Fashion Preferences and Store Patronage: a Longitudinal Study

ABSTRACT - This paper presents a preliminary introduction to a survey research methodology for integrating some new techniques of fashion trend monitoring and retail store image/patronage monitoring into a retailing management information system. Selected preliminary findings from a two year longitudinal study of four separate random samples of consumers are used to illustrate the methodology.


George B. Sproles (1978) ,"Fashion Preferences and Store Patronage: a Longitudinal Study", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 675-681.

Advances in Consumer Research Volume 5, 1978      Pages 675-681


George B. Sproles, University of Houston

[Field administration of this study was funded by the Agricultural Experiment Station, Purdue University. Data analysis and computer programming was funded by the Department of Human Development and Consumer Sciences (pending), University of Houston. Mrs. Jacqueline McCullough was a doctoral researcher in one phase of this investigation.]


This paper presents a preliminary introduction to a survey research methodology for integrating some new techniques of fashion trend monitoring and retail store image/patronage monitoring into a retailing management information system. Selected preliminary findings from a two year longitudinal study of four separate random samples of consumers are used to illustrate the methodology.


A new survey methodology for measuring retail store images and associated dimensions of store patronage has recently received increased use in retailing research (i.e. Arnold and Tigert, 1973; King, Tigert and Ring, 1975).

This methodology for market monitoring has been found to have a substantial degree of reliability and validity in describing the competitive positions of retail stores within a local market trading area. More recently the method has also proved useful for multivariate modeling and "mapping" of the structure of local retail markets (i.e. Ring and King, in press).

One use of this new methodology which has not yet been investigated involves integrating the methodology to some newly evolving techniques for measuring changing trends in consumers' product preferences and actual buying behavior. However, this represents the next logical step in the development of a comprehensive retail market monitoring system. Retailers need information not only on their competitive positions across general store image/patronage dimensions, but also on specific competitive products offered within their trading area. This type of data will ultimately be a key part of future retailing management information systems used by retailers as an aid in developing their general positions ("images") in the market and complementary product assortments.

Given these needs, the purpose of this paper is to describe a longitudinal survey research program which has merged the evolving market monitoring methodology with evolving techniques for measuring current and predicted future growth in consumer demand for specific products. The product category involved in this research is women's clothing fashions, which is one of the most difficult and yet crucial categories of products in which to measure consumers' current and predicted future preferences for consumption. This paper will present a brief overview on the development of the research methodology for the investigation. Some preliminary illustrative data on use of the integrated system will also be discussed.


The research was conducted in 1976 and 1977 in Lafayette, Indiana. In each year two separate random samples, one of college women at Purdue University and one of adult women in the Lafayette community, were mailed a questionnaire surveying their store patronage and fashion preferences. In the next paragraphs the store patronage measures, fashion preference measures, and administration of the study will be described.

The Store Image/Patronage Measures

The new retail image and patronage research methodology has been described in earlier cited references, and requires only brief introduction here. Basically, the methodology measures consumers' images of and preferences for retail store in their trading area across a variety of relevant patronage dimensions (i.e. store with the best location, best advertising, best assortment, most fashionable selections, store most often shopped, etc.). King, Tigert and Ring (1975) have developed a specific sequence of sixteen questions for use in clothing stores, and their sequence is used in the research reported in this paper.

A group of ten directly competitive clothing stores was selected for inclusion in the survey instruments, including 3 department stores, 6 specialty stores, and 1 chain store. The 10 stores had previously been identified in a pretest as among the most extensively patronized clothing stores in Lafayette. Each subject was asked to pick the best store on each image/patronage dimension from the list of 10 stores. Additionally, respondents were given a write-in for "other store," to measure instances when none of the 10 stores was perceived as best on a specific dimension.

The Fashion Preference Measures

The methodology for measuring consumers' fashion preferences and predicted buying behavior is relatively novel, and will be described in greater detail. The basic method was developed as a modification of techniques previously used by Baumgarten (1975). His method involved presenting subjects a series of slightly varied basic styles of dress (i.e. styles of pants, shoes)., and asking each subject a series of consumption-related questions on each style. Those questions included number of items of each style currently owned, current popularity of each style, predicted future popularity of each style, and future buying intentions for each style. Selected measures were then statistically combined into measures of consumer innovativeness, and were used to identify fashion innovative consumers. However, it appears this type of methodology, with the appropriate modifications, can be used as a simple system of market monitoring and prediction of forthcoming fashion trends.

Since the technique for monitoring consumers' fashion preferences used in the present study represents a substantially different application of Baumgarten's approach, the development of this application will be described in some detail. The principal difference is that the technique is used to comparatively measure the current and changing fashionability of specific styles. Therefore, it is first necessary to validate the use of such a technique in discriminating between styles which are at different levels of current and predicted future fashionability. A second major difference is that simple descriptive statistics (frequencies) are used to differentiate the fashionability of one style versus another. Thus the interpretation of findings is relatively uncomplicated and responsive to managerial needs.

To initially validate the use of this methodology in measuring stages and levels of fashionability for specific styles, a series of pilot pretests was conducted in the late Winter-early Spring of 1976. [McCullough was a co-researcher and doctoral student who conducted the pretests. Also see McCullough (1977) for an analysis of fashion-conscious college students identified in the 1976 phase of this study.] In the first test, female college students were presented illustrations of over 150 styles of dress taken from various fashion periodicals, and were asked to classify each illustration into four categories. [The concept of categorizing styles in a similar manner has previously been used by Reed (1973) in a study of clothing as a symbolic indicator of the self-concept of college women.] The categories can be described as follows:

#1 -- The "newest" general fashion trend

#2 -- Out of fashion

#3 -- The "newest" college age/youth-oriented fashion

#4 -- "Classic" (long established) fashion

For 42 styles which were placed in a single category by 50 percent or more of the first test subjects, a second test was run. This test used a group of junior-senior female students majoring in fashion retailing, who were used as "expert judges" to validate the categorizations done by the first test group. Based on the results given by this "expert" group, four dress styles and four pant styles representing the four categories were selected for use in the survey.

The selected styles were redrawn to the same size and scale on a neutral posed form, to control for any differences in appearances which could be attributed to anything other than the style itself (i.e. a model's pose, hair style, figure, facial expression). The redrawn styles were then subjected to a final validating test with female students. This test successfully validated each of the four drawings of dress and four drawings of pant outfits as fitting in one of the four categories of fashionability, though this test did indicate that minor redrawing of a few details on two outfits was needed.

To construct the questionnaire for this part of the survey, the four dress styles were randomly arranged in the left column of the page. The right column contained the four question sequence on consumption and perceptions of popularity for each style. The pant styles were then arranged in an identical manner on a separate page of the questionnaire.

Questionnaires containing the fashion preference and store patronage measures, as well as a variety of other measures not relevant to the present study, were mailed to separate random samples of 400 female students at Purdue, and 400 adults in Lafayette. The mailings were purposefully administered in the early part of the Spring 1976 fashion season, to obtain an "early season tracking" of fashion preferences. Additionally, the mailing was made to follow immediately after the pilot validation of categories for the four styles, thus avoiding the possibility that an abrupt change of fashion would invalidate any of the four categories.

In the early Spring of 1977, questionnaires containing identical measures of fashion preferences and store patronage were mailed to random samples of 400 female students at Purdue, and 400 adults in Lafayette. Thus the survey administrations in 1976 and 1977 were directly comparable in terms of samples, questionnaire designs, and times of administration within the Spring fashion seasons of both years.

Other methodological details were as follows. In both the 1976 and 1977 surveys, the college student subjects included in the surveys were randomly selected from the Purdue registrar's list of full-time undergraduate female students. The adult subjects were randomly selected from the Lafayette telephone book. Representatives of the telephone company indicated that probably less than 5 percent of the Lafayette area households had unlisted phones or no phones; therefore, it would appear that the Lafayette telephone book was a relatively representative sampling frame for the purposes of this research. Finally, one methodological difference between the 1976 and 1977 studies should be noted. In 1976 the surveys were followed 10 days later by a reminder postcard which was used to stimulate increased response rates; however, the 1977 study involved a single mailing of a shortened questionnaire without use of follow-up procedures. This resulted in a difference in response rates in the two years of the study. The final response rates, in terms of returned usable questionnaires, was as follows:

1976 - College students 250 returns (62.5%)

1976 - Adults 174 returns (43.5%)

1977 - College students 198 returns (49.5%)

1977 - Adults 122 returns (30.5%)

From these results it would appear that the postcard follow-up yielded an increased response in the order of 10 percent or more. This was found to be the case in analyzing the returns in the 1976 study by date of return as compared to the postcard mailing date.


The following discussion focuses on a preliminary analysis of data on consumers' preferences for the four selected dress styles, and patterns of preferences across store patronage dimensions. These data illustrate the validity of this new methodology for measuring fashion preferences and identifying the predicted buying behavior of consumers frequently patronizing specific stores.

Tables 1 - 4 present descriptive data for the fashion preference measures, in terms of raw percentages of subjects in each sample who responded to each consumption-oriented question. The numbers #1 through #4 in these tables are keyed to the categories of fashionability described earlier in this paper. In general, the findings in these tables would significantly appear to validate the use of this monitoring technique in discriminating current levels of fashionability and perhaps in forecasting the forthcoming fashion trends. However, the latter point will require further research for more formal evidence on that point. Another point of interest is the divergence in preferences between the adult and college subjects. There are substantial differences in orientations toward fashion consumption of these two broadly defined market segments, and the surveys appear to discriminate these differences. This is especially notable in the measures of preference for style #3, which was categorized as a college or youth-oriented fashion.

Some of the more notable findings in Tables 1 - 4 are as follows. In the Table 1 data on ownership of the four styles, style #4 (the classic fashion) would appear to be the most widely owned style, particularly among adults. College students were significantly less likely than adults to report owning #2 (out of fashion). Finally, growth in ownership of #1 (the newest general fashion) can be seen from the 1976 to 1977 studies for both the adult and student samples. Thus these ownership measures give a distinct picture of consumers' preferences in the market (this of course becomes even more true upon visual inspection of the line drawings for each style).









Tables 2 and 3 show consumers' judgments of current and predicted future popularity of each style. In the 1976 data of Table 2, it is clear that both the adult and college samples rated the category of fashionability of each style in the same general categories as identified in the pilot tests. This represents a significant validation of the methodology for discriminating current fashionability of styles based on using basic line drawings of the styles. In the 1977 data, results are similar, though there is a clear decline in popularity ratings for #1 and #3 (which were highly popular in 1976). The Table 3 data showing predicted popularity for each style in the next six.months is also of considerable interest. Among adults the prediction is for #1 (the newest general fashion) and #4 (classic) to grow most significantly in popularity. In comparison, students clearly pick #1 and #3 (the youthful fashion) for the most significant growth. Finally, both adult and college groups indicate that #2 (out of fashion) is leveling out or declining in fashionability.

Table 4 presents data on consumers' purchasing intentions for each style. Although the preceding tables indicated many consumers would seem to have relatively well formed perceptions regarding popularity of each style, there is uncertainty as to what will actually be purchased. This is indicated by the high percentage of consumers responding they would "possibly" purchase each style in the near future. Only for style #2 (out of fashion) is the trend clear, and in this instance approximately 75 percent or more of consumers indicated they would not purchase the style. The styles most likely to be purchased by adults appear to be #1 and #4, while college students appear to be most positive about #1 and #3. It is also interesting to note the significant growth in students' purchase intentions for #4 (classic fashion) from the 1976 to 1977 study, as well as their apparent decline in intentions regarding #1 (newest general fashion in 1976).

Tables 5 and 6 exemplify how data from the fashion preference monitoring system described in Tables 1 - 4 can be integrated with data from retail store image/patronage measures. Tables 5 and 6 contain cross-tabulations of three stores which were identified by subjects as "most frequently patronized" versus the style preferences expressed by consumers choosing each store. The three stores selected for this example illustration include a well-known local department store (store A), a fashionable local specialty store (store B), and a national chain store (store C). These three stores were found to be among the most frequently patronized stores in the Lafayette area, and they collectively controlled a major share of the total clothing market. The three stores also represent different organizational structures, marketing strategies, and store images in the local market.

The percentages in Tables 5 and 6 represent the proportion of consumers choosing each store as most frequently patronized who rated each style as "high in current popularity," "will increase in popularity in next six months," and "definitely will purchase" this style. These percentages, when compared across styles, give an indication of the specific styles which would appear to be highest or growing in popularity, and which therefore should be emphasized in each store's inventory. The data also give an approximation of the proportion of inventory which might be committed to each style. The general finding of these tables is that consumers who have a "favorite" (most frequently shopped) store can have different fashion preferences compared to those choosing other stores. A brief description of these differences in preferences, and their implications for the buying and assortment policy of each store, is as follows:

1) Store A - Department Store. The 1976 data indicate this store has substantial appeal to both adult and college markets. In adult-oriented departments, the data clearly indicate that emphasis should be placed on styles #1 and #4, perhaps including a balanced inventory of those styles. In comparison, the data indicate that college-oriented departments should carry styles #1, #3 and #4, with some weighting toward #3. The 1977 data indicate a general decline in styles #1 and #3, and a considerable growth in importance of #4, for both the adult and college segments of the market.

2) Store B - Specialty Store. The 1976 and 1977 data indicate this store appeals largely to the college market segment, with approximately 3/4 of the regular patrons coming from that market segment. The 1976 data indicate the store's inventory should be heavily weighted toward style #1, though a modest amount of #3 and #4 should be carried in stock. It is apparent that the emphasis on #1 would appeal not only to the college market, but also to the few adults favoring this store. In the 1977 findings there is a substantial shift in consumers' preferences, and it would appear that a relatively balanced inventory including styles #1, #3 and #4 would be best, though a slight over-representation of #3 and #4 would be acceptable.

3) Store C - Chain Store. In the 1976 data there are indications that this store has a substantial competitive appeal in both adult and college markets, and the market share of this store would appear to compare favorably to stores A and B. The data also indicate that this store should carry the widest variety of styles in order to appeal to its clientele. There are some clear indications that the most emphasis should be placed on style #1 (the newest fashion), but even style #2 (the out of fashion style) appears to achieve enough popularity and buying intentions to justify some modest representation in the store's inventory. The 1977 data also appear to support this policy of variety and balance for the chain store. It would thus appear that this chain store has some of the most complex inventory decisions to make, given its widespread appeal to the consumer market and the variety of styles appealing to those market segments. Considerable coordination between local store management and the centralized management of the chain would clearly be needed.


This research has validated a mail survey technique for monitoring trends in consumers' fashion preferences. The technique involves presenting consumers with line drawings of basic styles, and obtaining consumers' rating of their ownership of each style, popularity (current and future) of each style, and future purchasing intentions. The data indicate that responses to these types of questions can be used to discriminate the current fashionability of each style. Predictions of future demand can also be extrapolated from the data, though the validity of those predictions has not yet been fully explored.





The fashion preference monitoring technique can also be used to identify styles preferred by consumers who patronize specific stores most frequently. This type of information, when available to the retailer with the proper lead time, can be extremely useful as a merchandise management information system. Specifically, the data can be used to identify basic styles to buy, proportions of inventory to commit in each style, market segments for each style, and styles nearing the end of their period of fashionability. These types of information are crucial for effective and timely inventory management.

The fashion preference monitoring technique can be especially useful for longitudinal tracking of trends. In the present research a two year tracking was conducted, with studies conducted on an annual basis. However, to more precisely track changing trends, it is likely that more frequent studies would be necessary to fully respond to the needs of retailers. For example, a monitoring system keyed to major annual fashion seasons (Spring and Fall) might be more responsive to retailer needs. Furthermore, within any given season it would be useful to obtain an initial measurement of trends at the start of the season (as done in the present study), and then conduct a follow-up study at mid-season in order to monitor the trend and validate the earlier findings. Such accumulations of data, when compared to the store's actual results, could be very useful in precisely validating the predictive efficacy of the monitoring system for any given store.

Finally, it is important to note that the data presented in this paper are only one example illustration of how fashion preference and store patronage monitoring systems can be integrated into a management information system. In this example the focus was on the store patronage dimension of "store most frequently shopped." Similar comparisons such as those made in Tables 5 and 6 could also be made across other relevant store image/patronage dimensions (i.e. store last shopped, most fashionable store with best assortment, store with best prices, store with the best location, and so on). In this manner a complete profile of consumers' fashion preferences and their view of the competitive market can be obtained. This type of competitive information can be crucial for retailers who wish to assess their position in the local market, and it can be absolutely vital to those retailers who need to reposition their images in order to maintain their share of the local market.


Stephen J. Arnold and Douglas J. Tigert, "Market Monitoring Through Attitude Research," Journal of Retailing, 49 (Winter, 1973-1974), 3-22.

Steven A. Baumgarten, "The Innovative Communicator in the Diffusion Process, Journal of Marketing Research, 12 (February, 1975), 12-18.

Charles W. King, Douglas J. Tigert and Lawrence J. Ring, "Contemporary Fashion Theory and Retail Shopping Behavior: A Segmentation Analysis," Paper presented at the Marketing Educators' Conference of the American Marketing Association, Rochester, New York, August 17-20, 1975.

Jacquelyn Harden McCullough, A Multivariate Profile of Fashion Conscious College Women, Unpublished Ph.D. Dissertation, Purdue University, 1977.

Julia Ann Reed, Clothing As A Symbolic Indicator of The Self, Unpublished Ph.D. Dissertation, Purdue University, 1973.

Lawrence J. Ring and Charles W. King, "A Multiple Discriminant Analysis Approach to the Development of Retail Store Positioning," Advances in Consumer Research, 5 (1977), in press.



George B. Sproles, University of Houston


NA - Advances in Consumer Research Volume 05 | 1978

Share Proceeding

Featured papers

See More


H12. Does Economic Development Influence Consumer Innovativeness?

Fuchun Zhan, University of Wisconsin - Madison, USA
Nancy Wong, University of Wisconsin - Madison, USA
Julie Anne Lee, University of Western Australia

Read More


Does It Pay to Be Virtuous? Examining Whether and Why Firms Benefit From Their CSR Initiatives

Dionne A Nickerson, Georgia Tech, USA
Michael Lowe, Georgia Tech, USA
Adithya Pattabhiramaiah, Georgia Tech, USA

Read More


When News Gets Personal: The Evolution of Content in the Successive Retelling of Events

Shiri Melumad, University of Pennsylvania, USA
Yoon Duk Kim, University of Pennsylvania, USA
Robert Meyer, University of Pennsylvania, USA
Ani Nenkova, University of Pennsylvania, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.