Involvement, Attitude, and Behavior: on the Nature of the Relationships

ABSTRACT - Papers presented by Finn, Antil, and Schindler and Berbaum are discussed to emphasize their theoretical and empirical contributions, as well as their implications for future research. While all three papers are not specifically descriptions or discussions of involvement and attitude research, each contains implications for one or both of these areas.

Citation:

E. H. Bonfield (1983) ,"Involvement, Attitude, and Behavior: on the Nature of the Relationships", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 425-426.

Advances in Consumer Research Volume 10, 1983      Pages 425-426

INVOLVEMENT, ATTITUDE, AND BEHAVIOR: ON THE NATURE OF THE RELATIONSHIPS

E. H. Bonfield, Temple University

ABSTRACT -

Papers presented by Finn, Antil, and Schindler and Berbaum are discussed to emphasize their theoretical and empirical contributions, as well as their implications for future research. While all three papers are not specifically descriptions or discussions of involvement and attitude research, each contains implications for one or both of these areas.

INTRODUCTION

Finn, Antil, and Schindler and Berbaum have each approached the focal problems in consumer research--predicting and understanding consumer behavior--in different ways. Each makes a contribution to the understanding of how involvement and/or attitude interact with respect to behavior.

The title of Finn's paper is both perfectly accurate and misleading. Behavior and cognitive processing may well be different under conditions of low and high involvement, but low involvement research is not a low involvement activity for the researcher. Moreover, research thus far has not shown low involvement to lead to consistently less cognitive processing than high involvement.

Antil's research provides implications about the relationships among involvement, attitude, and behavior at one point in time. These implications are about the purchase decision stage of the single hierarchy model Finn recommends.

Schindler and Berbaum provide evidence which suggests that experimental situations can be high involving, thus affecting behavior.

THE DOMAIN OF INVOLVEMENT

Both Finn and Antil provide ample evidence that the involvement concept is, if nothing else, elusive.

Finn tells us that involvement has been studied as a person specific variable, a situation specific variable, a type of cognitive processing, and a behavioral variable. Antil links involvement with response certainty or confidence and with attitude intensity.

Finn (1982) and Smith and Swinyard (1982) have separately argued persuasively for a single hierarchy model. Each is similar, but Finn suggests more variables of interest while Smith and Swinyard focus on the belief-affect relationship. Finn's is the more general approach and may help prevent us from forgetting the total picture when we focus on a narrow portion of it.

Attitude is, indeed, formed prior to buying. Prior to the purchase decision stage of the single hierarchy model, this is an attitude toward some form of commitment to the product, with a negative attitude implying rejection of the product. Prior to, or during, the evaluation behavior stage of the single hierarchy model, the attitude formed is toward some form of trial of the product--including a trial purchase. Most of our consumer research attitude instruments have measured attitudes toward products. Consumer-respondents appear to treat these as measures of some form of commitment to the product. Under these conditions, when the consumer is at the evaluation stage, the low involvement hierarchy is spuriously supported.

Under this interpretation, Antil's results support the single hierarchy model. That is, low involvement respondents are at the evaluation behavior stage or earlier, while high involvement respondents are at the purchase behavior stage.

The point is that if we want to predict behavior, we must develop our instruments in terms of that behavior. Instruments designed, or associated with, purchase decision stage behavior cannot provide information about the evaluation stage--and -vice versa. This is the essence of the notion of attitude toward the act suggested by Fishbein (1967) and which I have discussed for different variables as anchoring (Bonfield 1981).

Finn's arguments that products cannot be differentiated as low or high involvement are sound as long as we confine analysis to the cognition stage--as Finn does. It may be desirable, however, to be consistent in concepts at all stages of the hierarchy; and more active cognitive processing is expected for high than low involvement products at the purchase decision stage. Thus, I am not quite ready to reject the involvement label for products.

As Finn points out, it is difficult to separate subjects from product/situation combinations. Since he does not reject interest/importance and goals and consequences as useful for understanding involvement, we turn to his rejection of commitment.

Finn rejects commitment, not because it fails to meet his two criteria for usefulness (I agree it is an open issue), but because he does not want to confuse the brand loyalty and involvement literature. But if brand loyalty and involvement are related, we should want to understand the nature of the relationship. Besides, anyone who studies consumer behavior, let alone focuses on involvement, already has a high tolerance for confusion. More seriously, brand loyalty would be expected at the purchase decision stage, not at the cognition stage. Thus, commitment and brand loyalty should not be confused at the cognition staRe.

Finally, Finn's single hierarchy model provides an improved conceptual framework for studying the numerous influences which affect its stages. Finn has provided a general research paradigm and one specific (i.e., response centered) research paradigm for cognition. We, including Finn, are challenged to provide the research paradigms for the other stages--and the research at all stages.

CERTAINTY AND INVOLVEMENT

Antil's results support the use of response certainty as an indicator of involvement when the population is relatively certain and attitudes are largely non-neutral. His results also support the use of response certainty-as a surrogate for response intensity--to estimate the true neutral point for the scale.

I was impressed that Antil developed a valid and reliable attitude instrument for his project, as well as his use of a moderator variable. Consumer researchers have been remiss in validating their attitude measures (Bonfield 1979) and also infrequent users of moderator variables (Sharma, Durand, and Gur-Arie 1981). Treating any involvement-linked variable as a moderator should enhance understanding of the concept.

Antil's use of a multiple-item behavior measure is also notable. It is supported by the correlation of .56 between attitude and behavior.

We have to recognize, however, that attitude-behavior correlations will be enhanced when multiple-item behavior measures are used. We should attend to correlations (point biserial if preferred)between attitude and each specific behavior. Because it appears that the attitude instrument is general, we should not expect high attitude-specific behavior correlations. But it would be instructive to know what behaviors were associated with high and low attitude-behavior correlations. For example, solar heating ownership might be found more frequently among groups with high attitude-behavior correlation, while a high incidence of membership in a car pool might be found for groups with low attitude-behavior correlation. In the former case, involvement is high, while among low involvement people, there might be social pressure from neighbors and/or co-workers to join a car pool.

Antil's research, it should be noted, is at the purchase decision stage of Finn's single hierarchy model. This is true, of course, for the research itself, not for the respondents.

THE IMPACT OF INVOLVEMENT IN EXPERIMENTS

For the Schindler and Berbaum experiment, we have to ask whether the subjects' "involvement" in the experiment itself had an impact on the results. That is, does the research have external validity--can the results be generalized to the marketing situations suggested by Schindler and Berbaum?

Moreover, what are the demand artifacts? The only cues the subjects had were the differential shading of the numbers which were salient to the experimenters--but which were not necessarily salient to the subjects. Not only are the found results expected, we might question why they were not even more drastically in the found direction.

Schindler and Berbaum are not unique in having to face this problem. The results of marketing studies on conformity and cognitive processing (e.g., Bettman, Capon, and Lutz 1975a,b.c) are also suspect because of potential demand artifact.

Formal debriefing following the Dulany (1968) paradigm on which the Fishbein Extended Model is based (Ryan and Bonfield 1975) might give a reasonable estimate of the effects of demand artifact. In Dulany's paradigm, the subject's assessment or the experimenter's expectations is obtained, as well as the degree to which the subject wanted to comply with those perceived expectations.

Even with the demand artifact problem, we should expect the results of the Schindler and Berbaum study to generalize to product purchase situations where it is difficult to differentiate among choices. In these cases, any very noticeable difference may be the cue needed to make a choice. This finding is compatible with the predictions of Krugman and those who treat involvement as he does.

CONCLUSIONS

The papers reviewed have made contributions for guiding consumer researchers. Finn, in particular, has provided an excellent exercise in the type of theoretical thinking we need. We do not have to agree with his positions to appreciate the value of the process he used in reaching them. Antil has reemphasized the importance of developing valid instruments and using moderator variables. While the use of response consistency as a moderator variable has not been proven, its use was strongly supported in the present set of circumstances. Schindler and Berbaum provide evidence that it does not take a great deal of difference, or even an important difference, to make a difference.

REFERENCES

Bettman, James R., Capon, Noel, and Lutz, Richard J. (1975a), "A Multimethod Approach to Validating MultiAttribute Models," in Advances in Consumer Research, Volume 2, ed. M. J. Sclinger, Chicago: Association for Consumer Research, 357-74.

Bettman, James R. (1975b), "Cognitive Algebra in Multi-Attribute Motels," Journal of Marketing Research, 12, 357-74.

Bettman, James R. (1975c), "Multiattribute Measurement Models and Multiattribute Attitude Theory: A Test of Construct Validity," Journal of Consumer Research, 1, 1-15.

Bonfield, E. H. (1981), "Different Influences on Consumer Decision Making," in Advances in Consumer Research, Volume 8, ed. Kent B. Monroe, Association for Consumer Research. 748-9.

Bonfield, E. H. (1979), "A Comment on the State of Attitude Measurement in Consumer Research: A Polemic," in Advances in Consumer Research, Volume 6, ed. William L. Wilkie Association for Consumer Research, 238-44.

Dulany, Don E. (1968), "Awareness, Rules, and Propositional Control: A Confrontation with S-R Behavior Theory," in Verbal Behavior and General Behavior Theory, eds. D. Horton and T. Dixon, New York: Prentice-Hall. 340-87.

Finn, David W. (1982), "It Is Time to Lay the Low-Involvement Hierarchy to Rest," in An Assessment of Marketing Thought & Practice, eds. Bruce J. Walker et al, Chicago: American Marketing Association, 99-103.

Fishbein, Martin (1967), "Attitude and the Prediction of Behavior," in Readings in Attitude Theory and Measurement, ed. Martin Fishbein, New York: Wiley, 477-92.

Ryan, Michael J. and Bonfield, E. H. (1975), "The Fishbein Extended Model and Consumer Behavior," Journal of Consumer Research, 2, 118-36.

Sharma, Subhash, Durand, Richard M., and Gur-Arie, Oded (1981), "Identification and Analysis of Moderator Variables," Journal of Marketing Research, 18, 291-300.

Smith, Robert E. and Swinyard, William R. (1982), "Information Response Models: An Integrated Approach," Journal of Marketing, 46, 81-93.

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