O .7 -. .r . _. . . .- .--- ...- Ion...II..I;I.h.MI.I. rHIHII1.m~m...[.I .. .. . , II... .I. ....... ._-.I .F... . . . , .....I. ..1. Start? . ... I - I . I I. . .. . ..h . . I l . n . I I I. O I I. O.’ I III. 'Io II. o I. II I I I . I I 9 n .I ' I. I....I.?I 5‘“. AorVI‘bIVO ) . - . . _ . .... ..-... .. £3315 .whnfié aim.” - . . - .. . , .. _ _ . . _ . . . -. .- . L35... .munhgbm dawn? . I "i .0 b . I I I . 1 . 6010.03. LIIIOTIPA' .19. o I A: u 1. 0 I I . . . . . _ . I . I I .I rvmw'I; ..Hu-THYIA r14... . II- I J" Iv . 1. . I . I Iowv I ’I -’t..'.' " 4'! I .. I‘ I I I +ng .... ..II If ;;!::jo_,' .l ' I ,u mans: 'III' I _ _ - -..— . 3 *v vvv‘fiv.q“.n.~mwnfv_qnol ‘ v - In 9.63.1361 A4525; mmma. Raymmd W Snowman Q 1956 '> £19152 mama gamma- to “F: . : MICHIGAN smire . II. — z. ...I I I I. .v n .I o. I, . . I- I I I . i I .. .1 . II In, I . . . I . . I I. 1 {I "I I I o. ... II IIINI gull-I. I I I IIII o 9 a . >0 II”! 3 .. I I I I I‘I I “4 ~ .. I. I. I III . I . II I. I. I .I I u I . I I I . . I I- II . S .. .. O .I.I I I .5 . , - I. , I I o . D ..I r ‘ . I . I.” I.“ ‘ I. O! m. .II . My I I I. . . .I I II n ; . I I I II I A II o . . ' 4 O I I I II . I I II I 0 «I0 I . . IIIFI -4 . o . I. I ..III .‘III I ,i‘ I II I . IIslIII-II I, I I II‘IIIII .Iflr'tull'IlNI‘-' II -- I III I. 'I II 'I. I I-..1I I. :IQI I I I. II . I ‘0' ..'%II. I! -- .. IIIII I II III-.. .IzzI'QIII I II ...II‘ .Io‘IOII 't’IIVIIII.’ - I IIol II.... II IIorIll:Io§“-II’II o'I I I. II -I.-OI.I igiII‘IOna. " I I III I I. II II . C’IIO'f‘ . . IO, .0. ‘3'. 0... ..III I’II’.‘ I. 'II oII I I . ’I. ‘IIIingil . . I 4 ll- , . n- I. IOII.’ III-I. r I.II III! . .IIIIIII :ir’s . . . III III. ..I .IPIIII'I I...II-IIIIO .I . I II. I III .III'IIJ-IIDIO III..."- I ' IIII 5. II III ..‘JI'III . "" I . . I II I..- . If I. (I ..I I%.I I . . ,II I III- I‘IIIIo"! III '6 I I I IIIIIO'III!‘ .I 'I‘Ii". J u I I o. I It. In. 0" I IIIIIII. I ifolngbv. I I I. II II I If arIrIILI'JI I I I I I ......) III-.. III- I I I I.“ I fj!r1 'III II I I I [I II. IIOIII ...I II 10- . III,"I.'IIC"I’II‘ "I‘I I I I I III. I ‘II‘I’IttI I . I Ir-I‘IIIIII ....{IIoI . II I I III '0 I. I}: II .. I III I I 'VIQ'IiIv .‘OIIOI I I I I I- IIIUIOIVIUIE. I I . II I II'IrIIr . I.’l. .IOI. . II VIII... 'I-r I .II‘I’I f0 I. I .1? II.I'IOI IDQ, I I 1“ I III. ’{0 '1""'i‘1 II I. II“.IIIOQII.~. I I I ' ‘III C I‘llIiI n‘ IEIIIUOIIIII I. I... V II iIII Ih ‘IOIII II III II II a -EJ§-II DI II.‘ I. I I I. II.-.\.II".I‘."II ‘. I I IIIIII'II’IOI I 0'0 III I Ia.III YIICO .Ii . I I I: I I Is XIII — I I III I II C’II'!‘ I I .III IIIIIOIIIWII I I I 0": II I I II IIIQ .-IIIIV‘ II I .49; D II III. ’5- A I. III 0 I I I I .I I-.. I I IOI I0 {I II POI. CI ' I II I “"OIO' . ‘- I I I" . II. I I ‘ II II... P.’ I.". O r ‘I 41 I... ‘. I II N I I III.II.I I u I ' II II II}I‘0 (I I l I I I1I‘I in t I I I I I .OI IOI IIogo, .I 0“ I I . II I II III. I. .“I. . I I P I . II III... II I ’II‘ £II. IIIH‘I O \... I | I I I I I, u. .. I. I III III ‘11‘; I. . IlogIOrjg III .0. 9". o I I“! I . . I 0 O I.-. O I I I u 3 I I ‘0. I I I | . u I. I I I . - .I I . I... .. aII I. I . .I _ IN: . . 5...: III II . .I. . I II c‘l- . Q¢I coI. I.II I] - l I I ‘I I III‘...‘OOOOO II I.. . . II C‘I I I V. U. ' II ‘I JQII; ..‘I'I I‘ll- .I ‘ PI. III. II.. ck!” “I- ..g I III II .k‘ u f I -‘ I1 IJI‘IRKI.£O‘ION{X.IJ:§. 0.-..‘yfllls III}. ”Jot. ‘ ‘ I, I ' I, III I.III'I . Olkl. 13 N I I% .. . o. , s0.‘!§flh‘&§opn ikfl‘IIJI". HI I; \l I \I‘ D . )‘I‘I I'll..." I I DI I. . I l- I ‘I I , I CHANGES IN MEAT PURCHASES DURING 1952 AND 1953 RELATED TO FAAILY CHARACzjISTISS By Raymond W; Gieseman A THESIS submitted to the College of Agriculture Michigan State University of Agriculture and Applied Sciences in partial fulfillment of the requirements for the degree of EASTER OF CIENCE Department of Agricultural Economics ACKNOWLEDGLENTS The author wishes to express his gratitude to Dr. Harold M. Riley for his supervision and encouragement in the prepara— tion of this manuscript. Special thanks are also extended to Dr. B. C. French who made helpful suggestions throughout the duration of this study. The author is indebted to Dr. L. L. Beger and the Depart- ment of Agricultural Economics for the financial assistance and valuable training rendered to him during his stay at Michigan State University. The help and cooperation of Mrs. Arlene King and the clerical staff has been a major contribution toward making this study a reality. The painstaking job of typing the final manu- script is the work of Mrs. Shirley Goodwin. Any errors that remain in the final manuscript are the sole reSponsibility of the author. _" ‘L‘L‘L‘L".‘L‘LV.V..‘ IC’A r\ l\ I\ A A l\ I\ I\ A CH.IGES IN MEAT PUXCHASES DURING 1952 AND 1953 RELATED TO FAMILY CHARACTEiISTICS 3:1 Raymond W. Gieseman AN ABSTRACT Submitted to the College of Agriculture Michigan State University of Agriculture and Applied Sciences in partial fulfillment of the requirements for the decree of EASTER OF SC IEI-ICE Department of Agricultural Economics Year 1956 Approved. WM 43% y ABSTEACT The purpose of this study was to determine whether consumer reSponse to price changes for selected cuts of red meats were related to family characteristics such as levels of income, age and education of the home- maker, and size of family. The two years, 1952 and 1953 were chosen as the time period for this study. This was a period of substantial change in prices for both beef and pork items. Substantial changes were made by consumers in t.eir purchasing patterns for these cuts. The basic data for this study were obtained from the purchase records of 131 families in the Michigan State University Consumer Panel. Three methods were used in analyzing the data. First, families were grouped according to each characteristic to determine whether families who were different with respect to any given characteriStic were also different in their response to price changes. Secondly, families wer grouped according to the degree of change in meat purchases to determine if these sub—groupings were also different with respect to family characteristics. Finally, multiple regression analysis was used to determine the net effects of family characteristics on consumer respon— siveness to price changes. When the various family characteristics were related to consumer response to price changes for meat items it was found that none of the characteristics were highly significant in determining consumer iv behavior. However, the results obtained were consistent in all three methods of analysis in pointing out the directional relationship between family characteristics and consumer responsiveness to price changes. It was generally found that age of the homemaker was positively related to consumer responses in meat purchases. Education of the homemaker and size of family were usually found to be negatively related to changes in meat purchases. The effects of income measures on response in meat purchases were not established as being important. From the results obtained it was concluded that age of the home- maker may well be the most important factor determining group differ- ences in reSponsiveness to meat price changes. TA3L3 CF ”CV” 38 CHAPTER I II; TiODUUTIOUOOOOOOOOOOO0.00.... Purpose of Study............ Source of Data.............. Nature of Previous Studies.. hethod of Study............. Usefulness of hesults....... II KETHODS CF ANALYSISoooooooooooo Choice of Time Period Studied........ Validity of the Data Used............... Selection of Cuts..................... Theoretical Considerations............ Preparation of Data............................... Interrelatedness of Family Characteristics........ Methods of Analysis.......................... ..... III CHANGES IN BE? W PUiClASSS BDTZDEN 1952 AND 1953 RILATDD TO FAiiILY CE—i‘qpbA'JhL-L 511105.000...O......OOOOOOOOOOOOOOOOOOO IntrOdUCtj—OHOOOOO0.0.0.0000...O......OOOOOOOOOOOOOOOOOO m1 sul Us From Sortinp Families Into Groups According to Family Characteristics......................... General COT-11111911138 O O 0 O O O O O O C C O O O O O O O O O O O C O O I O C O O 0 Age Of Holner\1a:l~er O O O O O O O O O O O C O C C C O O O O O O O O O O O O O O O O C O O Adllcation Of HOIHBW :GTOOOOOOOOOOOCOOO00.0.00... Size OfFa l‘Li-lVOOOOOOOOOO...........IOOOOOOOOOOOOOO ‘ Per Phalm 41y Incoll‘LeOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO Per Capita Income.............................. Families Grouped o;' Degree of Change in Purchases. General Co.Jnents............................... Age of Homemaker...................... Education of Homemaker......................... Size of Family........................ Per Family Income..................... Per Capita Income..................... Results from Regr IV CH GLS IN PO Li. PUL'CE IASBS DUB IN“ Intl‘Oduc—t’ion. O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 0 O O O 0 g Families into Groups According FaI'l:!—l:>r Cilaracue r:LSt1—CSOOOOCOOOOOOO......OOOOOOOOOOOOO Results From Sortin vi 1952 AND 1953 RELATED VATAILY CHAA1A01 £5 m1 0800000000000...0000000000000.0. ession Analysis......... Page *4 C0-\1I\)}—'}—" 10 10 10 1:3 17 18 21 2i; 28 29 29 29 32 3h 37 ;37 U1. ‘3 \H \f‘._ 03 -Q . ‘ . _ - .. _ ‘ r . . - . a > a — i ' , s > ,\ ' 4. . - - . — ‘ ~ a - ~ ' .. . ~ «I - - ~ .. . .. . v ‘ . n . a ’ a . q ~ A n a a . < . - . . - . . ‘ . a . ~ u -\ w ~ w ~ ~ . — .. ‘ w n a » n w , > \ - - . 1 x - n A 4 ‘ a - ‘ A ~ ~. . a .- n .7 . ,7 ~ - c A v i , r » a Q 1 ~ 1 5 . ‘ ~ - . a :- p - q a a 1 a . ~ a ~ c- 1 n - ‘ v I r ‘ a r ' s - I ‘ 'i " n .- . r! ‘ . A n i g . .. 1 u- a . . « a 5 >1 - a . — » . , .. . ’ u x - a 5 ~ 1 \ — ‘ i , \ 1 A , :- ~ . n 1 A - , - a 1 r ‘ '\ ~ - - a - 1 r -. n x y - - A . . . . . - - ‘ a , r a . ‘ . y ‘ -, - r .7 a A q . p A r I- - s s s ‘ V 1 < n r ‘ \ - a. . , - .. . - A n - a - -‘ - - ., q A , ‘ n A ~ . - a . .. . . w 3 \ . - ‘ TABLE or comm-ms - Continued CHAPTER Page Age of Homemaker................................. 58 Education of Homemaker........................... 60 Size of Family................................... 62 Per Family Income................................ 6h Per Capita Income................................ 6U Families Grouped by Degree of Change in Purchases... 67 General Comments................................. 57 Age of Homemaker................................. 5 Education of Homemaker........................... 69 Size of Family................................... 71 Per Family Income................................ 71 Per Capita Income................................ 73 Results from Regression Analysis.................... 73 V SUE-‘H'IAlth Alm COF'ICLUSICIJSOOOO......00.000.00.00...I0.0.0.... 82 BIBLICGRAPPIY.O..........OOOOOOOOOOOOO..........OOOOOOOOOOOOOOOOOO 8? APPEI‘IDICJESOOOCOOOOOOOO...............OOOOOOOOOOOOOOO00.00.0000... 88 vii TABLE III IV VI VII VIII KBTOFTMflHS age Comparison Between the United States, the Hichigan State University Consumer Panel and the Sub-Sample of Panel Families for Selected Measures of Beef Purchases, 1952- 195300000.........OOOOOOOOOOOO00...........OOOOOOOOOOOOOO 11 Comparison Between the United States, the Michigan State University Consumer Panel and the Sub-Sample of Panel Families for Selected Measures of Pork Purchases, 1952— 1953000000.0000000000.....OOOOOOOOOO......COOOOOOOOOOOOOO 13 Comparison of Family Characteristics for the hichigan State University Consumer Panel and a SubeSample of 131 Families Reporting hO Weeks or Kore During Both 1952 and 19S“..................................................... 15 Coefficients of Simple Correlation Between Family Characteristics for 131 Families in the Sub-Sample of Panel Data, 1952-1953-000000000000000.coo-000.000.000.000 2h Average Changes in Selected Measures of Beef Purchases Between 1952 and 1953 for Families Grouped According to Age Of Homemalcelaooooooo0......00.0.0.0...OOOOOOOOOOOOOOOO 30 Average Changes in Selected kcasures of Beef Purchases Between 1952-1953 for Families Grouped According to Education Of Homem}(er0.0.0.o............OOOOOOOO-OOOOOOOO 33 Average Changes in Selected Measures of Beef Purchases Between 1952 and 1953 for Families Grouped According to Size Of Fain-Lily.0.000.000.0000...O.OOOOOOOOOOOOOOOOOOOCOOO 36 Average Changes in Selected Measures of Beef Purchases Between 1952 and 1953 for Families Grouped According to Per Faraily InCOFLe, 1953.00.00.000000000000......OOOOOOOOO 38 Average Changes in Selected Measures of Beef Purchases Between 1952 and 1953 for Families Grouped According to Per CED-pita. Income, 195300.00...000............OOOOOOOOOOO 39 Average Age of Homemaker for Families Grouped According to the Degree of Change in Selected Measures of Beef PurChaSCS from 1952 to 1953000000000.00000000000000.0000. hz Viii LIST OF TABLES - Continued TABLE XI XII XIII XIV XV XVI XVII XVIII XIX Page Average Education of Homemaker for Families Grouped According to the Degree of Change in Selected Measures of Beef Purchases From 1952 to 1953...................... Average Size of Family for Families Grouped According to the Degree of Change in Selected heasures of Consumption From 1952 to 1953000000oooooooooooooooooooooooooooooooooo Average Per Family Income in 1953 for Families Grouped According to the Degree of Change in Selected heasures of consumption From 1952 to 1953.000acoooooooooooooooooooooo Average Per Capita Income in 1953 for Families Grouped According to the Degree of Change in Selected Measures of consumption From 1952 to 195300.0000000000000000000000000 Average Changes in Selected Measures of Pork Purchases Between 1952 and 1953 for Families Grouped According to Age Of HomemakerOOOOOOOOO......OOOOOOOOOOCOOOO00......... Average Changes in Selected Measures of Pork Purchases Between 1952 and 1953 for Families Grouped According to Education Of Home!!ml{er..............OCOOOOO0.00.0.0000... Average Changes in Selected Measures of Pork Purchases t a Between 1952 and 1953 for Families Grouped According to Size or E‘an‘iilyOOOOOOO......O...OO0.00000000000.00.0000... Average Changes in Selected Measures of Pork Purchases Between 1952 and 1953 for Families Grouped According to Per Family Income, 1953.................................. Average Changes in Selected Measures of Pork Purchases Between 1952 and 1953 for Families Grouped According to Per Capita Income, 19530000000000.0000...-000.000.000.000 Average Age of Homemaker for Families Grouped According to Degree of Change in Selected Measures of Pork Pur- chases from 1952 to 1953................................. Average Education of Homemaker for Families Grouped According to the Degree of Change in Selected Measures of Pork PurChaseS from 1952 to 19530000000000000000.0000...o U1 \0 61 68 70 LIST OF TABLES - Continued TABLE XXII XXIV Average Size of Family for Families Grouped According to the Degree of Change in Selected Measures of Change in Pork Purchases from 1952 to 1953......................... Average Per Family Income in 1953 for Families Grouped According to the Degree of Change in Selected Measures of Pork Purchases from 1952 to 1953......................... Average Per Capita Income in 1953 for Families Grouped According to the Degree of Change in Selected Measures of Pork Purchases from 1952 to 1953......................... 72 75 Purpose of Study The purpose of this study is to determine whether consumer resuonse to price changes for selected cuts of red meat is significantly related to family characteristics such as level of income, size of fami y, age ('1 and education of the homemaker. Lhe results of the study will provide evidence as to whether price elasticity of demand for meats varies gong economic sub-groupings of families. The basic hypothesis is that families grouped according to family characteristics differ significantly in their response to price changes for beef and pork items. An alternative hypothesis would suggest that families grouped according to the degree of change in er capita consump— tion and expenditure, for each of the items, are significantly different with respect to the family characteristics studied. The second hypothesis is the reverse of the first. Throughout the empirical testing of these hypotheses it is assumed that all families are subjected to the same price changes and that consumers have equal exposure to advertising and other merchandising practices. Source of Data T‘ o' ‘ L. c 44 'n ‘b ~—— . c 1 t ° (Ad I? ,. 41;; 9° }§-v 5+ 13 he basic data lor leo 5 any were 00 aine rom ens nicaigan Va e University Consumer Panel, a diary-type panel in continuous operation since early 1951.1 This Panel consists of about 250 families, repre- sentative of the city of Lansing, hichigan. The Panel is unique in that it provides weekly information on consumer food purchases which . 2 are suited to both time series and cross—sectional analysis. Consumer panel data has been found to be quite useful in estimating the short run price elasticities of demand for meat items. However, for these estimates to be reliable as a means of predicting behavior in geographic areas other than that from which the basic data were obtained, it is felt that information is needed as to the responsiveness of different groups of consumers to price changes. The selection and analy- sis of consumer panel records for families who reported meat purchases during a period of time when substantial price changes occurred seems to be an appropriate way to handle this problem. Nature of Previous Studies Several studies have been made which attempt to measure the price elasticity of demand for various meat items. These studies are generally time series analysis based on annual data for the entire United States for a period beginning in the early 1920's and extending up to the early 1950's, the war years generally being excluded. Some of the more 1 The organization and operation of the LSU Consumer Panel is under the direction of Dr. G. G. Quackenbush and Dr. J. D. Shaffe 21.. ror further information on the objectives and usefulness of this Panel, ee: G. G. Quackenbush, "Demand Analysis From the HSC Consumer ‘. t": o v "\ f a: '3 ’ "’ i '71 Panel," Journal of farm economics, Vol. go, no. ;, 195h, pp. hlj—d2(. \JJ widely known studies typifying this line 0* work include studies by 3 Fox and working. Fox has fitted single equations by the least squares m:%tliod in estimating price elasticities of demand for beef, pork, and all red meats. Thes eresults were obtained by expressing the average annual retail price as a function of the quantity of meat consumed per person and disposable income. The study is based on aggreg ate data for the period 1922-lyul. In a somewhat similar manner, werking has derived estimates of both price and income elasticities of dweland for beef and pork as well as all red meats. The actual technique for fitting single equations so as to obtain "unbiased" estimates of various demand parameters differs somewhat from those used by F x. In addition to estimating demand elasticities, based on annual adjustments in consump- tion, working also attempts to establish a longer run price elasticity of demand using a five-year lagged relationship between price and quantity. A recent study by Shepherd et al. uses similar data as that used by 5 Fox and working to point out shifts in demand between beef and pork. 3 Karl A. Fox, The Aialrs ’ Diiand for Farm Products, Tech. Bul. 1081, U. S. Department 0: A cultu ,Nashington, D. C., 83p ten ber, 19:” J). 4 lmer J. Werking, Demand for Leat, Institute oi‘ neat Pa cking, University of Chicago, Chicago, Illinois, l,3h. LU 5 C) eofirey Shepherd, J. C. Purcell, and L. V. n 1do rscheid, Econ01ic AnalZSl ' of Trex ids in Bee: Cattle and Hoe Prices, Res. Bul. hCS,- Iowa State Collece, Ames, Iowa, January,19§h. U In an earlier study, Shep133d made som3 estinat3s of price elasticities for all meat based on national average per capita consumption and United 0 O 6 States avera 33 retail neat prices. T113 studies re:3rred to above are useful insofar as they are able 3 the slope of the 3mand curve for riven COEL Odit3~ cl‘ to estirrr.~ HOJever, as Foote has pointed out, sev31al qu; st ion 8 in r35ard to pr ice elasticities of demand derived i rom conventional tt #3 series analysis a V V O 7 have been raised. Among tne more important proolems are: (1) Do demand functions differ among various economic groups? (2) How do ’1 "I demand func ions chan'e over time? (3) Do elasticity coeiiicients vary with the level of economic act:Lvity? Foote concludes that because conventional time 3ries estixates are based on aggregate market data, little information for answering these questions is provided by such 8 analyses. Studies such as have been cited earlier are of lixite d use in esti— mating shorter run demand paramet tors. Based on aggregate data over a period of twenty to thirty years they do not provide for the analysis of adjustments within a snort per iod of ime, particularly intra-year adjustments. 6Geoffrey Shepherd, ChaELZ§.i§ the Demand for heat and Daigy Md cts in the Uni ucd Sta tcs Since 1313, 338. Bul. 358, Iova State College, mes, Iowa, hovenber, 1949. 7 Richard J. Foote , Price £1 ticiti es oi Demand For Non-durable Goocs, With U3333Sis on Eood, A.S-/'o, U. S. U 3partn3nt of '5r Culture, Washington, U. 0., march, 1953, p. o. a laid. - ne crucial problem of determining how demand functions differ for economic sub-groups as they are distinguished my various socio- econ ric chaiacteristics has also remained untouched by conventional types of analysis. Do low inc me i'anilies r3spond more readily to price changes than comparable fanilies Jith hi3h3r lHLO‘E ? Is th3 size of family significantly related to responsiveness? Certain limitations on the applicability of standard time series estimates of elas ’ ticity have created an interest in th3 use of consumer panel data for estimating price elasticities. Qua ckenous:1 has pointed out the ap alicaoilitr of consuz 3r pan )1 inforn_ation in pr3dictin3 price 9 and cross elastic itics i'o1 major IQ on items. Luznets, who has done considerable work with consumer panel data, has made the following 10 statement aoout its useiulness in demand analysis: 5- L‘ r -1tne iou1 aieas singled out in hr. Foote's pa er as taose in wlfii ch sixiui icant ad_vanc es are b-eing marje, I would be inclined to rate the utilization of data fr on consumer panels as holding the greatest p101 1ise 01 obtaining valid estimates of short-run demand (purchase pacawete3s K znets seems to be in complete aL3re3nent TC th Foote as to the limitations of estimates based on annual data for the aggregate United States market. These and other statements seem to point toward greater use of consumer anel data to promote a more dynamic approach to the analysis of short run demand parameters in riarket analvsis. 9 , 3 G. G. Quackenbush, 92. cit., p. 313. 10 fl _ ' Georb e h. Luznets, UiscuSSion on p 1pm presented by uichaid J. Foote, "Demand and Prices," Journal of Farn_:conomics, Vol. 37, Lo. 2, 1955, p0 2:1"). An extensive study has been undertaken at Michigan State University to estimate price and cross-elasticities of demand for all major cuts 11 of meat from consumer panel data. A series of single equation models have been fitted by least squares in deriving demand elasticities for broad groups of meats as well as for a large number of retail cuts. However, these analyses are based on data for the entire panel and do not relate the elasticity estimates to family characteristics. Since the hichigan State University Cansumer Panel is located in a single city, differences in reSponsiveness of different groups at the retail level become important when attempts are made to apply these estimates to other areas. If family characteristics are related to consumer behavior, then some consideration of this relationship must be made in estimating price responsiveness for other areas whose population makeup may be different from the sample population. The only known attempt to determine if reSponsiveness to price . ' 12 changes is related to family characteristics was a study by Zwick. Using a consumer purchasing panel as the source of information, an attempt was made to analyze the effects of variations in family characteristics on purchase behavior by means of covariance analysis. Seven characteristics, family size, income, education, age, ethnic back— ground, religion and occupational status were used in the study. Age was the only factor found related to consumer responsiveness. 11 u . i - Harold A. Kiley, "Some measurements of Consumer hemand ror heats." Unpublished Ph. D. Thesis, Department of Agricultural jconomics, hichigan State University, 195a. Charles Zwick, "A Quantitative Study of the DLmand for heat," Econometrica, Vol. 2;, ho. 5, (Abstract), pp. g27—gzb. 1 A review of the aoove mentioned studies indicated that there are several available estimates on responsiveness to meat price changes for large aggregate consuming groups. bore recently, consumer panel data has been used to make estimates of short-run demand parameters. However, little empirical evidence is available to determine whether the degree of reSponse to price changes is related to family characteristics such as level of income, size of family, age and education of the homemaker. Lethod of Study There are several approaches that might be used for measuring the extent to which consumer reSponsiveness to price Changes is related to family characteristics. One approach would be to extend previous attempts to estimate price elasticities by sub-grouping families accord- ing to one or more characteristics. It would then be possible to compute a separate time series regression for each of the sub-groups. Such an equation would express quantity as a function of.the price of the item considered, the price of competing products, and other appropriate variables. The regression lines thus obtained would show the responsive— ness of different sub-groups of families to price changes. These regression lines could then be compared and differences in slope could be tested for significance. Such work is being carried out as part of the over-all project under which the meat purchase data from the Michigan State University Consumer Panel are being analyzed. Thus far, however, the only sub-grouping that has been used is a classification of families into three groups based on per capita income. EXperience with this procedure indicates that it may be impractical tO‘ further subdivide the 250 families into groupings involving additional family characteristics. Since this analysis requires rather costly recombinations of data, an attempt should be made to more clearly identify the relevant family characteristics affecting responsiveness to price change before attempting to proceed further with this approach. Another approach to the problem was, therefore, considered desir- able. The first step was to select those families who were in the panel during both 1952 and 1953. These families were then sorted into sub-groups based on family characteristics considered to be important, '1 and the mean chanees in purchases of majo cuts of meat between 1952 and 1933 compared. Using such a method it was possible to test these means to see if consumer responses differed between the sub-groups. This procedure was then followed by a series of regressions in which the change in level of purchases for each item of meat by each family was erpressed as a function of the family characteristics. The regression coefficients indicate which characteristics, if any, are significantly related to change in purchases and the extent of this relationship. This approach, while relatively simple, will explain differences in reSponse between consumers in considerable detail. A more complete account of the methods used and their limitations will be given in later chapters. Usefulness of Results Some of the uses which might be made of the results of this study could be listed as follows: l) The ap roach used in this study, while relatively simple, shows in considerable detail tae relationship between family characteristics and reSponsiveness to meat price changes. This information will be use- ful in determining the extent to which price elasticities of demand derived from the hichigan State University Consumer Panel will accurately predict similar adjustments in other markets where families are systematically dif erent from those in the panel. If reSponsiveness is related to one or more characteristics, the nature and extent cf‘ihe relationship will be useful in formulating additional analysis of Lichiren State University Consumer Panel data. 4) Various trade eroups mirtt well find the results helpful in carrying out pricing and. pro mtional pregrars that will lead to maxiwie zation of profits. 3) Finally, the results attained will add to the fundamental knowledge about the structure of denrfic for meat products, and might well lead to a more neaninvful approach in demand analysis. (a t: I .4, F1 F“! gEPECDS CF Choice oi Time Peri Durin the two yew period, 1932-1? one of the sharpest adjustments in meat recent history. During this period the creased its beeL purchases from 31.5 to pork consumption d.e or eased rom 71.5 to sharp chan5, es in levels of consumption i Sizabl " '3 J ments in livestock production. at all levels of livestock and neat di demand situations changed. The pattern of oeef and pork purcha national market are also revealed in the of the hichigan State UniversitJ Consume W ty of the D .1 -.iCll Table I summarizes the panel 6a J. , bile years 1952-1953. It provides str _L v’l L18. I. , ~ ,. .: .L 1. reader I'LL Lul ATALYSIS od Studied '_I p3, co Runers collectively made purchasing that has occurred in av: rage American iaaily in- ?o.7 pounds per capita while 62.? pounds per capita. These ef lect important cyclical adjust- price aCJusthan s were realize ibution as tie :Uly and ses descritnd above fo th weekly food purchase diaries r Panel. ata Used or beef consumption in the two ! ‘ +.~Q UA.-'J comparison between the total United States m1 ricet, th fiichi gan State Universit, Consuuer Panel, and the sub-sanple of ljl families selected for this study. Thee stihated United States average retail price for beef is based on choice grade retail cuts and is somewhat higher tie an the av er a3e retail price obtained from quantity and expenditure data reported by .ma;hmIHMMmH “.m.hn .copmcflamw3.nmpmmH:oflpm< Mo pcmfippwmma .m .3 «mm .02 moonccm: Hw%deSOHpm« .dnma pow pcmaoagmdm «mmImOmH «mm_w:m mmwwmm.ozp mm.mmoom.wm.COflpmfifimcoo .HmboH mammmaoaz @gp 9w pgmflmz mmmopmo mpweflxonmgdo .|.I .ma .9 «mmma A.0 .Q «copmcflgmmB «mpdesOHpm< mo pamaphmmma .m .3 «map mafimeHapsm HM mdomcmfiamomflz qwmmwu @mm.MMflmmwm.uflwpommmm.wamm .mpSQ moms Hflwpmk mmwpw moflono go ficdoa pmm mcflpm Hampmp mpHmOQEoon .quwm mgp pom mm mmowgm mswm esp saws wmpmopwcoo mum? mammwmIQSm map EH mmflaflawm pmnp @demmm ma pr m.oa m.p w.o ON.H ON.H w.qm H.m m.m mmmm mcflafiom sum mcfismpm w.mm H.HJ O.Hm N.mm NQ.HH Hw.w m.@m O.mH ©.m mgdmpm w.mm ®.mm m.mm @.HH wN.OH Qw.m O.m: O.mH H.MH - mpmmom m. - gm mam 0.3- $5 mde E «a: ma; Eméeam AmPSO JV w.b 3H.Nm Nw.mm W.mm O.mm m.H: Mmmm HQPOB mpwa Hmmmm mo mamgmm-psm b.3H w.w w.@ N.N I Hm. mm. m.mm J.m w.H P.0NI m.wm N.Hm momm wQHHHOm wad mcflzmpm w.mm m.oq w.om m.mm m@.m Fm.w m.nm w.ma m.m w.:Hu w.m> >.mw mxwmpm m.mm w.wm m.mm m.m mp.w mm.w n.am m.ma w.aa H.mmu o.mm ~.mw mpmmom N.NI w.wm H.mm O.wHI wfl.w Mm.m J.HH ©.>H ®.mH J.©0L fi.©: O.m© HmeSQEmm w.m wi.wm mm.0m N.wm N.®: @.Nm w.mHt 0.5m H.HN %m®m HwPOB Hmfimm .D.m.£ m.ma op.©w om.am m.ow- 2H.mo nm.wm wmom proe . mmpMpm vmpflcp mwflmwm fl - m 14.23 my "QMQUIA mhmflnmvmg mmflflflfldfi.fiu-mvcm@’onltmmwzflmmsm .Numswwnmoqv owqmgo cAamflMov A3383! pmmohmm mmma mmaa pcmopwm;su..mmma mmmw psmoaogxxu. mmaa mmmm pgwopom I, mmafi Nmaa pgusom mxmmB pcmopmm magpflvcmmmm mpw4du pmm CoaamHSmmou mpflmdo ham ) flpm depmm mmmmaammma «mmmHns meoa oammo Hos: exp pm oama mQHUSHoKo ;:w: AH hop: 1:-m o .00 «unto: a m> mmmoHuo chiiHoss: o lllll ll. 1" i (WHO, 4.4 .. a..anSoHam< mo .pmom .m .D “Haw .sz .omflA «mpuoo pow muwu:.: m:-wHoxpma atom m Loo Ho osaw> HflwpmH_ommama¢Q .Homdm mapco esp pom mm mmoHHQ osdm one new: cmpsopmcoo who; mademmundm map CH moflaflemm pmnp omEdmmw we pHm I H.0H- two” 1 3% e404... -..: Neva..- Idafilil-.a4?- mantel e. m kl , seesaw m.wHI w.m m.m a.HHn mw.m mm.m m.wan ~.n w.m mpmwom m.w H.e m.e m.m om.a m~.H e.m - m.m a.m moaeoam m.m . H.HH m.m e.oa- ww.n :m.© w.NH- m.w :.m ewe m.mHI m.mm ®.®m N.:cl ©O.m NN.© 0.3m1 ®.© 3.0H mxmmpm one mmogo m.w a w.mm m.©m J.w am.: mH.: m.mau o.~ H.w coomm m.oa- we.am em.mm m.wa- a.mm m.oe Ampeo 0v atom fleece mama Hocmm Ho magsmmunsm a.H - m.ea H.aa m.ma- ro.a ea.H w.aa- o.m e.m a. . H.mm H.3m memmemm H.0m- m.H m.m o.man mo.m mm. mm- H.m w.: o.m m.©m m.mm mammoe m.maa m.m m.: H.@ mn.a ms. w.m a w.m H.m o.£ m.m: :.>: moflsoflm a.s - m.aH a.eH a.m - ss.m mm.m s.aa- .m.a m.e i.a m.as m.qs ewe J.mau @.mm m.mm w.mfiu mq.3 @O.m @.mau H.@ i.w w.m ©.m> ©.®o mxmopm paw mmono e.ma- m.mm o.mm H.m c_-.: o .m H.mau m.e w.e o.mm o.eo w.om eooem ©.NI nw.wfi NN.ON m.mHI :.mm w.3m m.m N.m© m.®m xHom Hm+ OB Hmcwm .D.m.m N.NHI om.m© 00.HN 5.0H Qm.m© Qm.wm xhom HGQOB . mmpmem empaee nrznrmualzaldflw . mKMWmelngHmHHOU ldmwmmflmmqlrot.co£o Amtcsomqldmmmmwmq! mmmwso Ampcmmwfthmmuov peootmm mum mmaa mmmmmmm- mnoa osmotom mama Jmm ma pemoamm mMmH mmaa pgudom waooz psoopom mpopfiomo memo pom Immmmmmmmmdowprm301rcli. woewm Hempox mo magamlmbm SE. 9.? Hwiém od...5m Bu MBHWHQHZD ”agatm VQOHEOHZ nuke a._maummma mmmea:usem exam mo maeemea. . ameomqum mom mquHeaa amaem HH mgm¢ Per U9pitn.(pounds) Quantity “.1: (A; the restriction that only those 1a lilies be considered tho were in the “(W panel at least a0 13 3;:s in both years. Th3re were 1,0 families who met this requirement but s3ven were eliminated due to a lack of information. The data for tne renaining lfil families were then placed on summary cards especially designed for the study (See Appendix A). The five family characteristics needed for the studv, size of fanily, age and education of homemaker, per capita income ard per family income were summarized for each fa:nily and plac 3d at the top of t1_e summary card. Information pertai nin3 to the four cuts of beef and six of po~1c were obtained from panel records. Three measures of purchasing activity were considered, total quantity purclias-ed, total expenditures, and the p3rcent w331-cs oou"b It. T13 quantity and exp3nCiture figures were ' 1) I adjusted for all families who were in the panel less than 52 weeks in either year so that all purchase data was on an annual be sis. Per capita quantity and expenditure figures for each family were then obtained by dividing the adjusted total figures by the number of persons H. n tile fax dly. In twelve instances tm nunoer of persons in the family was adjust ted wh an a new child increased the family size dur ing the ry little (D period studied. Since children under two years of age he ve effect on the consumption of retail cuts of me at, it was felt that these adjustments were ne eded to insui e reliable measures of per capita consumption. es were obtained by dividing the The percent wee1:s bought figur number of weeks any given item was purchased by th3 number of wecks the family participated in the panel. -hei euaining columns on the summary card were used to inoi cate the changes that occurred in the three n1easures of activity oetween the two vears. This data oecame the primary source of information for the ana grsis of consumer re esponsive- ness to price changes. Interrelatedness of Fenilz Characteristics In errelationships betwee1 ianily characteristics are an inportant consideration in stueging the effects of these characteristics on responsiveness to price changes. They are important not only from the standpoint of correct interpretation of the results obtained from simple mean computations families are grouped according to characteristics, but also in terms of formulating and evaluating the multiple regression results. The scatter diagrams in Figures 2 and 3 provide some indi- cation as to the extent of interrelationships between characteristics. The scatter diagra1n for age of homer1alzer and family incore shown in rigure 2, suggests that tl1e level of income rises as age of homemaker increases from young to middle age, tut in the older age range there is a definite tendency for incomes to decrease as age increaSes. Parts B and C of the same figure show that education of honemabers and size of family were negatively correlated with age. The size of families and age of homemakers were definite”y correlated as a .riori reasoning (3;, would have sugges*§ e th n education and size of iamily were plotted avainst Tanily incomes as in Eiiure 3, Parts A an o B, there was general indication of v * ~ .0 :- a positive relationship. Ec- Mt- on anc Size oi iamily are shown in E13 INCOME PER FAMILY EwCATION (F HOMEUAKER SIZE OF FMILY 1 4o 1 50 AGE OF HOMEMAKER .. 1 l _;. 1 .. ’ \J ' - .... 90>— . a 80— . . . . ' 70— 1 ‘ I 60— . ' . ,, . 50— z... z . . o O .. I I o . . ' .. . .. .0 o .. o. . . . 4°" " - ° ' - .: . w__ o o '. . 20.. ' Im— - , - l I J l (1% 30 4o 50 616 70 AGE OF HOMEMAKER .1 I .— 3 I 'qb o '21—- : :Vg;.:. 30.33;; a .00.... .00. 3°: 0 0 '0L— oo o o o o o:- o o o. o ' 8 0 ' . 3 " 3 " :0”. i l l l I 0 3O 40 50 Eb 70 AGE OF HOMEMAKER 9 B . 7 . °. 5 : 5 .0 O 0.. o o I . o .0 ‘b— o ||looo 0‘. cl. .0 o o . T I o o I o : .‘000 o o 2+~4 - o oo o 3. o ...-3.3;";. ‘0. :..I;. I,- o 00. o o no 3. a. 5“.) R; 11—5 W— . 'z— ' 0 Vow OI ' [0.50 I! o of. o so on. o I. '0" . 0 I O O 0' a o o o o o o EMATION N HORMAKER INCOME PER FAMILY 1 SIZE OF FAMILY tz‘oibiWir—n‘a—str—m INCOME PER FAMILY SIZE OF FAMILY 1 1 1 1 1 I l 2 4 %fi IO 12 14 IS Te 20 22 24 aoucmou or HOMEMAKER PO A Part C of Figure 2. This scatter gave little indication of relation- ships between thes factors. (D A more exact measure of the relationships between the family characteristics and the direction of this relationship were the co- efficients of simple correlation. These comparisons are shown in Table IV. Marked interrelationships are particularly noted between family income when compared with size of family and age of homemaker mp‘w 1‘. T i111) It: I] COEFFICIiKTS UP SILPLE CDlijLATIOH BQTWEJR FAHILY ChAiAGTiiISTICS FDA 131 FAAILIES IN THE UB-SALPLE CF PANJL DATA, 1952-1953 Family Characteristics Family Income Age Education Family Income -- -O.5016 O.h037 ge -O.SOlé —- -C.3602 .--.. .. ~ n r/, education C.u0;7 -U.;eu2 —- Size 0.6396 ~O.6133 0.33a8 '.' l I n I 0 ~""" I L'AKELIOQS O I Anal ‘2‘ ol S The first procedure undertaken in the study was to sort the summary cards according to each family characteristic and compute the average changes in per capita quantity and expenditure for each group for each V r 3 st i ‘ cut, as well as for total beef and total pork. Changes in the percent of weeks bought were treated in a like manner for individual cuts to establish the effect of this measure on respoLsi 4 (V. P < u (1 U1 . 3 "Total beef and total pork" shall hereafter reier tc tae live beef cuts and six pork cuts used in the stiuy unless otherwise notad. as included in this part of the anal"sis were er capita income, per family income, family size, age and edu- cation of the homemaker. The means computed were tested for significance by use of tre Stucent "t" test. The 90 percent confidence limits were used throughout t as a basis for accepting or T‘J‘Lbln the various hgpotheses. The prin— cipal reason for using this level of significance was an areim rary decision that narrOJer confidence limits mi h‘ reie ct nvoooucsos that were actually true. A similar procecure as that descrioed above was followed W13n cap Dita consumption or C J C!) :_J o :3 FCl "S families we re grouped accordinw to Chan’fl Q (J n enditure for "eef and pork cuts. This method was designed to indicate the average level of each family characteristic associated with three groups of families sorted according to the degree of price reSponsive- ness. In each case, the class limits de'erminin; the middle group of essentially no r3sponse we 3 arbitrarilv chosen by placing a similar number of familie s in all three groups. Both analyses used 195; income figures when income was the characteristic examined. Since the di-soribu- tion of income remained quite stable, choice of the income period would have little eff ct on the results of these analyses. (D hultiple r3gr ession analysis ma ahes it possible to determine the xtent to which any given variable influences the predicted variable while other variables are beinv held constant at some known level. L.) 4 Fr 3de1ick E. Croxton and Dudl.ev J. Cowden, Applied General Statistics, Prentice-hdall, Inc., New York, N. Y., lane, Chapter 12. I‘D 1:)\ In this study it was possible to estimate the expected response of groups of families when all of the family characteristics were con- sidered in one equation. The inte rcorrelation between the family characteristics suggested that mean changes in purchases for families grouped according to one characteristic may lead to erroneous conclu- sions. The limited number of families prevented a more complete cross— classification of family characteristics to compare mean changes in consumption adjustments where several factors are considered. Therefore, regression analysis was used to provide useful indications of the net relationship between changes in consumption and each family characteris- tic. Since individual fanilies were sed as the unit of observation, the usual le els of signii icance were relaxed because primary concern he re was predicted behavior 01° groups of families. §_ ri ri reasoning suggested that t1w relationships be twe n the inde pend,nt variables and price reSponsiveness were not linear in actual 'values. Therefore, a semi—logarithmic function was chosen for expressing these relation— ships. The slope of the regression line resulting from this function assumes that larger proportionate differen es in responsiveness occur 5 at lower levels for the factors considered. Four characteristics were used in the regression equation, per capita income being the factor omitted. The general form of the equation used was as follows: 1r X1 = a + b2 log X2 + b3 log Ks 4- b4 log; L, + b51055 A5. Mordecai Ezekiel, lethods of Correl a? ion Au alysis, John Wiley and Sons, Inc., New York, N. 1., oeccnd ulelJfl, l9u7: pp. 75— 63. where X1 = changes in level of consumption .1? iamily income, 1953 N to II X = ace of homemaker 3 o X4 = education of homemaker X5 = size of family This equation assumes a linear relationship between the independent variables expressed in logarithms and the dependent variable in actual numbers. It remains to be pointed out that the methods of analysis used in the study are not designed to give precise measurements of the structural relationships for individual family demand. The testing of means for family groups indicated that there was wide variability’about the average level of responsiveness within each group. It was also found that the standard errors of the regression. coefficients were quite large in equations where changes in individual family purchases were related to the measures of family characteristics. In spite of the lack of pre- cision in estimating individual family behavior the analysis may provide estimates that are useful in predicting behavior for groups of families. CHAPTER III CHANGES IN BEEF PURCHASES BETVQEN 1952 AND 1953 RELATED TO FAMILY CHARACTERISTICS Introduction The intent of this chapter is to provide some measures of dif- ferences in consumer responsiveness to changes in beef prices. More Specifically, it is an attempt to provide a more objective measure of the direct or indirect effects of family characteristics on consumer responsiveness to price changes. The results of three types of analyses will be presented. The three types of analyses are: l) sorting the families according to traits and computing simple averages of response measures; 2) sorting families according to the changes in consumption of the various meat items to obtain the average level of each family characteristic associ- ated with different degrees of consumption change; and 3) multiple regression analysis showing the net relationship between responsiveness of families to price changes and family characteristics. Due to the interrelationships that exist between family character- istics, it is doubtful that the results from the first two methods of analysis will yield true measures of the impact that any one character- istic may have upon the structure of demand by groups of families. Since further cross classifications of families where two or more characteristics are considered simultaneously is restricted by the limited number of families in the study, the net relationships between 28 variables were estimated by multiple regression analysis. However, the results obtained from the first two procedures are useful. They are helpful in understanding the nature of the data as well as in facilitating the interpretation of more complete methods of analysis. Results From Sorting Families Into Groups According to Family Characteristics General Comments When this approach was used, families were sorted into three or more groups, once for each of the five characteristics, size of family, age of homemaker, education of homemaker, per capita income and per family income. The mean changes in three measures of consumer purchas- ing activity that occurred between the two years 1952 and 1953 were then computed. These measures included the change in percent weeks bought, change in per capita quantity purchased, and changes in per capita expenditures. Age of Homemaker Table V presents the mean changes in the three measures of purchas- ing activity considered when families were sorted according to age of the homemaker. This table is divided into three sections. Part A deals with the mean changes in percent weeks.bought for each item and each group of families. Parts B and C deal in a like manner with changes in per capita quantity and per capita expenditure. The change in percent weeks bought did not appear to be a factor closely related to the age of homemakers. None of the means tested for 30 TABLE V AVERAGE CHANGES IN SELECTED MEASURES OF BEEF BURCHASES BETWEEN 1952 AND 1953 FOR FAMILIES GROUPED ACCORDING TO AGE OF H0hEh."ERd Family'ége Retail Cut Group Hamburger Roasts Steaks Stewing and Total Beef Boiling Beef (h Cuts) . c A. Change In Percent Weeks Bought 35 Years and under -.79 6.h0 6.19 1.10 36-50 Years -.28 7.80 11.90 - .38 51 Years and over .09 6.80 11.26 .98 B. Change in Per Capita Quantity (pounds per year) 35 Years and under 2.08 3.h3 2.56 .67 8.7h 36-50 Years .93 5.31 6.17 .2h 12.6A 51 Years and over 1.00 6.98 5.95 1.27 . 15.20 C. Change in Per Capita Expenditure (dollars per year) 35 Years and under -1.h0 .28 .98 -.06 -.20 36-50 Years -1.53 1.35 3.h2 -.03 3.21 51 Years and over -1.83 1.27 3.6h .03 3.10 3The number of families in each age group are as follows: Age 35 and under, 323 36-50, h53 51 and over, 5h. b . .-. . Tests of SigniIicance for comparisons between group means are shown in Appendix B, Part I. CThese figures indicate the actual change in percent weeks bought from 1952 to 1953 for the various cuts. 31 this classification were significant. It is interesting to note, however, that all three groups of families made sizeable increases in the percent weeks bought for roasts and steaks. Significant differences between groups for changes in the quanti- ties purchased and per capita expenditures for total beef indicated that older homemakers were definitely more responsive to price changes.1 Most of the variation between groups for total beef purchases was attributable to roasts and steaks. For roasts, the older homemakers were more responsive to price changes than were those in the age group of 35 and under. The middle aged group of homemakers was found to be the most reSponsive to changes in the price of beef steaks. However, older home- makers were found to be somewhat more reSponsive than were young homemakers. The pattern of changes in total expenditures for beef indicated the same general tendency as for the quantity purchased, the older group of homemakers being most responsive. The same pattern of differences were eSpecially noted in steak expenditures where younger housewives showed less responsiveness than either of the other two groups. A Similar trend was noted for roasts but could not be established by statistical tests. Since estimates of price elasticities for beef have indicated that the demand for roasts and steaks is fairly elastic, it seemed logical 1 "Significant differences" as used herein pertains to the 10 per- cent level unless otherwise noted. to expect that consumer responsiveness to price changes should be largest for these cuts, which was the case. On the basis of differences in price reaction to these cuts it appears that there is a relationship between age of homemaker and the reSponse to changes in beef prices. This analysis indicated that the older homemakers were most reSponsive to beef price changes. Education of Homemaker All three of the measures of consumer purchasing activity indicated that the educational level of homemakers is related to price responsive- ness for beef. These relationships are shown in Table VI. The average changes in percent weeks bought generally indicated that families with homemakers having from 9-11 years of formal education made the greatest change in frequency of beef purchases. For hamburger, this group did not test significantly different from any of the others, but it was established to have made a greater change in frequency of roast purchases than did homemakers with more than high school training. Several groups were found to differ from each other in terms of the change in per capita quantity of total beef purchased. Families whose homemaker had from 9-11 years education were significantly more responsive to declining beef prices than were any of the other educational groupings. This was particularly true when they were compared with homemakers subjected to college training. The fact that homemakers with 8 years and less education did not test different from the 9-11 year classification probably further indicates that price responsiveness normally decreases as the level of education increases. 33 TABLE VI AVERAGE CHANGES IN SELECTED REASURBS 0F BEEF PUicuASEs BETWEEN 1952-1953 roe FAMILIES GHOUPED AgCCiDINC T0 EDUCATION OF HOMEHAKER Retail Cut Family Education Hamburger Roasts Steaks Stewing and Total Beef Group Boiling Beef (h Cuts) A. Change in Percent Weeks Bought 8 years or less - .59 5.81 8.21 1.08 9-11 years 2.01 10.00 lh.79 .86 12 years 1.13 7.32 9.91’ .28 13 years or more -h.lh h.35 7.33 .87 B. Change in Per Capita Quantity (pounds per year) 8 years or less .96 6.h3 5.35 1.39 1h.1h 9-11 years 1.79 8.09- 8.36 1.20 19.uh 12 years 2.61 n.39 b.52 .03 11.5A 13 years or more - .9h 3.79 3.55 .h5 6.85 Change in Per Capita Expenditure (dollars per year) 8 years or less -1.61 1.A3 2.76 - .Oh 2.5A 9-11 years -1.07 3.12 h.69 .26 6.99 12 years ~l.l6 .05 2.h5 - .20 1.1h 13 years or more -2.79 - .05 1.68 ,.0h -1.12 a . . . . - .. - The number of families in each education group 'was as f0110ws: 9—11 years, 293 12 years, h63 Education of 8 years or less, 293 13 years or more, 27. b . .-. A . Tests of Signiiicance Ior comparisons between group means are shown in Appendix B, Part I. The individual cuts most influential to the general trend in beef purchases again seemed to be roasts and steaks. While none of the means were significant for roasts, tests indicated the same pattern of variation between groups for hamburger and steaks as for total beef. The quantity reSponsiveness for stewing and boiling beef denoted Sig— nificantly larger changes for homemakers with the least education when compared with homemakers completing high school. The mean changes in per capita expenditures for all beef were generally in agreement with the results found for changes in total per capita quantity. Roasts and steaks were the only cuts showing significant- ly- different levels of expenditures for education groups. The 9-11 year education group generally made greater changes than groups with higher levels of education. There appear to be important differences in the price responsive- ness of families when classified according to education of the home- maker. It was generally found that as education increased, responsive- ness decreased. These differences seem to be most pronounced between 2 the high school groups and those with higher levels of education. Size of Family Size of family was found to have significant relationships with the changes made in percent weeks bought and per capita quantity. Some 2 Families with homemakers having 9-11 years of formal education were significantly different from homemakers with 13 or more years of education at the one percent level for both per capita quantity and expenditure Changes in total beef. significant differences between groups of families were also noted in the per capita expenditure patterns for beef. The largest increases in percent weeks bought for roasts were made by families with two persons. This group inoreased their frequency of roast purchases more than single member households and somewhat more than families with five or more persons. The two person families were also significantly different from families with five persons or more in their frequency of purchases for stewing and boiling beef. Families with five or more persons actually decreased the frequency of purchase for this cut. Per capita quantity changes for beef indicated that the most responsive groups were families with two persons and that this group differed significantly from families with five or more members. In general, two person families made greater per capita changes in roast purchases than did either of the other larger family size groups. A large increase in purchases of beef steaks, observed for single person families was not found to be significant when compared with other groups, but three and four person families were more responsive than were those having five or more persons in the family. The largest sized families reacted more to prices for hamburger than did small sized families. This situation was reversed for stewing and boiling beef, where smaller families were the most responsive. Some heterogeneity was noted between. groups of families in their per capita expenditure patterns for beef. Expenditure changes differed between the two large family size groups for steaks and total beef. TABLE VII Family Size Retail Cut Group Hamburger doasts Steaks Stewing and Total Beef Boiling Beef (h Cuts) A. Change in Percent Weeks Bought 1 person — .07 h.00 11.08 .2h 2 persons -1.15 7.87 8.73 .61 3 or u persons -l.5l 7.h3 11.89 .51 S or more persons -1.25 6.27 9.7M -1.37 B. Change in Per Capita Quantity (pounds per year) 1 person - .h7 5.11 7.29 1.60 13.52 2 persons 1.39 8.01 h.89 1.11 15.h0 3 or u persons 1.18 h.h2 6.38 .52 12.50 5 or more persons 2.22 2.28 2.7h - .13 7.11 C. Change in Per Capita Expenditure (dollars per year) 1 person -2.29 1.36 b.17 .08 3.32 2 persons -1.85 1.80 2.h5 .02 2.h2 3 or h persons -1.21 .51 3.59 .01 2.90 5 or more persons -1.25 .25 1.28 — .23 .05 a . . . . . . . . . . The number of families in eacn family Size group was as follows. family size 1, 1a; 2 persons, 50; 3 or N horrors, D7; 5 or more persons, 20. b . . . . ,. . TeSts of Significance ior comparisons between group means are shown in firperdix B, Part I. 37 Significant differences were also observed between the smallest and largest sized families for total beef. Per Family Income The sorting of families according to per family income and compu- tation of mean changes in consumer responses to price changes indicated that this trait was not a good measure for indicating differences between consumer groups in their reaction to price changes. Stewing and boiling beef was the only item where the degree of response showed some relation to family income level. In this case the medium income families were slightly more responsive than the low income families. Judging from any trends that might be shown in the data for other cuts, it would appear that there may have been some tendency for lower. income families to make greater changes in their beef purchases than did the higher income groups. However, due to the wide variations about these means, this relationship was not statistically significant. Per Capita Income Per capita income proved to be better family trait to indicate differences in consumer responsiveness than did per family income. The quantity and expenditure changes for beef indicated that families ' withkdéflland medium per capita incomes are generally more responsive to price changes for beef than the high income group. However, the medium and 1mm: income groups were the only group comparisons testing signifi- cantly different from each other. TABLE VIII AvaaAoE CHANGES IN SELECTED MEAsuass 0F BEEF PURCHASES Bzrwesu 1952 AND 1953 FOL FAHILIES cacuren ACCORDING TO PEEL". Mam 11100111, 10¢? l/I Family Income Retall Cut Group Hamburger Roasts Steaks Stewing and Total Beef Boiling Beef (h Cuts) A. Change in Percent weeks Bought $5,000 and over - .27 6.82 8.5h - .26 $h,OOl-Sh,999 -2.19 8.36 11.09 1.92 $h,OOO and under .58 6.28 11.71 .78 B. Change in Per Capita Quantity (pounds per year) $5,000 and over 1.78 8.78 h.58 .b9 11.59 9h,OOl-$h,999 .25 7.78 8.07 1.18 13.28 $h,000 and under 1.12 5.22 7.39 .92 1h.6h C. Change in Per Capita Expenditure (dollars per year) $5,000 and over -1.00 .70 2.20 — .08 1.83 $h,OOl-$h,999 -2.hé 1.88 2.21 .12 1.75 $8,000 and under _1.88 1.05 8.29 - .01 3.26 —— w? aThe number of families in each income group was as follows: per family income $5,000 and over, 633 $h,001—$h,999, 273 $h,000 and under, bl. bTests of significance for comparisons between group means are shown in Appendix B, Part I. b) \0 TABLE IX AVER GE CHAKGES IN SELECTED kaisuass or BEEF PUicuasas azrwzzv 1952 AND 1933 FOR FAHILIBS GROUPED Agcoxniwc T0 33 CAIITA IKCOIE, 1953 _- . Retail Cut Family Income GroupD Hamburger Roasts Steaks Stewing and Total Beef Boiling Beef (h Cuts) _- — __—_ A. Change in Percent weeks Bought $1,801 and over - .11 6.73 8.07 .7h $1,250-fl,800 - .32 8.61 12.69 - .22 91,2h9 and under -1.hh 6.95 9.99 1.03 B. Change in Per Capita Quantity (pounds per year) $1,801 and over 1.21 7.85 5.83 1.00 15.09 $1,250-F1,800 1.58 5.36 6.81 .57 1h.32 21,2h9 and under .98 3.16 3.79 .63 8.52 C. Change in Per Capita Expenditure (dollars per year) 91,800 and over -1.72 1.60 2.58 .03 2.h8 $1,250—91,800 -1.31 1.53 3.9 - .06 h.12 $1,2h9 and under -1.65 - .15 2.13 - .02 .30 —— a '. . . . . The number of families in each income group was as follows: per family income $1,2h9 and under, 533 $1,250-81,800, 39; $1,801 and over, 39. b 9 on. o Tests of Signiiicance for comparisons between group means are shown in Appendix B, Part I. The mean 0 anges in per capita quantity of roasts indicated that -g- the higher per capita income Class was most responsive to price changes for this cut. Other trends observed in per capita quantity and (D x enditure measures of group differences could not be established at d he chosen level of significance. Families Grouped by Degree of Change in Purchases General Comments To further delineate consuuer -esponsiveness to price changes associated with family characteristics, an alternative to the original hypothesis was tested when families were grouped by the degree of change in purchases. This alternative hypothesis suggested that families be sorted into three levels of quantity or expenditure change for each cut and that averages be computed for the family characteris- tics at each of the three levels. This was done arbitrarily by array- ing the 131 families by degree of change in purchases and then dividing them into three groups of approximately equal number. The families in Group 1 are those families who made the greatest reSponse to price changes, while many of those in Group 3 made unexpected decreases in their consumption levels. The middle group tends to be concentrated around little or no response, but in some cases may include families who made sizeable increases in purchases, depending on the variability of changes made by all 131 families in the particular sort. Within each reSponse group the means were computed for each family characteristic. In computing these means the measures for average size 311 of family, and age and education of homemaker were based on class intervals (See Tables X, XI and XII). Means for income levels were based on actual income rather than a class interval measure. gge of Homemaker Table X shows the average age of homemaker for three family group- ings based on their responsiveness to price changes. This table is quite similar with those of the previous section except that the group comparisons are shown horizontally rather than vertically. A total of ten different quantity and expenditure factors were used in comparing the groups of families. Substantial differences were noted in the average age of homemakers among families grouped by the degree of change in total beef purchases. These differences were noted for both the quantity and the expenditure measure of change. These measures indicated that Group 1 families were significantly older than Group 2 families. This difference was highly significant for the per capita expenditure measure.3 In both cases, Group 3 homemakers were slightly older than those in Group 2, but this difference was not significant. The age patterns corresponding to the quantity and expenditure changes for individual cuts followed quite closely that for total beef. The one exception to this rule was noted in hamburger purchases. The per capita quantity changes for hamburger indicated that Group 1 home- makers were generally younger homemakers than those in the other groups. Significant at the one percent level. 42 TABLE X AVERAGE AGE or HOMEEAKER FOR FARILIES GROUPED ACCORDING TO THE DEGREE OF CHANGE IN SELECTED KEASUREE 0F BEEF PURCHASES FdOM 1952 T0 1953 ‘ —' - “u‘fiMQ’fi. Retail Cut and Measure Degree of Chang in Class Limits of Change Consumption for Group 2 Group 1 Group 2 Group 3 “nil—"”- «Wu-«H ---e —-—-— ...—... . “fl .— Hamburger Per capita quantity 3.27 3.63 h.07 + 3.20 to - .75 lbs. Per capita expenditure 3.30 - 3.98 3.71 -§0.25 to -#2.30 Roasts Per capita quantity 3.96 3.h0 3.61 + 6.25 to + .50 lbs. Per capita expenditure 3.96 3.30 3.70 +§2.00 to -?0.75 Steaks Per capita quantity 3.89 3.5h 3.55 + 6.uo to + 1.10 lbs. Per capita expenditure 3.80 A 3.56 3.61 +£3.h0 to +£0.20 Stewing and Boiling Beef 3.33. 3.86 + .75 to - .1u lbs. «I U", Per capita quantity 3. 3.86 +90.10 to -§0.86 b) 0 (D N b.) O {\3 03 Per capita expenditure Total Beef (h Cuts) Per capita quantity 3.98 3.37 3.61 +16.70 to +h.80 lbs. Per capita expenditure b.05 3.12 3.80 +$h.70 to ~$l.55 -aThe number of families in each age group was as follows: age 25 years and under, 6; 26-35 years, 273 36-u5 years, 27; h6—55 years, 355 56-65 years, 18; 66 years and over, 18. bResults for tests of significance between means are shown in Appendix B, Part I. DB This group differed significantly in age from the Group 3 families who decreased hamburger consumption. The expenditure pattern for this cut indicated that the Group 2 families were the older homemakers, but these families did not test significantly different in age from the other groups. For roasts and steaks, the average age of homemakers appeared to be highest in Group 1 families. This group tested significantly dif- ferent from homemakers in Group 2 for roasts, but was not significantly different in age from other groups of families purchasing steaks. The per capita expenditure responses for stewing and boiling beef indicated that significant differences in age exists between family Groups 2 and 3. This difference suggested that older families are less responsive in the per capita quantity changes for stewing and boiling beef. From the evidence shown above, it appears that age is associated with group responsiveness to beef price changes. It was generally found that older homemakers made the largest changes in their beef purchase patterns. Education of Homemaker Considerable differences were observed in the education levels of homemakers associated with different degrees of change in consumption (Table XI). It was found that the education level of homemakers was significantly higher for Group 2 families than for those families in Group 1 when compared on the basis of changes in quantity and expendi- 4 ture for total beef. The individual cuts contributing the most to Significant at the one percent level. AVERAGE T7 1.1;. TABLE XI hh )UCATICE OF HORELAKER FOR FAhILIES GAOUPED ACCORDILG TO THE DEGREE or CHARGE Bi SELECTED xEAsuajs CF Bier PURCHASES FROM 1952 To 1953a Retail Cut and Measure of Hamburger Change “-9 -.. * -. -m --MM- —_ ~— Degree of Change in Consumption —.——-————. ‘ - —— a‘nom ——--— M'm—Mn—nvwwh~ ...,” Class Limits fof Group 2 Group 1 Group 2 Group 3 Per capita quantity 2.50 2.51 2.61 + 3.20 to - .75 lbs. Per capita expenditure 2.6a 2.33 2.66 -?0.25 to -?2.30 Roasts Per capita quantity 2.23 2.98 2.h3 +6.25 to + .50 lbs. Per capita expenditure 2.32 2.8M 2.h8 +?2.00 to -?0.75 Steaks Per capita quantity 2.39 2.70 2.55 +6.h0 to + 1.10 lbs. Per capita expenditure 2.39 2.77 2.h8 +$3.u0 to +90.20 Stewing and Boiling Beef Per capita quantity 2.u1 2.72 2.50 + .75 to - .lh lbs. Per capita expenditure 2.h6 2.72 2.h6 +20.10 to -90.86 Total Beef (h Cuts) Per capita quantity 2.20 2.88 2.55 +16.7O to b.80 lbs. Per capita expenditure 2.23 2.88 2.52 +fh.70 to -€l.55 aThe number of families in each education group was as follows: education of 8 years or less, 29; 0—11 years, 29; 12 years, h63 13 years or more, 27. b n . . . . desults ior tests of Significance betveen means are shown in Appendix B, Part I. LS this general pattern in total beef purchases were roasts and steaks. Lower levels of education were noted for the families most respon- sive to price changes for beef roasts, while higher educations generally predominated among Group 2 families who made little or no increase in the per capita quantity of roasts purchased. The Group 2 families had significantly higher educations than either of the other two groups.5 The Group 2 families for per capita expenditure changes in roast purchases were also found to have better educated homemakers than Group 1 families. The conclusions reached by this method seem to be in substantial agreement with the findings when families were sorted by education levels and the mean changes in consumption adjustment computed. Both procedures indicated high levels of significance between family groups, withless response being noted as education increased. This method was particus larly good in pointing out the stability of purchase patterns for the more highly educated families. Size of Family In this method of analysis, size of family was somewhat more related to beef purchase changes than was found in the previous method where families were first grouped according to size. Highly significant dif- ferences were found between the larger sized families of Group 1 and the tendency observed for smaller sized families to be prevalent in 6 Group 2 for both measures of change in total beef purchases. Significant at the one percent level. 6 . Idem. TABLE XII AVEnAGE SIZE or FAHILY Fca FAilLles GiOUPED ACCUhDILG TO THE DEGREE or CHANGE IN ssLscriD nsAsuass or COKSUIPTION FROM 1952 TO l95’3a -. m —-~.—.—--—.—— ~—--—& Class Limits for Group 2 Degree of Change in Consumption Group 1 Group 2 Group 3 Retail Cut and Measure of Change c5 Hamburger Per capita quantity 2.59 2.7a 2.30 + 3.20 to - .75 lbs. Per capita expenditure 2.52 2.70 2.h8 -§O.2S to -?2.30 Roasts Per capita quantity 2.32 2.84 2.n8 + 6.25 to + .50 lbs. Per capita expenditure 2.39 2.72 2.52 +P2.00 to -?O.7S teaks Per capita quantity 2.h8 2.70 2.b6‘ + 6.bo to + 1.10 lbs. Per capita expenditure 2.h6 2.70 2.h8 +33.h0 to +§O.2O ‘ Stewing and Boiling Beef Per capita quantity 2.hl 2.70 2.52 + .75 to - .lh lbs. Per capita expenditure 2.3M 2.8h 2.h6 +§O.lO to —$O.86 Total Beef (h Cuts) Per capita quantity 2.27 2.8b 2.52 +16.70 to b.80 lbs. Per capita expenditure 2.36 2.8h 2.h3 +§h.70 to -§l.55 a . . . . The number of families in each Size group were as follows: family size 1, lb; 2 persons, 50; 3 or h persons, D73 5 or more persons, 20. b . ... . . mesults for tests of Significance between means are snown in Appendix B, Part I. h? The differences in size of family observed in per capita quantity changes for individual cuts were most pronounced for hamburger and roasts. For roasts, Group 2 was definitely noted to have more large sized families than did Group 1.7 For hamburger, the relationships between Group 2 families and Group 3 families were somewhat less pro- nounced. Roasts and stewing and boiling beef were the cuts where differences in family size associated with the per capita expenditure measures were most noticeable. Group 1 in the per capita expenditure measure of responsiveness for stewing and boiling beef indicated that Group 2 families were definitely larger families than were those in Group 1.8 The general trend for the other cuts was for larger size families to be more stable. The families in Group 2 who made little or no reSponse in the quantity of hamburger purchased were larger than the Group 3 families. Larger size families again dominated Group 2 for quantity changes in roasts. The families in this group were significantly larger than families stepping up their roast purchases.9 Tests also indicated that the Group 2 families, as a general rule, would be larger than Group 3 families who decreased the quantity of beef purchased between 1952 and 1953. Changes in per capita expenditures for roasts showed the same general pattern for size of family as for quantity changes, but were 7 Significant at the one percent level. a Idem. 9 Idem. 148 less significant. Size of family was highly significant in explaining differences between families increasing expenditures for stewing and boiling beef and the families of Group 2 who made little or no changes expenditures for the cut. Some deviation was noted between the Group 2 and Group 3 families in the expenditure changes for the out. These tendencies apparently indicate that larger families shifted away from this cut somewhat and probably paid less for these purchases than did other groups. Per Family Income Some differences were noted between the different consumption groups when per family income was used as a factor for comparison. The higher incomes for the non-responsive families were significant for the total quantity changes of beef when this group was compared with Group 1. Changes in expenditures for beef showed that Group 2 families had significantly higher incomes than either of the other two reSponse groups.10 Differences in income between groups of families were noted for only one cut of beef. Group 1 families making increased expenditures for hamburger had somewhat larger incomes than did Group 2. The situ- ation was quite different for per capita quantity measures of roast purchases. Here, the less responsive families tended to have higher incomes and were different from either of the other two groups, particu- 11 larly the negative respondents. 10 Significant at the one percent level. Idem. AVERAGE PER FAMILY INC r ‘. '11 UL" 1.13 4,. ... IN 1953 r E XIII h9' '02 FAhiLIis Geoursu Accessing TO THE DEGREE OF CHANGE IN SELLCTID MEASURES OF CONSULPTION FROM 1952 TO 1953 Retail Cut and Measure of Hamburger Per capita Per capita Roasts Per capita Per capita Steaks Per capita Per capita Stewing and Per capita Per capita Change quantity expenditure quantity expenditure quantity expenditure Boiling Beef quantity expenditure Total Beef (h Cuts) Per capita Per capita quantity expenditure ‘—-—-—-_.—-—- Degree of Change in ‘_»Consumption Group 1' 5318.20 h6§6.80 h661.h0 u829.50 h938.6o h579.50 h629.50 h386.h0 h377.30 Group 2 5597.70 5186.00 Ox .2.80 C) S u83u.90 SlSl.ho 5258.10 5362.70 5662.80 Group 3 ih977.3o 9&967.h0 $u531.80 h6§2.30 h236.h0 h63h.10 h720.50 h595.50 h736.ho hh52.30 Class Limits for Group 2 to + .50 lbs. +§2.00 to ~§O.YS + 6.h0 to + 1.10 lbs. +23.ho to +eo.2o + .73 to — .lh lbs. +$0.10 to -?O.86 +16.70 to h.80 lbs. +fh.70 t0‘-?l.55 aResults for tests of significance between means are shown in Appendix B, Part I. TABLE XIV PER CAPITA INCOME IN 1953 FOR FAMILIES GROUPED ACCORDING TO THE OF CHANGE IN SELECTHD MEASUNES 0F CONSUNPTION FROM 1952 TO 1953 Retail Cut and measure Degree of Change in Class limits of Change Consumption for Group 2 Group 1 Group 2 Group 3 Hamburger Per capita quantity 91791.59 21970.70 21791.1h + 3.20 to -.75 . lbs. Per capita expenditure 2017.50 1665.35 1863.6h -?0.25 to -§2.30 Roasts Per capita quantity 1917.95 2032.09 160h.77 + 6.25 to +.50 lbs. Per capita expenditure 1992.73 1752.79 1802.95 +$2.00 to -¢0.75 Steaks Per capita quantity ' 1880.23 1853.02 1817.50 + 6.h0 to +1.10 lbs. Per capita expenditure 1915.h5 1778.1h 1855.h5 +93.h0 to +$0.20 Stewing and Boiling Beef 9 Per capita quantity 1886.59 1930.70 1735.23 + .75 to — .1h lbs. Per capita expenditure 1862.95 1932.09 1757.50 +90.10 to -90.86 Total Beef (h Cuts) Per capita quantity 18h5.u5 l930.b7 1776.59 +16.7O to n.80 lbs. Per capita expenditure 1805.68 1816.59 +9h.70 to -§1.55 1930.23 a. . . . . desults for tests of Significance between means are shown in Appendix B, Part I. Per Caoita Inco e When fauilies were grouped according to family characteristics it was found that per capita income was a more important factor related to consumption changes than per family incone. This was not found to be the case for the present method of sorting by changes in beef .9 1. purchases. Changes in the quantity 0 roasts purchased was ‘the only factor for which different groups exhibited any noteworthy variation in per capita incomes. The less-responsive roup had larger incomes C“ C) ) than Group 3 families. Results from Regression Analysis Four family characteristics were used in a series of multiple regression equations to determine the effects of thee characteristics (0 on consumer reSponsiveness to meat price changes. The inf rmation for all 131 families was fitted to measures of changes in per capita quantity and per capita expenditure for each of the individual cuts of beef and ’4- for total beef. The four ndependent variables for the equations were: per family income (X1); age of homemaker (X2); education of homemaker (KB); and size of family (X4). Thi equation was used to show the net relationship between family characteristics and the price responsiveness for the 131 individual family observations. When the necessary computations were made, the equation represent- ing the effect of family characteristics on the total quantity of beef was as follows: Change in Quantity (Y1) = 13.3817 + 9.87h8 log Kl + 6.7077 log X2 3.8823) (12.0923) (1) 83 log Xe - 13.757h log K4 ,83) ’ (8.373.) R = .2312 52 The average change in per capita quantity that was predicted for total beef was close to 13 pounds. The average per family income associated with this response was $h,2h0, age of homemaker, 88.8 years, education 10.8 years, and size of family 2.7 persons. These results give some evidence that perhaps two family traits, family incomes and size of family, affect consumer response to price changes for beef. The standard errors of the regression, indicated in parentheses below each regression coefficient,show that the level of significance for income and size of family approach 10 percent. The standard errors on the coefficients for age and education of the home- maker were quite large, indicating that the varied pattern of responsive- ness for individual families would not give a good prediction of the effect of these characteristics on average responsiveness for a large group of families. While the multiple correlation coefficients (R) are generally quite low throughout the study, it is maintained that these coefficients are an evaluation of the usefulness of the equations for predicting individual family behavior. Since the primary interest of this study was oriented ltoward predicting group action. this measure should rot be evaluated by the usual criteria. The equation for changes in expenditure for total beef was sug- gestive of similar patterns of response as that for total quantity measures. This equation was as follows: Change in Expenditure (Y2) = 6.7723 + 2.h2h7 log X1 + h.5700 log X2 (b.3381) (7.6409) (2) - 5.9088 log is - 2.6739 10g X4 (7.7713) (5.2902) 53 The results from this equation were in agreement with the results obtained in Equation 1 for per capita quantity changes in total beef. Family income and age of homemaker both tended to increase group expenditures for total beef. Education and size of family were vari- ables that tended to decrease beef expenditures. However, the high standard errors of the regression coefficients in Equation (3) indicated that the relative importance of these family characteristics could not be substantiated. The equations used in predicting the average responsiveness for all the families by individual cuts are shown below: Hamburger: Change in Quantity (Y3) = 12.7113 + .7h68 log X1 (2.6706) (3) - 7.1961 log K, - 1.7001 log X3 - 1.3556 log X4 (b.7039) (n.78u2) (3.2592) R = .1556 Change in Expenditure (Y4) = .9273 + .h258 log X1 (1.5668) (h) - 1.5931 10g X2 - 1.6957 log X3 + .7586 log X4 (2.7596) (2.8067) (1.9121) R - .1169 Roasts: Change in Quantity (Y5) = -9.0652 + 3.2799 log X1 + S.hl69 log X2 (h.6988) (8.2762) ) (5 - 3.2716 10g x3 - 6.7039 log X4 (8.h175) (5.73hu) R = .1758 5h Change in Expenditure (Y6) = b.9789 - .0707 log X1 - .3h26 log X2 (2.5735) (n.5328) (6) - 1.9776 10g x3 - 2.6809 10g x4 (8.6102) (3.1607) - R a .1190 Steaks: Change in Quantity (Y7) = - 1h.2526 + 5.2111 log K, + 9. 8963 log x, (3.5993) (6 3 97) (7) - 8.0399 log X3 - 2.359h log X4 (6.6h80) (8.3927) R = .2308 Change in Expenditure (Y8) = - 11.9017 + 2 .0h81 log X1+ 6. 7003 log X2 (2. 6938) (h. 7L6?) (8) (h. 8237 ) (3.2875) 3': .1713 Stewing and Boiling Beef: Change in Quantity (Y9) = 3.hl91 + .h668 log X1 - 1.0870 log X2 (l.hh52) (2.5h56) ' (9) - 1.2891 log X3 - n log X4 2. 85 0 (205890) (1. 7028) R = .1770 Change in Expenditure (Yio) a 1.3h38 + .0h88 log X1+ .0u82 log X2 ( .SSOh) (. 969D) (10) + 1,28hh log X3 - .5980 102 X4 (.9859) (.6717) R = .1397 55 The predicting equations for hamburger (Equations 3 and h) indicated slightly conflicting results with what was obtained for total beef ”quantity and expenditure changes. Equation (3) indicated that family incomes had practically no relationship with-hamburger purchases. The age of homemaker was found to be a factor decreasing per capita purchases for this cut. While this is contrary to the effects of age of homemaker observed for total beef, this deviation was also noted by other methods of analysis. The regression equations fitted to roast purchases were not good indications of the importance of family characteristics on consumer purchase behavior. The standard errors for all but one of the family characteristics were larger than the regression coefficients themselves. In general, the directional relationship was the same for all factors in the per capita quantity equation for roasts as for total beef. Family purchases of beef steaks were found to be positively corre- lated with family income and age of homemaker, but negatively related4 to education of the homemaker and size of the family. The lower relative standard errors of the regression coefficients for family income, age of homemaker, and education of the homemaker in Equation (7) indicated that these factors may be related to family responsiveness for steak purchases. The effect of family characteristics on purchases of stewing and boiling beef was quite varied (Equations 9 and 10). Families generally made very little change in the purchases of stewing and boiling beef from 1952 to 1953. It has generally been found in the regression analysis of beef purcha es that family income and age of homemaker are positively related to changes in beef consumption. This was true for roasts and steaks, the two beef cuts accounting for most of the change in beef purchases between 1932 and 1953. Education and size of family were usualli necativel related to chanres in beef urchases. (v; r L, CHAPTEd IV CHANGES IN PCEK PUmCHASSS DURING 1952 AND 1953 RELATED TO FAKILY CHAlACTEiISTICS Introduction During the two-year period, 1952-1953, retail pork prices were rising while beef prices were making a substantial decline. Sizeable adjustments in pork purchasing patterns were made. The average retail pork price dropped nearly 10 percent in the Greater Lansing area and similarly for the United States. The average consumption of pork in the consumer panel decreased over 15 percent in 1953, while average expenditures for pork dropped nearly 8 percent. These adjustments were also noted for the 131 families in this study that were taken from the consumer panel. Here it was found that the average decrease in the quantity of pork purchased was over 18 per- cent, while expenditures decreased slightly more than 10 percent. This pattern was not uniform for the families who reported their pork purchases during the two-year period. The range in the quantity of pork purchased by individual families was found to be from a decrease of nearly 55 pounds to an increase of over 20 pounds. These substantial changes that occurred in meat prices coupled with the varied Iaatterns of adjustment in pork purchases that were made by individual families provide an excellent situation for analyzing the impact of these price changes on pork purchases. 57 58 The analyses presented in this chapter are essentially the same as those discussed in Chapter III for beef purchases. The first section summarizes mean differences in purchase adjustments when families are sorted into groups according to selected characteristics. The second section shows the mean differences in family characteristics when the 131 families are sorted into three groups based on the degree of shift in pork purchases. The third section gives the results of a series of regression analyses where the degree of change in pork purchases is expressed as a function of four family characteristics. Results From Sorting Families into Groups According to Family Characteristics V Age of Homemaker Table XV presents the mean changes in the three measures of purchas- ing activity when families were sorted according to age of homemaker. Two cuts indicated that age of homemaker was significantly related to the change in frequency of percent weeks bought.) These cuts were bacon and ham. Families with homemakers who were 35 Years or under tended to buy bacon less frequently than those families with homemakers 51 years or over. However, for ham, the tendency was for older home- makers to purchase less frequently than the young age group. In both cases, the two extreme age groups were the only group comparisons established by test to be different. All age groups made large decreases in per capita quantity and expenditure measures for pork. No important group differences were observed for either per capita quantity or expenditure measures37? _\,__\,.\,_ ( 192 )::A. .\'.._\'._V. Roasts (233)" (1&2):;; (1a2);m& (1&2);;" M (233)""" (2&3)""" (2&j)““ (1&2)‘ Stewing and n Boiling Beef (192);M :UUL ,L a, Total Beef (2&3 “" (213)"“" (192)111 (2&3)‘" (2&3)“"“ Change in Per Capita Expenditure Hamburger ,QL ,ma 3, (2&3); ROEISJCS (2313 ) A A (23:3 )3: (23.3 A Steaks (2&3)“ Stewing and M a Boiling Beef (1&2); (1&2);&& (2a3);y 3,, (2&3);;" 11a: To tal Beef (151-2 11:" ( 2 3:3 ) " " ‘ ( 119:2 )_:,‘.:_\,_ (1.9-2 )_f,.j:_":_ (25r'*)""’r (2&3)”““ (2&3)”““ gLevels of significance are indicated as follows: ten ercent level, *3 five percent level, %%3 one percent level, %*%. APPENDIX C PART I RESULTS FOR TESTS OF SIGNIFICANT DIFFEEENCES BETWEEN MJASURES OF CHANGE IN PORK PURCHASES FOR FAMILIES GROUPED ACCORDING TO ” FAfiILY CHARACTERISTICS Classification of Families , Age Education Size of Per Family Per Capita Group a 0, (Years) (Years) bamily Income Income (Persons) (Dollars) (Dollars) l 35 and 8 or less l @5000 and $1301 and under over over 2 36-50 9-11 2 $h001-h999 fi250-1800 3 51 and 12 3-h $hOOO and $12h9 and over under under h 13 or more 5 or more r . , . . . . a uroup Comparison Testing Significant A ....- ___- “A..- AJ‘L; __ __._‘ ...“ A Age Education Size of Per Family Per Capita Family ‘Income Income Change in Percent weeks bought: M ,QH, 9% Bacon (1&3)” (2&h)“‘“ (1&3);1i _v..~<. y. (2&3), “A Chops and Steaks (2&n)"“ (1&2); (1&3 )3. (16h) " Ham (1&3)n x ,Q, iHQL Roasts (1&2); (i&h)“‘ (1&3x:;‘ (18:3 ) 1‘, (28:3 ) y Sausage (1&‘)( (lB‘X;: (233), Change in.Per Capita Quantity: xx Ham ‘ (189 )_’;' (1&2);h Picnics (l&')” a Roasts (1&3)" aLevels of significance are indicated as follows: ten percent level, *3 five percent level, **3 one percent level, ***. RESULTS FOR TESTS OF SIGNIFIC NT DIFFERFUCES BETJEflN LEVELS OF FAMILY CHAfiACTERISTICS FOR FAHILIES GflOUPED ACCORDIEG TO CHANGES IN PORK PURCHASES BETWEEN 1952 AND 1953 Age qucation Size of For Family Per Capita Family Income Income Change in Per Capita Quantity Bacon (132 )j‘ (2&3 )"" Pork Roasts (2&3)“ (1&2)" (1&3);uu (r) O_' )‘n-A‘n' €6- LK») $33.)? Sausage (l&3)x (1&3) (257:3) Change in Per Capita Expenditure Bacon (233 W ’ ) t 3 v \ ' a \‘I t \ I“ Pork Roasts AAA Five 3 \JJL) [\J Sausage Total Pork v V V )5 A A re cs k ) V v. 4‘ gLevels of significance are indicated as follows: ten percent level, %3 five percent level, **3 one percent level, %**. ROOM USE can ‘ ..'-»‘-. a» u «..., “‘-“-’“‘“0¢bw.h-£«' ,2, '3 a U d : ‘-. .- " b 5 c . s=~ - MICHIGAN STATE UNIVERSITY II I I IIIIIIIIIIIIES 1 636 293 0306] 3 3