The Communication Structure of Consumer Opinions

ABSTRACT - In the literature, a distinction between opinion leaders and opinion followers has been made. In this study, both concepts were measured by multiple item scales in student samples in the USA and the Netherlands in seven consumption domains. It appeared that the communication structure of opinions was characterized more by clusters of consumption domains than by the distinction of leaders and followers. Three clusters of consumption domains were found with respect to the communication of opinions: (1) clothings, food articles and external care products, (2) vacation trips, audio/video equipment and leisure time/hobbies, and (3) media. The findings suggest that the theory of consumer communication flows should be revised.

Citation:

Gerrit Antonides and Gulden Asugman (1995) ,"The Communication Structure of Consumer Opinions", in E - European Advances in Consumer Research Volume 2, eds. Flemming Hansen, Provo, UT : Association for Consumer Research, Pages: 132-137.

European Advances in Consumer Research Volume 2, 1995      Pages 132-137

THE COMMUNICATION STRUCTURE OF CONSUMER OPINIONS

Gerrit Antonides, Erasmus University

Gulden Asugman, Bogazici University

[Jack Smeins is gratefully acknowledged for creating the Dutch data set.]

ABSTRACT -

In the literature, a distinction between opinion leaders and opinion followers has been made. In this study, both concepts were measured by multiple item scales in student samples in the USA and the Netherlands in seven consumption domains. It appeared that the communication structure of opinions was characterized more by clusters of consumption domains than by the distinction of leaders and followers. Three clusters of consumption domains were found with respect to the communication of opinions: (1) clothings, food articles and external care products, (2) vacation trips, audio/video equipment and leisure time/hobbies, and (3) media. The findings suggest that the theory of consumer communication flows should be revised.

INTRODUCTION

It is generally believed that opinion leadership plays an important part in the communication between consumers. The two-step communications model (Katz and Lazarsfeld 1955) distinguishes between opinion leaders and opinion followers. Rogers (1962) considers opinion leaders as individuals who are looked to for advice by many of their peers. Consequently, opinion followers may be considered as individuals who relatively often seek for advice on consumption items from their peers, in particular from opinion leaders. Shiffman and Kanuk (1994) use the term opinion receiver to identify those who actively seek information and those who receive unsolicited information. Although receiving information does not formally imply forming or changing an opinion, we will use the term opinion followership to stress the complementary nature of the concept in relation to opinion leadership. The two-step communication model assumes that opinion leaders are influenced by the media directly (step A1 in figure 1) and pass the information on to the opinion followers (step A2 in figure 1). Troldahl (1971) assumes that opinion leaders also may ask for information from professionals and that they change their opinion relatively often on the basis of this. One problem in this respect concerns the question of who is to be considered a professional or an expert. [For example, a graduate from an art school could be an expert in fashion but need not be considered a professional. We owe this remark to Jan Schoormans.] Opinion followers also are exposed to the mass media (B in figure 1) but relatively often they change their opinion due to advice from the opinion leaders (A2 in figure 1).

Several research questions regarding the concepts of opinion leadership (OLS) and opinion followership (OFS) may be posed. First, one may question whether OLS and OFS are two poles of a one-dimensional concept or whether they are two dimensions which may be correlated. In Troldahl (1971) and Myers and Robertson (1972) the two concepts were measured by only one survey question each. Reynolds and Darden (1971) use a three-items information-seeking scale in the area of clothing fashions. In King and Summers (1970) and Flynn, Goldsmith and Eastman (1994) an OLS scale has been developed but OFS has not been measured directly. From the cited research, the dimensionality of the concepts cannot be determined. To study this question, we include two scales to measure the concepts.

Second, the generality of the concepts should be ascertained. OLS and OFS may not be the same across consumption domains. An individual may be an opinion leader in one domain and an opinion follower in a different domain. Myers and Robertson (1972) clearly find two orthogonal OLS factors, each for different consumption domains. Although King and Summers (1970) and Summers and King (1971) claim considerable overlap of the concept across consumption domains, Sudman (1971) questions the generality of the concept in their research. It should be mentioned that King and Summers (1970) include consumption items that fall in only one domain (the 'endogenous interests' domain) of the Myers and Robertson (1972) analysis. Thus, the overlap found in King and Summers (1970) may be due to a limited variety of consumption items. Since opinion leadership seems to be related to product involvement, and involvement varies with the product concerned, it seems natural to consider opinion leadership in relation to specific consumption domains rather than as a general trait (cf. Muncy and Hunt 1984; Shiffman and Kanuk 1994).

Third, the cultural aspects of the concepts should be considered. Communication flows could be different across cultures or the information from the mass media could be different, inducing different communication structures. Here, two samples from different countries (USA and The Netherlands) will be used to explore the cultural aspect of the concepts.

The next section describes the survey measures and the samples. Then, the results of the analysis will be reported. Finally, a discussion follows.

DATA

Both the OLS and the OFS concepts were measured by means of survey questions. The OLS scale has been taken from Flynn, Goldsmith and Eastman (1994). The OFS scale has been constructed by reframing the OLS items, i.e. giving information has been restated as asking for information generally (see appendix 1). Furthermore, both OLS and OFS have been measured regarding consumption domains of interest to students: (1) media, (2) clothings, (3) food articles, (4) external care products, (5) vacation trips, (6) audio/video equipment and (7) leisure time/hobbies. Several other questions were asked concerning innovativeness, self-monitoring, consumption, media usage and socioeconomic circumstances. The results of these questions will be reported elsewhere.

Two surveys were conducted among students in the Fall of 1994. The first survey included 91 juniors and seniors of a Business College at a Midwest University. The survey was administered in class during a class period. Students were offered credit for participating in the survey. Debriefing was provided after the students filled out the questionnaires. The second survey included 102 Dutch undergraduates in economics at Erasmus University. The survey was conducted during a break in a lecture in the classroom. The students were not familiar with the concepts measured.

RESULTS

It appeared that the two samples were different regarding sociodemographic composition to some extent. The Dutch sample was slightly older, more students lived with their parents and the income was considerably lower than in the USA sample (see table 1). On the other hand, education is much cheaper in the Netherlands than in the USA.

FIGURE 1

ONE-STEP AND TWO-STEP COMMUNICATION FLOWS

TABLE 1

DESCRIPTIONG OF THE SAMPLES

The first analysis dealt with the dimensionality of the constructs of opinion leadership and opinion followership. For each consumption domain, principal components analyses were run separately on the leadership and followership variables in the combined sample. It appeared that each scale was explained by only one factor, the percentage of explained variance ranging from 58-74% (see table 2).

The USA sample showed more opinion leadership than the Dutch sample in the domains of clothings and food articles, and less in the media domain. Also, opinion followership was stronger in the USA sample for the clothings, food and leisure time/hobbies domains. Both opinion leadership and followership seem to be more prevailing among students in the USA than among the Dutch students. This points to a general disposition to communicate rather than to opposite tendencies to lead or to follow opinions. Although this may seem inconsistent with the idea of complementary communication concerning giving and receiving information in a dyad, it points to the possibility of giving information to one person and receiving information from a different person concerning a particular consumption domain.

The correlations between factor scores of opinion leadership and opinion followership are shown in table 3 for each consumption domain. It appeared that the two types of scales were positively correlated in each consumption domain. This result is consistent with Myers and Robertson (1972) and Reynolds and Darden (1971), who found positive correlations between measures of these constructs. A separate analysis, not reported here, dealt with the correlations between the multiple OLS and OFS items in each domain. Two positively correlated factors of OLS and OLS were found. Secondary analysis of these factors yielded similar results as in the remaining part of this paper.

Furthermore, a principal components analysis was conducted on the 14 X 14 matrix of component scores used in table 2. Five components with eigenvalues exceeding one were found. A scree test indicated a drop in the explained variance up to and including the third component. The three components explained 52% of the variance of the component scores. The varimax rotated factor matrix is shown in table 4.

The three factors seem to fall apart according to the type of consumption domain, not according to opinion leadership and opinion followership. This finding indicates that the consumption domain is more important in classifying communication structures than the communication flows. It appears that the first factor deals with communication in the domains of (2) clothings, (3) food articles and (4) external care products. This domain seems to be equivalent to the 'endogenous interests' domain distinguished in Myers and Robertson (1972). The second factor deals with communication in the domains of (5) vacation trips, (6) audio/video equipment and (7) leisure time/hobbies. This seems to be somehow related to the 'exogenous interests' domain in Myers and Robertson (1972). The third factor deals with commmunication in the media domain. Also, it seems to explain some variation in the communication concerning leisure time and hobbies. This factor was not included in Myers and Robertson (1971), although a similar domain of home entertainment (possibly including the media domain) loaded both on the endogenous interests and the exogenous interests factors in their study. These findings do not seem to be related to the domain clusters regarding opinion leadership found in Montgomery and Silk (1971).

TABLE 2

EXPLAINED VARIANCES OF OPINION LEADERSHIP AND FOLLOWERSHIP VARIABLES BY PRINCIPAL COMPONENTS, AND AVERAGE FACTOR SCORES IN THE TWO SAMPLES

TABLE 3

CORRELATIONS BETWEEN OPINION LEADERSHIP AND OPINION FOLLOWERSHIP FACTOR SCORES

Next, the component scores from the analysis in table 4 were compared between the two samples (see table 5). It appeared that significantly more communication concerning clothings, food and external care products occurs in the USA than in the Netherlands. In the other domains, no significant differences were found.

Finally, we look at the media usage associated with opinion leadership and followership. The three factors in table 4 were correlated with the reported number of newspapers read on average each day, the number of magazines read on average each month, and the frequency of listening to the radio and watching television relative to other people (see table 6).

It appeared that the first communication factor, concerning the domains of clothings, food and external care products was negatively related to newspaper reading, whereas it was positively related to radio listening. A possible explanation could be that the newspapers report little news concerning the 'endogenous interests' domain, in contrast to the radio. Communication concerning the media domain was positively related to newspaper reading but not to the usage of the other media. Apparently, newspapers are the most important sources of communication in the media domain. Using the separate principal components of the OLS and OFS scales in each of the seven domains yielded similar results, i.e. opinion leaders tended to use the same type of media as the opinion followers.

TABLE 4

FACTOR MATRIX CONCERNING OPINION LEADERSHIP (OLS) AND OPINION FOLLOWERSHIP (OFS) COMPONENT SCORES IN SEVEN CONSUMPTION DOMAINS

TABLE 5

COMMUNICATION DOMAIN FACTORS IN THE TWO SAMPLES

TABLE 6

CORRELATIONS OF MEDIA USAGE WITH THE COMMUNICATION FACTORS

DISCUSSION

This study attempted at the empirical distinction of the opinion followership (OFS) concept by using a multiple item scale. It appeared that the items could be explained by one common factor which was positively related to the opinion leadership scale. This result is not consistent with the two-step flow communication model in which the leaders are assumed to pass their opinions on to the followers. Shiffman and Kanuk (1994) classify consumers who are both opinion leaders and opinion seekers in a particular domain as socially integrated, those high on opinion leadership and low on opinion seeking as socially independent, those low on opinion leadership and high on opinion seeking as socially dependent, and those low on both dimensions as socially isolated. Their classification is based on arbitrary splits of the distributions of OLS and OFS scores. Our result suggests a common communication factor explaining both opinion leadership and opinion followership. Myers and Robertson (1972, p.41) interpret such a relationship as two-way opinion leadership: ".. people who influence others are themselves influenced by others in the same topic area". One might as well denote this as two-way opinion followership, and the validity of the conceptual distinction of leaders and followers may be questioned. For this reason, the theory of consumer communication flows should be revised.

APPENDIX 1

OPERATIONALIZATION OF THE CONCEPTS

A more fruitful approach to studying communication among consumers concerns the distinction of communication domains. It appears that communication among consumers is not generally related across consumption domains. It seems that consumers specialize in particular areas of consumption and within these areas they use particular media to obtain their information. It appears that in the 'endogenous interests' domain, communication is based on information from the radio rather than from the newspapers. This finding may be used in marketing communication concerning this consumption domain.

In the USA sample, more communication occurs in the 'endogenous interests' domain than in the Dutch sample of students. In the other domains, no such differences were found. Since basically the same communication structure was found in the two samples, this points to the generality of the findings across the two cultures. Concerning the generality of our results, we notice that they are in agreement with results reported in the literature, although the study was limited to samples of students.

REFERENCES

Flynn, Leisa R., Goldsmith, Ronald E. and Eastman, Jacqueline K. (1994), "The King and Summers opinion leadership scale: Revision and refinement," Journal of Business Research, 31, 55-64.

Katz, Elihu, and Lazarsfeld, Paul F. (1955), Personal Influence, Glencoe: The Free Press.

King, Charles W. and Summers, John O. (1970), "Overlap of opinion leadership across product categories," Journal of Marketing Research, 7, 43-50.

Montgomery, David B. and Silk, Alvin J. (1971), "Clusters of consumer interest and opinion leaders' spheres of influence," Journal of Marketing Research, 8, 317-321.

Muncy, James A. and Hunt, Shelby D. (1984), "Consumer involvement: Definitional issues and research directions," in Advances in Consumer Research, 11, Editor Thomas C. Kinnear, Provo, UT: association for Consumer Research, 193-196.

Myers, James H. and Robertson, Thomas S. (1972), "Dimensions of opinion leadership," Journal of Marketing Research, 9, 41-46.

Reynolds, Fred D. and Darden, William R. (1971), "Mutually adaptive effects of interpersonal communication," Journal of Marketing Research, 8, 449-454.

Rogers, Everett M. (1962), Diffusion of Innovations, New York: Free Press.

Schiffman, Leon G. and Kanuk, Leslie L. (1994), Consumer Behavior, Englewood Cliffs, NJ: Prentice Hall.

Summers, John O. and King, Charles W. (1971), "Overlap of opinion leadership: A reply," Journal of Marketing Research, 8, 259-261.

Sudman, Seymour (1971), "Overlap of opinion leadership across consumer product categories," Journal of Marketing Research, 8, 258-259.

Troldahl, Verling T. (1971), "A field test of a modified 'Two-step flow of communication' model," Public Opinion Quarterly, 35, 609-623.

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