Assessing Self Concept

J. Michael Munson, University of Santa Clara
W. Austin Spivey, University of Santa Clara
ABSTRACT - Two approaches to self measurement are compared: 'expressive self' which pertains to either ideal self or looking-glass self, and 'product expressive self' which either relates one's own image to a product preference or looking-glass image to a product preference. Results show people evaluate various self-concept constructs differently; but the differences are not generalizable to all product classes. Evaluations are dependent upon the product symbolism and conspicuousness. The various self constructs are also affected by specific covariates, such as sex and occupation.
[ to cite ]:
J. Michael Munson and W. Austin Spivey (1980) ,"Assessing Self Concept", in NA - Advances in Consumer Research Volume 07, eds. Jerry C. Olson, Ann Abor, MI : Association for Consumer Research, Pages: 598-603.

Advances in Consumer Research Volume 7, 1980     Pages 598-603

ASSESSING SELF CONCEPT

J. Michael Munson, University of Santa Clara

W. Austin Spivey, University of Santa Clara

ABSTRACT -

Two approaches to self measurement are compared: 'expressive self' which pertains to either ideal self or looking-glass self, and 'product expressive self' which either relates one's own image to a product preference or looking-glass image to a product preference. Results show people evaluate various self-concept constructs differently; but the differences are not generalizable to all product classes. Evaluations are dependent upon the product symbolism and conspicuousness. The various self constructs are also affected by specific covariates, such as sex and occupation.

INTRODUCTION

One's environment is seen in ways which are acceptable to self image even if perceptual distortion of the environment is necessary to make it consistent with self needs. Thus, self concept plays a crucial role in motivation by organizing the wants and goals of the individual. The urge to preserve a relatively stable, somewhat flattering, and steadily more satisfying self image is the driving and organizing force behind a large share of human activity (Douglas, Field and Tarpey 1967). Some important goals emerge which enhance and defend self (Kretch, Crutchfield and Ballachey 1962). Grubb and Grathwohl (1967) observe that the more valued one's self, the more organized and consistent one's behavior. In fact, Hayakawa (1963) notes that self concept is the fundamental determinant of all behavior. Indeed, Douglas, Field and Tarpey (1967) suggest that self image is "man's most valuable possession and is the key to his behavior."

Importance to Communicators

These notions have led many to conclude that self is an important construct in consumer decision making. Some believe that successful promotion of a particular product requires the development of an unambiguous brand image that differentiates the product from the competition and imbues it with its own character. Some practitioners and academicians believe that a brand's image must be made congruent with the self concepts of consumers in order for them to find a brand appealing and identify with it. Numerous studies (Grubb and Grathwohl 1967; Birdwell 1968; Dolich 1969; and Munson 1973) have suggested that the key concern for communicators about products is the linkage between self image and product image. Considering only the classic tripartite view of self (Boone and Kurtz 1974)--actual self, ideal self, and looking-glass self--is insufficient. A communicator must somehow link product to self; a communicator must recognize the existence of "product expressive" selves. These are self constructs which make this key linkage explicit. Although previous studies have not explicitly defined 'product expressive' selves, a few studies have operationalized self concept along these lines (e.g., Sommers 1964; Munson 1973; and Landon 1974). Two possible such 'product expressive' self constructs are as follows:

1.  SP (self perception given a product preference) is how a person perceives oneself given a preference for a specific product.

2.  OSP (others perception of self, given a product preference) is how a person believes other people view him given a preference for a specific product.

The first image, SP, provides an indication of how the individual perceives himself based upon his own interpretation of the relevance of the product's symbolic cues for one's self image. The second image, OSP, is an indication of how the individual perceives himself based upon his interpretation of the symbolism attached to the product by others. However, both constructs reflect the notion that the individual attempts to assess the match between the product and one's self image and determine its suitability as a viable vehicle for expression of one's self.

Despite numerous studies and the continued interest in self concept research, there has been little research of an integrating nature to help understand the relationship between self and product. Previous studies have involved brand preference (Dolich 1969; Hamm and Cundiff 1969; Ross 1966, and Munson 1973) purchase intention (Landon 1974) and actual product ownership (Birdwell 1968; Grubb and Hupp 1968). Findings are equivocable; for example, a communicator cannot yet discern whether actual or ideal self image is more important in influencing brand preferences, and the role of looking-glass self has not yet been investigated.

Thus, a basic question remains unanswered--which self construct is most relevant to communicators? Given that the key concern for marketers is the linkage between product image and self image, actual self is theoretically of less concern than any other construct. Theoretically, each person's self conception is a selective working compromise between an ideal image and the images forced upon him by his imperfect behavior in actual situations (Turner 1968). Also, one could argue from a conceptual standpoint that individuals define themselves through feedback from others; thus, there is little difference between actual self and looking-glass self. On the other hand, distinctive role demands or situations may cause either ideal self or looking-glass self to be the more important. Of particular interest is the likelihood that one owns or prefers a product in order to be closer to one's ideal self or to express a particular looking-glass self.

Other questions are introduced by this research--is there more than one "product expressive" self, and if so, which "product expressive" self is more relevant? On the one hand, product expressive images could be very similar for much the same reasons that looking-glass self and actual self could be similar. If true, the notion of product expressive self would be an unnecessary complication in any attempt to link self image to product image. As such, the question becomes moot. On the other hand, role demands or situations could cause either product expressive self (SP or OSP) to be more important in a consumer's decision. As such, recognition of the differences between SP and OSP could provide insight into the process by which self image influences product consumption.

HYPOTHESES

Surely the answer to these three questions is hidden by the confusion surrounding the linkage between self image and product image. Possibly, the answer lies in studying the discrepancies between different stimuli involving:

Expressive Self Constructs

I = Ideal Self

LG = Looking-glass Self

Product Expressive Self Constructs

SP = Self image given a product preference

OSP = Self's perception of how one is viewed by others given a product preference

One might expect a systematic variation in the particular self constructs which are important in product decisions depending upon situational and social factors. For example, for highly conspicuous, socially consumed items, a product may be evaluated with respect to its ability to enhance or express LG. Furthermore, one may be unable to separate one's own image from the stereotype image; thus, causing SP to merge with OSP. Of course, there are products where user stereotypes are neither well-developed nor well-known. For these products, the typical individual will be less certain about the meaning behind a product preference. Here, ideal self may be a more appropriate baseline. For these products the individual need not be as worried about reprisals from others, or negative sanctions for deviation from group norms. One has a chance to be "one's self." These individuals will respond to measures of OSP based primarily upon their own needs and values. That is, SP influences OSP, causing them to merge. Essentially, a single hypothesis is implied from these ideas about self image and product image.

H1:  Depending upon product conspicuousness and symbolism, the discrepancy between various self stimuli will vary.

But this study entails more than a single global hypothesis because actual self is of more than tangential interest. A primary concern is that the data from self studies is artifactual depending upon the inclusion of a measure of "actual" self. It seems intuitively reasonable to expect someone to be more accustomed to thinking about one's "actual" self. Responding to "actual" measures, therefore, should be easier than responding to any other measure of self. But having an individual assess only "actual self" could result in their projecting some aspects of "ideal self" into the measure. Thus, the report of actual self would be less accurate. On the other hand, having individuals respond about actual self then respond about other self images without a preview of all measures involved could distort the response pattern for all measures. Actual image could be distorted because of the projection of ideal self into the response, ideal self could be distorted because the respondent feels compelled to make it even "more perfect," etc. Finally, individuals not completing a measure of actual self could respond to other measures as though actual self were being indexed. It is hypothesized, therefore, that :

H2:  Without previewing all measures of self involved, having a person index "actual" self image prior to indexing other images of self will affect the response pattern between self image and product image.

METHODOLOGY

Subjects

In order to test these hypotheses, a sample was chosen primarily from businessmen and businesswomen working at high-technology companies located in Northern California. Of these, 19 were skilled manual employees or clerical workers; 68 were administrative personnel; 25 were business managers; 67 were executives or professional; 18 were unemployed; and 13 failed to respond to the occupational question. Of these respondents, 69 were women and 141 were men. All were asked to participate in a study designed to obtain their evaluations of some proposed new products. Participation was voluntary.

Instruments

Five measures were included:

1.  Actual self (A) is "the sort of person you actually are or the way in which you actually see yourself."

2.  Ideal self (I) is "the sort of person you would most like to be or the way in which you would like to see yourself."

3.  Looking-glass (LG) is "the way you think other people see you, or the sort of person you think other people see you as."

4.  Self-image, given self has a product preference (SP) is "how you see yourself assuming you owned or preferred (product name inserted, eg., Brand M)."

5.  Others' image of self, given self has product preference (OSP) is "how you think others would see you if you preferred or owned (product name inserted, e.g., Brand M)."

All were measured by having subjects respond to seven-point bipolar semantic differential scales. Each subject, however, completed only a specific combination of self measures, depending upon their treatment group in the experimental design (discussed in Procedure section).

The approach to reassuring self concept was a modification of that used by Ross (1966) and Munson(1973) which in turn was based on the work of Wells, et al. (1957) and Bills (1953). The instrument contains 15 semantic differential scale items (e.g., dominating-submissive, successful-unsuccessful, informed-uninformed). These items were culled from over 17,000 traits of Allport (1937), and represent ". . . items which occurred frequently in client centered interviews and seemed to present clear examples of self-concept definition" (Bills 1953). The scale items demonstrated adequate test-retest reliability over a three week interval on sample of 64 subjects when used to measure actual and ideal self (Munson 1973). Ross (1971) performed specific converging operations in demonstrating the construct validity of this measure.

In addition to the self measure, classic demographic characteristics, such as age, sex, and occupation were recorded.

Product Class

Two product categories, automobiles and tennis rackets, were investigated, Previous research has shown the automobile to be highly symbolic and conspicuous, while the tennis racket is less conspicuous and symbolic (e.g., Cohen and Barban 1970; Munson 1973; Vanderwicken 1975; and Fiott 1979). Tennis rackets are perceived as less appropriate to the communication of one's image and more suitable for satisfying functional needs. Fiott (1979) suggests that today both men and women judge a tennis racket more on the basis of its physical features than on other criteria, such as who endorses it. It is likely that a few people have developed strong stereotypes of users of specific types of tennis rackets (e.g., players who use wooden rackets may be viewed as traditionalists; those using oversized rackets may be seen as mediocre in playing ability, etc.). But such stereotypes are less well developed at the brand level, compared to those for automobile brands.

Product Messages

Choosing new or existing products within these product classes is problematic. To test the research questions by using an existing product introduces two possible sources of bias: product ownership and existing stereotypes. If an existing product is used, ownership must be controlled for, otherwise product-expressive self could be confounded by owner's projections of their self image into the product. Even if this bias is accounted for, the existing user stereotype may preclude using SP in the product choice decision. For symbolic products, whether new or old, it is probable that OSP will be more relevant than SP. For example, asking someone to complete measures of SP and OSP for a "Cadillac" or "Jimmy Connors Autograph Racket" could lead to a parroting of the established image, thereby causing OSP to become relevant for product evaluation. On the other hand, if the product is not symbolic and conspicuous, OSP could be less relevant. The net result is that irrespective of product symbolism or conspicuousness, an individual may respond in the same way to either measure of product-expressive self. But if an existing product is used, variation in response is even less likely; therefore, the most important consideration seems to be whether or not the product is conspicuous and symbolic. Thus, only "new" products were included. This was felt to provide a more conservative test of the hypotheses.

An identical message about the two new products was given to each respondent. The presentation format was similar to that used in Consumer Reports. Based on a pretest sample of 54 people, information was provided about both product attributes and the life-style of those preferring the product. The product attributes mentioned were those which were judged to be the most salient. The life-style characteristics were described in the same way for both products.

Procedure

The research hypotheses were tested by using ANCOVA. Basically, the data were generated within the 2 X 2 factorial, design shown in Figure 1. Factor E is Expressive Self, with one level measuring looking-glass self (LG) and the other ideal self (I). Factor FE represents Product Expressive Self. One level measures self image, given self has a product preference (SP), and the second indexes one's perception of self as seen by others, given a product preference (OSP). This design generated 4 unique treatment cells, each requiring different combinations of the various self-concept constructs. For example, subjects in cell 1 completed instruments measuring only % and SP, while those in cell 4 completed LG and OSP. To implement the design, 8 different questionnaires were administered, with 4 reflecting unique sets of constructs and the remaining 4 representing variations in the order of presentation of the constructs. Only one-half of the subjects in each cell were asked to complete a measure of actual self. Thus, the experiment may be viewed as a 23 factorial, if one regards the existence or absence of the actual measure as a third factor.

FIGURE 1

2 X 2 FACTORIAL DESIGN FOR INVESTIGATING VARIOUS CONSTRUCTS OF SELF CONCEPT

Subjects were assigned to cells at random and each subject was exposed to only one treatment condition. Measures of SP and OSP were obtained from most subjects for each of two products (Brand "M" automobile and Brand "M" tennis racket). All 210 subjects completed measures of either SP or OSP for Brand "M" automobile; however, subjects who indicated they "never play tennis" (n = 101) were not asked to complete SP and OSP for Brand "M" tennis racket. Within each treatment condition, one-half of the subjects received one order of instruments. For the other one-half, the order was essentially reversed.

Response variable

The response variable for analysis in the ANCOVA design (Figure 1) is the D2 statistic, where D2 is a generalized measure. When measured over scales, each self construct can be regarded as a profile, and each represented as a point in k-dimensional space, with each scale corresponding to a dimension. Dissimilarity of any two profiles is the linear distance between their respective points in k-dimensional space. As a method of comparing profiles, the distance measure (D2) has some advantages over procedures based on correlation (Cronbach and Gleser 1953; and Munson 1973). The latter method tends to ignore mean differences between profiles, whereas the distance measure reflects both mean and shape (covariance) similarity between profiles.

All D2 measures represent discrepancies between levels of Factor E (Expressive Self) and Factor PE (Product Expressive Self). That is, the following discrepancies were compared: I vs. SP; I vs. OSP; LG vs. SP; LG vs. OSP. Discrepancy between any pair can be obtained by summing the squared differences over scales between a subject's scores on each profile. The legitimacy of this technique for use in self concept research is well documented (e.g., Cronbach and Gleser 1953; Evans 1968; and Munson 1973).

RESULTS AND DISCUSSION

ANCOVA results clearly show that the concern for artifactuality in self studies is warranted. Regardless of product type, including a measure of actual self increases the D2 between any pair of expressive-self constructs and product-expressive self constructs: I vs. SP, I vs. OSP, LG vs. SP, and LG vs. OSP. For the automobile this result is significant at the O.10 level; for the tennis racket, it is significant at the 0.01 level. The associated means are:

TABLE

There may be several reasons behind this result as noted in the "Hypotheses" section.. For this study, the most probable explanation is that both the ideal and the looking-glass measures are particularly poor representatives of I and LG for those who completed an actual measure. Actual self was measured at the beginning of the questionnaire, and no one knew that other self measures would follow. Believing that the inclusion of actual self led to increased awareness about other possible self constructs is, therefore, unrealistic. It seems more likely that actual's inclusion led to distortion.

Global Hypothesis

In spite of this possibility for distortion, the results in Tables 1 and 2 imply that there is no practical difference in the product anchored constructs of SP and OSP. [Essentially the data was analyzed by using ANCOVA within the framework of a 23 experiment where the inclusion or deletion of the actual measure formed the two levels of the third factor. Consequently, all the results shown are purged of the effect of including a measure of actual self.] The main effect for factor PE is not significant for either product. While conceptually one may argue for the existence of both SP and OSP, the instruments used in this study are unable to discriminate between them. The respondents essentially merge the two constructs.

TABLE 1

ANALYSIS OF VARIANCE SUMMARY TABLE: AUTOMOBILE

TABLE 2

ANALYSIS OF VARIANCE SUMMARY TABLE: TENNIS RACKET

Even though the mechanisms for the merger may depend upon product type, the net result is the same. For a highly conspicuous product (e.g., automobiles), one may know the symbolic meaning attributed to the product by various sources, such as the commercial media or reference groups. Knowing these, one evaluates self image along the lines of stereotyping prevalent among significant others. The process by which this occurs is unknown; perhaps one executes a self-talk where one asks "If I use this product, what will others think of me?" For inconspicuous products, however, there is less chance for a reference; therefore, one simply projects one's own beliefs as being the beliefs of others.

Or perhaps people cannot distinguish cognitively between the two constructs?

On the other hand, the results in Tables 1 and 2 clearly show that respondents are able to distinguish between ideal self and looking-glass self in some situations. As expected, Expressive Self is significant only for automobiles. The reason is because automobiles are a better vehicle for value expression than tennis rackets which are more instrumental in achieving utilitarian goals (Fiott 1979), This view is supported by noting that the average D2 for automobiles was about 68 compared to an average D2 for tennis rackets of about 79.

Generally, therefore, one may conclude that some products are more amenable to image expression than others. When evaluating the suitability of an automobile, individuals could be expected to be more concerned with "how they looked to others or with what others would think" rather than with its functional characteristics, per se. Given the strong stereotypes which may have developed over the years for automobiles, that they are an integral part of LG seems likely. In such cases, LG is very similar to both SP and OSP for specific brands. That is, products, such as houses, suits and vacations tend to be used to stereotype, and one uses these same products to formulate LG. Hence, it is very similar to both product-expressive selves; the discrepancies among them are small.

However, for tennis rackets, just the opposite is true. In the absence of strongly developed user stereotypes (the media has yet to become saturated with ads promoting rackets based on symbolic cues), there is little homogeneity for OSP among respondents. In fact, postulating that one is less confident about both SP and OSP seems reasonable. The lack of homogeneity and confidence operates to increase not only the discrepancy between LG and PE, but also the variance associated with the discrepancy. In addition, in view of their more utilitarian nature, tennis rackets are poor instruments for facilitating the movement of either SP or OSP toward one's ideal self. Consequently, the D2 between not only I and PE, but also LG and PE is large in both magnitude and variance; therefore, Factor E is insignificant for tennis rackets.

Covariates

The significant covariates in Table 2 further support the notion of heterogeneity about the image of tennis rackets. The D2 among the various combinations of self constructs were related to both sex and occupation. Males tend to perceive tennis rackets as even more utilitarian than do women. That is, men are more concerned about performance criteria, such as racket vibration, power, and control, while women are more concerned about who uses the racket.

For occupation, respondents who hold full-time jobs exhibited significantly lower D2 than the unemployed respondents. This suggests that those with limited funds perceive tennis rackets more on a utilitarian dimension than on a value-expressive dimension. Presumably, when income is limited, one is more "objective."

IMPLICATIONS

The findings of this study have significant implications for both researchers into self concept and marketing strategists.

Research Considerations

For theorists, noting that respondents discriminate between ideal self and looking-glass self is important. Just as noteworthy is the possibility that respondents may not be able to distinguish between their "own" feelings about a product and their beliefs about how they are liable to be viewed by others if they own or prefer a particular product. These constructs may be inseparably interwoven. Even if individuals can discriminate between them, the practical differences seem slight. For conspicuous, symbolic products one's assessment of OSP seems to overpower one's evaluation of SP. And vice-versa for less conspicuous, less symbolic products. The net result is that researchers need not be overly concerned about which product-expressive self is used.

However, researchers should be concerned about whether or not a measure of actual self is included. Its inclusion dramatically increases the D2 regardless of the pair of expressive selves and product-expressive selves being considered. The reason is unclear. On the one hand, it may be easier to think of one's self in "actual'' terms. On the other hand, when responding to a measure of actual self, one may be tempted to project aspects of ideal self into the measure. Consequently, subsequent measures of either ideal or looking-glass self are distorted. More study is needed on this notion. Until then, researchers would be well advised to preview all the measures for a respondent, then begin the measurement process with actual self, Doing so could sensitize the respondent about the need to distinguish among the different constructs and allow them to "practice" on perhaps the easiest measure to understand.

Strategy Considerations

These methodological and theoretical implications have strategic impact. Marketers should recognize that products vary in their ability to project image cues to others; therefore, segmentation strategy could be formulated around the differences. Specifically, these results show the problems a marketer could have in using a purely demographic segmentation base. For some products, user stereotypes may be so well established or effectively communicated that homogeneous viewpoints are held regardless of age, occupation or sex.

This idea, however, is not meant to diminish the importance of demographics; rather it is meant to highlight their proper role. For example, marketers of tennis rackets could apply the results of this type of study directly. Assuming that the results obtained from an externally valid sample showed men to be more utilitarian toward tennis rackets than women, males would be more trial oriented. The diffusion of a new tennis racket would be made more effective by implementing a demonstration program for men, with opportunity for trial. For women, however, using testimonials stressing cues related to social desirability would seem more effective.

In conclusion, this study has been accomplished primarily as an aid to conducting and interpreting research about self concept. Results indicate that one's self evaluation of specific instruments meant to index different constructs of self are affected not only by which constructs are measured, but also by which products are evaluated. If advances are to be made in understanding and predicting consumers' behavior, the role of self concept must be better specified. This study represents an initial step toward this end.

REFERENCES

Allport, G. W. (1937), Personality: A Psychological Interpretation, New York: Holt.

Bills, R. E. (1953), "A Comparison of Scores on the Index of Adjustment and Values with Behavior in Level-of Aspiration Tasks," Journal of Consulting Psychology, 17, 206-212.

Birdwell, A. E. (1968), "A Study of the Influence of Image Congruence on Consumer Choice," Journal of Business, 41 (January), 76-88.

Boone, Louis E. and Kurtz, David L. (1974), Contemporary Marketing, Hinsdale, Illinois: The Dryden Press.

Cohen, Joel B. and Barban, Arnold (1970), "An Interactive Consumer-Product Typological System: A Progress Report and Partial Evaluation," Working Series in Marketing Research, 12 (December), College of Business Administration, The Pennsylvania State University.

Cronbach, L. J. and Gleser, G. G. (1953), "Assessing Similarity Between Profiles," Psychological Bulletin, 50, 465-474.

Dolich, I. M. (1969), "Congruence Relations Between Self Images and Product Brands," Journal of Marketing Research, 6 (February), 80-84.

Douglas, John; Field, George A.; and Tarpey, Lawrence X. (1967), Human Behavior in Marketing, Columbus, Ohio: Charles E. Merrill Books, Inc., 64-67.

Evans, Franklin B. (1968), "Ford versus Chevrolet: Park Forest Revisited," Journal of Business, 41 (October), 445-449.

Fiott, Steve (1979), "Equipment," Tennis USA, (April), 11-12.

Grubb, Edward L. and Grathwohl, Harrison L. (1967), "Consumer Self-Concept, Symbolism, and Market Behavior: A Theoretical Approach," Journal of Marketing, 31 (October), 22-27.

Grubb, Edward L. and Hupp, Gregg (1968), "Perception of Self, Generalized Stereotypes, and Brand Selection," Journal of Marketing Research, 5 (February), 58-63.

Hamm, B. Curtis and Cundiff, Edward W. (1969), "Self-Actualization and Product Perception," Journal of Marketing Research, (November), 470-72.

Hayakawa, S. I. (1963), Symbol, Status, and Personality, New York: Harcourt, Brace, Javanovich, Inc., 37-38.

Kretch, David; Crutchfield, Richard S. and Ballachey, Edgerton L. (1962), Individual in Society, New York: McGraw-Hill Book Co., Inc., 79-83.

Landon, E. Laird, Jr. (1974), "Self-Concept, Ideal Self-Concept and Consumer Purchase Intentions," Journal of Consumer Research, (September), 44-51.

Munson, J. Michael (1973), "A Typological Investigation of Self Concept Congruity and Brand Preferences: Toward a Predictive Model," unpublished doctoral dissertation, University of Illinois, Urbana, Illinois.

Ross, Ivan (1966), "The Relationship Between Self-Concept and Judgments Concerning Selected Consumer Products," unpublished doctoral dissertation, Ann Arbor, Michigan: University Microfilms.

Ross, Ivan (1971), "Self Concept and Brand Preference," The Journal of Business, University of Chicago Press, 44 (January), 38-50.

Sommers, M. S. (1964), "Product Symbolism and the Perception of Social Strata," in Toward a Scientific Marketing, ed. S. A. Greyser, Chicago: American Marketing Association, 200-216.

Turner, Ralph H. (1968), "The Self-Conception in Social Interaction," in The Self in Social Interaction, Vol. 1, ed. C. Gordan and K. J. Gergen, New York: John Wiley and Sons, 93-106.

Vanderwicken, Peter (1975), "What's Really Wrong at Chrysler," Fortune, (May), 176.

Wells, W. D.; Andriuli, F. J.; Goi, F. J.; and Seader, S. (1957), "An Adjective Checklist for the Study of 'Product Personality'," Journal of Applied Psychology, 41, 317-319.

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