An Exploratory Study Comparing Amount-Of-Search Measures to Consumers' Reliance on Each Source of Information

Jeff Blodgett, Indiana University
Donna Hill, Bradley University
[ to cite ]:
Jeff Blodgett and Donna Hill (1991) ,"An Exploratory Study Comparing Amount-Of-Search Measures to Consumers' Reliance on Each Source of Information", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 773-779.

Advances in Consumer Research Volume 18, 1991      Pages 773-779


Jeff Blodgett, Indiana University

Donna Hill, Bradley University

Researchers measure external information search by counting the number of times each particular behavior was undertaken. For example, respondents are asked how many stores were visited, how many friends one talked to about the product, etc. A serious limitation of this type of measure is that it provides only limited information as to how much each consumer relied on a particular source of information when making the purchase decision; i.e. their search strategy. This paper introduces a new measure (instrumentality) of consumers' reliance on each source of information, and suggest that this measure can complement the traditional amount-of-search measures. By relating the antecedents of search to the instrumentality measures researchers can gain more insight into consumers' search strategies.

Researchers investigating the external information search process have made much progress in the past decade. For example, Kiel and Layton (1981) identified several different dimensions of external search, while Furse, Punj, and Stewart (1984) created a topology of individual search strategies. Punj and Stewart (1983) and Duncan and Olshavsky (1982) investigated the effects of several different determinants of external search on a measure of total search. Beatty and Smith (1987) expanded upon these studies by investigating the effects of purchase involvement, ego involvement, time availability, and product class knowledge across the different dimensions of search (identified by Kiel and Layton 1981). This latter approach investigating the effects of individual and environmental differences across the different types of search - is particularly informative.

Despite these conceptual and methodological advances, researchers have yet to explain much of the variance of external il formation search. One reason for this empirical shortcoming has been noted by Bloch, Sherrell, and Ridgway (1986). They noted that much search activity is understated because it is part of an ongoing search process, therefore, most studies - which focus on prepurchase information search - actually measure only a subset of consumers' total search activity. This paper argues that another factor limiting our understanding of the search process lies in the definition of external search; these studies define external search solely as amount-of-search. The problem with this definition (and measurement) is that it does not take into account the extent to which an individual relied on a particular source of information. Because one of the main goals of research on the external search process has been to do just that (i.e. to explain why different individuals rely on different search strategies) this issue is important. The purpose of this paper is to elaborate on this problem and to offer a new measure of search as a possible solution. An exploratory study was undertaken to provide a first step towards validating this new measure of search.


Researchers have measured the different types of search (e.g. in-store search, interpersonal search, neutral source search, etc.) by asking the respondent to recall the number of times each particular behavior was undertaken. For example, respondents are asked how many stores were visited, how many friends one talked to about the product, how many buying guides were read, etc. (see Kiel and Layton 1981; Duncan and Olshavsky 1982; Punj and Staelin 1983; Furse, Punj, Stewart 1984; Bloch, Sherrell, and Ridgway 1986; Beatty and Smith 1987). As just mentioned, a serious limitation of this type of measure is that it provides only limited insight as to how much each consumer relied on a particular source of information when making the purchase decision. For example, imagine two consumers who both consult one issue of Consumer Reports during the search process: one consumer studies this buying guide in detail and relies solely on this information when making the purchase decision, while the other consumer also looks at this buying guide but discounts the information and instead relies on another source of information. In this situation it is quite obvious that the search behavior of these two consumers differed significantly (with respect to the buying guide), however, studies using amount-of-search as the dependent measure would not make this distinction. This shortcoming holds true across the other dimensions of external search also: number of friends talked to, number of retailers visited, etc. In other words, amount-of-search measures do not always accurately reflect individual consumers' search strategies.

Engel, Kollat, and Blackwell (1968) recognized long ago that amount-of-search - or exposure to an information source - is not necessarily the same as the importance of an information source. They state that (p. 404):

One way of expressing the importance of an information source is in terms of exposure. For example, an information source could be judged more important than other sources in that a greater percentage of consumers report being exposed to it.... There are, however, fundamental problems involved in using exposure in this manner [italics added]. Consumers can be exposed to information sources without using them or finding them helpful in making purchasing decisions. As a consequence, there may be a significant difference between exposure and effectiveness, and an information source that is most important in terms of exposure may be of lesser importance when the criterion of effectiveness is employed.

Indeed, Engel et al. (1968) list several studies which found that consumers often relied more heavily on low-exposure information sources than on high-exposure information sources when making their purchase decisions (see LeGrand and Udell 1964; Sargent 1959; Katz and Lazersfeld 1955). Despite these findings, recent studies have relied solely on exposure to measure external search.

This paper argues that external search can be better understood by measuring what we will call the instrumentality of the information source (i.e. effectiveness), in addition to amount-of-search. The causal effects of the determinants of search could then be compared across these two measures, thus providing more insight into individuals' search strategies. Used in this manner, the instrumentality measures could be a useful complement to the traditional measures of external search. Before the instrumentality measures can be considered to have any practical importance, however, they must first be shown - empirically - to be somewhat different than their respective amount-of-search measures. The purpose of this pilot study, then, is to study the relationship between these two measures of external search to determine whether the instrumentality argument is valid. At issue is: first, how to measure instrumentality and second, whether or not measures of instrumentality will differ from measures of exposure (or amount-of-search). Thus, this paper is an attempt to develop a reliable measure of instrumentality and to take a preliminary step toward establishing the validity of this measure.


Amount of-Search

This study will measure five different dimensions of search (see Duncan and Olshavsky 1982; Kiel and Layton 1981; Bennett and Mandell 1969). These five dimensions, and their traditional amount-of-search measures, are: 1) in-store search, which is measured by the number of stores visited and the number of models examined; 2) interpersonal search is a measure of how many friends, relatives, and/or neighbors who were consulted; 3) an unsponsored (neutral) sources factor is a measure of how many buying guides (such as Consumer Reports) the consumer read; 4) a fourth factor reflects the number of salespeople or other store employees who the consumer talked to; and 5) a media factor measures the lumber of TV, radio, magazine, and newspaper advertisements seen, heard, or read. The Furse et al. (1984) study provides support for these five dimensions. This classification is more precise than Kiel and Layton's (1981) in that it explicitly includes a salesperson dimension. Previous research (Olshavsky 1973) has shown that salespeople can be important sources of information. Like Beatty and Smith (1986), it also separates the media and neutral sources dimensions.

Instrumentality of the Information Source

In addition to the amount-of-search measures, we will also assess the instrumentality of each dimension of search. Again, we intend for the instrumentality construct to reflect the degree to which the consumer relied on a particular source of information when making the purchase decision. As such, the domain of this construct might also encompass the relative helpfulness, usefulness, and importance of the particular source of information.

In order to develop a reliable measure of instrumentality multiple items are needed (Churchill 1979). Therefore, we measure each of the instrumentality constructs with three 7-point interval scales, using agree/disagree, not at all useful/extremely useful, and did not rely on at all/I relied most heavily (on this information) as scale endpoints. Examples of items (one for each dimension of search) that were used include:

To what extent did you rely on the advice of your friends, relatives, or neighbors, etc., in making this purchase decision?

I found consumer rating guides, such as Consumer Reports, or hobbyist magazines, such as Modern Photography to be very useful when deciding which brand to buy.

To what extent did you rely on the advice of the different salespeople, or other store employees, who you talked to?

When deciding which brand to buy, I found advertisements in newspapers or magazines, and/or on TV or the radio, to be very useful.

Compared to these other strategies, when deciding which brand to buy to what extent did you rely on just visiting different stores and comparing the various brands that were available?


The data for this study was collected as part of another study being conducted by the authors. A convenience sample of 114 adults participated in the study; 64% were white-collar workers enrolled in evening MBA courses, while 36% were staff employees (mainly secretaries and clerks) at a midwestern university. Forty-four percent were male while 56% were female. Respondents were provided with a list of major durable products and asked which ones they had purchased within the last twelve months. This list included VCR's, stereo equipment, camera equipment, televisions, and personal computers (this product group is similar to that used by Beatty and Smith 1987). They were then asked to use the product they purchased most recently (from among this list) as the focal point of this study. To avoid problems of recall among subjects who had purchased the particular product before only responses from first time purchasers were utilized. As recommended by Feldman and Lynch (1988), the 15 instrumentality items and the amount-of-search items were spread throughout the questionnaire, with reverse wording on many of the instrumentality items. This practice reduces subjects' propensity to retrieve one response as the basis for another, thus providing a more stringent test of reliability.



The means and standard deviations of both types of measures are shown in Table 1. Respondents, on average, visited 3.11 stores and considered 2.88 brands, for an in-store mean of 5.99. They asked for advice from 2.19 friends, or relatives, on average, and from 1.67 salespeople. Only .62 neutral sources were used, on average, while respondents obtained product information from an average of 3.41 radio, television, newspaper, and/or magazine ads. Respondents relied most heavily on in-store search, followed by interpersonal sources and information from salespeople. Surprisingly, respondents did not rely very heavily upon neutral sources.

Reliability of the Instrumentality Measures

Before proceeding to empirically assess the discriminant validity of the instrumentality measures, one should first check to see whether these measures are reliable (Churchill 1979). Coefficient alpha is the most common indicator of reliability, and was computed for each of the instrumentality measures. In addition to being reliable, the three items measuring each dimension of search should load together, with each set of three items loading on separate factors (Schwab 1986). A principal components factor analysis was performed. The factor loadings, and the reliability (Cronbach's alpha) of each measure, are shown in Table 2.

The items measuring each dimension of search load cleanly on separate factors. The factor loadings are all high (above .80), and the cross loadings are all small (none over .35). These results surpass the standards suggested by Nunnally (1978) for criterion validity. In addition, each instrumentality measure is fairly reliable (the lowest alpha is .78 and the highest is .94), surpassing the standard for reliability suggested by Nunnally (1978) for exploratory research. Overall, the results indicate that we have indeed measured what we set out to measure - instrumentality, or one's reliance on a particular source of information. With this assurance in mind, we can now set out to compare the instrumentality measures to the amount-of-search scores.


The correlations within the amount-of-search measures and within the instrumentality measures can be seen in Table 3. An interesting finding is that all of the significant amount-of-search correlations are positive while two of the instrumentality correlations are negative. The latter findings imply that, for many people, interpersonal and neutral sources information are substitutes for in-store information, while the correlations for the amount-of-search measures imply that consumers who undertake greater amounts of in-store search also undertake greater amounts of all other types of search. Although it is a bit difficult at this point to reconcile these findings, it is important to note that these findings are not necessarily contradictive. Rather, these findings lend credibility to the argument that the information gained from using the instrumentality measures might complement that gained from the traditional amount-of-search measures. Future studies that investigate the causal effects of the determinants of search might be able to use these findings to better explain consumers' search strategies.

Table 4 shows the correlations between the amount-of-search scores and the instrumentality scores. In order to convincingly argue that the instrumentality measures might complement the amount-of-search measures these two types of measures should not be too highly correlated with one another. The correlations are low enough to suggest that the instrumentality measures are related, but not identical, to the amount-of-search measures. Therefore, the instrumentality measures might indeed provide researchers with additional information.





One of the goals of previous researchers has been to explain what causes different consumers to undertake different types of search. To do so, researchers have tried to model the effects of various antecedents of search on the different dimensions of search (see Beatty and Smith 1987). Having established that the instrumentality and amount-of-search measures are somewhat different, it is argued that researchers might find some interesting results when they compare the effects of the various determinants of search across these two types of measures (i.e. across the different dimensions). That is, our interpretation of the causal effects of the f various determinants of search might well depend on which type of dependent measure is used. Consider again two fictitious consumers: consumer number one is highly confident in her ability to judge new cars (i. product class, or subjective knowledge) and undertakes an extensive search process. In addition, she asks for advice from one friend, but does not rely on this advice when deciding which brand to buy. Consumer number two, who has little confidence in his ability to judge new cars, undertakes very little search and instead relies entirely on the advice of one relative. In this situation because both consumers undertook the same amount of interpersonal search it would not appear that product class knowledge had any effect upon search when regressed upon this measure. However, while it would be correct to say that this variable had no effect on the amount of interpersonal search, it would not be correct to say that this determinant had no effect on these two consumers' search strategies. Obviously, researchers would reach different conclusions regarding the effect of product class knowledge depending on whether the amount-of-search or the instrumentality measure is used. Again, we do not suggest that one measure is better than the other, rather, we feel that the information gained from using the instrumentality measures would certainly complement that from the amount-of-search measures.


To illustrate the potential contribution of the instrumentality measures, Beatty and Smith (1987) were able to explain only 15%, 7%, and 5% of the variance of the amount of interpersonal, media, and neutral sources search, respectively. Because the instrumentality measures appear to better reflect consumers' search strategies, we feel that researchers who use these measures will be able to explain a significantly greater percent of the variance of external search. For example, Beatty and Smith (1987) found the effects of time availability and ego involvement to be .17 and .14 when regressed on the amount of neutral source measure. Again, these two determinants (plus purchase involvement and subjective product class knowledge) explained only 5% of the variance of this type of search. Going back to the example cited earlier in this paper (regarding Consumer Reports), researchers might find these determinants to have a much greater effect on one's reliance on these sources. In addition, other variables that had no effect on the amount of neutral search - such as product importance - might have a significant effect when instrumentality is used as the dependent variable.

A related-area of research where it might pay to use the instrumentality measures concerns the stages of the decision process. It has generally been considered that different types of information sources will be used in different stages of the decision process. For example, when consumers are in the early stages of the decision process (problem recognition and refinement) they tend to seek information from the various media sources (Mowen 1987). During the purchase specification stage consumers turn to friends, relatives, and neutral sources such as Consumer Reports. As the consumer enters the purchase respecification stage and begins to visit different stores he or she is more likely to seek information from a salesperson. If the in-store and salesperson information confirms the consumer's prior beliefs he/she may decide not to seek any more information (Wilkie and Dickson 1985) and to make the purchase. By examining the sequence of search, and by determining consumers' reliance upon different types of information at these different stages, we can better understand the effects of the various individual and situational determinants of search. With this information we can create more precise profiles of consumers' search strategies.


Future research should focus on the relationships between the antecedents of search, the instrumentality of search sources and the amount of search. Figure 1 provides a conceptual diagram of such a model. Rather than aggregating the various measures of search it is proposed that in-store search, media search, interpersonal search, neutral sources search, and salespersons' advice search all be treated as separate dependent variables. The rationale for treating each dimension of search separately is that when all of the different measures of search are aggregated many interesting effects are averaged out. For example, one consumer might undertake a high amount of retailer search and very little media and neutral source (buying guide) search while another consumer does the exact opposite. Although these two consumers undertook different types of search an aggregate measure would reflect only the amount of search undertaken, obscuring the independent-dependent variable relationship. Data regarding the type of search undertaken is much more interesting and informative than data regarding how much search occurred overall.



By introducing the instrumentality measures into the model the researcher will have information on the extent to which the consumer relied on each source of information. It is hypothesized that each of the antecedent variables will have a different impact on the type of sources consulted for additional information. For example, higher levels of subjective knowledge (and ability to judge) should lead to less reliance on interpersonal sources and more reliance on retailer/in-store information (Selnes and Gronhaug 1986; Duncan and Olshavsky 1982). Higher levels of satisfaction with previous brands might lead to less dealer and media search while not having a significant effect on the other types of search (Kiel and Layton 1981; Katona and Mueller 1955; Newman and Staelin 1972). If little time is available the consumer might rely heavily on interpersonal sources and salespersons' information, and undertake little retailer search.

Although the effects on retailer search and interpersonal search seem fairly straightforward the same is not true of the other dimensions of search. Overall, it is difficult to go beyond generalizations at this point; there are too few empirical studies and little theory to rely upon to develop specific hypotheses. Nonetheless, one of the purposes of model building is to develop realistic pictures of the world, with or without adequate theoretical justification.


The purpose of this paper has been to propose that our knowledge of the externalinformation search process can be better understood by taking into account the instrumentality of the various sources of information. In the future, we plan to compare the causal effects of several determinants of search across these two different measures of search. Only then can the validity of these measures accurately be assessed. The current study is encouraging in that it provides a strong justification for this type of inquiry. Again, we believe that researchers will be better able to interpret the effects of the various determinants when the instrumentality measures are used, thus enhancing our knowledge of consumers' search strategies.


Beatty, S.E. & Smith, S.M. (1987). "External Search Effort An Investigation Across Several Product Categories," Journal of Consumer Research, 14 (June), 83-95.

Bennett, J.H. & Mandell, R.M. (1969). "Prepurchase Information Seeking Behavior of New Car Purchasers - The Learning Hypothesis," Journal of Marketing Research, 6 (November), 430-33.

Bloch, P.H., Sherrell, D.L. & Ridgway, N.M. (1986). "Consumer Search: An Extended Framework," Journal of Consumer Research, 13 (June), 119-126.

Churchill, G. (1979). "A Paradigm for Developing Better Measures of Marketing Constructs," Journal of Marketing Research, 16 (February), 64-73

Duncan, C.P. & Olshavsky, R.W. (1982). "External Search: The Role of Consumer Beliefs," Journal of Marketing Research, 19 (February), 3243.

Engel, Kollat, & Blackwell (1968). Consumer Behavior, Chicago, Il.: Holt, Rinehart, and Winston, - Inc.

Feldman, J.M. & Lynch, J.G. (1988). "Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention, and Behavior," Journal of Applied Psychology, 73, no. 3, 421-435.

Furse, D.H., Punj, G.N. & Stewart, D.W. (1984). "A Topology of Individual Search Strategies Among Purchasers of New Automobiles," Journal of Consumer Research, 10 (March), 417-31.

Katona, G. & Mueller, U. (1955). "A Study of Purchasing Decisions," in Consumer Behavior: The Dynamics of Consumer Reaction, L. H. Clark (ed.), New York, NY: New York University Press, 30-87.

Katz, E. & Lazersfeld, P. (1955). "Personal Influence, New York, NY.: The Free Press of Glencoe.

Kiel, G.C. & Layton, R.A. (1981). "Dimensions of Consumer Information Seeking Behavior," Journal of Marketing Research, 18 (May), 23339.

LeGrand, B. & Udell, J.G. (1964). "Consumer Behavior in the Market Place," Journal of Retailing, Fall, 3240, 47-48.

Mowen, J. (1987), Consumer Behavior, New York: Macmillan Publishing Company.

Newman, J.W. & Staelin, R. (1972). "Prepurchase Information Seeking for New Cars and Major Household Appliances," Journal of Marketing Research, 17 (November), 460-69.

Nunnally, J.C. (1978). Psychometric Theory, New York: McGraw Hill Book Co.

Olshavsky, R.W. (1973). "Customer-Salesman Interaction in Appliance Retailing," Journal of Marketing Research, (May) 10, 208-212.

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

Punj, G. and W. Stewart (1983), "An Interaction Framework of Consumer Decision Making," Journal of Consumer Research, 10 (September), 181-196.

Sargent, H.W. (1959). "Consumer Product Rating Publications and Buying Behavior, Urbana, Il.: University of Illinois Bureau of Economic and Business Research.

Schwab, D.P. (1986) . "Construct Validation in Organizational Behavior," in L.L. Cummings and B. Staw, eds., Research in Organizational Behavior, 12, Greenwich, CT: JAI Press, 41-50.

Selnes, F. & Gronhaug, K. (1986). "Subjective and Objective Measures of Product Knowledge Contrasted," in R. Lutz (ed.) Advances in Consumer Research, 13, Provo, Utah: ACR, 6171.

Wilkie, W. and P. Dickson (1985), "Shopping For Appliances: Consumers' Strategies and Patterns of Information Search," Report No. 85-108, Cambridge Mass: Marketing Science Institute.