Influence of Expertise and Purchase Experience on the Formation of Evoked Sets

ABSTRACT - The role of expertise in consumer decision-making is explored in relation to the dynamics of evoked set formation and transformation. Investigation of the relationship of knowledge about alternatives, task knowledge, width and depth of previous purchase history provides a more detailed picture of the various influences on the brands considered during the purchase process.


Girish Punj and Narasimhan Srinivasan (1989) ,"Influence of Expertise and Purchase Experience on the Formation of Evoked Sets", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 507-514.

Advances in Consumer Research Volume 16, 1989      Pages 507-514


Girish Punj, University of Connecticut

Narasimhan Srinivasan, University of Connecticut


The role of expertise in consumer decision-making is explored in relation to the dynamics of evoked set formation and transformation. Investigation of the relationship of knowledge about alternatives, task knowledge, width and depth of previous purchase history provides a more detailed picture of the various influences on the brands considered during the purchase process.


Expertise and its role in the consumer decision making process has been receiving increasing research attention lately (Alba and Hutchinson 1987; Chi, Glaser and Rees 1982). As noted by several researchers, expertise primarily consists of two types of knowledge. The first type deals with how knowledge is structured and processed, and includes the related functions of cognition, encoding, retrieval, and information usage in the performance of a choice task. The second type deals with the content relating to the decision problem, i.e. knowledge about the different alternatives available for choice (Brucks 1986; Dacin and Mitchell 1986).

The concept of the evoked set has received continued attention since its introduction into marketing by Howard (1963). Several terms, including available set, feasible set, awareness set, consideration set and choice set have been used to depict the various sets of alternatives which a consumer uses while making a choice.

The two constructs of expertise and evoked set are clearly related. Increased product knowledge implies a greater awareness of alternatives and the ability to assimilate more product information, thereby influencing the size and composition of the brands considered for purchase. The exact nature of this relationship is, however, not fully understood. This paper seeks to investigate the relationship between these two important constructs, thereby furthering our understanding of an important phase of the consumer decision making process. We propose and test several hypotheses relating the two constructs with a view toward formulating a preliminary model of the relationship between them.

Importance of the topic

Expertise plays a critical role in structuring the choice problem and in the consideration of alternatives for purchase. A major consequence of a consumer's expertise is the formation of the evoked set soon after the problem recognition stage in the consumer decision making process. As the consumer progresses through the choice process, expertise plays a role in how the evoked set is transformed into the final choice set, as a result of the processing of new information. Eventually an alternative is selected. Research which focuses on the interplay between expertise and evoked set is thus likely to enhance our understanding of the how a brand enters and is retained in the evoked set. From a practical standpoint, this is a significant topic for study, given that much of the task of practitioners is to seek inclusion and retention of the company's brand in the consumer's choice set.

Definitions of Expertise

Consumer researchers have long recognized the importance of expertise, though not in an elaborate manner. The construct, in its various forms of product knowledge, attention to new information, etc., is a part of the comprehensive models of consumer behavior proposed Howard and Sheth (1969), Hansen (1972) and Bettman (1979).

Chi, Glass and Rees (1982) defined expertise as "the possession of a large body of knowledge and procedural skill." Following a similar perspective, Johnson and Russo (1984) suggested that familiarity with a product class results in three different kinds of expertise: (1) a superior knowledge of existing alternatives (2) a superior ability to encode new information and (3) a superior ability to discriminate between relevant and irrelevant information.

Punj and Staelin (1983) refer to two aspects of knowledge, related to expertise. One is the amount of prior knowledge directly associated with the available alternatives, and the other is information stored in memory which helps a consumer obtain and process newly acquired information. Brucks (1985) also proposed two components of expertise: objective and subjective knowledge. Subjective knowledge can be thought of as including an individual's degree of confidence in his/her product knowledge while objective knowledge refers to only what the individual actually knows about the choice alternatives.

Alba and Hutchinson (1987) define expertise as "the ability to perform product related tasks successfully (p 411)." They explicitly distinguish expertise from familiarity and define the latter as "the number of product related experiences that have been accumulated by the consumer."

Multi-dimensionality of Expertise

Though conceptualizations of the domain of expertise have been varied, two important components of the construct may be recognized and identified: (1) the possession of alternative specific information, and (2) a more abstract level ability to perform the choice task. These two aspects of expertise exist across the various definitions of the construct. The first aspect represents the information which the consumer possesses on the choice alternatives or information which is directly and immediately suitable for the choice decision. The second aspect deals with the consumer's knowledge structure and represents the ability to deal with the procedural aspects of the choice task, such as drawing inferences from information which may not be exactly suitable for the current decision. In the present study, we use two dimensions of expertise similar to the ones outlined above. These components may be referred to as "alternatives knowledge" and "task knowledge" respectively.

Another element of expertise, relates to the knowledge gained through past purchase experiences in a product category. While the exact feedback mechanism from previous purchases to current knowledge is neither simple nor clear, there can be little doubt that there is such an influence.

One approach to understanding this influence is to partition its effect along two dimensions. In this research we use the term "width of purchase experience" to denote the diversity of an individual's purchase history in terms of the number of different brands the person has owned in the past. A complement variable "depth of purchase experience" is used to signify the total number of purchases an individual has- made in the product category.

Definitions Of Evoked Set

Since the introduction of the term by Howard (1963), the concept of an evoked set has become an accepted part of consumer decision making models (Bettman 1979; Howard and Sheth 1969). Howard (1977) defined evoked set as "the subset of brands that a consumer considers buying out of the set of brands that he or she is aware of in a given product class." Campbell (1973) defined evoked set as "the set of brands of a product which the buyer actually considers when making a specific brand choice." Silk and Urban (1978) use the term "relevant set" of alternatives to describe the "subset of available brands which are familiar to the respondent regardless of whether they are judged favorably or unfavorably as choice alternatives. A more comprehensive conceptualization of the concept is provided by Narayana and Markin (1975) who proposed three different sets of brands relevant to a choice decision: evoked set, inert set and inept set. The evoked size consists of the select brands the consumer considers while making a purchase as a result of having given them a positive evaluation.

The above definitions fail to be precise about the point in time at which the evoked set is being measured and also tend to view it as a static construct. Myers (1978) has suggested the need for a longitudinal analysis of changes in the evoked set to model its dynamic nature.

Multiple Conceptualizations of Evoked Set

The definitions given above vary from one another, in terms of the interpretation that may be given to the notion of a brand being "acceptable" for purchase. Also, it is clear that the time dimension is an important element in the definition of an evoked set. Of the various characteristics of evoked sets that have been studied, it appears that the size and composition of the evoked set and how they change over time capture the essence of the construct.

One way of modeling the dynamics of evoked set formation is to use two or more measurements of the construct during the course of the decision process. In this paper, we conceptualize the "initial evoked set" to represent the set of brands which a consumer considers soon after the onset of problem recognition. The term "final evoked set" is used to refer to the set of brands which a consumer considers just prior to purchase. Such a conceptualization while being meaningful is, however, hard to operationalize. One potential difficulty is identifying the duration of the purchase decision in natural settings, such as in new car buying. The dependence on recall information appears to be unavoidable. A second potential problem is that the size of the evoked set is easier to measure than its composition, which has earlier been reasoned to an important dimension.


The hypotheses regarding the effects of expertise on evoked set are based on prior empirical findings using these constructs. They are exploratory and should be viewed as contributing to theory development since the available previous research is not conclusive enough to permit definitive predictions concerning the relationships between expertise and evoked set. The hypotheses center around the two (alternative) explanations that have been forwarded about the effects of product knowledge on the subsequent processing of information.

According to the "facilitating explanation," consumers with greater expertise are more capable of considering a broader range of choice options because they possess the requisite information processing skills. Thus we would expect these consumers to have large initial choice sets. It is possible to argue that only a particular type of product knowledge, i.e. that directly related to the alternatives is likely to facilitate the formation of a large initial evoked set. A statement of this proposition is:

H1: Higher levels of knowledge about alternatives are more likely to be associated with larger initial evoked sets.

An alternate view of the effects of product knowledge on subsequent information processing is provided by the "efficiency explanation." Under this justification, consumers with a greater degree of expertise are likely to have developed superior abilities to discriminate among alternatives and are thus likely to have a lesser need to consider several of them, avoiding the need for additional information search and deliberation. Hence, one can postulate that experience with the procedural aspects of making a choice decision is likely to result in the formation of a smaller initial evoked set due to efficiency considerations. Hence.

H2: Higher levels of task knowledge are more likely to be associated with smaller initial evoked sets.

Once the initial evoked set has been formed, consumers are likely to add and/or delete alternatives to their choice sets as a consequence of additional information processing. Few (if any) research findings are available on how the initial evoked set is transformed to arrive at the final evoked set. At one extreme is the possibility that there is little difference between the initial and final evoked sets in terms of both size and composition. At the other extreme it is plausible that the final choice set bears little resemblance to the initial evoked set in terms of content or size. In the absence of any guidance, the hypotheses relating expertise to final evoked set should be viewed as speculative. Under one likely scenario, consumers with a higher level of expertise are likely to have small final evoked sets, because of their ability to narrow down choice options to a final few. Consumers with little product knowledge may not have this ability, leading to an increase in the size of the alternatives considered during the decision making process. Such a reasoning would be consistent with both the "facilitating" and "efficiency" explanations of product knowledge discussed earlier. Hence:

H3: Higher levels of alternative knowledge and task knowledge are likely to lead to smaller final evoked sets.

As mentioned earlier, the diversity of a consumer's previous purchase experiences is another element which influences evoked sets. Perhaps reflecting a brand loyalty effect, the narrower the assortment of brands a consumer has owned in the past, the more likely it is that s/he will consider a smaller set of alternatives, leading to:

H4: Greater depth of purchase experience is likely to be associated with smaller initial evoked sets.

Another aspect of the influence of expertise on evoked sets relates to the diversity of the consumer's previous purchase experiences. The broader the assortment of alternatives a consumer has purchased, the more likely it is that s/he will consider a wider range of choice options at the start of the decision process, provided that preferences for the most recent brand bought are not dominant. Hence,

H5: Greater width of purchase experience is likely to be associated with larger initial evoked sets.

The impact of previous purchase experience on the final evoked set is more difficult to discern. One possibility is that consumers with a narrow (i.e. less diverse) purchase history are not likely to alter their evoked sets from the initial to the final stages of the decision process thus causing them to have smaller final evoked sets. However, those with a diverse purchase history are also likely to arrive at smaller final evoked sets, as a result of pruning down their choice sets through an elimination and selection process, rather than selection alone as in the former case. In fact, the only type of consumer likely to have a large final evoked set is the one who is unable to narrow down the choice options either because of a lack of previous purchase experience or expertise (implied in H3). Hence:

H6: Limited purchase experience is likely to be associated with larger final evoked sets.

Taken together, the hypotheses postulate relationships between expertise, previous purchase experience and evoked sets. The relationships are listed and depicted pictorially in Figure 1, which may be viewed as a flowchart for representing the effects of expertise on the formation of evoked sets.


The data used for this research were collected as part of a study of new car buyers conducted in three geographically separate major metropolitan markets. Potential respondents were first contacted by telephone to solicit their participation in the study, with a view towards enhancing response rate. Approximately 2400 questionnaires were mailed and 1046 usable responses were received. Prior to their submission for analyses, the responses were subjected to an extensive series of checks to identify possible biases, which could be controlled for during the analysis stage (for details, see Punj and Staelin 1983). In almost all instances the data held up well to the verification tests. However, like in most survey research, it is difficult to assess the respondent's ability to retrospectively recall the information requested in the data collection instrument, given that some of the questionnaires were completed up to six months after the purchase. A test to reveal any systematic bias due to forgetting (across subsamples) did not turn up any differences.

Information was requested on the two aspects of expertise (alternatives knowledge and task knowledge), purchase histories and evoked sets identified earlier. The exact operationalizations used are provided in Table 1. In this context it should be mentioned that the two expertise variables, while conceptually meaningful, are only psychological scale approximations of a complex construct. Thus the results corresponding to them should be interpreted with caution. For the purchase experience and evoked set variables there is a better correspondence between the measures employed and the underlying constructs. However, it should be kept in mind that the distinctions between "initial" and "final" should be viewed only in a relative sense, because of the difficulty of anchoring the consumer decision process in very precise terms.


The plan of analysis adopted for testing the hypotheses called for the sequential use of simple correlation and canonical correlation analysis. It was felt that these techniques were particularly suited given the nature of the data, the type of relationships being tested, and the desire to test all the hypotheses collectively, rather than in a piecemeal fashion.



Table 2 displays the means and standard deviations for all the variables employed in the analysis. The variables used to measure expertise appear to exhibit the most variation, while those used to measure the size of the final evoked set and the depth of a consumer's past purchase experience appear to vary the least. The pearson correlations for the variables of interest are shown in Table 3. Most of the coefficients are significant at the p < 0.05 level.

An examination of the table shows the two expertise variables to be correlated with each other (r = .30, p < .001). This was to be expected due to the definitional overlap in these variables. Also, the two experience variables are positively correlated with the expertise measures, while the latter are negatively correlated with the evoked set variables. Thus, one might be led to expect a negative correlation between the evoked set and the experience variables. However, this appears not to be the case. The experience variable representing the diversity of a consumer's previous purchase experiences is positively correlated with both the evoked set variables. Broadly speaking, the interpretation which may be ascribed to the correlation coefficients is that both depth and width of previous purchase experience affect the development of expertise. But they are not singular influences, since only the expertise variables appear to be related to the formation of smaller evoked sets.

In order to further understand the nature of the relationships among the constructs of interest and to provide a collective test of the hypotheses, the matrix of correlation coefficients were subjected to a canonical correlation analysis (using BMDP6M) with the evoked set variables comprising the criterion set. The strength of the relationships between the predictor set of variables (expertise and purchase experience) and the evoked set variables can be assessed by examining the canonical correlation coefficients, canonical weights and loadings and the canonical cross loadings depicted in Table 4.







The canonical correlation coefficient of (r = 0.29) provides a measure of the overall association between two linear combinations of variables (canonical variates) constructed from the predictor and criterion sets respectively. The square of this statistic (0.084) can be interpreted as the amount of variance explained in the evoked set variables by the purchase experience and expertise variables. The canonical weights are indicative of the relative contribution of each variable to the linear combination (canonical variate) of which it is a part. The canonical loadings can be interpreted as within set variable-variate correlations, while the canonical cross loadings represent the between set variable-variate correlations.

An examination of the entries in Table 4 shows that both the expertise variables appear to be inversely related (cross loadings = - 0.18 and - 0.11) to the two evoked set variables. Hence there is support for hypotheses 3 and 5 but not for hypothesis 1, thus indirectly advancing the efficiency explanation of the role of knowledge on the subsequent processing of information. The relationship between the purchase experience and evoked set variables shows that purchase experience (width) has a positive (cross loading = 0.20) while purchase experience (depth) has a negative (cross loading = - 0.01) relationship, thus providing support for hypotheses 2 and 4. However, the effect of the depth of experience is very weak and the results appear to be driven by the width of consumer's past purchase experience. There appears to be only marginal support for hypothesis 6, with purchase experience (depth) showing a mild negative relationship with the evoked set variables.

Although not explicitly stated, a positive relationship between the initial evoked set and final evoked set is implied by the hypotheses, subject to the relative magnitude of the effects in hypotheses 1 through 4. Such a relationship can be observed in the criterion variables set (the simple correlation of 0.58 is significant; the canonical loadings are 0.91 and 0.87 for the initial and final evoked sets respectively).


Overall, the results appear to support the relationships postulated in the Figure. However, as indicated earlier the hypotheses are exploratory and can best be viewed as contributing to theory development, rather than providing confirmatory evidence of one. In light of the results obtained, it seems that there is strong support for the efficiency explanation of how expertise influences the evoked set formation process. Consumers appear to use their knowledge to narrow down their consideration set, even at the very initial stages of the decision process.

If expertise facilitates the consideration of additional choice alternatives, this probably occurs only after the initial choice set has been formed. The influence of past purchase experience is less certain. For one, it appears that only the diversity of one's purchase history influences the formation of the initial evoked set. Furthermore the effect appears to be counter to the influence of expertise, thus supporting the facilitating explanation of knowledge on the subsequent processing of information. Here it appears that consumers use their past purchase experiences to broaden their choice set of alternatives at the early stages of the choice process. Thus, if past experience is used to limit the consideration set, this occurs only after the purchase process is well underway.

A simple explanation which is consistent with the results (but is not necessarily the cause of them) is that consumer's rely on the diversity of their purchase experiences to form their initial evoked sets. Once this is accomplished, they use their expertise to narrow down their choice options as they proceed along the decision process.



The nature of the transformations occurring on the evoked set from the initial to the final stages is a fruitful area for additional research. Of particular interest would be a conceptualization which accounts for both size and composition changes in the evoked set over the course of the decision process. Such a model would permit a more precise characterization of the various expertise and purchase experience effects explored in this study.

Future research directions include generalizable studies using different samples and different product categories. Two critical issues to be borne in mind are (1) the need to develop/test alternate and possibly improved measures of expertise and experience using scale development procedures, and (2) the need for process research, rather than the traditional static, cross-sectional analyses. The latter approach permits integration with research studies in message comprehension, persuasion, recall and product choice.


Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise," Journal of Consumer Research, 13 (March), 411-54.

Bettman, James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley

Brucks, Merrie (1985), 'The Effects of Product Class Knowledge on Information Search Behavior," Journal of Consumer Research 12 (June), 1-16.

Brucks, Merrie (1986), "A Typology of Consumer Knowledge Content," in Richard J. Lutz (ed.) Advances in Consumer Research, Vol. 13, 58-63.

Campbell, Brian M. (1973), 'The Existence of Evoked Set and Determinants of Its Magnitude in Brand Choice Behavior," in John A. Howard and Lyman E. Ostlund, eds. Buyer Behavior: Theoretical and Empirical Foundations, NY: Alfred Knopf

Chi, Michelene T. H., Robert Glaser and Ernest Rees (1982), "Expertise in Problem Solving," in K. J. Sternberg (ed.), Advances in The Psychology of Human Intelligence, Vol. I, Hillsdale, NJ: Lawrence Erlbaum Associates.

Dacin, Peter A. and Andrew A. Mitchell (1986), "The Measurement of Declarative Knowledge," in Richard J. Lutz (ed.) Advances in Consumer Research, Vol. 13, 454-59.

Hansen, Fleming (1972), Consumer Choice Behavior: A Cognitive Theory, NY: Free Press

Howard, John A. (1963), Marketing Management: Analysis and Planning, Homewood, IL: Richard Irwin

Howard, John A. (1977), Consumer Behavior: Application of Theory, NY: McGraw Hill.

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

Johnson, Eric and J. Edward Russo (1984), "Product Familiarity and Learning New Information," Journal of Consumer Research, 11 (June), 542-50.

Myers, James H. (1978), "Methodological Issues in Evoked Set Formation and Composition," in William L. Wilkie (ed.) Advances in Consumer Research, Vol. 6, 236-37.

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

Punj, Girish N. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," Journal of Consumer Research, Vol. 9 (March), 366-80.

Silk, Alvin J. and Glen L. Urban (1978), "Pre-Test Market Evaluation of New Packaged Goods: A Model and Measurement Methodology," Journal of Marketing Research, 15, 171-91.



Girish Punj, University of Connecticut
Narasimhan Srinivasan, University of Connecticut


NA - Advances in Consumer Research Volume 16 | 1989

Share Proceeding

Featured papers

See More


How the Unconstructed Identity Relieves Consumers of Identity-Relevant Consumption

Tracy Rank-Christman, University of Wisconsin - Milwaukee, USA
Lauren Poupis, Iona College

Read More


Why is 1 out of 20 Riskier than 5%? Effect of Representing Unlikely Events as Frequency versus Percentage on Risk Perceptions

Nevena T Koukova, Lehigh University
Joydeep Srivastava, Temple University, USA

Read More


The “Upper Limit Framing” Effect: Upper Limit Framing of a Cost Estimate Influences Consumption Choices

Sudipta Mukherjee, Virginia Tech, USA
Frank May, Virginia Tech, 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.