An Examination of an Alternative Form of the Behavioral Intention Model's Normative Component

Michael J. Ryan, Columbia University
ABSTRACT - An argument for expanding Fishbein's Behavioral Intention Model that posits non-additivity, the introduction of a new variable, and restructuring of the model by way of explicitly identifying components as antecedent, independent, and dependent variables is put forth. Evidence providing limited support is described and a modified expansion of the model is suggested as a guideline for future research.
[ to cite ]:
Michael J. Ryan (1978) ,"An Examination of an Alternative Form of the Behavioral Intention Model's Normative Component", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 283-289.

Advances in Consumer Research Volume 5, 1978      Pages 283-289


Michael J. Ryan, Columbia University


An argument for expanding Fishbein's Behavioral Intention Model that posits non-additivity, the introduction of a new variable, and restructuring of the model by way of explicitly identifying components as antecedent, independent, and dependent variables is put forth. Evidence providing limited support is described and a modified expansion of the model is suggested as a guideline for future research.


The popularity of applying and investigating the Fishbein Behavioral Intention or Extended Model within the context of consumer behavior is evidenced in a recent review (Ryan and Bonfield, 1975). More recently, efforts have focused on refinements or extensions of the original model (Bennett and Harrell, 1975; Lutz, 1975; Ahtola, 1975; Glassman and Fitzhenry, 1975). This research examines a portion of a causal chain proposed by Ryan and Bonfield (1975) as a framework for identifying relationships and specifying components as antecedent, independent, moderator, and dependent variables. The major focus regards non-additivity of attitudinal and social influences and the specification of an independent variable form of social influences.


The Fishbein behavioral intention theory states that intention to perform a behavior in a given situation and subsequent performance of that behavior is a function of (1) attitude toward the performance of a specific behavior in a particular situation and (2) social norms influencing the individual's performance of the behavior. The central equation in the theory is commonly modeled as follows:


In previous research, the operational model has taken the form of a standardized regression equation and the magnitude of the beta weights has in turn been interpreted as an indication of the relative strength of attitudinal and social influences.

By combining attitudes and social influences Fishbein attempts to overcome a probable reason for the traditional lack of empirical support for an attitude-behavior relationship. Namely, that perceived social group influence variables must be considered in conjunction with attitudes to predict and understand behavior (Triandis, 1965; Warner and DeFleur, 1969; Wicker, 1969). Although it may be useful to separate these two sources of influence in regard to understanding behavioral intention formation and change strategies, Fishbein does not provide a conceptual basis for the independence stated in the additive model. This is a crucial point since the use of beta weights to indicate the relative influence of the variables on behavioral intention rests upon the additivity assumption. Many studies (Ajzen and Fishbein, 1973; Ryan and Bonfield, 1975) have produced statistically significant w0 and w1 beta weights thus yielding empirical support for the additivity notion. However, other theorists (Banton, 1965; Krech, et al., 1962; Karlins and Abelson, 1970) have argued on a conceptual basis and furnished empirical evidence (Janis and King, 1954; King and Janis, 1956; Siegal and Siegal, 1957; Newcomb, 1958; Sherif, 1958; and Janis and Gilmore, 1965) indicating that attitude and social influences are related. Although these studies did not use expectancy-value models and conceptual differences make their total assessment difficult, this evidence together with the lack of a conceptual foundation, suggests that the additivity notion be questioned.

A Reconceptualization

The research carried out by Fishbein and his associates has usually incorporated a measure of Aact as the attitudinal component instead of SBiai. Since these studies (Ajzen and Fishbein, 1973) were primarily concerned with prediction of behavior and behavioral intentions, this seems appropriate from an operational standpoint. In all cases the measurement of Aact was patterned after the multi-item, bipolar adjective, evaluative dimension of the semantic differential (Osgood, Suci, and Tannenbaum, 1957; Fishbein and Raven, 1962). This technique seems appropriate for predictive purposes since it. is generally considered to produce reliable and valid measures of attitude. However, the bipolar adjective set does not have much explanatory power since it merely indicates the strength and direction of attitude. It does not reveal the elements comprising the attitudinal structure and therefore does not furnish diagnostic power useful in examining attitude formation and change. In those cases where predictive power or variable associations are the sole interest this is appropriate since a determination of structural elements is not necessary, but a reliable and valid measurement is essential. Given a choice, Aact is also easier to operationalize than SBiai since there is no need to determine salient outcomes, encounter problems arising from an erroneous or incomplete list, or run the risk that correlations with multiplied non-ratio scaled scores may be invariant (Schmidt, 1973).

The major thrust of Fishbein's work has been to conceptualize the attitude formation process as the summation of beliefs times evaluations (Fishbein, 1963).The relationship between Aact and SBiai has been supported in a number of studies (Ajzen and Fishbein, 1970; Bruce, 1971; Bright and Stummers, 1971; Jaccard and Davidson, 1972; Ajzen and Fishbein, 1972) and the interaction of Bi and ai manipulations has produced changes in Aact (Lutz, 1975b). Thus, SBiai appears appropriate for diagnosing Aact whereas Aact seems more appropriate than SBiai for predicting behavioral intention. Empirical support for this notion is found in evidence provided by Lutz (1973). When he replaced Aact scores with SBiai scores the multiple correlation on BI dropped from .647 to .240 and the attitudinal beta weight dropped from .646 to .238. This evidence was strengthened when Lutz (1975a) also observed changes in BI following changes in Aact that were induced by manipulating Bi and ai values. In a pilot study for the present research it was found that SBiai -BI correlations disappeared when Aact was partialled out (Ryan, 1974).

Although Fishbein does not use the term, this evidence indicates that SBiai fits the use prescribed to antecedent variables which in this case is attempting to discover a first cause by clarifying the influences which precede the Aact-BI relationship. Thus, SBiai is recast as an antecedent and Aact as an independent variable.

Following the same reasoning used to describe the two forms of attitude as antecedent and independent variables, the formulation of the normative component appears incomplete. The SNBjMCj formulation fits the use ascribed to explanatory variables but an independent form of social influence has not been specified that avoids operational problems and is more appropriate for prediction. In fact, within the context of Fishbein's theory the present understanding of the normative component is limited (Ajzen and Fishbein, 1973; Fishbein, 1975). Consequently, it seemed useful to look elsewhere for a framework within which to develop an independent form of the normative variable. Kelman's (1961) three processes of social influence; compliance, identification, and internalization, were used for this purpose.

Compliance occurs when an individual accepts influence from another person or group because he believes it will result in a satisfying social effect which may be the attainment of rewards or avoidance of punishments under the other's control. Identification occurs when an individual adopts the role of another. It differs from compliance in that the individual is not primarily motivated to please the other but to meet the other's own role expectations since they fit the individual's self image. Internalization occurs when an individual accepts influence because it is congruent with his value system. The reward is the content of the induced behavior.

Since internalization and identification involve rewards internal to the individual, these sources of social influence seem related to attitude. On the other hand, compliance involves an external reward since the individual assumes an actor's role, that is, a role not necessarily congruent with his internal values, in order to obtain an expected reward. Consequently, compliance may well be independent of attitude. However, Kelman holds that all three processes are present and interact in a given social situation and that the nature of the situation determines which particular process predominates and the type of interaction among them. Thus, some relationship between attitude and social compliance is expected with the strength of the relationship determined by the degree to which internalization and identification are present. In the intentions model such effects would be manifested in Aact whereas the social compliance dimension by itself, hereafter referred to as SC, seems independent from Aact. Thus, SC is posited as an independent variable form of social influence that is separable yet, in contrast to the additivity notion, related to Aact.

The notion of a separable influence variable based on external rewards also has some parallel with the development of Dulany's (1968) verbal response model from which Fishbein borrowed heavily in developing the Intention's Model. Dulany manipulated his variables in laboratory experiments that employed negative and positive reinforcers administered by an experimenter after subjects verbally responded to statements that were regarded as neutral in terms of the respondent's value system.

In summary, this reconceptualization views BI as a dependent variable, Aact and SC as independent variables related to each other, and SBiai and SNBjMCj as antecedent variables. Subsequent to the initial work on this reconceptualization (Ryan, 1974), Fishbein and Ajzen (1975, pp. 16, 334, 407) have put forth a similar restructured model with three primary differences. First, attitudinal and social variables continue to be viewed as independently affecting BI. Second, although positing the same general flow of effects, identification of antecedent-independent variable relationships is not made explicit. Third, a new form of social influence is introduced as a generalized social norm (SN) identified as the person's perception that most people who are important to him think he should or should not perform the behavior in question (Fishbein and Ajzen, 1975, p. 302). Evidence supporting an association between SNBjMCj and SN and between SN and BI has been furnished (Glassman and Fitzhenry, 1975) although SN has yet to be shown as separable or independent from Aact. Fishbein (1975) introduced the importance concept based on an assumption that individuals would have little or no motivation to comply with unimportant others. This notion raises questions about both the separability of SN from Aact and the need to include MC in the antecedent variable. First, importance is likely to be based on value Judgments or emotional feelings thereby indicating that "important others" would be internalized and reflected in Aact. Second, if importance guarantees a strong motivation to comply, this removes the necessity for including MC in the normative structure thereby rendering the theory less general since it would exclude motivations not to comply with a perceived social influence. For example, a referent expectation would be negatively related to behavioral intention if the individual was motivated not to comply with a positive influence. In the case of social compliance, such a situation may arise if the individual places little or no value on the expected reward from the referent as a consequence of performing the behavior. This argument parallels the thinking behind the original Fishbein (1967) expectancy times evaluation approach to attitude.

Research Objectives

This research tested the proposed associations among SBiai, SNBjMCj, Aact, SC, and BI, It was not possible to empirically compare SC and SN since the research was executed prior to the appearance of SN. In addition to the value of replication, attitudinal variable relationships were included since it seemed unrealistic to test social influence variables without including their relationships with all other variables in the system. More specifically, support was sought for the following relationships as specified by the reconceptualization.

In terms of the attitudinal variables, attitude toward the act (Aact) should be directly related to its underlying structure (SBiai) and behavioral intention (BI) should be directly related to attitude toward the act (Aact). Behavioral intention (BI) should be indirectly associated with attitudinal underlying structure (SBiai). That is, behavioral intention (BI) should be more strongly associated with attitude toward the act (Aact) than it is with attitudinal underlying structure (SBiai).

In terms of social influence variables, social compliance (SC) should be directly related to its underlying structure (SNBjMCj) and behavioral intention (BI) should be directly related to social compliance (SC). Behavioral intention (BI) should be indirectly associated with social compliance (SC) underlying structure (SNBjMCj). That is, behavioral intention (BI) should be more strongly associated with social compliance (SC) than it is with social underlying structure (SNBjMCj).

Concerning the relationship between attitudinal and social influence variables, attitudinal underlying structure (SBiai) and normative underlying structure (SNBjMCj) should be related, attitudinal underlying structure (SBiai) and social compliance (SC), normative structure (SNBjMCj) and attitude toward the act (Aact) should not be related. Finally, attitude toward the act (Aact) should be related to social compliance (SC).

In addition, a better fit and more meaningful weights should be obtained when the posited independent Aact and SC rather than antecedent variables SBiai and SNBjMCj are used as predictors of BI.


A series of simple and first order partial correlations were employed to test the variable associations following a method promulgated by Blalock (1968). In addition, the posited antecedent and independent variables were placed in the central equation in order to compare R2s and beta weight values where the relative strength of the predictors was assumed a priori.

Behavioral Intention Criterion and Sample. The behavioral intention criterion were young adult purchase intentions toward two major brands of toothpaste, Crest and Ultra Brite. It was a priori expected that since this product is privately purchased and consumed, attitudinal influences should be stronger than social influences for both brands (Bourne, 1957). However, since Ultra Brite promotional activities emphasize social interactions, social influences were expected to be relatively stronger for Ultra Brite than for Crest. The subjects were 97 undergraduate college students enrolled in the business college of a large southeastern university.

Operationalization of the model involved two steps. First, the salient outcomes and referents were identified. Second, instruments to measure the variables were developed.

Salient Outcomes and Referents. Salient outcomes and referents were determined with an elicitation technique (Ryan and Etzel, 1975). The notion behind this technique is not to identify a long list of belief statements but rather to identify only those that determine attitudes and social influence. The obtained free responses were content analyzed and separated into groups based on common meanings. Natural breaks in the frequency of mentioned items were used to separate salient from non-salient items. The salient attitudinal outcomes and normative referents are presented in Table 1.

Structural Components. A set of belief (Bi and NBj), evaluation (ai), and motivation to comply (MCj) statements were constructed for each of the salient outcomes and referents by modifying the semantic differential approach used in the previous pilot study (Ryan, 1974). In order to obtain reliability estimates and avoid bias from adjective specificity, this approach used multiple item scales. The bipolar adjectives used in the Bi and NBj scales involved the addition of two sets of adjectives to the single set commonly used in previous research (c.f., Jaccard and Davidson, 1972). The adjectives used in the ai and MCj scales involved the addition of two sets of adjectives from the Fishbein and Raven (1962) AB scale. In the present study, these instruments were modified by separating the adjectives with separate but similar statements and reversing the scale directions in order to minimize response set bias. For example:

Ultra Brite B2

My purchase and use of Ultra Brite leads to whiter teeth:

improbable _____:_____:_____:_____:_____:_____:_____:probable

My purchase and use of Ultra Brite causes tooth whiteness:

possible _____:_____:_____:_____:_____:_____:_____:impossible

My purchase and use of Ultra Brite provides me with whiter teeth:

likely _____:_____:_____:_____:_____:_____:_____:unlikely

All items were placed in the questionnaire so that none of the items indicating the same construct were adjacent. Bipolar (+3 to -3) summative scoring was used.



Behavioral Intention, Attitude Toward the Aact, and Social Compliance. BI and Aact measures involved similar scaling and modifications as described above. However, in order to obtain operational distinction between Aact and SC, SC was operationalized with a 14 item Likert type, 7 point scale. In line with Kelman's reasoning, the social compliance statements referred to expected social rewards. The 14 statements approximated the following example:

My purchase and use of Ultra Brite because other people expect me to do so will cause these other people to reward me:

Strongly Agree ___: Agree___: Slightly Agree ___: Uncertain___: Slightly Disagree___: Disagree___: Strongly Disagree___:

An agree to disagree scale followed each of the 14 statements, eight of which referred to rewards and six to sanctions.

Reliability estimates, employing Cronbach's (1951) estimate of internal consistency, for all measures met acceptable standards (Nunnally, 1967) ranging from .61 to .98. This indicates that each instrument measured a single construct and suggests that beta weights in the central equation would not be attenuated due to Aact or SC measurement error.

Statistical Procedures. Direct associations were tested with simple correlations and indirect associations with partial correlations holding the proposed moderating variable constant. Correlation coefficient size indicated the magnitude of the association and Fishers Z transformation was used to test for statistical significance at the .05 alpha level. In order to support the proposed relationships, variables posited to have direct associations should produce positive, statistically significant correlations between their respective measurement scores. Indirectly linked variable correlations should lower in magnitude and become statistically insignificant when the moderator variable score is partialled out. Variables posited to have no relationship should produce statistically insignificant correlations between their respective measurement scores.

Further testing of the antecedent-independent variable designation involved an examination of beta weight and R2 values when the respective variables were placed in the central equation. In order to support this notion the model employing Aact and SC should produce larger R2s than those produced using SBiai and SNBjMCj as predictor variables. In addition, it was expected that the Aact attitudinal beta weights would predominate for both brands and that the SC weight would be larger in the Ultra Brite versus the Crest model. Variations from these beta weight expectations when using SBiai and SNBjMCj as predictors would indicate that these variables are not appropriate independent variable forms.

Findings and Discussion

The simple and partial correlations used to test the attitudinal and normative variable associations are presented in Table 2. Correlation Sets i through 4 test attitudinal variable associations. The positive, statistically significant correlations obtained for both brands in Sets 1 and 2 support the direct associations between the attitudinal variables SBiai -Aact and Aact-BI. The statistically significant correlations in Set 3 for the indirectly linked variables SBiai -BI became statistically insignificant (Set 4) when Aact was partialled out. This evidence supports the associations posited among the attitudinal variables.

Correlation Sets 5 through 8 tested normative variable associations. The positive, statistically significant correlations for both brands in Sets 5 and 6 support the direct associations between the social influence variables SNBjMCj -SC and SC-BI. The statistically significant correlations in Set 7 for the indirectly linked variables SNBjMCj -BI became statistically insignificant when SC was partialled out for Ultra Brite but remained significant for Crest (Set 8). The Crest partial correlation was smaller than its indirect correlation (Set 7) indicating that the removal of SC variance did lower the BI-SNBjMCj relationship. These findings largely support the proposed associations among the social influence variables. However, the failure of the statistical test to indicate that the entire Crest indirect association was due to the moderating effects of SC mitigates these findings and renders the support tentative.





Correlations testing the relationships among social influence and attitudinal variables are presented in Table 3. Correlation Sets 1 and 4 tested hypothesized associations. The positive statistically significant correlations in Set 4 support the associations between the independent variables Aact and SC. However, in Set 1 only the Crest correlation between the antecedent variables is significant and its value indicates the strength of the relationship was only moderate. This finding is unexpected, especially in view of the strong support for the independent variable relationship. It may be that the correlation of the two variable scores, each computed through the multiplication of non-ratio variable scores, may have attenuated these correlations.

Correlation Sets 2 and 3 tested hypothesized lack of associations among variables. The positive statistically significant correlations in both sets provide evidence disconfirming this hypothesis. In retrospect these findings are not surprising given the demonstrated collinearity between Aact and SC and the previous discussion concerning internalization. If certain specific sources of social influence are internalized it seems logical that these sources would become a part of attitude. This could happen, for example, if the focal person assumes a relevant other's expectation is congruent with his feelings or if a commonality of information about the product exists between the focal person and relevant others.

The findings concerning the use of the posited independent and antecedent variables in the central equation are shown in Table 4. The R2s were stronger for both brands and the social influence beta weight values were congruent with expectations when Aact and SC were used as independent variables whereas the weights were incongruent with expectations when SBiai and SNBjMCj were used as predictors. These findings lend additional support to the notion that Aact and SC are more appropriate for use as independent variables than are SBiai and SNBjMCj. In spite of these findings, however, the independent variable collinearity suggests that the central equation beta weight interpretations may not be straight forward. First, an assumption must be made that collinearity does not cause beta weight instability. Second, the joint contribution of the predictors (c.f., Ferber, 1949) to BI variance should be considered since this effect may be stronger than the direct effects indicated by the beta weight values.



Summary and Conclusions

The findings from this research are summarized diagrammatically in Figure 1. The clear cut findings concerning the attitudinal variables are congruent with a large body of research investigating the expectancy value approach to attitude and a smaller but growing body of knowledge regarding relationships among cognitive, attitudinal, and behavioral intention variables. However, the findings concerning social influence variables and crossover effects were not straightforward and cannot be viewed within a framework provided by a body of research. The tentative evidence reported here constitutes the only support for the SC variable. More evidence is needed regarding SC measurement validity and applications across other products, brands, and groups.



The latter suggestion is especially appropriate given the limited generalizability of the sample used in this research. Also, other independent forms of social influence have been hypothesized (Ahtola, 1975; Lutz, 1975c) in addition to SN. Empirical comparisons of these different approaches are yet to be made. Consequently, the specification of meaning for a social influence independent variable is not a settled issue.

Given the state of knowledge, the suggested relationships in Figure 1 concerning crossover and social influence variable relationships are best viewed as tentative suggestions to guide future research. It is hypothesized that the indirect paths between the antecedent and outcome variable, indicated by dotted and dashed lines, represent weaker relationships than the direct paths, indicated by solid lines. The associations in this direct and indirect network seem susceptible to empirical testing through the use of path analysis (Nygreen, 1971). However, the examination of associations is considered only a starting point rather than a complete investigation of what is proposed as a possible causal chain. Additional testing of causal order and lack of spuriousness would follow before causality could be supported or disconfirmed (Popper, 1959). Causal ordering, for example, could be tested with experimental design procedures outlined by Perreault and Darden (1975) that would consider the antecedent variables as factors and attitude, social compliance, and behavioral intention as multiple criterion variables.

More evidence supporting these notions would lead to a reduced form model since there are many variables in addition to those included that have been hypothesized to explain buyer behavior (Andreason, 1965; Nicosia, 1966; Howard and Sheth, 1969). However, there are two major advantages to the reduced form model. First, the relationships among variables are explicitly structured. Second, the level of abstraction allows the development of operational definitions for all variables. Consequently, the model can be shown to fit or not fit empirical regularities, a requirement of all good theories (Kaplan, 1964).


O. T. Ahtola, "Toward a Vector Model of Intentions," in B. B. Anderson (Ed.), Advances in Consumer Research, Volume III (Cincinnati: The Association for Consumer Research, 1975), 469-76.

A. R. Andreason, "Attitudes and Customer Behavior: A Decision Model," in L. Preston (Ed.), New Research in Marketing (California: University of California at Berkeley, Institute for Business and Economic Research, 1965), 1-16.

I. Ajzen and M. Fishbein, "Attitudinal and Normative Variables as Predictors of Specific Behaviors," Journal of Personality and Social Psychology, 27(1973), 41-57.

M. Banton, Roles: An Introduction to the Study of Social Relations (New York: Basic Books, 1965).

P. D. Bennett and G. D. Harrell, "The Role of Confidence in Understanding and Predicting Buyers' Attitudes and Purchase Intentions," Journal of Consumer Research, 2(September, 1975), 110-117.

H. M. Blalock, Jr., "Theory Building and Causal Inferences,'' in H. M. Blalock and A. B. Blalock (Eds.), Methodology in Social Research (New York: McGraw-Hill, 1968).

Francis S. Bourne, "Group Influence in Marketing and Public Relations," in R. Likert and S. P. Haynes (Eds.), Some Applications of Behavioral Research (Paris: UNESCO, 1957), 208-224.

B. E. Bright and H. R. Stammers, "Advertising, Promotion, and Medical Detailing," Proceedings, Seminar on Pharmaceutical Research (Amsterdam: European Society of Market Research, 1971).

J. Bruce, "First Experiences with Fishbein Theory and Survey Methods," Proceedings, Seminar on Translating Advanced Advertising Theories into Research Reality (Amsterdam: European Society of Market Research, 1971, 25-39.

A. B. Cowling, "Use of Elicitation Technique for Producing Dimensions of Brand Choice," Proceedings, British Market Research Society, 1973, 139-156.

L. J. Cronbach, "Coefficient Alpha and the Internal Structure of Tests," Psychometrika, 16(1951), 297-334.

D. E. Dulany, "Awareness, Rules, and Propositional Control: A Confrontation with S-R Behavior Theory," in D. Horton and T. Dixon (Eds.), Verbal Behavioral and General Behavior Theory (New York: Prentice-Hall, 1968).

R. Ferber, Market Research (New York: McGraw-Hill, 1949).

M. Fishbein, "A Consideration of Beliefs, and Their Role in Attitude Measurement," in M. Fishbein (Ed.), Readings in Attitude Theory and Measurement (New York: Wiley, 1967), 257-266.

M. Fishbein, "An Investigation of the Relationships Between Beliefs About an Object and the Attitude Toward that Object," Human Relations, 16(1963), 233-240.

M. Fishbein, "Extending the Extended Model: Some Comments,'' in B. B. Anderson (Ed.), Advances in Consumer Research, Volume III (Cincinnati: The Association for Consumer Research, 1975), 491-497.

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

M. Fishbein and B. H. Raven, "The AB Scales: An Operational Definition of Belief and Attitude," Human Relations, 13(1962), 35-44.

M. Glassman and N. Fitzhenry, "Fishbein's Subjective Norm: Theoretical Considerations and Empirical Evidence," in B. B. Anderson (Ed.), Advances in Consumer Research, Volume III (Cincinnati: The Association for Consumer Research, 1975, 477-480.

J. A. Howard and J. N. Sheth, The Theory of Buyer Behavior (New York: Wiley and Sons, 1964).

J. J. Jaccard and A. R. Davidson, "Toward an Understanding of Family Planning Behaviors: An Initial Investigation," Journal of Applied Social Psychology, 2(1972), 228-235.

I. L. Janis and J. B. Gilmore, "The Influence of Incentive Conditions on the Success of Role Playing Modifying Attitudes," Journal of Personality and Social Psychology, 1(1965), 17-27.

I. L. Janis and B. T. King, "The Influence of Role Playing on Opinion Change," Journal of Abnormal and Social Psychology, 49(1954), 211-218.

M. Karlins and H. I. Abelson, Persuasion (New York: Springer, 1970).

A. Kaplan, The Conduct of Inquiry (San Francisco: Chandler Publishing, 1964).

H. C. Kelman, "Process of Opinion Change," Public Opinion Quarterly, 25(1961), 57-78.

B. T. King and I. L. Janis, "Comparison of the Effectiveness of Improvised vs. Non-improvised Role Playing in Producing Opinion Changes," Human Relations, 9(1956), 177-186.

D. Krech, R. S. Crutchfield and E. L. Ballachey, Individual in Society (New York: McGraw-Hill, 1962).

R. J. Lutz, "An Experimental Investigation of Causal Relations Among Cognitions, Affect and Behavioral Intention,'' Working Paper No. 34, Center for Marketing Studies, University of California, Los Angeles (October, 1975a).

R. J. Lutz, "Changing Brand Attitudes Through Modification of Cognitive Structure," The Journal of Consumer Research, 1(March 1975b), 49-59.

R. J. Lutz, Cognitive Change and Attitude Change: A Validation Study. Unpublished doctoral dissertation, University of Illinois-Urbana, 1973.

R. J. Lutz, "Conceptual and Operational Issues in the Extended Fishbein Model," in B. B. Anderson (Ed.), Advances in Consumer Research, Volume III (Cincinnati: The Association for Consumer Research, 1975c), 469-476.

T. M. Newcomb, "Attitude Development as a Function of Reference Groups: The Bennington Study," in E. E. Macoby, et al. (Eds.), Readings in Social Psychology (3rd ed.) (New York: Holt, Rinehart and Winston, 1958).

F. M. Nicosia, Consumer Decision Processes (Englewood Cliffs, New Jersey: Prentice-Hall, 1966).

J. C. Nunnally, Psychometric Theory (New York: McGraw-Hill, 1967).

G. T. Nygreen, "Interactive Path Analysis," The American Sociologist, 6(February, 1971), 3-37.

C. E. Osgood, G. J. Suci and P. H. Tannenbaum, The Measurement of Meaning (Urbana: University of Illinois, 1957).

William D. Perreault, Jr. and William R. Darden, "Unequal Cell Sizes in Marketing Experiments: Use of the General Linear Hypothesis," The Journal of Marketing Research (1975), 333-342.

K. R. Popper, The Logic of Scientific Discovery (New York: Basic Books, 1959).

M. J. Ryan, "An Empirical Test of a Predictive Model and Causal Chain Derived from Fishbein's Behavioral Intention Model and Applied to a Purchase Intention Situation," unpublished doctoral dissertation, Department of Business Administration, University of Kentucky, 1974.

M. J. Ryan and E. H. Bonfield, "The Fishbein Extended Model and Consumer Behavior," Journal of Consumer Research, 2(September 1975), 118-136.

M. J. Ryan and M. J. Etzel, "The Nature of Salient Outcomes and Referents in the Extended Model," in B. B. Anderson (Ed.), Advances in Consumer Research, Volume III (Cincinnati: The Association for Consumer Research, 1975), 485-490.

R. L. Schmidt, "Implications of a Measurement Problem for Expectancy Theory Research," Organizational Behavior and Human Performance, 10(1973).

M. Sherif, "Group Influences upon the Formation of Norms and Attitudes," in E. E. Macoby, et al. (Eds.), Readings in Social Psychology (3rd ed.) (New York: Holt, Rinehart and Winston), 1958.

A. E. Siegal and S. Siegal, "Reference Groups, Membership Groups, and Attitude Change," Journal of Abnormal and Social Psychology, 55(1957).

L. G. Warner and M. L. DeFleur, "Attitudes as an Interaction Concept: Social Distance as Intervening Variables Between Attitudes and Action," American Sociological Review, 34(1969), 153-169.

A. W. Wicker, "Attitudes Versus Actions: The Relationship of Verbal and Overt Behavioral Responses to Attitude Objects," Journal of Social Issues, 25(1969), 41-78.