Individual Differences As Moderating Variables: Issues in the Development and Use of Personality Variables

David J. Moore, University of Michigan
[ to cite ]:
David J. Moore (1995) ,"Individual Differences As Moderating Variables: Issues in the Development and Use of Personality Variables", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 111-112.

Advances in Consumer Research Volume 22, 1995      Pages 111-112

INDIVIDUAL DIFFERENCES AS MODERATING VARIABLES: ISSUES IN THE DEVELOPMENT AND USE OF PERSONALITY VARIABLES

David J. Moore, University of Michigan

The recent resurgence of interest in personality research in consumer behavior has been a result of careful conceptualizations and a stronger commitment to rigorous measurement (Bagozzi 1993). Indeed, the real contribution to be derived from personality concepts lies not in their power as predictors of main effects but rather in their power as moderator variables (Bagozzi 1993; Haugtvedt, Petty, and Cacioppo, 1992). As more and more consumer behavior researchers recognize the potential of personality variables in theory testing and applied research, it is important that certain basic guidelines and perspectives be made salient. Thus, one objective of this session was to provide researchers with examples of the development and the use of personality variables in a manner that enhances understanding of the basic underlying processes.

In the first paper of the session "A General Approach for Representing Personality Constructs," Richard Bagozzi outlined the advantages and the limitations of representing a personality variable as a unidimensional or multidimensional construct. Results of analyses on the need for cognition construct (Cacioppo & Petty, 1982) were used to illustrate a unidimensional scale. Analyses on the Carver and Scheier's self-conciousness measure were used to illustrate a multidimensional scale approach. In addition to these illustrations, Bagozzi's presentation included a discussion of four levels of abstraction at which a given personality construct may be modeled.

The total aggregation model is the level at which the personality construct is represented as either a single composite made up of the sum of several items measuring that construct, or a single factor with each measure loading on that factor. The need for cognition was shown as a representation of total aggregation. The main advantages of the total aggregation model are its simplicity and its ability to capture the essence of the underlying meaning of a personality construct. On the other hand, the disadvantage is that it fails to represent the unique properties of subdimensions, if any, and overlooks both the differential dependence and the effects of subdimensions on other constructs of theoretical interest.

A second level of representation is the partial aggregation model where the dimensions of the construct are not formally modeled as indicators of it but rather are treated as separated subscales loosely tied to the overall construct, termed the discrete components case. Two approaches can be taken with discrete components. Either the items within the components are summed to form a total subscore or else the components are treated as individual factors with their respective items loading on them. The self-monitoring scale has been treated this way, where three lowly correlated dimensions have been identified: extraversion, other-relatedness, and acting ability.

The third level of representation is the partial disaggregation model. Here each dimension of the scale can be represented by a unique latent variable measured by a composite of items. The main advantage of this approach is the ability to specify and test for the existence of multiple dimensions of a personality construct. One unresolved issue associated with this approach is the method to be used to group items to form composite indicators.

The fourth level of representation is the total disaggregation model which is similar to the partial disaggregation model, but instead of using indicators comprised of composites of items, each item is employed as a measure of its respective dimension in a confirmatory factor analysis. In practice, this model can be unwieldly because of likely high levels of random error in typical items and the many parameters that must be estimated.

As illustrations, two well-known scales in consumer research were examined with the proposed framework. For the need for cognition scale, the 18 items from the shortened version were treated as indicators of a single latent variable in a confirmatory factory analysis on a sample of 1700 respondents. The model fit well and was found to generalize across gender, education, and age. For the self-consciousness scale, two composite indicators were formed as the sum of appropriate items on each of three subdimensions: private self-consciousness, public self-consciousness, and social anxiety. The partial disaggregation model was found to fit the model welll and to generalize across samples of Americans and Japanese consumers.

Having been exposed to Bagozzi's general framework for representing personality constructs, two relative new individual difference measures were presented- the Propensity to Self-Reference scale and the Affect Intensity Measure (Larsen and Diener 1987). Curt Haugtvedt introduced the Propensity to Self-Reference individual difference measure, a personality construct developed for use in attitude change and persuasion research. Theoretical justification for the scale was consistent with the perspectives outlined in a JCP article by Haugtvedt, Petty, & Cacioppo (1992)-that is, the idea that useful individual difference measures can be developed that mirror the impact of situational manipulations. The PSR scale was thus developed to reflect the kinds of processes associated with situational manipulations of self-referencing (e.g., Burnkrant & Unnava, 1989). His presentation included a brief discussion of the psychometric properties of the scale and the results of a study in which the scale has been used. Interestingly, this study revealed a curvilinear pattern of self-referencing effects. That is, the highest degree of message elaboration was oberserved at moderate levels of self-referencing. The results of the study thus have implications for resolving the seeming inconsistency between results previously reported by Burnkrant & Unnava (1989) and Sujan, Bettman, & Baumgartner (1993). That is, Burnkrant and Unnava (1989) found high self-referencing to be associated with high levels of elaboration whereas Sujan et al. (1993) found high levels of self-referencing to be associated with the lowest levels of elaboration. In summarizing his presentation, Haugtvedt discussed how self-relevant thinking was posited to play important and unique roles in persuasion processes and suggested that future consumer research might benefit by giving consideration to the theoretical as well as practical value of well-defined unidimensional personality constructs. He concluded his presentation by discussing research projects underway in which he is attempting to replicate the curvilinear pattern using three levels of a situational manipulation as well as assessment of attitude strength consequences of self-referencing processes.

The final paper of the session,"Emotional Intensity: An Individual Difference Perspective" by David Moore, demonstrated how the Affect Intensity Measure (AIM) can be used as an important moderator in explaining consumer response to advertising stimuli. The AIM is defined at the upper end of the continuum by those who experience their emotions with fairly strong intensity and by those at the lower end of the continuum who experience their emotions mildly with only minor emotional responses (Larsen 1984). Even when exposed to similar levels of emotional stimulation, high AIM individuals are expected to manifest more intense emotional reactions than their low affect intensity counterparts, regardless of the hedonic tone of the emotion evoked (Larsen, Diener and Emmons 1986). However, when the stimulus is neutral or non-emotional, the significant differences between high and low AIM individuals are expected to disappear.

To offer further confirmation of the affect intensity hypothesis, Moore (1994) randomly exposed subjects to six public service TV ads (three negative emotional ads and three non-emotional). The results can be summarized as follows: (1) Significant differences in the intensity of emotional responses between high and low AIM subjects occurred only in response to the three emotional ads, but not in response to the non-emotional ads. (2) When exposed to emotionally charged advertising appeals, high AIM individuals reported significantly higher levels of emotional discomfort and lower levels of enjoyment than their low AIM counterparts; (3) Affect Intensity was significantly related to the affective component of the empathy scale, but showed no relationship to the cognitive component of the empathy contruct (Davis 1983). (4) High and low AIM subjects showed no significant differences in attitude toward ads which stimulated emotionally uncomfortable feelings. These results are consistent with previous findings of Larsen et al. (1987).

The studies reported by both Moore and by Haugtvedt demonstrated the fact that individual difference variables can offer additional insight to consumer researchers when these variables are linked to a rigorous theoretical framework. The work presented at this session underscored the notion that the real contribution to be derived from personality concepts lies not in their power as predictors of main effects, but rather in their role as moderator variables (Bagozzi 1993; Haugtvedt et al. 1992). The future contribution of personality research can be significantly enriched if more attention is given to the refinement of the dimensional structure of these scales along the lines suggested in Richard Bagozzi's presentation.

REFERENCES

Bagozzi, Richard P. (1993), "Assessing Construct Validity in Personality Research: Applications to Measure Self-Esteem, Journal of Research in Personality, 27, 49-87.

Baumgartner, Hans, Mita Sujan and James R. Bettman (1992), "Autobiographical Memories, Affect and Consumer Information Processing", Journal of Consumer Psychology, 1 (1), 53-82.

Davis, Mark, H. (1983), "Measuring Individual Differences in Empathy: Evidence for a Multidimensional Approach," Journal of Personality and Social Psychology, 44, 113-126.

Haugtvedt, Curtis P., Richard E. Petty, John T. Cacioppo (1992), "Need For Cognition and Advertising: Understanding the Role of Personality Variables in Consumer Behavior," Journal of Consumer Psychology, 1 (3), 239-260.

Larsen, Randy J. and Ed Diener (1987), "Affect Intensity as an Individual Difference Characteristic: A Review," Journal of Research in Personality, 21, 1-39.

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