Beliefs About Others As Determinants of Purchase Behavior

Citation:

Robert E. Burnkrant (1972) ,"Beliefs About Others As Determinants of Purchase Behavior", in SV - Proceedings of the Third Annual Conference of the Association for Consumer Research, eds. M. Venkatesan, Chicago, IL : Association for Consumer Research, Pages: 807-811.

Proceedings of the Third Annual Conference of the Association for Consumer Research, 1972      Pages 807-811

BELIEFS ABOUT OTHERS AS DETERMINANTS OF PURCHASE BEHAVIOR

Robert E. Burnkrant, Illinois State University

Understanding and predicting purchasing and the decision making which underlies it is an objective of interest to everyone in consumer behavior. Much recent work directed at furthering this understanding and ability to predict has centered around expectancy-value models of attitude and intention formation and change with notable emphasis recently being placed on the Fishbein model.

A theoretical approach which offers considerable insight into the processes underlying the formation of attitudes and intentions with respect to products is attribution theory. This theory, although receiving considerable attention in psychology, has received only minor consideration thus far in consumer behavior. Settle (1972) and Settle, Faricy and Warren (1971) have recently done some work in this area applying an approach to attribution theory which draws heavily on the work of Kelley (1967).

Another approach to attribution theory which is being taken by Calder and Burnkrant (in final preparation) is derived primarily from the work of Jones and his associates (Jones and Davis, 1965; Jones and Harris, 1967). This espousal of attribution theory focuses more directly and explicitly on the formation and change of beliefs about other users of products based on observations of their behavior with respect to these products. Questions that might be asked of consumers and are currently being investigated at the University of Illinois are to what extent consumers attribute characteristics to others (acquire beliefs) based on the others' use of certain products or brands, and to what extent consumers are aware of and behave in accordance with these attributions.

In the initial study of a series of research projects currently being conducted to investigate these phenomena an attempt was made to determine if the types of products people use are significant factors in determining the characteristics others attribute to them. Although analysis of this data is not yet complete, preliminary results indicate that people make significantly different (p < .01) attributions to users of brands which are low in social desirability than they do to users o, brands which are high in social desirability (Calder and Burnkrant, in final preparation). To state this more simply they attribute distinct characteristics to individuals based on their use of brands of a product which vary in social desirability.

The study to be discussed here was undertaken as a follow-up to the initial project to see if these results which were obtained using college students in a relatively structured homogeneous environment could be replicated in a field survey using people taken from the community at large and if the respondent's purchase of a given brand influences the attributions he makes about other users of that product. Specifically, it was hypothesized that users of a product low in social desirability would be perceived as having characteristics significantly different from users of a product high in social desirability. It was also predicted that purchase by the respondent of one of these types of products would not significantly affect the attributions made on the basis of another's use of one of these product types.

TABLE 1

TESTS OF THE MULTIVARIATE F TEST

TABLE 2

WOMAN EVALUATED - MAIN EFFECT

TABLE 3

FACTOR 1

TABLE 4

FACTOR 2

TABLE 5

FACTOR 3

METHODOLOGY

A probability sample of 100 respondents was selected from the Champaign-Urbana area. The Champaign Urbana City Directory (1972) which lists streets and addresses alphabetically and numerically was sampled systematically to provide 20 blocks. Within each selected block a complete listing of housing units was obtained. Five people were systematically selected from within each block and they became the respondents for the sample.

Each selected respondent was personally interviewed by trained student interviewers who provided subjects with a structured questionnaire and written instructions. The critical items for the purpose of this paper were a series of 22 semantic differential scales on which the respondents rated either a woman who "normally uses a regular phosphate detergent in doing her laundry," or a woman who "normally uses a non-phosphate detergent in doing her laundry." The determination of which woman the respondent evaluated was made by placing one description systematically in half the questionnaires and the other description in the other half of the questionnaires. Thus each successive respondent received a different treatment level. Each respondent was also asked to indicate whether she usually bought a regular phosphate detergent, a non-phosphate detergent or didn't know.

RESULTS

The semantic differential scales were factored using principal components analysis on a correlation matrix adjusted for the treatment effects. Three factors were extracted and rotated using a varimax rotation. Those items loading highly on a given factor were selected as defining that factor. The individual's score on that factor became the sum of his scores on the selected scales. The scales selected to represent each factor are listed as follows:

TABLE

Using these factor scores as dependent variables a 2 X 2 factorial analysis of variance was performed. The two treatments were whether the woman evaluated by the respondent bought regular detergent (low) or non-phosphate detergent (high) and whether the respondent purchased a regular detergent (low) or a non-phosphate detergent (high). A significant multivariate F test (p < .01) was obtained for the main effect of the "woman evaluated" treatment. The multivariate F test obtained for the "product purchased" treatment was not significant. The interaction effect was also tested and found not to be significant.

Looking at the univariate F tests for the three factors on the "woman evaluated" treatment we find the first factor marginally significant at p < .054, the second factor significant at p < .05 and the third factor significant at p < .001.

DISCUSSION

A look at Tables 3, 4 and 5 which contain the cell and marginal means provides an indication of the direction of the effects. Clearly, women who bought nonphosphate detergents were evaluated as being higher on the first and second factors and lower on the third factor than women who normally bought regular detergents. This appears to make sense since we might label the first factor as being something of a social concern or leadership factor. Thus the results indicate that people who purchase non-phosphate detergents are significantly higher in social concern or leadership than those who purchase regular detergents. Similarly, we might call the second factor a success factor where success is viewed in terms of goal attainment. The results would then indicate that people see users of non-phosphate detergents as being more successful than those who purchase regular detergents. Finally, the last factor which is an uninformed impulsiveness factor indicates that people who purchase non-phosphate detergents are seen as being more informed and less impulsive than those who buy regular detergents.

The absence of an interaction effect between the "purchase" factor and the "woman evaluated'' factor has interesting implications. If a person's purchase and use of a product would influence his perception of users of that product along the dimensions relevant to that perception we would expect a significant symmetric interaction effect. In other words, we would expect those who buy regular detergent to rate other women who buy regular detergent higher on the social leadership and success factors and lower on the uninformed impulsiveness factor than women who buy non-phosphate detergents since the former two traits are presumably good and the last is probably seen as bad. Likewise we would expect the reverse to occur for those people who regularly purchase non-phosphate detergents. That this didn't occur provides evidence to indicate that the characteristics people attribute to others are stable percepts held in general by the population from which the sample was selected.

The lack of an interaction also provides some inferential evidence indicating that the characteristics we attribute to users of these two product types are veridical. If they were not we would not expect the characteristics attributed to users of a given product type by people who do not purchase it to be reinforced by people who do purchase it.

SUMMARY

A study was conducted in which people were asked to attribute characteristics to users of regular detergent or users of non-phosphate detergent. It was found that significantly different characteristics were attributed to users of non-phosphate detergents than to users of regular detergents. The attribution of these traits was found not to be influenced by whether or not the respondent purchases the product in question.

REFERENCES

Calder, B. J. & Burnkrant, R. E. Attribution Theory and Consumer Behavior. In Final Preparation. Urbana-Champaign, Illinois.

Champaign-Urbana City Directory. Loveland, Colorado: The Johnson Publishing Company, 1972.

Jones, E. E. & Davis, R. E. From Acts to Dispositions. In L. Berkowitz' (ed.), Advances in Experimental Social Psychology, Volume 2. New York: Academic Press, 1965.

Jones, E. E. & Harris, V. A. The Attribution of Attitudes. Journal of Experimental Social Psychology, 1967, 3, pp. 1-24.

Kelley, Harold H. Attribution Theory in Social Psychology. In David Levine (ed.), Nebraska Symposium on Motivation. Lincoln, Nebraska: University of Nebraska Press, 1967, pp. 192-238.

Settle, Robert B. Attribution Theory and Acceptance of Information. Journal of Marketing Research, 1972, 9, pp. 85-88.

Settle, Robert B., Faricy, J. H. & Warren, G. T. Consumer Information Processing: Attributing Effects to Causes. Proceedings: 2nd Annual Conference. Association for Consumer Research. 1971.

----------------------------------------