An Exploratory Investigation of Mediating Factors in Retail Store Image Responses

Robert A. Peterson, University of Texas at Austin
ABSTRACT - This paper proposes a general paradigm for investigating retail store images which takes into account the context of the research. An example illustrating several of the concepts set forth is provided.
[ to cite ]:
Robert A. Peterson (1981) ,"An Exploratory Investigation of Mediating Factors in Retail Store Image Responses", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 662-664.

Advances in Consumer Research Volume 8, 1981      Pages 662-664

AN EXPLORATORY INVESTIGATION OF MEDIATING FACTORS IN RETAIL STORE IMAGE RESPONSES

Robert A. Peterson, University of Texas at Austin

ABSTRACT -

This paper proposes a general paradigm for investigating retail store images which takes into account the context of the research. An example illustrating several of the concepts set forth is provided.

INTRODUCTION

A considerable amount of research in the area of retail patronage has been concerned with the concept of store image. For example, research has bean conducted regarding the role of store image in determining store loyalty and choice (e.g., Anderson and Scott 1970; Lessig 1973; Schiffman, Dash and Dillon 1977). Likewise, research has been conducted to investigate possible relationships between consumer self-images and retail store images (e.g., Bellenger, Steinberg and Stances 1976; Dornoff and Tatham 1972), and differences between consumers' images and retailers' images (e.g. Pathak, Crissy and Sweitzer 1974-75). Indeed, an entire issue of the Journal of Retailing was devoted to store image research. [Journal of Retailing, 50, 4 (Winter 1974-75).] Finally, the importance of store image research as a managerial strategy was recently documented by King, Tigert and Ping (1979).

Moreover, because of its importance, a substantial amount of research has been dedicated to improving the measurement of retail store image. Empirical studies focusing on both validity and reliability issues have been reported in the literature (e.g. Hawkins, Best and Albaum 1975-76; McDougall and Pry 1974-75; Menezes and Elbert 1979), and several contributions made. In addition, progress has recently been made in determining the manner in which store images are formed (Reich, Ferguson and Weinberger 1977).

Purpose

The purpose of this paper is not to review and/or summarize the (ponderous) research on retail store image. Nor is it to offer new findings on store image determinants or methodological approaches to measuring store image. Rather, the purpose is more modest. It is to propose and illustrate a paradigm for more fully evaluating and interpreting retail store image data. While this paradigm can be applied in many research situations, it is particularly appropriate in store image research since there is a paucity of both physical and conceptual criteria against which to evaluate research conclusions.

Because of the desire to stimulate discussion as well as thought, no attempt is made to present an exhaustive documentation of appropriate literature. That task will be left to a later, more systematic exposition of the insights offered here. It is assumed that researchers interested in store images will already be thoroughly familiar with the appropriate literature.

A PARADIGM

Investigating consumer images of retail stores is a relatively straightforward and standardized research activity. In the typical store image study a sample of consumers is selected, questionnaires administered, resulting data analyzed, and inferences drawn about image similarities and differences among the stores investigated. Few researchers, however, are naive enough to believe that image differences observed to exist among stores---as reflected by one or more scale or question responses---can be interpreted at face value. Instead, differences must be considered within the context in which the data were collected.

In particular, image differences observed between retail stores may not arise solely as a result of perceived differences between the stores (the stimulus objects). Although the majority of the variance in consumer image responses may be due to differences between stores, other, frequently confounding, factors may account for some of the variance as well. Conceptually, responses to an image question can be thought of as reflecting a plethora of factors and influences, only one of which is stimulus object-related. Symbolically

R = f(SO, SC, MI, M, EN, E)   (1)

where

R represents an image response (in general or in particular);

SO represents stimulus object characteristics;

SC represents subject (consumer) characteristics;

MI represents measurement instrument characteristics;

M represents mode of data collection;

EN represents the data collection environment; and

E represents extraneous or error factors.

Subject characteristics consist of a wide variety of factors and influences. Many of them are relatively permanent and general in nature, such as demographic or personality characteristics. Others, like mood, are more transitory, while still others are specific to the research situation (e.g., response syndromes like haloing or yea-saying).

Three categories of factors are methodological in nature. Measurement instrument characteristics include type of question (open or closed end), nature of questionnaire, and so forth. Data collection mode refers to whether data are collected by means of telephone, mail or personal interviews, and whether questionnaires are self- or other-administered. Environmental factors include physical, social, and temporal (time of day, etc.) influences. The list of factors is nearly endless.

Factors within each category as well as the entire categories themselves interact and collectively result in an image response. Many of the nonstimulus object factors can be considered mediating factors in that they intervene between a stimulus and a response. Such factors can confound, bias, accentuate, or otherwise adversely influence an image response. In the extreme case an image response may actually be more an artifact of the mediating factors than be reflective of stimulus object characteristics.

In general, the above paradigm can be viewed as merely an application of concepts derived from SOR learning theory and environmental and gestalt psychology. Thus, underlying assumptions and axioms are well described elsewhere and need not be reiterated here. The remainder of this paper consists of an example illustrating, in a nonabstract manner, some of the notions expressed in the paradigm, Although the research is exploratory, it is hoped that insights will be obtained into, and an appreciation gained of, factors in store image research.

ILLUSTRATION

Method

To collect requisite study data a 2 x 2x 2 experimental design was employed. Two classes of undergraduate marketing students, 347 individuals in total, served as subjects.

Subjects were requested to complete a three-page questionnaire consisting of retail store image questions, shopping behavior questions, and several self-perception and demographic questions. The dependent variables investigated were 10 common retail store image dimensions. These dimensions were presented in the form of 7-point rating scales and addressed such characteristics as prices, merchandise quality, sales personnel, and physical layout. The specific scale content is irrelevant for the present discussion. However, approximately half of the subjects were administered the scales in a bipolar (semantic differential) format while the other half were administered the scales in a unipolar (stapel scale) format. This treatment, labeled V1, was designed to serve as a measurement instrument (MI) characteristic. In addition, half of the subjects evaluated Foley's Department Store, a large component of the Federated group, while the other half evaluated Penney's. This treatment was labeled V2; it served as a measure of stimulus object characteristics.

The third treatment vas environment. About two-thirds of the subjects completed their questionnaires in a classroom with controlled conditions at one point in time. The remaining third were allowed to complete their questionnaires at their leisure, outside of a classroom environment, and return them at their convenience. [The low return rate of the latter group accounted for the size differences in the treatment levels.] This treatment was labeled V3.

Four other variables in the subject characteristic category were also investigated. These variables were, respectively,

---sex (V4);

---shopping frequency at ___________ department store (V5);

---general interest in department stores (V6); and

---mood at time of questionnaire completion (V7).

Shopping frequency and interest were measured using k-point rating scales. Mood was measured by means of a 6-item Likert-type summated rating scale. Scale scores ranged from 5 (good mood) to 30 (bad mood), with a mean of 16.2, and standard deviation of 3.5. Because of its temporary nature mood was deemed a "nuisance" factor.

Analysis

Initially relationships between the set of 10 dependent variables and the set of 7 independent variables were investigated through canonical analysis. This resulted in 3 canonical roots (functions) being statistically significant at the .07 level or greater. Canonical loadings for the 7 independent variables are given in Table 1. The first root was by far the most significant, with p < .0001; it accounted for 71 percent of the explainable variance. Together, the next two roots only accounted for 19 percent of the explainable variance.

Function I is clearly dominated by V2; 76 percent of its communality is accounted for by stimulus object characteristics. Function II is essentially an interest-frequency function since these variables load most heavily. The third function can be interpreted as reflecting the influence of mediating factors since all independent variables but stimulus object characteristics possessed significant canonical loadings. [Terming a factor mediating or determinant is essentially a function of analytical purpose.] With regard to the variance explained by each independent variable, stimulus object characteristics accounted for the largest percentage of explained variance in the dependent variable set while mood accounted for the smallest percentage of explained variance.

TABLE 1

INDEPENDENT VARIABLE CANONICAL LOADINGS

To obtain greater insights into specific variable interactions an analysis of variance was conducted for each of the 10 dependent variables. However, to make each analysis more manageable, only four independent variables were retained for investigation:

---V2 (score)

---V4 (sex)

---V5 (frequency)

---V6 (interest).

Using a significance criterion of .05, only 27 of 150 possible effects were statistically significant. Of these 27 effects, 22 were main effects, broken down as follows:

Variable   Number of Main Effect Significant

V2 (store)                 7

V4 (sex)                   8

V5 (frequency)        6

V6 (interest)            1

There was no consistent pattern to the 5 interaction effects that were statistically significant. Moreover, the substantive significance of any effect was marginal at best.

CONCLUSION

This paper has suggested a paradigm for empirically investigating retail store images. By means of this paradigm a researcher can evaluate store image differences in a holistic sense, simultaneously taking into account perceived differences between stimulus objects and mediating (intervening) factors. This would allow, in turn, quantitative evaluation of the various factors influencing image perceptions apart from store perceptions per se.

While the results of the example are not extremely significant, they are sufficiently strong to both illustrate the concepts espoused and warrant further research. Other possible influencing factors need to be empirically explored and categorized as to their importance. Ultimately, a typology of influencing factors needs to be constructed so that store image data can be better evaluated with regard to the manner in which they are collected. The benefits of so doing are both widespread and obvious.

REFERENCES

Andersen, Clifton R. and Scott, Richard A. (1970), "Supermarkets: Are They Really Alike?" Journal of Retailing, 46, 16-24.

Bellenger, Danny N., Steinberg, Earle and Stanton, Wilbur W. (1976), "The Congruence of Store Image and Self-Image," Journal of Retailing, 52, 17-32.

Dornoff, Ronald J. and Tatham, Ronald L. (1972), "Congruence Between Personal Image and Store Image," Journal of the Market Research Society, 14, 45-52.

Hawkins, Del I., Best, Roger and Albaum, Gerald (1975-76), "Reliability of Retail Store Images as Measured by the Stapel Scale," Journal of Retailing, 52, 31-38.

King, Charles W., Tigert, Douglas J. and Ring, Lawrence J. (1979), "Pragmatic Applications of Consumer Research in Retailing," in William Wilkie (ed.), Advances in Consumer Research, Vol. VI, Ann Arbor: Association for Consumer Research, 20-26.

Lessig, V. Parker (1973), "Consumer Store Images and Store Loyalties," Journal of Marketing, 37, 73-74.

McDougall, G. H. G. and Fry, J. N. (1974-75), "Combining Two Methods of Image Measurement," Journal of Retailing, 50, 53-61.

Menezes, Dennis and Elbert, Norbert F. (1979), "Alternative Semantic Scaling Formats for Measuring Store Image: An Evaluation," Journal of Marketing Research, 16, 80-87.

Pathak, Dev S., Crissy, William, J. E. and Sweitzer, Robert W. (1974-75), "Customer Image Versus the Retailer's Anticipated Image," Journal of Retailing, 50, 21-28, 116.

Reich, John W., Ferguson, Jeffery M. and Weinberger, Marc G. (1977), "An Information Integration Analysis of Retail Store Image," Journal of Applied Psychology, 62, 609-614.

Schiffman, Leon G., Dash, Joseph F. and Dillon, William R. (1977), "The Contribution of Store-Image Characteristics to Store-Type Choice," Journal of Retailing, 53, 3-14.

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