Identifying the Characteristics of Single Retail (Dealer) Visit New Automobile Buyers

Philippe Cattin, University of Connecticut
Girish Punj, University of Connecticut
ABSTRACT - Several recent marketing research studies have reported that 25 to 50% of all new car buyers purchase from the first automobile dealer they visit. In terms of the domestic market, these buyers account for approximately thirty billion dollars in sales every year. The present research focuses on this lucrative segment. Using a sample of 1056 new automobile buyers, an attempt is made to isolate the demographic, psychological and behavioral characteristics of the single retail visit new car buyer. A-discriminant analysis methodology is used. The findings have implications for the marketing strategies of both automobile manufacturers and dealers
[ to cite ]:
Philippe Cattin and Girish Punj (1983) ,"Identifying the Characteristics of Single Retail (Dealer) Visit New Automobile Buyers", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 383-388.

Advances in Consumer Research Volume 10, 1983      Pages 383-388

IDENTIFYING THE CHARACTERISTICS OF SINGLE RETAIL (DEALER) VISIT NEW AUTOMOBILE BUYERS

Philippe Cattin, University of Connecticut

Girish Punj, University of Connecticut

ABSTRACT -

Several recent marketing research studies have reported that 25 to 50% of all new car buyers purchase from the first automobile dealer they visit. In terms of the domestic market, these buyers account for approximately thirty billion dollars in sales every year. The present research focuses on this lucrative segment. Using a sample of 1056 new automobile buyers, an attempt is made to isolate the demographic, psychological and behavioral characteristics of the single retail visit new car buyer. A-discriminant analysis methodology is used. The findings have implications for the marketing strategies of both automobile manufacturers and dealers

INTRODUCTION

The purpose of this research is to determine the characteristics of new automobile buyers who buy from the first dealer they visit. There are two reasons why such knowledge would be useful. First, single retail visit buying is a viable shopping strategy. In fact, it is the most frequently adopted strategy for Routinized Response Behavior (RRB) in consumer decision situations. In Limited and Extensive Problem Solving (LPS and EPS), the single retail visit strategy is less called for theoretically, but nevertheless used. Thus an understanding of the variables and conditions associated with single retail visit buying might add to existing research on shopping strategies, which has experienced a large increase in recent years (see next section). The research question in the present paper is distinct, but ye. related to this body of knowledge.

A second reason for studying new automobile buyers who buy from the first dealer they visit is more managerial. Numerous marketing research studies have reported that 25 to 50% of all new car buyers purchase from the first automobile dealer they visit (for example, see U.S. News & World Report, 1974, 1975, Newsweek, 1977, 1980). In terms of the domestic market alone, these buyers account for about $30 billion in sales every year The present research focuses on this lucrative segment. Information on the characteristics of the consumer who buys from the first dealer he visits can be used by marketers at various levels and in a variety of ways (see concluding section)

RELATED PREVIOUS RESEARCH

There are a large number of studies on information search in general, and on the morphology of shopping patterns in particular. Notable among the studies in the first category are those of Katona and Mueller (1955), Bennett and Mandell (1969), Newman and Staelin (1972), Claxton, Fry and Portis (1974) and Newman and Lockeman (1975) The findings of these studies are summarized in Newman (1977) and Bettman (1979) and shall not be reiterated here. Studies that can be classified in the second category (i.e. those that deal with the morphology of shopping patterns) include Dommermuth (1965), Bucklin (1966), Darden and Reynolds (1971), Tauber (1972), Williams and Dardis (1972), MacKay and Olshavsky (1975), Peter and Tarpey (1975), Blattberg, Peacook and Sen (1976), and Blattberg, Buesing, Peacock and Sen (1978). Several of these studies attempt to identify shopping orientations or profiles and are summarized in Granbois (1977). For example, Bucklin (1966) developed two shopping orientations: "intratrip shopping" (number of interstore comparisons made during a single trip) and "intertrip shopping" (number of shopping trips made for the purchase of a product). In analyzing data on the shopping behavior of 506 women who had purchased items worth five dollars or more, he found a strong positive relationship between the extent of intratrip and intertrip shoPPing.

Along similar lines, Dommermuth (1965) developed a matrix typology of shopping strategies based on a study of 845 recent purchasers of refrigerators, TV sets, washing machines, vacuum cleaners and irons. The typology classified buyers along two dimensions: number of retail outlets shopped and number of brands examined. One extreme of this typology consisted of one store-one brand shoppers and the other of buyers who examined several brands in several stores. The distribution of these shopper profiles varied across product categories, with 58% of the refrigerator buyers shopping in more than one retail outlet as opposed to 18% of the iron buyers These results indicate that the proportion of consumers who buy from the first store they visit is quite large (for products which most of us would put into Limited or Extensive Problem Solving Situations). Dommermuth did not collect any data to identify the characteristics of consumers who buy from the first store they visit.

Darden and Reynolds (1971) used a series of psychographic statements to categorize shoppers into five "shopping orientations," originally proposed by Stone (1954). The labels used to describe the categories were "local store personalizing," "shopping apathy," "economic shoppers," "ethical support for local merchants" and "chain store depersonalization." This (and other) classifications have not been validated. Blattberg, Peacock and Sen (1976) studied the extent to which shoppers use similar brand choice and store choice strategies across product categories. They found that a relatively large proportion of their sample used identical store choice strategies across the product classes studied.

Despite this body of empirical research, there has been relatively little theory development in the area of information search in the context of shopping strategies. Perhaps, the most significant attempt at constructing a theory of information search is the work of Bettman (1979). His model ties together previous research from economics and psychology on the subject. According to Bettman's model consumers weigh the costs of obtaining information against the benefits which might be expected from using that information. The trade-offs are assumed to be made in a heuristic rather than optimizing fashion. Numerous types of costs have been suggested in the literature. These include the costs of a perceived delay in the decision [see Engel, Blackwell and Kollat (1978)], frustration such as fighting traffic and dealing with incompetent sales persons [see Downs (1961)] and out-of-pocket monetary expenses such as tolls, gasoline, etc. The benefits of search include increased satisfaction with the purchase, the psychological feeling or having done a thorough job [see Engel, Blackwell and Kollat (1978)] and the monetary benefits of getting a good deal. The cost-benefit model of information search has been directly or indirectly tested by Burnkrant (1976) and Pollay (1970).

HYPOTHESES

In the present study, the cost-benefit model of information search presented above, was used to generate several testable hypotheses about the behavior of interest. An application of the model would suggest that the incidence of single dealer visit shopping would be greatest for consumers with a high cost of obtaining information and/or low perceived benefit of information usage Consumers with high costs of information search would be those who tend to place a greater value on their time. Thus we would expect high levels of income and education to be associated with the behavior of interest. Older consumers are more likely to have higher psychological costs of search (e;g., frustration). Hence, age is likely to be a factor as well. In addition, the cost of information search is likely to be higher for consumers who have professional, technical, managerial positions or other positions of responsibility, but lower for households that have a large number of adults (because of division of labor).

Consumers with a smaller choice set of alternatives are likely to have a low perceived benefit associated with information search. Hence, they are more likely to engage in single dealer shopping. In similar fashion, consumers with a high level of confidence in their ability to judge cars are less likely to visit more than one dealer. The perceived benefits of information search are also likely to be low for consumers who were satisfied with a car they purchased from a dealer before, or who know somebody at a dealership. In addition, consumers who have more experience with cars (e.g. who purchased a car recently or owned more cars over the last several years), or who need a car urgently (e.g., because their car was in an accident or needs repairs that are too expensive to be worthwhile) are also likely to spend less time searching for information, and thus likely purchase from the first dealer they visit.

In summary, a consumer is more likely to purchase a car from the first dealer he visits

(a) if his choice set of alternatives is relatively small,

(b) if he has confidence in his ability to judge cars and if he feels he knows a lot about cars,

(c) if he has had more purchase experience with cars recently.

(d) if he has earlier purchased a car from the first dealership he visits and was satisfied with it, or if he knows someone at the first dealership he visits

(e) if he needs a car urgently,

(f) if the number of adults in his household is high, if he is older or more educated, if he has a high income or is a professional

DATA AND SAMPLE

The data were collected from households, living in the SMSA's of Buffalo, Milwaukee and Phoenix, who purchased a new automobile for their personal use in the period September to November, 1978. Three stratified random samples of sizes 1000, 978 and 984 respectively were drawn from the study population. The make of the car was used as the stratification variable. The sampling frame was obtained from R. L. Polk & Co. in Detroit.

The selected sample elements were first contacted by telephone to solicit their cooperation in the study. This was done to achieve a dual purpose. Firs c, it was felt that an initial screening of the sample was desirable to identify the chief decision maker(s). Second, due to length of the research instrument, prior notification of the sample was believed to be necessary to generate an acceptable response rate.

The telephone prenotification resulted in a total of 1561 consumers agreeing to participate in the study. These 1561 consumers were mailed a questionnaire, as were the 813 who could not be reached during the telephone prenotification. Hence, a total of 2374 questionnaires were sent out. 1056 useable responses were received. 61% of the 1561 respondents who agreed to participate in the study and 135 of the 813 who could not be prenotified returned the completed questionnaire.

Several analyses were performed to check the consistency and reliability of the data. The first type of check involved a comparison of the responses to those items in the research instrument which requested essentially the same information. More than 952 of the respondents displayed consistency across the multiple responses. In order to estimate effect of "forgetting," a variable to measure the elapsed calendar time between the purchase of the new scar and the return of the questionnaire was constructed. 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 selected questions in the research instrument. The null hypothesis was that the four subsamples were drawn from four identically distributed populations. A chi-square statistic was used to determine rejection/non-rejection of the null hypothesis. Using a 95: significance level, the null hypothesis could be rejected in only a few instances. While there is no way to guarantee that the respondents filled out the research instrument before "forgetting" took place, the evidence does appear to point in that direction. In similar fashion, another set of reliability checks were performed to compare the prenotified responses with the nonprenotified responses. No appreciable differences were found. The last set of reliability checks involved comparing the sample distribution; obtained in this study (for variables like number of retail (dealer) visits made, number of previous car purchases, etc.) with those obtained in previous studies (U.S. News & World Report, 1974, 1975, Newsweek, 1977, 1980). Once again no appreciable differences were discernible. In light of the above validity and reliability checks, it was felt that the data was of high enough quality to merit their use in answering the research question.

ANALYSIS

The variables used in the study and their operational definitions are shown in Table 1. The number of dealers visited was used to create the dependent variable (variable 16 in Table 1). The consumers who visited more than one dealer were classified in group 1, and those who purchased from the first dealer they visited in group 2. Fifteen variables were used as predictor variables. These variables were classified into six categories. The six categories and the variables included in each category are as follows:

(a) Prior set of feasible cars: size of the feasible set (variable 1) and knowledge of the manufacturer to purchase from (variable 2);

(b) Prior confidence and expertise about cars: confidence in ability to judge cars (variable 3), perceived ability to make a wise purchase (variable 4) and car expertise in whole household (variable 5);

(c) Previous experience with cars: length of time since purchasing a new car (variable 6) and number of cars purchased in last ten years (variable 7);

(d) Previous experience with first dealer visited: purchased car from dealer before and was satisfied with it (variable 8) and knows someone at first dealer visited (variable 95:

(e) Situational Variable: urgency of purchase (variable 10):

(f) Demographic Variables: number of adults in household (variable 11), age (variable 12), education (variable 13), income (variable 14), and occupation (variable 15).

Table 1 shows how these 15 variables were measured.

TABLE 1

VARIABLES AND THE MEASURES USED

Five of the predictor variables are discrete (0,1) variables. They are: knowledge of manufacturer to purchase from, car expertise in household, purchased car from dealer before and was satisfied with it, knows someone at first dealer visited, and urgency of purchase. The other variables are measured on 5 to 11 point scales, except for size of the feasible set (an integer), and length of time since purchasing a new car (months). The sample means obtained for groups 1 and 2 are shown in Table 2, along with the t-values for the difference between means. For the five discrete (0,1) variables, Table 9 shows the sample proportions (instead of means) and the t-values of the difference between proportions. Six variables are significant in the expected direction beyond the 12 level: size of feasible set, knowledge of manufacturer to purchase from, perceived ability to make a wise purchase, purchased a car from dealer before and was satisfied with it, knows someone at first dealership visited, and age. Two variables are significant in the wrong direction at the 1% level: education and number of adults in household. One more variable is significant at the 5% level in the expected direction: length of time since purchase or new car.

TABLE 2

MEAN OF THE VARIABLES IN EACH GROUP AND T-VALUES OF DIFFERENCE

The predictor variables can be incorporated in a linear discriminant analysis. The advantages of a discriminant analysis over means (and proportions) are that it provides:

(a) a measure of the strength and significance of the differences between two groups using all predictor variables at once, instead of one variable at a time;

(b) a linear combination of the predictor variables which "best" discriminates between two groups, and predicts group membership:

(c) a measure of the relative importance of each predictor variable, indicated by the standardized discriminant coefficients.

Three discriminant analyses were run: one using all the predictor variables, the second with the nondemographic variables only (variables 1 through 10) and the third with the demographic variables only. The discriminant analysis program used was the forward stepwise procedure or SPSS Version H, Release 9A. To be included in the final step, a predictor variable had to have a partial F-value of at least 1.0.

The results or the three discriminant analyses are shown in Table 3, which reports:

(a) the standardized discriminant coefficients of the variables included in the last step,

(b) w2, a measure of the variance explained by a discriminant function (Winkler & Hays, 1975, p. 676), and

(c) the significance level of each discriminant function

All three discriminant functions are highly significant. However, each one does not explain the same proportion of variance between the two groups. All the variables together explain 19.9% of the variance between the two groups (as indicated by the w2 at the bottom of Table 3). The nondemographic variables alone explain 17.7%, thus almost as much as all the variables. On the other hand, the demographic variables explain only 4.8%. Hence, the nondemographic variables explain much more of the variance than the demographic variables.

The higher a standardized coefficient, the more "important" the corresponding variable in a discriminant function. The results in Table 3 (first two columns) show that the importance ranking of the variables is quite similar whether all the variables are used, or the nondemographic variables only. Moreover, the ranking or the first four standardized coefficients is the same as the ranking of the t-values in Table 2, thus strengthening the results. Among the variables that are significant beyond the 1% level in Table 2, only one does not appear in Table 3 when all variables are included: purchased car from dealer before and was satisfied with it. This is because this variable is somewhat correlated with others, especially "knows someone at first dealership visited" (r= .26). The main results in Tables 2 and 3 can be summarized as follows: a consumer is more likely to purchase from the first dealer he visits mostly if:

(a) he knows somebody at the first dealer he visits, and/or he purchased from this dealer before and was satisfied with it,

(b) he feels he is able to make a wise purchase,

(c) he knows which manufacturer he will purchase from,

TABLE 3

DISCRIMINANT ANALYSIS RESULTS

(d) The size of his feasible set is small,

(e) he is older, less educated, and if there are fewer adults in his household.

Finally, confusion matrices were obtained to check the predictive validity of the discriminant functions. These matrices are shown in Table 4. They were obtained after splitting the data randomly into an estimation sample and a holdout sample, and predicting group membership in the holdout sample. The predictions based on the nondemographic variables alone are significantly and far superior to the predictions based on the demographic variables alone (70.3% vs. 61.7% hit rate), and almost as good (and not significantly inferior to) the predictions based on all variables (71.5% hit rate).

TABLE 4

CONFUSION MATRICES

CONCLUDING COMMENTS

The study confirmed that there is a substantial segment of car buyers who purchase from the first dealer they visit. Bettman's (1979) cost-benefit theory of information search was used to generate several testable hypotheses concerning the characteristics of this segment. For the most part, empirical results do support the cost-benefit model. Nevertheless, additional research is needed to further validate the theory. Education was significant in the wrong direction. The cost-benefit model does not seem sufficient to explain the relationship between this variable and single dealer buying. It may be that less educated consumers do not search for as much information as more educated consumers because these are less able to understand and use it in their purchase decisions. The number of adults in the household was also significant in the wrong direction. Perhaps, there is a dysfunctional effect and the cost of search increases with the number of adults.

The results of this study also have managerial implications. For instance, a manufacturers' and dealers' marketing mix can be better targeted at the segment of first dealer buyers. Salespersons can also use the results of this study since they can observe certain demographic and behavioral characteristics of a prospect the moment he walks into a showroom. Once communication commences the salesperson can determine (or infer) other characteristics, and adapt his sales presentation and the product-service package, depending upon whether the prospect is a Potential "single visit buyer" or not.

REFERENCES

Bennett, Peter D., and Mandell, Robert M. (1969), "Prepurchase Information Seeking Behavior of NewF Car Purchasers--The Learning Hypothesis," Journal of Marketing Research, 6, 430-33.

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

Blattberg, Robert, Buesing, Thomas, Peacock, Peter and Sen, Subrata (1978), "Identifying the Deal Prone Segment," Journal of Marketing Research, 15, 369-77.

Blattberg, Robert C., Peacock, Peter and Sen, Subrata R. (1976), "Purchasing Strategies Across Product Categories," Journal of Consumer Research, 3, 143-54.

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

Burnkrant, Robert E. (1976), "A Motivational Model of Information Processing Intensity," Journal of Consumer Research, 3, 21-30.

Claxton, John D., Fry, Joseph N., and Portis, Bernard (1974), "A Taxonomy of Prepurchase Information Gathering Patterns," Journal of Consumer Research, 1, 35-42.

Darden, W. R., and Reynolds, F. D. (1971), "Shopping Orientations and Product Usage Rates," Journal of Marketing Research, 8, 505-8.

Dommermuth, William P. (1965), "The Shopping Matrix and Marketing Strategy," Journal of Marketing Research, 2, 128-32.

Downs, A. (1961), "A Theory of Consumer Efficiency," Journal of Retailing, 39, 6-12.

Engel, James F., Blackwell, Robert D., and Kollat, David T. (1978), Consumer Behavior, Hinsdale, IL: The Dryden Press.

Granbois, Donald M. (1977), "Shopping Behavior and Preferences," in Selected Aspects of Consumer Behavior, Washington, D.C.,: U.S. Government Printing Office, pp. 259-98.

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

MacKay, David B., and Olshavsky, Richard W. (1975), "Cognitive Maps of Retail Locations: An Investigation of Some Basic Issues," Journal of Consumer Research, 2, 197-205.

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 Publishing Company, pp. 79-94.

Newman, Joseph W., and Lockeman, Bradley D. (1975), "Measuring Prepurchase Information Seeking," Journal of Consumer Research, 2, 216-22.

Newman, Joseph W., and Staelin, Richard (1972), "Prepurchase Information Seeking for New Cars and Major Household Appliances," Journal or Marketing Research, 9, 249-57.

Newsweek (1977a), "Buyers of New Domestic Cars: 1977," research report. New York. NY.

Newsweek (1977b), "Buyers of New Imported Cars: 1977," research report, New York, NY.

Newsweek (1980a), "Buyers of New Domestic Cars: 1980," research report. New York. NY.

Newsweek (1980b), "Buyers of New Imported Cars: 1980," research report, New York. NY.

Peter, J. Paul and Tarpen, Lawrence X. (1975), "A Comparative Analysis of Three Consumer Decision Strategies," Journal of Consumer Research, 2, 29-37

Pollay, Richard W. (1970b), "The Structure of Executive Decisions and Decision Times," Administrative Science Quarterly, 15, 459-71.

Stone, G. P. (1954), "City Shoppers and Urban Identification: Observation on the Social Psychology of City Life," American Journal of Sociology, 60, 36-45.

Tauber, E. M. (1972), "Why Do People Shop?" Journal of Marketing, 36, 46-49.

U.S. News & World Report (1974), "The Buyers of New 1974 Automobiles," Research Report, New York. NY.

U.S. News & World Report (1975), "1975 Study of American Markets: Automobiles," Research Report, New York, NY.

Williams, J. and Dardis, R. (1972), "Shopping Behavior for Soft Goods and Marketing Strategies," Journal of Retailing, 48, 32-41, 126.

Winkler, Robert L. and Hays, William L. (1975), Statistics: Probability, Inference, and Decision, New York, NY: Holt, Rinehart and Winston.

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