What If Opinion Leaders Didn't Know More? a Question of Nomological Validity

ABSTRACT - Given the numerous investigations involving opinion leadership, it is surprising to find that the construct validity of this important concept has yet to be firmly established. While a few studies present evidence for conversant and discriminate validity, no data appear to exist bearing on nomological validity. The present investigation addressed this issue by examining a predicted relationship between opinion leadership and expertise, results show a strong positive correlation between these two concepts, thereby providing support for the concept of opinion leadership, While this study represents only a brief methodological note, had the hypothesis not been confirmed, the implications would have been substantial.

Citation:

Jacob Jacoby and Wayne D. Hoyer (1981) ,"What If Opinion Leaders Didn't Know More? a Question of Nomological Validity", in NA - Advances in Consumer Research Volume 08, eds. Kent B. Monroe, Ann Abor, MI : Association for Consumer Research, Pages: 299-303.

Advances in Consumer Research Volume 8, 1981      Pages 299-303

WHAT IF OPINION LEADERS DIDN'T KNOW MORE? A QUESTION OF NOMOLOGICAL VALIDITY

Jacob Jacoby, Purdue University

Wayne D. Hoyer, Purdue University

ABSTRACT -

Given the numerous investigations involving opinion leadership, it is surprising to find that the construct validity of this important concept has yet to be firmly established. While a few studies present evidence for conversant and discriminate validity, no data appear to exist bearing on nomological validity. The present investigation addressed this issue by examining a predicted relationship between opinion leadership and expertise, results show a strong positive correlation between these two concepts, thereby providing support for the concept of opinion leadership, While this study represents only a brief methodological note, had the hypothesis not been confirmed, the implications would have been substantial.

INTRODUCTION

The concept of opinion leadership has been the subject of numerous investigations in a variety of fields, including consumer behavior, social psychology, communications, marketing, and sociology. Most of these studies have attempted to isolate social activity. general attitudes. demographic, personality, and life style characteristics in order to gain a better understanding of just who opinion leaders are and how they influence others in society (Engel, Blackwell, & Keller 1978).

A major problem with this body of research is that the concept of opinion leadership is accepted uncritically; not much empirical evidence exists to establish the construct validity of this important concept (Jacoby 1974). In order to establish construct validity, one needs to show the extent to which the measures of the construct "fit into a network of expected relations" (Nunnally 1978). This requires demonstrating that: (1) different measures of the same concept are highly related (convergent validity); (2) measures of the same concept are correlated more highly with each other than with measures of totally different concepts (discriminate validity); and (3) the measures of this concept bear some relationship to measures of other constructs which are hypothesized to be related to the concept in question (nomological validity; Campbell 1960).

Virtually all efforts at operationalizing opinion leadership can be reduced to three basic techniques. The self-designation method asks the respondent to judge whether or not he himself is an opinion leader. The key informant approach involves, first, the identification of individuals who are assumed to be knowledgeable of the social interactions within a particular group and, second, having these key informers indicate which members of the group are opinion leaders. Finally, the sociometric method involves interviewing all members of a particular group in order to determine the nature of the social interactions within the group and thus identify the opinion leaders.

The first step in construct validation is to establish the extent to which these measures are related (i.e., convergent validity). Some evidence for convergent validity already exists. Katz and Lazarsfeld (1955) found a strong positive relationship between the self-designating and sociometric approaches. Rogers and Catarno (1962) described data from an unpublished doctoral dissertation which revealed a positive correlation across all three measures. The problem with these studies, however, is that evidence was only provided for convergent validity and no attempt was made to establish discriminate validity.

A subsequent study by Jacoby (1974) applied the Campbell and Fiske (1959) multi-method multi-trait approach to determine both the convergent and discriminant validity of opinion leadership as this was manifested by the women in four large campus sorority organizations. Results showed that "the four coefficients pertinent for assessing convergent validity were surprisingly high in view of the limits that reliability places on validity." Discriminate validity was adequately established for only two of the four groups tested in the study. In general, however, there would appear to be at least some evidence for both the convergent and discriminate validity of opinion leadership.

What still needs to be determined, however, is the extent to which measures of opinion leadership are related as predicted to measures of other conceptually specified constructs (i.e., nomological validity). For example, it seems to be generally assumed that opinion leaders actually possess more knowledge or expertise in the product category of interest than do members of the general population. That is, it seems logical that individuals who are sought out and considered useful sources of information regarding an issue or product category would, in fact, possess a somewhat higher level of knowledge regarding the issue or product category. Evidence for some logical validity would therefore be supplied if a measure of opinion leadership in a particular product category would correlate positively with a measure of actual knowledge regarding that product category. The purpose of the present study was to determine the extent of this relationship and to hopefully be able to provide some degree of evidence for nomological validity which had been previously lacking.

Consumer researchers (e.g., Abelson, et al. 1974, Valenzi and Andrews 1973) have usually measured "knowledge level" or expertise by asking respondents to provide a self estimate of the degree of their knowledge. Researchers are increasingly finding, however, that individuals are not accurate judges of their own knowledge level or ability. For example, De Nisi and Shaw (1977) found that self ratings of ability correlated very poorly with scores on an actual ability test and that self-ability ratings were unable to differentiate between those of high and low ability.

The present study avoided this problem by using a 30 item stereo knowledge test developed by Jacoby and Williams-Jones (1973). This instrument was pretested on 50 undergraduates. Results showed "that scores fell into 3 distinct groups, with approximately one tenth of the scores being clearly higher than the rest and another one fifth falling much lower than the majority of subjects whose scores formed the center of the distribution" (Williams-Jones 1974, p. 16). For purposes of validation, this test was next administered to a sample of 12 stereo repairmen and technicians, all 12 of which scored in the "expert" range. Given these results, it was felt that testing expertise via this instrument was preferable to asking respondents to provide a judgment regarding their level of expertise.

In summary, the present study attempted to examine the nomological validity of opinion leadership by testing the relationship between this construct and another construct believed to be logically related, namely expertise. The specific hypothesis being tested was: Opinion leadership in regard to stereo equipment is positively related to knowledge about the product category.

METHOD

Subjects - One hundred thirty one students (72 males and 59 females) enrolled in an introductory psychology course at a large Midwestern university served as subjects in this study. Participation served as partial fulfillment of a course requirement.

Procedure and Measures - Each subjects filled out a three part questionnaire. The first part consisted of Jacoby's (1972) seven item self-designating measure of opinion leadership. This index differs from the more traditional measures (e.g., King and Summers 1970) in having seven rather than six questions and more response options (for improved reliability) for each question. Apropos of our earlier consents regarding the frailty of verbal report instruments and our comments elsewhere regarding the necessity for multiple indicant research (Jacoby 1976, p. 7), it should be noted that the convergent validity between this index and separate sociometric and key informant indices was found to be quite high (cf. Jacoby 1974). This previous research provides the rationale for employing only a single indicant approach in the present investigation.

To provide additional perspective on the findings, the second section of the questionnaire consisted of five questions which asked subjects: whether they had ever owned any stereo equipment both in the past and presently, whether they had ever been in the process of looking for stereo equipment, the last time they purchased any equipment, and how much they felt they knew about stereos.

The final portion of the questionnaire contained the stereo knowledge test developed by Jacoby and Williams-Jones (1973). The test consisted of 16 multiple choice and 14 matching questions covering a variety of facts about stereo equipment (see Appendix A).

RESULTS

The Pearson product-moment correlations between the measure of opinion leadership, stereo knowledge, and the other assessed variables are presented in Table 1.

Opinion Leadership and Expertise

The most significant finding is that opinion leadership scores correlate as one would expect rich expertise (r =.69, p<.001). That is, the tendency to be an opinion leader for stereo equipment is highly related to one's knowledge about stereo equipment.

Opinion Leadership and Past Experience

Also as might be expected, a positive correlation (r =.32, p<.01) was found between past stereo ownership and opinion leadership. In other words, opinion leaders were somewhat more likely to have owned stereo equipment in the past than were non-opinion leaders. Further, opinion leaders were more likely to currently be in the process of looking for stereo equipment than non-opinion leaders (r =.50, p<.01). There was either a low or no correlation between opinion leadership and either present ownership, number of years owned, years since most recent purchase, or whether the respondent believed he had greater knowledge at the time of the first purchase or now.

Expertise and Past Experience

A low positive correlation was found between expertise and past ownership (r = .25, p<.05). In addition, "experts" were somewhat more likely to be currently looking for stereo equipment than were "nonexperts" (r = .33, p<.01). Consistent with the findings for opinion leadership, the other dependent measures exhibited little or no correlation with expertise.

Sex Differences

As expected, expertise correlated negatively with sex of subject (r = .69, p<.001). Females were far less likely to be "experts" for stereo equipment than were males.

Sex of subject was also correlated negatively with opinion leadership (r =-.44, p<.01). That is, females were also less likely to be opinion leaders with regard to stereo equipment.

In view of these findings, the main analysis was further subdivided according to sex. Table 2 presents the Peerson correlations for males and Table 3 for females. It can be seen from these Tables that, in the case of males, the correlation between opinion leadership and expertise was almost identical to the previous result (r =.66, p<.001). The female correlation was substantially lower (r =.27, p<.05). Thus, the relationship between opinion leadership and expertise for the stereo product class appears to be sex-specific.

Product Involvement

Finally, it is always possible that the link between opinion leadership and expertise is a reflection of a chard variable. One logical possibility for such a third variable is involvement with the product category and several of the variables that were assessed (e.g., looking to purchase, ownership) appear to tap such a dimension. Table 4 presents partial correlations which control for chess "involvement" variables. As shown in Table 4, the relationship between opinion leadership is not significantly lover when these variables are held constant. Thus, those data do not support the third variable hypothesis insofar as the two variables examined are concerned.

DISCUSSION

As hypothesized, opinion leadership was found to be highly related to expertise. This finding is noteworthy for two reasons. First, it has long been assumed that opinion leaders were more knowledgeable in their area of leadership but this key assumption seems to have never been tested. The present findings provide the missing empirical support.

More importantly, this finding can be viewed as evidence for the nomological validity of opinion leadership. In other words, the measure of opinion leadership did in fact bear a predicted relationship to the conceptually related concept (expertise). If one also includes the findings that opinion leaders were more likely to have owned stereo equipment in the past and that they were more likely to be presently looking for such equipment, even more confidence can be placed in the nomological validity of the concept. Since stereo equipment is a more male oriented area, the finding that opinion leadership and expertise are strongly related to sex can also be interpreted as support for nomological validity.

TABLE 1

TABLE 2

MALES

TABLE 3

FEMALES

TABLE 4

PARTIAL CORRELATIONS: OL VS. EXPERTISE

Several considerations, however, must be kept in mind when interpreting the results. First, this assessment was performed for only one product category. Research is needed to determine whether this relationship holds for other product categories and subject matters as well. More importantly, stereo equipment is an area which can be considered highly technical. As a result, the difference between expert and non-expert becomes much more distinct. There may be areas or product categories where this distinction is not clear and these areas may not produce as strong results. Relatedly, the product category is highly male oriented. Research is needed to determine if the same pattern exists in female dominated and more neutral or non-sex typed areas.

Another issue concerns the fact that technical knowledge may be only one of several factors which jointly determine expertise. Other expertise oriented factors which may be equally or more important for the opinion leader to possess include such things as knowledge regarding where to purchase, how much to pay, suitability for various purposes, etc. These factors also warrant empirical attention.

In conclusion, the results of this study, when coupled with the results of the other studies mentioned earlier, seem to indicate a satisfactory degree of validity for the construct of opinion leadership. While it hardly comes as a surprise to find that those who rate themselves as a good source of advice regarding stereo knowledge also have a basis in knowledge for that advice, had this relationship not materialized, the implications would have been substantial. Thus, while only a brief methodological note, this study provides evidence for a critical, but heretofore empirically neglected assumption.

APPENDIX A: STEREO KNOWLEDGE TEST

PART 1

PART 2

REFERENCES

Abelson, H., Schroyer, D. and Gunzelman, S. (1974), Food and Nutrition -- Knowledge and Beliefs. Report prepared for the Division of Consumer Studies, Bureau of Foods, Food and Drug Administration. New Jersey, Response Analyses Corp.

Campbell, D. T. (1960), "Recommendations for Test Standards Regarding Construct, Trait, and Discriminant Validity." American Psychologist, 16, 546-553.

Campbell, D. T., & Fiske, D. W., (1959), "Convergent and Discriminant Validation by the Multitrait-multimethod Matrix," Psychological Bulletin, 81-105.

De Nisi, A. S. & Shaw, J. B. (1977), "Investigation of the Uses of Self Reports of Abilities," Journal of Applied Psychology, 62(5), 641-644.

Engel, J. F., Blackwell, B. D., & Kollat, D. T. (1978), Consumer Behavior, Third Edition, Hinsdale, Illinois: The Dryden Press.

Jacoby, J. (1972) "Opinion Leadership and Innovativeness: Overlap and Validity." In M. Venkatesan (Ed.) Proceedings, Third Annual Conference, The Association for Consumer Research, 2, 632-649.

Jacoby, J. (1974), "The Construct Validity of Opinion Leadership," Public Opinion Quarterly, 26(1), 81-89.

Jacoby, J. (1976), "Consumer Research: Telling it Like it is." In B. B. Anderson (Ed.), Advances in Consumer Research, 3, 1-11.

Jacoby, J. & Williams-Jones, J. (1973), "Price Cue Utilization in Quality Judgments as a Function of Consumer Expertise.'' Unpublished manuscript.

Katz, E., & Lazarsfeld, P. F. (19.55), Personal Influence: The Part Played by People in the Flow of Mass Communication, Illinois: The Free Press, 137-61.

King, C. W., & Summers, J. O. (1970), "Overlap of Opinion Leadership Across Consumer Product Categories." Journal of Marketing Research, 7, 43-50.

Nunnally, J. C. (1978), Psychometric Theory, Second Edition, New York: McGraw-Hill,

Rogers, E. M. & Catarno, D. G. (1962), "Methods of Measuring Opinion Leadership;" Public Opinion Quarterly, 441.

Valenzi, E. R. & Eldridge, L. (1973), "Effect of Price Information, Composition Differences, Expected, and Rating Scales on Product Quality Ratings." Proceedings of the 81st Annual Convention of the American Psychological Association, 8, 820-830.

Williams-Jonas, J. (1974), "Price Cue Utilization in Quality. Judgements as a Function of Expertise: A Replication." Unpublished Master's thesis, Purdue University.

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