External Variables and the Fishbeinmodel: Mediation, Moderation, Or Direct Effects?

Lawrence A. Crosby, University of Michigan
Darrel D. Muehling, University of Nebraska
ABSTRACT - Contrary to the Theory of Reasoned Action, external variables directly impacted intentions, were only partially mediated by the attitudinal and normative components, and had no moderating role. Among college students, intentions to attend arts events were predicted by attitude toward the act, normative structure, self-concept, past behavior, awareness, arts interest. and demographics.
[ to cite ]:
Lawrence A. Crosby and Darrel D. Muehling (1983) ,"External Variables and the Fishbeinmodel: Mediation, Moderation, Or Direct Effects?", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 94-99.

Advances in Consumer Research Volume 10, 1983      Pages 94-99

EXTERNAL VARIABLES AND THE FISHBEINMODEL: MEDIATION, MODERATION, OR DIRECT EFFECTS?

Lawrence A. Crosby, University of Michigan

Darrel D. Muehling, University of Nebraska

ABSTRACT -

Contrary to the Theory of Reasoned Action, external variables directly impacted intentions, were only partially mediated by the attitudinal and normative components, and had no moderating role. Among college students, intentions to attend arts events were predicted by attitude toward the act, normative structure, self-concept, past behavior, awareness, arts interest. and demographics.

INTRODUCTION

In an attempt to provide an all-encompassing explanation of behavior Ajzen and Fishbein have recently elaborated on the Theory of Reasoned Action (Ajzen and Fishbein 1977; 1980; Fishbein and Ajzen 1975). These extensions sake the theory more general and testable in terms of its application to consumer behavior. In particular, the theory specifies distinct roles for other variables that have been used to explain and predict consumer behavior, namely demographics, attitudes toward targets, and personality traits. These variables are designated as "external and are said to influence behavioral intention in one of two ways (only): through the mediating effect of the normative and attitudinal components and/or by moderating the relationship between the normative and attitudinal components and behavioral intention. This is true with the exception of situation variables (Belk 1975; Kakkar and Lutz 1981) which are controlled by specifying the action, target, time, and context of the measures.

The purpose of this paper is to test the prescribed role of external variables in a consumer behavior context. Application is mate using (college) student attendance at performing arts events as the behavior of interest. Though numerous studies dealing with performing arts attendance have been conducted (Andreasen and Belk 1980; Belk and Andreasen 1979; Currim, Weinberg, and Wittink 1981; DiMaggio, Useem, and Brown 1978; Semenik and Young 1979; Sexton and Britney 1979) few have investigated this behavior within Fishbein's conceptual framework. This study expands the framework to include other variables relevant to performing arts situations.

THEORETICAL PERSPECTIVE

The Theory of Reasoned Action views a person's behavior as following logically and systematically from whatever information he happens to have available (Ajzen and Fishbein 1980, p. 244). The theory posits that people consider the implications of their actions before deciding to engage in a behavior. Various kinds of behavior can be accounted for by incorporating a relatively small number of concepts embedded within this single theoretical framework.

Figure 1 provides a graphical representation of the Theory of Reasoned Action. If prediction of behavior is the objective, Ajzen and Fishbein claim that only a measure of behavioral intention is needed. If understanding the behavior is the research objective, then the determinants of behavioral intention must be investigated. The immediate determinants, according to the theory, are attitude toward the behavior and subjective norm (normative expectations of various important others in the social environment).

The theory assumes that the relative importance of the attitudinal and normative factors is dependent upon the nature of the behavior under investigation. Therefore, for some intentions either the attitudinal or the normative component may predominate. In other situations beth may be equally important. The explanatory value of the theory is increased by the empirical assignment of relative weights to the two determinants of intention.

Ajzen and Fishbein note that in many applications attitude and subjective norm provide a sufficient explanation of behavioral intention. If a more complete explanation of behavioral intention is sought, the determinants of the attitudinal and normative factors must be investigated. The immediate determinant of attitude is the cognitive structure. The immediate determinant of subjective norm is the normative structure.

Ajzen and Fishbein relegate demographics, attitudes toward targets (e.g. people, objects, institutions), and personality characteristics to the role of external variables. The influences of these variables are indicated as dashed lines in Figure l. An integral part of the theory is that external variables are completely mediated by the set of endogenous variables or serve to moderate the influence of the endogenous variables on behavioral intention. In statistical terms, this implies that (a) the causal paths linking the external variables to behavioral intention will be entirely indirect, and/or (b) the external variables will interact with the endogenous variables to affect behavioral intention. Presumably, this argument would also hold for any other type of external variable including, for example, information and past experience.

By providing such an explicit statement of their theory, Ajzen and Fishbein make empirical verification possible. This is important since other, competing i theories and research findings suggest a much different role for the (so-called) external variables. Despite their claim of an all-encompassing explanation of behavior, the possibility exists that some or all of these variables may also exert a direct effect on behavioral intention for some people in some situations. Since empirical evidence indicates that external variables are relevant to attendance at performing arts events, there is both theoretical and practical value in understanding the exact nature of their influence.

THEORETICAL EXTENSIONS AND HYPOTHESES

Self-concept would most logically fall within the personality category of external variables proposed by Ajzen and Fishbein (see Figure 1). However, Triandis (1977) offers a competing view that a person's self-concept may, at times, be inconsistent with attitude and thus have an independent influence on behavioral intention. Empirical evidence indicates that the self-concept is a determinant of consumer behavior (e.g. Dolich and Shilling 1969; Green, Maheshwari and Rao 1979; Grubb and Stern 1971; Ross 1971; Landon 1974). It is reasonable to expect that measures of image congruence would be relevant to entertainment choice given the highly segmented appeal and stereotypes which surround the choice alternatives. The symbolic nature of entertainment choice involving the fine arts was noted by Levy (1959).

FIGURE 1

THE THEORY OF REASONED ACTION (ADAPTED FROM AJZEN AND FISHBEIN 1980)

H1: A measure of self-concept will have a direct effect, as well as mediating and moderating effects, on intentions to attend a performing arts event.

Past behavior, considered an "other variable" in the Figure 1 diagram, has for years been considered the best predictor of future acts by behavioral theorists (e.g., Mischel 1968). Recently, Bentler and Speckart (1979) proposed that past behavior may have a direct impact on intentions and subsequent behavior. They argue that attitudes are a partial self-generated inference from behavior, but not a perfect reflection, and that past behavior has an independent role in the prediction of future behavior. These ideas are founded in the concepts of learning and reinforcement theory. Consumer researchers such as Bagozzi (1981) have reached similar conclusions. Bagozzi finds that the effect of including past behavior is to attenuate the impact of attitudes on intentions. This is predicted from the work of Landis, Triandis, and Adamopoulos (1978) who argue that the habit component increases in importance as a behavior is repeated. For these reasons a relationship between past behavior and behavioral intention in a performing arts context is hypothesized.

H2: A measure of past behavior will have a direct effect, as well as mediating and moderating effects, on intentions to attend a performing arts event.

In Figure 1, "awareness" would be categorized as an "other external variable. It is possible that under low involvement, a condition hypothesized by Krugman (1965) and tested by consumer researchers (Ray 1973; Tyebjee 1979), awareness may be the direct antecedent of behavior, without the usual mediating effect of attitude. This might also occur under high involvement, if product performance judgments are deemed a matter of taste and trial seems to provide the best source of product information. Since some consumers are likely to approach attendance as a low involvement decision while others may trust only their own judgment on matters of entertainment, a direct link may exist in the performing arts context.

H3: A measure of awareness will have a direct effect, as well as mediating and moderating effects, on intentions to attend a performing arts event.

A generalized attitude toward the performing arts falls within the external variable category "attitudes toward targets". Earlier theorizing by Fishbein (1963) addressed the explanatory value of an "attitude toward the target' measure. Since a behavioral measure of attitude is a better predictor, the utility of other attitude measures has been discounted in recent years. However, for those who lack specific information about choice alternatives, a generalized response based on attitudes toward the class of objects may be predictive of behavioral intention. In a recent consumer research study (Andreasen and Belk 1980), general attitude measures (interest in theater or classical music) were significantly related to the likelihood of attendance at performing-arts events.

H4: A measure will have mediating intention event of attitude toward the target a direct effect, as well as and moderating effects, on to attend a performing arts

Finally, various demographics have been included to represent this category of external variables. Accepting the premise that demographics to not directly "cause" behavior, the possibility still exists that some intervening variable, the nature of which is unknown, is missing from the model. A direct link from measures of demographics to behavioral intention would indicate model misspecification (Pindyck and Rubinfeld 1981, p. 128).

H5: Demographics will have direct effects, as well as mediating and moderating effects, on intentions to attend a performing arts event.

METHOD

Sample

Data were obtained from students enrolled in fall classes at a large Midwestern university in 1980. To obtain a probability sample, 21 classes (i.e. clusters) were systematically selected from the university course schedule and questionnaires were distributed to all students in each class. Questionnaires were returned by 293 students which represented a 50% response rate. A comparison with school records indicated the sample was very representative of the university population.

Further precision was obtained by assigning weights to individual cases based on a cross-classification of major, sex, and year in school. The analysis is based on 164 cases with complete data on all the variables. Case-wise deletion did not appear to have a biasing effect on any of the variable means or variances.

Measures

Multiple measures were taken on most of the constructs under investigation. Indices were formed by summing responses to the theoretically relevant items. All indices formed in this way had acceptable standardized item alphas for basic research (Nunnally 1978).

Endogenous Variables, To capture the attitudinal and normative influences affecting behavioral intentions, this study employed Form 3 (Ryan and Bonfield 1975) of the Fishbein Extended Model. This formulation is:

EQUATION    (1)

where B = overt behavior; BI = behavioral intention; Aact = attitude toward the behavior; NBj = degree of belief or disbelief that a specific act is expected of the individual by the jth person or group; MCj = the individual's motivation to comply or not comply with the expectation of the jth person or group; and Wo and W1 are empirically determined weights. Measures were obtained for all the constructs in (1) except B.

This version of the Fishbein Model has considerable empirical support (Ajzen and Fishbein 1973; Ryan and Bonfield 1975) and is often used in consumer research. However, it deviates from Figure 1 in two respects. First, the model operationalizes only part of the Theory of Reasoned Action. Specifically excluded endogenous variables are behavioral beliefs, outcome evaluations, and subjective nor;a. Second, the model mixes causal levels in the theoretical network. Normative structure (SNBjMCj) is of a higher causal order than attitude toward the behavior (Aact). These differences, while important to note, should not seriously impair the hypothesis-testing aims of the study. According to Figure 1, any indirect influence of the external variables must at some point pass through either attitude toward the act or normative structure and should, therefore, be fully captured by these measures. Also, a recent application of structural equation methodology which considered all of the endogenous variables found that subjective norm is strictly a function of the normative structure index (Ryan 1981). Therefore, possible moderating effects due to the external variables should be observed regardless of whether subjective norm or the normative structure index is used as the measure of social influence.

An index of behavioral intention was created which corresponds to the notion of a behavioral category (Ajzen and Fishbein 1980). Students were asked to indicate the probability that they would attend each of eight performing arts events scheduled for the remainder of the school year. The eight responses were then summed. Despite the varied nature of these performances, from drama to chamber music, the resulting index was found to be internally consistent (alpha =.86). It should be noted that behavioral categories are typically somewhat general in terms of action, target, time, and context.

The measurement of attitude toward the act followed the procedures recommended by Ajzen and Fishbein. Students rated future attendance behavior on four a-scales (Fishbein and Raven 1962). Item responses were coded +3 to -3 and summed, providing an index with high reliability (alpha = .92). The normative structure index was formed by multiplying normative beliefs by the corresponding motivation to comply, then summing across salient referents (e.g. teachers, classmates, friends, etc.). Referents were identified through group discussion with students. Normative beliefs and motivation to comply were operationalized with single iter measures for each referent, again following the recommended procedures. Responses to the NBj and NC1 items were coded +3 to -3 as is the convention. The reliability of the normative index was .67. This lower alpha level was expected since the index is a composite of diverse referents.

Recognizing that a lack of correspondence between measures can limit the predictive and explanatory value of the model, the equivalence of the behavioral intention, attitudinal, and normative measures was analyzed. As can be seen in Table 1, the correspondence requirement is met in terms of action and context but appears to he somewhat violated for the target and time characteristics.

TABLE 1

CORROSPONDENCE BETWEEN MODEL COMPONENTS

Knowledge of the situation suggests the difference in targets is relatively minor. The "eight major performing arts events" that constitute the intentions measure were very representative of the Center's offerings. They were also part of a highly publicized program that gives the Center its identity and contributes much of its patronage. It is most probable that students responded to the attitudinal and normative influence questions either in terms of this specific program or the types and caliber of events it represents. Likewise, though the time element in seemingly more specific for the behavioral intentions measure, the nature of respondents' planning decisions must be taken into account. For entertainment decisions concerning University programs, students are likely to equate future with the coming school year since that is the planning horizon for most school activities.

Taken together, these considerations suggest there is sufficient correspondence so as not to attenuate the findings. Empirical evidence is that for the regression of behavioral intentions on attitude and normative structure, R = .53 which compares favorably with other tests of the Form 3 Fishbein Model (Ryan and Bonfield 1975). It might be added that, to the extent differences exist, the bias is against finding direct effects of the external variables on behavioral intentions. The external variables are most general, attitude and normative structure are more specific, and behavioral intentions is most specific. Therefore, if any of the relationships are likely to be attenuated for correspondence reasons it would be those involving the external variables and intentions.

Exogeneous Variables. A self-concept variable measured the congruence of the student's (actual) self image with the perception of the type of person who attends performing arts events at the University Center. Images were compared on nine, 9-point semantic differential scales (e.g. idealistic vs. practical). Scales were selected through group discussions with students as being personality dimension relevant to entertainment choice. The Osgood-Suci D2 statistic was used as the measure of profile similarity. Higher scores indicate more incongruity. Reliability on this index was .75.

Factor analysis of the set of 13 awareness items indicated three orthogonal dimensions. These were interpreted as: awareness of the Center's pricing and discount policies, awareness that some events are perform.ed by internationally known touring artists, and awareness that other events are performed by students and faculty. Indices were formed by computing factor scores for each respondent on each of the three awareness dimensions.

Another index was created to measure the student's general interest in the performing arts or 'attitude toward a target.- Ten different forms of entertainment were rated on 7-point, "like-dislike" scales. Five were arts-related (e.g. art, classical music) and five were taken from the popular culture (e.g. collegiate sports, rock music). Ratings of the "arts-related" items were sunned as were the ratings of the "non arts-related" items. These summations were used to form a ratio of arts to nonarts ratings - the larger the quotient, the greater the interest in performing arts relative to other entertainment forms.

All other variables in the study involved single item scales. These included sex (male/female), major (business or engineering/all others), age, whether the student was enrolled at the University the previous year (yes/no), and number of fine arts performances attended anywhere in the last three years.

ANALYSIS AND RESULTS

It was determined that a hierarchical regression would provide a simultaneous test of all the hypotheses (Cohen and Cohen 1975). The logic behind this analysis is as follows:

1. if the Fishbein Model variables of attitude toward the act and normative structure completely mediate the effects of the (so-called) external variables, then the inclusion of the external variables should add nothing to the variance explained in behavioral intentions unless...

2. the external variables moderate the effects of the Fishbein Model variables. In that case, significant interactions should exist between the external and model variables.

The starting point for this hierarchical regression is the 'complete' model which includes all potential main and interaction effect terms. The complete model is then compared to a "reduced" model containing only the main effect terms. The hierarchical F statistic is then computed to determine if any of the interactions are significant. In the present study, all two-way interactions of external with model variables were considered. The regression results for this stage of the analysis appear in Table 2.

As the results indicate, the R2 improvement after including the interactions is not significant. The difference in the R2 values for the two models can be attributed to differences in the degrees of freedom. Since an issue can be raised whether the hierarchical test is overly conservative, it might be noted that the R difference obtained is non-significant at virtually any alpha level. Other statistical procedures that are less conservative were also used to identify possible interactions. These included a stepwise regression of the interaction variables after the main effects were forced to enter and a simple examination or the betas for the interaction terms in the complete model. These procedures failed to reveal any interactions significant at the .z)5 level. The conclusion, therefore, is that none of these particular external variables exhibit the hypothesized moderating effects.

TABLE 2

TEST OF THE MODERATING ROLE OF THE EXTERNAL VARIABLES

The next phase of the hierarchical regression concerns the mediating role of attitude and social norm. As would be evident from path analysis, a regression of behavioral intention on attitude, normative structure, and the external variables is sufficient to determine whether the Fishbein Model variables completely mediate the external influences.- If the mediation is complete, the external variables will not appear as significant predictors in the regression (i.e. will have no direct effect on behavioral intentions). The needed regression is simply the reduced model of Table 2. The detailed results for this model are summarized in Table 3.

TABLE 3

REGRESSION OF EXTERNAL VARIABLES AND FISHBEIN MODEL VARIABLES ON BEHAVIORAL INTENTION

As suggested by Ajzen and Fishbein, behavioral intention is dependent on attitude toward the act and normative influence. However, the role of the external variables does not appear as they proposed. In fact, all categories of external variables were found to have a direct influence on behavioral intention. External variables having significant direct effects on behavioral intention included self-concept, past behavior, awareness of pricing, age, and interest in performing arts. The conclusion, therefore, is that the Fishbein Model variables do not completely mediate the influence of the external variables.

To test whether the Fishbein Model variables have any mediating effect, separate regressions of attitude and normative structure on the external variables were performed. The results appear in Table 4. Based on these regression results, it appears that the effects of past behavior, awareness of pricing, and interest in performing arts are partially mediated by attitude toward the act. Likewise, age and interest in performing arts are partially mediated by normative structure. Only in the case of the variable student last year is the mediating effect of normative structure complete.

In terms of the specific hypotheses, none are entirely supported by the findings. Moderating effects were not observed for any of the external variables. Mediating effects were observed for some of the external variables but not all. However, the hypothesized direct effects were observed in all external variable categories. The conclusion, therefore, is that the two-component Fishbein Model (while parsimonious) does not adequately describe the multiple influences on behavioral intention.

DISCUSSION

When viewed in the context of arts attendance behavior the findings of this study raise some questions about the Theory of Reasoned Action. Though Ajzen and Fishbein's model significantly explains behavioral intention through measures of attitude and social influence, all classes of external variables were found to have direct effects. However, it is not our intent to advocate a further extension of the already extended Fishbein Model. Rather, our point is that further theoretical refinement is needed to specify when behavior is primarily under attitude/normative control or when other variables are more important. We would argue too for a revitalization of some traditional consumer behavior concepts (e.g. self-concept, habit) often ignored during the decade-long domination of the multiattribute models. We also support the trend in research which recognizes that decision making differs by level of involvement and that trial can be an important source of consumer information.

Despite the Model's limitations, Ajzen and Fishbein's approach to theory development is to be commended. Consumer behavior needs to adopt a similar approach to become a cumulative science. A few simple ideas described in a parsimonious model can be readily tested. The Theory of Reasoned Action is one explanation; what is needed are competing explanations of the same phenomena that provide different predictions and can empirically challenge the established viewpoint.

From an applied perspective, the findings contain implications for marketing the performing arts. For example, the self-concept measure was negatively related to behavioral intention, i.e., as a student's self-image becomes less congruent with his image of fine arts attenders, the likelihood of attendance is reduced. No mediating or moderating influences were observed. This confirms the hypothesis that attendance is a symbolic act with social and personal meanings for these consumers. To increase image congruence marketers could communicate to potential attenders the heterogeneous composition of performing arts audiences.

The influence of past behavior on intentions was partially mediated by an attitude toward the act. Attendance at past performing arts events leads to a more positive attitude about attending future events.

TABLE 4

TEST OF THE MEDIATING ROLE OF THE FISHBEIN MODEL VARIABLES

As behavior is repeated the effect of attitude on intention is attenuated and a direct positive influence between past and future behavior emerges. This suggests that a successful marketing strategy would include frequent reminders to past attenders, quantity discounts via season tickets, convenient scheduling and parking arrangements, ticket sales for future events during intermission and immediately after performances, and other tactics appropriate for habitual buying.

An increased awareness of pricing and discount policies was associated with a more positive attitude. In addition, the findings suggest awareness can directly influence behavioral intention. This might imply that, for some, attendance is a low involvement decision. If true, marketing tactics aimed at inducing product trial may have application to the arts, e.g. simple but frequent messages, opportunities for free sampling, wide distribution of tickets, coupons and deals, and the use of symbols, imagery, and visuals in promotion.

Interest in the performing arts positively influenced both the attitude and normative components. But, interest also had a direct influence on behavioral intention. Individuals with a general interest in the performing arts were more likely to attend specific arts events. Especially in situations where no specific attitudes or norms are present, a general attitude (interest in the arts) may explain why students plan to attend an event. Arts interest is believed to be a product of one's up-bringing and family socialization. Those who score very low on this variable represent "hard-core nonattenders" (Andreasen and Belk 1980) and very little promotional effort should be wasted on them. Recognizing the importance of interest, proposals have been made to introduce young children to the arts (Andreasen and Belk 1980). At the very least, a concerted effort should be made during the college years to expose all students to the arts through course requirements or innovative programs for marketing the arts on campus.

The normative component totally mediated the effect of "being a student last year" on intentions to attend an arts event. Students more familiar with the university environment were negatively influenced by these norms. This finding may be peculiar to the University where the study was conducted. More likely, it implies that norms exist which constrain college students from attending arts events. Strategies to make attendance more socially acceptable might involve the identification of opinion leaders to participate in various arts center activities.

Age, though partially mediated by the normative component, was also found to have a direct link to intentions. As a student ages he feels that important others will regard negatively his intention to attend a performing arts event (ref: previous paragraph). The direct negative relationship between age and behavioral intention can be interpreted as either the absence of some endogenous variable or a possible suppression effect.

REFERENCES

Ajzen, Icek, and Fishbein, Martin (1973), Attitudinal and Normative Variables as Predictors of Specific Behaviors, Journal of Personality and Social Psychology, 27, 51-57.

Ajzen, Icek, and Fishbein, Martin (1977), "Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research," Psychological Bulletin, 84, 888-918.

Ajzen, Icek, and Fishbein, Martin (1980), Understanding Attitudes and Predicting Social Behavior, Englewood Cliffs, New Jersey: Prentice-Hall, Inc.

Andreasen, Alan R., and Belk, Russell W. (1980), Predictors of Attendance at the Performing Arts, Journal of Consumer Research, 7, 112-120.

Bagozzi, Richard P. (1981), Attitudes, Intentions, and Behavior: A Test of Some Key Hypotheses," Journal of Personality and Social Psychology, 41, 607-627.

Belk, Russell W. (1975) "Situational Variables and Consumer Behavior," Journal of Consumer Research, 2, 157-163

Belk, Russell W., and Andreasen, Alan R. (1979), "De Cuscibus Non Ese Disputandum: A Study of the Potential for broadening the Appeal of Performing Arts, in Advances in Consumer Research, cd. Jerry C. Olson, 7, 109-113.

Bentler, P. M., and Speckart, George (1979), "'Models of Attitude - Behavior Relations," Psychological Review, 86, 452-404.

Currim, Imran S., Weinberg, Charles B., and Wittink, Dick R., (1981), "Design of Subscription Programs for a Performing Arts Series," Journal of Consumer Research, 8, 6,-75.

Cohen, Jacob, and Cohen, Patricia (1975), Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences, Hillsdale, New Jersey; Lawrence Erlbaum Associates Publishers.

DiMaggio, Paul, Useem, Michael and Brown, Paul (1978), Audience Studies of the Performing Arts and Museums: A Critical Review, Washington, D.C.: National Endowment for the Arts.

Dolich, Ira J., and Shilling, Neb (1969), "Congruence Relationships Between Self-Images and Product Brands," Journal or Marketing Research, 6, 80-84.

Fishbein, Martin (1963) "An Investigation of the Relationship Between Beliefs About An Object and the Attitude toward that Object," Human Relations, 16, 233-240.

Fishbein, Martin, and Ajzen, Icek (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, Mass: Addison-Wesley Publishing CompanY.

Fishbein, Martin, and Raven, Bertram, H. (1962), "The AB Scales: An Operational Definition of Belief and Attitude,' Human Relations, 15, 35-44.

Green, Paul E., Makeshwari, Arun, and Rao, V. R. (1969), Self-Concept and Brand Preference: An Empirical Application of Multidimensional Scaling," Journal of the Marketing Research Society, 11, 343-360.

Grubb, Edward I., and Stern, Bruce L. (1971), "Self-Concept and Significant Others," Journal of Marketing Research, 8, 382-385.

Kakkar, Pradeep, and Lutz, Richard J. (1931) "Situational Influences on Consumer Behavior: A Review," in Perspectives in Consumer Behavior, 3rd Edition, eds. Harold H. Kassarjian and Thomas S. Robertson, Glenview, Ill: Scott, Foresman and Company.

Krugman, Herbert E., (1965), 'The Impact of Television Advertising: Learning Without Involvement, Public Opinion Quarterly, 29, 349-356.

Landis, D., Triandis, H. C., and Adamopoulos, J. (1978), "Habit and Behavioral Intentions as Predictors of Social Behavior," A Journal of Social Psychology, 106, 297-237

Landon, E. Laird, Jr. (1974), "Self Concept, Ideal Self L Concept, and Consumer Purchase Intentions." Journal of Consumer Research, 1, 44-51.

Levy, Sidney J. (1959), "Symbols for Sale", Harvard Business Review, 117-124.

Mischel, Walter (1968) Personality and Assessment, New York: Wiley.

Nunnally, Jum C. (1978) Psychometric Theory, New York: McGraw-Hill.

Pindyck, Robert S., and Rubinfeld, Daniel L. (1981), Econometric Models and Economic Forecasts, 2nd Edition, New York: McGraw-Hill Book Company.

Ray, Michael L. (1973), Marketing Communication and the Hierarchy of Effects," Working Paper, Marketing Science Institute

Ross, Ivan (1971), "Self Concept and Brand Preference,' Journal of Business, 44, 38-SO.

Ryan, Michael J. (1981) "Behavioral Intention Formation: The Interdependency or Attitudinal and Social Influence Variables," Working Paper, Division of Research, Graduate School of Business Administration, University of Michigan.

Semenik, Richard J., and Young, Clifford E. (1979), 'Correlates of Season Ticket Subscription Behavior, in Advances in Consumer Research, ed. Jerry C. Olson, 7, 114-118.

Sexton, Donald E., and Britney, Kathryn (1979) "A Behavioral Segmentation or the Arts Market," in Advances in Consumer Research, ed. Jerry C. Olson. 7. 119-120.

Triandis, Harry C. (1977) Interpersonal Behavior, Monterey, California: Brooks-Cole Publishing Company.

Tyebjee, Tyzoon T. (1979), Response Time, Conflict and Involvement in Brand Choice, Journal of Consumer Research, 6, 295-304.

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