Duivis Lon ot kesearcn Graduate School of Business Administration University of Michigan MARKET SEGMENTATION ANALYSIS: EXAMINING SHOPPING AND BUYING DECISIONS Working Paper No. 70 by Claude R. Martin, Jr. Assistant Professor of Marketing University of Michigan FOR DISCUSSION PURPOSES ONLY None of this material is to be quoted or reproduced without the express permission of the Division of Research. January 1973 "jA;kr.F

BACKGROUND This paper is based on research sponsored by Research Group B of the Division of Research, Graduate School of Business Administration, University of Michigan. A group of department stores from seven midwestern states provided the financial and logistical support for a series of studies into consumer behavior. This paper is one in a series of reports on this behavioral research. ABSTRACT This paper examines the factors considered by women consumers in the purchase of personal wearing apparel. Seventy-five in-depth interviews with consumers at the point of purchase were followed by 356 mail survey interviews that asked women to retrace a most recent purchase of an item of clothing they had purchased for themselves. The women were interrogated concerning the behavior, predispositions, information and product cues, demographics, and buyer goals associated with that purchase. The data were then subjected to the MCA computer analysis to determine those variables that had the greatest impact on choice of store type for both shopping and buying. These variables were then processed through the AID analysis to formulate tree diagrams that graphically depict both the shoppers and buyers for each particular type of store. Both analyses show a greater predictive power for predispositional and behavioral characteristics in market segmentation and a relative nonimportance of standard demographic characteristics., 482 323A2 a 13 on W,. Xt

I i CONTENTS Introduction Research Design First-Level Analysis Second-Level Analysis Design Results Conclusions 1. 1 5 5 5 9 1I

I i i TABLES 1. 1970 Census Characteristics of Joplin and Springfield, Missouri 2. MCA Analysis-Correlation Coefficients 3. Independent Variables with Eta Greater than.02 4 7

FIGURES 1. Buyer construct. 4 2. Explanatory variables —first MCA run. 6 3. Aid tree for high-fashion women's specialty store shopper. 13 4. Aid tree for mass merchandiser shopper. 14 5. Aid tree for department store shopper. 15 6. Aid tree for high-fashion women's specialty store buyers. 16 7. Aid tree for mass merchandiser store buyer. 17 8. Aid tree for department store buyers. 18

Introduction Several recent studies have shown that two methodological techniques — AID (Automatic Interaction Detector) and MCA (Multiple Classification Analysis) —can be powerful tools for examining consumer behavior and for 1/ market segmentation.- This paper demonstrates the use of these two computer tools with a well-connected hypothetical construct of buyer behavior in an attempt to identify those dimensions that are the best predictors of women's fashion-buying behavior. Research Design The first step was to conduct 75 in-depth interviews with women consumers in three southwestern Missouri stores. The women were approached as they completed an apparel purchase and were interviewed concerning that purchase decision. This interview provided the basis for the design of a mail questionnaire which examined specific variables that emulated the 9, 3/ behavioral modeling work of Howard-/ and Howard and Sheth. The variables -1/Dennis Gensch and Richard Staelin, "The Appeal of Buying Black," Journal of Marketing Research, IX (May, 1972), 141-48; Joseph Newman and Richard Staelin, "Prepurchase Information Seeking for New Cars and Major Household Appliances," Journal of Marketing Research, IX (August, 1972), 249-57; Henry Assael, "Segmenting Markets by Group Purchasing Behavior: An Application of the Aid Technique," Journal of Marketing Research, VII (May, 1970), 153-58; William Wilkie, "Extension and Tests of Alternative Approaches to Market Segmentation," Working Paper No. 323 (Lafayette, Ind.: Institute for Research in the Behavioral, Economic, and Management Sciences, Purdue University, September, 1971); William Peters, "Using MCA to Segment New Car Markets," Journal of Marketing Research, VII (August, 1970), 360-63. 2/ - John A. Howard, Marketing Management (Rev. ed.; Homewood, Il.: Richard D. Irwin, Inc., 1963), Chapters 3 and 4. 3/ 3/John A. Howard and Jagdish Sheth, The Theory of Buyer Behavior (New York: Harper & Row, Inc., 1969).

- were categorized as: behavior, predispositions, information and product cues, demographics, and buyer goals. Each woman was asked to retrace her most recent purchase of an item of apparel for herself. Then she was questioned about the variables shown in Figure 1 as they related to that purchase((P. 4). Two retail trade areas in Missouri —Joplin and Springfield —were chosen for this study. The main reason for this selection was merchant cooperation, but another reason was that the two areas showed differences in socioeconomic 4/ status and growth. Additionally, the two areas are geographically close, thereby controlling for regional differences. The major factors differentiating the two markets are population growth, educational levels attained, median and mean income levels, and median value of housing (Table 1, p. 5). The mail survey obtained 356 usable responses. The distribution of the respondents was compared to the age, marital status and employment Idistributions of the 1970 census to check for nonrepresentative samples, and it was found that the distribution of the respondents was similar in configuration to that of the general population. The original in-depth interviews led to the conclusion that there are differences in consumer behavior associated with different types of retail stores. The respondents were asked to identify, by name, the stores in which they had shopped and the store in which the purchase of a particular item of apparel had been made. With the assistance of a five-member retailer panel in both cities, the 96 stores mentioned were classified into general categories. Among these categories were three distinctive types, which were studied further: 41Bureau of the Census, Census of the Population 1970, PC (1), A27, B27, and C27 (Washington, D.C.: Government Printing Office, 1971).

High-fashion women s -- Principally selling apparel specialty stores for women in the middle and upper price ranges Mass merchandiser - Restricted to Sears, Penney's, stores and Montgomery-Ward only Independent -- Full-line department stores department stores other than those cited in the mass merchandiser category First-Level. Analysis The objective of the first level of analysis was to determine whether the variables shown in Figure 1 contribute to understanding where women 5/ shop and buy clothes for themselves. The MCA program- was used to test six dependent variables: Department store buyers Department store shoppers Buyers in high-fashion women's specialty store Shoppers in highEfashion women s specialty store Buyers in mass merchandiser store Shoppers in mass merchandiser store There were 33 explanatory variables used in this analysis (Figure 2)..When the model for all variables was used, significantly high correlation coefficients were obtained for each dependent variable (Table 2), which led to the conclusion that many of the predictors of women's fashion-buying behavior had been measured in the study. (See pp. 6 and 7.) Second-Level _Analysis Design Bass, Tigert, and Lonsdale recognized the need for a multivariate analysis to examine the variations in such grouped data.-i Certainly the 5/Frank Andrews, James Morgan, and John Sonquist, Multiple Classification Analysis (Ann Arbor: Institute for Social Research, University of Michigan, 1967). 6/Frank Bass, Douglas Tigert, and Ronald Lonsdale, "Marketing Segmentation: Group Versus Individual Behavior," Jurna of Marketing Resea.rh. V (August, 1.968), 264-70.

Demographics: Marital status Age Employment status of respondent Employment status of husband of respondent, if married Number and ages of children City of residence Predispositions: Negative colors —garment colors respondent would not buy Negative fabric characteristics —fabrics respondent would not buy Garment care characteristics wanted Wardrobe accessory matching Upper and lower price limits to purchase Had charge account where shopping and buying reported Previously bought apparel in store of purchase Prepurchase planning: General Specific —positive color wanted positive fabric wanted Product and Information Cues: Comparison shopping at alternate stores Utilization of price limitations Method of payment Sought out particular sales clerk Use of "shopping pals" Used sales clerk evaluations of style and fit of garment Evaluation of mass media helpfulness in purchase decision Buyer's Goals: Self-evaluation of fashion awareness Factors used in developing level of fashion awareness Shopping enjoyment in buying clothes for self Behavior: Coordinating items purchased Type of garment purchased Number of stores shopped Number of stores shopped on day of purchase Color of garment purchased Fabric of garment purchased Garment care requirement for item purchased Fig. 1. Buyer construct.

TABLE 1 1970 Census Characteristics of Joplin and Springfield, Missouri I- -II I - -- -- - ---— - -I — — ___ __..__ __..____ _ _ _.II____.__.___,_ __ __,__..._____.....,._. _ ___,..__ — -- - Census Characteristic Springfield Standard Metropolitan Statistical Area (SMSA) Joplin Retail Trade Area Jasper County Newton County Population growth (in percentage) 8.5 6.9 8.1 Median years of education for males Percentage of population having completed high school 12.3 years 11.8 years 11.5 years 47.3!T r 58.9 48.7 Median income Mean income $8,215 $7,312 $6,887 $7,785 $9,310 $8,410 Owner-occupied household's median value Renter-occupied household's median rent $13,900 $9,000 $9,800 $55 $73 $55 Source: Bureau of the Census, Census of the Population, 1970, PC(1), A27, Government Printing Office, 1971). B27, C27 (Washington, D.C.:

-6 - City of residence of respondent Life cycle-pcombination of marital status and age variables Employment status of respondent Employment status of respondent's husband Number of children Item purchased Number of total stores shopped Number of stores shopped on day of purchase Prepurchase planning Positive color preference Positive fabric preference Lower and upper price limitations Method of payment Charge account in store of purchase and/or shopping Previous purchase of personal apparel in store Evaluation of sales clerk assistance with style and/or fit Predisposition tonuse a male sales clerk Self-evaluation of fashion consciousness or awareness Amount of shopping enjoyment in buying for self Frequency with which respondent uses newspaper advertising in fashion purchasing and evaluation of helpfulness of such ads Method most helpful to respondent in developing her fashion awareness Prepurchase discussion of buying decision with others Frequency of shopping with other persons Tendency of respondent to shop alone Type of store shopped Type of store where purchase was made Fig. 2. Explanatory variables —first MCA run.

I_ \I TABLE 2 MCA Analysis —Correlation Coefficients -... _. Dependent Variable Shopper in high-fashion women's specialty store Department store shopper Shopper in mass merchandiser store Buyer in high-fashion women's specialty store Department store buyer Buyer in mass merchandiser store.50635.38177.41322.50264.48303.40457

limits imposed by 356 respondents and 33 predictive variables argue against the more simplistic multiple cross-classification suggested by those authors. Limitations of AID. Assael has demonstrated the effective use of the 7/ AID algorithm in market segmentation.- There are, however, several substantive limitations to such use, AID computes the ratio of the between sum of squares for each variable by the total sum of squares for the group to be split. It then selects the highest ratio for the binary split of the respondents. All of the subsequent splits are contingent on the subgroups formed by the first split; yet it is possible that there is a small difference in discrimination between the variable split and the second variable; The resultant tree diagram produced by a split on the second variable could be 8/ 9/ quite different. Assael- and Newman and Staelin- have suggested that a sensitivity analysis consisting of subsequent AID runs be used. Under this technique the analyst eliminates the first split, reruns the analysis without the initial variable, and compares the structures produced. The major drawbacks to this technique are the expense of programming and computer time charges and the arbitrary judgments used in comparing the different tree diagrams. Another limitation to the use of AID is that the independent variables may be closely interdependent and there may be high intercorrelation among the predictors. Finally, there is the problem of judging where to terminate the AID tree, i.e., what criteria are to be used for aborting the iterative process? 7/Assael, "Segmenting Markets", pp. 157-58. 8/Ibid., pp. 155-56 9/Newman and Staelin, "Prepurchase Information Seeking", pp. 250-52.

-9 - Overcoming the limitations. It is suggested that the MCA program be used to partially overcome these limitations. In its elementary form the MCA program produces measures of simple associations —pairwise ) correlations —between the dependent variables and the independent variable. This is reflected in the Eta score output of the program. Eta2 indicates the "ability of the predictor, given the categories given, to explain the variation in the dependent variable." 10/ The scores provide a foundation for the need to undertake the type of sensitivity analysis that has been suggested and the content of such analysis. Certainly they form the basis for halting the tree diagram construction to some defined limit, e.g., in 2 this study only those variables with an Eta score of beyond.02 were used. Finally, MCA can aid in indicating the presence of possible high intercorrelations among predictors. The MCA iteration process fails to converge in/ or converges very slowly with oscillations when such a possiblity exists. The analyst would then stop the process and examine the variables for interdependence rather than proceed with the costly, subsequent AID analysis. Results MCA analysis. When the MCA analysis was used with each of the six 2 dependent variables and the limitation of an Eta >.02 was imposed, the result was the identification of the applicable independent variables (Table 3). These then formed the basis for the subsequent AID analysis. AID analysis. The results from the use of the AID algorithm are shown in the tree diagrams in Figures. 3-8. The numerical figure is the percentage of that type buyer or shopper occupying that particular branch -Andrews, Morgan, and Sonquist, Multiple Classification Anal ysi s, p. 22. ll/Ibid. p. 32.

TABLE 3 2 Independent Variables with Eta Greater Than.02 Shopper in High- Buyer in HighFashion Women's Department Store Shopper in Mass Fashion Women's Department Store Buyer Specialty Store Shopper Merchandiser Store S._ecialty Store Buyer Merchand Location: Joplin Location: Joplin Age of buyer Location: Joplin Location: Joplin Age o or Springfield or Springfield Woman's employment or Springfield or Springfield Woman Husband's employment Age of buyer status Age of buyer Age of buyer men status Woman' employment Husband's employment Husband's employment Woman's employ- Numbs Number of stores status status status ment status chi shopped Number of children Number of children Number of stores Husband's employ- Numbs.Prepurchase planning in family Number of stores shopped ment status shb Price limitations Number of stores shopped Prepurchase planning Number of children Prepu Method of payment shopped Prepurchase planning Prepurchase positive Number of stores shopped pla Previously purchased Prepurchase planning Prepurchase positive color preference Prepurchase planning Prepu in store Prepurchase positive fabric preference Price limitation Prepurchase positive poa Use of sales clerk in color preference Price limit Method of payment color preference * col purchase decision Prepurchase positive Method of payment Previously pur- Prepurchase positive fev Enjoyment in shopping fabric preference Store charge account chased in store fabric preference Nethc for self Method of payment Use of sales clerk in Use of sales clerk Method of payment me: Helpfulness of newspaper Store charge account decision in purchase Store charge account Store advertising in buying Use of sales clerk in Enjoyment in shopping decision Predisposition to use acc decision purchase decision for self Enjoyment in shop- a male sales clerk Use c Fashion awareness Predisposition to use Helpfulness of newspaper ping for self Enjoyment of shopping cle a male sales clerk ads in purchase decision Helpfutness of news- for self - put Enjoyment of.shopping Fashion awareness paper ads in buy Helpfulness of news- -dec for self decision paper ads Enjoj Helpfulness of newspaper Fashion awareness Fashion awareness she ads Discuss with others sel Fashion awareness Shop with others Helpi int dec 2aht1 in Mass riser Store f buyer 'a empl yit status kr of Idren ir of store wpped irchase mainning irchase itive Lor prerence Md of payIt t charge:ount if sales irk in *chase:ision ment of pping for Lf uliness of rspaper ads purchase itsion Lon avarener

-11 -of the analytical tree to that point. For example, the farthest right-hand branch of the tree for department store shoppers shows that 55.9 percent of those shoppers enjoy shopping for themselves, find newspaper advertising helpful in their purchase decision for personal clothing, and shop in more than one store. In fact, when department store shoppers were examined, the maximum reduction in the unexplained sum of squares is obtained by splitting this cadre of shoppers by the number of stores in which they shopped. The tree diagrams show graphically the characteristics of the customersy either shoppers or buyers, of particular types of retail stores. Conclusions The tree diagram should be useful to the retail marketing manager in identifying major market segments for women's apparel. Certainly the factthat there are differences in the tree diagrams of buyers and shoppers of similar types of stores can lead to some understanding of the differences between those who actually buy and those who are lured into, but do net buy, in a store.': Interestingly the three different types of store shoppers are all initially split on the basis of the number of stores shopped. Certainly the department store manager should find the fact that 58.4 percent of his shoppers comparison shop (two or more stores) and find newspapers helpful in making purchase decisions useful in his evaluation of the media. The additional recognition that most of those shoppers (55.9 percent) find shopping for their own clothing enjoyable should be a valuable contribution in determining the content of the advertising message. Similarly, the Sear's, Ward's, or Penney's retailer should be interested in the fact that the best predictor of his buyers is customer use of sales clerks, and that 75.9 percent of his market makes limited or

no use of the sales person in his purchase decision making. On the basis of these findings the retailer could decide to hire and train sales clerk5: more for the order-taking function than for the order-getting function, and serious questions can be raised about the viability of paying commissions to sales clerks based on ordersi:taken. Other analyses concerning sales force usage, advertising strategy, and product "extras" can be gleaned from the results of this study. 2 The relatively high R, when all the variables were used in the initial MCA analysis, lends support to the behavioral construct of Howard and Sheth. The greater predictive power of the predispositional and behavioral characteristics and the relative nonimportance of standard demographic characteristics is indicative of cross-classification analyses that use these demographic dimensions as the primary basis for market segmentation.

.1 Husband is Salaried Ei (Non.selfemployed) Fig. 3. Aid tree for high-fashion women's specialty store shopper. I

NO Prercha 15.1 _ rSO Oyed I oried Employe 1113 sb-odis plOyl "~~~~\ ~ ~ Fig. 4. Aid tree for mass merchandise store shopper.

I~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 9S1 I Fig. 5. Aid tree for department store shopper.;

Fig. 6. Aid tree for high-fashion women's specialty store buyers...

Charge ount with re of Purchase. 21.3 Ii* High Positive Color iton I Predisposition 9[. 112.2 -30 iddte-.Aq~d Age Groups: omen Urvder'30 0-60yeors) Ov-60 12.2 0.0~ o Fig. 7. Aid tree for.ass F erchandi,e store buyer.

F -: i-I; Extremes of Median Shopping En. Shopping joyment(Avfid Enjoyment n r NO Positive Positive!or None)"!.e II in Color Color tS' 8 Preference Preference Fig. 8. Aid tree for department store buyers.