of the values obtained for these parameters is unwarranted in the present research given the chi-square values and probability levels obtained for each theoretic construct. More specifically, a close inspection of Tables 17-20 reveals the following chi-square values, degrees of freedom, and probability levels for each theo- retic construct: Theoretic variable chi—square df probability level 5 139.00 44 .0000 nl 213.34 54 .0000 n2 73.95 5 .0000 6.59 2 .0370 132 These results indicate that the specified models are too restrictive. In other words, the indicators for each theoretic construct do not fit single factor structures. Consequently, any further discussion of the specific results obtained from the confirmatory factor analyses would have marginal utility. Nonetheless, to further examine the extent to which the theoretic constructs were erroneously specified, canonical correlation analyses were performed. Using the indicators within each factor, this procedure derives a series of linear combinations from each set of indicators so as to maximize the correlation between these linear combinations. Each linear combination extracted from a set of indicators is referred to as a canonical variate, and the correlation between each (corresponding) pair of canonical variates is called the canonical correlation. Further, the square of the canonical correlation represents the amount of variance in one canonical variate accounted for by the other. Tables 21-24 report the canonical correlations among the indicators of the predictor and predicted constructs. As can be observed in these tables, there are two signifi- cant canonical correlations between the indicators of E and n1; there are two significant canonical correlations between the indicators of ml and n2; there are no significant canon- ical correlations between the indicators of ml and n3, and ca H H WH xflocomm< mom .muoumoflocfl mcu mo mummmon m mom .OH manna wow u m on N new A x 0» x muoumoflocfl now mouoom ooEHOM Imcmuu co ommmo ma mfimxamcm mace .uouwoflocfl comm mo muwm meEmm on» How ca wanna mom «muoumofl©Cw Ham mmouom mumo mcammfle ou mop coflumcwfiwao mmmo mo uaammu 0:» mafia mHQEmmm 133 Nmm. _ N mmo. mam. hNo. Hoo. Ha mom. 0 me.N mmm. HON. owe. 0H mmm. NH vmm.v mom. NMN. vmo. m mom. 0N mmm.m mNm. mom. Nmo. m mmm. om va.ba Nah. mwm. mma. h Nmm. Ne «mm.hN Hmm. va. vma. m NNm. mm OHh.Hv mvv. vmw. va. m mow. Nb mmv.ao mom. com. mam. v hmm. om mmN.mm oma. 5mm. woe. m oma. OHH 0mm.mNH mmo. was. mam. N oao. Nma moN.N>H mmo. nub. mom. H oocmoflMflcmam up mumsqm ch moQqu m.xafi3 cofiuwamuuou Hmowcocmu moam>comwm umnEsz If mAvmuzv a: mo muoumofiocH on» :ufl3 u no muoumowOCH on» mc05< mcofiuwaouuou Hmoficocmo HN manna 134 Table 22 Canonical Correlations Among the Indicators of ml with the Indicators of n2 (N=64)a Canonical Wilk's Chi Signif- Number Eigenvalue Correlation Lambda Square df icance l .562 .750 .120 115.297 60 .000 2 .462 .680 .276 70.246 44 .007 3 .302 .549 .513 36.430 30 .194 4 .177 .421 .734 16.863 18 .533 5 .108 .328 .892 6.220 8 .623 aSample size the result of case elimination due to missing data across all indicators; see Table 10 for the sample size of each indicator. This analysis is based on the transformed scores for indicators y to y , y S' and ylz; see Table 10. For a glossary of the indiégtors, see Appendix I. 135 Table 23 Canonical Correlations Among the Indicators of n with the Indicators of n3 (N=64)a 1 Canonical Wilk's Chi Signif- Number Eigenvalue Correlation Lambda Square df icance 1 .285 .534 .469 41.689 48 .728 2 .224 .473 .656 23.227 33 .897 3 .109 .330 .844 9.303 20 .979 4 .053 .230 .947 2.978 9 .965 3Sample size the result of case elimination due to missing data across all indicators; see Table 10 for the sample size of each indicator. This analysis is based on trans- formed scores for indicators y to y and ylg to YZO; see , ee Table 10. For a glossary of the indigators Appendix I. 136 Table 24 Canonical Correlations Among the Indicators of n with the Indicators of n3 (N=64)a 2 —. a- __ Canonical Wilk's Chi Signif- Number Eigenvalue Correlation Lambda Square df icance l .912 .955 .046 179.913 20 .000 2 .466 .683 .525 37.654 12 .000 3 .010 .102 .984 .958 6 .987 4 .006 .097 .994 .348 2 .840 aSample size the result of case elimination due to missing data across all indicators; see Table 10 for the sample size of each indicator. This analysis is based on trans- formed scores for indicators y , and y19 to yzo; see Table 10. For a glossary of thg indicators, see Appendix I. 137 there are two significant canonical correlations between the indicators of n and n3. These results provide addi- 2 tional evidence that the model has been erroneously speci- fied. This is because if the model were correctly specified, we should have obtained only one_significant canonical correlation between the indicators of each set of predictor and predicted constructs. Further, these results, when examined in light of the previous discussion concerning the correlations among indicators both within and across theoretic constructs, strongly suggest that: (l) the pro- posed indicators for E are probably two sets of indicators for two different underlying (theoretic) constructs; (2) the proposed indicators for n1 are probably the indicators of at least two, and possibly three or four underlying (theoretic)' constructs; (3) the proposed indicators for n1 have little relationship with the indicators of n and (4) the proposed 3; indicators for U3 are probably two sets of indicators for two different underlying (theoretic) constructs. When the kind of model specification problems we are experiencing are encountered, Joreskog and Sdrbom recommend an analysis of the covariance residuals and/or an analysis of the first order derivatives of the parameters with respect to the fitting function (F) that was used to derive the maximum likelihood estimates for the parameters (1978: p. 15). According to these authors, such analysis of- ten will suggest ways to alter the model so that the relative 138 fit may be improved, typically through the introduction of new (i.e., additional) parameters to be estimated. However, while such analyses would be useful to improve the fit of the indicators to each theoretic construct, we have yet to examine the overall model as originally proposed. While it is unlikely that the relative fit of the sample data to the overall model will be very good, such an investigation (since it is a full information approach) might provide ad- ditional insights to the problems of the model that might otherwise go unnoticed. Test of the Full Model Table 25 reports the maximum likelihood estimates for all parameters in the full model, and the re- sults of the chi-square goodness of fit test. As expected, the results of the goodness of fit test indicates that the overall fit of the data to the model is not very good (x2 = 1222.35, df = 457, probability level = .0000). However, the ratio of the chi-square value to degrees of freedom (2.67) indicates that the model, while implausible in its present form, may likely be improved via sensitivity analy- sis which focuses on the addition or deletion of specific indicators and/or parameters. In addition to the poor fit of the data to the model, the major problems with the full model rest in: (a) the high levels of measurement error in the indicators, par- x x x x ticularly the errors associated with x x 4' 5' 7' 8' 9' 10' 139 upwusmeoo on no: “ONx Ou ma» can cllxmmlw.mai co momma ma mflm>amcm mans new: X0“ .uoumofiocfi mocouowou m mm w>uom ou omxwm ma osHm> mace acne ms» mo xummmon m uom .oH manna mom x muoumowocw no“ mmuoom omEHOMmcmuu .HOumofiocw comm mo mafia wHaEmm on» How ca manna womm .HHH “mummco mom oasoo muouuo oumocmum .anooua cOwumofimfiucmofi m.amoos on» 0» mono a ADV Hm.m om©.N AUV hNh.MH HHX luv mo.og mmm.~ luv mmm.m~ cflx loo mm.hfi «Hm.H lo. ~¢o.p ax luv hv.eH Hvo.~ lo. opm.ofl ax luv mm.~H FNm.m loo HF~.mH bx ADV NN.F omo.N ADV mHm.OH wk luv NQ.HH hem.- luv max.fl- ma luv ~m.mH mmm.- Au. amm.H- ex loo He.H and. lot was. mx luv cv.~ mmo.- loo «ed. Nx luv oe.v mag. luv nooo.H Hx Auouum MO uouuw Umnwcum Auouuw @wNMDHm MonmowvcH uouum .cumv socmum pumocmumv nocmumcs ucmEousmme Mo wouum pmnwpumpcmumca mN mHQmB mammnzv umwB cam COwusHOm Hmcoz Hank mo meMEwumm 140 luv ma.~s Hum. luv va.H luv om.“ 54H.H- loo mmm.m- om» loo AH.- som.au luv mom.nu ma» on mH.H pom. luv nooo.a ma» luv om.m www.mu luv omo.mu has luv mm.H mam. luv mam. WM» 10c No.8 pmm.au luv om~.m- «as “us «5.4 om~.H- Au. Hom.~- ma» luv mm.~m nae. luv nooc.H » luv o~.H Hoe. luv mac. MM» loo ma.m mmm.- loo mmo.- ca» luv mm.sm mam. luv soo. a» luv ~a.a ~m~.~ luv mmm. m» luv -.m mmm.~ luv «mm. as its a~.- mao.v .o. nooo.a ms luv ev.ma mom.m luv mmv.a m» luv ~m.v ~mm.~ loo mmo. «5 luv qm.oH avH.m luv mph. m» luv sa.m sva.m lo. who. mm luv mm.m~ mmo.~ luv mmo. a» luv om.m mos.m roe mom. a Amouuw mo uouuo powwoum Amouuo nonwoum uoumoHoGH nouum .oumv Iocmum Uumocmumv locmumc: acoEouzmmoz mo uouum pmuwpumpcmumco An.ucoov mN manna 141 H H h ma . c o» w Eouw ucmfloflwmmoo wcu we > u c ou c Eouw ucofloflmwmoo mcu ma mo oooo. u Hm>mH suflcfinmnoua mm.-~a u x Ame u mm mam. on NVH.AH > Ham.u luv hm~.- mmm- omo. luv woo. Hmm- amp. luv mmo. Hmmu comflpumccmum “nouuw conwcumpcmumc: ucwflofiwwmoo comm ucmfiowmwmou c0wmmmumom G AU.MCOUV mN wacme 142 ND " m ooo.c luv nmo. mg m u «on. luv soo. mm m N < u U mow. AUv ovo. w m H < “Ho. ice «as. u so 3,. mqm. ADV NHN. m N HN< U u mvm. ADV mom. 0 as. va. ADV hhw.m No prflpumccmum Auouum .cumv cwuflnumocmumcs noumEmumm mumEflumm A©.ucoov mN mHnt 143 yz. Y4, Y6, Y7: le' yl3, and y21; (b) the apparently inad- missible value for the variance of the measurement error of yl9 (-.l7l); (c) the apparently inadmissible values for the standardized regression coefficients of Ax6 to Axll' Ayl to AY9. Ay“, Ayls. Ay17, Ayl9 and Ayzo, which are all greater than i 1.00; 10 and (d) the fact that the model may not be identified. More specifically, while the model meets the counting rule for identification, it may not meet all nec- essary and sufficient conditions for identification (see footnote 2). Substantively, this implies that we are unsure as to whether the data reported in Table 25 provide a unique solution, or simply one solution from a (possibly infinite) set of solutions which could be obtained, given the present model specification and variance-covariance matrix.11 Finally, while there are several problems with the overall model, the coefficients among the theoretic vari— ables 5, n1, n2 and n3 (y, 821, 831 and 832) were all in the expected (positive) direction, which supports the three hypotheses presented earlier.12 However, these coefficients must be interpreted with extreme caution for three reasons. First, in light of the problems previously mentioned with the overall model, the extremely small unstandarized coeffi- cients for 821 and 831 may be unstable. Second, the un- standardized and standardized coefficients provided by the LISREé:>program for -821 and were originally "B31 .088 and .004, respectively. The direction of these 144 coefficients are interpreted by the author after analyzing the regression coefficients of the indicators of Th, T5 and n As can be seen from Table 22, eleven of the twelve 3. coefficients for the indicators of n1 are positive, three of the five indicators of Hz are negative, and two of the four coefficients of n3 are negative. This mix of coeffi- cients suggest that the coefficients which describe the effect of ml or n2 and n3 should be Opposite in direction to that which is obtained from the computer print out. How- 1' 02 and 03 is an artifact of the identification or specification prob- ever, if the direction of the coefficients of n lems mentioned earlier, this "direction interpretation" procedure may be erroneous. Third, because of the identifi- cation problem, there was no way to "test" the theoretic model and verify the magnitude and/or direction of the beta and coefficients independent of the model (and all its asso- ciated problems). A Discussion of the Overall Results The primary purpose of the data analysis procedures was to test the theory presented in Chapter I. To accom- plish this objective, the theoretic model generated on the basis of the text of the theory (Figure l) was translated into a Operational model via the creation of a set of indi- cators which served to operationalize each theoretic con- struct contained in the theoretic model. Table 10 simply 145 described the summary statistics which characterized each indicator contained in the full model. The primary means by which these data were to be used was to test the model using a full information maximum likelihood (linear regres- sion) estimation technique. This technique was selected primarily because it provided: (a) maximum likelihood estimates for the model parameters, and (b) a goodness of fit test which would evaluate the plausibility of the model. However, before this technique was employed, some of the sample data were transformed so as to meet certain statis- tical assumptions concerning the distribution of variables. The transformation and summary statistics for each indica- tor in the model are also reported in Table 10. Prior to testing the model, the correlations both within and across theoretic constructs were examined. Tables 11-16 reported the correlations among the indicators of the model. An analysis of these correlations indicated that the initial expectation for these correlations to all be characterized by moderate to high significant coeffi- cients was not upheld. More specifically, the analysis of the correlations revealed many correlations to be near zero and/or nonsignificant. Realizing that a multiple-indicator approach was being employed in the present research, it was suspected that these findings were a result of "zero-sum" relationship existing among some indicators of each theo- retic construct, wherein increases in one indicator 146 necessitates (or requires) decreases in another (much like increases in hours awake is associated with decreases in hours asleep). To assess this assumption, a series of confirmatory factor analyses were performed on the indica- tors of each theoretic construct. Table l7-20 reports the results of the confirmatory factor analyses which were performed on the indicators of each theoretic construct. We interpret these results to indicate that there are specification errors in the model for each indicator. More specifically, the indicators of the theoretic variables have been erroneously combined into single factor structures. This finding is consistent across all four factor structures contained in the model. In hindsight, it appears that the attempt to use macro- level13 theoretic constructs resulted in the selection of indicators which were unrelated to each other but which still may be considered an indicator of the underlying macro-level theoretic construct. To help clarify this problem, consider the following example concerning the vari- able of "health." In one sense, a person's health may be assessed by examining his/her respiration rate, heart rate and skin coloring. However, when one is ill, the symptoms (indica- tors) of the illness may be uncorrelated with the above mentioned indicators of health. Consequently, to measure one's relative health, one cannot use a macro-level approach 147 and simply combine indicators which only in spme circum- stances reflect variation in the underlying theoretic vari- able. Substantively, this implies that the theoretic constructs must be refined by: (a) creating multiple con- structs to refer to the broad categories of stimuli which are now only represented by a single construct and (b) creating more precise measures of each construct. To examine the relationship between the factor structures of each theoretic construct, a canonical corre- lation analysis was performed on the indicators of the model. Tables 21-24 report the results of the canonical correlation analysis. The major findings were: (1) two significant canonical correlations between the canonical variates of 5 and n (2) two significant canonical corre- 1' lations between the canonical variates of n1 and n2, (3) no significant canonical correlations between the canonical variates of ml and n3, and (4) two significant canonical correlations between the canonical variates of n2 and n3- On the basis of the correlation analyses, it was concluded that: (l) the indicators of g were probably the indicators to x ; the 1 5 second with x6 to x11); (2) the indicators of n1 were prob- ably the indicators of (at least) two separate factor of two separate factor structures (one with x structures (the "clearest" factor structure being the one which accounts for y1 to yg); (3) the indicators of n2 were probably the indicators of (at least) two separate factor 148 structures (but there were no clear means for determining which indicators should be combined); and (4) the indica- tors of n were probably the indicators of two separate 3 factor structures (with yl8 to y20 loading on one, and y21 loading on the other). However, these conclusions were not empirically tested. Even though there were serious specification errors in the model, it was decided to examine the maximum likeli- hood estimates of the overall model, and the overall fit of the data to the model. Table 25 reports the unstandard- ized and standardized coefficients, and the chi-square goodness of fit test statistics obtained from the LISRE6:) program. These results suggest three important conclusions. First, the model may not be identified. Consequently, the parameter estimates which were obtained may not be a unique solution. Second, in light of the identification problem, the overall fit of the data to the full model is fairly good. While the probability is .0000, the X2 value of 1222.35 with 457 degrees of freedom suggests that this model can very likely be respecified (via sensitivity an- alysis) so as to fit this set of data, and then retested using a new set of data. Finally, there were three hypoth- eses specified in Chapter I. Hypothesis One: the more ratio dominant the individual's processing style, the greater the relative complexity of the individual's cognitive structure. 149 Hypotheses Two: the greater the relative complexity of the individual's information environment, the greater the relative com- plexity of the individual's cognitive structure. Hypotheses Three: the greater the relative complexity of the individual's information environment, the more likely the individual is to emply a ratio- dominant (processing) style. An examination of the (interpreted) beta coefficients in the model show positive relationships between Hz, and n3, n1 and n3, and ml and 3, respectively. Further, the coefficient between g and n3, which supports Hypotheses l, 2, and n1 (Y) is also in the expected direction (see foot- note 12). In sum, while there were some problems with respect to model specification and identification, the three hypoth- eses postulated in Chapter I were supported by the results of the analytic procedures described in this chapter. 150 FOOTNOTES 1The following set of assumptions are the assumptions which must be met by a structural equation model to be estimated using a maximum likelihood estimation procedure. The reader is encouraged to compare this set of assumptions with the assumptions required by a least-squares procedure (e.g., see Nie et al., 1970). 2The present research employs a multiple equation model with unobserved variables. Models of this type require the ad- ditional assumption that the model meets the conditions of identification. For a discussion of these conditions for identification, see Goldberger, 1964, pp. 316-317. 3The formula for calculating the skew of a distribution of scores is given below: skewness = [(xi-x')/s]3 P-MZS n As the distribution of scores more closely approximates a gaussian distribution, the numerical value of the skew approaches zero. A negative numerical value indicates that most scores are greater than the mean, with some scores which are extremely smaller than the mean also in the distribution of scores. Conversely, a positive numerical value for the skew indicates that most scores are less than the mean, with some scores which are extremely larger than the mean also in the distribution of scores. Since the skew of a distribution reflects deviations from symmetry, it is a good indicator of the extent to which a measure is normally distributed about is mean. The values of i 2.0 were arbitrarily selected by the author as an acceptable "confidence interval" to reflect normally dis- tributed sample measures. 4It should be noted that the transformed skew values are not "optimum" values. That is, data analysis concerned with transforming the raw data stopped when an acceptable 151 skew value was obtained. A costlier method would have been to iterate to the value closest to zero for every measure. 5Due to the exploratory nature of the present research, many correlations which are near zero (e.g.rx x =.018) and/or nonsignificant are nonetheless discussed as being simply "low" correlations. While caution will be used when interpreting these correlations, the reader is alerted that this strategy will be employed throughout the correlation analysis. 61n this discussion, the term "direct relationship" will refer to a positive correlation among variables, while the term "inverse relationship" will refer to a negative cor- relation among variables. 7The term "absolute magnitude" refers to the absolute value (lxl) of the correlation coefficient (i.e., magnitude ir- respective of direction). 8Briefly, the confirmatory factor analyses reported here require the following three key assumptions: (1) that the number of factors is correctly specified; (2) that the relations are linear and correctly specified, and (3) that the errors in measurement associated with the indicators of each factor are uncorrelated. 9For a good discussion of the results provided by LISRE£E> and how they are derived, see Joreskog and Sorbom, 1978, pp. 13-15. 10For a brief discussion on the interpretation of standard- ized regression coefficients larger than i 1.0, see Fink and Mabee (1978). 11For a discussion of identification problems of the present type, see Joreskog and Sorbom, 1978, pp. 10-11. 12While no formal hypothesis was offered in Chapter I regard- ing the relationship between 5 and n , it was expected that increases in the value of the social structural indicators would be associated with increases in the relative complex- ity of the individual's information environment. 13The term "macro-level" (theoretic construct) is used here to refer to a very general theoretic construct. That is, a theoretic construct which defines a broad category of stimuli. CHAPTER IV CONCLUSION The purpose of this chapter is to: (l) summarize the objectives, methodology and results of the present research, (2) review the problems which were encountered and discuss the limitations on interpreting the findings, and (3) suggest directions for future research. Theoretic summary. Chapter I presented a theory of information processing which explained the process by which individual differences in cognitive structure occur. The theory focused on the relationships between four macro- level theoretic constructs: social structural factors, in- formation environment, processing style, and cognitive structure. The theory contended that there were four key relationships among these variables that accounted for indi- vidual differences in cognitive structure. First, the theory explained how an individual's social structural back- ground is related to the relative complexity of the individ- ual's information environment. More specifically, it was argued that social structural factors (most notably parental socioeconomic status and the extent to which the individu- al's life style was conducive for development) served as the 152 153 generating mechanism for exposure to an information environ- ment which was characterized by varied interpersonal and media experiences. That is, as social structural factors increased, the greater the likelihood that the individual would interact with a wide array of media and other individ- uals who would convey information about a wide range of tOpics, from a variety of perspectives. Second, the theory explained how the relative complexity of the individual's information environment was related to the kind of informa- tion processing style the individual develops. Two distinct types of processing style were identified, nominal dominant and ratio dominant. A nominal dominant style was defined as a style in which an individual made categorical and/or polarized judgments of phenomena. That is, an individual who employes a nominal dominant style to define the phenom- ena of his/her experience tends to use a limited set of descriptors and/or tends to judge similar but non-identical stimuli as the same (i.e., stereotypes). Alternatively, a ratio-dominant processing style was defined as a style in which the individual makes extremely fine discriminations among phenomena. That is, an individual who employs a ratio dominant style tends to use a large array of descriptors to define the stimuli of his/her experience, and tends to articulate the differences between similar but non-identical stimuli. Given these two processing styles, the theory argued that as the relative complexity of the individual's 154 information environment increased, the individual's tend- ency to employ a ratio-dominant processing style increased. The third and fourth key relationships expressed by the theory explained how the relative complexity of the indi- vidual's information environment, and the processing style the individual employs, are related to the relative com- plexity of the individual's cognitive structure. Very simply, the theory contended that there were direct (i.e., positive) relationships between: environmental complexity and cognitive complexity, and processing style and cogni- tive complexity. By "relative complexity of the individ- ual's cognitive structure," we mean: (a) the number of independent dimensions the individual employs to define the stimuli of his/her experience, (b) the extent to which the individual arrays phenomena along these dimensions, and (c) the extent to which the individual establishes relationships among the phenomena defined along the dimensions. These three aspects of cognitive structure were defined as cogni- tive differentiation, discrimination, and integration, respectively. The four key relationships outlined above were cast into the following model: 155 ‘ ———————5. n2 where: E = social structural factors HI = the relative complexity of the indi- vidual's information environment n2 = the individual's processing style n3 = the relative complexity of the individual's cognitive structure From this model, three hypotheses which focused on the relationships among the endogenous variables were presented. These hypotheses are presented below: Hypothesis One: The more ratio dominant the individual's processing style, the greater the relative complexity of the individual's cognitive structure. Hypothesis Two: The greater the relative complexity of the individual's information environment, the greater the relative complexity of the individual's cognitive structure. Hypothesis Three: The greater the relative complexity of the individual's information environment, the more likely the individual is to employ a ratio-dominant (processing) style. Methodological summary. To test the proposed theory of information processing and cognitive structure, a quasi- eXperimental research design was employed. First, each theoretic construct contained in the model was 156 operationalized using ratio scaled items. Second, these items were pretested in June of 1980 using a sample of twenty-nine undergraduates enrolled in a communication course at Michigan State University. Third, using infor- mation obtained from the pretest, a final set of question- naire items were developed. Fourth, because some of the items which were included in the final questionnaire con- cerned an event which was undergoing change, a second pre- test was conducted in October of 1980 to insure that the items were properly selected. Fifth, the final question— naire was administered to a sample of ninety-nine juniors and seniors enrolled in the Department of Communication at Michigan State University. Appendix C contains the full questionnaire which was administered, while Appendix I contains the specific items which were used to operationalize each theoretic construct. As can be observed in Appendix I, for the exception of the one measure of cognitive structure derived from the Scott "Listing and Comparing" task, each theoretic construct was operationalized using ratio scaled items. The validity of these measures will be discussed later in terms of the va- lidity of the overall model. Summary of findings. A variety of analytic tech- niques were utilized to examine the data. The results of the data analysis allowed for the following conclusions to be drawn. First, the data yielded by many of the 157 ratio scaled items was positively skewed and required trans- formation to meet certain theoretical and statistical as- sumptions. Second, contrary to the expectation that all correlations among the indicators within each theoretic construct would be significant and reflect a single theo- retic construct, many of these correlations were nonsig- nificant, and did not seem to reflect a single construct. This finding was particularly surprising for the relation- ship between the measures of coqnitive structure derived from the MMDS instrument and the cognitive structure meas- ure derived from the Scott "Listing and Comparing" task. Third, contrary to the expectation that all correlations among the indicators across each theoretic construct would be significant and moderate to high in absolute magnitude, many of these correlations were nonsignificant and/or near zero. Fourth, a canonical correlation analysis indicated that the indicators from the theoretic constructs were erroneously specified as single factor structures. That is to say, the indicators which were used for each theoretic construct were, in actuality, the indicators of several orthogonal.construct31nisspecified as the indicators of a single factor. Fifth, even though there were specification errors in the measurement model, a structural equation an- alysis yielded coefficients which tended to support each of the three hypotheses that were presented earlier. More specifically, a full information maximum likelihood 158 estimation technique yielded positive coefficients for the paths between all theoretic constructs in the model. This finding supports the relationships proposed by the theory presented in Chapter I. Finally, a chi-square goodness of fit test indicated that while the overall fit of the data to the model was poor, an "improved" fit might be obtained via a sensitivity analysis of the factor structures. Problems and Limitations While it is always a humbling task, no research effort would be complete without a discussion of those aspects of the research which limit its interpretation and generalizability. Basically, our discussion here will focus on theoretic and methodological considerations. Further, while some of these considerations are offered in light of the results, others are concerned with issues which are not directly tied to the particular findings of the present re- search. Consequently, it will be left to the reader to interpret the following considerations vis-a-vis the find- ings. Theoretic considerations. First, the theoretic con- structs were not specified clearly enough to avoid the specification problems cited earlier with respect to the factor structures. Second, while four direct relationships were proposed among the theoretic constructs, no "threshold levels" were discussed. That is to say, the theory did not discuss the different "levels" of each theoretic construct 159 such that one might determine when: (a) social structural factors would be conducive or non-conducive for development, (b) an information environment was simple or complex, (c) an individual's processing style was nominal dominant or ratio dominant, or (d) an individual's cognitive structure was simple or complex. While the specification of such "threshold levels" may not be theoretically necessary to account for differences in cognitive structure, future research should investigate this matter more thoroughly. Methodological considerations. First, while two of the three endogenous variables were concerned with how indi- viduals responded to ratio scaled items, due to the complex- ity of the questionnaire, no attempts were made to vary the presentation of items across respondents. Ideally, item presentation would have been manipulated to minimize the extent to which: (a) "training" on ratio scaled items pre- sented early on in the questionnaire biased responses to ratio scaled items presented later on in the questionnaire, and (b) fatigue factors biased responses to particular items toward the end of the questionnaire. Second, while the concepts concerning the Iranian hostage situation used in the MMDS and Scott portions of the instrument were pretested twice for stability, several dramatic events took place in Iran during the data collec- tion period. Consequently, some of the variability in these measures may (in part) be a function of history 160 rather than individual differences. Ideally, it would have been desirable to present "stable" concepts to the respond— ents, i.e., concepts whose meaning was "stable" across time. Third, because of scheduling problems, not all participants filled out the questionnaire under identical circumstances. Hence, while some participants completed the questionnaire independently (alone), others filled out the questionnaire in the presence of other participants. Further, there were differences in such "situational factors" as the date of administration, time of day, weather conditions, etc., which were beyond the control of the author. Ideally, these "situational factors" would be controlled so as to minimize their impact on participants' responses. Fourth, certainly a larger and more heterogeneous sample would serve to strengthen the internal and external validity of the pre- sent research. Fifth, this project basically employed a static design. Future research should consider the use of a time series design. Such a design would allow for: (a) an opportunity to examine the relative stability of the model’s parameters, and (b) the capability of tracing the relationships among the theoretic constructs over time (i.e., the lifespan). Sixth, in the study a sensitivity analysis of the model, which would have informed us of the effect subtle changes in the specification of the model would have on the fit of the data to the model, was not performed. 161 In light of the findings and these theoretic and methodological considerations, several recommendations can be made for future research in this area of scientific inquiry. These recommendations, which range from addition- al analyses using the present data set to radical theoretic and methodological modifications, are detailed in the next section. Directions for Future Research As was mentioned earlier, a sensitivity analysis of the full model was not conducted. Given the findings yielded by the data analysis reported in Chapter III, it is apparent that such an analysis would be useful, partic- ularly if it focused on the refinement of each factor structure proposed by the model. More Specifically, the sensitivity analysis of the model should be focused on re- fining these factor structures in two ways. First, by the simple addition or deletion of indicators, and/or second, the respecification of each single factor structure as a multiple factor structure. An example of each of these modification suggestions for the full model would be to: (a) delete indicators le-yl2 from n1 and re-examine that factor structure using only yl-yg, and (b) respecify g as Egg factor structures, one which was reflected by indicators xl-xs, and a second which was reflected by indicators x6-x11. Using confirmatory factor analysis, these new factors may be inspected, and the relative improvement these modifications 162 provide may be examined.1 Once each factor structure had been improved, addi- tional sensitivity analyses could be conducted to examine how well a set of sample data fit various portions of the model (as well as the entire model). In this form of sen- sitivity analysis, key theoretic assumptions concerning: (a) the causal paths specified in the model, and (b) the correlation among errors of prediction can be re-examined. For example, in the present model, the analyses indicate a very low path coefficient between ml and n3, and no signif- icant path from ml to n3. Certainly, a sensitivity analy- sis should be conducted to see if the effect of mi on n3 is indirect (i.e., channeled through n2) rather than direct. An additional analysis which would be interesting to perform would be to examine the effect of simply delet- ing n3 from the model. That is to say, instead of attempt- ing to derive measures of cognitive structure from information processing tasks, it may be worthwhile to simply more carefully scrutinize the processing style data, and to consider processing style as the "ultimate" dependent meas- ure. While such a strategy diverts attention from the development of a model which represents internal (cognitive) organization, it does place emphasis on the development of a model which can predict individual performance, which is a central concern of many cognitive theorists. 163 Beyond the sensitivity analysis prescribed above, attempts should be made to employ a time series design which uses larger sample sizes and a more heterogeneous sample of respondents. Three important advantages would result from the adoption of this procedure. First, the generalizability of any findings which were obtained would be greater. Second, the stability of model parameters could be better assessed. Third, changes in the key theo- retic constructs may be examined over the lifespan, and more SOphisticated "lagged" models may be deve10ped. While the first two advantages are extremely desirable when de- scribing gpy process, the third advantage may be the most important. This is because theoretical advancement in the area of cognitive development will most likely require a greater understanding of how: (1) social structural fac- tors during one stage of the lifespan influence exposure patterns to information environments during the later stages of the lifespan; (2) exposure patterns to information environments during early stages of lifespan influence ex- posure patterns to information environments during later stages of the lifespan; and (3) how processing style during early cognitive development influences processing style during later stages of the lifespan. These "over-time" considerations can only be examined through the utilization of time series designs. 164 Finally, future research efforts will have to con- centrate on the development of more precise measures for each theoretic construct. For example, future measures of information environment should be developed so as to better assess: (l) aspects of the individual's interpersonal and media experiences, and (2) the individual's perception of these interpersonal and media experiences. Further, with respect to the theoretic construct of processing style, future work should focus on Operationalizing this construct so as to better assess the individual's capability to make judgments among phenomena, both within and across domains of experience. Finally, it appears that a major problem with research concerning cognitive structure is the lack of an adequate processing task which can be used to assess general processing capabilities. While it is unclear whether or not such a task can even be developed, new ef- forts might attempt to create tasks which assess a wider array of processing skills. For example, future research might focus on the development of a processing task which assessed "matrix thinking" capabilities, i.e., which as- sessed the individual's capability to consider 2 factors simultaneously when making a decision (or judgment). If successful, such a "matrix" task might later be expanded to include the presentation of 2 factors which have varying probabilities of occurrence (or success). Nonetheless, because measures of cognitive structure are derived from 165 processing tasks, any improvement in the Operationalization of processing style would most likely be associated with improvements in the Operationalization of cognitive struc- ture.3 That is to say, with more varied processing tasks, new capabilities to develop better measures of differentia- tion, discrimination, and integration are likely to follow.4 Dissertation Summary The present research developed a theory of informa- tion processing and cognitive structure which explained how social-structural factors influence exposure patterns to an information environment, exposure patterns to informa- tion environment influence processing style development, and how environmental exposure and processing style influ- ence an individual's cognitive structure. A structural equation model and three formal hypotheses were generated from the theory, and submitted to a variety of analyses. The results of the data analysis indicated that, while there were specification errors in the proposed model, the three hypotheses were supported. The dissertation then concluded with a discussion of the problems and limitations of the present research effort, and suggestions for the direction of future research. 166 FOOTNOTES 1See Joreskog (1974) and J6reskoq and Sorbom (1978) for a discussion of how the chi-square goodness of fit test can be interpreted to judge the extent to which modified models "fit" better than previous models. 2Further, the use of interactive software can greatly enhance the utility of a time series design, by providing the opportunity to: (1) be more flexible with respect to administration (of questionnaire) procedures, and (2) better control the effects of history and item presenta— tion bias. 3See Woelfel and Fink (1980) for a discussion on the close relationship between fundamental and derived measures. 4Additional research which will examine the relationship between ACT test scores, SAT test scores and cognitive structure was proposed here but not performed. Future research would examine these relationships. APPENDIX A The Pre-Test Instrument APPENDIX A The Pre-Test Instrument Dear Participant, please consider each of the following questions carefully and respond thoughtfully. (1) List the two most important news events of the past year. Try to select events about which most persons have a large amount of information and a firm opinion. (1) (2) (2) For the first news event listed above, free-associate and list as many words or phrases as possible which you feel are related to (or describe) this news event. Then, in a paragraph or two, state your opinion of the news event. 167 (3) (4) 168 For the second news event listed on page one, free- associate and list as many words or phrases as possible which you feel are related to (or describe) this news event. Then, in a paragraph or two, state your opinion of the news event. On the next page you will find a list of 21 emotions. Consider the entire list of emotions and pick out some which are alike in an important way. Write their line numbers in the left-hand box and, at the bottom of the box, write whatever it is that the emotions in that group have in common—-that is, why you put them to- gether. Then, in the right-hand box, write the line numbers of any emotions on the list that are clearly different from the first group in this respect. That is, include in the right-hand box emotions which do not possess the characteristic written at the bottom of the left-hand box. It is not necessary to include all emotions on the list in either box. If a particular emotion cannot be evaluated on this characteristic, omit it from both boxes. LINE NUMBERS OF ALL LINE NUMBERS OF EMOTIONS SIMILAR EMOTIONS (ON THIS WHICH DIFFER IN THIS CHARACTERISTIC) RESPECT FROM THE GROUP ON THE LEFT CHARACTERISTIC: 169 Here are the emotions you are tO consider: (1) ACTIVE (2) ANGER (3) ANXIETY (4) BAD (5) DEPRESSION (6) ENVY (7) EXCITEMENT (8) FEAR (9) GOOD (10) GUILT (11) HAPPINESS (12) HATE (l3) INDIFFERENCE (14) JEALOUSY (15) JOY (16) LOVE (17) PASSIVE (18) SADNESS (l9) SELFISH (20) STRONG (21) WEAK CHARACTERISTIC: On the following pages are more pairs Of boxes. Consider the entire list Of emotions again and pick out another group which are alike in an important way. Write their line numbers in the left-hand box, and at the bottom, write in the characteristic they have in common. Then, in the right- hand box, write the line numbers Of the emotions which clearly differ in this respect from the left hand group. Continue this process, using as many boxes as you need to make groups Of emotions which are similar in an important way. Any particular emotion may be included in more than one group if you wish. 170 Note: Nine of these pages appeared in the questionnaire CHARACTERISTIC: CHARACTERISTIC: CHARACTERISTIC: CHARACTERISTIC: 171 (5) Below you will find a list of words that are paired together. What we would like you to do is tell us how different each word is from the other word using a number. For example, if you think two words are very similar in meaning to each other, or are closely associated with each other, then you would report a small number. On the other hand, if you think the two words are very different in meaning from each other, or are hardly related at all, then you would report a very large number. REMEMBER, SMALL NUMBERS INDICATE SIMILARITY AND LARGE NUMBERS REFLECT DIFFERENCES. ZERO WOULD MEAN THAT THERE IS NO DIFFERENCE IN MEANING BETWEEN THE TWO WORDS. YOU MAY USE ANY NUMBER YOU WISH. TO help you make these judgments of similarity and difference, consider the following example. Think about the words NEW TECHNOLOGY and SCIENTIFIC RESEARCH OBVIOUSLY, these words are somewhat related, but they clear- ly do not mean the same thing, so lets assume the number 100 is a number which reflects the difference between these two words. That is to say, the difference between NEW TECH- NOLOGY and SCIENTIFIC RESEARCH is 100 units. Try to keep this simple example in mind when making your judgments about the other pairs of words. In review, the important things tO remember are: (a) small numbers indicate similarity, large numbers reflect differences, (b) you may use any number you wish, and (c) the difference between NEW TECHNOLOGY and SCIENTIFIC RESEARCH is 100 units. WHAT IS THE DIFFERENCE BETWEEN NEW TECHNOLOGY and MONEY WELL SPENT NEW TECHNOLOGY and INDUSTRY IN SPACE NEW TECHNOLOGY and COLONIES IN SPACE NEW TECHNOLOGY and SCIENTIFIC RESEARCH NEW TECHNOLOGY and EXPLORING SPACE NEW TECHNOLOGY and NATIONAL DEFENSE NEW TECHNOLOGY and NATIONAL PRESTIGE NEW TECHNOLOGY and YOU 172 WHAT IS THE DIFFERENCE BETWEEN MONEY WELL SPENT and MONEY WELL SPENT and MONEY WELL SPENT and MONEY WELL SPENT and MONEY WELL SPENT and MONEY WELL SPENT and MONEY WELL SPENT and INDUSTRY IN SPACE COLONIES IN SPACE SCIENTIFIC RESEARCH EXPLORING SPACE NATIONAL DEFENSE NATIONAL PRESTIGE YOU WHAT IS THE DIFFERENCE BETWEEN INDUSTRY IN SPACE and COLONIES IN SPACE INDUSTRY IN SPACE and SCIENTIFIC RESEARCH INDUSTRY IN SPACE and EXPLORING SPACE INDUSTRY IN SPACE and NATIONAL DEFENSE INDUSTRY IN SPACE and NATIONAL PRESTIGE INDUSTRY IN SPACE and YOU WHAT IS THE DIFFERENCE BETWEEN COLONIES IN SPACE and SCIENTIFIC RESEARCH COLONIES IN SPACE and EXPLORING SPACE COLONIES IN SPACE and NATIONAL DEFENSE COLONIES IN SPACE and NATIONAL PRESTIGE COLONIES IN SPACE and YOU WHAT IS THE DIFFERENCE BETWEEN SCIENTIFIC RESEARCH SCIENTIFIC RESEARCH SCIENTIFIC RESEARCH SCIENTIFIC RESEARCH and EXPLORING SPACE and NATIONAL DEFENSE and NATIONAL PRESTIGE and YOU WHAT IS THE DIFFERENCE BETWEEN EXPLORING SPACE and NATIONAL DEFENSE EXPLORING SPACE and NATIONAL PRESTIGE EXPLORING SPACE and YOU 173 WHAT IS THE DIFFERENCE BETWEEN NATIONAL DEFENSE and NATIONAL PRESTIGE NATIONAL DEFENSE and YOU NATIONAL PRESTIGE and YOU (6) This part Of the survey asks for some information about you. (a) (b) (C) (d) (e) (f) (g) (h) (i) (j) (k) (1) (m) (n) (o) (p) (q) (r) (s) (t) SEX: MALE FEMALE BIRTHDATE: MAJOR: GPA: CLASS: NUMBER OF PERSONS IN YOUR FAMILY: APPROXIMATE FAMILY INCOME: per year FATHER'S OCCUPATION: MOTHER'S OCCUPATION: YOUR OCCUPATION: Now PLANNED YOUR RACE: HOW MANY HOURS OF TV DO YOU WATCH PER WEEK: HOW MANY HOURS PER WEEK DO YOU SPEND LISTENING TO THE RADIO AND TO RECORDS: HOW MANY HOURS PER WEEK DO YOU SPEND READING A NEWSPAPER OR NEWS MAGAZINE: DO YOU WATCH THE NATIONAL NEWS ON TV: YES NO HOW MANY CLOSE FRIENDS DO YOU SPEND TIME WITH ON A REGULAR BASIS ABOUT HOW MANY HOURS A WEEK DO YOU SPEND CONVERSING WITH YOUR CLOSE FRIENDS: ABOUT HOW MANY HOURS A WEEK DO YOU SPEND STUDYING: If 10 represents the amount of understanding the average person has about emotions, what number best reflects your understanding Of emotions (Remember 10 is average). If 10 represents the amount of information the average person has about scientific development in space, what number best reflects your under- standing Of this topic (Remember 10 is average). 174 (u) If 10 represents the amount of diversity in the average person's life, how diverse have your experiences been thus far (Remember 10 is average). THANK YOU FOR TAKING THE TIME TO COMPLETE THIS SURVEY. IF YOU HAVE ANY COMMENTS ABOUT THE TASKS YOU HAVE JUST COM- PLETED, PLEASE FEEL FREE TO WRITE THEM DOWN BELOW. WE ARE PARTICULARLY INTERESTED IN THOSE COMMENTS WHICH DETAIL DIFFICULT OR TROUBLESOME SECTIONS, AND WHICH SUGGEST WAYS TO IMPROVE SURVEY TECHNIQUES. THANKS AGAIN FOR YOUR HELP. APPENDIX B Concept Categories and Terms Extracted From the Pre-Test Instrument which were Concerned with the United States - Iranian Hostage Situation (N=29) APPENDIX B Concept Categories and Terms Extracted From the Pre—Test Instrument which were Concerned with the United States - Iranian Hostage Situation (N=29) Concept Category Iran United States Hostages Terms * Iran Iranians Iranian Crisis Iran Crisis Iranian Situation Tehran This Issue This Situation Crisis in Iran Iran Situation Crisis Tehranian Situation Their Government This Event This Subject Problems in Iran The Situation in Iran The Issue The Problem Iranian Government * US Americans American Government The Country America Lack Of Patriotism Administrations USA US Image Renewed American Patriotism Cultural Identification * Hostages Embassy Hostages in Iran American Hostages Captured the Embassy Hostage Situation in Iran The Hostage Situation 175 Frequency Of Occurrence (44) 14 HPJPJHravahwakawrdhdNFJNJNLAUJW (25) |-’ FJHFHFJHP#FJH+4U1H (58) N |._a NNNNb-b 176 Frequency Of Concept Category Terms Occurrence Hostages (cont'd.) Our People Day Of Hostage Captivity American Personnel Captured the Embassy in Tehran Seized the US Embassy Seized Seizure of Our Embassy in Iran Overthrow Of the Embassy 50 American Hostages 60 American Hostages Take the Hostages Captives All Of Its [Embassy] Officials Hostage Crisis Iranian Hostage Crisis Hold 53 People Hostage Capture Of Americans in Iran Iran Hostages Prisoners of Iran Held Captive Holding the Hostages in Tehran HPJPJHPHPJHPHPJHPJPJHrahawrahuahaw Khomeini * (l4) Khomeini 6 Ayatollah Khomeini 4 Ayatollah 4 Carter * (19) Carter 17 President Carter 2 Shah * (15) Shah 12 Shah Of Iran 3 Militants * (23) Militants 8 Students 7 Small Band Of Terrorists 1 Terrorism 1 Terrorists l Captors l Fanatics 1 Mobs l 177 Frequency Of Concept Catggory Term§_ Occurrence Militants (cont'd.) Rebels l Revolutionary Council 1 Economic Sanctions * (15) Blackmail Oil Economic Sanctions Embargo Punish Them Boycott Economic Business Related Sanctions Iranian Oil American Concern for Oil Pahawrdkawrahawcuu: Military Inter- vention * (43) Rescue Attempt Rescue Rescue Mission Shouldn't Have Failed War Military Draft HelicOpters Invasion Drop an Atom Bomb on Iran Endangering Lives Ill Fated Rescue Attempt Should Have Tried to Move in Right Away Plan to Rescue Prisoners Rescue Mission Failed Misson Failed Surprise Rescue Mission Rescue Failed Failure 8 Men Lost Their Lives Daring Technical Mishap in Machinery It was Necessary Not Carefully Planned Kill Them Killing GO to War HrdhaHrdhuthHPahuahuhnd HFJPHHBJNEQAJNFQUJb 178 Frequency of Concept Category Terms Occurrence Military Inter- vention (cont'd.) The Navy Major Power Revolution [Men] Can be Taken Anytime Nice Safe Little World PJHPHF‘H Media Coverage * (31) Media News News Coverage Coverage Media Event Mediazed Media Coverage News Event Read or Hear Inform Covered Well Newspaper Publicity Facts Display Walter Cronkite Hush it Up Not so Much B.S. Should Have Been Played Down Over Carried Away Losing Importance raPJHPHPJHrAPJHPJPawrahawhahawrooaq Diplomacy * (66) Should Have Taken a Firm Stand Situation Calls for Immediacy Release Let Our People GO Issues Should Have Been Taken Care of Some Time Ago Help the Prisoners Something has tO be Done We May Have Been at Fault Internal Affairs Lack Communication Don't Know What I'd DO Games Countries Play )APJHPahuaPJH hwahawra 179 Frequency of Concept Category Terms Occurrence Diplomacy (cont'd.) TOO Many Things to Con- sider 1 Hard to State the Actual Events 1 In a Position it [Us] Has Never Had TO Face Before They [Iranians] Should do Their Share Two Children Fighting over Dolls Something Must be Done A Test Precautions Measures NO Answers or Solutions to the Problem Not Sure Of Best Solution We Seem in nO Hurry tO Get Our Citizens Back 1 NO Control Over Situation by Hrahua H PJH FJH Either Country 1 142 (etc.) Days 1 SO Many Months 1 Fed Up with the Whole Crisis Deal 1 Carried on tOO Long 1 Gone too Far 1 Sick and Tired Of Hearing How Many Days Have Passed 1 They Have to be Held Account- able for What They've Done 1 Passive Point Of View 1 Protest 1 Don't Know a Lot About it 1 Don't Really Have a Firm Opinion 1 Many Wrongs in Foreign Policy 1 Should Come to an End 1 We Need to Take Action 1 Try to StOp it Should be Remedied as Soon as Possible 1 Have to Pull Together Behind Leaders 1 Supports Actions 1 Wish They were Free 1 Gone on tOO Long 1 Long Term 1 180 Frequency Of Concept Category Terms Occurrence Diplomacy (cont'd.) Drawn Out Long NO Progress Government Moving Slow Power Hungry Political Unrest NO Reason Being in Power Hamper US Relations with Russia An Issue That Affects More Than Iran and the Hostages Set Off a Fuse That the World Better Put Out Before its the Cause of a Big Blow Up The Front of US Foreign Policy Not Doing the Job Don't Like Actions UN Judicial Body Criticized by Many Mishandled Mishandling Careless Inept rd HFHPJHFHFJH g... g... Hrahawrdhuahaw Irrational * (26) Stupid Wrong Irrational NO Valid Reason Unreasonable Mistake Sad When You Think About it Foolish Uncouth Foolish Reaction Don't Feel it Was Right Not Justified Justified Unjust Unfair Violation Of International Law Unlawful Outside of Law At Fault The Way They Went About it Wrong Approach HrahawrahawrahudhawruhaHrahuahamcu 181 Frequency Of Concept Category Terms Occurrence Irrational (cont'd.) Thought These Kind of Things Don't Happen 1 Two Wrongs Don't Justify a Right H v Weakness * ( Weakness Feel Helpless Helplessness Feeling of Helplessness Vulnerability Seems to Show Weakness in Government Felt Terrible Weakness US Looks Weak Humbled Hrahwdram HP‘F‘H Hostile * (10) Anger Hostility Intense Feelings On Both Sides Upset Outraged Outrage Humiliated Hatred Disgusted by the Situation HPJPJNFHPHHPJH Religion * ( Religion Religious Freak Tradition l-‘I—‘wU'l Other * (56) Empathetic Toward Captives Sad TOO Bad Interdependence Brainwashing Bigotry Spies Tried as Spies Olympics NFHP‘NFHPHONJH Concept Category Other (cont'd.) 182 Terms Vacillating Fear Disgrace Rage and Terror Support My Fiance Loud Liberal Refugees Foreigners Out Of Their Hair Other Countries Canada USSR Bani-Sadar Ghotzbadeh Families Empathetic Physical and Mental Stress Russia Afghanistan Despair Horrifying Shameful Shocking Shocked by the Event Confusion Torment Suffering Devastating Tragedy Hits the Heart Pity Empathy Dismay Deport Frequency Of Occurrence PJHF‘FJHFHBJHPHFJHFHFJHrdFJHFJFJbPHF‘HPHFJHBQF‘HFJNJHFJFJHto APPENDIX C The Final Instrument APPENDIX C The Final Instrument THE IPACS PROJECT A STUDY OF INFORMATION PROCESSING AND COGNITIVE STRUCTURE STUDENT NUMBER: DATE: TIME STARTED: TIME ENDED: 183 184 MESSAGE TO THE PARTICIPANT: THANK YOU FOR VOLUNTEERING TO PARTICIPATE IN THIS STUDY. ON THE FOLLOWING PAGES YOU WILL FIND A SERIES OF QUESTIONS WHICH ASK YOU TO TELL US ABOUT YOUR BACKGROUND, AND HOW YOU PERCEIVE CERTAIN THINGS. PLEASE READ ALL IN- STRUCTIONS BEFORE YOU BEGIN A NEW SECTION, AND PLEASE CON- SIDER EACH QUESTION CAREFULLY. WHILE IT IS TRUE THAT THERE IS NO SUCH THING AS A "RIGHT" OR "WRONG" ANSWER WHEN IT COMES TO PUBLIC OPINION, THE MORE ACCURATELY YOU TRY TO ANSWER, THE MORE LIKELY THE RESULT OF YOUR EFFORTS WILL BE USEFUL. IF YOU HAVE ANY QUESTIONS ABOUT HOW TO FILL OUT ANY OF THE ITEMS ON THE QUESTIONNAIRE, CONSULT THE RESEARCH ASSISTANTS; THEY WILL BE GLAD TO HELP. PLEASE ANSWER ALL QUESTIONS; DO NOT LEAVE ANY ITEMS BLANK. THANK YOU SECTION I INSTRUCTIONS: THE ITEMS IN THIS SECTION ARE DESIGNED TO GATHER INFORMATION ABOUT YOUR BACKGROUND. PLEASE TRY TO ANSWER ALL ITEMS AS ACCURATELY AS POSSIBLE. (l) SEX: MALE FEMALE (2) BIRTHDATE: / / month day year (3) BIRTHPLACE: / / city state country (4) CITY WHERE YOU SPENT MOST OF YOUR LIFE: / / city state country (5) CITY WHERE YOU WENT TO HIGH SCHOOL: / / city state country (6) (7) (8) (9) (10) (ll) (12) (l3) (14) 185 WHEN YOU WERE GROWING UP, WHAT WAS THE AVERAGE YEARLY INCOME YOUR FAMILY LIVED ON? $ / YR (estimate in 1980 dollars) FATHER'S OCCUPATION? (WHAT HE DOES/DID, NOT WHERE HE WORKS) MOTHER'S OCCUPATION? (WHAT SHE DOES/DID, NOT WHERE SHE WORKS) WHAT IS THE HIGHEST GRADE OR YEAR IN SCHOOL THAT YOUR FATHER COMPLETED? WHAT IS THE HIGHEST GRADE OR YEAR IN SCHOOL THAT YOUR MOTHER COMPLETED? WHAT IS YOUR RELIGION? DURING THE LAST YEAR, HOW MANY TIMES HAVE YOU GONE TO A PLACE OF WORSHIP? DURING THE LAST YEAR, HOW MANY RELIGIOUS HOLIDAYS HAVE YOU OBSERVED? WHAT IS YOUR RACE? (15) (l6) (17) (18) (19) (20) (21) (22) (23) 186 DURING THE LAST YEAR, HOW MANY TIMES DID YOU OBSERVE OR PARTICIPATE IN EVENTS RELATED TO YOUR ETHNIC BACKGROUND? INCLUDING YOUR PARENTS, HOW MANY PEOPLE ARE IN YOUR FAMILY? INCLUDING YOURSELF, HOW MANY CHILDREN ARE IN YOUR FAMILY? CONSIDERING THE OLDEST CHILD IN YOUR FAMILY AS NUMBER ONE, AND THE SECOND OLDEST CHILD IN YOUR FAMILY AS NUMBER TWO, ETC., WHAT NUMBER CHILD ARE YOU? ARE YOU A TWIN OR A TRIPLET? TWIN TRIPLET NEITHER A TWIN NOR A TRIPLET HOW MANY BROTHERS DO YOU HAVE? HOW MANY OF YOUR BROTHERS ARE STEP-BROTHERS? HOW MANY SISTERS DO YOU HAVE? HOW MANY OF YOUR SISTERS ARE STEP-SISTERS? (24) (25) (26) (27) (28) (29) (30) (31) 187 HOW MANY OF YOUR BROTHERS AND/OR STEP-BROTHERS ARE OLDER THAN YOU? HOW MANY OF YOUR SISTERS AND/OR STEP-SISTERS ARE OLDER THAN YOU? HOW MANY OF YOUR BROTHERS AND/OR STEP-BROTHERS ARE YOUNGER THAN YOU? HOW MANY OF YOUR SISTERS AND/OR STEP-SISTERS ARE YOUNGER THAN YOU? IN ADDITION TO YOUR PARENTS, YOURSELF, AND YOUR BROTHERS AND SISTERS, HOW MANY OTHER RELATIVES LIVED WITH YOU WHEN YOU WERE GROWING UP? WITHIN YOUR IMMEDIATE FAMILY, HOW MANY PERSONS HAVE ANY KIND OF A HANDICAP THAT REQUIRES SPECIAL ATTEN- TION OR CARE? WHEN YOU WERE GRADUATED FROM HIGH SCHOOL, APPROXIMATE- LY WHAT WAS YOUR GRADE POINT AVERAGE? (PLEASE EXPRESS YOUR GRADE POINT AVERAGE USING THE MSU GRADING SYSTEM.) IF THE PERSON WITH THE HIGHEST GRADE POINT AVERAGE RANKED IN THE 99EH PERCENTILE, AND THE PERSON WITH THE LOWEST GRADE POINT RANKED IN THE 0 EH PERCENTILE, WHAT WAS YOUR PERCENTILE RANK IN HIGH SCHOOL? %ti1e 188 (32) WHAT IS YOUR CURRENT MSU GRADE POINT AVERAGE? (33) WHAT IS YOUR CURRENT CLASS STANDING? FRESHMAN SOPHOMORE JUNIOR SENIOR (35) INDICATE HOW MANY COLLEGE COURSES YOU HAVE TAKEN IN THE AREAS LISTED BELOW:* AREA NUMBER OF COURSES TAKEN COURSES WHERE YOU LEARN ABOUT AND ACQUIRE LANGUAGE SKILLS COURSES WHERE YOU LEARN ABOUT AND ACQUIRE MATHEMATICAL SKILLS COURSES WHERE YOU LEARN ABOUT AND ACQUIRE READING SKILLS COURSES WHERE YOU LEARN ABOUT AND ACQUIRE SOCIAL SCIENCE SKILLS COURSES WHERE YOU LEARN ABOUT AND ACQUIRE PHYSICAL SCIENCE SKILLS * A COURSE MAY BE USED IN MORE THAN ONE CATEGORY 189 SECTION II INSTRUCTIONS: THE ITEMS IN THIS SECTION ARE DESIGNED TO GATHER INFORMATION ABOUT YOUR FAMILY'S LIFESTYLE, YOUR LIFE- STYLE WHEN YOU WERE GROWING UP, AND YOUR CURRENT LIFESTYLE. EACH QUESTION WILL FIRST DESCRIBE A PARTICULAR LIFESTYLE, AND THEN ASK YOU TO ESTIMATE YOUR EXPERIENCES AS COMPARED TO THE AVERAGE PERSON'S EXPERIENCE. FOR OUR PURPOSES HERE, THE NUMBER 100 WILL DESCRIBE THE AVERAGE PERSON'S EXPERIENCE WITH THE PARTICULAR LIFESTYLE, AND THE NUMBER 0 WILL MEAN THAT THE PARTICULAR LIFESTYLE WAS NOT EXPERIENCED AT ALL. YOU MAY USE ANY NON-NEGATIVE NUMBER YOU WISH TO DESCRIBE YOUR OWN EXPERIENCES. ALL WE ASK IS THAT YOU MAKE YOUR ESTIMATES CAREFULLY. REMEMBER, O MEANS NOT AT ALL, AND 100 IS THE AVERAGE PERSON'S EXPERIENCE. IF YOUR EXPERIENCES ARE TWICE THAT OF THE AVERAGE PERSON'S, YOU WOULD WRITE 292_(because 2 X 100 = 200). ON THE OTHER HAND, IF YOUR EXPERIENCES ARE ONLY HALF THAT OF THE AVERAGE PERSON'S, THEN YOU WOULD WRITE 29 (because 1/2 Of 100 = 50). AGAIN, YOU MAY USE ANY NUMBER YOU WISH. PLEASE CONSIDER EACH QUESTION CAREFULLY, AND ANSWER EACH ITEM AS ACCURATELY AS POSSIBLE. (36) AN ACTIVE LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL DOES ENERGETIC, ANIMATED, AND LIVELY THINGS. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW ACTIVE WERE YOU WHEN YOU WERE GROWING UP? HOW ACTIVE WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW ACTIVE ARE YOU NOW? 190 (37) A SPATIALLY ORGANIZED LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL'S LIVING SPACE IS COORDINATED, TIDY, AND WELL ARRANGED. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW SPATIALLY ORGANIZED WERE YOU WHEN YOU WERE GROWING UP? HOW SPATIALLY ORGANIZED WAS YOUR FAMILY? HOW SPATIALLY ORGANIZED ARE YOU NOW? (38) A TEMPORALLY ORGANIZED LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL IS PUNCTUAL, METHODICAL, AND RHYTHMIC IN HIS/HER BEHAVIORS. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW TEMPORALLY ORGANIZED WERE YOU WHEN YOU WERE GROWING UP? HOW TEMPORALLY ORGANIZED WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW TEMPORALLY ORGANIZED ARE YOU NOW? (39) A STIMULATING LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL CAN LEARN AND SATISFY HIS/HER CURIOSITY. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW STIMULATED WERE YOU WHEN YOU WERE GROWING UP? HOW STIMULATED WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW STIMULATED ARE YOU NOW 191 (40) A SOCIALLY RESPONSIVE LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL'S BEHAVIORS ARE ACKNOWLEDGED AND RESPONDED TO BY OTHERS. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW SOCIALLY RESPONSIVE WAS YOUR LIFESTYLE WHEN YOU WERE GROWING UP? HOW SOCIALLY RESPONSIVE WAS YOUR FAMILY"S LIFESTYLE WHEN YOU WERE GROWING UP? HOW SOCIALLY RESPONSIVE IS YOUR LIFESTYLE NOW? (41) A DIVERSE LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL DOES (42) THINGS WHICH ARE VARIED, AND UNUSUAL. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW DIVERSE WAS YOUR LIFESTYLE WHEN YOU WERE GROWING UP? HOW DIVERSE WAS YOUR FAMILY'S LIFESTYLE WHEN YOU WERE GROWING UP? HOW DIVERSE IS YOUR LIFESTYLE NOW? AN INDEPENDENT LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL IS SELF-DIRECTED, AUTONOMOUS, AND FREE TO DO WHATEVER HE/SHE LIKES. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW INDEPENDENT WERE YOU WHEN YOU WERE GROWING UP? HOW INDEPENDENT WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW INDEPENDENT ARE YOU NOW? 192 (43) A CONSTRAINED LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL'S BEHAVIORS ARE CONFINED, RESTRICTED, AND LIMITED. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW CONSTRAINED WERE YOU WHEN YOU WERE GROWING UP? HOW CONSTRAINED WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW CONSTRAINED ARE YOU NOW? (44) A NURTURED LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL'S EMOTIONS, FEELINGS, AND WELL-BEING ARE CARED FOR BY OTHERS. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . HOW NURTURED WERE YOU WHEN YOU WERE GROWING UP? HOW NURTURED WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW NURTURED ARE YOU NOW? (45) A COHESIVE LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL ESTABLISHES CLOSE RELATIONSHIPS AMONG FRIENDS AND FAMILY. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW COHESIVE WAS YOUR LIFESTYLE WHEN YOU WERE GROWING UP? HOW COHESIVE WAS YOUR FAMILY'S LIFESTYLE WHEN YOU WERE GROWING UP? HOW COHESIVE IS YOUR LIFESTYLE NOW? 193 (46) AN INTELLECTUAL LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL IS SURROUNDED BY INTELLIGENT, KNOWLEDGEABLE, AND COMPETENT PEOPLE. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW INTELLECTUAL WAS YOUR LIFE- STYLE WHEN YOU WERE GROWING UP? HOW INTELLECTUAL WAS YOUR FAMILY'S LIFESTYLE WHEN YOU WERE GROWING UP? HOW INTELLECTUAL IS YOUR LIFESTYLE NOW? (47) A GOOD LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL ENJOYS (48) LIFE, IS SATISFIED WITH HOW THINGS ARE, AND IS GENERALLY HAPPY. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW GOOD WAS YOUR LIFESTYLE WHEN YOU WERE GROWING UP? HOW GOOD WAS YOUR FAMILY'S LIFE- STYLE WHEN YOU WERE GROWING UP? HOW GOOD IS YOUR LIFESTYLE NOW? A STRONG LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL CAN WITHSTAND ADVERSITY, STRESS, AND BAD LUCK. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW STRONG WERE YOU WHEN YOU WERE GROWING UP? HOW STRONG WAS YOUR FAMILY WHEN YOU WERE GROWING UP? HOW STRONG ARE YOU NOW? (49) IF ZERO MEANS NOT AT ALL, 194 AVERAGE PERSON'S EXPERIENCE . . . (50) IF 100 DESCRIBES THE AVERAGE PERSON'S SKILLS, WHAT HOW SUPPORTIVE WERE YOUR PARENTS OF YOUR AMBITIONS AND ASPIRATIONS WHEN YOU WERE GROWING UP? AND 100 DESCRIBES THE HOW MUCH ENCOURAGEMENT DID YOUR PARENTS PROVIDE FOR YOUR IDEAS WHEN YOU WERE GROWING UP? NUMBER BEST DESCRIBES YOUR . . . (51) HOW (52) TEN YEARS AFTER YOU ARE GRADUATED FROM COLLEGE, WHAT LANGUAGE SKILLS MATHEMATICAL SKILLS READING SKILLS SOCIAL SCIENCE SKILLS PHYSICAL SCIENCE SKILLS COMMUNICATION SKILLS MANY YEARS HAVE YOU BEEN . . . PLAYING A MUSICAL INSTRUMENT DANCING PAINTING SCULPTURING WRITING PLAYING SPORTS DO YOU EXPECT YOUR ANNUAL INCOME TO BE? $ / YR. (estimate in dollars) 195 (53) WHAT OCCUPATION DO YOU THINK YOU WILL BE WORKING AT TEN YEARS AFTER YOU ARE GRADUATED FROM COLLEGE? SECTION III INSTRUCTIONS: THE FOLLOWING SET OF ITEMS ARE DESIGNED TO GATHER INFORMATION ABOUT YOUR COMMUNICATION HABITS AND MEDIA EXPERIENCES. IN PARTICULAR, WE ARE INTERESTED IN FINDING OUT ABOUT YOUR INTERPERSONAL COMMUNICATION NETWORKS, AND YOUR MASS MEDIA EXPERIENCES. BY INTERPERSONAL COMMUNICATION NETWORKS, WE MEAN THE FACE-TO-FACE INTERACTIONS THAT YOU HAVE WITH OTHER PERSONS. BY MASS MEDIA EXPERIENCES, WE MEAN THE EXPERIENCES YOU HAVE HAD WITH TELEVISION, RADIO, PERIODICALS, THE CINEMA OR ANY OTHER MEDIA WHICH DISTRIBUTE MESSAGES ON A MASS SCALE. EACH QUESTION WILL FIRST DESCRIBE A PARTICULAR INFORMATION ENVIRONMENT, AND THEN ASK YOU TO ESTIMATE YOUR EXPERIENCES AS COMPARED TO THE AVERAGE PERSON'S EXPERIENCE. FOR OUR PURPOSES HERE, THE NUMBER 100 WILL DESCRIBE THE AVERAGE PERSON'S EXPERIENCE WITH THE PARTICULAR INFORMATION ENVIRONMENT, AND THE NUMBER ZERO WILL MEAN THAT THE PARTICU- LAR INFORMATION ENVIRONMENT WAS HOT EXPERIENCED AT ALL. YOU MAY USE ANY NON-NEGATIVE NUMBER YOU WISH TO DESCRIBE YOUR EXPERIENCES. ALL WE ASK IS THAT YOU MAKE YOUR ESTI- MATES CAREFULLY. REMEMBER, ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE. IF YOUR EXPERIENCES ARE TWICE THAT OF THE AVERAGE PERSON'S, YOU WOULD WRITE 200 (because 2 X 100 = 200). ON THE OTHER HAND, IF YOUR EXPERIENCES ARE ONLY HALF THAT OF THE AVERAGE PERSON'S, YOU WOULD WRITE £9 (because 1/2 of 100 = 50). AGAIN, YOU MAY USE ANY NUMBER YOU WISH. PLEASE CONSIDER EACH QUESTION CAREFULLY, AND ANSWER EACH ITEM AS ACCURATELY AS POSSIBLE. 196 (54) AN ACTIVE INFORMATION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PROCESSES A LOT OF INFORMATION. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW ACTIVE WERE YOUR INTERPERSONAL COMMUNICATION NETWORKS WHEN YOU WERE GROWING UP? HOW ACTIVE ARE YOUR INTERPERSONAL COMMUNICATION NETWORKS NOW? HOW ACTIVE WERE YOUR MEDIA EXPERI- ENCES WHEN YOU WERE GROWING UP? HOW ACTIVE ARE YOUR MEDIA EXPERI- ENCES NOW? (55) A STIMULATING INFORMATION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PROCESSES A LOT OF INFORMATION WHICH TEACHES HIM/HER NEW THINGS AND STIMULATES HIS/HER CURIOSITY AND IMAGINATION. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . HOW STIMULATING WERE YOUR INTER- PERSONAL COMMUNICATION NETWORKS WHEN YOU WERE GROWING UP? HOW STIMULATING ARE YOUR INTER- PERSONAL COMMUNICATION NETWORKS NOW? HOW STIMULATING WERE YOUR MEDIA EXPERIENCES WHEN YOU WERE GROW- ING UP? HOW STIMULATING ARE YOUR MEDIA EXPERIENCES NOW? 197 (56) A SOCIALLY RESPONSIVE INFORMATION ENVIRONMENT IS ONE (57) IN WHICH AN INDIVIDUAL PROCESSES A LOT OF INFORMATION WHICH ACKNOWLEDGES OR RESPONDS TO HIS/HER BEHAVIORS. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW SOCIALLY RESPONSIVE WERE YOUR INTERPERSONAL COMMUNICATION NET- WORKS WHEN YOU WERE GROWING UP? HOW SOCIALLY RESPONSIVE ARE YOUR INTERPERSONAL COMMUNICATION NET- WORKS NOW? A DIVERSE INFORMATION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PROCESSES A LOT OF INFORMATION ABOUT A VARIETY OF TOPICS, AND WHICH INFORMS HIM/HER ABOUT DIFFERENT THINGS ABOUT THE TOPIC. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW DIVERSE WERE YOUR INTERPERSONAL COMMUNICATION NETWORKS WHEN YOU WERE GROWING UP? HOW DIVERSE ARE YOUR INTERPERSONAL COMMUNICATION NETWORKS NOW? HOW DIVERSE WERE YOUR MEDIA EX- PERIENCES WHEN YOU WERE GROWING UP? HOW DIVERSE ARE YOUR MEDIA EX- PERIENCES NOW? (58) (59) (60) (61) (62) 198 A NURTURING INFORMATION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PROCESSES A LOT OF INFORMATION WHICH IS DIRECTED AT HIS/HER FEELINGS, EMOTIONS, AND WELL-BEING. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW NURTURING WERE YOUR INTER- PERSONAL COMMUNICATION NETWORKS WHEN YOU WERE GROWING UP? HOW NURTURING ARE YOUR INTER- PERSONAL COMMUNICATION NETWORKS NOW? A COHESIVE INFORMATION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL INTERACTS WITH A SET OF PEOPLE WHO ARE HIGHLY INTEGRATED AND WHO EXCHANGE A LOT OF INFORMA- TION WITH EACH OTHER. IF ZERO MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE HOW COHESIVE WERE YOUR INTER- PERSONAL COMMUNICATION NETWORKS WHEN YOU WERE GROWING UP? HOW COHESIVE ARE YOUR INTER- PERSONAL COMMUNICATION NETWORKS NOW? WHEN YOU WERE GROWING UP, HOW MANY CLOSE FRIENDS WOULD YOU SAY YOU HAD? ON THE AVERAGE, ABOUT HOW MANY HOURS PER WEEK DID YOU SPEND TALKING WITH THESE FRIENDS? HRS/WK HOW MANY CLOSE FRIENDS WOULD YOU SAY YOU HAD NOW? 199 (63) ON THE AVERAGE, ABOUT HOW MANY HOURS PER WEEK DO YOU SPEND TALKING WITH THESE FRIENDS? HRS/WK (64) ON THE AVERAGE, ABOUT HOW MANY HOURS PER MONTH HAVE YOU SPENT TALKING ABOUT . . . THE HOSTAGES IN IRAN HRS/MONTH THE IRAN/IRAQ WAR HRS/MONTH (65) INDICATE HOW MANY HOURS PER WEEK YOU SPEND READING EACH OF THE FOLLOWING NEWSPAPERS . . . THE STATE NEWS HRS/WK THE LANSING STATE JOURNAL HRS/WK THE DETROIT PRESS HRS/WK THE NEW YORK TIMES HRS/WK THE WALL STREET JOURNAL HRS/WK OTHER (PLEASE SPECIFY) HRS/WK OTHER (PLEASE SPECIFY) HRS/WK (66) ON THE AVERAGE, ABOUT HOW MANY HOURS PER WEEK DO YOU SPEND . . . LISTENING TO THE RADIO HRS/WK PLAYING RECORDS OR TAPES HRS/WK READING PERIODICALS (MAGAZINES) HRS/WK (67) (68) INDICATE HOW 200 TV NETWORK NEWS TV LOCAL NEWS TV MORNING NEWS PROGRAMS TV LATE NIGHT NEWS IF ZERO MEANS NOT INFORMED AT ALL, HOW INFORMED THE AVERAGE PERSON IS, HOW INFORMED WOULD YOU SAY YOU ARE ABOUT . . . THE UNITED STATES IRAN AYATOLLAH KHOMEINI PRESIDENT CARTER DIPLOMACY MILITARY INTERVENTION ECONOMIC SANCTION THE SHAH THE AMERICAN HOSTAGES IN IRAN THE STUDENT MILITANTS IN IRAN IRRATIONALITY HOSTILITY FREEDOM WEAKNESS MULTINATIONAL CORPORATIONS CAMP DAVID ACCORDS INTERNATIONAL TRADE OPEC THE IRAN/IRAQ WAR YOURSELF MANY HOURS PER WEEK YOU SPEND WATCHING HRS/WK HRS/WK HRS/WK HRS/WK AND 100 DESCRIBES 201 SECTION IV INSTRUCTIONS: ON THE FOLLOWING PAGES YOU WILL FIND A LIST OF WORDS THAT ARE PAIRED TOGETHER. WHAT WE WOULD LIKE YOU TO DO IS TELL US HOW DIFFERENT EACH WORD IS FROM THE OTHER WORDS USING A NUMBER. IF YOU THINK TWO WORDS ARE VERY SIMILAR TO EACH OTHER IN MEANING, OR ARE CLOSELY ASSOCIATED WITH EACH OTHER, THEN YOU WOULD REPORT A SMALL NUMBER. ON THE OTHER HAND, IF YOU THINK THAT THE TWO WORDS ARE VERY DIFFERENT FROM EACH OTHER IN MEANING, OR ARE HARDLY RELATED TO EACH OTHER AT ALL, THEN YOU WOULD REPORT A LARGE NUMBER. REMEMBER, SMALL NUMBERS INDICATE SIMILARITY, LARGE NUMBERS INDICATE DIFFERENCES, AND ZERO MEANS THAT THERE IS NO DIFFERENCE BETWEEN THE TWO WORDS AT ALL TO HELP YOU MAKE THESE JUDGMENTS OF SIMILARITY AND DIFFERENCE, CONSIDER THE FOLLOWING EXAMPLE. THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100. OBVIOUSLY, THESE TWO WORDS ARE SOMEWHAT RELATED, BUT THEY CLEARLY DO NOT MEAN THE SAME THING. HENCE, THE NUMBER 100 WAS ASSIGNED TO INDICATE THE DIFFERENCE BETWEEN THESE TWO WORDS. TRY TO KEEP THIS SIMPLE EXAMPLE IN MIND WHEN YOU CONSIDER EACH OF THE FOLLOWING PAIRS OF WORDS. YOU MAY USE ANY NON- NEGATIVE NUMBER YOU WISH TO DESCRIBE THE DIFFERENCES BETWEEN THE WORDS, THE ONLY THING WE ASK IS THAT YOU TRY TO BE AS ACCURATE AS YOU CAN. USE EXTREMELY LARGE NUMBERS ONLY WHEN THERE IS A VERY LARGE DIFFERENCE BETWEEN THE TWO WORDS, AND USE EXTREMELY SMALL NUMBERS ONLY WHEN THERE IS A VERY GREAT SIMILARITY BETWEEN THE TWO WORDS. AGAIN, ZERO MEANS THAT THERE IS H9 DIFFERENCE WHATSOEVER BETWEEN THE TWO WORDS. REMEMBER, SMALL NUMBERS INDICATE SIMILARITY, LARGE NUMBERS INDICATE DIFFERENCES, ZERO MEANS THAT THERE IS NO DIFFERENCE BETWEEN THE TWO WORDS, AND THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 19g NOW TURN THE PAGE AND MAKE YOUR JUDGMENTS: IF YOU HAVE ANY DIFFICULTY ASK THE EXPERIMENTAL ASSISTANTS FOR HELP. 202 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . THE UNITED STATES AND IRAN THE UNITED STATES AND KHOMEINI THE UNITED STATES AND CARTER THE UNITED STATES AND DIPLOMACY THE UNITED STATES AND MILITARY INTERVENTION THE UNITED STATES AND ECONOMIC SANCTION THE UNITED STATES AND THE SHAH THE UNITED STATES AND THE HOSTAGES IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . THE UNITED STATES AND THE MILITANTS THE UNITED STATES AND IRRATIONALITY THE UNITED STATES AND WEAKNESS THE UNITED STATES AND HOSTILITY THE UNITED STATES AND FREEDOM THE UNITED STATES AND HH (i.e., yourself) IRAN AND KHOMEINI IRAN AND CARTER 203 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 IRAN AND IRAN AND IRAN AND IRAN AND IRAN AND IRAN AND IRAN AND IRAN AND HOW DIFFERENT IRAN AND IRAN AND IRAN AND KHOMEINI KHOMEINI KHOMEINI KHOMEINI KHOMEIHI THE SHAH HOW DIFFERENT ARE . . . DIPLOMACY MILITARY INTERVENTION ECONOMIC SANCTION THE HOSTAGES THE MILITANTS IRRATIONALITY WEAKNESS IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 ARE . . . HOSTILITY FREEDQH £123. AND CARTER AND DIPLOMACY AND MILITARY INTERVENTION AND ECONOMIC SANCTION AND THE SHAH 204 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . KHOMEINI AND THE HOSTAGES KHOMEINI AND THE MILITANTS KHOMEINI AND IRRATIONALITY KHOMEINI AND WEAKNESS KHOMEINI AND HOSTILITY KHOMEINI AND FREEDOM KHOMEINI AND HE CARTER AND DIPLOMACY IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . CARTER AND MILITARY INTERVENTION CARTER AND ECONOMIC SANCTION CARTER AND THE SHAH CARTER AND THE HOSTAGES CARTER AND THE MILITANTS CARTER AND IRRATIONALITY CARTER AND WEAKNESS CARTER AND HOSTILITY 205 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . CARTER AND FREEDOM CARTER AND HE DIPLOMACY AND MILITARY INTERVENTION DIPLOMACY AND ECONOMIC SANCTION DIPLOMACY AND THE SHAH DIPLOMACY AND THE HOSTAGES DIPLOMACY AND THE MILITANTS DIPLOMACY AND IRRATIONALITY IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . DIPLOMACY AND WEAKNESS DIPLOMACY AND HOSTILITY DIPLOMACY AND FREEDOM DIPLOMACY AND HE MILITARY INTERVENTION AND ECONOMIC SANCTION MILITARY INTERVENTION AND THE SHAH MILITARY INTERVENTION AND THE HOSTAGES MILITARY INTERVENTION AND THE MILITANTS 206 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . MILITARY INTERVENTION AND IRRATIONALITY MILITARY INTERVENTION AND WEAKNESS MILITARY INTERVENTION AND HOSTILITY MILITARY INTERVENTION AND FREEDOM MILITARY INTERVENTION AND ME ECONOMIC SANCTION AND THE SHAH ECONOMIC SANCTION AND THE HOSTAGES ECONOMIC SANCTION AND THE MILITANTS IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . ECONOMIC SANCTION AND IRRATIONALITY ECONOMIC SANCTION AND WEAKNESS ECONOMIC SANCTION AND HOSTILITY ECONOMIC SANCTION AND FREEDOM ECONOMIC SANCTION AND HE THE SHAH AND THE HOSTAGES THE SHAH AND THE MILITANTS THE SHAH AND IRRATIONALITY IF THE DIFFERENCE IS 100 HOW DIFFERENT ARE SHAH AND THE THE SHAH AND SHAH AND THE THE SHAH AND THE HOSTAGES THE HOSTAGES THE HOSTAGES THE HOSTAGES IF THE DIFFERENCE IS 100 HOW DIFFERENT ARE THE HOSTAGES THE HOSTAGES THE MILITANTS 207 BETWEEN DIPLOMACY AND ECONOMIC SANCTION WEAKNESS HOSTILITY FREEDOM M_E. AND THE MILITANTS AND IRRATIONALITY AND WEAKNESS AND HOSTILITY BETWEEN DIPLOMACY AND ECONOMIC SANCTION AND HHEEDOM AND HE AND IRRATIONALITY THE MILITANTS AND WEAKNESS THE MILITANTS AND HOSTILITY THE MILITANTS AND FREEDOM THE MILITANTS AND Hp IRRATIONALITY AND WEAKNESS 208 IF THE DIFFERENCE BETWEEN DIPLOMACY AND ECONOMIC SANCTION IS 100 HOW DIFFERENT ARE . . . IRRATIONALITY AND HOSTILITY IRRATIONALITY AND FREEDOM IRRATIONALITY AND HE WEAKNESS AND HOSTILITY WEAKNESS AND FREEDOM WEAKNESS AND ME HOSTILITY AND FREEDOM HOSTILITY AND HE FREEDOM AND HE 209 SECTION V INSTRUCTIONS: ON THE FOLLOWING PAGES YOU WILL FIND A LIST OF WORDS AND A SET OF BOXES. WHAT WE WOULD LIKE YOU TO DO IS TO FIRST, CONSIDER THE LIST AND SEE IF YOU CAN PICK OUT SOME WORDS THAT ARE ALIKE IN AN IMPORTANT WAY. WHEN YOU HAVE SELECTED SOME WORDS FROM THE LIST, WRITE THEIR LINE NUMBERS IN THE LEFT-HAND PORTION OF ONE OF THE BOXES, AND WRITE WHATEVER THEY HAD IN COMMON AT THE BOTTOM OF THE BOX. FINALLY, IN THE RIGHT-HAND PORTION OF THE BOX, WRITE THE LINE NUMBERS OF ANY WORDS ON THE LIST WHICH ARE CLEARLY DIFFERENT FROM THE WORDS IN THE LEFT-HAND PORTION. IT IS NOT NECESSARY TO INCLUDE ALL THE WORDS ON THE LIST IN EITHER PORTION OF THE BOX. IF A PARTICULAR WORD CANNOT BE EVALU- ATED IN TERMS OF THE OTHERS, SIMPLY OMIT IT. USE AS MANY BOXES AS YOU WISH. ALL WE ASK IS THAT YOU MAKE YOUR JUDGMENTS CAREFULLY. YOU NEED NOT USE ALL THE BOXES: SIMPLY STOP WHEN YOU HAVE GROUPED ALL THE WORDS WHICH ARE ALIKE IN SOME IMPORTANT WAY. BEFORE YOU TURN THE PAGE AND BEGIN GROUPING THE WORDS, CONSIDER THE EXAMPLE THAT IS PRESENTED BELOW. HERE IS A LIST OF FIVE WORDS. IN THE FIRST BOX, THE RESPONDENT HAS LISTED LINE NUMBERS 3, 4, AND 5 IN THE LEFT HAND PORTION, AND WRITTEN "MADE WITH STRAIGHT LINES" AT THE BOTTOM. IN THE RIGHT-HAND PORTION OF THE FIRST BOX THE RESPONDENT HAS LISTED LINE NUMBERS 1 AND 2. IN THE SECOND BOX, THE RESPONDENT HAS LISTED LINE NUMBERS 1 and 5 IN THE LEFT-HAND PORT, AND WRITTEN "ROUND-SHAPED" AT THE BOTTOM OF THE BOX. IN THE RIGHT-HAND PORTION THE RESPONDENT HAS LISTED LINE NUMBERS 2, 3, AND 4. (1) CIRCLE (2) CONE BOX 1 7 (3) SQUARE ' 3 4 5 l 2 (4) RECTANGLE . ' ’ . ' (5) PENTAGON hgde With straight ines BOX 2 l, 5 2, 3, 4 ound-shaped PN3W TURN THE PAGE AND MAKE YOUR JUDGMENTS; IF YOU HAVE ANY IIIFFICULTY ASK THE EXPERIMENTAL ASSISTANTS FOR HELP. 210 Note: Five Of these pages appeared in the questionnaire (1) THE UNITED STATES (2) IRAN (3) KHOMEINI (4) CARTER (5) DIPLOMACY (6) MILITARY INTERVENTION (7) ECONOMIC SANCTION (8) THE SHAH (9) THE HOSTAGES (10) THE MILITANTS (11) IRRATIONALITY (12) WEAKNESS (13) HOSTILITY (14) FREEDOM (15) ME APPENDIX D Concept Categories and Terms Extracted From the Re-Administration of the Request For Information Regarding the United States - Iranian Hostage Situation (N=38) APPENDIX D Concept Categories and Terms Extracted From the Re-Administration Of the Request For Information Regarding the United States - Iranian Hostage Situation (N=38) Frequency Of Concept Categogy Terms Occurrence Khomenini * (30) Ayatollah 8 Ayatollah Khomeini 2 Khomeini l9 Bastard l Hostages * (99) The Event The Situation A Situation The Problem These PeOple This Whole Thing Takeover Taken Control Hostages 3 Hostage Crisis Fifty Fifty-Two Hostages Fifty-Three Hostage Situation Hostage Taking Americans American Hostages Iranian Embassy Iranian Hostages Iran Crisis Iran Hostage Crisis Iranian Crisis Iranian Situation Irans Capture Crisis Hold Them Hostage One Hostage One Guy That is Home with Disease One Came Home Our Hostages Prisoner Seizure Richard Queen US Embassy 211 HF‘NHHPJFHHFJHFHPJUHHOJHPQU1bLfiPHHFHPJHF‘U1H NLQFJHF‘FJH Concept Category Hostages (cont'd.) Iran Carter Diplomacy 212 Terms Embassy Listening Station Capture Capturing * Tehran Tehran University Two Bit Country Small Country Third World Country Iran Iran Parliament Iranian Country Of Wimps Revolutionary Government * Carter President Carter Presidential Candidate * Apology Diplomatic Ties Diplomatically Make a Swap Major Concern Of Government Lack Of Progress Negotiated Negotiations World Relations Solved Peacefully Solved Peaceably Politics Political Value Political Gains Should be Worked Out Solve the Problem Sensible Solution Dragged Its Feet Use Words Pawn Political Advantage Political Riff/Raff Frequency of Occurrence Hraham (52) w HPHPJH~UPJH+aP4q (34) 32 (27) PaH(aN2HPAPJHrdhwahaHPaundPJHPAPaHrs 213 Frequency Of Concept Categogy» Terms Occurrence Diplomacy (cont'd.) Soviet Move 1 Slow Moving 1 Freedom * (43) Free Freed Freedom Free Women Getting Them Back Lost no Matter What Release Return Seeing Them Again Aborted Rescue Mission Aborted Rescue Attempt Attempted Rescue Mission Attempt to Release Escape Helicopter Rescue Rescue Mission Rescue Efforts Rescue Attempt Rescue H GCDF‘NFJUJHPJPHdFJKJbF‘FJHLuFJN Weakness * (15) Weak Weakness Victim Of Circumstance Pushover Pushed Around Strong Intimidation Decline Power Move Power Struggle Oppression Oppressed Overpower Flopping World Power HF‘F‘HFHHJHFJPHHPJHFHPJH Economic Sanctions * (16) Economic Restrictions Economic Sanctions Economy Embargo Of FOOd and Supplies HPHPJH Concept Category Economic Santions (cont'd). Hostility Military Interven- tion Students 214 Terms Resource Embargo Freeze on Iran's Holdings Oil Oil Price Money * Deserve What They Get Hostility Hostile Makes Me Sick Rebellious Rebelling Again Revolution Sneaky Plan Torture Threat Pain Kill Violence Violent Approach Aggression * War Attack Attack Of Iraq Eight-Day War Military Action Military Strength Make an Attack Use Force Raids * Students Rebels Militants Terrorists Iranians Iranian Students Frequency Of Occurrence Hruaawra (l6) FJH HPaPJHPAPJNPHPJHPAPJH (15) wrahawrahwakam (l7) l-‘xl HUll-‘N Concept Category US Irrational Shah Other 215 Terms * US US Government America American Power Our Country Our Position The Government * Blunder Foolish Fool Ridiculous Unfair Unfairly Kept Unjustified Stupid Crazed Fanatic Panic Competent Correct Hopeless Mishandling Stunt Tyrant Right Wrong Shah * ABC Niteline Action Afghanistan Alienated Bad Back Of People's Minds Backs on Issue Biggest Messes Blamed Bleak Blacks Blindfolds Frequency Of Occurrence (22) 16 HPHFHAPJH (l9) HF‘P‘HF‘F‘HF‘F‘HPAFJHFHPJHPHPJH (16) (135) HP‘PHHPJHFJFHHFJHFH Concept Category Other (cont'd.) 216 Terms Brainwash Be Home Canada Coup Camp Campaigns Control Death Dead Draft Draft Registration Disgraced Election Time Election Day Embarrasment Exhausted Exploit Fear Family Ford Free Nation Forgotten About Funding Gold Grief Getting out Of Hand Help Helping Hodding Carter Hero Held in Fear Planes Marines Spy Trial Spies Suffer Super Power Supply Of Weapons Ted Koppel Triumph The West Tired Tried as Spies UN UnprOVOked Wait and Hope Waiting Screw Up Screw Up Escape Plan Visas Frequency Of Occurrence )HFJHFHFJthhflfldePdF‘HFHB)NFJFJH+4FJHPHFJHFJFJHFHFJHFHFJHF‘FJHFJPJHFHUJHFHPJHPOFJH Concept Category Other (cont'd.) 217 Terms. Women World Opinion Worst Thing Ever Worst Things Thats Happened Year Long Helicopters Downed Helicopters Time Third World Thanksgiving Taking Hopes for Election Increase Indecision Instable Conditions Imperialistic Imperialism Islam Juvenile Kremlin Leader Loss Lost Interest Love Media Mid-East Murder Multiple Sclerosis Moslem Muslim Negative Nixon November Niteline News Foiled Failed Boni Sadr Failure Iraq Nationalism October Persian Gulf Prayers Protest Power Powell Pressure Safe Frequency of Occurrence HFJPHHPJPWHFJdeO\HFHFJHFJFJHF‘PHHPJFHHPJHPJFJHFHFJHPHFJHFJFJHPHFJNPdPJHF‘F‘HP‘ Concept Category Other (cont'd.) 218 Terms Still There Should Have Never Happened We must Face an Action Salience Sensible Authority Sacrificing Frequency Of Occurrence Hrakdwraha AP PENDI X E A Brief Description of the Data Assessed From Student Records APPENDIX E A Brief Description Of the Data Assessed From Student Records Testing Bulletin NO. 3 Prepared by (Revised, 2-75) The Office Of IENaluation Services University College Michigan State University THE USE OF ORIENTATION TEST DHI‘A All new undergraduate students at Michigan State University take a set Of examinations generally known as "Orientation Tests," the results Of which are distributed four times a year to all departments. Although scores on these tests are used regularly by Admissions Officers, Counselors, and others, this bulletin is particularly directed to the instructional faculty, administrative and academic advisers, who may find the scores helpful. As measures Of ability important in academic work, the test scores are Of value in ascertaining the ability patterns Of individual students or groups Of students. Knowledge Of this kind has proved to be useful in academic advising, class section- ing, and understanding student progress. Brief Description Of the Tests The NBU Reading Test, Form A-73,1(IVSU-R) is a 112-item test of reading comprehension and vocabulary. TwO sub-test scores and a total score are reported. The reading vocabulary sub-test (RV) is 72-item test asking for definitions Of words in context. Words and meanings were first selected from academic and leisure reading material carmonly available to freshman students, and then revised by successive try-outs on the 1970, 71, and 72 entering student groups. For a group of 1000 fall, 1973, new freshman students, the K—R #20 reliability of the Ibading Vocabulary sub-test was .92. The Reading Comprehension sub—test (RC) is a 40-item test measuring comprehension of college-level reading passages. The score is based upon the student's ability to answer questions based on reading passages repre- sentative Of several basic academic areas at MSU. The test is not re- stricted tO the simple mechanics Of reading, but rather the score provides some measure Of factors involved in critical thought. Reliabil- ity Of the K2 sub—test was .84 in fall, 1972. (The fall, 1972, K-R #20 reliability Of the entire 112-item test was .94.) lOsmond E. Palmer, Ed. PBU Readirg Test, Form A—73, OES, DBU, 1973. 219 220 The MSU—Reading Test is useful to faculty members in any decision requiring some knowledge about the student's verbal ability. It is routinely Leed as one basis for assigning students tO the American Thought and Language Comprehensive English Program (ATL 101-102-103) . 'Ihe MSU Arithmetic Placerent Te8t and the MSU Mathematics Est (algebra) are also administered as a part of the Orientation Tea bat- tery. All freshman students are required tO take both the 40-item basic arithmetic test and the 30-item high school level algebra test. Trans- fer students, in SCIIE cases, are required to take the Mathematics Test. The Mathematics test was designed by the Department of Mathematics specifically for course placement in IVSU rmath courses. The Arithmetic test is Of value in detecting students who are deficient in basic arithmetic, and is used for placerent in Quantitative Techniques (IS 194) . The K—R #20 reliability was .89 for the DBU Mathematics Test and .83 for the MSU Arithmetic Test. Reliability for the Total Math score (Aritl'mretic plus Algebra) was .92. In addition tO locally-developed test instruments, scores fram two national tests may also be included in a student's test score profile (these are generally taken by students as juniors or seniors in high school) . The American College Test (ACI‘) assesses developed academic abili- ties in four areas Of high school and college curricula: English, mathe- matics, social studies, and natural sciences. Four standard scores in each Of these areas plus a Composite Score are reported. Scores range from 1 to 36 with a national college-bomd high school senior mean Of 20. The composite is a weighted total Of four subject-matter areas. The Scholastic Aptitute Tea (SAT) also measures developed academic ability. It reports two standardized three—digit part—scores: Verbal (SATLV) and Mathematics (SAT-M) . These standard scores mere originally computed to yield a national college bound high school senior mean Of 500 with a range from 200 to 800. The total score (SAT-T) therefore has a possible range from 400 to 1600. Beginning in fall, 1974 the verbal and mathematical sections were shortened by 15 minutes and a 30 minute Tea Of Standard Written English was included. Scores range from 20 to 80. The TSNE was designed to select students "who need some degree Of additional instruction in written English usage. " These additional scores may be available in fall, 1975. The Method of Reporting Scores Scores on the MSU Reading Test, the Arithmetic and Mathematics Teas, and one national test are available on the Orientation Record which each new freshman student must carry whei reporting to his academic advisor for planning his first term's program Of studies. The scores reported on the Record are raw scores (number Of questions answered cor- rectly) , and percentile ranks (PR) in graphic form. The PR "score" 221 specifies directly the percentage Of freshmen entering 160 who receive scores which are lower on a given test than that reported for a given student. A PR Of 85 on the MSU Reading Comprehension sub—test (RC), for example, means that a student was quite superior in reading comprehension since his score on the test was higher than the scores Of 85 percent Of the freshmen entering MSU that year. It also means that he ranked among the highest 15 percent Of students on reading carprehension. In contrast, a PR Of 08 is a rather low score since the large majority (92 percent) Of freshmen received scores as high or higher. Scores fram the Orientation Record are used to specify class placement levels in beginning English and mathematics courses. Each term Academic Advisors receive the formal report, Test Scores By Entering Students, through their departmental Offices. This report contains test scores for all students who participated in the most recent Orientation testing program, reported as percentile ranks (PR) . Other test records available to faculty and advisors, and maintained by the Office Of Evaluation Services include Cawarative Standings Of Various College and Curriculum Grows on the Orientation Week Examina- tions, Orientation Itast and High School Grade Norms and GPA Expectancy Tables for Freshmen. For large research projects, actual raw test scores can be provided on request. The "Comparative Standings" report presents summary test data for both new fresl'mren and new transfers in different curricula. It reports "typical" performance Of students within each curricular grow. Since ability levels on different tests vary consider- ably among curricular grows, a student with a given score may be expected to perform at different levels in different cm‘ricula. The expectancy tables report, by sex and total grow, the relative first term grade achievement Of new students ordered by their orientatim test score levels. The Predictive Value Of the Tests The Orientation Tests serve many purposes. Each test must satisfy different criteria to be considered a valid measure. Tb be useful for many problems in the diagnosis of individuals or grows, however, a test mmist measure a relatively stable ability that is indicative Of the later quality Of college work. A canton method Of evaluating tests for pur— poses of this kind is to compare the standings Of students on the tests to their later academic attainment as reflected in grade point averages. These comparisons form the basis for the GPA Expectancy Tables for Freshmen. Results from studies Of this kind have demonstrated that all Of the tests are Of some value in predicting grades. The degree Of relationship does vary, however. The total score on either national test (ACT or SAT) has generally proved to be the best single predictor Of the grade point average (GPA) for all students in general, followed closely by the 160 Reading Test total score. While this pattern is evident for both sexes, grades for women are usually predicted more accurately. 222 The meaningfulness Of prediction as a factor in determining the degree to which test data can be relied wai in student advising is illustrated in Table I, which shows academic attainment Of women the first term at NBU relative to scores earned on the NBU reading mst Total score for 1973-74. The percentage Of stuients earning "satis- factory" (2.0 or higher) or "honors" (3.0 or higher) first term GPA is reported for successive percentile score intervals on the NED-RT. TABLEI PEIKIENTAGE OF FIRSTTER‘TFRESIMNWAT INDICATEDPRIE/EISONTHE PBU-RI‘ WITH A SATISFACIORY FIRS'IUIERM GPA PSU—RI‘ PR Level 0-19 20-39 40-59 60-79 80-99 % with 2.0 and 3.0 _ GPA and w 81-04 82-07 87-10 93-22 95-38 Table I shcwrs that four-fifths (81 percent) Of mien students whose scores ranked them among the lowest 20 percent on NBU-RI‘ made satisfac- tory progress, but only 4 percent earned a 3.0 or higher. In contrast, 95 percent Of waren with PR's on the NBU—RT which ranged from 80-99 earned satisfactory grades and almost two-fifths (38 percent) earned honors level grades. Table II illustrates how EB relevant tests considered together in- crease the precision Of prediction. As before, the mmmbers in the body Of the table give the percentage Of students at specified test levels garnering a 2.0 or 3.0 GPA or higher, but the probability figure to use is dependent on both the MSU-PCP and one of the national tests (ACT or SAT). A student whose tested ability on the MSU-RP falls in the 40-59 PR range would seem at a level where typically 87 percent Of waren earn a 2.0 or higher their first term. Depending on the level Of tested ability for the second test, she may be at an overall level where frcm 80 to 95 percent Of women typically earn a first term GPA Of 2.0 or higher. TABLEII PW OFFRESIWWQ’IENAT'IHE40-59 PRIMOF'IHEIVSU-RI‘ANDIN- DIGXTED LEVELS OF THE NATIONAL TEST SCDRE EARNING A 2.0 or 3.0 (or higher) FIRST TERM G.P.A. MSU-RI National Test Percentile Rank %—ile T—19 20-39 40-59 60-79 80-99 40-59 80—04 85-07 90-10 95-21 95-30 Knowledge of high school GPA further increases precision Of pre- diction. In 1973 a female student at the 40-59 PR level on both the MSU- RI‘ and the naticnal test total score, who earned a high school GPA in the range 0.0-2.9 was at a level where typically 83 percent Of students earn a 2.0 GPA or higher their first term. Had her high school GPA been in the range 3.0-3.4, the expectancy percentage would have been 88, and 98 had her high school GPA been in the range 3.5-4.0. 223 It is clear from these data that students at all levels of tested ability have a good chance tO succeed in college. meledge of the nature Of the tests, study Of test score patterns, and student back- ground Often make it possible to isolate reasons for a student's succeeding or failing, especially when a strong academic background is desired for entrance tO upper college Of graduate education. Caution. - The illustrative data above are not directly applic- able tO all students. Male studmts tend to exhibit lower levels Of attainment at Specified ability levels than do women students. The magnitude Of the GPA tends to vary from curriculum to curriculum. Other patterns Of test scores are more predictive in certain curricula than scores used in the illustration. The Office Of Evaluation Services has further data which can be provided for persons interested in making more detailed analyses. Potential Applications Of Test Data Specific applications for test data will be suggested in this section. Applications tO the classroom, Applications to Student Advising, and Other Applications will be considered. A. Applications to the Classroom Example 1. - You have two students in class who seem to be outstand- ing students. You feel they Should be encouraged to pursue independent work and to plan a long-range program. The Orientation Score profiles (PR'S) for Students A and B are: PISU Test Percentiles ACI‘ Percentiles BY. 53. ET. A E E F: .1! 5—5 B.S. m A 84 71 84 99 97 99 89 99 95 96 97 B 24 42 30 46 52 51 08 57 40 4O 37 Scores for Student A confirm your initial hypothesis. His per- formance On the tests is outstanding. He may have even more ability than he has shown in class. Tea data for Student B are much less consistent with your beliefs. When data from several sources lead to the same conclusion, as with Student A, one can feel more confident initiating a proposed plan Of action. Where inconsistencies are found, as with Student B, additimal stirly is necessary before action is taken. 1e 2. - One Of your classes seers lackadaisical. Techniques and procedures which have worked well with previous classes seem to "fall flat." You tabulate the scores from Test Scores py Entering Students and find the following pattern: National (PR) fists Reading (PR) Tbtal 90-99 90—99 u 80—89 I 80-89 70-79 70-79 . 60-69 :1 60-69 W 50-59 M 1‘ 50-59 mm 40-49 rm m 40-49 m M 30-39 M m. 30-39 rm nu 20-29 M.) (N ) 20-29 N W " 10-19 mm M 10-19 IN (N 1 0-09 m 0-09 N Median Score 32.0 28.3 Both distributions Show the same pattern. The students, as a grow, score unusually low on the two tests. The data certainly point to this grow being atypical, and suggest that the procedures used might be "over the heads" Of the class. Had the analysis showed the grow to be "very superior," a related hypothesis could be suggested. Lethargy can also accompany instruction which is keyed below the general level Of the Class. Class analysis Of this kind is most effective when several classes are studied and comparisons made. Reference to Cognarative Standings Of Various College and Curriculum Grows on the Orientation- Week Examinations could also make an analysis of this type more pene- trating by focusing on the general level Of tested ability to be expected in certain classes. Em 1e 3. - Additional actions which might be suggested by refer- ence to the Orientation Tests include: a. Special aid to students deficient in Specific areas. b. Referral to developmental or supportive services, i.e., learning Resources Center, Center for Swportive Services, Improvement Services, Comprehensive English, or depart- mental tutoring programs. c. A search for special programs for students who seem not to be working w to their abilities. B. Applications to Student Advising Advising or counseling is always a complex process where ability, interest, emotions, and other personality factors must be considered. The suggestions which follow must be considered only as clues coming from one source, and must not be followed mechanically. Case I. A student comes in tO plan his next term's program. Grades from previOLs terms have been on the 1.5 - 0.0 borderline. His Orientation Test scores are: 225 IVEU Test Percentiles RV RC RT A 14. '94. £3. E 9.8. MS Gee 21 12 15 11 ll 09 08 07 15 14 09 The scores are uniformly low. This is consistent with his perfonmance in college. The scores do not suggest any special need for specialized remedial programs since nO specific disability is suggested. A com- plete reevaluation Of his educational and vocational plans would seem advisable. Ieferral to the Counseling Center, where facilities for service Of this kind are available, sfould certainly be considered. Until a more intensive analysis is made, temporary provisions such as reducing class or extra-class activities might be suggested. Case II. A student has exhibited near-average work in University College courses but has failed both the chemistry and mathematics courses in his major area Of concentration, which is Engineering. His Orientation Test scores are: MSU Test Percenti les .5! .BE .32 :E .E. 12! E2 11 §§. NS .EETE 52 43 48 45 66 43 37 64 55 55 52 His quantitative ability is about average for students planning to elect a course in mathematics but well below the average score for students majoring in technical curricula. The general test score pattern is one which would describe an average NBU student with some- what greater than average strength in the quantitative area. Success- ful students in the technical curricula, however, are generally average or above in most tested abilities and higly superior in the quantita- tive areas. Students with ACILM and Total Math scores that do not rank arong the highest 20 percent Of all freshman norms, and with ACPNS scores that are not substantially above average, do not have a high likelihood Of success. A change in curriculum preference may be con- sidered in this case, or if this is not desirable, erphasis could be placed on having the student work earnestly at strengthening possible weaknesses in his mathematical and basic science background. Case III. A student transferring 90 credits from a community college wishes to major in Accounting in the College Of Business. His general community college GPA was 2.42, but work in courses related to MSU Business Curricula was less than 2.0. He wants to re-take these courses at PSU to meet Business Admissions requirements. His DBU test score percentiles were: .! 5!. .BE. .32 42 20 33 25 Again referring to the Report Of Corparative Standings, typical trans- fer students in his major preference have reading scores at a slightly 226 higher level, and freshmen majors have math scores at a considerably higher level. Competition in Accounting might be especially frustrating for this student. Business-oriented majors in other colleges may be more appropriate to his tested abilities. Case IV. A student is very submissive and seems to lack self- confidence. He looks upon his inferior past achieverent as a major calamity and considers himself to be worthless in a number of ways. He seers to lnave withdrawn within himself and participates in no college activities. His test scores are as follows: NSU Test Percentiles ACT Percentiles $3951; A M'IM E M_S_S__N_S_COHp 50 48 50 57 63 62 56 67 65 63 60 These symptoms suggest a general problem in adjustment. In cases of this type a simple diagnosis or solution is usually unlikely. While test scores may yield some clues on the problem, they can seldom be used in a simple prescribed manner. In cases of this kind referral to the Counseling Center is usually advisable, but an understanding faculty member working in cooperation with professional counselors can be doubly effective. C. Other Applications The data from the Orientation TeSm are available for individual or departmental research projects. The scores have been used widely as control data in experiments on learning and for inquiries into the characteristics Of students found in a given curriculum. When desired, members Of the Office Of Evaluation Services are available for consulta- tion on evaluation methodology or research design. Trans fer Students While percentile rank scores are reported for all new stuients, freshmen and transfers, the reference grow used in cotputing these PR'S is new freshmen only. The result is that transfer students with pre- vious college experience may appear to score higher on the tests than advanced NBU students who took the tests as freshmen. High scores for transfer students should not, as a consequence, be regarded as indexes of swerior ability without some adjustment, although low scores are equally as useful in revealing deficiencies for transfer students as they are for freshmen. In the latter sense academic advisors may find it particularly helpful to recommend a reredial service for transfers with low test scores and borderline academic attainment. New freshmen are routinely assigned to improvement services on the basis Of test scores but transfer students are not screened for possible disabilities in reading. 227 A Note Of Caution Tea scores must never be considered infallible. While the scores on carefully constructed tests are much more dependable than impressions secured from casual classroom experiences or individual conferences, any one test score must be regarded only as indicative and never final. In this regard, it is usually advisable to View a score as a possible range Of scores, i.e., a PR Of 55 is considered as any Of the PR's in the range from 40 to 70, or a PR Of 95 in the range from 90-99. The range suggested above is deliberately larger near the average, since one can generally place more confidence in extremely high or low scores that are reported as percentile ranks than in scores near the PR Of 50. Further- more, when inconsistencies are found among similar tests or between tests and attainment, retesting is Often advisable. A Few Quick Guides Routine procedures which others who work with students have found tO be fruitful include the following: 1. The scores Of advisees can be recorded on a convenient record sheet. This sheet can also include other easily summarized background information, such as previous grades. Sometimes information Of this kind is secured for small classes where individualized instruction is possible. 2. Tea scores can be corpared to actual scholastic attainment. Students with marked discrepancies in the two sets Of measures are noted for further study when the Opportunity arises. 3. Before beginning a conference with a student, a moment spent in scanning the record sheet may provide a useful orientation for the conference. 4. When a new student reports to his advisor the first time, the student should have a cOpy Of his Orientation Record available with his test scores profiled. These scores may be useful in recomending the size Of class load to carry, or the types Of courses to select. 5. The average test scores for students in a class may be used to help determine the relative nmber of extreme grades (4.0's, 3.5's and 1.5's or less) to be assigned to a class. However, grades for an individual student should never be influenced by these scores. APPENDIX F Instructions for Calling Participants APPENDIX F Instructions for Calling Participants Hello, my name is and I'm calling from Dr. Ralph's Office in the Department Of Commun- ication. We are currently working on an important project which is examining how people process information. And we would like tO know if you could help us out. All you would have to do is stop by at your convenience, and fill out a questionnaire. DO you think you could help us out? When would it be convenient? If they say» no." It is really important for us to get as many people tO participate, because some Of the results from this study will likely be used to improve the undergraduate curriculum. Are you sure you can't find time? 228 229 Some Common Questions How Was I Selected? all juniors and seniors in the Department Of Communication will be given the Opportunity to participate. What Will I have TO DO? all we would like you to do is come in and fill out a questionnaire; it shouldn't take any more than one-half hour Of your time. What's The Study Trying tO DO? basically, the study is trying to assess public Opinion toward a few current news topics, and we want to see how your communication activity and information processing habits influence your Opinion. How Will These Data Help The Undergraduate Curriculum? some Of the results may be used to recommend that certain skills he acquired during your collegiate career. AS you are probably well aware, it is a difficult task tO determine what the best "mix" of classes is for each individual. Hopefully, the results will provide some insights into this problem. .ALL RESULTS WILL BE CONFIDENTIAL. NO ONE WILL KNOW HOW YOU RESPONDED. APPENDIX G Procedures that were Followed to Greet Participants and Introduce Them to the Study APPENDIX G Procedures that were Followed to Greet Participants and Introduce Them to the Study Instructions for When the Participant Arrives # # Introduce yourself and ask them to sit down. Wait a few minutes for all persons to arrive. Don't wait tOO long (maybe 5 minutes). Explain that the study is examining information proces- sing, and what we would like them to do is fill out a questionnaire. Tell them that: The first part Of the guestionnaire asks a little bit about their background; the second part asks them tO state their opinion about a few topics, and the thirdppart asks them togperform a simple grouping task. HOWEVER, before they can begin, you must present them consent Form A, and have them Sign it, date it, and put the time of day on it. Explain to them that the consent form is required by the University, and that it is simply a statement of their agreement to participate. Tell them once they finish the questionnaire, you will debrief them and tell them more about the nature of the study. After they Sign the consent form put the time Of day (to the nearest minute) on the questionnaire and let them begin. Tell them that if they have any questions, or experience any confusion over how to answer any question, or are just not sure about what to do-—pg EEK ygp. Tell them you'll be right outside, and that when they are done to just come outside and give you the questionnaire. Once they finish the questionnaire and they come to hand it in, note the time Of day (to the nearest minute) and write it on the questionnaire. Thank them for participating, and ask them to sit down so that you can explain the nature Of the study. Tell them: 230 231 l. The study is concerned with how individual's process information. 2. The major question being tested is the extent to which a person's environmental conditions will determine how he/she will process information. 3. The first part Of the questionnaire asked them to tell us about what things were like when they were growing up, and how things are now. In particular, we were interested in what kind Of information environment they were exposed to, because differences in environment Should cause differences in processing. TO test this: 4. Our theory predicts that there should be differ- ences in the kinds Of numbers pegple use in the second part Of'theyguestionnaIre, depending upon what the environmental conditions were. Of course, we may be wrong, but this is what we are interested in finding out. 5. Finally, tell them the Communication Department is interested in the results because the findings may suggest how to "track" students. If our "guesses" are supported, it is possible that in- stead Of putting people in "remedial" or "acceler- ated" tracks, we might want to put them in courses that teach them new symbol sets and/or different ways to package information or view the world. For example: in math courses, logic courses, foreign languages, etc. Ask them if they have any questions. If not, tell them that we would like to access their student records tO extract ACT 8 SAT Test Scores MSU Entrance Exam Scores High School GPA, and College GPA Tell them that they dO not have tO grant us access to their student record information, but thatgit would really be beneficial to our study if they did SO. Assure them that: (1) their information will be treated carefully, (2) con- fidentially, and (3) there will be no way for anyone else to find out what their student record has in it, or how they performed on this questionnaire. 232 If they say "OK," present them with consent Form B. Have them Sign it and date it. Thank them again and ask them to cooperate by not explaining to anyone else the nature Of the study or what they did. Tell them the study is going to run for two weeks, and we don't want to bias any future participants. Tell them we are asking a lot Of communication majors tO participate, and we don't want to screw-up what work we have already done by having people know what's going on, and try tO respond either: (a) in ways tO help the study, or (b) in ways to hurt the study. SO, mums the word OKAY? OKAYII T H A N K S! APPENDIX H Consent Forms A and B APPENDIX H Consent Forms A and B Michigan State University Department Of Communication Form A 1. I have freely consented tO take part in a scientific study being conducted by: n, J, Stoyanoff under the supervision of: Dr. David C. Ralph Academic Title: Director of Undergraduate Studies The study has been explained to me and I understand the explanation that has been given and what my participation will involve. I understand that I am free to discontinue my participa- tion in the study at any time without penalty. I understand that the results Of the study will be treated in strict confidence and that I will remain anonymous. Within these restrictions, results of the study will be made available to me at my request. I understand that my participation in the study does not guarantee any beneficial results to me. I understand that, at my request, I can receive additional explanation Of the study after my participation is completed. Signed: Date: 233 234 Michigan State University Department Of Communication Form B I have been informed on the nature of the research being conducted, and agree to let the Department Of Communication provide access to my student records to the persons involved in this research effort. I recognize that these investigators will employ certain procedures tO combine my student record data with the questionnaire data, and understand the procedures that will be utilized to protect the confidentiality Of my person- al data. I understand that I may withdraw my participation in this study at any time. N. J. Stoyanoff is in charge Of this specific activity. Signed Date APPENDIX I Model Indicators, Labels, and Corresponding Questionnaire Items APPENDIX I Abdel Indicators, Labels, and Corresponding Questionnaire Items Theoretic Construct Indicator label Questionnaire Iters Social Structural Factors (6) x FINCOME WHEN YOU WERE GROWING UP, 2 WHAT WAS THE AVERAGE YEARLY - INCOME YOUR FAMILY LIVED ON? 5 /YR (estimate in 1980 dollars) FOCII FATHER'S CECUPATIO‘I? (WHAT HE IDES/DID, NOI‘ WHERE HE MDRKS) Lsee Appendix J) X3 MIX: NUI‘HER'S CECUPATION? (WHAT SHE mES/DID, NOI‘ WHERE SHE VDRKS) (see Appendix J) X FED WHAT IS THE HIGHEST GRADE OR YEAR IN SCH(I)L 'I'HAT YWR FATHER CDMPLE'IED? (see prendix J) X MED WHAT IS THE HIGHEST GRADE OR YEAR IN SQKXDL THAT YOUR MINER CDMPIE’IED? (see Appendix J) x SRNCM A SCXIIALLY RESPONSIVE LIFE- STYLE IS ONE IN WHICH AN INDI- VIDUAL'S BEHAVIORS ARE ACKNGVL- EIIEED AND RESPONDED '10 BY OTHERS. IF ZERO MEANS I‘UI‘ AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIEMZE . . . I'm SCEIALLY RESPONSIVE IS YOUR LIFESTYLE NON? 235 236 Theoretic Construct Indicator Iabel Questionnaire Items Social Structural Factors ( 5,) ( cont ' d) x 10 DIVNG‘V INDIW A DIVERSE LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL DOES THINGS WHICH ARE VARIED, DIFFERENT, AND IMJSUAL. IF ZERO MEANS NOT 4 AT AIL, AND 100 DESCRIBES THE T' AVERAGE PERSON"S EXPERIENCE . . . r 3: HCW DIVEIBE IS YOUR LIFESTYLE NON? AN INDEPENDENT LIFESTYLE IS ONE IN WHICH AN INDIVIDUAL IS SELF- DIREC'IED, AU'IONOMIXJS, AND FREE '10 II) WHATEVER HE/SHE LIKE. IF ZEK) MEANS NOT AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HCWINDEPENDENTAREYCXJNGN? ANURTUREDLIFESTYIE ISONEIN MHQIANNDMDLRL'S EMOTIONS, FEELINGS, ANDWELL-BEINGARE CAREDEORBYOIHERS. IFZERO MEANSNOTATAIL,AND100DES— CRIBESTHEIWERACE PEIBON'S EXPERIENCE . . . HWNUR’IUREDAREYOUMW? ACDHE‘SIVELIFE‘STYLEISONEIN MCI-IANINDIVIDUALI‘STABLISHEB CIDSEREIATIONSHIPSADD‘JG FRIENIBANDFAMILY. IFZERO MEANSNOTAT ALL, ANDlOO IES- CRIBES'IHEAVERAGEPERSON'S EXPERIENCE . . . 237 'Iheoretic Construct Indicator label Questionnaire Items Social Structural Factors (5) (cont ' d) x11 Information Hwironment (n ) 1 Y1 HON CDHESIVE IS YOUR LIFESTYLE NON? AN mIEIIEc-mAL LIFESTYLE F IS ONE IN WHICH AN INDI- E 'u VIDUAL IS SURKDUNDED BY INIEILIGENT, W, AND COMPETENT PEOPLE. IF ZEK) MEANS NOI‘ AT ALL, AND 100 II‘SCRIBES 'IHE AVERACE PERSOJ'S EXPERIEMIE . . . HON WW IS YOUR LIFES'I‘YIENCW? AN ACTIVE INFORMATION EN- VIHJNMFNI‘ IS ONE IN WHICH AN INDMDUAL PKXIBSES A IDI‘ OF INFOMTION. IF ZEK) NEANS NOT AT ALL, AND 100 IESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . WACI‘IVEAREYOURIN’IER— PERSONAL WICATIONNEP- VDRKS W? HWACI'IVEAREYOURMEDIA EXPERIEI‘CESW? 238 Theoretic Construct Indicator Label Questionnaire Items Information Environment (”1) (oont'd) y3 STNTNOW STMENOW SRNTNOW A.STIMUDNEH«3]1EIE®EHTON ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PRO- CESSES A.LOT'OF INFORMA- TTON WHICH TEACHES HIM/HER NEW THINGS AND STIMULATES HIS/HER CURIOSITY AND IMAGINATION. IF ZERO MEANS NOT AT.ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW STIMULATING.ARE YOUR INTERPERSONAL OOMMUNICAF TION NETWORKS NOW? HOW STIMULATING ARE YOUR MEDIA EXPERIENCES NOW? A.SOCIALLY RESPONSIVE IN- FORMATION ENVIRONMENT IS ONE IN”WHICH AN INDIVIDUAL PROCESSES A.LOT OF INTORP MATTON WHICH ACKNOWLEDGES OR RESPONDS TO HIS/HER.BE- HAVIORS. IF ZERO.MEANS NOT.AT.ALL,.AND 100 DES- CRIBES THE AVERAGE PERSON'S EXPERIENCE . . . HOW SOCIALLY RESPONSIVE ARE YOUR.INTERPERSONAL COMMUN- ICATION NETWORKS NOW? 239 Theoretic Construct Indicator label Questionnaire Items Infonnation Environment (n1) (oont'd) Y6 A DIVERSE INFOM'TION ENVIRONMENT IS ONE IN WHICH AN INDIVIDUAL PED- CESSES A IOT OF INFORMA- TION ABOUT A VARIETY OF TOPICS, AND WHICH INFORIVB HIM/HER ABOUT DIFFERENT 'IHINCS ABOUT THE TOPIC. IF ZERO MEANS INUI‘ AT ALL, AND 100 DESCRIBES THE AVERAGE PERSON'S EXPERIEML‘E . . . HCM DIVERSE ARE YOUR INTERPEIGONAL MINICA- TION NETWORKS W? HOV DIVERSE ARE YOUR MEDIA EXPERIEDCES W? A NURTURING INFORMATION ENVIKDNMENT IS ONE IN WHIG'I AN INDIVIDUAL PRO- CESSES A IDT OF INEOIMA- TICN WHICH IS DIREC'ED AT HIS/HER FEELINGS, ENDTIOINB, AND WELL-BRIDE. IF ZERO MEANS WT AT ALL, AND 100 IESCRIBES THE AVERAGE PERSON'S EXPERIM . . . HOV NURTURING ARE YOUR INTERPERSOML WICK- T'ION NETWORKS MW? 240 ‘Iheoretic Construct Indicator label Questionnaire Items Information Environment (*1 ) ( oont ' d) Y9 NHRSTKT PRINTIME ACI)HESIVE INFORMATION ENVIIU‘J- MENTISONEINWHICHANINDI- VIDUAL INTERACTS WI'IHASETOF PEDPLEWHOAREHIGILY INTE- GRATE'DANDWIDEm-IANGEAIOT OF INPOHVIATION WITH EACH CHER. IFZEH) MEANSNUTATALL, AND 100 DESCRIBES'IIEAVERACE PERSON'S EDCPERIEI‘KZE . . . HOV CIIESIVE ARE YOUR INTER- PEIGONAL (INMUNICATION NEIVDRKS W? ON THE AVERAGE, AmU'T HCM MNY HOURS PER WEEK II) YOU SPEND TALKING WITH THESE YOUR FRIENIB? INDICATEHCWMANYHOURS PERWEEK YOUSPENDREADII‘BEACHOFTIE W: THE STATE NEWS HRS/WK THE LANSING JOURNAL HRS/WK THE DETROIT FREE PRESS HIE/WK THE DETKDIT NEWS HIS/WK THE NEW YORK TIMES HRS/9R 'ITE WALL STREET JOUINAL HRS/WK PERIODICALS (MAGA- ZINES) HIE/WK aLPRIN'I'IME was an indicator which was computed using the following equation : PRINrIME = (-.0005) (SN) + (.0018) (IJ) + (.0333) (PP) + (-.1029) (DN) + (.0039) (NYT) + (-.0695) (WSJ) + (-.1569) (MAG) 241 Theoretic Construct Indicator Label Questionnaire Items Information Environment (n1) b (Cont' d) yl2 NEWSTIME INDICATE HCW MANY HGJRS PER WEEKYOU SPENDMTCHING ... . TV NETWORK NEWS HRS/WK TV 1m NEWS HRS/WK TV MORNING NEWS PRDGRAD'S HRS/WK TV LATE NIGIT NEWS HRS/WK where: number of hours per week spent reading the State News II II II Lansing Journal II II II II II mtmit Free Press II II II II II II II II mtmit Ms II II II II II II II II M York T111185 II II II II II II II II wall Street Journal II II II II II II II II Pericxiicals (Nagazines) 'I'nese coefficients were the factor score coefficients obtained from a confirmatory factor analysis which examined the extent to which these six questionnaire items loaded on a single factor. See Tables I-1 and I-2 for the full results of this analysis. hNE{~7:'-.7I'IIVIEZ was an indicatory which was computed using the following equation: E $32 a as NEWSTIME = (1.323) (TVNN) + (-.642) (IDNW) + (-.509) (ANNA?) + (-.229) (PMNW) where: m = number of hours per week spent watching TV network news IDNW = " " " 'IV local news AMIW = " " " " " " " 'IV morning news PM = II II II II II II II IIV late night news 'Ihese coefficients were the factor score regression coefficients obtained from a confirmatory factor analysis which examined the extent to which these four questionnaire items loaded on a single factor. See Tables I-3 and I-4 for the :full results of this analysis. 242 Theoretic Construct Indicator label Qiesticnnaire Item Indicators y to y were all derived from the MMDS data. Indicators y to y 7 8.1153 simpihoz descriptive statistics which describe the paired- cogpar' n scores generated by each individual, while indicators yl to y describe the summary statistics obtained from the factoring 8f the agriance—covariance matrix generated by the paired-comparison SOOI'GS . Processing Style (03) yl3 NOS Total nurber of unique scores used yl4 POP Total number of scores = 0, 50, or an integer multiple of 50 y15 C.V. Coefficient of variation among scores used yl6 NEW Total number of scores : lO yl7 NEXMAX Total number of scores 3 1000 Cognitive Structure (n3) (differentiation) y18 NDR Number of real dimensions (discrimination) y19 TRACE Trace of the eigenvector matrix (integration) y20 WARP Sum of the positive eigenroots sum of all (positive and nega- tive) eigenroots y21 was derived from the Scott "Listing and Comparing Task." (differentiation) y21 H Computed using the formula: 1 log2 n-n [Zni log:2 ni] where: n = the number of stimuli presented (19) n. = the number of elements in each group- ing pattern. 243 5 8 0950353 8m: 33 93 ._~_ A 30x,” can: . ucmumcoo + wannabes .H 3 wow .mHOumowc—hn 05 m0 363on m “cm .33 mfimmds 0» man cog—Egan". mmmo mo “:5me may mum woman a Hancocbm ooo.m~ ooo.o 35d mmhfi ~36 mm mm: ooo.m ooo.s main 0mm. vmv. mm ems 08.2.. 25.0 mmmé mam; mom. mm E 08.5 ooo.o omoé. oSJ mow. mm 20 08.3 80.0 «.24 mmm.~ mmm.~ mm mm oom.m ooo.o vom.m aNHA mmo. 3 B oooé ooo.o 0mm. manta omm.m mm 2m Boom 5558.: 88m E2552 Exm :03"...ng bgm 5mm: 2 HBMOECH mfimouzv gunman mo muOumofiucH 05 ecu moflmflmfim gm TH 033. 244 omm.I hmbm. mmvm. mm. was Hmm.a Hmvm. mmem.I mm. nmz mmm.a moma.~ mmmm.MI Ho. 9&2 ooo.a mama. hmo¢.I mm. 2D mmv.I anom.a mahm. mm. mm moh.a vmmm. hmm¢.I mm. ha I I I I 2m mgxm mom-58289 mwuoom pgemcmb. mo meoom Eggs ucBmcoo ~303ch mo 32m ? SEEK/8 Buccfim no can). AU.u:00v HIH mHnt 245 £3885 853mg n. no 95mm cu gm ma 8am.» mind 6030mm.” on uoccmo can saga 889.8 @5qu 05 8 88.93 83523 mafia 05 88090 .888 38:3... 0&5 .H x: mom .muBmown—J m5 mo #33on w Mom .HIH wanna. mom “muoumoflg Ham new mmuoom EnoMguu so comma mun mwmxamcm mg £38.69." comm mo mnwm 398m 93 Mom HIH manna. wwmm Noam. u Hm>oH suaaanmnoua mm.¢~ u x can? aemH.I AMH.V km. mv. inc mm.vm~ mg: mmoo.- AHH.C me. am. inc hm.akfl ems mmoo. Ask.c om.m pa. inc om.omm ssz mmog.- AHH.V me. an. imm.vovmc mm.q- zo mmmo. i-.c mN.H we. inc mq.smm as mass. ANH.V am. so. loo.oc ooo.H ha” mooo.I iom.c Hm.H Ho.I imp.oavc H~.m- 2m ucmflofluwoo 30me mo echo 830.8 38.3 pong “Booed—CH 88m “BUM“ Haw oBMv Imvcsm ““ng Inga uggmz mo nonum Eggg mimmuzv memucnua no muoumoeucH may no mfimsamca uouoma smoumsnnmcpu.< NIH magma 246 5 3 289mb mums» 33 m5 ._o.N_A Rim cwcz . 3:388 + manmflgv .H xfiucwmg mam .muoumoflg m5 m0 >33on 0 uom .38 93mm? 3 mac Sflggam mmmo mo ugmwu wfi mum mmawm z aflawwcpm ooo.m ooo.s mafia momé mmoé mm 32m ooo.~.~ ooo.o mmmé 3min omm. mm 3793 oom.m ooo.o nmvé v2.4 mmmé mm E 80.3 ooo.o Hmo.m mmmd Hmo.m am 3,59 800m gm: 800m 553:: 3wxm :oUmfi/mo fingcmum chz z uoumoflflH mamnzv 95952 no $8865 m5 “8 moflmflfim gm TH 22E. 247 8H. mnmoé 3mm... mm. 33.5 mmmJ mmmm. Q¢Hm.l mm. 393 th.I moafia ommo. mm. 5 mm... @264 mmov. mm. 3759 mwam BBCmmgfi. mmuoom ogOMmcm‘fi. mo wwuoom @8985 ucmumcoo HBBHUCH m0 3wxm :oflumfi>mo cumncmum “05% 6. £53 mIH mafia 248 .uoumog 8§mmmu 6 mm E8 on EKG ma Sag ma n .H xfig 8m .muoumoflx: m5 m0 bmwmgm m Mom .3382 SM How 8.80m Gnu—50mg :0 memn mum maimed. mg £38an sumo mo 3% 393m $5 How mlH man—fl. 8mm :5. u H93 ESEBBQ s M H mm mmmf ANH; on. mm. SH; 2.. 372m mom... 2:; mo. Hm. SH; mm. 32¢ woof “ma; ow. mm. 898 284 g mmmé 3H; mu. mu. 8H; mm. 359 ucmfloflwmoo 38.8 mo nonuw “Mafia—m 38m can? noumofivcH 988 .833 “ohm .Bmv uvcmum Egg ngmg naggmmflz mo nouum 8228535 Mammy: gmzmz mo muoumog wfi co 3%:ng wouomm bougwmcoo d vIH manna. APPENDIX J Codebook for x2 - x5 249 APPENDIX J Codebook for x - x 2 5 2 and x3 from Duncan (1961). For x the occupational prestige score is derived The Siegel NORC scale was used. For x4 and x5 the following scheme: the participants' responses were coded using EDUCATION Missing = xx None = 00 Kindergarten = 01 lst Grade = 02 2nd Grade = 03 3rd Grade = 04 4th Grade = 05 5th Grade = 06 6th Grade = 07 7th Grade = 08 8th Grade = 09 9th Grade = 10 10th Grade = 11 11th Grade = 12 B.S. Grad. 12th Grade = 13 1 Yr. Coll. = 14 2 Yr. Coll. = 15 3 Yr. Coll. = 16 4 Yr. Coll. = 17 2 Yr. Degree = 18 4 Yr. Degree = 19 MA, MBA, MFA, MS = 20 PHD = 21 MD = 22 DDS = 23 JD = 24 lst Yr. Adv. Degree = 25 2nd Yr. Adv. Degree = 26 3rd Yr. Adv. Degree = 27 4th Yr. Adv. Degree = 28 BIBLIOGRAPHY BIBLIOGRAPHY Adorno, T.W., Frenkel-Brunswick, E., Levinson, D. and Sanford, R.N. 1950 The authoritarian personality. New York: Harper and Row. Anderson, N.H. 1971 Integration theory and attitude change. Psycho- logical Review, 78:171-206. Asch, 5.3. 1952 Social psychology. New York: Prentice Hall. Attneave, F. 1959 Applications of information theory to psychology. New York: Holt, Rinehart and Winston. Baird, J.C., Lewis, C. and Romer, D. 1970 Relative frequencies of numerical responses in ratio estimation. Perception and Psychophysics, 8:352-362. Baird, J.C. and Noma, E. 1978 Fundamentals of scaling and_psychophysics. New York: John Wiley and Sons. Barker, E. N. 1963 Authoritarianism of the political right, center and left. Journal of Social Issues, 19:63-74. Bayley, N. 1940 Factors influencing the growth of intelligence in young children. In G. Whipple (Ed.), Thirty- ninthgyearbook of the national society for the study of education, Vol. III. Bloomington, 111.: Public School Publishing Co. Berglund, B., Berglund, U. and Lindvall, T. 1971 On the principle of odor interaction. Acta Psy- chologica, 35:255-268. 250 251 Bieri, J. 1955 COgnitive complexity--simp1icity and predictive behavior. Journal of Abnormal and Social Psy- chology, 51:263-268. 1961 Complexity--simplicity as a personality variable in cognitive and preferential behavior. In D.W. Fiske and S.R. Maddi (Eds.), Functions of varied experience. Homewood, 111.: Dorsey. 1966 Cognitive complexity and personality development. In O.J. Harvey (Ed.), Experience structure and adaptibility. New York: Springer. 1971 Cognitive structures in personality. In H. M. Schroder and P. Suedfeld (Eds.), Personality theory and information processing. New York: Ronald Press. Bieri, J., Atkins, A.L., Briar, S., Leaman, R.L., Miller, H. and Tripodi, T. 1966 Clinical and socia1_iudgment: The discrimination of behavioral information. New York: Wiley. Brooke, J.B. and MacRae, A.W. 1977 Error patterns in the judgment and production of numerical proportions. Perception and Psycho- physics, 21:336-340. Bruner, J.S., Shapiro, D. and Taguiri, R. 1958 The mean of traits in isolation and in combination. In R. Taguiri and L. Petrullo (Eds.), Person perception and interpersonal behavior. Stanford: Stanford University Press. Brown, D.R. and Datta, L. 1959 Authoritarianism, verbal ability, and response set. Journal of Abnormal and Social Psychology, 58:131-133. Campbell, N.R. 1938 Symposium: Measurement and its importance for philosophy. Proceedings of the Aristotlean Society Supplement, 17:121-142. Campbell, V.N. 1960 Assumed similarity, perceived sociometric balance, and social influence. Unpublished doctoral dissertation, University of Colorado. 252 Carroll, J.D. 1972 Individual differences in multidimensional scaling. In A.K. Romney, R.N. Shepard and 8.8. Nerlove (Eds.), Multidimensional scaling: Theory and applicationsITVOlume 1). New York: Seminar Press. Chapman, L.J. and Bock, R.D. 1958 Components of variance due to acquiescence and content in the F-scale measure of authoritarian- ism. Psychological Bulletin, 55:328-333. Chapman, L.J. and Campbell, D.T. 1959 The effect of acquiencence response-set upon relationships among F-scale, ethnocentrism, and intelligence. Sociometry, 22:153-161. Christie, R. 1954 Authoritarianism re-examined. In R. Christie and M. Jahoda (Eds.), Studies in the scopgand method of "The Authoritarian Personality." New York: Free Press. Clayton, M.B. and Jackson, D.N. 1961 Equivalence range, acquiescence, and over- generalization. Educational and Psychological Measurement, 21:371-382. Cohn, T.S. 1953 The relation of the F-scale to a response set to answer positively. American Psychologist, 8:335. 1956 Relation of the F-scale to a response set to answer positively. Journal of Social Psychology, 44:129-133. Couch, H. and Keniston, K. 1960 Yeasayers and naysayers: Agreeing response set as a personality variable. Journal of Abnormal and Social Psychology, 60:151-174. Crockett, W.H. 1965 Cognitive complexity and impression formation. In B.A. Maher (Ed.), Progress in experimental personality research (volume 1). New York: Academic Press. Cronbach, L.J. 1955 Processes affecting scores on "understanding of others," and "assumed similarity." Psyghological Bulletin, 52:117-193. 253 Curtis, D.W. 1970 Magnitude estimations and category judgments of brightness and brightness intervals: A two- stage interpretation. Journal of Experimental Psychology, 83:201-208. Curtis, D.W., Attneave, F. and Harrington, T.L. 1968 A test of a two-stage model of magnitude judgment. Perception and Psychophysics, 31:25-31. Curtis, D.W. and Rule, S.J. 1972 Magnitude judgments of brightness and brightness difference as a function of background reflectance. Journal of Experimental Psychology, 95:215-222. Danowski, J.A., Stoyanoff, N.J. and Barnett, G.A. 1976 A simple-minded view of television: A look at users' cognitive complexity. Paper presented to the Annual Convention of the International Com- munication Association, Berlin, West Germany. DiRenzo, G.J. 1968 Dogmatism and presidential preferences in the 1964 elections. Psycholggical Reports, 22:1197-1202. Duncan, O.D. 1961 A socioeconomic index for all occupations. Pp. 109-138 and appendix in A.J. Reiss, Jr., Occupations and social status. New York: The Free Press. Duncan, O.D., Featherman, D.L. and Duncan, B. 1972 Socioeconomic background and achievement. New York: Seminar Press. Duncan, O.D., Haller, A.O. and Portes, A. 1968 Peer influences on aspirations: A reinterpreta- tion. American Journal of Sociology, 74:119-137. Ekman, G., Hosman, B., Lindman, R., Ljungberg, L. and Akesson, C.A. 1968 Interindividual differences in scaling performance. Perceptual and Motor Skills, 26:815-823. Elder, G.H., Jr. 1968 Adolescent socialization and development. In E.F. Borgatta and W.W. Lambert (Eds.), Handbook of personality theory and research. Chicago: Rand McNally & Co. 254 Elrado, R., Bradley, R. and Caldwell, B.M. 1975 The relation of infants' home environments to mental test performance from six to thirty-six months: A longitudinal analysis. Child Development, 46:71-76. 1976 The relation of infants' home environments to mental test performance at fifty-four months: A follow-up study. Child Development, 47:1172-1174. Essen, J., Foggleman, K. and Head, J. 1978 Childhood housing experiences and school attain- ment. Child: Care, Health and Development, 4: 41-58. Feather, N.T. 1969 Preference for information in relation to con- sistency, novelty, intolerance of ambiguity, and dogmatism. Australian Journal of Psychology, 21:235-249. Pestinger, L. 1957 A theory of cognitive dissonance. Evanston: Row, Peterson and Co. Fink, E.L. 1980 Unobserved variables within structural equation models. In P. Monge and J.N. Cappella (Eds.), Multivariate techniques in human communication research. New York: Academic Press. Fink, E.L. and Mabee, T.I. 1978 Linear equations and nonlinear estimation: A lesson from a nonrecursive example. Sociological Methods and Research, 7:107-120. Frenkel-Brunswick, E. 1949 Intolerance of ambiguity as an emotional and perceptual personality variable. Journal of Personality, 18:108-143. Gibson, R.H. and Tomko, D.L. 1972 The relation between category and magnitude esti- mates of tactile intensity. Perception and Psy- chophysics, 12:135-138. Goldberger, A.S. 1964 Econometric theory. New York: Wiley. Goldstein, K.M. and Blackman, S. 1978 Cognitive style. New York: John Wiley and Sons. 255 Granberg, D. and Corrigan, G. 1972 Guyton, 1971 Haller, 1973 Haller, 1969 Authoritarianism, dogmatism and orientations to- ward the Vietnam War. Sociometry, 35:468-476. A.C. Textbook of medicalyphysiology (4th. ed.). Philadelphia: Saunders Co. A.O. and Portes, A. Status attainment processes. Sociology of Educa- tion, 46:51-91. A.O. and Woelfel, J., with Fink, E.L. ‘ The Wisconsin significant other battery. Madison, f Wisconsin: University of Wisconsin, Department ' of Rural Sociology. «cm Hammond, K.R. 1972 Inductive knowing. In J.R. Royce and W.N. Rozeboom (Eds.), The psychology of knowing. London: Gordon and Breach. Hanushek, B.A. and Jackson, J.E. 1977 Statistical methods for social scientists. New York: Academic Press. Harris, J.D. 1952 Pitch discrimination. Journal of the Acoustical Society of America, 24:750-755. Harvey, O.J. 1966 System structure, flexibility and creativity. In O.J. Harvey (Ed.), Experience, structure and adaptability. New York: Springer. Harvey, O.J., Hunt, D.R. and Schroder, H.M. 1961 Conceptual systems and personality organization. New York: Wiley. Hebb, D.O. 1966 The organization of behavior: A neuropsycho- logical theory. New York: Wiley. Heider, F. 1946 Attitudes and cognitive organization. Journal of Psychology, 21:107-112. 1958 The psychology of interpersonal relations. New York: Wiley. 256 Hession, E. and McCarthy, E. 1975 Human performance in assessing subjective prob- ability distribution. Irish Journal of Psychology, 3:31-46. Joreskog, K.G. 1970 1974 A general method for the analysis of covariance structures. Biometrika, 1970, 57:237-251. Analyzing psychological data by structural analy- sis of covariance matrices. In D.H. Krantz, R.C. Atkinson, R.D. Luce and P. Suppes (Eds.), Cog- temporary developments in mathematicalgpsychology Volume II: Measurement, psychophysiCs and neural information processing. San Francisco: W.H. Freeman Co. Joreskog, K.G. nd Sorbom, D. 1978 LISREL IV: Analysis of linear structural rela- tionships by the method of maximum likelihood. Chicago: National Educational Resources Inc. Karabenick, S.A. and Wilson, E. 1969 Dogmatism among war hawks and peace doves. Psy- chological Rgports, 25:419-422. Kelly, G.A. 1955 The psychology of personal constructs, volume 1: A theory of personality. New York: W.W. Norton. Kent, E.W. 1981 The brains of men and machines. New York: McGraw-Hill. Kerlinger, F.N. and Pedhazur, E.J. 1973 Multiple regression in behavioral research. New York: Holt, Rinehart, and Winston. Kirtley, D. and Harkless, R. 1969 Kmenta, 1971 Some personality and attitudinal correlates of dogmatism. Psychological Reports, 24:851-854. J. Elements of econometrics. New York: The MacMillan Co. Korn, H.A. and Giddan, N.S. 1964 Scoring methods and construct validity of the dogmatism scale. Educational and Psychological Measurement, 24:867-874. 257 Kretch, D. and Crutchfield, R.S. 1948 Theory and_prob1ems of social psychology. New York: McGraw-Hill. Larsen, K. 1969 Authoritarianism, hawkishness and attitude change as related to high and low status communications. Perceptual and Motor Skills, 28:114 Lawson, C.A. 1967 Brain mechanisms and human learning. Boston: Houghton Mifflin Co. Levin, H. and Fleischmann, B. 1968 Childhood socialization. In E.F. Borgatta and W.W. Lambert (Eds.), Handbook of personality theory and research. Chicago: Rand McNally & Co. Levinson, D.J. 1949 An approach to the theory and measurement of ethnocentric ideology. Journal of Psychology, Levinson, D.J. and Sanford, R.N. 1944 A scale for the measurement of anti-semitism. Journal of Psychology, 17:339-370. Lewin, K. 1935 A dynamic theory of_personality. New York: McGraw-Hill. 1936 Principles of topological psychology. New York: McGraw-Hill. 1948 Resolving social conflicts. New York: Harper. 1951 Field theory in social science. NeW‘York: Harper. MacDonald, A.P. 1970 Revised scale for ambiguity intolerance: Reli- ability and validity. Psychological Reports, 26:791-798. McGuire, W.J. 1960 Cognitive consistency and attitude change. Journal of Abnormal and Social Psychology, 60: 345-353. 258 McLeod, J.M. and O'Keefe, Jr., G.J. 1972 The socialization perspective and communication behavior. In F.G. Kline and P.J. Tichenor (Eds.), Current perspectives in mass communication research. Beverly Hills: Sage Publications. Messick, S. and Fredericksen, N. 1958 Ability, acquiescence, and authoritarianism. Psychological Rsports, 4:687-697. Miller, A.G. and Harvey, O.J. 1973 Effects of concreteness-abstractness and anxiety on intellectual and motor performance. Journal of Consulting and Clinical Psychology, 40:444-450. Murphy, G. 1947 Personality: A biosocial approach to origins and structure. New York: Harper. Newcomb, T.N. 1953 An approach to the study of communicative acts. Psychological Review, 60:393-404. Nidorf, L.J. and Argabrite, A.H. 1970 Cognitive complexity and the tendency to make extreme judgments. Percepsual and Motor Skills, 31:478. Nie, N.H., Hull, C.H., Jenkins, J.G., Steinbrenner, K., and Brent, D.H. 1970 Statistical package for the social sciences (2nd ed.). New York: McGraw-Hill. Nuttal, E.V., Nuttal, R.L. Polit, D. and Hunter, J.B. 1976 The effects of family size, birth order, sibling separation, and crowding on the academic achieve- ment of boys and girls. American Educational Research Journal, 13:217-223. Ono, H. 1967 Difference threshold for stimulus length under simultaneous and nonsimultaneous viewing con- ditions. Perception and Psychophysics, 2:201-207. Osgood, E.E., and Tannenbaum, P.H. 1955 The principles of congruity in the prediction of attitude change. Psychological Review, 62:42-55. Parrott, G. and Brown, L. 1972 Political bias in the Rokeach dogmatism scale. Psychological Reports, 30:805-806. 259 Peabody, D. 1961 Attitude content and agreement set in scales of authoritarianism, dogmatism, anti-semitism, and economic conservatism. Journal of Abnormal and Social Psychology, 63:1-11. Piaget, J. 1926 The language and thought of the child. New York: Harcourt Brace. 1932 The moral judgment of the child. London: Routledge and Kegan Paul. 1952 The origins of intelligence in children. New York: International Press. 1954 The construction of reality in the child. New York: Basic Books. Pribram, K.H. 1969 On the biologysof learning. New York: Harcourt, Brace and World Inc. Restak, R.M. 1979 The brain: The last frontier. New York: Warner Books. Roberts, A.H. 1962 Intra-test variability as a measure of generalized response set. Psychological Rsports, 11:793-799. Rokeach, M. 1951 A method for studying individual differences in "narrow-mindedness." Journal of Personality, 20:219-233. 1951 Narrow-mindedness and personality. Journal of Personality, 20:234-251. 1954 The nature and meaning of dogmatism. Psycho- logical Review, 61:194-204. 1956 Political and religious dogmatism: An alternative to the authoritarian personality. Psychological Monographs, 70 (18, Whole No. 425). 1960 The open and closed mind. New York: Basic Books. Rule, B.G. and Hewitt, D. 1970 Factor structure of anti-semitism, self-concept and cognitive structure. Personality, 1:319-332. 260 Rule, S.J. 1971 Discriminability scales of number for multiple and fractional estimates. Acta Psychological, 35:328-333. Rule, S.J. and Curtis, D.W. 1973 Conjoint scaling of subjective number and weight. Journal of Experimental Psychology, 86:343-349. Rule, S.J., Curtis, D.W. and Markley, R.P. 1970 Input and output transformations from magnitude estimation. Journal of Experimental Psychology, 86:343-349. Rule, S.J., Laye, R.C. and Curtis, D.W. 1974 Magnitude judgments and difference judgments of lightness and darkness: A two-stage analysis. Journal of Experimental Psychology, 103:1108-1114. Russell, B. 1938 Principles of mathematics (2nd ed.). New York: W.W. Norton. Sawatzky, D.D. and Zingle, H.W. 1971 Level of conceptual functioning and polarization of judgments. Perceptual and Motor Skills, 33:358. Schneider, B.S., Parker, 8. Valenti, M., Farrell, G. and Kanow, G. 1978 Response bias in category and magnitude estimation of difference and similarity for loudness and pitch. Journal of Experimental Psychology, 4:483-496. Schroder, H.M., Driver, M.J. and Streufert, S. 1967 Human information_processing. New York: Holt, Rinehart and Winston. Schroder, H.M. and Streufert, S. 1962 The measurement of four systems of personality structure varying in level of abstractness (sentence completion method). Technical Report No 11, ONR. Princeton: Princeton University Press. Scott, W.A., Osgood, D.W. and Peterson, C. 1979 Cognitive structure: Theory and measurement of individual differences. Washington, D.C.: V.H. Winston and Sons. Sewell, W.H. and Shah, V. 1967 Socioeconomic status, intelligence and the attain- ment of higher education. Sociology of Education, 40:1-23. 261 Sheikh, A.A. 1968 Stereotyping in interpersonal perception and inter- correlation between some attitude measures. Journal of Social Psychology, 76:175-179. Sherif, C., Sherif, M. and Nebergall, R. 1965 Attitude and attitude change. Philadelphia: W.B. Saunders Co. Shils, E.A. 1954 Authoritarianism: "Right and left." In R. Christie and M. Jahoda (Eds.), Studies in the scope and method of "the authoritarian personality." New York: Free Press. Shower, B.G. and Biddulph, R. 1931 Differential pitch intensity of the ear. Journal of the Acoustical Society of America, 3:275-287. Steininger, M.P., Durso, B.E. and Pasquariello, C. 1972 Dogmatism and attitudes. Psychological Reports, 30:151-157. Steininger, M. and Lesser, H. 1974 Dogmatism, dogmatism factors, and liberalism- conservatism. Psychological Reports, 35:15-21. Stevens, 8.8. 1951 Mathematics, measurement, and psychophysics. In 5.8. Stevens (Ed.), Handbook of experimental psyshology. New York: Wiley. 1966 Matching functions between loudness and ten other continua. Perception and Psychophysics, 1:5-8. 1971 Issues in psychophysical measurement. Psycholog- ical Review, 78:426-450. Stimpson, D.V. and D'Alo, J. 1974 Dogmatism, attitude extremity, and attitude intensity as determinants of perceptual displace- ment. Journal of Psychology, 86:87-91. Stoyanoff, N.J. 1979 A theory of information processing. Unpublished manuscript, Department of Communication, Michigan State University. 262 Stoyanoff, N.J. 1980 A brief note on the application of metric multi- dimensional scaling analysis to the study of cognitive structure/complexity. Paper presented at MMDS TRES, an international conference on metric multidimensional scaling, at the Annual Convention of the International Communication Association in Acapulco, Mexico. Streufert, S. and Schroder, H.M. 1965 Conceptual structure, environmental complexity and task performance. Journal of Experimental Research in Personality, 1:132-137. Streufert, S. and Streufert, S. 1978 Behavior in the complex environment. Washington, D.C.: V.H. Winstons and Sons. Suedfeld, P. 1964 Conceptual structure and subjective stress in sensory deprivation. Perceptual and Motor Skills, 19:896-898. Supnick, J.J. 1964 An examination of change in categorization of others following a college course in personality development. Unpublished senior honors thesis, Clark University. Thompson, R.C. and Michel, J.B. 1972 Measuring authoritarianism: A comparison of the F and D scales. Journal of Personality, 40:180- 190. Torgerson, W. 1958 Theory and methods of scaling. New York: Wiley. Travers, R.M.W. 1970 Man's information system: A primer for media specialists and educational technologists. Scranton: Chandler Publishing Co. Tuckman, B.W. 1962 Integrative complexity: Its measurement and relation to creativity. Educational and Psycho- logical measurement, 26:369-382. Wachs, T.D., Uzgiris, I.C. and Hunt, J. McV. 1971 Cognitive development in infants of different age levels and from different environmental back- grounds: An explanatory investigation. Merrill Palmer Review, 17:283-317. 263 Walberg, H.J. and Majoribanks, K. 1976 Family environment and cognitive development: Twelve analytic models. Review of Educational Research, 46:527-551. Ward, L.M. 1972 Category judgments of loudness in the absence of experimenter-induced identification function. Journal of Experimental Psychology, 94:179-184. Warner, W.L. 1960 Social class in America. New York: Harper & Row. Werner, H. 1957 Comparitive psychology of mental development (3rd ed.). New York: International Universities Press. Wexler, K.N. and Romney, A.K. 1972 Individual variations in cognitive structures. In A.K. Romney, R.N. Shepard, and 8.3. Nerlove (Eds.), Multidimensional scaling: Theory and applications (volume 2). New York: Seminar Press. Wheaton, B. Muthén, B., Alwin, D.F., and Summers, G.F. 1977 Assessing reliability and stability in panel models. In D.R. Heise (Ed.), Sociological methodology. San Francisco: Jossey-Bass. White, E.J. and Alter, R.D. 1965 Dogmatism, authoritarianism and contrast effects in judgment. Perceptual and Motor Skills, 20: 99-101. White, E.J. and Harvey, O.J. 1965 Effects of personality and own stand on judgment and production of statements about a central issue. Journal of Experimental Social Psychology, 1:334- 347. wOelfel, J. 1974 A theory of occupational choice. Unpublished manuscript, Department of Communication, Michigan State University. Woelfel, J. and Fink, E.L. 1980 The measurement of communicationsprocess: Galileo theory_and method. New York: Academic Press. Woelfel, J.C., Woelfel, J.D. and Woelfel, M.L. 1977 Standardized versus unstandardized data matrices: which type is more apprOpriate for factor analysis. Paper presented at the convention of the American Sociological Association, at Chicago. Wolf, R. 1964 Wright, 1934 Wyer, R. 1964 264 The measurement of environments. In A. Anastasi (Ed.), Testing problems in perspective. Washington D.C.: American Council on Education. S. The method of path coefficients. Annals of Mathe- matical Statistics, 5:161-215. S. Assessment and correlates of cognitive differ- entiation and integration. Journal of Personality, 32:395-509. Yum, J.O. 1979 Zajonc, 1960 1968 Zippel, 1966 The communicationepatterns and information acqui- sition of Korean immigrants in Hawaii. Doctoral diSsertation, University of Southern California. R. The process of cognitive tuning in communication. Journal of Abnormal and Social PsycholoQYr 61: 159-164. Cognitive theories in social psychology. In G. Lindzey and E. Aronson (Eds.), Handbook of social psychology (volume 1, 2nd ed.). Reading, Mass.: Addison-Wesley. B. and Norman, R.D. Party switching, authoritarianism, and dogmatism in the 1964 election. Psychological Reports, 19:667-670. Zuckerman, M., Norton, J. and Sprague, D.S. 1958 Acquiescence and extreme sets and their role in tests of authoritarianism and parental attitudes. Psychiatric Research Reports, 10:28-45.