A Comparative Analysis of Determinant Attributes in Retail Store Selection

ABSTRACT - Results from twelve studies provided the opportunity to identify determinant attributes in retail store selection. Generalizations of the findings appeared possible in light of the variation among the studies in terms of retail environment, market differentiation, and national boundaries. Location and price appeared determinant in food stores selection while value for the money, assortment, and quality were determinant in the selection of fashion clothing stores. The results suggest implications for both super and box store strategies.

Citation:

Stephen J. Arnold, Sylvia Ma, and Douglas J. Tigert (1978) ,"A Comparative Analysis of Determinant Attributes in Retail Store Selection", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 663-667.

Advances in Consumer Research Volume 5, 1978      Pages 663-667

A COMPARATIVE ANALYSIS OF DETERMINANT ATTRIBUTES IN RETAIL STORE SELECTION

Stephen J. Arnold, Queen's University

Sylvia Ma (Student), University of Toronto

Douglas J. Tigert, University of Toronto

ABSTRACT -

Results from twelve studies provided the opportunity to identify determinant attributes in retail store selection. Generalizations of the findings appeared possible in light of the variation among the studies in terms of retail environment, market differentiation, and national boundaries. Location and price appeared determinant in food stores selection while value for the money, assortment, and quality were determinant in the selection of fashion clothing stores. The results suggest implications for both super and box store strategies.

INTRODUCTION

Review studies such as that of Lindquist (1974-75) attest to the continued search for determinant attributes in the retail store selection process. Is price the major determinant or is it shopping convenience? Or is determinancy dependent upon some combination of other attributes such as assortment, service and quality?

The availability of results from twelve studies provided the opportunity to address these questions. A review of these results is first preceded by a discussion of the concept of determinancy. The results are further placed in context by a brief review of comparable studies. Additional questions are raised and discussed.

Attribute Determinance

The notion of attribute determinance was discussed by Myers and Alpert (1968) and Alpert (1971). In order for an attribute to be determinant in the choice process, or, in other words, predispose consumers to action, the attribute must exhibit two characteristics. First, the attribute must be regarded as important. By important, it is meant that the consumer is "extremely offended" (Myers and Alpert, 1968, citing Foote, 1961) by the attribute's absence. It also appears that "important" is related to the multi-attribute attitude notion (e.g. Fishbein, 1972) that the attribute is "highly evaluated" or that its presence provides the consumers "much satisfaction.''

The second characteristic exhibited by a determinant attribute is that alternatives in the choice process are perceived as "differentiated" with respect to the presence of the important attribute. In the multiattribute attitude context, consumers state "different beliefs" or provide "different likelihoods" about the presence or absence of the attribute in each of the alternatives.

According to the preceding, determinant attributes would appear to be identifiable by either of two general approaches. Directly, they can be determined by relating the degree of presence of the attribute to the results of the choice process (Myers and Alpert's, 1968 "covariate analysis"). Less directly, they can be determined by identification of those attributes that consumers say are important, as well as identification of those attributes that consumers say differentiate the alternatives (Myers and Alpert's, 1968 "dual questioning"). Either approach, however, has limitations related to problems of causality and consumer vocalization of "real" motivations and attitudes (Myers and Alpert, 1968). Following Alpert's (1971) conclusion that regression coefficient determination (covariate analysis) and direct dual questioning are the best methods, it is concluded here that combinations of both general approaches (direct and indirect) are necessary in the search for determinant attributes.

"Important" Attributes in Retail Store Selection

A review by Lindquist (1974-75) of 26 empirical and theoretical studies of retail store selection led him to observe that the following attributes were mentioned in at least 25% of the studies:

TABLE

While he provided a caveat, Lindquist suggested that this relative frequency of mention is a "valuable indicator" of the important attributes.

In a study not reviewed by Lindquist, Jolson and Spath (1973) found that price/value relationship, store specialization, quality of merchandise, salesclerk service, and location were the factors considered most important in the selection of eight stores at a local shopping center.

How valid it is to aggregate different retail environments (e.g. food, department, clothing, etc.) in order to identify important attributes is not clear. Presumably, each type of retail store exists to satisfy different consumer needs. Hence, the "evoked set" of attributes from which arise the "determinant" attributes would be different in each situation. Conversely, if all retail environments are characterized by the same evoked set, it would suggest that consumers would be indifferent in their pursuit of need satisfaction as to whether they went to a food, department or clothing store. Casual observation suggests that this indifference is not the case and it is thus concluded that each retail environment must be examined separately.

A separate analysis of department and grocery stores characterized a recent study by Hansen and Deutscher (forthcoming) of important attributes. In this investigation, it was found that the three top-ranked dimensions in grocery store selection were physical facilities (clean, easy to move through, easy to find items, fast checkout), store atmosphere (friendly store personnel), and merchandise (dependable products, high quality products, high value for the money, wide selection, fully stocked, numerous brands, well-known brands, low prices vs. competition, many specially-priced items).

In another recent study of men's clothing stores by James, Durand and Dreves (1976), it was found that price, assortment and personnel were the attributes deemed most important by respondents.

"Determinant" Attributes in Retail Store Selection

While replication of the preceding studies in similar retail environments might identify consistently-rated "important" attributes, it is not clear that the same subset of attributes would be consistently identified as "determinant." To the extent that markets differ in the extent of differentiation on each of the important attributes, it would be expected that studies carried out in each market would also differ in their reports of determinant attributes. For example, a Toronto, Canada study of retail food stores (Arnold and Tigert, 1973-74) reported that Dominion Store increased its market share from 30% to 45% when it suddenly created a wide gap in the pricing structure by dropping its prices by about 12%. In other words, price appeared determinant among a large group of shoppers.

A year later, there was evidence that the key determinant attribute had changed. With the price war over, "best customer service" was found to be the variable most closely correlated with whether or not the respondent last shopped at Dominion. However, "lower price" still remained most highly correlated with Miracle Food Mart shopping behavior while "easiest to get to" characterized Loblaws shoppers. That is, even in the same geographic market, determinancy appeared to change over time.

With changes even in the same market, it might be expected that determinant attributes would again be different when national boundaries are crossed and the search for determinant attributes is carried out in different cultures. Thus, Doyle and Fenwick (1974-75) found in a London, U.K. study of food stores that "most individuals preferred to move up the vertical or 'quality' dimension toward Sainsbury's and Marks and Spencer" (as opposed to the "greater variety" or "lower prices" dimensions in that study). Quality is a different determinant attribute than the determinant service, location, and price attributes found in the Toronto study.

It must be further noted that different measurement methodologies characterized the latter two studies. In the Arnold and Tigert (1973-74) study, respondents selected the store best on an attribute. That selection was then correlated against store last shopped. In the Doyle and Fenwick (1974-75) study, however, stores were both ranked by preference and rated on semantic differential scales. These data were then subjected to multidimensional scaling. Thus, different methodologies may also account for the different results in these two studies. It is consequently recognized that this factor may tend to confound the search for consistently rated important and determinant attributes.

In view of the preceding considerations, an analysis was made of results obtained in 12 different commercial studies. The usefulness of these studies arose from the fact that although there were differences in methodologies, national boundaries, and retail environments, there were enough replications of each condition to permit tentative conclusions about important and determinant attributes.

RESEARCH METHODOLOGY

Attribute importance and determinancy in retail store selection were investigated in twelve separate studies characterized by: i) three different countries (U.S., Canada, Netherlands); (ii) three different methodologies (open-ended questions, Likert type statements, forced-choice scales); iii) three different interviewing techniques (telephone, in-home interview, in-home self-administered questionnaires); and iv) two different retail environments (supermarkets, fashion). All studies were completed in 1976-77 and all involved random samples of between 1,000 and 3,000 female household heads drawn from Census Metropolitan Trading Areas. All studies involved a measurement of not only attribute importance but also attribute scores of retail outlets, shopping behaviour across outlets, and external validity data on actual prices, market share, performance/productivity, etc. One of the best comparative performance measures across stores and across markets is sales per square foot of gross leasable space (GLA) or sales per square foot of selling area, on either a weekly or annual basis. All ten studies reported here used that measurement, as well as others.

Retail Food

Seven of the twelve studies were surveys of retail food store patronage. In each of these seven studies, the attribute importance question was asked in an identical fashion using the open-ended question, "What is the single most important reason you shop at (________) for most of your food shopping?" Some 35 specifically recorded verbatim responses were post-coded into eight major components. Six of the seven studies were completed by telephone while the seventh utilized an in-home interview at the same time the telephone interview was being completed in the same market (i.e. the sixth study).

Clothing Fashions

Five separate measures were made of attribute importance in the selection of stores for women's fashions. Two of the studies involved the use of five point importance statements while the remaining three utilized an i1 point forced choice format. Unlike the supermarket studies, the fashion studies involved preselected store attributes that had been determined on the basis of a large number of focussed group interviews probing on the importance of various store characteristics. The derived list of 11 attributes included a number of economy/utility dimensions (price, value, location) as well as a number of fashion specific store characteristics. Three of these dimensions relate to how a store is positioned on the "fashion spectrum" from low (everyday, conservative wear) to high (latest, most fashionable women's wear). Respondents completed the questionnaires in their homes on a self-administered basis.

RESULTS

Retail Food

Table I reports on the seven separate studies of supermarket shopping behavior. While the commercial nature of the data dictates disguise of each city's identity, it is nonetheless clear that location and price ranked either first or second in importance among the eight major attribute groupings.

With regards to determinancy, it was observed that in all seven studies, the chain with the highest market performance, as measured by sales per square foot of GLA, was also the market price leader, both in terms of consumer perceptions and in terms of an actual market basket measurement, or was within two percentage points of the price leader in the shopping basket measure. Price was thus concluded to be determinant.

There are several differences in the columns of Table 1 that relate primarily to differences in competitive activity and hence market differentiation but which reinforce the above conclusions about importance and determinacy.

TABLE 1

COMPARATIVE ATTRIBUTE IMPORTANCE DATA FOR SUPERMARKETS: SHARE OF MENTIONS GOING TO EACH RESPONSE TO THE QUESTION..."ALL THINGS CONSIDERED, WHAT IS THE SINGLE MOST IMPORTANT REASON YOU SHOP AT (_____ ) FOR MOST OF YOUR FOOD SHOPPING"

The first four cities (A, B, C, D) all show location/ convenience to rank first by a wide margin with low price in second position but only slightly ahead of attributes like assortment, meat, quality and service. In these four markets, the price differential across the four leading chains in market share is less than four percentage points in the price basket. Thus, while there is limited but noticeable price differentiation, price clearly plays a lesser role as a determinant of patronage and the non-price variables play a more prominent role, particularly location.

In North American city "E" and in the two Netherlands studies, the low price attribute jumps into first place and by a wide margin over location. In these markets, the price leader is 12 to 15 percent below the next nearest competitor. Furthermore, the price leaders are also the market share leaders in these markets and have by far the highest performance records in sales per square foot. Clearly, the competitive structure of the market has forced price into the leading role as a determinant store attribute. In addition, but not shown in this report, low price is the single most important determinant of patronage for the market leaders in these markets. Finally, on store attributes like cleanliness, friendly personnel, etc., that received high ratings in the Hansen and Deutscher study, the leading chains in city "E" and the Netherlands cities have the lowest ratings in the market. They have traded away those dimensions, as well as quality on such items as meat and produce, in favor of total market dominance on price and assortment. Their bottom line profitability indicates their strategy is working well.

In the two Netherlands cities, meat quality/variety received no mentions on the attribute importance question. That result should hardly be surprising given that upwards of 80 percent of all Dutch consumers buy their meat at a specialty butcher shop. Pleasant shopping environment, however, ranks fourth in the Netherlands while receiving almost no mentions in the North American studies.

Clothing Fashions

The results of the comparative investigation of important attributes in the selection of women's fashion clothing stores are found in Table 2. Value for the money, assortment/selection, and quality dominate the first three rank positions. Rank correlations above 0.99 for all 10 pairs of columns indicate that these results are highly consistent across the five studies.

Across national boundaries and assuming equivalent translation of the questions, it was found that there were significant differences in the mean score levels. For example, in the two forced choice scales in Canada in 1976 and 1977, "best value for the money" totally dominated the data with mean scores of 2.5 and 2.6 in the two years. The gap between this dimension in first place and the second ranked dimension, largest overall assortment, is enormous, about 20 standard errors. In the Netherlands, the top three store dimensions received almost identical mean scores on the importance scale and "highest quality" was in first place on the forced choice. In fact, highest quality ranked first or second in both Dutch studies.

TABLE 2

COMPARATIVE ATTRIBUTE IMPORTANCE DATA FOR WOMEN'S CLOTHING FASHIONS: MEAN AND RANK SCORES ON STORE CHARACTERISTICS FOR LIKERT SCALES AND FORCED CHOICE SCALES..."HOW IMPORTANT IS EACH SPECIFIC STORE CHARACTERISTIC TO YOU IN CHOOSING A STORE TO SHOP FOR WOMEN'S FASHIONS?"

An interesting trend was also observed in the two year period separating the two Canadian studies. "Best for current, up-to-date women's fashions" moved up from fifth to third position while "knowledgeable, helpful salesclerks" dropped from fourth to sixth position. While such a change could be attributed to sampling error, focussed group research suggested that salesclerk service was becoming universally poor across major fashion outlets with the result that the consumers perceived they would have to service themselves in the stores. Thus, importance of good in-store service was consequently declining as an important store attribute.

DISCUSSION AND IMPLICATIONS

What conclusions can be reached on the basis of these results and in the context of other studies devoted to the identification of important and determinant attributes? First, price/value for the money is an important attribute and will be determinant, it appears, with even small amounts of differentiation across competitors.

This finding, characterizing both the retail food and fashion clothing environments, clearly matches the results of Jolson and Spath (1973) and James et-al (1976) and closely parallels those of Lindquist (1974-75) and Arnold and Tigert (1973-74). Only in the results of Hansen and Deutscher (forthcoming) does the effect of price appear absent. It is hypothesized that the latter result was obtained because there was little pricing differentiation in the Columbus, Ohio market at the time of their study and that respondents to the questionnaire were, in fact, identifying the determinant as opposed to the important attributes as suggested. Thus, overall, the economic model of the consumer is suggested as having strong validity in the retail store selection process.

Attributes either equally important or secondary to price appear to differ depending upon the environment. Among retail food stores, according to these studies, it is locational convenience. Among fashion clothing stores, however, it is assortment and quality.

In other words, there is considerable evidence that whenever there is noticeable dispersion across supermarket chains on the economy/utility dimensions (price, location), those store characteristics dominate the store choice process. Fashion shoppers, on the other hand are not as concerned about physical convenience. They search out value, quality, wide assortments, and up-to-date fashions. They appear to be willing to travel to whatever location is required to find what they want.

This second conclusion supports Arnold and Tigert (1973-74) in the retail food environment and James et al (1976), Lindquist (1974-75), and Jolson and Spath (l973) to the extent that their findings apply to fashion clothing stores. These two conclusions appear to hold implications for both retail food superstores and box stores. The supermarket chains that are now moving into the "superstore" game with 35-100,000 square foot outlets are implicitly trading on the assortment attribute which was not concluded to be as important as price and location. Thus, in order to be successful, superstores will at least have to match the competition on price. But the very nature of the large capital investment in physical plant, however, implies fewer stores and therefore less convenience. Their only apparent option in order to satisfy the important convenience attribute would be, as some have done, to add specialty bakeries, delicatessens, floral shop and general merchandise departments, which would then provide one-stop shopping for the consumer and thereby justify the greater distance they would have to travel on average. Superstores would then be perceived as satisfying the convenience criterion.

At the other end of the spectrum are the box stores such as Aldi in Iowa which are attempting to take a share of the market on the basis of only limited assortments of basic groceries (400 stock keeping units). Furthermore, prices are at least ten percent below the next nearest competitor.

While the box store strategy has worked well in Germany, the evidence is not yet available for North America. However, the results of this study appear to support the strategy. Pricing well below the competition automatically makes the important price attribute also determinant. In addition, low capital investment per outlet should also make them more amenable to greater numbers of units and hence more likely to satisfy the convenience attribute.

If the future brings a showdown between the superstore and the box store, the superstore may win. It is conceivable that over time, the cost advantages that accrue to the box store will be wiped out due to advances in technology. For example, no price stamping of merchandise in the box store could be countered by optical scanning of label codes. Thus, if price differentiation is erased and consumers perceive the fewer superstores just as convenient, the superstores will then completely dominate on assortment, the other important and consequently determinant attribute in retail food store selection.

REFERENCES

Mark I. Alpert, "Identification of Determinant Attributes: A Comparison of Methods," Journal of Marketing Research, 8(May, 1971), 184-91.

Stephen J. Arnold and Douglas J. Tigert, "Market Monitoring Through Attitude Research," Journal of Retailing, 49(Winter, 1973-74), 3-22.

Peter Doyle and Ian Fenwick, "How Store Image Affects Shopping Habits in Grocery Stores," Journal of Retailing, 50(Winter, 1974-75), 39-52.

Martin Fishbein, "The Search for Attitudinal-Behavioral Consistency," in Joel B. Cohen (ed.), Behavioral Science Foundations of Consumer Behavior (New York: Free Press, 1972).

Nelson N. Foote, Consumer Behavior: Household Decision-Making, Vol. 4 (New York: New York University Press, 1961).

Robert A. Hansen and Terry Deutscher, "An Empirical Investigation of Attribute Importance in Retail Store Selection," Journal of Retailing, forthcoming.

Don L. James, Richard M. Durand and Robert A. Dreves, "The Use of Multi-Attribute Attitude Models in a Store Image Study," Journal of Retailing, 52(Summer, 1976), 23-32.

M. A. Jolson and Walter F. Spath, "Understanding and Fulfilling Shopper's Requirement: An Anomaly in Retailing?'' Journal of Retailing, 49(Summer, 1973), 39-49.

Jay D. Lindquist, "Meaning of Image: A Survey of Empirical Hypothetical Evidence," Journal of Retailing, 50(Winter, 1974-75), 29-38.

James H. Myers and Mark I. Alpert, "Determinant Buying Attitudes: Meaning and Measurement," Journal of Marketing, 32(October, 1968), 13-20.

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