Personality As a Determinant Factor in Store Choice


Stewart W. Bither and Ira J. Dolich (1972) ,"Personality As a Determinant Factor in Store Choice", in SV - Proceedings of the Third Annual Conference of the Association for Consumer Research, eds. M. Venkatesan, Chicago, IL : Association for Consumer Research, Pages: 9-19.

Proceedings of the Third Annual Conference of the Association for Consumer Research, 1972      Pages 9-19


Stewart W. Bither, Dartmouth College

Ira J. Dolich, The Pennsylvania State University

[This research was supported by a National Science Foundation institutional grant to the two authors by The Pennsylvania State University.]

[Stewart W. Bither is Visiting Associate Professor at the Amos Tuck School of Business Administration, Dartmouth College and Associate Professor of Marketing at The Pennsylvania State University.]

[Ira J. Dolich is Associate Professor of Marketing at The Pennsylvania State University.]

In working toward a better understanding of consumer behavior, researchers have turned their attention to theories of perception, attitude formation, learning, personality and other social-psychological variables that influence the individual's decision making process. Because the monetary stakes are high in applied marketing problems, much of the research has focused on factors important in brand differentiation. Personality has not proven to be as successful a tool for predicting brand differentiation as other social psychological variables (Massy, Frank, & Lodahl, 1968). Perhaps one of the pervasive reasons for lack of success is the fact that most personality instruments used in marketing focus on interpersonal response traits. Although many brands are consumed in social situations involving more than one actor, the brand choice decision is seldom made in interpersonal social settings. Thus, for many products, individual differences of importance in interpersonal response tendencies are only indirectly related to brand choice.

There are, however, two major consumer decisions which have a direct effect on brand purchases and usually occur prior to brand choice decisions. These two decisions are the consumer's choice of product class ant his choice of store. The former decision variable has received little attention in marketing although it appears, on the surface, to be an interesting and fruitful area for personality oriented research. The latter decision variable has received much attention but most of the published research focuses on demographic data and the economic analysis of store location. The present study involves an investigation of individual personality differences and their influence on store choice.


The decision to investigate the relationships between personality and store choice was predicated on the belief that store selection, although complex, involves many considerations that might be influenced by individual response characteristics of an interpersonal nature. The modern supermarket does not present the opportunity for close personal relationships between owner and patron, yet, the supermarket shopping trip is characterized by a number of interpersonal contacts. The selection ant cutting of meat is often a personalized service. Check-out counters involve personal interrelationships as do check cashing situations. In addition, the occasional search for the unusual item or the item apparently out of stock often requires the help of store personnel. Finally, supermarket shopping involves maneuvering among fellow customers as well as the chance or planned meeting with friends and acquaintances. This latter characteristic has been taken into account by shopping mall planners where facilities for social activities have been incorporated in several new shopping establishments.

Despite the fact that the store choice situation is a consumer decision where present personality instruments may have predictive power, it must be acknowledged that no personality instruments have been developed specifically for the study of this, or any other consumer behavior process. In addition, previous marketing studies utilizing personality instruments have analyzed markedly different measuring instruments, consumer groups, and heterogeneous choice situations. Thus, without a meaningful research tradition it is difficult to generalize about the potential predictability of present instruments (Kassarjian, 1971).

The personality instruments used in this study were the Edwards Personal Preference Schedule (EPPS), Jackson's Personality Research Form (PRF), and Canon's Social Adroitness Index (SAI). The EPPS was selected based on Claycamp's success with this instrument in predicting savings bank versus savings and loan patronage behavior. Using this instrument, Claycamp was able to correctly classify 72% of his subject's patronage behavior (Claycamp, 1965). Jackson's PRF and Canon's SAI were employed in this study to ascertain the applicability of newer instruments to tap individual differences relevant to marketing choice situations. These latter instruments were specifically developed for the study of nonpathological behavior and each has been validated on subjects similar to the ones used in this studY.


A panel of 134 housewives was established to analyze the supermarket shopping behavior of new arrivals to a community. Members of the panel were contacted prior to their arrival to the new community. Cooperation in prearrival identification of prospective newcomers was obtained from industry, the Chamber of Commerce, real estate agencies and colleges located in the area. Pre-arrival shopping factors were measured through a mail questionnaire and post-arrival data was obtained by personal interview three months after arrival. During the post-arrival interview, subjects completed the Edwards EPPS, the Jackson PRF, the Canon SAI, and a comprehensive questionnaire about their supermarket shopping since their arrival.

Panelists were classified into three groups according to their usage of a most preferred supermarket (Farley, 1968). Those women who made 60% or more of their shopping trips to their most preferred supermarket were classified as regular shoppers. Irregular shoppers were defined as those women who made less than 40% of their shopping trips to their most preferred supermarket. The remainder of the panel were designated as intermediate shoppers. Since the major thrust of the investigation was to study regular versus irregular shoppers, the intermediate shoppers were set aside. The entire battery of personality traits and a standard collection of socioeconomic variables were then utilized as independent variables in building discriminant functions.

A stepwise discriminant analysis was utilized to establish the functions (Dixon, 1971). The stepwise procedures were used because of the exploratory nature of this study. Although the EPPS had been used in previous research (cf. Claycamp, 1965~ Evans, 1959; Keuhn, 1963), insufficient evidence existed to identify relevant traits for supermarket choice behavior on an a Priori basis. Further, the Canon SAI had not been previously used in marketing research and the Jackson PRF had not been used in studies of this kind

The Jackson and Canon instruments were separated from the Edwards Schedule to illustrate the discriminating ability of the two sets of traits individually as well as in combination. This was a procedure utilized by Claycamp to indicate relative importance of classes of variables. Separation of trait instruments in this study is not intended to indicate importance, but to show the loss of discriminating power of individual instruments relative to combined trait usage.


The results of this research project are subject to the following limitations:

1. The community in which the study was conducted is dominated by a large university. Thus, the panel of newcomers differed from average United States demographic characteristics on level of education.

2. The panel differed from established adult norms on several traits of the personality instruments used.

3. The study included store choice behavior only during the first three months of a newcomers arrival.

4. The community studied is a medium sized community with approximately 12 full line supermarkets.

All of the above factors limit the extent to which the results of this study may be generalized.


There are numerous studies which have reported improved explanation of regression models and increased classification in discriminant models when socioeconomic variables were "tossed into the pot." Demographic variables were also included in the present analysis. However, the stepwise procedures did not provide socioeconomic variable usage meeting several important criteria: (1) parsimonious use of variables, (2) useful increases in proper classifications, and (3) significant differences in discriminant mean values. Thus, the discriminant functions were not improved by adding socioeconomic variables. It should be noted, however, that the newcomer panel did not have substantial variation on the demographic variables of income, education, and age.

The Multiple Discriminant Analysis of Regular and Irregular Shoppers

The 41 regular and 39 irregular shoppers were randomly divided into split half samples. Subsample 1 utilized 21 regular and 20 irregular subjects and subsample 2 contained 20 regular and 19 irregular shoppers. Each discriminant function was derived from subsample 1 and tested in subsample 2 (cf. Wilson, Mathews & Sweeney, 1971; Cooley & Lohnes, 1962; Overall & Klett, 1971; Frank, Massy & Morrison, 1965; Morrison, 1969). A total of five discriminant models were derived and tested. Model 1 consists of the best possible model using only the EPPS traits to predict store patronage. The second model uses the combined traits of the Jackson PRF and Canon SAI with store patronage as the dependent variable. The third model uses all three personality instruments as independent variables with patronage behavior as the dependent variable. Model 4 indicates the discriminating ability of the EPPS variables, and Model 5 indicates the discriminating ability of the PRE and SAI variables used in Model 3. Table 1 shows the traits included in each of the 5 models as well as the canonical coefficients and F tests for each of the traits. Table 2 shows the classification matrices for each of the five models for the subsample from which the model was derived and for the subsample on which the model was tested. Table 3 is a summary of the percentage of properly classified cases by each of the 5 models and Table 4 shows the approximate F tests for each of the 5 models.

Model 1. Model l, using the EPPS, predicts 85 percent correct classification (Tables 2 & 3), using the traits of achievement, deference, autonomy, succorance, dominance, endurance, and aggression. However, validation with subsample 2 provides only 54 percent correct assignments. This occurs even with the strong significance test results (P[F] = < .01) of actual differences existing between the two discriminant means (Table 4). It is interesting to note that Claycamp found (among 6 discriminating traits) the three traits of achievement, deference, and autonomy in his stepwise discriminant models of savings depositors (Claycamp, 1965). And, Evans (Evans, 1959) used achievement, deference, autonomy, dominance, aggression, and six other traits in his often quoted study of Ford and Chevrolet owners.

Model 2. The best possible classification model utilizing the PRF and SAI is shown as Model 2 (with 6 traits). This model suffered the same fate as Model 1 in terms of precipitous drops in classification ability for the test group. Tables 1 through 4 provide most of the detail necessary for further evaluation.

Model 3. Model 3, with 5 traits, is the most parsimonious configuration obtained for the entire set of traits. No improvement in the 78 percent correct classification is obtained until the addition of a 15th trait. Testing the model against subsample 2 provides 70 percent correct classification, the best obtained for any model. The F values of the variables are all in reasonable ranges (Table 1) as well as approximate F test for Wilks Lambda (Table 4). A summary table of mean values is also provided as Table 5. Figure 1 provides plots of subsample 1 and Figure 2 provides plots for subsample 2 based on Model 3 coefficients.

The combined instrument model utilizes 2 EPPS (dominance and intraception), 2 PRF (social approval and order), and the SAI scale. It is interesting to note that Evans and Claycamp (Claycamp, 1965; Evans, 1959) both used intraception and dominance but the latter eliminated intraception in his stepwise procedures. It would appear that the EPPS Dominance trait has relevance in marketing across several behavioral measures. Claycamp's best results were with 6 personality variables alone (autonomy, achievement, recession, nurturance, deference, and heterosexuality) where 79 percent were properly classified. However, he did not have a test of the model to determine the magnitude of the upward bias. It should be noted that Model 1, a model with 7 EPPS traits and significant F at .01, rather than Claycamp's .05, can drop over 30 percent in correct assignment on testing with a split-half sample.

Models 4 and 5. Models 4 and 5 are provided to show the relative capabilities and characteristics of the individual sets of personality traits. Although neither model performs as well as Model 3, they do not fair badly nor indicate the presence of the large upward biases found with the more numerous trait functions of Models 1 and 2.
















This study involved an investigation of the relationship between individual personality differences and store patronage behavior. A rationale for investigating the role of personality in this area of consumer decision making was developed based on the interpersonal aspects of the consumer shopping trip. A panel of 134 newcomers to a community participated in the study. Regular and irregular usage of the consumers most preferred store was used as the dependent measure of patronage. Independent variables were the individual trait scales of 3 personality instruments, the Edwards Personal Profile Schedule, Jackson's Personality Research Form, and Canon's Social Adroitness Index. Discriminant models were developed using these instruments singly and in combination. Each model was derived and tested on separate halfs of the panel data.

The most parsimonious and predictive of the models developed used a combination of traits from the three personality instruments. Seventy percent of the subjects were properly classified in the test of this model. Although use of the Edwards instrument in consumer studies has been reported, two of the personality instruments have seen little use in marketing research (the Canon instrument has not been previously used). Some similarity between the present study and earlier studies in the prediction ability of individual traits on this instrument was observed.


Claycamp, H. J. Characteristics of owners of thrift deposits in commercial banks and savings and loan associations. Journal of Marketing Research, 1965, 2, 163-170.

Cooley, W. W. & Lohnes, P. R. Multivariate Procedures in the Behavioral Sciences, New York: John Wiley, 1962.

Dixon, W. J. (Ed.), BMD: Biomedical Computer Programs, Berkeley: University of California Press, 1971.

Evans, F. B. Psychological and objective factors in the prediction of brand choice. Journal of Business, 1959, 32, 340-369.

Farley, J. U. Dimensions of supermarket choice patterns. Journal of Marketing Research, 1968, 5, 206-208.

Frank, R. E., Massy, W F., & Morrison, D G. Bias in multiple discriminant analysis. Journal of Marketing Research, 1965, 2, 250-258.

Kassarjian, H. H. Personality and consumer behavior: a review. Journal of Marketing Research, 1971, 8, 408-418.

Keuhn, A. A. Demonstration o, a relationship between psychological factors and brand choice. Journal of Business, 1963, 36, 237-241.

Massy, W. F., Frank, R. E., & Lodahl, T. Purchasing behavior and personal attributes, Philadelphia: University of Pennsylvania Press, 1968 and Journal of Advertising Research, 1969, 9.

Morrison, D. G. On the interpretation of discriminant analysis. Journal of Marketing Research, 1969, 6, 156-163.

Overall, J. E. & Klett, C. J. Applied multivariate analysis. New York: McGraw-Hill Book Company. 1971.

Wilson, D. T., Mathews, H. L. & Sweeney, T. W., Industrial buyer segmentation: a psychographic approach. Presented at The American Marketing Association's Fall Conference in Minneapolis, Minnesota, August 30 - September 1, 1971.



Stewart W. Bither, Dartmouth College
Ira J. Dolich, The Pennsylvania State University


SV - Proceedings of the Third Annual Conference of the Association for Consumer Research | 1972

Share Proceeding

Featured papers

See More


Anticipated Interpersonal Feedback Reshapes Other-oriented Intertemporal Choices

Adelle Xue Yang, National University of Singapore, Singapore
Oleg Urminsky, University of Chicago, USA

Read More


Felt Status, Social Contagion, and Consumer Word-of-Mouth in Preferential Treatment Contexts

Brent McFerran, Simon Fraser University, Canada
Jennifer Argo, University of Alberta, Canada

Read More


P2. The Upside of Myopic Loss Aversion

Daniel Wall, Carnegie Mellon University, USA
Gretchen Chapman, Carnegie Mellon University, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.