A post by David A. Kenny and Megan Goldring
The initial ideas for our article, The Extended Social Relations Model: Understanding Dissimilation and Dissensus in the Judgment of Others were developed in the middle of the 1990s, even before one of us was born. Our coauthor, Taeyun Jung, was interested in how much people’s perceptions agree with each other about another person’s personality. He wanted to investigate differences in the level of agreement between those making judgments (the perceivers) on those being judged (the targets). For instance, do Bob and Alice agree more when they rate Ted and less when they rate Carol? To investigate this topic, Taeyun realized that he needed to focus on disagreement. That is, he needed to look at the degree to which perceivers disagree for a given target. He investigated this disagreement in his dissertation, which was completed in 1999.
Dissimilation: Perceivers See Others Differently
In November of 2019, one of us, Dave Kenny, was lecturing at Columbia University about his new book, Interpersonal Perception: The Foundation of Social Relationships. That book discusses how the Social Relations Model (SRM) can be used to answer basic questions in interpersonal perception. The SRM states that a perceiver’s view of a target is based on how a perceiver generally views others (perceiver effect), how a target is generally viewed by the perceivers (target effect), and how a perceiver uniquely views the target (relationship effect). Dave began by discussing the similarity in a perceiver’s judgments of others, something called Assimilation in the SRM. In the audience was another one of us, Megan Goldring. She suggested that instead of focusing on the similarity of a perceiver’s judgments, it would also be interesting to focus on dissimilarity or the extent to which a perceiver saw others differently or what might be called Dissimilation. Thus, Megan, like Taeyun some twenty years earlier, believed that instead of looking at consistency or agreement, it would also be interesting to examine inconsistency or disagreement.
We then began a discussion of how to study Dissimilation that led us to develop a new model. In our conversations on disagreement, we realized that Dissimilation actually refers to two components in the SRM. The first is the target effect, or how a target is generally viewed by the perceivers. We realized that some perceivers base their judgments more on the target effect than other perceivers; some perceivers’ judgments of targets are more strongly predicted by the target effect than other perceivers. We named this individual difference Sensitivity. In essence, Sensitivity implies that some perceivers have more variance in the target effect than others. The second factor determining Dissimilation, called Differentiation, is measured by the standard deviation of the relationship effect, or how a perceiver uniquely views a target. Some perceivers may have greater variability in their judgments (i.e., more variance in the relationship effect) than other perceivers.
Dissensus: Targets Are Seen Differently
In the 1990s, Taeyun had proposed a parallel analysis for the people being judged. He argued that in looking at targets, some raters would show more agreement and others less. He called disagreement about a target Dissensus. He suggested that the perceiver effect, or how a perceiver generally views others, was used more for some targets and used less for others, something that we call Prototypicality. In essence, some targets have more variance in the perceiver variance than do others. He also proposed greater relationship variance for some targets than others, called Volatility.
A New Model: The Extended Social Relations Model
The above analysis suggested a major modification of the Social Relations Model: which we titled the extended Social Relations Model (eSRM). Some perceivers may use the target effect more than other perceivers (i.e., greater Sensitivity) and the judgments of some targets may be based more on the perceiver effect than others (i.e., greater Prototypicality). Moreover, controlling for the target effect may cause variety in responses amongst some perceivers (i.e., greater Differentiation). Additionally, controlling for the perceiver effect may lead to less agreement about some targets than others (i.e., greater Volatility).
This new model has opened the door for the study of perceiver and target moderators of SRM variances. For instance, say a researcher wants to show that female perceivers have more target variance than male perceivers. Such an analysis was not as easy beforehand, but now the researcher can use the eSRM and look at the levels of Sensitivity among female compared to male perceivers. In a similar fashion, if a researcher wants to show that targets who were less familiar to perceivers have lower consensus, they would use the eSRM.
The extended SRM has latent variables, perceiver and target effects, which vary across targets and perceivers, respectively, as well as heterogeneity of relationship variances that vary by perceiver and target. In our article, we develop some initial ideas, but not a complete operationalization of the eSRM framework. Our preliminary solution is relatively simple (at least it seems that way to us!), involving slopes to compute Sensitivity and Prototypicality and standard deviations to compute Differentiation and Volatility. We are optimistic that methodologists will soon discover a parametric solution for the eSRM.
Illustration
To illustrate the new model, we used a dataset gathered by Taeyun in his dissertation. The targets in this study were 40 famous celebrities in the 1990s (e.g., Whitney Houston and Jerry Seinfeld). The perceivers were 160 University of Connecticut undergraduates. Targets were judged on 20 different personality traits. Note that the study was not preregistered, so conclusions from our analyses should be viewed cautiously, although the study does show the potential of eSRM analyses.
Correlations between eSRM Components and Consistency Across Traits
We found that the two parts of Dissimilation, Sensitivity and Differentiation, correlated positively. Thus, perceivers with higher levels of Sensitivity (i.e., strongly viewed targets as other perceivers did) also had greater Differentiation (i.e., viewed targets more idiosyncratically). This correlation of Sensitivity and Differentiation is consistent with prior research that some perceivers use a wider range of the scale than do other perceivers, something called an extreme response style. Prototypicality and Volatility were essentially uncorrelated, indicating that they measured something different.
We further examined the consistency of the eSRM elements across the traits. For instance, if Sensitivity was consistent across traits, then someone who was a good judge for one personality trait would also be a good judge for other traits. The greatest consistency was for Differentiation, and the least was for Volatility, followed by Prototypicality. The consistency for Sensitivity fell in between. The low consistency for Prototypicality and Volatility calls into question whether some targets were generally easier to judge than others; rather it appears that some people are easier to judge for some traits and harder to judge for others.
Relation of eSRM Components with Familiarity and Liking
We also examined correlations of the eSRM components with Familiarity (how much was each perceiver familiar with the targets?) and Liking (how much did each perceiver like the celebrities?). Looking first at perceivers, we found that perceivers who were more familiar with the celebrities showed higher levels of both Sensitivity and Differentiation, the two major parts of Dissimilation. Presumably, perceivers who were more familiar with the celebrities varied their responses more than those who were less familiar with the celebrities.
Then, we examined Familiarity and Liking for each target and used the target effect of Familiarity and Liking. For Dissensus, we found that there was greater consensus for targets who were liked more. Perceiver effects were more noticeable with targets who were generally familiar and liked. For Prototypicality, the correlation with Liking was positive, whereas for Volatility it was negative. The negative correlation for Volatility makes sense in that we would expect less disagreement for those celebrities who are liked. The positive correlation with Prototypicality, albeit only marginally significant, is intriguing, which we discuss below.
Several sources of evidence point to the measure of Prototypicality having poor construct validity. Nonetheless, there is evidence, admittedly exploratory, that the component might be meaningful. There is the finding that the Liking correlates positively with Prototypicality. We had expected a negative correlation in that more Liking would lead to less use of the perceiver effect. However, we found the opposite, which suggests, very preliminarily, that the perceiver effect might reflect the perceiver’s view of the ideal other. Prior literature has shown that the two major parts of the perceiver effect are positivity, a general liking or disliking of targets, and trait-specific effects (Rau et al., 2020). If positivity effects were weaker for some targets and stronger for others, we would expect to see consistency across traits in Prototypicality. However, if Prototypicality were due to trait-specific effects, we would not expect to find consistency. Given the relatively low consistency, the tentative implication then is that Prototypicality taps trait‑specific effects.
Conclusion
The SRM is already a difficult model to understand with all of its terms, but for the eSRM we added eight more, only six of which we have discussed in the blog! Nonetheless, we are optimistic that the development of the eSRM has opened the door to answers to many important questions in the field of personality. In particular, it gives researchers a tool to study factors that moderate the extent to which perceivers agree about a person, the extent to which a perceiver sees others as similar, and the extent to which perceivers see others uniquely.
References
Biesanz, J. C. (2010). The social accuracy model of interpersonal perception: Assessing individual differences in perceptive and expressive accuracy. Multivariate Behavioral Research, 45, 853–885. https://doi.org/10.1080/00273171.2010.519262
Funder, D. C. (1995). On the accuracy of personality judgment: A realistic approach. Psychological Review, 102, 652–670. https://doi.org/10.1037/0033-295X.102.4.652
Jung, T. (1999). A new look at moderator variables of agreement: The role of target standing. Doctoral dissertation. Retrieved from ProQuest Dissertations and Theses. (Order No. 9906550).
Kenny, D. A. (2019). Interpersonal perception: The foundation of social relationships, 2nd ed. New York: Guilford Press.
Rau, R., Carlson, E. N., Back, M. D., Barranti, M., Gebauer, J. E., Human, L. J., Leising, D., & Nestler, S. (2020). What is the structure of perceiver effects? On the importance of global positivity and trait-specificity across personality domains and judgment contexts. Journal of Personality and Social Psychology, in press.doi.org/10.1037/pspp0000278