The Relationship Between Prior Knowledge and External Search

ABSTRACT - A number of studies have explored the relationship between prior knowledge and search. These studies have produced results which indicate conflicting hypotheses. Part of this conflict may be attributable to the lack of clear and concise definitions of relevant constructs and the use of different measures to define them. In this study we examine the relationship between prior knowledge and external search, using different measures for each construct We demonstrate that the relationship between the two constructs is sensitive to the measures employed.


Narasimhan Srinivasan and Jagdish Agrawal (1988) ,"The Relationship Between Prior Knowledge and External Search", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 27-31.

Advances in Consumer Research Volume 15, 1988      Pages 27-31


Narasimhan Srinivasan, University of Connecticut

Jagdish Agrawal, State University of New York at Buffalo


A number of studies have explored the relationship between prior knowledge and search. These studies have produced results which indicate conflicting hypotheses. Part of this conflict may be attributable to the lack of clear and concise definitions of relevant constructs and the use of different measures to define them. In this study we examine the relationship between prior knowledge and external search, using different measures for each construct We demonstrate that the relationship between the two constructs is sensitive to the measures employed.


In any buying situation, a consumer may acquire information from two sources. One source, commonly termed 'prior knowledge' refers to the inventory of information acquired prior to the purchasing situation and stored in memory. The second source of information, external search, embodies any source external to memory, such as store visits, friends and advertisements. Thus, the total amount of information available to a consumer making a purchase decision is the combination of both internal and external search. Intuitively, in a purchase situation, if a consumer already has a substantial amount of relevant information stored in memory, that consumer may be expected to engage in less external search compared to another consumer with a smaller inventory of prior knowledge. However, the empirical findings regarding the relationship between internal and external search have often yielded conflicting results.

A number of studies have examined the impact of the prior knowledge on external search (Punj and Staelin, 1983; Brucks, 1985) and choice processes (Bettman and Park, 1980; Park and Lessig, 1981; Johnson and Russo, 1984; Srull, 1983). Though these studies have made significant contributions to our understanding of the relationship between internal and external search, efforts to develop a theory which defines and explains the relationship between external search and prior knowledge are hampered by the multiplicity of measures used by different researchers and contradictory findings (Brucks, 1985). The purpose of the present study is to report some preliminary findings relating to the relationship between different measures of prior knowledge and external search. We intend to show that, in the absence of any succinct definition of prior knowledge and external search, the findings are highly measure sensitive.

First we present a number of hypotheses reflecting the nature of the observed relationships between knowledge and search found in various empirical studies. Then we discuss the various measures used to operationalize knowledge and search. Finally, we present empirical results showing the relationships between knowledge and search, using different measures for each construct.


The empirical studies on the relationship between prior knowledge, with or without experience, and external search have produced very inconsistent findings. A review of the results of these studies leads us to extract the following hypotheses:

Hypothesis 1: Prior knowledge increases the ability to process new information, and, therefore, facilitates search. (Jacoby, 1978; Johnson and Russo, 1984; Jacoby et al., 1978)

Hypothesis 2: Prior knowledge reduces the motivation to search i.e. consumers having a high inventory of prior knowledge engage in less external search relative to those consumers with a low inventory of prior knowledge. (Katona and Mueller, 1955; Bucklin, 1966; Newman and Staelin, 1971, 1972; Moore and Lehmann, 1980; Bettman and Park, 1980)

Hypothesis 3: Prior knowledge about a product class generates efficiency in information processing. i.e. consumers having a high level of prior knowledge know which attributes to use in evaluation and are, therefore, quicker at eliminating irrelevant alternatives. The result would be less external search for those with high prior knowledge (Brucks, 1985).

Hypothesis 4: Prior knowledge exhibits an inverted U shaped relationship with external search. Consumers with little prior knowledge find processing of new information to be overwhelming Consumers with some prior knowledge find that it facilitates the processing of further information and hence, the relationship between knowledge and search is positive at first. Consumers with high prior knowledge have less motivation to process new information. Consumers with high prior knowledge are also efficient and thus, engage in less external search. The overall relationship is, therefore, positive at first and then negative. (Bettman and Park, 1980; Johnson and Russo, 1984; Brucks, 1985).

These competing hypotheses, drawn from the empirical studies reviewed, clearly show that the relationship between prior knowledge and search is not straight forward. Knowledge affects search through ability, efficiency, and motivation to search. In an excellent discussion of the nature of the relationship between knowledge and search, Alba and Hutchinson (1987) propose that knowledge and search are multidimensional constructs. Perhaps, the hypotheses presented above would not be in conflict with each other if the dimensions of knowledge and search under study were well defined and measured. Hence, one approach to understanding the seemingly conflicting results of different studies is to examine the measures used to operationalize knowledge and search. ID the following section, we present a review of some major studies with specific reference to the measures employed.

Prior Knowledge

The construct 'prior knowledge' has been conceptualized in a variety of ways. Its domain has included familiarity with the product class, knowledge acquired without experience and knowledge acquired as a result of experience. Prior knowledge has been dichotomized into subjective knowledge and objective knowledge. The corresponding measures of this latent construct differ significantly across studies.

Bettman and Park (1980) have defined prior knowledge as knowledge acquired both through experience and knowledge acquired without experience. In their study, a sample of 62 housewives were classified into three categories. The low knowledge group consisted of those respondents who had never searched for, used or owned the product (microwave ovens). The moderate knowledge group had searched for and used the product but had never owned one. The high knowledge group consisted of respondents who had searched for, used and owned the product.

Punj and Staelin (1983) used experience and knowledge as two distinct constructs in their model of information search. Experience was captured in one construct, termed 'usable prior information'. This construct included experience with either the product class (automobiles) or with the purchase task. The construct was operationalized by using three measures (I) satisfaction with previous cars, (2) time elapsed between purchase of the two most recently purchased cars and (3) number of used and new cars bought in the previous 10 years. The second construct, 'prior memory structure', reflected knowledge of the buying process and the product class. Prior memory structure also had three measures: (1) number of cars bought in the last 10 years, (2) knowledge about cars acquired through magazine readership, and (3) education level.

It is interesting to note that Punj and Staelin (1983) did not find a high correlation between the two constructs of experience and knowledge. Other researchers (Bettman and Park, 1980) have used a single construct to capture both experience and knowledge. Yet the desirability of distinguishing between experience and knowledge is illustrated by Punj and Staelin's (1983) finding that experience had a negative influence upon external search, whereas, knowledge had a positive but insignificant influence on external search.

Punj and Staelin (1983) tested the impact of knowledge alone on search through OLS regression. Prior memory structure was measured as the weighted average of (a) a linear combination of the three measures outlined earlier, and (b) a single item self-report measure of ability to judge cars. This second component is similar to Brucks' (1985) measure of subjective knowledge. Both the linear and inverted U relationship were tested. The curvilinear relationship reported by Bettman and Park (1980) did not come out to be significant, whereas prior memory structure, which was not significant in the multivariate setting (causal model), turned out to have a significant positive influence upon the amount of external search. Given the different implicit definitions of the construct developed by each of the authors cited above, and the different measures employed, the conflicting results are not surprising.

Johnson and Russo's (1984) study deals with learning. i.e. the amount of information remembered after search, rather than search itself. The authors have labeled the knowledge construct 'familiarity', but the measures employed reflect knowledge, either with or without experience. The impact of learning, as a consequence of search, examined in this study provides some insights into the relationship between knowledge and search. Familiarity was operationalized using three measures: (1) self-report about product class knowledge measured on a 5-point rating scale (2) the number of cars owned and (3) the number of cars ridden in. Thus, Johnson and Russo's construct, "familiarity" reflects both subjective knowledge (Brucks, 1985) and experience with the product class.

Brucks (1985) argued that knowledge has two correlated dimensions - subjective knowledge and objective knowledge (Park and Lessig, 1981; Rudell, 1979) - and that these dimensions would have differing degrees of impact upon external search depending upon the usage situation. Objective knowledge was measured as the summated scale of responses to a variety of questions relating to the following: terminology relevant to the product (sewing machine), available attributes, criteria for evaluation, attribute covariation with prices and usage situations. Subjective knowledge was measured as the summed score of two items - one indicating self rating of product class knowledge and the other indicating product class familiarity. Brucks argued that subjective knowledge generates more confidence in an individual and helps eliminate alternatives, whereas objective knowledge increases one's ability to process attributes. In this context, it is relevant to note that Pun; and Staelin (1983) found a correlation of 0.03 between 'usable prior information' and 'prior memory structure'. Based on the low correlation, the authors observed that "consumers who possess the ability to process new information do not necessarily have a substantial amount of directly relevant information already stored in memory" (p 377).

The measures of prior knowledge presented above capture different components of the construct of prior knowledge. Bettman and Park (1980) and Johnson and Russo (1983) measure a construct which reflects knowledge with or without experience whereas Punj and Staelin (1983) include ability in addition to knowledge with or without experience. Brucks' (1985) measures include experience with the product class in general but not with any specific brand, since the experimental setting was designed to control for the effect of substituting internal search for external search. To the best of our knowledge, there has been no replication of any of these studies. So, despite their importance, the contributions of the research which has been conducted to date are limited.

External Search

The crux of the difficulty in studying prior knowledge is the lack of a clear definition of the underlying construct. In the absence of a clear definition, measurement varies from study to study. The same is true of external search. In this section we present a brief review of the ways in which the external search has been operationalized in the relevant literature.

The Bettman and Park (1980) study was based upon protocol analysis. External search was measured by providing a matrix of information through an information display board (IDB) and asking the respondents to verbalize while they engaged in search. The analysis of the protocols provided the basis for identifying the use of either prior knowledge or new information. However, as Brucks (1985) has pointed out, the use of IDB limits the number of brands and provides information in such a narrowly defined structure that it could potentially obscure the linkage between knowledge and search. The IDB methodology also limits external search to a single source of information. Johnson and Russo's (1983) study, also employing the IDB methodology suffers from similar limitations.

Punj and Staelin's (1983) operationalization of external search was based on a linear composite of 5 different measures. These measures are related to the time spent by the main shopper and other members of the household in different search activities, number of visits to dealers and the total number of search activities. Generally, researchers have typically used a limited number of components of the measure of search (see review by Newman, 1977). However, Punj and Staelin's construct of external search captures a larger domain of search.

Brucks' (1985) used four measures of search to test several different hypotheses. The first measure was related to the number of attributes. The second measure was related to the percentage of total inquiries which were directed towards dealer evaluation. The third measure, called the variability of search, was the standard deviation of the number of inquiries across alternatives. This measure was an indicator of efficiency in search; high variability meant concentration of search among fewer brands. The last measure, percentage of search spent on alternatives which were inappropriate for the intended use, was an indicator of inefficiency.

The review of the literature illustrates the fact that a number of different components are used to measure and implicitly define external search. Without a standard measure of search, one can expect conflicting results.


In this section, we propose to demonstrate the sensitivity of the relationship between knowledge and search to different measures of the two constructs. The data used for this study came from a set developed for a larger study of external search and its determinants. A mail survey was conducted in a metropolitan area in the north-east. A random sample of new car buyers were contacted three times by mail - prenotification, questionnaire and follow-up. Subjects were asked to provide information about different aspects of their knowledge and search activities prior to their new car purchase. The average time lag between purchase of a new car and responding to the questionnaire was four and a half month. The total number of usable responses was 1401.

In the following section, we outline measures of amount of external search and prior knowledge. Second, we examine the correlations between various measures. Finally, we present the empirical results showing the nature of the relationship between prior knowledge and external search using OLS regression.

Measures of External Search and Knowledge

Three measures of the amount of search were computed. The first measure of the amount of search was a comprehensive measure of the total time spent in various search activities (TOTTIME). Search activities included in the measure were: (i) talking with friends and relatives (ii) reading books/magazine articles (iii) advertisements on TV/radio (iv) car ratings in magazines (v) reading manufacturer brochures/pamphlets (vi) showroom visits (vii) talking to sales people, and (viii) test-driving cars. Respondents estimated the time spent in each activity. The estimates were summated to yield the cumulative time spent in various search activities.

The second and third measures were single item measures frequently used by researchers in this area (see review by Newman, 1977; Punj and Staelin, 1983; Brucks,1985): the number of dealers visited (NDRAL) and the number of models test-driven (NTEST).

To assess the performance of the comprehensive measure of search, a comparison was made with two weighted indices of search efforts reported in the literature (Bennett and Mandell, 1969; Duncan and Olshavsky, 1982). A weighting process, based upon the percentage of total time spent on each of the eight activities, yielded weights which were consistent with those reported by Bennett and Mandell (1969), and Duncan and Olshavsky (1982).

Three multiple-item measures of knowledge were computed. Brucks (1985) discussed the conceptual distinction between two components of knowledge subjective and objective. The reasoning behind making this distinction is that a difference might exist between "what individuals perceive they know" and "what is actually stored in memory". This distinction is accepted and extended in this study. Following Brucks' definition, Subjective knowledge (SKNOW) is the measure of what individuals perceive they know. The same two measures used by Brucks and an additional statement using a 7-point Likert type scale formed the subjective knowledge scale.

However, we treat Objective knowledge as having two components - technical knowledge (OKNOWT) and product class knowledge (OKNOWG). The multiple-item technical knowledge scale measures respondents' expertise regarding cars. The scale reflects whether the respondents understand how a car functions, whether they work on cars themselves and friends' opinion of their expertise. The second dimension of Objective Knowledge a general product class knowledge reflects the familiarity resulting from usage, such as paying attention to a car's maintenance and mileage.

Table 1 shows the reliabilities of the three measures of knowledge. All three scales are unidimensional and the Cronbach alpha values are above the acceptable level for basic research (Nunnally, 1968).


The correlation between the different measures of search and knowledge are shown in Table 2. It may be observed that the comprehensive measure of search, based upon the time spent on various activities, correlates only moderately (.32) with the two commonly used single item measures. All the correlation coefficients of different measures of knowledge and search are positive and significant indicating, in general, a positive relationship between different dimensions of search and knowledge. Subjective knowledge and technical knowledge have relatively lower correlation with the time measure of search than with the two single item measures.

The relationship between knowledge and search was first tested for curvilinearity. The equation used for the regression runs is:

Search = a + b1(Knowledge) + b2(Knowledge)2 + e

If the inverted U hypotheses is valid, b1 should be positive and significant and b2 should be negative and significant, simultaneously. Table 3 summarizes the results of the inverted-U tests.

The number of dealers visited is significantly affected in an inverted U fashion by subjective knowledge at alpha < 0.05. The number of models test driven is also significantly related to subjective knowledge in an inverted U fashion ( alpha < 0.10). However, the time spent in various search activities is not significantly influenced in an inverted U fashion by any of the three measures of knowledge. It is apparent that the inverted U hypothesis may be either accepted or rejected, depending on the measure of knowledge and search employed. In general this study is not -supportive of the inverted U relationship put forth as an integration of the facilitating and efficiency/complementary explanations.

It is important to recognize that a measure of search based upon the total time spent in various search activities, although operationally quite appealing, does not recognize the difference in the amount of information extracted from various sources for the same unit of time. It also does not reflect the differential effort required to collect information on beliefs regarding the relative reliability of different sources. Next, a simple linear regression model:

Search = a + b (Knowledge) + e

was run for those cases where the inverted U hypothesis was rejected at alpha < 0.05. A summary of the results of the simple regression runs is presented in Table 4. The beta values are positive and significant in all cases, lending support to the facilitating hypothesis. It will be recalled that OKNOWG represents what Brucks (1985) refers to as an experienced based measure. Should different people learn differently from the same experience, it is reasonable to expect that their behaviors will also be different. Hence, as compared to the other measures, OKNOWG should be less directly related to search behavior. ID our investigation this idea is supported when search was operationalized in terms of NDEAL and NTEST.

Although F values for each of the eight equations are significant at alpha < .05, the adjusted R2 values are extremely low. The low degree of variance explained by knowledge is not inconsistent with the other widely cited studies (e.g. Punj and Staelin, 1983). We take these findings to indicate that although the various hypotheses presented in the literature capture some significant aspects of external information search, prior knowledge alone cannot explain search behavior.










The purpose of this investigation was to examine the relationship between prior knowledge and external search using multiple measures of these two constructs extracted from the relevant literature. In this study, the facilitating hypothesis, which posits a positive relationship between knowledge and search fails rejection consistently, irrespective of the measures used. Findings regarding the inverted U relationship between knowledge and search, however, are mixed. None of the several tests produced any finding supportive of Motivation or Efficiency hypotheses.

In view of the conflicting findings reported in the literature, including those of the present study, future researchers may consider incorporating the intervening constructs, viz. Ability, Motivation, and Efficiency, explicitly in the model in order to get a better understanding of the relationship between prior knowledge and external search.


Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of Consumer Expertise", Journal of Consumer Research, 13 (March), 411-454.

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

Bettman, R. James and C. Whan Park (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes", Journal of Consumer Research, 7 (December), 234-248.

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

Bucklin, Louis P. (1966), "Testing Propensities to Shop", Journal of Marketing, 30, 22-27.

Jacoby, Jacob, Roben W. Chestnut, and William A. Fischer (1978), "A Behavioral Approach in Non Durable Purchasing", Journal of Marketing Research, 15 (November), 532-544.

Johnson, Eric and J. Edward Russo (1984), "Product Familiarity and Learning New Information", Journal of Consumer Research, 11 (June), 542-550.

Katona, George and Eva Mueller (1955), "A Study of Purchase Decisions", in Consumer Behavior: The Dynamics of Consumer Reaction, ed. Lincoln H. Clark, New York: NYU Press, 30-87

Moore, William L. and Donald R. Lehmann (1980), "Individual Differences in Search Behavior for a Nondurable", Journal of Consumer Research, 7 (December), 296-307.

Newman, Joseph W. (1977), "Consumer External Search: Amount and Determinants", in Consumer and Industrial Buying Behavior, eds. Arch G. Woodside, Jagdish N. Sheth, and Peter D. Bennett, New York: Elsevier North-Holland, 79-94.

Newman, Joseph, W. and Richard Staelin (1971), "Multivariate Analysis of Differences in Buying Decision Time", Journal of Marketing Research, 8 (May), 192-198.

Newman, Joseph W. and Richard Staelin (1972), "Prepurchase Information Seeking for New Cars and Major Household Appliances", Journal of Marketing Research, 9 (August), 249-257.

Park, C. Whan and V. Parker Lessig (1981), "Familiarity and its Impact on Consumer Decision Biases and Heuristics", Journal of Consumer Research, 8 (September), 223-230.

Punj, Girish N. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," Journal of Consumer Research, 9 (March), 366-380.

Rudell, Fredrica (1979), Consumer Food Selection and Nutrition Information, New York: Praeger.

Srull, Thomas K. (1983), "The Role of Prior Knowledge in the Acquisition, Retention, and Use of New Information", in Advances in Consumer Research, Vol. 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for Consumer Research, 572-576.



Narasimhan Srinivasan, University of Connecticut
Jagdish Agrawal, State University of New York at Buffalo


NA - Advances in Consumer Research Volume 15 | 1988

Share Proceeding

Featured papers

See More


How Employees Relate to Their Brand Online: A Critical Visual Analysis of Hollister

Stephanie Kogler, University of Innsbruck, Austria

Read More


Vicarious Pride: When Gift Customization Increases Recipients’ Appreciation of the Gift

Marta Pizzetti, Università della Svizzera Italiana
Michael Gibbert, Università della Svizzera Italiana

Read More


The “Break-in” Effect: A Token Gesture Can Increase Task Initiation and Prevent Goal Abandonment

Adelle Xue Yang, National University of Singapore, Singapore
Babu Gounder, University of Miami, USA
Rajesh Bagchi, 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.