Vous êtes sur la page 1sur 16

Journal of Consumer Research, Inc.

A Typology of Individual Search Strategies Among Purchasers of New Automobiles Author(s): David H. Furse, Girish N. Punj, David W. Stewart Reviewed work(s): Source: Journal of Consumer Research, Vol. 10, No. 4 (Mar., 1984), pp. 417-431 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2488911 . Accessed: 19/04/2012 03:37
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press and Journal of Consumer Research, Inc. are collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Research.

http://www.jstor.org

Typology

of

Individual of New

Search

Strategies

Among

Purchasers

Automobiles

DAVID H. FURSE GIRISH N. PUNJ DAVID W. STEWART*


Cluster analysis of questionnaire data was used to identify six distinctive external informationsearch patterns among purchasers of new automobiles. Two of the shopper clusters had not been clearly specified in prior research-namely, an advisor-assisted shopper group and a highly self-reliant shopper group. An effort to cross-validate the typology using data obtained from automobile sales personnel was partiallysuccessful. It is hypothesized that these strategies are reflections of heuristic decision processes which reflect both individualdifference characteristics and the purchase situation.

Previous researchon consumerbehaviorhas found that


consumers employ information search strategies which can be distinguished by the amount of external search effort and decision time. Early efforts to identify distinct groups of consumers characterized by differing information search strategies used unidimensional measures of aggregate search activity (Katona and Mueller 1955, Newman and Staelin 1972). Such aggregate measures have been criticized for failing to adequately define differences in patterns of information gathering (Claxton, Fry, and Portis 1974). Very different patterns of search activity may produce similar aggregate search measures, with the result that potentially important differences in search activity may be missed. More recently, Newman and Staelin (1973), Claxton et al. (1974), Westbrook and Fornell (1979), and Kiel and Layton (1981) have sought to more fully examine the richness of the pattern of information search. This literature suggests that there are rather distinct patterns of information search, at least among purchasers of durable goods. All of the research to date has examined search for high-cost items such as appliances and automobiles. Despite differences in measurement, samples, and method of data analysis, there is a remarkable degree of agreement in the findings. Patterns of search emphasizing retail outlets, interpersonal sources, little active search, extensive search of many
*David H. Furse is President, Nashville Consulting Group, 2200 Hillsboro Road, Nashville, TN 37212. Girish N. Punj is Assistant Professor of Marketing, University of Connecticut, Storrs, CT 06268. David W. Stewart is Associate Professor of Management, Owen Graduate School of Management, Vanderbilt University, Nashville, TN 37203. This study was supported by a doctoral dissertation grant from the American Marketing Association, a grant from the University of Maine at Orono, and the 1981 and 1982 Dean's Funds for Faculty Research of the Owen Graduate School of Management, Vanderbilt University.

sources, and a more modest amount of search of several sources have been identified in one or more previous studies. Yet much of the prior research has been restricted to an examination of search activities, and so has not provided a detailed characterization of individuals who use a particular search strategy on other, non-search-related variables. The present study was designed to consolidate and extend previous efforts in three ways: 1. First, we attemptto focus more extensivelyon the participation of others in the decision-makingunit. Previous studies have not considered the involvement of anyone but the principaldecisionmaker the information in search process, althoughothers have been examined as sources of information. 2. Second, we attemptto validatefindingsbased on the analysis of consumerself-reportwith data gatheredindependentlyfrom sales personnel. While studies to date have relied on dataprovidedsolely by the buyer, characteristic searchpatternsshould also be evident to sales personnel. Not all search activities of shoppers may be visible to sales personnel, but evidence of prominentshoppingactivity patternsshould be obtainable. 3. Our thirdobjective is to cross-validateand synthesizethe findings of previous research and offer a speculativeexploratory framework for the processes which produce characteristicsearch patternis.The largersample size of the present study provided an opportunityfor split-half validationof the derived searchpatternsand for detailed of analyses of the characteristics smaller, yet significant, consumerclusters. We also examineda very large set of data on non-search-related characteristics the clusters, of which allowed for a richerand more completedescription of the types of individualswho use particular patternsof search. This additionaldescriptive informationprovides a basis for theorizing about the determinants search of strategiesamong consumers. 417
?) JOURNAL OF CONSUMER RESEARCH e Vol. 10 e March 1984

418 The study was carried out in two stages: (1) an extension of earlier findings using buyer-generated data, and (2) an effort at consensual validation of the findings from buyergenerated data using data obtained from sales personnel.

THE JOURNAL OF CONSUMER RESEARCH Table 1 provides a comparison of the variables used in the classification phase of the present study with those used in the Claxton et al. (1974) and Kiel and Layton (1981) studies, which are the two studies most similar to the present one in methodology. Data were obtained from 1,056 respondents, and 1,031 responses were actually used. The survey response rate was 38 percent. However, 61 percent of respondents who were reached by telephone returned a completed questionnaire, while only 13 percent of households that could not be contacted by phone responded. Twenty-five respondents were eliminated because of excessive missing and/or improbable responses. For other respondents, missing responses were replaced by the mean for the entire sample. To ensure the internal validity of the data, a series of preliminary analyses was carried out. First, responses to separate items that requested essentially the same information were compared. This multipleresponse feature was deliberately included in the questionnaire to allow a check for internal consistency. The detailed results of these comparisons are too numerous to report here (about 15 consistency checks were performed), but in each case better than 98 percent of the respondents displayed consistency across the multiple responses. These checks included determining actual agreement between several measures of the same event-e.g., number of decisionmakers, whether the consumer was a first-time buyer-as well as the more frequently used reliability coefficients among multiple measures of time spent, preference, and so on. To analyze the effect of "forgetting," a variable labeled "duration" was constructed to measure the elapsed calendar time between the purchase of the car and the return of the questionnaire. The total sample was split into four subsamples corresponding to the four quartiles on this variable. The Kruskal-Wallis test (one-way analysis of variance by ranks) was used to compare the four subsamples on a selective set of variables related to information search. The null hypothesis was that the four subsamples were drawn from four identically distributed populations. A chi-squared statistic was used to determine rejection/nonrejection of the null hypothesis. Using a 0.05 confidence level, the null hypothesis could be rejected for only five of the 23 variables examined. Although this number is greater than what would be obtained by chance, an examination of the mean ranks showed no systematic directional relationships. The lack of association of the variable "duration" with the distribution of responses seems to indicate that respondents were able to recall the sources used-or at least that if there was forgetting, this forgetting did not take place systematically during the period when any of the respondents were asked to fill out the questionnaires. Next, a set of analyses compared the sample distributions for the variables "dealer visits," "total purchases," and "total makes" with those reported in previous studies (see Staelin 1969; U.S. News and World Report 1974, 1975; and Newsweek 1977a, 1977b). No appreciable differences were found. Finally, an analysis similar to the one reported

NEW CAR BUYER STUDY'


Subjects/Data Collection
Data for the study were generated by consumers who had purchased a new automobile during the period of September to November 1978, in the cities of Buffalo, Milwaukee, and Phoenix. Stratified random samples taken from a complete census obtained from R.L. Polk of all households that had purchased new automobiles during a specified time period were used in the study. Most respondents were contacted by telephone and asked to participate in the study. The individuals identified as the principal decisionmakers were mailed a questionnaire between late February and May 1979. Average time from purchase to receipt of the questionnaire was 28 weeks. The questionnaire solicited information on the activities and decisions undertaken by the respondent or others in selecting their new automobile. The questionnaire requested information such as the number of dealer visits, activities at each dealership visited, out-ofstore information search behavior (e.g., talking to friends, reading magazines), number and types of previous cars owned, satisfaction with car purchased, and so on. Information was also requested concerning the automobile purchased, the dealership from which it was purchased, and the price paid for the automobile. Demographic information concerning age, sex, education, income, and so on was also collected. The variables of primary interest were those concerned with the amount of time the purchaser or someone else spent on a variety of information search activities, ranging from reading advertisements in the newspaper to test driving automobiles. There were 18 of these items. A six-point scale was employed to obtain estimates of the time spent on each activity: (1) no time; (2) up to '/2 hour; (3) more than 1/2 hour but less than 2 hours; (4) more than 2 hours but less than 5 hours; (5) more than 5 hours but less than 10 hours; and (6) more than 10 hours. Other questions asked for the total number of visits to dealers and the number of different dealers visited. Twenty-four items were used in the classification phase of the study. To be consistent with earlier studies, the items selected for classification were related to search activities, rather than to the content or outcome of the search. Item selection was based on the findings of previous studies of search patterns and on other studies of search behavior.
'A preliminary report of the findings of this study was made at the 1981 meeting of the Association for Consumer Research. The present report represents a reanalysis of earlier findings using a larger data set and additional variables. Donald Granbois, Roger Layton, and two anonymous reviewers of the earlier report made useful suggestions for further analyses, which we have tried to incorporate in the current report.

SEARCH STRATEGIES
TABLE 1 RELATIONSHIPS BETWEEN SEARCH ACTIVITIES EXAMINED IN THREE DIFFERENT STUDIESa Claxton, Fry and Portis (1974) 1. Considered 2. Considered 3. Considered alternate use of money alternate brands alternate price levels Kiel and Layton (1981) Number of ads recalled Number of owners contacted Number of opinion leaders contacted Number of trips to dealers Number of retailers visited Time spent at retailers Total introspection time Total search time *Introspection and search time factor Number of makes considered Number of phone calls made Number of items of written information used Number of other dealers considered *Retailer search factore *Media search factore *Interpersonal search factore Furse, Punj, and Stewart (1984) Time spent talking to salespersons (hours)b Time spent looking at ads (hours)b

419

Time spent talking to others (hours)b Purchase pal help (hours)c Total number of visits to dealer Number of different dealers visited Time spent walking around dealer showrooms (hours)b Total number of different search activities Time spent reading books and magazines (hours)b Time spent reading manufacturer's pamphlets (hours)b Time spent reading about car ratings (hours)b Regularly reads consumer magazines, such as Consumer Reportsd Time spent test driving cars (hours)b *Retailer visits factore *Out-of-store search factore *Interpersonal search factore *ln-store search factore *Involvement of othere

4. Salesmen mentioned as source of information 5. Stores mentioned as source of information 6. Advertisments mentioned as source of information 7. Family mentioned as source of information 8. Friends mentioned as source of information 9. Other sources mentioned

10. Price mentioned as important feature 11. Brand mentioned as important feature 12. Style mentioned as important feature 13. Quality mentioned as important feature 14. Size mentioned as important feature *15. Total store visits 16. Number of stores visited 17. Maximum visits to single store *18. Deliberation time (duration of search) 19. Number of alternatives considered (sum of 1-3 above) *20. Number of sources 4-9 above) used (sum of

21. Number of features considered important (sum of 10-14 above)


aAsterisks indicate variables serving as the basis for classification. bThese activities included separate measures for the respondent and others in the household; responses ranged from 1 (no hours) to 6 (more than 10 hours). cPurchase pal help involved specific assistance of others within or outside the household who had specific expertise; responses ranged from 0 (no assistance) to 5 (more than 10 hours). dDichotomous variable. eVariables derived from factor analysis.

for "duration" was performed by comparing the responses for the telephone precontacted and noncontacted subsamples. The null hypothesis was rejected for only one of the 23 variables, a number just about equal to chance. In light of the above validity and reliability checks, it was felt that the data were of high enough quality to merit their use in the analysis. Two key assumptions concerning the data should be noted, however. First, it is assumed that the amount of information sought is related to time spent in each of the search activities. Second, it is assumed that

auto buyers can provide reasonable estimates of the amount of time spent in various search activities. Previous studies have made similar assumptions, although it is not clear whether they are strictly true. Obviously, some consumers make more efficient use of their time and obtain greater information than others. Time spent appears to be a reasonable measure of effort expended, but it is not a perfect one. Since the emphasis of the present study is on the relative amounts of time spent on various search activities rather than on absolute time spent, even if consumers distort

420 time estimates, the same results will hold unless that distortion varies with each activity.

THE JOURNAL OF CONSUMER RESEARCH


TABLE 2 INDIVIDUAL ITEMS RELATEDTO DIMENSIONSOF SEARCH: SHOPPER DATA Factor 1 Number of different dealers visited (.89) Number of hours spent at dealers (.82) Number of visits to dealers (.81) Time you spent driving to and from dealerships (.60) Factor 2 Time you spent reading books and magazine articles (- .81) Time you spent reading about car ratings in magazines (-.80) Number of different search activities (-.75) Time you spent reading advertisements in newspapers and
magazines (-.74)

Analysis
As a prelude to cluster analysis for identifying search patterns, a factor analysis was carried out on the 24 items related to various search activities. The initial factor analysis was designed (1) to reduce the problem of multiple measures of similar constructs being more heavily weighted than constructs measured by fewer items, and (2) to attempt to replicate dimensions of search activity identified in earlier research. A principal components analysis was carried out and a five-factor solution was indicated by both the eigenroots > 1.0 criterion and a plot of the roots. These five factors were submitted to both varimax and oblique rotations. Although the oblique rotation (direct oblimax, delta = 0.2) did not appreciably change the hyperplane count (the number of factor loadings between ? 0.10) obtained with the VARIMAX rotation, it did serve to reduce moderate factor loadings and to increase those loadings already high. For this reason the oblique rotation was retained in the final solution. Table 2 identifies items loading each factor. Factor 1 is characterized by items related to dealership visits: time spent driving to dealerships, the number of different dealers visited, total dealer visits, and the number of hours spent at dealer showrooms. This factor is very similar to the retailer search factor identified by Kiel and Layton (1981) and was probably represented in the Claxton et al. (1974) and Westbrook and Fomell (1979) studies by the composite measure "total store visits." Factor 2 is associated with the level of personal participation in out-of-store search activities, particularly time spent reading ads, car ratings, and manufacturer brochures. This factor is also related to the total number of different search activities. It was probably represented in the Claxton et al. study by a composite measure of number of sources used. Kiel and Layton also identified a media search factor which is similar to the present one. Factor 3 is associated with the participation and involvement of others in out-of-store search activities. Factor 3 has not been identified previously, because earlier studies have not explicitly considered the active involvement of persons other than the principal decisionmaker in particular search activities. Factor 4 is an interpersonal search factor characterized by the amount of time the principal decisionmakers reported spending in talking to others about cars and by the involvement of a "purchase pal" in the search process.2 This search dimension was also identified by Kiel and Layton. Factor 5 is characterized by amount of search activity while at the dealership. This factor is related to the search process at dealers: time spent looking around showrooms,
2A purchase pal is someone assisting in the search process who is known to the principal decisionmaker and is believed to have some expertise in evaluating automobiles (see Bell 1967 for the role played by purchase pals in automobile purchasing).

Time you spent reading automobile manufacturer brochures and


pamphlets (-.59)

Regularly reads consumer magazines such as Consumer Reports (-.52) Time you spent talking to friends/relatives about new cars or dealers (-.40) Factor 3 Time others spend reading about car ratings in magazines (.70) Time others spent reading books and magazine articles (.66) Time others spent talking to friends/relatives about new cars or dealers (.59) Time others spent reading automobile manufacturer brochures and pamphlets (.55) Time others spent reading advertisements in newspapers and magazines (.46) Factor 4 Purchase pal help (.63) Time you spent talking to friends/relatives about new cars or dealers (.59) Time others spent talking to friends/relatives about new cars or
dealers (-.48)

Factor 5 Time others spent test driving cars (-.76) Time others spent talking to salespersons (- .71) Time others spent looking around dealer showrooms (-.61) Time you spent test driving cars (-.54) Time you spent talking to salespersons (- .51) Time you spent looking around dealer showrooms (-.39)
NOTE: Factor pattern coefficients are in parentheses.

talking to sales personnel, and test driving automobiles both by the principal decisionmaker and by others. It appears to differ from Factor 1 in that it is related to the amount of time spent in activities while at dealerships, rather than to the process of finding dealerships. Factor scores3 for each of the 1,031 subjects were computed and provided the basis for a clustering procedure. A
3The use of oblique factors does not fully eliminate the problem of multiple measures of the same construct. However, a cluster analysis using orthogonal factors produced almost identical results to those reported here. The results of the oblique rotation are presented because we believe they provide a richer and more realistic picture of the dimensions of search activities.

SEARCH STRATEGIES
TABLE 3 CLUSTER MEANS FOR DERIVEDFACTORS (SELF-REPORT DATA) Cluster 1 Factor 1-Dealer visits
(-

421

Cluster 2 .24
.25 -.27 .97

Cluster 3 1.07
-.72 3.42 .48

Cluster 4 1.19
-1.50 -.29 - .03

Cluster 5 1.38
.34 -.13 .19

Cluster 6 -.40
-.43 -.01 - .27

-.70
1.10 -.26 - .50

Factor 2-Out-of-store activity Factor 3-Other involvement search Factor 4-Interpersonal Factor 5-In-store activity (_)a

.53

.01

-1.29

-.30

-2.10

.22

aBoth factors 2 and 5 are negatively loaded factors; negative factor scores indicate high activity for clusters.

procedure suggested by Punj and Stewart (1983a) was followed. Ward's hierarchical clustering method with Euclidean distances was used with half of the respondents to obtain an initial description of potential clusters within the data. This initial analysis suggested five to seven clusters. A k-means clustering procedure (BMDP-KM with Euclidean distances) was then used to develop five-, six-, and seven-cluster solutions based on seed points suggested by the earlier hierarchical clustering. These solutions were developed for two independent subsets of the data, which were generated by random assignment of cases to subset. Group centroids obtained from one subset of the data were used to classify cases in the other subset, and vice versa. This cross-validation procedure was carried out for the five-, six-, and seven-cluster solutions. Coefficients of agreement (Kappa) were then computed. Kappa is a chancecorrected measure of agreement for nominal scales that may be interpreted as an intraclass correlation coefficient (Fleiss and Cohen 1973). The six-cluster solution produced a Kappa coefficient of 0.83. Coefficients of agreement obtained for the five- and seven-cluster solutions were smaller and were particularly poor for the seventh cluster. A high degree of confusion among cluster assignments existed in the five-cluster solution. On the basis of these findings, the six-cluster solution was accepted. A final six-cluster solution based on all 1,031 cases was then developed. Cluster means for the derived factor scores are in Table 3.

5. A retail-shopper group with heavy other-involvement,

classifying about 5 percentof respondents


6. A moderate-search group-the largest group-with 32

activity by percentof respondents,characterized moderate on all search factors, althoughslightly above averageon out-of-store search activities and slightly below average on numberof visits to car dealers. Cluster 1 is similar to the low-search groups identified in prior studies. These studies have also identified a highsearch pattern similar to Cluster 3 and an in-store shopper pattern similar to Cluster 5, but none have isolated the heavy involvement of others in these shopper patterns. Cluster 6, the moderate shopper, has also been identified in earlier studies (it resembles the "balanced, thorough shopper" of Claxton et al. 1974). The other two search patterns-Cluster 2, the purchase-pal-assisted shopper, and Cluster 4, the highly self-reliant shopper-have not been clearly identified in prior research. Earlier studies have pointed to the existence of additional clusters beyond those "selective replicated here (Clusters 1, 3, 5, and 6)-e.g., searchers" (Kiel and Layton 1981) and "balanced, thorough" shoppers (Claxton et al. 1974)-but limitations of sample size and the type of variables measured made it difficult to define these additional clusters clearly. Both Claxton et al. and Kiel and Layton did find some evidence of an interpersonal search pattern, but it was not as clearly defined as in the present study, nor was the involvement of a purchase pal associated with this pattern. Comparisons among the six clusters were conducted on a variety of descriptive variables, including demographic characteristics and past purchase experience. Table 4 provides a thumbnail sketch of the characteristics that differentiate each cluster from other clusters, based on tests of statistical significance.4 Cluster 1, the group involved in the least information search, is the most experienced of the groups. Members of this group are older and have, on average, owned more cars and been more satisfied with previous purchases than members of other groups. They are more likely to know in advance the manufacturer and dealer from whom they want to purchase, and they spend less time than any other group in search-related activities. This group appears to exhibit
4A complete listing of the variables by cluster is available from the authors upon request.

Findings
Based on the variables from which they were derived, the six clusters were labeled as follows: 1. A low-search group, classifying 26 percent of respondents, with below average search on all factors-especially out-of-storesearch activities
2. A purchase-pal-assisted search group, classifying 19 per-

cent of respondents 3. A high-searchgroup, with only 5 percentof respondents, characterized above averageactivity on all searchfacby of tors-especially the participation others
4. A high-self-search group, classifying 12 percent of re-

spondents, who are above average on all out-of-store search activities in additionto the total numberof visits to differentcar dealers

422
TABLE 4

THE JOURNAL OF CONSUMER RESEARCH

THUMBNAIL SKETCHES OF CLUSTERS DERIVEDFROM CONSUMER SELF-REPORT DATA

Cluster 1 (Low Search) Spend least time of all clusters in search-related activities Greatest prior purchase experience Have owned more cars than average More satisfied with previous purchases Most certain would get a good deal without informationsearch More likely to know in advance the manufacturer and dealer from whom they wish to purchase Reason for purchase more likely to be feeling that it is good to trade cars every few years Older Highest income of all clusters Search for and purchase cars in a higher price range Most likely of clusters to consider full-sized, four-door models More likely to consider products made by Ford and General Motors (e.g., Cadillac) Less likely to consider Chrysler products or imports Cluster 2 (Purchase Pal Assisted) Least experienced car shoppers Have owned the fewest cars previously Most likely to indicate a father was involved in decision Tend to involve another who is perceived as knowing cars (purchase pal) Express little confidence in their ability to judge cars Likelyto be less satisfied with most recent car purchase May know manufacturerbut not dealer fromwhom they willpurchase Largest percentage of single respondents in this cluster Tend to work in clerical and sales jobs Most likely to be buying because had no car or because they feel it is good to trade cars every few years More likely to purchase a two-door model Car purchased more likely than for any other cluster to be outside of original size and price set of models considered at the outset of formal search Cluster 3 (High Search) Spend the greatest amount of time (their own and others') in search activity Have lowest confidence of any cluster in their ability to judge cars Believe extensive informationsearch is necessary to get a good buy Least satisfied of all clusters with previous purchase Post-purchase satisfaction with new car is below average Tend to involve others in search activities, but these other individuals may have no particularexpertise Car actually purchased has the lowest average sticker price of all clusters Less likely to have a trade-in or get a high trade-in price Best educated of clusters but of moderate income More likely than other groups to be female (although over half are male) More likely to consider subcompacts, compacts, and hatchbacks Likely to consider popular imports Least likely of the clusters to select General Motors as the preferred manufacturer, although a majoritystill prefer GM products

Cluster 4 (Self-Reliant Shopper) Spend the greatest amount of own time in search process but do not involve others in search Consider a large number of automobile makes and models Less likely to know in advance the dealer from whom car is purchased Less likely to have a trade-in Well educated with moderate to high income Male Most likely to be purchasing new car for fuel efficiency or because they want a car for a different purpose from that of previous car Most likely to consider subcompacts and compacts Most likely to consider imports, Ford products, and Dodge Less likely than average to consider General Motors products Cluster 5 (Retail Shopper) Largest number of decisionmakers involved-especially the wife when she is not the principal decisionmaker (this group has the highest percentage of married individuals) Unlikely to know dealer in advance Less likely to have a trade-in Consider a large number of makes Large amount of "other"involvement in the search process Well educated but not necessarily high income Common occupations are managers, government officials, or proprietors Principal reasons for new car purchase are desire for greater fuel efficiency or the fact that the old car quit working and needed replacement Pay highest average price of all clusters for car Prefer intermediate-sized sedans made by GM or Ford (Oldsmobile and Pontiac are particularfavorites) More likely to buy outside of initial manufacturer set but less likely to buy outside of original price set Cluster 6 (Moderate Search) Devote below-average amount of time to search activities High certainty that they could obtain a good deal without information search Very likely to know manufacturer in advance but not necessarily the dealer Least likely to involve others in search process Tend to be older males with higher income than average Most likely to receive a high trade-in price Principal reasons for purchase are desire for greater fuel efficiency and feeling that it is best to trade cars every few years Most likely to consider four-door models Most likely to buy outside of initialprice set Preferences for manufacturers well distributed among members of this group

SEARCH STRATEGIES a shopping pattern similar to what Bettman and Zins (1977) have labeled "preprocessed choice." These individuals are the most certain that they would make a good car purchase without any information search. They are also more likely than members of other groups to say that their reason for purchase was that they trade cars every few years. This group has, on average, the highest income, and they search for and purchase cars in a higher price range than do other new car buyers. They are the most likely to consider a fullsize, four-door automobile-not subcompacts or hatchbacks. Cluster 1 is more likely to consider favorably the products of Ford and General Motors. They are most likely to consider buying a Cadillac, and are less likely to consider a Datsun, Honda, Toyota, Volkswagen, Chrysler, Plymouth, Pontiac, or Ford. Cluster 2, the group that depends on the assistance of a purchase pal, is the least experienced of new car shoppers. They have owned the fewest cars previously, and are the most likely to indicate that a father was involved in helping make the decision. They express little confidence in their ability to judge cars, and they are likely to indicate that they are less satisfied with their most recent car purchase. This group is more likely to know the manufacturer (although not necessarily the dealer) from whom they will purchase, perhaps reflecting the influence of the purchase pal (probably the father). This group has the highest percentage of single respondents and those working in clerical and sales jobs. The reason for purchase is more likely to be the fact that they currently have no car-or that it is best to trade cars every few years, perhaps reflecting the influence of an older purchase pal, such as a parent. The car type purchased is more likely to be a two-door, and the final purchase is more likely than for other groups to be outside the original size and price set of models considered. Cluster 3, the most involved shopper group, has the greatest number of hours (their own and others) involved with search activities. They are less likely to have a tradein and to get a high trade-in price. They also buy, on average, the new car with the lowest sticker price compared to the purchases of other groups. They have the highest number of total hours devoted to search activities, both by themselves and others, but have the lowest confidence in their ability to judge cars. They are less satisfied than other groups with their previous car, and they are less certain that they would get a good buy without extensive information search activity. This group appears to be compensating for lack of prior positive experience by devoting a large number of hours to search. They do not appear to particularly enjoy the search process, and their post-purchase satisfaction with their current automobile is below average. They appear to compensate by involving others in the decision, although not necessarily others with any particular expertise. They may be attempting to limit their risk by purchasing a new car with the lowest sticker price. This group is the best educated (largest percentage with post-graduate study) and of moderate income. They are more likely than other groups to be female, although over half are male. The car types

423 most likely to be considered are subcompact and compact of any body type, but especially hatchback. They are least likely to select General Motors as the preferred manufacturer, although the majority still prefer GM products. The specific makes they are most likely to consider are Datsun, Toyota, Volkswagen, Dodge, and Mercury. Cluster 4, the self-reliant shopper group, considers a larger total number of automobile makes and is higher on total number of personal hours devoted to search than other groups. A member of this group is likely to be male, is less likely to know in advance the dealer from whom he will purchase, and is less likely to have a trade-in. He is well educated, has a moderate to high income compared to members of other groups, and is more likely to be employed as a craftsman or foreman. He is less likely to involve others, except for his wife, in the search process. The reason for purchase is more likely to be that new cars are more fuelefficient or that he wanted a car for a different purpose. Car types most likely to be considered are subcompact or compact, and he is less likely than members of other groups to consider General Motors products. He is most likely to consider Datsun, Fiat, Honda, Toyota, Volkswagen, Dodge, Ford, and Mercury. Cluster 5, the in-store shopper group, has the largest number of decisionmakers involved-especially the wife, when she is not the principal decisionmaker. This group has the highest percentage of currently married individuals. They are less likely than other shoppers to know in advance the dealer from whom they will purchase or to have a tradein. They are higher than other groups (except Cluster 3) on the number hours that others spend in the search process, more likely to consider a large number of makes, and more likely to buy a new car with a higher sticker price. This group is relatively well educated, but does not necessarily earn a large income. They are more likely than other groups to be employed as managers, government officials, or proprietors. The principal reasons for purchase are that new cars are more fuel-efficient or that the old car quit working and needed replacement. The car types they consider are intermediate-sized sedans; the makes they consider favorably are Chevrolet, Oldsmobile, Pontiac, Buick, and Ford, but not Datsun, Volkswagen, or Honda. Oldsmobile and Pontiac, in particular, are disproportionately favored by this group. Cluster 5 shoppers are more likely to buy outside of their original manufacturer set, possibly due to their high level of dealer search, but they are less likely to buy outside of their original price set. Cluster 6, the moderate-search group, devotes a belowaverage number of total hours to search activities, and members of this group exhibit higher certainty on average that they could get a good buy without prior information search. They are more likely to know the manufacturer in advance of purchase, but not necessarily the dealer. They are also more likely to receive a high trade-in value. Along with Cluster 1 (the low-search group), Cluster 6 exhibits higher income (31 percent with incomes over $30,000). This group is older, has the second highest proportion of

424 males, and is less likely than other groups to involve other individuals in the search process. The principal reasons given for purchase are that it is best to trade cars every few years and that new cars are more fuel-efficient. The car type most likely to be considered is a four-door, and this group, along with Cluster 1, is more likely to consider Ford but not GM products. Members of Cluster 6 are less likely to specify particular makes that they would consider favorably, and more likely than other groups to buy outside their original price set. The results of the self-report data are generally consistent with prior research. It is worth noting that no simple relationship was found between the amount and type of search and such variables as confidence in one's ability to judge the product, product experience, and satisfaction with previous or current purchases. Consistent with the findings of Duncan and Olshavsky (1982), there does appear to be a relationship between the amount of search activity and consumers' certainty about their ability to get a good deal without information search. This finding reinforces the importance of consumer beliefs as mediators of search activity. Although the study is cross-sectional, there is little in the results to suggest a longitudinal development of shoppers through the patterns-except that younger, less experienced consumers appear to rely more heavily on the expertise of others, while more experienced buyers do not. This is particularly evident in the purchase-pal-assisted cluster (Cluster 2). A purchase pal is another individual perceived by the purchaser to be knowledgeable about the product category. Information obtained from a purchase pal is clearly interpersonal communication, but it is not merely opinion or prior experience which is being communicated; it is also a statement of how the purchase process should be carried out. While other clusters may have involved others in the search and decision process, these other interpersonal sources are not perceived as more experienced or knowledgeable than the primary purchaser. Purchase-pal-assisted shoppers employ a strategy that appears to substitute the expertise of another person for their own lack of expertise. The effect is a rather modest amount of search activity. This strategy may be contrasted with that employed by the high-search group (Cluster 3). These purchasers have low confidence in their ability to judge cars but believe extensive information search is required to get a good buy. Thus, this high-search cluster spends a great deal of time (their own and that of others) in search activity. Interpersonal sources of information are important for this group, but they are only one of many sources of information. Few empirical studies to date have examined the development of these strategies. Prior studies do seem to suggest the existence of characteristic search strategies similar to those identified here, but few efforts have been made to explain why these patterns may develop. Further understanding of this developmental process may be possible by examining the perspective of the seller. Accordingly, the second phase of this research studied consumer search patterns as viewed from the other side of the buyer-seller dyad-a sample of automobile dealer sales personnel.

THE JOURNAL OF CONSUMER RESEARCH

NEW CAR DEALER SALESPERSON STUDY


If consumers use different search strategies in the automobile purchase process, car salespersons should be aware of at least some of these strategies. Obviously, much search behavior cannot be observed directly by sales personnel, so some of the dimensions of search behavior apparent from consumers' self-reports may not be reported by sales personnel. Nevertheless, salespersons provide the best opportunity for cross-validation of at least some of the characteristics of different shopper types. To gain an initial perspective on salespersons' perceptions of their customers, a series of personal interviews was conducted with sales personnel and sales managers at car dealerships. These initial interviews suggested the search behaviors that would be observable and replicable. Based on these interviews, a structured questionnaire was developed for use with sales personnel.

Method
Questionnaire. A set of 74 items formed the basis of the salesperson questionnaire. These items described potential characteristics of car buyers-e. g., "first-time buyer," "visit prompted by advertisement," "comes to showroom alone." Each item was measured on a sevenpoint scale ranging from "definitely does not apply" to "definitely applies." The questionnaire instructions requested that the salesperson respondents think of a particular customer or type of customer with whom they had personal experience. They were asked to describe that customer briefly in their own words in the space provided at the top of the questionnaire and then describe that customer using the 74-item scale. Respondents were also asked to indicate whether or not they liked to sell to this type of customer. Upon completion of this description, respondents were asked to think of another customer or type of customer whom they felt was different in some important aspect, and to describe that customer in their own words. Again, respondents were asked to describe this customer using the 74-item scale and to indicate whether or not they liked to sell to that type of customer. This procedure was continued until the respondent indicated that he or she could think of no other types of customers to describe. Each respondent also provided personal demographic and sales experience data. Respondents. Forty-eight sales personnel representing eight Nashville, Tennessee automobile dealerships participated in the study. Three of the dealerships had separate import and domestic sales staffs, which were treated as separate dealerships, bringing the total number of "dealerships" represented in the study to 11. These dealerships represented AMC, Buick, Cadillac, Chevrolet, Datsun, Fiat, Honda, Jaguar, Mercedes-Benz, Oldsmobile, Pontiac, Renault, Subaru, Toyota, and Volvo. The number of salespersons participating in the study per dealership ranged

SEARCH STRATEGIES from two to 10, and was largely a function of the size of the dealership and the proportion of experienced sales personnel employed. Only the "most experienced" sales personnel, as identified by the sales manager, were asked to participate in the study. Respondents' experience selling cars ranged from five to 31 years. Procedure. Each respondent was approached by one of the authors and invited to participate in the study. This initial contact provided an opportunity to give verbal instructions for completing the questionnaire and to answer any questions respondents might have. Respondents were left with a copy of the questionnaire and were requested not to discuss their responses with others. Questionnaires were picked up from respondents at a prearranged time several days later. Salespersons described from two to eight different customer types; typically, each one described three or four types. For purposes of data analysis, each description provided by a respondent was treated as a separate observation. A total of 194 such observations formed the basis for data analysis.
TABLE 5

425

INDIVIDUAL ITEMS RELATEDTO DIMENSIONSOF SEARCH: SALESPERSON DATA Factor 1 Making decision alone (-.88) More than one person makes decision to buy (.84) Others (family, etc.) involved in decision (.81) Comes to showroom alone (-.75) Has advisor (spouse, friend, parent, etc.) along (.74) Factor 2 Spent a lot of time looking at models on the showroom floor (.76) Returns to dealership several times before purchasing (.75) Spent a lot of time test driving cars (.75) Spent a lot of time talking to sales personnel (.74) Obtained manufacturers' brochures describing cars (.55) Has visited many other dealerships (.47) Factor 3 Visit prompted by advertisement (.70) Telephones for informationon model availability,prices, etc. (.62) Referred by previous customers (.58)
NOTE: Factor loadings are in parentheses.

Analysis and Findings


Factor Analysis. Fourteen of the 74 total items were specifically related to information search behavior. These were submitted to a factor analysis, which resulted in three factors that accounted for 50 percent of the total variance. These factors were submitted to both VARIMAX and OBLIMIN rotations, and the VARIMAX solution was retained when the oblique solution failed to increase the hyperplane count or to make the results of the orthogonal rotation more interpretable. Table 5 indicates which items loaded specific factors. Not surprisingly, the factor solution is considerably simpler than the one obtained from the consumer self-report data. Factor 1 appears to represent the participation, or lack of participation, of multiple decisionmakers in the purchase process. In the self-report data, self- and other-in-store search loaded on the same factor and appeared to vary together, but car sales personnel seem to treat self- versus other-involvement as bipolar. This difference is probably due to differences in frames of reference and objectives between customers and sales personnel. Customers are information gatherers and undoubtedly use as sources (perhaps extensively) other persons whose influence will not be apparent to salespersons. Sales personnel attempt to close the sale as expeditiously as possible; they are likely to be aware of and react to only those other persons who are physically present in the showroom or otherwise actively involved in the final decision. Factor 2 represents a retail search pattern and is loaded by items related to both the number of visits to dealerships and the amount of time spent at the dealership. This factor appears to correspond to two retail search factors in the consumer self-report data-one related to the number of visits to dealer showrooms, and the other related to time

spent on in-store activities. (However, these two factors were positively correlated.) Given that sales personnel can only infer the number of total visits to different dealerships and time spent at other dealerships, the differences among the retail dimensions in the two solutions is not surprising. Factor 3 is a general out-of-store search factor. Unlike the results obtained from the consumer data, where three separate out-of-store search factors emerged, only one emerged from the salesperson analysis. This is not unreasonable, since sales personnel have limited opportunity to learn of out-of-store search activities. Cluster Analysis. Factor scores for each observation were computed and used as the basis for cluster analysis. An initial hierarchical clustering procedure was employed to obtain a candidate number of clusters and seed points for a k-means cluster analysis. Although the limited number of observations prohibited split-sample cross-validation, the same six-cluster solution was obtained with random and nonrandom seed points, lending additional credibility to the final solution. Table 6 provides the cluster means on each factor. Clusters were then compared on all 74 variables measured in the questionnaire. Table 7 is a thumbnail summary of these results.5 Only variables for which statistical differences among groups were obtained are reported. Cluster 1 appears to be a retail search group seeking a good price, hence the label "the negotiator." Cluster 2 is composed of inexperienced shoppers who may be assisted in the purchase process by an "advisor." Cluster 3 is composed of shoppers who engage in little search activity and make their purchase decision alone. Cluster 4 is also made up of low-search shoppers, but appears to involve the family in what in-store search does take place. Cluster 5 is
5A complete set of results is available from the authors upon request.

426 TABLE 6

THE JOURNAL OF CONSUMER RESEARCH

CLUSTER MEANS OF DERIVEDFACTORS:SALESPERSON DATA Cluster 1 Factor 1-Participation of others Factor 2-In-store activities Factor 3-Out-of-store activities .94 .80 -1.17 Cluster 2 .44 .42 .75 Cluster 3 -1.10 -.21 -1.22 Cluster 4 1.24 -1.07 - .24 Cluster 5 -.86 .94 .19 Cluster 6 -.76 -1.21 .62

TABLE 7 THUMBNAIL DESCRIPTIONOF CLUSTERS DERIVEDFROM SALES FORCE DATA

Cluster 1 (The Negotiator) Spends much time at dealers (talkingto salespeople, looking at models, test driving cars, etc.) Returns to dealership several times before purchasing Gets a "good deal" Higher income Works as manager or professional Better educated A "hard sell" Least liked of all clusters by sales personnel Likely to be married and have wife involved in purchase decision Looking for a family car Seeks good gas mileage, good dealer service department, and good price Cluster 2 (The Inexperienced Shopper) Likely to be a first-timebuyer Equally likely to be female as male Not an efficient shopper Has an unrealistic, ideal car in mind Visits many dealerships and returns several times before purchasing Looks at many different models, attains many price quotes, test drives many cars Unlikely to make a purchase (at a given dealership) Telephones for information Likelyto be referred to dealership by previous customer Obtains manufacturer brochures Most likely to be assisted by an advisor Conernd bou fnanin Concerned about financing Lower income, less educated, younger than average More likely to be single Has trouble qualifying for purchase Neither liked nor disliked by sales personnel Cluster 3 (The Lone Shopper) Makes purchase alone Knows exactly what he wants Does not visit many dealers or consider many models Most experienced shopper Engages in little informationsearch (does not telephone, obtain brochures) Visit not prompted by either a referralor an advertisement Most concerned about dealer service department Older An "efficient"shopper Sales personnel like this shopper best (tie with Cluster 6)

Cluster 4 (The Family Shopper) Most likely to have other family members involved in the decision Likely to have small children along when visits the showroom Spends relatively littletime at dealership Looks at fewest models of any cluster Not a hard bargainer but does not get a "bad deal" Above average likeability Cluster 5 (The Pain-in-the-Neck) Hardest bargainer Hardest to please Least likely to make a purchase (at a given dealership) Not well liked by sales personnel Visits many dealers (and communicates this to sales personnel) Returns to dealerships several times More likely than average to order a car rather than buy off the lot Middle-aged Most likely of all clusters to have an unrealistic, ideal car in mind Product attributes of concern are gas mileage and dealer service department Does not necessarily get a "good deal" Cluster 6 (The Moderate-Search Shopper) Most likely to make a purchase (from a given dealership) Visits few dealerships Easy to please Spends least time on in-store search activities of any cluster Snds east time on nanyu Knows exactly what he wants
Upper income

Well educated Weledated Mieaged Well dressed Comes to showroom alone Self-assured Visit most likely to be triggered by an advertisement Engaged in out-of-store search Best liked of all clusters (tie with Cluster 3)

SEARCH STRATEGIES
FIGUREA OF RELATIONSHIP CUSTOMER SELF-REPORT CLUSTERS AND CLUSTERS BASED ON SALES PERSONNEL REPORTS

427

Self-Report
Purchase-Pal-Assisted Moderate-Search Retail Shopper (5) Shopper (2)

Sales-Personnel
Inexperienced Shopper (2) Shopper (6)

Shopper (6)

Moderate-Search Negotiator (1)

amily Shopper (4) Low-Search Shopper (1) Lone Shopper (3) High-Search Shopper (other) (3 Pain-in-the-Neck Self-Reliant Shopper (4)
NOTE: Numbers in parentheses are the cluster numbers (see Tables 4 and 7).

Shopper (5)

composed of high-search individuals who are difficult to sell and hard to please. Members of Cluster 6 manifest a moderate amount of search activity, relying more heavily on out-of-store information sources, such as advertising, than on in-store sources of information. It is not possible to compute an objective measure of congruence between the clusters obtained from the self-report data in Phase 1 and the sales personnel data in Phase 2, but there does appear to be a reasonable degree of congruence between the two independent sets of clusters. Figure A is a comparison of the clusters with the most apparent similarities. There are definitely differences (many of them probably due to differences in perspective), but both data sets yield six clusters that are surprisingly similar. Both sets of data manifest low-search clusters, retail shoppers, inexperienced shoppers, high-search shoppers, and moderatesearch shoppers, confirming what has been found in previous studies. Yet the two sets of factors are complementary, in that differences do exist in the two sets of clusters. For example, the cluster that appears as a low-search cluster in the self-report data seems to be composed of two of the clusters derived from the sales personnel: a family shopper who may find it difficult to shop because there are small children present, and a lone shopper who knows exactly what he or she wants. The two high-search clusters in the self-report data-one entailing heavy involvement of others

and the other confined to the principal decisionmaker-appear to collapse into a single high-search cluster in the sales personnel data. From the perspective of sales personnel, it probably matters little how many persons are actively involved in the search process; these are simply difficult customers. The sales personnel data also shed further light on the retail shopper and the moderate-search shopper in the self-report data. It appears that the strictly retail shopper is price shopping, seeking the best buy at the lowest price. This strategy requires visiting numerous retailers, comparing prices, and actively negotiating price. The moderate shopper is apparently less concerned with price and more likely to use out-of-store search, particularly advertising. It is interesting to note that the sales personnel who participated in the study did not tend to readily offer types of customers based on search strategy. Rather, they appeared to respond to specific characteristics of customers. Informal discussions with sales personnel in the early phases of the study suggested that they tended to use cues such as "pipesmoker," "carries a clipboard," and so on as indicators of a time-consuming or difficult sale, while recognizing that people with small children tended to be rather easy or quick sales. Additional study of how sales personnel process and use information would be interesting. Based on anecdotal evidence, it appears that sales personnel do not make the most

428 efficient use of the information about consumers that is available to them. Potentially useful information about consumers appears to be rather nonsystematically processed. This obviously has important normative implications for sales management. Also of significance is the degree to which sales personnel reported liking or not liking particular customer groups. It is quite clear that sales personnel prefer selling to customers who are involved in lesser amounts of search activity and who have a clearer idea of the product they wish to purchase. Least liked are those customers who actively engage in negotiation. Such customers appear to consume large amounts of a salesperson's time. The results of the sales personnel data provide additional support for the existence of systematic search patterns similar to those identified in prior research. The fact that the results are not a perfect replication suggests either differences in frame of reference, considerable noise in both data sets, or both. Further work on the seller side of this puzzle, as well as on the buyer side, would be illuminating.

THE JOURNAL OF CONSUMER RESEARCH Petty, and Schumann (1982) has supported the suggestion that heuristic processing is more likely in low involvement settings. Wright (1974) and Chaiken and Eagly (1976) have found increased heuristic processing under conditions of distraction. This work on heuristic versus systematic processing provides a framework for explaining some of the search patterns identified in the two studies reported here. In the absence of significant prior experience and when there is low self-confidence in one's ability to judge a product, a simple heuristic may be employed, such as asking for advice from someone perceived as knowledgeable about the product. This appears to be the process at work among members of the purchase-pal-assisted cluster identified in the self-report data. When shoppers are under time pressure or are distracted, heuristics are more likely to be employed. This may explain the low level of search among members of the family-shopper cluster identified in the sales personnel data. In contrast, the moderate-search groups spend relatively little time on active search because they are more experienced car buyers who have been satisfied with prior purchases. Their most involved (i.e., systematic) activity is focused on out-of-store search, particularly advertising. Advertising appears to trigger a dealer visit for this group. Anderson (1980, 1982), Neves and Anderson (1981) and others (Shiffrin and Dumais 1981; Schiffrin and Schneider 1977; Schneider and Shiffrin 1977) have argued that, with experience, cognitive processes may be overleamed and become automatic. This automaticity occurs as a result of integration of information and procedures in a "chunking" process. In effect, more elaborate processing of earlier experiences is replaced by cognitive shortcuts, which require less attention and function as heuristics. This notion has found support in studies demonstrating that experienced consumers tend to use fewer attributes to evaluate products and rely more heavily on more global evaluations (Bettman and Park 1980; Edell and Mitchell 1978; Johnson and Russo 1980, 1981; Park and Lessig 1981; Russo and Johnson 1980). It is reasonable to assume that the two lowest-search groups found here are exhibiting automaticity of choice. Indeed, both groups' extensive experience with the product class is consistent with this hypothesis. In addition, the low-search group reports a greater likelihood of knowing in advance both make and dealer, while the moderatesearch group is more likely to know only the make(s) of interest in advance of the search process. This may suggest that the low-search group is further along in the development of automated decisionmaking. This hypothesis, while speculative, is more appealing than the notion that low search among these consumers is due to lack of ability to process information or to laziness. The extensive search exhibited by two of the clusters identified here may be attributed to (1) a high degree of involvement with the product class, (2) a nonautomatic decision process-i.e., systematic processing, due possibly to lack of experience or to a conscious effort to avoid the use of decision heuristics and thereby satisfy oneself that one has done all one

DISCUSSION
While it now seems clear that systematic search strategies are common among consumers and identifiable both from the self-report data of shoppers and from sales personnel, it is less clear why such patterns emerge. The amount of search activity a consumer actually engages in is a function of numerous factors. Newman (1977), Bettman (1979), and Moore and Lehmann (1980) have all suggested categories of variables that are related to amount of external search. These categorizations generally include product knowledge and experience, including satisfaction with prior purchases; individual differences, such as ability; situational variables, such as time pressure; and product importance. Punj and Stewart (1983b) suggest that an interaction of situational and individual difference characteristics may produce distinctive patterris of search and decisionmaking. Recently, Chaiken (1980, 1982) has suggested that a distinction should be made between heuristic and systematic processing in choice behavior, in order to understand how various factors influence decisionmaking and information search strategies. In the systematic processing mode, decisionmakers actively attempt to comprehend and critically evaluate information about relevant attributes of alternatives. In the heuristic mode, decisions are based on a more superficial assessment of cues. The systematic mode requires detailed processing of information content, whereas the heuristic approach emphasizes the role of simple schemas or cognitive heuristics. Chaiken (1982) has suggested that heuristics are most likely to be used by individuals with low involvement in the decision, by those who do not have the ability or expertise to engage in systematic processing, and by those faced with tasks or distractions that are difficult for them. This view is consistent with the interaction framework of decisionmaking proposed by Punj and Stewart (1983b). Research by Chaiken (1980) and by Caccioppo,

SEARCH STRATEGIES FIGURE B AS PATTERNS OF CONSUMER DECISIONMAKING A FUNCTION OF INVOLVEMENT, ABILITY, DISTRACTION, AND LEVELOF AUTOMATICITY
High Involvement | (e.g. durable products) |
s |

429

Low Involvement (e.g., package goods)

Inability to

Distraction

Ability to Process

Heuristic Processing AdvisorAssisted

Heuristic Processing Family (in-store)

Automaticity ("chunking") Low-Search Shopper

Systematic Processing

Heuristic Processing

Preferred Mode(s) of Information Acquisition

Shopper

Shopper

&
Moderate-Search Shopper (out-of-store)

out-of-store search

involvement of others

in-store search

Moderate-Search Shopper (in-store); Self-Reliant Shopper; & High-Search Shopper

could to understand the alternatives by a thorough search, or (3) the existence of preferred modes of information acquisition. Punj and Staelin (1983) have offered an empirical test of the proposition that product knowledge in memory consists of two unique components: (1) knowledge of specific attributes associated with product alternatives, as well as general shopping procedures for the particular product, and (2) a general knowledge structure about the product and/or purchase decisions in general. The former construct would tend to decrease external search, while the latter would tend to increase external search-at least up to a point, since it provides a frame of reference for new product information. In the context of a consumer decision task, these findings add further weight to the systematic/automatic processing explanation of the results of the present study. A significant amount of empirical evidence now exists on consumer search strategies. This empirical evidence has yet to be integrated within an explanatory framework. It is not premature to offer such an explanation, even if speculative. The automaticity/systematic processing notion advanced by Chaiken (1982) appears to have promise, although it remains to be tested and other theories may be put forth. Figure B provides a schematic diagram of the relationship among the patterns of search behavior identified here and the constructs of involvement, ability, distraction, and automaticity. This figure represents but one of many possible schematics that might capture the inter-

action of situational and individual difference characteristics. Several of the patterns identified in the present study have been observed by other researchers, not only for automobiles, but for a variety of durable goods. Since all such goods tend to carry high costs and risks, it would be unreasonable to assume that the same patterns of search obtained in research with durable goods would necessarily be manifest with other types of products that are less involving. Indeed, it is likely that heuristic processing is even more dominant for low-cost, low-risk, frequently purchased consumer goods.

SUMMARY AND CONCLUSION


This paper replicates and extends previous research on consumer search patterns and lends additional empirical evidence to support the existence of such patterns. It suggests that there is considerable similarity in the perceptions of buyers and sellers regarding search strategies employed by consumers, but identifies enough differences in these perceptions to warrant further exploration of their sources. We have also presented a tentative framework that integrates and suggests an explanation for observed patterns of search. A number of questions for future research are suggested by this framework: * Do the heuristics used by consumers under the various branchesof Figure B differ?If so, how and why?

430 * How do preferredmodes of information searcharise? * What changes occur in the decision process over time? Does an automated,high-involvement processeventually become a low-involvementprocess?How do searchstrategies change with situations-for example, whathappens if the distractedconsumeris not distractedon otherpurchase occasions? * To what extent do heuristicsdeveloped underone set of conditions, such as inabilityto processor distraction, generalize to other purchasesituations? An interesting and potentially useful finding of the studies reported here is the ability of experienced sales personnel to identify types of shoppers, although they are currently less systematic about this process than might be desirable. Informal discussions with sales personnel suggested that they tend to use different sales approaches with different shopper types. This has potential relevance for sales training and management. It should be possible to teach sales personnel to recognize characteristic search patterns and adapt their sales approaches to these types. This could take the form of formal training rather than "learning by doing." In addition, sales productivity may be increased by identifying those customer types that engage in protracted information search and deliberation prior to purchase. To the extent that alternative modes of information presentation may be substituted for the salesperson, the time of the sales personnel could be used more productively. Further, it may prove unnecessary to provide certain types of information to particular sets of customers. These managerially relevant hypotheses are equally worthy of pursuit in future research. [Received March 1983. Revised October 1983.]

THE JOURNAL OF CONSUMER RESEARCH


Chaiken, Shelly (1980), "Heuristic Versus Systematic Information Processing and the Use of Source Versus Message Cues in Persuasion," Journal of Personality and Social Psychology, 39 (November), 752-766. (1982), "The Heuristic/Systematic Processing Distinction in Persuasion," paper presented at the Symposium on Automatic Processing, Society for Experimental Social Psychology, Nashville, TN. and Alice H. Eagly (1976), "Communication Modality as a Determinant of Message Pervasiveness and Message Comprehensibility," Journal of Personality and Social Psychology, 34 (October), 605-614. Claxton, John D., Joseph N. Fry, and Bernard Portis (1974), "A Taxonomy of Purchase Information Gathering Patterns," Journal of Consumer Research, 1 (December), 35-42. Duncan, Calvin P., and Richard,W. Olshavsky (1982), " External Search: The Role of Consumer Beliefs," Journal of Marketing Research, 19 (February), 32-43. Edell, Julie A. and Andrew A. Mitchell (1978), "An Information Processing Approach to Cognitive Response," in Research Frontiers in Marketing: Dialogues and Directions, ed. Subhash C. Jain, Chicago: American Marketing Association, 178-183. Fleiss, Joseph L. and Jacob Cohen (1973), "The Equivalence of Weighted Kappa and the Intraclass Correlation Coefficient as Measures of Reliability," Educational and Psychological Measurement, 33(3), 613-619. Johnson, Eric J. and J. Edward Russo (1980), "The Organization of Product Information in Memory Identified by Recall Times," in Advances in Consumer Research, Vol. 5, ed. H. Keith Hunt, Chicago: Association for Consumer Research, 79-86. and J. Edward Russo (1981), "Product Familiarity and Learning New Information," in Advances in Consumer Research, Vol. 8, ed. Kent B. Monroe, Ann Arbor, MI: Association for Consumer Research, 151-160. Katona, George and Eva Mueller (1955), "A Study of Purchase Decisions in Consumer Behavior," in Consumer Behavior: The Dynamics of Consumer Reaction, ed. Lincoln H. Clark, New York: New York University Press, 30-87. Kiel, Geoffrey C. and Roger A. Layton (1981), "Dimensions of Consumer Information Seeking Behavior," Journal of Marketing Research, 18 (May), 233-239. Moore, William L. and Donald R. Lehmann (1980), "Individual Differences in Search Behavior for a Nondurable," Journal of Consumer Research, 7 (December), 296-307. Neves, D. M. and John R. Anderson (1981), "Knowledge Compilation: Mechanisms for the Automation of Cognitive Skills," in Cognitive Skills and Their Acquisition, ed. John R. Anderson, Hillsdale, NJ: Lawrence Erlbaum. 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, Amsterdam: North-Holland, 79-94. and Richard Staelin (1972), "Prepurchase Information Seeking for New Cars and Major Household Appliances," Journal of Marketing Research, 9 (August), 249-257. and Richard Staelin (1973), "Information Sources of Durable Goods," Journal of Advertising Research, 13 (April), 19-29. Newsweek (1977a), "Buyers of New Domestic Cars, 1977," company publication.

REFERENCES
Anderson, John R. (1980), Cognitive Psychology and Its Implications, San Francisco: W.H. Freeman and Co. (1982), "Acquisition of Cognitive Skill," Psychological Review, 89 (July), 369-406. Bell, Gerald D. (1967), "Self Confidence and Persuasion in Car Buying," Journal of Marketing Research, 4 (February), 46-52. Bettmnan,James R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA: Addison-Wesley. and C. Whan Park (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis," Journal of Consumer Research, 7 (December), 234-248. and Michael A. Zins (1977), "Constructive Processes in Consumer Choice," Journal of Consumer Research, 3 (September), 75-85. Cacioppo, John T., Richard E. Petty, and David Schumann (1982), "Message Factors, Cognitive Responses, and Advertising," paper presented to the Division of Consumer Psychology of the American Psychological Association, Washington, D.C.

SEARCH STRATEGIES
(1977b), "Buyers of New Foreign Cars, 1977," company publication. Park, C. Whan and V.P. Lessig (1981), "Familiarity and Its Impact on Consumer Decision Biases and Heuristics," Journal of Consumer Research, 8 (September), 223-230. Punj, Girish and David W. Stewart (1983a), "Cluster Analysis in Marketing Research: Review and Suggestions for Application," Journal of Marketing Research, 20 (May), 134-148. and David W. Stewart (1983b), "An Interaction Framework of Consumer Decision Making," Journal of Consumer Research, 10 (September), 181-196. and Richard Staelin (1983), "A Model of Consumer Information Search Behavior for New Automobiles," Journal of Consumer Research, 10 (March), 366-380. Russo, J. Edward and Eric J. Johnson (1980), "What Do Consumers Know About Familiar Products?" in Advances in Consumer Research, Vol. 7, ed. Jerry C. Olson, Ann Arbor, MI: Association for Consumer Research, 417-423. Schneider, Walter and Richard M. Shiffrin (1977), "Controlled and Automatic Human Information Processing: I. Detection, Search, and Attention," Psychological Review, 84 (January), 1-66.

431
Shiffrin, Richard M. and Dumais, S. (1981), "The Development of Automaticism, " in Cognitive Skills and Their Acquisition, ed. John R. Anderson, Hillsdale, NJ: Lawrence Erlbaum. and Walter Schneider (1977), "Controlled and Automatic Human Information Processing: II. Perceptual Learning, Automatic Attending, and a General Theory," Psychological Review, 84 (March), 127-190. Staelin, Richard (1969), "Information Seeking and Duration of the Purchase Decision Process for New Cars and Major Household Appliances," unpublished Ph.D. dissertation, University of Michigan. U.S. News and World Report (1974), "The Buyers of New 1974 Automobiles," Marketing Department of U.S. News and World Report. (1975), " 1975 Study of American Market Automobiles," Marketing Department of U.S. News and World Report. Westbrook, Robert A. and Claes Fornell (1979), "Patterns of Information Source Usage Among Durable Goods Buyers," Journal of Marketing Research, 16 (August), 303-312. Wright, Peter (1974), "The Harassed Decision Maker: Time Pressures, Distraction, and the Use of Evidence," Journal of Applied Psychology, 59 (October), 555-561.

Vous aimerez peut-être aussi