Individual and Product Correlates of Evoked Set Size For Consumer Package Goods

Michael Reilly, Montana State University
Thomas L. Parkinson, Lehigh University
ABSTRACT - This study is the most comprehensive study of the determinants of evoked set size for consumer package goods to date, focusing on eight product categories and based on the responses of experienced adult female consumers of those products. The results indicate that the size of consumer's evoked sets is affected by both individual and product characteristics, specifically brand loyalty, brand awareness, situational factors, and educational level. Implications for further research and marketing practice are identified.
[ to cite ]:
Michael Reilly and Thomas L. Parkinson (1985) ,"Individual and Product Correlates of Evoked Set Size For Consumer Package Goods", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 492-497.

Advances in Consumer Research Volume 12, 1985      Pages 492-497

INDIVIDUAL AND PRODUCT CORRELATES OF EVOKED SET SIZE FOR CONSUMER PACKAGE GOODS

Michael Reilly, Montana State University

Thomas L. Parkinson, Lehigh University

ABSTRACT -

This study is the most comprehensive study of the determinants of evoked set size for consumer package goods to date, focusing on eight product categories and based on the responses of experienced adult female consumers of those products. The results indicate that the size of consumer's evoked sets is affected by both individual and product characteristics, specifically brand loyalty, brand awareness, situational factors, and educational level. Implications for further research and marketing practice are identified.

INTRODUCTION

With the proliferation of brands that has occurred in the marketplace for consumer package goods, it is well known and widely accepted that buyers don't give equal consideration to all available brands. In order to simplify the decision-making task, many brands are eliminated from consideration early in the decision process and the final selection is made from a relatively small number of brands. This final subset of brands actually considered for purchase when making a specific brand choice is called the evoked set (Howard & Sheth 1969).

In light of the fact that inclusion in the evoked set is a necessary condition for final purchase of a brand, the process of evoked set formation should be of great interest to both marketing practitioners and public policy makers. From the marketing manager's point of view an understanding of how consumers finally select the brand(s) that they purchase is only relevant if the firm's brand is among those finally considered, i.e., is in the evoked set. Yet, while the process by which consumers evaluate and make final brand choices has been a subject of great interest and the focus of a large body of research, the process of evoked set formation has received only recent attention (Belonax 1979, May 1979, Parkinson and Reilly 1979, Brisoux and Laroche 1980) . Since the size of the buyer's evoked set is a limiting condition on the breadth of buyer choice, it is basic to the understanding of evoked set formation to first analyze those factors that determine the size of that set.

The early research into the determinants of evoked set size is characterized by its sparsity of results and focus on a single or very few products (Belonax 1979, Belonax and Mittelstaedt 1978, Campbell 1969, Gronhaug 1973-74, Jarvis and Wilcox 1973, Maddox et.al. 1978, May and Homans 1977, Ostlund 1973, Prasad 1975, Williams and Etzel 1976) . This study reported investigates the relationship between evoked set size and a large number of individual and product characteristics across eight different consumer package goods. Specifically, evoked set size will be correlated with a variety of variables related to the purchase and use of the products, the personal characteristics of the buyers, and a number of demographic variables.

The Determinants of Evoked Set Size

The term evoked set was first introduced into the marketing literature by John Howard in 1963 (Howard 1963). However, the concept was not generally accepted until the publication of Howard and Sheth's (1969) comprehensive model of buyer behavior some years later. Howard and Sheth defined the evoked set as "the brands that the buyer considers as acceptable for his next purchase...", and postulated that the magnitude of a buyer's evoked set was related to the importance of the purchase and personality traits such as Pettigrew's (1958) category width. However, they also hypothesized that when the number of available brands is numerous, there would be no relationship between evoked set size and the number of known brands.

Campbell (1968) provided the first evidence of the actual existence of evoked sets. Participants did, in fact, consider only a small subset of the brands known to them purchasing toothpaste and laundry detergent. Campbell found that only two variables, brand loyalty and "importance of price differences among brands," were associated with evoked set size. Evoked set size was found to be related negatively to the former and positively to the latter. Campbell, however, found no relationship between the evoked set size and any demographic characteristic. The results of this study further supported the proposition that evoked set size is an individual phenomenon as buyers were consistent in their relative evoked set size across both product categories.

Ostlund (1973) replicated Campbell's initial study several years later using sub-compact cars as the product of interest and a subject pool made up of adult males. Again brand loyalty was found to be negatively correlated with the magnitude of the respondent's evoked set. However, category width (Pettigrew 1958), perceived risk, perceived price importance, amount of search, and previous ownership experience did not appear to effect evoked set size. In the same year Jarvis and Wilcox (1973) investigated the relationship between perceived importance of the product class and evoked set size utilizing three household products and female adults as subjects. Jarvis and Wilcox hypothesized that individuals who were more ego-involved with a product class would have smaller evoked sets. Their results supported this relationship, and in addition, indicated that an individual's evoked set size in one product class was correlated with his evoked set size in another. This second finding would appear to suggest that individual differences are important determinants of evoked set size.

Narayana and Markin (1975) expanded the concept of evoked set by further classifying all of the known brands in the consumer's "awareness set." Their conceptualization suggested the division of the brands in the awareness set into the evoked set, the inept set (rejected brands), and the inert set (brands evaluated neither positively nor negatively).

The evoked set size concept was examined in the consumer durables category by Williams and Etzel (1976) who examined the purchase behavior of married students and their spouses. The results indicated that evoked set size for consumer durables was positively related to price, but not to the "perceived importance of the purchase." This study also indicated that the evoked set concept could be extended to consideration of retail outlets.

Thus, with the exception of Gronhaug's (1973-74) study, which demonstrated a positive relationship between evoked set size and education level, there exists very little evidence of any relationship between evoked set size and traditional marketing and demographic variables except brand loyalty. This paucity of significant results led to several studies designed to explore the relationship of individual cognitive variables and evoked set size.

Prasad (1975) hypothesized that the relationships between personality variables and evoked set size was mediated by intervening variables. In his study o f toothpaste, shampoo, and headache remedies, evoked set size was shown to be related to need for cognitive clarity only in situations where performance risk and product specific self-confidence was high. When either performance risk or self-confidence was low, no relationship was observed between evoked set size and need for cognitive clarity. Thus, personality variables appear to manifest themselves differently in different decision situations. Prasad's results would appear to indicate that investigating direct relationships between personality variables and evoked set size would be difficult, and that investigation of the indirect links would be more appropriate.

May and Homans (1976) investigated the relationship between the abstractness of an individual's information processing, which is also positively correlated with education level, and the size of the individual's evoked set for automobiles. They found that the level of abstraction in the comprehension-of the product class and the level of abstractness in the choice criteria were both positively related to the size of the evoked set. This finding would appear to indicate that evoked set size might be related to the complexity of a person's cognitive processes. Maddox, et.al. (1978) extended the research on evoked set size for automobiles to the international arena with their study of evoked set size correlates in Norway and the United States. Their results indicated that evoked set size in both countries is negatively associated with age, and positively correlated with educational level and the amount o f search.

Further evidence justifying the examination of both individual and purchase situation variables and their effect on evoked set size is provided by the recent work of Belonax and Mittelstaedt (1978, 1979). Their study investigated the decision making process of adults and reported that the size of the evoked set was negatively related to both the number of choice criteria used and the variability of information available.

The most recent examinations of the evoked set concept have focused on the process of evoked set formation (Parkinson and Reilly 1979, Brisoux and Laroche 1980). However, there remains a gap in the evoked set size research stream for a comprehensive study focusing on a significant number o f products and both individual and product characteristics. This study attempts to fill that gap.

METHODOLOGY

To investigate the relationships between the number of brands in a consumer's evoked set and both individual and product characteristics, a written questionnaire was designed and administered to 111 adult females. The participants in the study were drawn from the women's organizations of four different churches in a university community. A contribution of $5.00 per participating member was made to these organizations upon completion of each questionnaire.

This self-administered questionnaire was distributed to each respondent at one of their organizations regular meetings and returned within a week to the authors. The questionnaire consisted of three parts and covered eight product categories. These categories were selected to represent a cross-section of household items commonly purchased by women homemakers. The products included were bath soap, ground coffee, hand lotion, kitchen paper towels, laundry detergent, margarine, shampoo, and toothpaste.

The first section of the questionnaire consisted of eight pages (one for each product category) and required the respondents to evaluate the following eleven characteristics related to the purchase and use of brands in each product category:

- frequency of purchase

- difficulty of purchase decision

- purchase pattern (brand loyalty)

- importance of purchase decision

- perceived risk (2 items) likelihood of satisfactory performance importance of satisfactory performance

- product complexity

- peer influence

- confidence in purchase ability

- situational variation in brand choice

Each subject evaluated all eight products; however, the order of presentation was rotated to offset any fatigue or experience effects. The second section of the questionnaire was designed to measure the subject's evoked set size in each product category. To accomplish this, the subjects were first presented with a list of all of the national and private brands in each product category that were available in the community. The total number of brands listed ranged from a high of 19 (laundry detergent) and a low of 10 (ground coffee, kitchen paper towels, and toothpaste). In each product category, the subjects were first asked to cross-out any unfamiliar brands, and then to do likewise with any which were familiar to them by name only. This established the "inert set" (Narayana and Markin 1975). The evoked set was then identified by asking the subjects to select from among the remaining brands, only those brands that they "would consider buying." In addition, the subjects were given the opportunity and encouraged to write in any additional brands that they would consider which were not on the original list. This procedure was repeated for each product category with the order of presentation again being rotated throughout the sample. The final section of the questionnaire contained the dogmatism measure (Troldhal and Powell 1965), a measure of the subject's cognitive style (Cox 1967), generalized self-confidence, (Day and Hamblin 1964), and finally a cognitive complexity measure (Bieri et.al. 1966). The questionnaire ended with the collection of demographic information.

Analysis Procedure

In view of the fact that each of the subjects was asked about all eight products, a repeated measures analysis of covariance model was utilized. However, since eight of the 111 subjects failed to complete one or more of the evoked set size instruments, a final data matrix containing responses from 103 subjects was used. The data were first analyzed by a simple repeated measure analysis of variance to determine if there were any significant differences in evoked set size across subjects and product. Stepwise analyses of covariance was then used to identify the various individual and product characteristics that significantly contributed to the differences in evoked set size in the first analysis. BMDP2V (Dixon and Brown 1979) was utilized in all three analyses.

RESULTS AND DISCUSSION

Table 1 contains the results of the initial repeated measures analysis of variance across product categories, while controlling for subject effect. The significance of the F-Ratio associated with the product effect indicate that there were significant differences across product categories. To further explore the significance of these differences, Fisher's test was used to determine which means were statistically significant from each other. The results of this analysis showing the mean evoked set size for the eight product categories appear in Table 2. As shown in the table, the mean evoked size ranged from a high of 4.09 brands for bath soap to a low of 2.83 brands for coffee. Although significant differences exist across the entire eight categories, the results also indicate that many of the means are not statistically different from each other.

TABLE 1

ANALYSIS OF EVOKED SET SIZE ACROSS PRODUCT CATEGORIES

TABLE 2

COMPARISON OF MEAN EVOKED SET SIZE BY PRODUCT CATEGORY

Product Factors

In order to identify the product factors that contribute to the differences in evoked set size between categories, a stepwise analysis of covariance was carried out with each of the eleven product factors included as covariates, while controlling for subject differences. The resulting linear model was as follows:

Yi = m + Ti + aj + Ti aj + bk(Xkij - Xk), where (1)

bk = the regression coefficient for covariate k

Xk = the mean across subjects and products on covariate k

Xkij = subject i's rating of covariate k for J product j.

As shown in Table 3, three covariates were significant at or beyond v = .10 and entered the analysis. In order of significance they were the number of brands that the women were aware of (counted up), their stated tendency toward brand loyalty (measured by loyalty in Appendix 1), and the degree to which different brands were utilized in different consumption situations (measured by situate in Appendix 1).

These results bolster Campbell (1968), Ostlund (1973) and Gronhaug's (1973-74) conclusions that brand loyalty is an important determinant of evoked set size. However, there is a tautology implicit in this finding. Brand loyal consumers, by definition, must only buy from a small subset of available brands. As a result, it is not surprising that they have smaller evoked sets; evoked set size being defined in terms of the number of brands considered. Small evoked sets and brand loyalty may actually be different measures of the same behavior, i.e., selecting from a tightly constrained subset of available brands.

TABLE 3

ANALYSIS OF COVARIANCE RESULTS FOR PRODUCT FACTORS

The finding that evoked set size is positively correlated with the size of the awareness set is, indeed, more interesting. This empirical finding which is consistent with earlier results published by Markin and Narayana (1975), would appear to contradict Howard and Sheth's (1969) prediction that no relationship would exist between the number of known brands and the number of brands in the evoked set. This result which suggests that the size of consumer's evoked sets may expand as their awareness set is enlarged should be of considerable interest to those managers concerned with the introduction of new brands to the marketplace. Their efforts to gain membership in target consumer's evoked sets would appear to be greatly facilitated if, in fact, consumers simply expand their evoked set to accommodate new brands rather than replacing an old brand with the new one as implied by Howard and Sheth.

The finding that evoked set size is related to the degree to which the individual feels that different brands are appropriate; different situations is new to the evoked set literature. Respondents who felt that there were different brands suitable for different consumption situations appeared to have larger evoked sets. Belk has repeatedly argued the importance of situational influences on consumption (Belk 1974, 1975; Clarke and Belk 1979), but little work has been done on exactly how situational factors influence consumer decision process. This study would appear to support Belk's contention in the case of evoked sets.

The three previously mentioned product factors taken together appear to account for most of the differences in evoked set size across all eight produce categories. This is evidenced by the fact that inclusion of these three covariates in the model results in a non-significant F-ratio associated with the product effect in Table 3. However, it should be noted that this finding is limited by the degree to which the observed relationships may reflect underlying causal variables that were not measured, but are correlated with both the covariates and the criterion variables.

Individual Factors

Given the significant F-ratio associated with the subJect effect in Table 1, it was also appropriate to explore the underlying individual factors related to evoked set size. For this purpose a second stepwise analysis of covariance was carried out. In this case the covariates were added to the linear model which was as follows:

Yi = m + ti + aj + tiaj + bk (Xki-Xk), where (2)

bk = the regression coefficient for covariate k

Xk = the mean across subjects on covariate k

Xki = subject i's score on covariate k

As shown in Table 4, only two individual factors, education and family size were significant at or beyond a = .10. These results support the earlier work by Gronhaug (1973-74) that demonstrated that there was a positive correlation between educational level and the number of brands considered for purchase. It is interesting, however, that the other cognitive measures included in the study were not significantly related to evoked set size in the produce categories measured. It is possible that past studies which found these types of variable useful in explaining evoked size may well have been capitalizing on the intercorrelation between most of these measures and educational level. Further research is needed to replicate these findings with a broader sample of respondents and across a broader range of product categories.

TABLE 4

ANALYSIS OF COVARIANCE RESULTS FOR INDIVIDUAL FACTORS

The significant correlation between family size and evoked set size is interesting when viewed in light of the previously reported finding regarding situational influences. One could postulate that within larger families, the products included in this study would have a variety of uses among different family members. Since different family members may be using products differently and may be seeking different benefits from their use; family size may be another measure of situational influence.

The significance of the F-ratio associated with the subject factor, even after the removal of the covariate effects, indicates, however, that additional work is needed to find individual factors that account for individual differences in evoked set size

CONCLUSIONS AND FUTURE RESEARCH

The study reported here is the most comprehensive study of evoked set size among consumer package goods reported to date. The study covered a larger number of product categories than any previous work, and was based on the responses of subjects who were familiar and experienced purchasers of these products (female homemakers.)

The results, while not abundant, appear to support the proposition that the number of brands considered by consumers is affected by both individual and product characteristics specifically, brand loyalty, awareness set size, situational and use factors and education level. However, additional research is needed especially in the area of individual factors to complete the picture.

However, as intriguing as the current findings may be, they have not resolved the fundamental issue associated with the concept of evoked sets. Howard and Sheth viewed the evoked set for a given brand as essentially fixed. When the time came for the consumer to make a choice, she/he referred to the evoked set to see which brands needed to be considered. The evoked set itself existed as essentially a static representation of the consumers past decision making and experience with the product category. Not much theorizing went into the creation and modification of the evoked set, or even into the notion of whether an evoked set would even exist for infrequently purchased items.

A less static view comes from researchers studying consumer information processing. In this research stream, the consumer is viewed operating much like a computer following the dictates of some algorithmic information processing strategy. Evoked set, if it exists at all, is simply the current contents of memory in the processor. Brands that are still under consideration (the evoked set) are those still remaining in the buffer. Other brands have not yet been considered (the inept set) or have been rejected (the inept set). In one version of this line of thinking, the evoked set is merely the middle phase of a decision strategy where the first step is to select some subset of available brands and the next step is to choose the best from among the subset. The evoked set from this process is fundamentally different from that suggested by the static view

A Key piece of research which needs to be done is to longitudinally measure the evoked set as a sample of consumers go through their decision processes, optimally for durable as well as package goods. If the brands under active consideration remain fairly consistent, there would be some support for the static notion of the evoked set as stored positive reactions to brands. If, however, there is variability in the contents of "brands currently under consideration" then an information processing view might be more attractive. It may even be that both views are supported for different products, and/or by different consumers.

APPENDIX 1

MEASURES USED

REFERENCES

Belk, Russell W. (1974), "An Exploratory Assessment of Situational Effects in Buyer Behavior," Journal of Marketing Research, 11 (May), 156-63.

Belk, Russell W. (1975), "Situational Variables and Consumer Behavior," Journal of Consumer Research, 2 (December), 157-67.

Belonax Jr., Joseph J. (1979), "Decision Rule Uncertainty, Evoked Set Size, and Task Difficulty as a Function of Number of Choice Criteria and Information Variability," in Advances in Consumer Research, 6, William L. Wilkie, ed., Ann Arbor: Association for Consumer Research, 232-5.

Belonax Jr., Joseph J., and Robert A. Mittelstaedt (1978), "Evoked Set Size as a Function of Choice Criteria and Information Variability," in Advances in Consumer Research, 5, H. Keith Hunt, ed., Ann Arbor: Association for Consumer Research, 48-51.

Bieri, J. and A. L. Atkins, S. Briar, R. L. Leaman, H. Miller, and T. Tripodi (1966), Clinical and Social Judgement: The Discrimination of Behavioral Information, New York: John Wiley and Sons.

Brisoux, Jacques E. and Michel Laroche (1980), "Evoked Set Formation and Composition: An Empirical Investigation Under a Routinized Response Behavior Situation," in Advances in Consumer Research, 8, Kent B. Monroe, ed., Ann Arbor: Association for Consumer Research, 357-361.

Campbell, Brian M. (1969), "The Existence of Evoked Set and Determinants of It's Magnitude in Brand Choice Behavior," unpublished Ph.D. dissertation, Columbia University.

Clarke, Keith and Russell W. Belk (1979), "The Effects of Product Involvement and Task Definition on Consumer Efforts," in Advances in Consumer Research, 6, William L. Wilkie, ed., Ann Arbor: Association for Consumer Research, 313-18.

Cohen, T. and P. Cohen ( 1975), Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences, Hillsdale, NJ: Lawrence Erlbaum Associates.

Cox, Donald F. (1967), "The Influence of Cognitive Needs and Styles," in Risk Taking and Information Handling in Consumer Behavior, Donald F. Cox, ed., Boston: Harvard University, 370-93.

Day, Robert C. and Robert L. Hamblin (1964), "Some Effects of Close and Punitive Styles of Supervision," American Journal of Sociology, 69 (March), 499-511.

Dixon, W. J. and M. L. Brown (1979), Biomedical Computer Programs P-Series, Los Angeles: University of California Press.

Gronhaug, Kjell (1973-74), "Some Factors Influencing the Size of the Buyer's Evoked Set," European Journal of Marketing, 7, 232-41.

Jarvis, Lance P. and James B. Wilcox (1973), "Evoked Set Size - Some Theoretical Foundations and Empirical Evidence," in Combined Proceedings: Fall Conference of the American Marketing Association, Thomas V. Greer, ed., Chicago: American Marketing Association, 236-40.

Howard, John A. (1963), Marketing Management: Analysis and Planning, rev. ed., Chicago: Richard D. Irwin.

Howard, John A., and Jagdish N. Sheth (1969), The Theory of Buyer Behavior, New York: John Wiley and Sons.

Maddox, R. Neil, Kjell Gronhaug, Richard D. Homans, and Frederick E. May (1978), "Correlates of Information Gathering and Evoked Set Size for New Automobile Purchasers in Norway and the U. S.," in Advances in Consumer Research, 5, H. Keith Hunt, ed., Ann Arbor: Association for Consumer Research. 167-70.

May, Frederick E. and Richard E. Homans ( 1977), "Evoked Set Size and the Level of Information Processing in Product Comprehension and Choice Criteria," in Advances in Consumer Research, 4, William D. Perreault, Jr., ed., Ann Arbor: Association for Consumer Research, 172-5.

Narayana, them L. and Rom J. Markin (1975), "Consumer Behavior and Product Performance; An Alternative Conceptualization," Journal of Marketing, 39 (October), 1-6.

Ostland, Lyman E. (1973), "Evoked Set Size: Some Empirical Results," in Combined Proceedings: Fall Conference of the American Marketing Association, Thomas V. Greer, ed., Chicago: American Marketing Association, 226-30.

Parkinson, Thomas L. (1979), "An Information Processing Approach to Evoked Set Formation" in Advances in Consumer Research, 6, William L. Wilkie, ed., Ann Arbor: Association for Consumer Research, 227-31.

Pethgrew, Thomas F. (1958), "The Measurement and Correlates of Category Width as a Cognitive Variable," Journal of Personality 26 (December), 532-44.

Prasad, U. Kanti (1975) "Evoked Set Size-Personality Correlates and Mediating Variables," Journal of the Academy of Marketing Science, 3 (Summer), 272-99.

Troldahl, Verling C. and Frederic A. Powell (1965), "A Short-Four Dogmatism Scale for Use in Field Studies," Social Forces, 44 (December), 211-14.

Williams, Robert G. and Michael J. Etzel (1976), "An Investigation and Extension of the Evoked Set Concept Applied to Consumer Durables," in Proceedings of the Southern Marketing Association, Henry W. Nash and Donald P. Robin, eds., Starkville, MI: Mississippi State University.

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