The Social Accuracy Model of Interpersonal Perception: Assessing Individual Differences in Perceptive and Expressive Accuracy

Abstract

The social accuracy model of interpersonal perception (SAM) is a componential model that estimates perceiver and target effects of different components of accuracy across traits simultaneously. For instance, Jane may be generally accurate in her perceptions of others and thus high in perceptive accuracy—the extent to which a particular perceiver's impressions are more or less accurate than other perceivers on average across different targets. Just as well, Jake may be accurately perceived by others and thus high in expressive accuracy—the extent to which a particular target is accurately perceived on average across different perceivers. Perceptive and expressive accuracy can be further decomposed into their constituent components of normative and distinctive accuracy. Thus the SAM represents an integration of Cronbach's componential approach with Kenny's (1994) social relations model. The SAM is illustrated using both a half-block as well as a round-robin design. Key findings include reliable individual differences in several specific aspects of interpersonal perceptions.

Notes

¹Ipsatization of measures across a range of different traits is needed to make this fully equivalent to Cronbach's (1955) measure of differential accuracy. See Kenny and Winquist (2001) for detailed examples and more discussion.

²The normal terminology for the SRM refers to perceivers as actors and targets as partners to help emphasize the dyadic nature of the social interaction present in research such as a round-robin design.

³Estimation of the variance components was conducted in R using the lmer package under restricted maximum likelihood.

⁴See Human (2009) for the development and more extensive discussion of the terms perceptive and expressive accuracy.

⁵Estimates in this example are carried to four decimal places in order to illustrate several mathematical relationships.

⁶Although distinctive accuracy can be estimated for a single perceiver-target dyad, normative accuracy is confounded with the target's actual normativeness. Estimating a single perceiver's normative accuracy requires determining the normativeness of that perceiver's impressions on average across a large number of targets. In other words, how does the perceiver's impressions of others, on average, correspond to the average validity measure?

*p < .05.

**p < .01.

⁷Define β _VN = E(b VN_j ) as the expected unstandardized regression equation predicting the target's validity measures from the normative validity measures (i.e., = b _0j + b VN_j β _mu ) and β _NV = E(b NV_j ) as the reciprocal relationship (i.e., = b _0j + b NV_j V_jk). All expectations are taken across targets. Because = β _mu , β _VN = E(b _VNj ) = = 1. In other words, the expected unstandardized regression coefficient predicting the validity measure from the average response on that validity measure across k measures is exactly 1.00. Consequently the expected unstandardized relationship between the validity measure predicting the average validity measure will be β _NV = E(b NV_j ) = = E(r _{Vkj μk} )². The expected unstandardized relationship between the target's validity measure and the normative profile is, asymptotically, the intraclass correlation coefficient.

⁸ Cronbach (1958) later focused on analyzing components of accuracy separately for each trait; although useful for certain questions and designs, a narrow trait-focused approach precludes examination of target effects. When the analysis is restricted to a single trait, there exists but a single validity measure value for the target on that trait. It is not possible to estimate relationships for the target based on a single predictor value.