The Cad Instrument in Behavioral Diagnosis


Jerome B. Kernan (1971) ,"The Cad Instrument in Behavioral Diagnosis", in SV - Proceedings of the Second Annual Conference of the Association for Consumer Research, eds. David M. Gardner, College Park, MD : Association for Consumer Research, Pages: 307-312.

Proceedings of the Second Annual Conference of the Association for Consumer Research, 1971     Pages 307-312


Jerome B. Kernan, University of Cincinnati

[These studies were supported by a grant from the Educational Foundation of the American Association of Advertising Agencies. The assistance of graduate students Dan Daily, Thomas Elliott, Richard Findlay, John Geers, Jeffrey Levenberg, Judith Moneyhon, George Robb, Donald Taplits, and Leonard Walke, who collected the data reported here, is gratefully acknowledged.]

[Professor of Behavioral Analysis, University of Cincinnati. Requests for reprints should be sent to the author at the College of Business Administration, University of Cincinnati, Cincinnati, Ohio, 45221]

One of the more relevant criticisms of empirical research in consumer behavior centers about the contention that we sometimes play a "hit-and-run game" in an effort to develop a body of behavioral knowledge. Generally, this takes one of two forms. First, we fail to adequately replicate studies--we fail in what Kollat, Engel, and Blackwell (1970) have dubbed the "replication tradition"--and generalize on the strength of isolated studies. And second, we too commonly use unproved, unvalidated instruments to generate data--we "lift" standard personality tests to measure consumer behavior. for example.

Cohen (1967) has recently offered some potential relief from the latter criticism by publishing his CAD instrument. Its unique feature among personality tests is that it was designed specifically to diagnose consumer behavior. Accordingly, it deserves especial attention by marketing researchers. It is the purpose of this paper to relate some empirical experiences using the CAD instrument in an effort to suggest its usefulness and limitations.

The CAD instrument consists of three scales, measuring a person's compliance, aggressiveness, and detachment. A thirty-five item, self-administered Likert-type battery, the instrument is derived from Karen Horney's (1945) tripartite interpersonal model. It suggests that individuals can be placed into three groups, which reflect their predominant mode of response to others: (1) those who move toward other people (compliant); (2) those who move against people (aggressive); and those who move away from people (detached).

Since Cohen's exposition of the CAD instrument includes the actual scales and certain psychometric data attendant to them, this paper does not dwell on these points. Rather, we wish to relate some hands-on experience in using the instrument in the hope that others whose research counsels the use of a response-tendency instrument might better be able to judge the CAD's efficacy. Instrument statistics for these CAD applications appear in the Appendix.


Five unpublished studies are considered here. Each of these is noted only briefly, the emphasis being on those elements of the study particularly germane to CAD.

Message Advocacy

Using an ex post facto design, a probability sample of 87 University-of-Cincinnati student Ss was examined relative to whether their self-reported behavior was consistent with antismoking TV commercials. CAD scores for Ss were obtained and analyzed categorically: scores at or above the empirical median were counted as "high;" below the median as "low." [Median scores were: for C, 41; for A, 48; and for D, 29. The theoretic medians for the CAD are: for C, 36; for A, 53; and for D, 36.] Among smokers, :2 analysis showed that high compliance and "cutting down or stopping smoking as a result of the commercials" was significant at the .20 level. Neither aggressiveness nor detachment was indicated to be related to this phenomenon. Among the same Ss, high compliance was also significantly associated (p = .09) with persuading others to reduce or stop smoking. Aggressiveness and detachment again proved unrelated. These findings are interesting in their own right, but perhaps especially so since smokers' initial reaction to the commercials proved to be independent of their compliance scores. Apparently, then, message advocacy is to some extent a function of one's compliance, but the advocated behavior does not occur immediately. Moreover, affective forecasts of it (initial reactions to the message) might be very misleading.

Group Influence

The comparison-shopping department of a large department store was analyzed, using the CAD, for group influence. Ss comprised ten women, age 22-51. Five of these, whose group average on compliance was 40.6, were designated as the low-compliance group. The remaining five, whose group average on compliance was 47.8, served as the high-compliance group. Group differences on both aggressiveness and detachment were insignificant. "Information" was then transmitted through the (consensually designated) opinion leader--with her collaboration [The opinion leaders' aggressiveness score was 60. This was not only the highest in the department but also significantly different from the department's average on aggressiveness of 45.3 at the .001 level.]--to both groups. The information was to the effect that "values" on certain items were greater at competitive stores than at the employing store, in spite of the 10% employee discount. Purchasing behavior was measured both before and (two weeks) after the message receipt in both groups. In the low-compliance group, no one's behavior changed. In the high-compliance group, four of five women changed their purchasing behavior in the direction of the advocated behavior--i.e., to competitive stores. The difference in mean compliance scores between these two groups is significant at the .05 level (2-tailed). The more compliant, then, the more susceptible to group norms a person appears to be.

Fashion Possession and Approval

A convenience sample of 57 seventh- and eighth-grade girls was examined for possession and approval of 16 items of clothing and cosmetics fashion. The mean number of items possessed was 7.39: a = 2.53; range = 1-12 items. Those Ss who possessed at least 10 of the items were found to be uniformly "high" in compliance; these 11 girls were all above the theoretic median of 36 on this scale, whereas the sample as a whole (all 57 Ss) showed a highly-compliant proportion of 91 percent. The "high-possession" group showed a mean compliance score of 53.18; the "low-possession" group (fewer than 10 items possessed), however, had a mean compliance score of 47.27. The t statistic of 2.51 indicates these group means to be significantly different at the .05 level (2-tailed test). Differences in aggressiveness and detachment means between these same groups proved insignificant

After a principal-components analysis failed to materially reduce the 16-item set, a canonical analysis was performed, using the CAD as predictor variables and approval scores for the 16 items as the criterion set (all variables standardized because of the skewed distributions). The canonical correlation was .69 (p = .03); and the canonical index was .48. [Successive roots were not significant, even at the .10 level.] The predictor coefficients were: C = .94; A =.32; and D = .08. Criterion coefficients ranged from a high of .62 for face makeup to a low of .01 for costume rings. In general, the greatest contributions in this latter set were made by cosmetic items. Thus, approval of fashion (especially cosmetic) items appears strongly associated with high compliance.

New-Product Information Sources

A convenience sample of 332 housewives was queried as regards their primary information source (product samples, coupons, price deals, friends' recommendation, or TV advertising) for new products (detergents and presoaks, instant coffees, and foods) they had tried. Using empirical medians as input to x2 analysis, [Median scores were: for C, 44; for A, 47; and for C, 30] and pooling data across product categories, it is clear that the use of product samples is associated with high aggressiveness (p < .01), and that compliance and detachment are insignificant; and that reliance on TV advertising is associated with high detachment (p < .01), with compliance and aggressiveness being insignificant. Reliance on coupons and/or price dealings seems to be totally independent of CAD. It would appear, then, that in securing information about these new products, the compliant person is vulnerable to peer influence while the aggressive person "goes it alone" with a product sample. The detached person, consistent with a priori notions, seems to prefer the "impersonal" source of TV advertising.

Brand Loyalty

A probability sample of 172 households was examined for brand loyalty (in an admittedly simplistic sense). "Loyalty" was a self-designated notion of "whether you have a brand preference" in a variety of product categories. It should be noted that neither the nature nor the degree of this "loyalty" was analyzed. Using Ss' standardized CAD scores, multiple discriminant analyses were run to determine the extent to which loyalty (or no loyalty) can be accounted for by CAD. The results suggest that it can for mouthwashes (p = .01), hairsprays and air fresheners (p = .05), deodorants, headache remedies, and razors (p = .10); almost can for cigarettes and cigars (D = .25); and cannot for bath soaps, soft drinks, shampoos, teas, pipe tobaccos, coffees, and toothpastes (all not significant). Compliance was the largest contributor to the discrimination of the mouthwash, air-freshener, razor, and cigar preference functions; aggressiveness contributed most to partitioning the hairspray, deodorant, headache remedy, and cigarette categories; detachment was the largest contributor in none of the significant discriminant functions.

These findings, while hardly identical to Cohen's, tend more to support his earlier results than to refute them.


The empirical studies briefly noted each show encouraging results with the CAD instrument. This is not to argue that the instrument, in toto or any one of its scales, will necessarily explain behavior. The evidence seems reasonably compelling, however, that CAD is a highly useful way to "first-cut" a data batch, so long as the behavior under analysis is likely to be a function of interpersonal disposition.

For those consumer researchers who choose to follow an interpersonal orientation to behavior (and this may be a minority) Cohen's instrument appears to offer much. It is easily administered(neither adolescents nor adults seem to have any problems understanding it), short, and contains no socially-taboo statements. And if the studies cited are any indication, it appears that it possesses a most desirable, albeit pragmatic, of instrument attributes: It works. Of course only extensive use of the instrument can truly prove its efficacy. Its prognosis even at this incipient stage, however, would surely seem to be good.


Cohen, J. B. Interpersonal response traits and consumer behavior. Unpublished doctoral dissertation, University of California, Los Angeles, 1966.

Cohen, J. B. An interpersonal orientation to the study of consumer behavior. Journal of Marketing Research, 1967, 4: 270-8

Cronbach, L. J. Coefficient alpha and the internal structure of tests. Psychometrika, 1951, 16: 297-334.

Horney, K. Our Inner Conflicts. New York: W. W. Norton & Company, Inc., 1945.

Kollat, D. T., Engel, J. F. and Blackwell, R. D. Current problems in consumer behavior research. Journal of Marketing Research . 1970 7 327-32.





Jerome B. Kernan, University of Cincinnati


SV - Proceedings of the Second Annual Conference of the Association for Consumer Research | 1971

Share Proceeding

Featured papers

See More


Memory-Based Models of Predicting Inferences about Brand Quality

Yvetta Simonyan, University of Bath, UK
Dan Goldstein, Microsoft Research

Read More


I9. From Childhood Toys to Grownup Choices: Understanding the Gendered Appeal of Violent Media

Martin A. Pyle, Ryerson University

Read More


Explaining the Attraction Effect: An Ambiguity-Attention-Applicability Framework

Sharlene He, Concordia University, Canada
Brian Sternthal, Northwestern University, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.