An Exploratory Study of the Influences of Usage Situation on Perceptions of Product-Markets

Rajendra K. Srivastava, (student), University of Pittsburgh
Allan D. Shocker, University of Pittsburgh
George S. Day, University of Toronto
ABSTRACT - Recent research has illustrated the importance of environmental influences in determining consumer behavior. A typology of relevant "objective" usage-situations is developed, and subsequently, the composition of competitive product markets determined based on the similarity of the usage patterns. Results appear to show the feasibility of creating meaningful product specific situational taxonomies.
[ to cite ]:
Rajendra K. Srivastava, Allan D. Shocker, and George S. Day (1978) ,"An Exploratory Study of the Influences of Usage Situation on Perceptions of Product-Markets", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 32-38.

Advances in Consumer Research Volume 5, 1978      Pages 32-38

AN EXPLORATORY STUDY OF THE INFLUENCES OF USAGE SITUATION ON PERCEPTIONS OF PRODUCT-MARKETS

Rajendra K. Srivastava (student), University of Pittsburgh

Allan D. Shocker, University of Pittsburgh

George S. Day, University of Toronto

ABSTRACT -

Recent research has illustrated the importance of environmental influences in determining consumer behavior. A typology of relevant "objective" usage-situations is developed, and subsequently, the composition of competitive product markets determined based on the similarity of the usage patterns. Results appear to show the feasibility of creating meaningful product specific situational taxonomies.

INTRODUCTION

Consumer research in marketing, heavily influenced by a psychological tradition, has typically followed a paradigm where consumer choice in the market has been examined largely as a function of personal characteristics and preferences. In recent years, both the psychological tradition (e.g., Fredriksen (1972), Wicker (1972), Bowers (1973), and Endler and Magnusson (1976)) and consumer behavior research, (Belk (1975), Lutz and Kakkar (1975), Bearden and Woodside (1976), and Hansen (1976)) have examined the evidence for situational effects moderating consumer behavior and have proposed conceptual models of these effects. This work has generally supported the hypothesis of important situational effects but raised problems of measurement and classification.

Theoretical development in the area of situational influence on consumer behavior is still in an early stage. Although there appears little doubt that environmental factors affect behavior (see reviews by Lutz and Kakkar (1975), Belk (1975, and Endler and Magnusson (1976)), there exist differing viewpoints as to the line of inquiry which would best illustrate the interaction between individual characteristics and situations. Perhaps one of the more limiting drawbacks at the current stage of development is the lack of consensus on a classification schema or typology for situations within which consumer behavior may be systematically studied (Wicker (1972), Fredriksen (1972), Belk (1975), Russell and Mehrabian (1976)). Without such a comprehensive typology, systematic analysis of the influence of situations on choice behavior cannot be achieved. The two primary approaches to the development of such typologies have not been totally satisfactory. The first has sought to analyze objective situations (those meaningful to respondents) to find situational factors which appear to affect behavior (e.g., Belk (1975) proposed five groups of objective characteristics-- physical surroundings, social surroundings, temporal perspective, task definition, and antecedent states). The second has sought to classify situations by the nature of the psychological processes to which they give rise (e.g., Lutz and Kakkar (1975) attempted to classify situations in terms of Mehrabian and Russell's dimensions of pleasure, arousal, and dominance). Allen (1965), Barker (1975) and Wicker (i975), among others, have discussed variants of these approaches.

Typologies which retain an objective character have the advantage of greater meaningfulness (i.e., they provide some usable explanations for situational effects on behavior). Its major disadvantage is that the potentially large number of dimensions may lead to a correspondingly large number of distinct situational types. The psychological situation in turn offers parsimony at the possible expense of relevance. However, for most managerial marketing purposes, one might expect that the psychological situation would prove less usable (Lutz, personal communication).

The focus of this paper is on the usage or application situation; that is, the objective circumstances for which the product or service is purchased. Specifically, the objective is to develop a situational typology that can account for a comprehensive array of usage situations. In other words, we are seeking a method for extracting the minimum number of underlying dimensions which account for the variability in the effects of a large number of usage situations. Not only is such a typology useful in clarifying the general concept of usage situation, it can also be used to suggest additional situations to incorporate in the analysis.

In discussing alternative criteria for the development of a classification of situations (centered upon various taxonomic procedures such as cluster and factor analysis), Fredriksen (1972) suggested that one could be based upon elicited behavior. The consideration of products or brands for use under various situational contexts would appear to be a worthwhile organizing criterion for such a classification. In this study the setting is consumer perceptions of competitive product-markets.

The composition of competitive product-markets is of considerable practical relevance for a number of managerial and public policy questions (Day and Shocker (1976)). Several researchers have argued for definitions of markets based on consumer behavior or judgment (e.g., Jain and Etgar (1975); Bourgeois, Haines and Sommers (1975)). By and large, these procedures are based upon measures of overall perceived product similarity and have not explicitly considered the effect of usage situations. Since competitive products are more or less appropriate for the same usage situations, a refinement of the approach to include situational effects would be desirable. Thus a corollary purpose of this paper is to develop methodology to implement this concept.

Though this is an exploratory study, there is some precedent for this general line of thought in the work of Stefflre (1971) who found analysis of product-usage associations beneficial in operationally defining pro-duct-markets for the purpose of new product development. His purpose was not to create a situational typology but rather to find groupings of products which were similar in terms of their associations with a common set of uses. Nonetheless, his efforts would appear to provide some legitimization for our present efforts which can be considered a refinement and extension of his. As justification for further research lies in the validity of our findings, care has been taken at various stages to assess this. Of course, validation efforts need replication over a broader array of product-market contexts and over more diverse market segments.

METHODOLOGY

The Methodology progressed through two stages: (1) Determination of varieties of possibly related products and usage-situations and development of a products-by-usage situations matrix (such that each cell (ij) of the matrix provides a measure of the degree of appropriateness of product i for situation j). Subsequently, taxonomic procedures are used to examine the similarities between pairs of situations (in terms of products judged appropriate), to produce an initial situational taxonomy (and definitions of the competitive product-markets). (2) Refinement of the above taxonomy by development of morphological situational descriptions, testing whether these descriptions were interpreted by a new set of respondents in a manner similar to the researchers' conceptualization, and finally replicating the procedure in the previous step with the more complete morphological usage-situational descriptions.

The "breath freshener" market was chosen for the purpose of this exploratory study because: it (1) showed promise in its ability to show variability in product choice across usage situations, and (2) would be a product class familiar to a convenience sample of socially-conscious graduate students in business administration. The latter consideration is especially relevant as familiarity/experience with the product class would be important in the development of meaningful, realistic usage-situations and for minimizing "guesswork" on the part of respondents in indicating appropriateness of products for specific usage-situations (thus reducing the error component of individual differences between subjects).

Stage 1

A procedure adapted from Stefflre (1971) was used to determine the initial set of products and usage situations. Ten graduate students in administration at a major Eastern university were asked by questionnaire to go through an iterative procedure: from this starting point of a "core" of several breath freshening products and examples of usage situations, they were asked to identify additional situations where use of the products would be relevant and other products appropriate to the expanded set of situations, etc. This led not only to lists of commercial products/brands, but also numbers of home remedies (e.g., rinsing mouth with saltwater, sucking a lemon). An alternative data generation method involved group interviews with an additional 8-10 students which, while leading to additional products and situational descriptions, also explored the choice criteria used in determining appropriateness. Generally, the interviews provided a richer data source than had the questionnaires.

Our own judgments supplemented the insights gained from the interview and questionnaire. Duplications and redundancies in products and situations were eliminated. Additional products were added by surveying the product racks of supermarkets and drug stores. In identifying products, an attempt was made to consider all distinctions between brands that might prove relevant (e.g., breath sprays were distinguished from drops, brands such as Close-up--a brand with the flavor and clarity of a mouthwash--were explicitly identified, different flavors were distinguished). Further, a few "extra" products were included in the list to provide some validation for subsequent procedures (the expectation being that such products would be judged inappropriate). This process yielded a set of 18 situations and 46 products.

EXHIBIT 1

EXAMPLES OF PRODUCTS

EXHIBIT 2

EXAMPLES OF SITUATIONS, STAGE 1

An additional sample of 46 graduate student respondents was then asked for each of the 18 usage situations to indicate whether each product was (a) appropriate, (b) inappropriate, or (c) no opinion. Given the large number of judgments required (18 x 46 = 828) and to relieve tedium, this was accomplished by presenting respondents with 46 cards (with product names and identification numbers). After shuffling the cards (once for each situation) and sorting them (products) into the above categories, this information was coded on a questionnaire form.

A Measure of Appropriateness of Product i for Situation j.  By aggregating responses across subjects, it was possible to obtain a products-by-usages matrix such that within each product-situational cell we would have counts for the number of respondents judging (1) appropriate (Aij), (2) inappropriate (Iij), and (3) no opinion (Nij). Define:

EQUATION

as a measure of the degree of appropriateness of product i for situation j. This measure excludes the "no opinion" responses (which were very infrequent) and merely reflects the proportion of people who considered a product appropriate for a particular situation.

The Test for Perceptual Homogeneity.  If there were complete agreement among the subjects, virtually all (or none) would rate the product as a appropriate. Let:

EQUATION

be a measure of the level of agreement between respondents for cell ij of the product-use matrix. By estimating the mean and variance of this measure (across all cells), one can compute whether the level of agreement (homogeneity) is "adequate" (for example, significantly different from 80% or 90%, i.e., vij = 0.40 or 0.45 respectively). For Stage 1, vij has a mean of 0.31 and a variance of 0.0208. This allows us to reject at the a = .01 level the hypothesis that the entries in each cell occur purely by chance. At the same time, we cannot reject the hypothesis of nearly complete agreement (say, vij = 0.45).

Taxonomic Approach. In order to develop a taxonomy of usage-situational influences, we analyze the similarity between the situations (in terms of the products/brands associated with them). One such measure of similarity would be the correlation between columns (situations) in the products-by-usages matrix. This correlation matrix could be used as input into a variety of dimension reducing (taxonomic) techniques. Principal components analysis was the method adopted for the following reasons:

(1) it was felt that since the input measures (degree of appropriateness) were ratio-scaled, a metric scaling technique was applicable.

(2) it allowed us to examine the interrelationships between situations and products in the same reduced space (see below).

(3) because some correlations between situations were very high, the use of a technique such as nonmetric multidimensional scaling would be problematic due to the associated problem of degeneracy.

(4) further, scaling techniques are primary informative about large distance similarities {each point in a scaling configuration is located by reference to all other points, so that large distance similarities dominate) while clustering techniques are more informative about small distance relationships (Kruksal (1972)).

Since for the purpose of this exploratory study, we are more interested in the interrelationships between generic product types and the association between a set of usage situations, it would be more beneficial to use principal components analysis (metric scaling) to examine the large distance similarities. Subsequently, situational clusters could be formed (based on patterns of factor loadings) and product clusters generated (based on the pattern of factor scores). In addition to the development of a taxonomy of the objective usage-situational influences, these situational clusters would serve as a reasonable basis for product-market definition i.e., product clusters closely associated with the same situational cluster being viewed as competitive and constituting a relevant market.

Analysis (Stage 1). Principal components analysis was conducted with the (18) usage situations as variables and the (46) products as cases using the BMDO8M algorithm. The raw data were the appropriateness scores (aij's of products for situations). Varimax rotation was performed to yield a rotated factor matrix that approximated simple structure, and unstandardized factor scores obtained for each of the 46 products. This analysis suggested two or three factors (accounting for 52%, 35%, and 6% of the variance, respectively, in the unrotated factor structure). Interpretation of the rotated factor was as follows:

Component 1 - Social versus Personal Concern (i.e., did the situation involve others or just self-e.g., stale taste in mouth)

Component 2 - Away versus at home (implying privacy and availability of alternatives)

Only one situation ("when you have a sore throat") had high factor loadings on the third component. Since this situation had very low loadings on the first two dimensions it seemed to have little in common with the other situations. Examination of the product-use matrix revealed that products that were generally inappropriate for the other situations (e.g., cough syrups, sodas, and the "extra" products) had comparatively higher measures of appropriateness for this situation. It thus seems that this situation was picking up the variance for the products that were in general considered inappropriate for breath freshening. Accordingly, only two factors were retained.

The unstandardized factor scores for these 46 products were plotted, while not shown here, the results were highly intuitive with toothpastes and mouthwashes having high scores on component 2 (home) and low scores on component 1 (personal concerns). Similarly, candies, mints and gums loaded low on component 2 (away from home) and high on component 1 (social concerns). The "extra" products were not strongly associated with any of the situations thereby indicating their general inappropriateness as breath fresheners.

Discussion. The results of principal components analysis were to a large extent congruent with prior expectations. Commercial products as opposed to "home remedies'' were generally associated with breath freshening. Toothpastes and mouthwashes were closely related in terms of their general association with virtually all situations. Breath sprays and drops appeared to be a distinct category. However, there were some things that were not anticipated. An interesting finding was that for several situations (generally centering around social occasions) mint-flavored gum and candy products were seen as more appropriate breath fresheners than fruit-flavored gums and candies. Flavor appeared to be a better indication of product-market differentiation than the nominal product types (candies or gums).

The purpose of this initial stage was to develop a crude taxonomy of situations (and products). The ad hoc elicited situations had been of varying levels of specificity (e.g., to get rid of stale taste in the mouth - anytime) and completeness, and some more clearly called for breath freshening than others. For example, the situation "when you have a sore throat" was not associated to an appreciable extent with breath freshening products. In addition, four situations resulted in possible confusion regarding their interpretation (as indicated by a low level of agreement (vij) across respondents). In view of these shortcomings, it appeared useful to create a morphological set of usage situations of more uniform levels of specificity and completeness, and then to replicate the procedures described above.

Stage 2

A debriefing session was held with the 46 respondents who had participated at stage 1. They confirmed the ambiguity of several of the troublesome situations, suggested that current descriptions of situations were deficient in not disclosing the amount of time available to freshen breath, and indicated some lack of identification with certain situations. These comments called for changes in the procedures.

Interpretation of the two-factor solution at Stage 1 provided the major guidance for the development of a situational taxonomy. The comments of the students at the debriefing together with our own judgment suggested two additional situational dimensions. Four dimensions were posited:

1. Personal vs. Social Concerns: "Self" - the primary concern is with the subjects personal reactions to either their own mouth freshness or to the fresh- ness of others' breath; "Other" - the subjects primary concern is with others' reaction to their (the subjects') breath freshness. (2 levels)

2. Privacy of Use and Access to Alternatives: "Home"- usage of a breath freshener will be in a location affording considerable privacy of use and ability to use a wide variety of breath freshening alternatives; "Away" - usage of a breath freshener will be in a location affording less privacy of use and more limited ability to use a variety of alternatives. (2 levels)

3. Risk of Being Noticed: "High Noticeability" - a high expectation that the subjects' state of breath will be noticed by others or by self; "Lower Noticeability" - a lower expectation that the subjects state of breath freshness will be noticed by others. (2 levels)

4. Amount of Time to Prepare: "Time to Prepare" - the subject has adequate time to freshen breath; Limited Time to Prepare" - the subject has limited time to freshen breath. (2 levels)

The first two dimensions of the typology were obtained from the interpretation of the Stage 1 factor structure. The third and fourth were added because of their apparent plausibility.

New situational descriptions were generated to correspond to all possible (morphological) combinations of the levels of the four characteristics. 27 such descriptions were developed providing duplication for several of the possible (2 x 2 x 2 x 2=16) cases. In order to determine whether these descriptions validly represented their respective cells, successive groups of six to eight students were given each description together with the typology above and asked to code each description back into the typology. After each iteration an attempt was made to understand those instances where there was substantial disagreement among the subjects. Some situational descriptions needed to be rewritten. Two iterations of this process produced substantial convergence. All but three situational descriptions produced a high level of consensus and these three were dropped from further analysis. Unfortunately, some of the changes resulted in increased representation for several cells of the typology and no representation in a few others. In all further analysis, situational descriptions are assigned to the cells of the typology using this "back coding" of the Judges. Exhibit 3 shows examples of situations in stage 2.

38 respondents in the Graduate School of Management at a major West Coast University completed the task of indicating appropriateness (appropriate/inappropriate/ undecided/product unfamiliar) for the 48 products [Two products were added to the set of products at the first stage.] for 14-15 randomly assigned (from the list of 24) situations. In addition data were collected on the overall effectiveness of each product as a breath freshener and the importance, frequency of occurrence, and "realism" of the situation.

EXHIBIT 3

EXAMPLES OF SITUATIONS, STAGE 2

Measure of Appropriateness. A measure of the degree of appropriateness similar to the one in Stage 1 was designed to account for the inclusion of the undecided category. If Aij = count for appropriate, Iij = count for inappropriate, and Uij = count for undecided, for each cell ij, then:

EQUATION

The counts for Uij (undecided) for each cell were very low compared to Aij (appropriate).

Test for Perceptual Homogeneity. vij, the measure of the level of agreement, was again calculated. It had a mean of 0.258 and a variance of 0.019. This allowed us again to reject (at the a = 0.05 level) the hypothesis that the entries in each cell were due purely to chance. The average level of agreement is lower in this stage. However we must keep in mind that each of the 38 respondents indicated product appropriateness for only 14-15 randomly selected (from the list of 24) situations. Thus on an average, the appropriateness measure (aij) and consequently the level of agreement (vij) were calculated on the basis of 25 responses. Consequently, a somewhat lower level of agreement may be expected as each "aberrant" response will. have a larger effect on the measure of appropriateness.

Analysis. Again principal component analysis using the situations as variables was conducted. The first three unrotated components accounted for 57%, 30%, and 6% of the variance, respectively. Varimax rotation was again performed to improve interpretability. Examination of the rotated factor structure led to an interpretation of the first two factors as at Stage 1. (The third factor was more ambiguous with only two situations loading greater than 0.4 on it and one of those loaded quite heavily on factor 1). From this we conclude that the two additional dimensions in the typology added little. It appears in retrospect that dimension 4 (time for planning) would be closely related to dimension 2 (home vs. away) in that time needed would logically depend on whether a breath freshener needed to be first acquired or not.

EXHIBIT 4

SITUATIONS BY FACTOR LOADINGS AND PRODUCTS BY UNSTANDARDIZED FACTOR SCORES

Exhibit 4 shows the 24 usage situations and 48 products of the second stage plotted in the space of the first two components. Points denoting the usage situations are merely plots of corresponding factor loadings (correlation coefficients with the principal components) which have directional properties only. The points denoting products are unstandardized factor scores, which have properties of both magnitude and direction.

Usage situation and product clusters were derived on the bases of patterns of factor loadings and factor scores, respectively, by visual clustering, in order to determine the relative appropriateness of product clusters for the situational clusters two methods could be used: (1) the projection of points representing products on the directional vector from the origin to the location of the centroid of each cluster of usage situations in reduced space would provide a measure of the relative appropriateness. The greater the projection along this directional vector, the more appropriate the product for the situational cluster. Thus in Exhibit 4, for situation "A", toothpastes and mouthwashes are considered more appropriate than breath sprays and drops. This procedure may be repeated for each cluster of products and usage situations, (2) Alternatively, the original products-by-use matrix provides an explicit measure of appropriateness.

Situational clusters may be formed comprising situations that have similar product (cluster) appropriateness patterns and may be viewed as constituting a single market/submarket. This analysis revealed four submarkets (corresponding to the situational clusters shown in Exhibit 4).

For submarkets 2 and 4, product cluster 1 (mouthwashes and toothpastes) appears to compete with product group 5 (gums and candies). The extent of such competition, however, is an issue requiring further research.

Finally in order to assess the usefulness of the typology, average correlations were computed between pairs of situations coded in terms of the typology. These pairs had either zero to four typological dimensions in common (e.g., suppose situation A was coded 1, 0, 1, 1 (in terms of levels 1 or 2 of the four typological dimensions) and situation B was coded 1, O, O, 1; then the two situations would have three dimensions in agreement). Correlations were across the original product appropriateness measures aij. The results are shown in Exhibit 5. As expected, as the number of typological elements in common increases, the extent to which pairs of situations have appropriate products in common also increases.

EXHIBIT 5

CONSISTENCY OF STAGE 2 USAGE SITUATIONS WITH TYPOLOGY

Discussion

Space did not permit the graphical representation results for Stage 1 (comparable to that reported for Stage 2 (see Exhibit 4)). If it had, the similarity of results between stages 1 and 2 would have been striking-this despite the several differences in stimuli (number of products, substantially different situational descriptions), the samples (eastern and western MBA candidates), and methods of data collection (card sorting versus partially overlapping questionnaires). All this is some testimony to the robustness of the procedures discussed.

As can be noted in Exhibit 4, brands within each generic product type (e.g., toothpastes, mouthwashes, breath sprays, cough drops) are generally close together in the space, indicating that subjects perceived fewer distinctions between brands than between product types and that such brands were generally perceived similarly in terms of their appropriateness to most usage situations (geometrically, the compactness of a cluster should connote a higher level of perceived substitutability with respect to all usage situations). It was particularly interesting to note, however, that flavor served to define the generic product rather than candies or gums. Cough drops were perceived differently from other candies (at least for breath freshening).

One particularly noteworthy consequence of our findings is the fact that just two principal components accounted for approximately 87 percent of the variation in association (93 percent for 3 components). This indicates that it may be possible to generate a parsimonious, yet meaningful situational typology for product types. Also, we observed that the first two factors in the two stages were quite similar. This may imply that "natural" situational descriptions serve about as well as morphologically-created ones for purposes of explaining systematic product-situation associations.

The relatively high measures of agreement among respondents would argue strongly against the presence of significant demand effects in the research. If respondents felt they should respond differently to each situation simply because of the task, it is unlikely that aij would be so high.

In stage 1, the appropriateness measure for each cell of the product-use matrix was based on 46 entries (respondents). On average, the number of entries of each cell was only 25 for stage 2. Yet, the results were strikingly similar. Again, this indicates the stability of the procedure for low (n = 20 to 40) sample sizes. A study by Rosenberg et al (1968) also reports stability of similarity matrices (based on concurrence/ association measures) for small sample sizes (n = 37).

Conclusions

The results of the study are consistent with the fol- lowing conclusions:

(1) situational influences appear to affect the choice of products/brands in a systematic manner.

(2) a parsimonious typology of objectively meaningful situations can be constructed based on behavioral criteria (the brands considered appropriate by consumers).

(3) "ad hoc" (stage 1) descriptions of usage-situations serve about as well as descriptions based on a morphology (stage 2) in determining the composition of product-markets. In fact, the results of the two stages are strikingly similar.

(4) product-market composition based on perceived consumer behavior across usage-situations would appear to stretch across generic product types (although the bases for defining a product-type may not always be apparent).

(5) small sample sizes (30-40), using the procedure adopted, produced stable results.

Though we cannot claim to have developed a general situational taxonomy (although there is some similarity between our two dimensions and those of Belk (1975)), the procedures used here at least appear useful for a variety of scientific and managerial purposes. Further research is necessary with more significant sample sizes representing a broader cross-section of consumers and across a wider range of product-markets before generalizations can be claimed. But we have at least suggested that with product familiarity and usage situation controlled for, people's perceptions of the nature and composition of competitive markets are relatively stable. Individual differences in market behavior (e.g., in preferences and choice). which have become an important topic in consumer research, can more productively be explored within such a context.

REFERENCES

Allen, B. L. "Situational Factors in Conformity," in Leonard Berkowitz, Advances in Experimental Social Psychology, Vol. 2, New York: Academic Press, 1965.

Barker, R. G. "Commentaries on Belk, 'Situational Variables and Consumer Behavior'". Journal of Consumer Research, Vol. 2, Dee. 1975, p. 165.

Bearden, W. O. and A. G. Woodside "Interactions of Consumption Situations and Brand Attitudes," Journal of Applied Psychology, Vol. 61, No. 6, 1976, pp. 764-769.

Belk, R. W. "Situational Variables and Consumer Behavior,'' Journal of Consumer Research, Vol. 2, Dec. 1975, pp. 157-164.

Bourgeois, J.C., Haines, G. H. and M. Sommers "Defining an Industry," paper presented at the joint national conference of ORSA and TIMS, Las Vegas, Nevada: Nov. 17-19, 1975.

Bowers, K. S. "Situationism in Psychology: An Analysis and Critique," Psychological Review, Vol. 80, No. 5, Sept. 1973, pp. 307-336.

Day, G. S. and A. D. Shocker "Identifying Competitive Product Market Boundaries: Strategic and Analytical Issues." Working Paper #156, University of Pittsburgh, 1976.

Endler, N. S. and D. Magnusson "Toward an Interactional Psychology of Personality," Psychological Bulletin, Vol. 83, No. 5, 1976, pp. 956-974.

Frederiksen, N. Toward a Taxonomy of Situations," American Psychologist, Feb. 1972, pp. 114-123.

Hansen, F. "Psychological Theories of Consumer Choice," Journal of Consumer Research, Vol. 3, Dec. 1976, pp. 117-142.

Jain, A. K. and M. Etgar "How to Improve Antitrust Policies with Marketing Research Tools," in Edward M. Mazze (Ed.), 1975 Combined Proceedings of the American Marketing Association, Series #37, 1915, pp. 72-75.

Kruskal, J. B. "Relationship of Clustering to Scaling and the Interpretation of Neighborhoods," Bell-Penn Workshop on Multidimensional Scaling, June 7-10, 1972.

Rosenberg, S., Nelson, C. and P.S. Vivekananthan "A Multidimensional Approach to the Structure of Personality impressions," Journal of Personality and Social Psychology, Vol. 9, 1968, pp. 283-294.

Russell, J. A. and A. Mehrabian "Environmental Variables in Consumer Research," Journal of Consumer Research, Vol. 3, June 1976, pp. 62-63.

Stefflre, V. J. New Products and New Enterprises: A Report of an Experiment in Applied Social Science, Irvine, Calif.: University of California, March, 1971.

Wicker, A. W. "Processes Which Mediate Behavior - Environment Congruence," Behavioral Science, Vol. 17, May 1972, pp. 265-277.

Wicker, A. W. "Commentaries on Belk, 'Situational Variables and Consumer Behavior'", Journal of Consumer Research, Vol. 2, Dec. 1975, pp. 165-167.

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