Self-Image/Product-Image Congruence Models: Testing Selected Models

ABSTRACT - Self-image/product-image congruence models employed in consumer self-concept studies were identified and criticized for their atheoretical foundation. These models include the absolute difference, the simple difference, the difference squared, the Euclidean distance, and the divisional models. This study is designed to compare the predictive strength of a theoretically-based congruence model (referred to as the Interactive model) with those traditional congruence models in relation to purchase motivation (i.e., product preference and purchase intention). The results showed that the Interactive congruence model predicted consumers' purchase motivation only slightly better than the traditional models.

Citation:

M. Joseph Sirgy and Jeffrey E. Danes (1982) ,"Self-Image/Product-Image Congruence Models: Testing Selected Models", in NA - Advances in Consumer Research Volume 09, eds. Andrew Mitchell, Ann Abor, MI : Association for Consumer Research, Pages: 556-561.

Advances in Consumer Research Volume 9, 1982      Pages 556-561

SELF-IMAGE/PRODUCT-IMAGE CONGRUENCE MODELS: TESTING SELECTED MODELS

M. Joseph Sirgy, Virginia Polytechnic Institute and State University

Jeffrey E. Danes, Virginia Polytechnic and State University

ABSTRACT -

Self-image/product-image congruence models employed in consumer self-concept studies were identified and criticized for their atheoretical foundation. These models include the absolute difference, the simple difference, the difference squared, the Euclidean distance, and the divisional models. This study is designed to compare the predictive strength of a theoretically-based congruence model (referred to as the Interactive model) with those traditional congruence models in relation to purchase motivation (i.e., product preference and purchase intention). The results showed that the Interactive congruence model predicted consumers' purchase motivation only slightly better than the traditional models.

INTRODUCTION

Self-concept studies in consumer behavior used a variety of methods to measure degree of congruence between self-image (actual self-image, ideal self-image, etc.) and product image in relation to product preference, purchase intention, and product ownership. These methods include algebraic models (e.g., distance models, multiplicative models), product-anchored measures. correlational indices, measures utilizing mean scores and frequency breakdowns, and other grouping techniques such as factor analysis and multidimensional scaling (see Sirgy, (forthcoming) for a comprehensive literature review).

This paper (1) reviews the measurement models used to index the congruity between consumer's self-concept and product image in relation to consumer behavior, (2) introduces a theoretically-derived self-image/product-image congruence model, and (3) compares the predictive strength of the theoretically-derived congruence model with those traditionally used in the consumer self-concept literature.

LITERATURE REVIEW

Birdwell (1968) in his classic self-concept study attempted to determine the differential effects of self-image/product-image congruence on car ownership. A generalized Euclidean distance model was used to test the self-congruity hypothesis.

EQUATION    (1)

where

Dj = overall discrepancy between product images (Pi)s and actual self-concepts (ASi)s for individual (j).

i = a particular personality image (i) as it is associated with a particular product and consumer self-concept.

Pij = a specific product image (i) of individual (j).

ASij = the corresponding actual self-image (i) of ] individual (j).

The relationship between actual self-image/product-image congruence and product preference, purchase intention, or product ownership has been referred to in the consumer self-concept literature as the "self-congruity hypothesis." Self-concept and product image were measured using a semantic differential scale. Results were significant and supportive of the self-congruity hypothesis. Other applications of the Euclidean distance model in consumer self-concept studies include Delozier (1971), Delozier and Tillman (1972), Green, Maheshwari, and Rao (1969), and Maheshwari (1974).

The Delozier study (Delozier, 1971; Delozier & Tillman 1972), which analyzed the data at an individual level, [Individual-level analysis refers to statistical-analysis performed using the subject as the unit of analysis. This is usually accomplished by correlating the behavior scores (e.g., preference, intention) with self image/product-image congruity scores for each individual. This is to be compared with the traditional aggregate-or group-level analysis in which the behavior scores are correlated with congruity scores among individuals. That is, each subject gets a behavior score and a generalized congruity score. These two variables are then correlated across sample subjects.] found additional support for the self-congruity and ideal congruity hypotheses. [The "ideal congruity hypothesis" refers to the relationship between ideal self-image/product-image congruence and a consumer behavior variable such as product preference, purchase intention, or product ownership. The ideal self-image is defined as "what I ideally like to be."] The Green, Maheshwari, and Rao's (1969) study analyzed the data also at an individual level, and contrary to expectations, the results were not supportive of the self-congruity hypothesis.

Maheshwari's (1974) doctoral dissertation study analyzed the data at both the aggregate and individual levels. His results provided some support to the ideal social-congruity hypothesis [The "ideal social congruity hypothesis" refers to the relationship between ideal social self-image/product-image congruence and a consumer behavior variable such as product preferences purchase intention, or product ownership. The ideal social self-image is defined as "what I ideally like to be seen by others."] but little support for the social Congruity hypothesis. [The "social congruity hypothesis" refers to the relationship between social self-image/product-image congruence and a consumer behavior variable such as product preference, purchase intention, or product ownership. The social self-image is defined as "what I believe others see me as."]

Dolich (1969) hypothesized that product conspicuousness moderates the extent to which the actual self-image influences product preference relative to the ideal self-image. Similar to most of the studies discussed above, actual self-image, ideal self-image, and product image were measured using a semantic differential scale, and congruence scores were derived using a generalized Absolute Difference Model:

EQUATION    (2)

where ISij = the corresponding ideal self-image (i) of individual (j).

The data provided support for both self- and ideal-congruity hypotheses but failed to substantiate the moderating effects of product conspicuousness on the relationship between self-image/product-image congruity and product preference.

Maheshwari (1974) also used the generalized Absolute Difference motel (which was referred to as the "City-Block Model") in comparison with the generalized Euclidean Distance model. The results indicated no significant predictive differences in preference scores between the two congruence models.

Sirgy (1979, 1980) used a similar generalized Absolute Difference model to test hypotheses involving self-congruity, ideal congruity, social congruity, and ideal-social congruity. It was expected that ideal and ideal-social congruity would be more pronounced in relation to product preference than purchase intention, and self- and social-congruity would be more influential on purchase intention than product preference, and that these relationships are moderated by "product personalization." Product personalization was defined as the extent to which a product has personality symbols and associations. Self-Concept as well as product image were measured [In comparison to the traditional semantic differential scale which utilized bipolar adjectives, the format used in this study incorporated unipolar adjectives. This procedure was argued to better suited for self-concept measurement since it allows for the measurement of the applicability or certainty attached to each adjective. It therefore avoids the forced associations of the bipolar format (Sirgy, 1980b).] using a unipolar form of the semantic differential scale. The results provided support to the hypotheses but only with respect to some products. .. Ross (1971) used a generalized Difference Squared motel to test the moderating effects of product conspicuousness on the relationship between self-image/product-image congruity ant product preference. The study employed a semantic differential to measure actual-, ideal-self-image, and product image. This congruity model can be represented as,

EQUATION    (3)

The results provided support for the self- and ideal-congruity hypotheses, but failed to support the moderating role hypothesis of product conspicuousness.

Schewe ant Dillon (1978) used a Simple Difference motel (Pij-ASij) and (Pij-ISij) as a congruence measure. However, compared to other studies, the index was not "generalized" or summed across all attributes (images) but treated at the image level. Similar to most other studies, self-concept and generalized stereotype of product user together with significant others were measured using a semantic differential scale. The analysis provided some support for the self- ant ideal-congruity hypotheses.

In a direct test of various congruence motels, Hughes and Naert (1970) compared the predictive effectiveness (dependent variable being probability that subject will replace present car) of the following congruence models:

TABLE

The results showed that the weighted (by importance) versions of self-image/product-image congruence models were more predictive than the unweighted ones, and also the self-congruence models were found to be less predictive of consumer choice compared to attitudinal model.

The congruence models employed by Birdwell (1968), Delozier (1971), Delozier and Tillman (1972), Green, Maheshwari, and Rao (1969) Maheshwari (1972), Dolich (1969), Ross (1971), Schewe and Dillon (1978), and Hughes and Naert (1970) are all criticized for being atheoretical. No theoretical self-concept rationale was advanced to justify the utilization of any of these motels.

Contrary to this trend, Sirgy (1981a, 1981b) develop !t a congruence model based on a self-image/product-image congruence theory (Sirgy, 1981c). One version of the model was tested and the results provided support for the model (Sirgy, 1981a, 1981b). The following section will present a brief exposition of this theoretically-based congruence motel. For more theoretical detail, the reader may consult Sirgy (1981a, 1981b).

AN INTERACTIVE SELF-CONCEPT CONGRUENCE MOTEL

The interactive self-concept congruence model presented here is based on the self-esteem motive. The interaction between a product-image and a self-image is argued to elicit self-esteem motivation. This interaction results into at least four discrete conditions, namely positive self-congruity, positive self-incongruity, negative self- congruity, and negative self-incongruity.

Positive self-congruity occurs when a positively valued self-image matches that of a positively valued product image. For example, "this car seems to have an image of social outgoingness and dominance" can match the consumer's self-image as being "socially outgoing and dominant." In this case, the theory predicts that the individual will be motivated to approach the product to maintain a desirable level of self-esteem.

Positive self-incongruity occurs when a negatively valued self-image is compared with a corresponding positive product-image. This would induce a high motivational state to approach that product. This is mainly because approaching that product becomes instrumental in approaching an unmet ideal image and therefore increases the person's self-esteem.

Negative self-congruity comes about when a negative product-image is matched against a negative self-image. Here, the person would not be motivated to maintain a state which he/she views in a negative light, since by doing so his/her self-esteem would decrease.

Finally, negative self-incongruity refers to the comparison between a negative product-image and a positive self-image. In this situation, the theory argues that the consumer would avoid the product, since the product does not serve to enhance or maintain his/her self-esteem and in fact may threaten his/her self-esteem given purchase and usage.

The degree of self-esteem motivation activated in relation to a particular product is argued to a primary determinant of purchase motivation (e.g., product preference and purchase intention) toward that product.

A continuous mathematical function of the described interactive congruence model can be shown as follows:

PMij = F(EMij) = f (2PIVij-SIVij)   (4)

where

PMij = purchase motivation as induced by image (i) of individual (j).

EMij = self-esteem motivation as induced by image (i) of individual (j).

PIVij = product-image value of image (i) of individual (j).

SIVij = self-image value of image (i) of individual (j).

But produce-image value (PIVij) (i.e., a positive or negative product image) according to the congruence theory is a function of the product-image belief (Pij) and the desirability weight placed on that image as reflected by the ideal self-image (ISij).

PIVij = Pij ISij   (5)

And similarly, self-image value (SIVij) (i.e., a positive or negative self-image) is a function of both actual self-image (ASij) and ideal self-image (ISij).

SIVij = ASij ISij   (6)

Therefore,

PMij = f (EMij) = f (2Pij ISij - ASij ISij)   (7)

The constant "2" is introduced in the function to support the theoretical argument that a positive congruity state (positive self-congruity) contributes to a different level of self-esteem motivation than a negative congruity state (negative self-congruity). Positive self-congruity is argued to induce a moderate level of positive (approach motivation) whereas negative self-congruity elicits a moderate level of negative (avoidance) motivation. Without this constant, this theoretical distinction would not be reflected.

In factored form, we have:

PMij = f (EMij) = f (2Pij - ASij) ISij   (8)

And summing across all activatable images, the interactive model can be mathematically represented as:

EQUATION    (9)

PURPOSE

The purpose of this study is to compare the predictive strength of the Interactive model with those traditional models such as the Absolute Difference model, the Simple

Difference model, the Difference Squared model, the Euclidean Distance model, and the Divisional motel. These are mathematically represented as follows: n

EQUATION  (10) through (15)

METHOD

Subjects

One hundred and sixty-eight female students enrolled in undergraduate psychology and consumer behavior classes at the University of Massachusetts and Virginia Tech were recruited as subjects. Since this is an explanatory (theoretical) study and not a descriptive one, the use of students and convenience sampling is justified in the context of this stud;

Products

Two brands of automobiles (MGB and VW RABBIT) and two brands of magazines (PLAYGIRL and GLAMOUR) were used in this study.

Preliminary Procedure

In an effort to obtain a highly consensual set of images associated with each of the designated automobiles and magazines that would be as salient as possible for the population under consideration, 23 female subjects in an independent sample set were asked prior to the development of the questionnaire what characteristic images and stereotypes they thought would be associated with driving each of the designated automobiles and with reading each of the designated magazines. Subjects' responses were subjected to a content-analysis procedure and those images which were found to be highly consensual were presented in the final questionnaire (30 images or attributes).

Procedure

The 168 female subjects were run in groups of 5-15. The products were displayed in pictures removed from magazines and posted on 8 1/2" x 11 1/2" white bond paper. The subjects were instructed to look at these products and to become familiar with them before proceeding to respond to the questionnaire.

Questionnaire

The questionnaire was divided in two parts. Part 1 included questions concerning product preferences and purchase intentions for each product. Product preference was measured on a 5-point rating scale varying from very-much-dislike to very-much-like using the following question

TABLE 1

BETA WEIGHTS OF THE VARIOUS CONGRUENCE MODELS IN RELATION TO PRODUCT PREFERENCE AND PURCHASE INTENTION

To what extent to you like ___, or to what extent does it appeal to you? (Note that the question is about liking not buying ___).

Purchase intention was measured on a 5-point rating scale varying from definitely-would-intend-to-buy-it to definitely-would-not-intent-to-buy-it. The precise question was:

Suppose that you have become aware of the need to buy ___, and suppose that you can reasonably afford ___ of your choice, would you intent to buy ___ in the near future.

Part 2 included questions on product images, actual self-images, and ideal self-images, respectively. The personality adjectives which were elicited from the preliminary procedure were used in a unipolar-type semantic differential format to measure these variables. Product images pertaining to PLAYGIRL, MGB, GLAMOUR, and VW RABBIT were independently assessed by the personality adjectives using 5-point likelihood-type scales.

A representative question used to elicit product image responses for a specific product is:

Imagine yourself driving or owning a MGB automobile. What kind of image to you think others would have of you driving or owning this car? For example, if I imagine myself driving or owning a CADILLAC, the kind of image others would have of me would be that of being wealthy, upper class, powerful, and dominant. Now driving or owning a MGB automobile may elicit a certain type of image. Describe this image by checking the likelihood of the personal characteristics listed below:

The actual self-image variable was measured using 5-point likelihood type scales ranging from very-much unlike me to very-much-like-me using the following question:

How to you see yourself? To what extent do you think of yourself as having the following personal characteristics listed below?

The ideal self-image variable was measured using a 5-point desirability-type scale ranging from very-much-dislike to very-much-like using the following question:

How would you ideally like to see yourself? To what extent would you ideally like to see yourself as having the following personal characteristics listed below?

RESULTS AND DISCUSSION

Product preference (and purchase intention) scores were regressed on each of the different models. The beta weights (or correlation coefficients) of each congruence model is shown in Table 1.

The pattern of correlations as shown in Table 1 point to the fact that overall the Interactive congruence model performed only slightly better than the Absolute Difference, Simple Difference, Difference Squared, Euclidean Distance, and Divisional models.

To compare the relative predictive validity of the Interactive model against the other models, incremental F-tests were performed. [e.g./ F-incremental (see Kerlinger and Pedhazur. 1973)

EQUATION

where

I = Interactive congruence scores.

DS = Difference Squared congruence scores.

N = Number of total observations,

k1 = Number of independent variables in the reduced model.

k2 = Number of independent variables in the full model.] Product preference (and purchase intention) scores were regressed with one of the traditional congruence model scores (restricted model) and then with both the traditional and the Interactive congruence models scores (full model). In each case, it was expected that the incremental R2 would be significant. In other words, the variance added by including the Interactive congruence model with the traditional model would be significant, and therefore attesting to the predictive strength of the Interactive model.

In this light, 80 incremental F-tests were performed through a step-wise regression procedure. The results revealed that only 47 out of the 80 tests showed that the Interactive congruence model to be more significantly predictive (p < .05) than the Absolute Difference, Simple Difference, Difference Squared, Euclidean Distance, and Divisional models.

Although these results provide little support for the predictive validity of the theoretically-based Interactive congruence model relative to its traditional counterparts, it is not clear whether differences in the model's predictive performance was due to construct validity or product and model variation. It should also be noted that the reliability and validity of the measures used in this study are unchecked and may be a source for error. Future research is underway to investigate these issues.

REFERENCES

Birdwell, A. E. (1968), " study of the influence of image congruence on consumer choice." Journal of Business, 41, 76-78.

Delozier, W. & Tillman, R., (1972), "Self-image concepts: Can they be used to design marketing programs." Southern Journal of Business, 7, 9-15.

Delozier, W., (1974), "A longitudinal study of the relationship between self-image and brand image." Unpublished doctoral dissertation. University of North Carolina at Chapel Hill.

Dolich, I. J., (1969), "Congruence relationships between self-images and product-brands." Journal of Marketing Research Society, 6, 80-84.

Green, P. E., Maheshwari, P., and Rao, V. R., (1969), "Self-concept and brand preference: An empirical application of multi-dimensional scaling." Journal of Market Research Society, 11, 343-360.

Hughes, G. D. & Naert, P. A., (1970), "A computer-controlled experiment in consumer behavior." Journal of Business, 43, 354-372.

Kerlinger, F. N. & Pedhazur, E. J., (1973), Multiple Regression in Behavioral Research, N. Y.: Holt, Rinehart & Winston.

Maheshwari, A. D., (1974), Self-product image congruence: "A Micro-level Analysis. Ann Arbor, Michigan, University Micro-film.

Rosa, E., (1971), "Self-concept and brand preference." Journal of Business. 44, 38-50.

Schewe, C. D. & Dillon, W. R., (1978), "Marketing information system utilization: An application of self-concept theory." Journal of Business, 6, 67-79.

Sirgy, M. J., (1979), "Self-concePt in consumer behavior." Doctoral dissertation. University of Massachusetts, Amherst.

Sirgy, M. J., "Self-concept in consumer behavior: A critical review and theoretical extensions." Journal of Consumer Research (forthcoming).

Sirgy, M. J., (1980), "Self-concept in relation to product preference and purchase intention." In D. Ashton (ed.), Development in Marketing Science, vol. 3. Academy of Marketing Science.

Sirgy, M. J., (1981a) "Testing a consumer self-concept model using tangible products." Paper presented to the American Psychology Association Convention, Consumer Psychology Division, Los Angeles.

Sirgy, M. J. (1980a), "An interactive model of selfconcept/ product-image self-concept congruity: Predicting product preference." Working paper, Department of Marketing, Virginia Tech, Blacksburg.

Sirgy, M. J. (1981), "Introducing A "Self-Theory" to Consumer Personality Research." ISAS. Catalog of Selected Documents in Psychology, 11, 33, Ms. 2250.

Wylie, R. C., (1974), The self-concept: A review of the methodological considerations and measuring instruments. Lincoln: University of Nebraska Press.

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