Contribution of Instrumentality and Value to Attitudinal Affect

Olli T. Ahtola, University of Florida
ABSTRACT - This article suggests a new way of looking at and testing the relative contributions of the two components in expectancy-value attitude models. The results suggest that very little if any predictive power is lost by greatly simplifying the most commonly used models.
[ to cite ]:
Olli T. Ahtola (1977) ,"Contribution of Instrumentality and Value to Attitudinal Affect", in NA - Advances in Consumer Research Volume 04, eds. William D. Perreault, Jr., Atlanta, GA : Association for Consumer Research, Pages: 117-120.

Advances in Consumer Research Volume 4, 1977   Pages 117-120

CONTRIBUTION OF INSTRUMENTALITY AND VALUE TO ATTITUDINAL AFFECT

Olli T. Ahtola, University of Florida

ABSTRACT -

This article suggests a new way of looking at and testing the relative contributions of the two components in expectancy-value attitude models. The results suggest that very little if any predictive power is lost by greatly simplifying the most commonly used models.

Rosenberg's (1956) finding that perceived instrumentality of an attitude object to attain or block salient values controls more variance in attitudinal affect than does value importance (i.e., how satisfactory or dissatisfactory the value is), has led, together with the results of some marketing researchers, to a suggestion that the value importance component should be dropped from the attitude model developed by Rosenberg (Sheth and Talarzyk, 1971; 1972). This argument has been supported by empirical findings that weighting perceived instrumentality by value importance suppresses the prediction of affect as compared to un-weighted perceived instrumentality (Sheth and Talarzyk, 1971; 1972; Moinpour and MacLachlan, 1971; Lutz and Howard, 1971). This issue has created considerable attention and numerous studies which have recently been reviewed by Wilkie and Pessemier (1973). No consensus has been arrived at and both theoretical and empirical support has been presented in favor of both points of view.

However, it has been pointed out by Ahtola (1971, 1973) and Cohen, Fishbein, and Ahtola (1972) that the method used to measure the components of the Rosenberg model in these marketing studies was not conceptually in line with the one used by Rosenberg (1956). It was shown that the perceived instrumentality measure used by marketing researchers actually confounded and combined both perceived instrumentality and value importance. Consequently, it is not surprising that nothing was gained by adding value importance to the model again. Furthermore, the value importance measure used by marketing researchers was not conceptually equivalent to the one used by Rosenberg.

One has to conclude that, so far, marketing researchers have failed to investigate which cognitive component, perceived instrumentality or value importance, accounts for more variance in attitudinal affect or whether they are equally important in that respect.

When investigating this issue it has to be made clear that it is intuitively and logically nonsensical to argue that either of the two components could be totally excluded from the model. It is counterintuitive to argue that it does not make any difference if some salient satisfactory/dissatisfactory value (or product attribute) is either attained (possessed) or blocked (not possessed, avoided) by an attitude object. It is equally counterintuitive to argue that it does not make any difference whether an attitude object attains/blocks satisfactory or dissatisfactory values.

For example, assuming that rust in a car is a negatively evaluated salient attribute, it is illogical to argue that it does not make any difference whether some detergent to wash cars tends to rust them or tends to prevent rust. Also, assuming that certain pills cause you to lose weight, it is illogical to argue that it does not make any difference whether you want to lose weight or whether you want to gain weight, you will like the pills equally well.

Even though it must be concluded that both components, perceived instrumentality and value importance, have to be incorporated into the model implicitly or explicitly, it is of interest to investigate the contributions of these two components as determinants of attitudinal affect.

In reviewing the attitude literature in social psychology at least four conceptualizations emerge.

First, the conceptualization originated by Heider (1946) considers only the direction of association (e.g., perceived instrumentality) and only the direction of liking (i.e., value importance) in deriving attitude from an individual's cognitive organization. This model, consequently, assumes both components to be equally important but scales them only as +1 or -1. Cartwright and Harary (1956) added the third scale value by including the absence of relationship, i.e., scale value O.

Second, the conceptualizations by Rosenberg (1956) and Fishbein (1967) differ from that by Heider in that also the magnitude of association and liking in addition to the direction was taken into consideration. Fishbein (Fishbein and Raven, 1962) measured both components on +3 to -3 scales. Rosenberg (1956) measured perceived instrumentality on a +5 to -5 scale and value importance on a +10 to -10 scale. However, no reason was given why value importance was weighted twice perceived instrumentality by Rosenberg.

Third, Osgood and Tannenbaum's (1955) congruity model measures the liking (evaluation) relationship on +3 to -3 scales, that is, it takes both the direction and the magnitude of liking into consideration, while only the direction of association (e.g., instrumentality) is included. The model considers associations only positive (+1) or negative (-1). Similar conceptualization seems to underlie the Thurstone (1928) scale where subjects either endorse or do not endorse a belief statement which is scaled on affect.

Fourth, even though it .is difficult to find an obvious example of a conceptualization where only the direction of liking is considered while both the direction and the magnitude of association are taken into consideration, it could be said that the Likert (1932) scale comes very close to that. In the Likert scale, subjects' agreement/disagreement with belief statements is measured on a 5-point scale. The liking relationship seems to be taken into consideration when the scale values are reversed if the correlation of a statement with the battery of statements 'utilized is negative. (For a detailed discussion see Green, 1954.)

In summary, four different attitude models can be derived from the above well-known attitude measurement instruments and conceptualizations.

Model 1. Only the direction of liking (i.e., value importance) and association (e.g., perceived instrumentality) of salient values/attributes deter- mines the attitude.

Model 2. Both the direction and the magnitude of liking and association of salient values/attributes determine the attitude.

Model 3. Both the direction and the magnitude of liking but only the direction of salient values/attributes determine the attitude.

Model 4. Only the direction of liking but both the direction and the magnitude of association of salient values/ attributes determine the attitude.

METHOD

Because Models 1, 3, and 4 are reduced versions of Model 2, data designed to test Model 2 was used for all the four models. Thus, the data for Models 1, 3, and 4 was derived from the data collected for Model 2. 190 undergraduate students at Western Illinois University and 339 undergraduate students at the University of Florida were used as subjects. Students' attitudes toward three brands of soft drinks, shampoos, underarm deodorants, and three fast food restaurants were measured. The same subject never responded to more than one product group. Three somewhat different versions of Model 2 were used to collect the data. Each subject was exposed to only one of the three versions. The first version followed closely the procedure outlined by Fishbein (1967; Fishbein and Raven, 1962). Only this procedure was used to measure attitudes toward the three soft drinks. For the rest of the products, attitudes were also measured using the procedure outlined by Rosenberg (1956) with the modification that both value importance and perceived instrumentality were measured on +3 to -3 scales. The third procedure modified the Rosenberg model by using product attributes instead of personal values as salient concepts associated with the attitude object. (For detailed explanation of the procedure see Ahtola, 1973; and Mazis, Ahtola, and Klippel, 1975).

The salient values and the product attributes associated with each product selection were elicited in separate pilot studies.

In the questionnaires, the first portion contained the evaluation or value importance questions. The second portion contained the instrumentality questions. And in the final portion, the respondent was asked to provide his overall attitude toward the three brands in question using a set of four scales found to load highly and consistently on the evaluative dimension of the Semantic Differential (Osgood, Suci, and Tannenbaum, 1957).

To derive data for Models 1, 3, and 4 every positive score in the evaluation and/or instrumentality questions was transformed to +1 and every negative score in the evaluation and/or instrumentality questions was transformed to -1 as required by each model. Zero scores were not transformed.

RESULTS

To calculate each subject's attitude score for each model the two components, evaluation and instrumentality, were multiplied and then summed over. Thus twelve attitude scores were calculated for each respondent, corresponding to the three attitude objects and the four models. These attitude scores were then correlated with the direct measure of overall attitudes to find out how well they converged.

From Table 1. it becomes immediately clear that the differences between the models are not strikingly large. To test the null hypotheses that there are no significant differences between the four models, the Friedman two-way analysis of variance by ranks for related samples was used (Siegel, 1956).X for the thirty comparisons was 35.32 with 3rd.f. which is significant beyond the .001 level. The average ranks (1 lowest, 4 highest) were 1.63; 3.30; 2.00; and 3.07 for Models 1, 2, 3, and 4, respectively.

When comparing Models 1 and 2 it was found that 26 times correlations went down and 4 times up when the magnitudes of both instrumentality and evaluation were eliminated. This result gives the Sign test z of 3.834 which is significant beyond the .001 level in the two-tailed test (Siegel, 1956).

When comparing Models 2 and 3 it was found that, again, 26 times correlations went down and 4 times up when only the magnitude of instrumentality was eliminated. The Sign test z of 3.834 is significant beyond the .001. level.

When comparing Models 2 and 4 it was found that 17 times correlations went down and 13 times up when only the magnitude of evaluation was eliminated. The Sign test z of .548 is not significant.

TABLE 1

COMPARISON OF THE MODELS

From above, one might conclude that Model 4 gives far better predictions than Model 3. However, it is theoretically still quite possible that Model 3 had better predictions more often than Model 4. However, when comparing Models.3 and 4 the expected relationship was found. In 22 out of 30 comparisons, Model 4 gave higher correlations than Model 3. The Sign test z of 2.3734 is significant at the .02 level.

The above results indicate that the predictive power of the instrumentality-value attitude model is somewhat reduced when the magnitude of instrumentality is eliminated. The above results failed to prove that the elimination of the magnitude of evaluation also reduced the predictive power. It should be noted that one cannot conclude that the magnitude of evaluation never has any predictive power based on these studies. It may be that in some more ego-involving situations the magnitudes of evaluations vary more from individual to individual and/or among the product attributes which could have an effect on the overall attitudes. Moreover, because the inclusion of the magnitude of evaluation did not reduce the predictive power (actually, it more often than not increased it, but not often enough for statistical significance) there are no strong reasons to draw a conclusion that only the direction and not the magnitude of evaluation determines the attitude.

From practitioners' point of view it is interesting to notice that Model l's predictions are almost as good (even though statistically significantly smaller) as the predictions of the best model, Differences in the magnitudes of correlation coefficients between the models are strikingly small. If this were generally true the practitioner might want to select Model 1 for pragmatic reasons. It would be fairly fast and easy to conduct telephone surveys using this model because the interviewer needs to ask only if the respondent likes, dislikes, or is indifferent toward a value (e.g. product attribute) and whether the respondent thinks that the attitude object (e.g., a given brand in a product class) possesses, does not possess, or he does not know if it possesses the value.

REFERENCES

O. T. Ahtola, An investigation of cognitive structure within expectancy x value response model. Unpublished manuscript, (University of Illinois, 1971).

O. T. Ahtola, An investigation of cognitive structure within expectancy-value response models. (Doctoral dissertation, University of Illinois Ann Arbor, Michigan, University Microfilms, 1973), No. 74-5517.

D. Cartwright and T. Harary, Structural balance: a generalization of Heider's theory. Psychological Review, 63 (1956), 277-93.

J. B. Cohen, M. Fishbein and O. T. Ahtola, The nature and uses of expectancy-value models in consumer attitude research. Journal of Marketing Research, 9 (November 1972), 456-60.

M. Fishbein, A consideration of beliefs and their role in attitude measurement. In M. Fishbein (ed.), Readings in attitude theory and measurement. (New York: John Wiley & Sons 1967).

M. Fishbein and B. H. Raven, The AB scales: an operational definition of belief and attitude. Human Relations, 15 (1962), 35-44.

B. F. Green, Attitude measurement. In G. Lindsay (ed.), Handbook of Social Psychology. (1st ed.) (Reading, Mass.: Addison-Wesley 1954).

F. Heider, Attitudes and cognitive organization. Journal of Psychology, 21 (January 1946) 107-12.

R. Likert, A technique for the measurement of attitudes. Archives of Psychology, 140 (1932) 44-53.

R. J. Lutz and J. A. Howard, Toward a comprehensive view of the attitude-behavior .relationship: the use of multiple-set canonical analysis. Proceedings of the Social Statistics Section of the American Statistical Association, (1971), 215-24.

M.B. Mazis, O. T. Ahtola and R. E. Klippel, A comparison of four multi-attribute attitude models in the prediction of consumer attitudes. Journal of Consumer Research, 2 (June 1975), 38-52.

R. Moinpour and D. L. MacLachlan, The relations among attribute and importance components of Rosenberg-Fishbein type attitude model: an empirical investigation. Proceedings of the 2nd Annual Conference of the Association for Consumer Research, (1971), 365-75.

C. E. Osgood and P. H. Tannenbaum, The principle of congruity in the prediction of attitude change. Psychological Review, 62 (1955), 42-55.

C. E. Osgood, G. J. Suci and P. H- Tannenbaum, The measurement of meaning. (Urbana, Illinois; University of Illinois Press, 1957).

M. J. Rosenberg, Cognitive structure and attitudinal affect. Journal of Abnormal and Social Psychology, 53 (November 1956), 367-72.

J. N. Sheth and W. W. Talarzyk, Relative contribution of perceived instrumentality and value importance components in determining attitudes. Faculty working paper No. 15, College of Commerce and Business Administration, (University of Illinois, 1971).

J. N. Sheth and W. W. Talarzyk, Perceived instrumentality and value importance as determinants of attitudes. Journal of Marketing Research, 9 (February 1972), 6-9.

S. Siegel, Nonparametric statistics for the social sciences. (New York: McGraw-Hill, 1956.

L. L.. Thurstone., Attitudes can be measured. American Journal of Sociology, 33 (1928), 529-54.

W. L. Wilkie and E. A. Pessemier, Issues in marketing's use of multi-attribute attitude models. Journal of Marketing Research, 10 (November 1973), 428-41.

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