This model is introduced in Chapter 4 of Interpersonal Perception: The Foundation of Social Relationships and is also discussed on Chapters 5, 6, 7, and 10. Appendix C contain the technical details of the model.
The original version of the PERSON Model was a model called the Weighted Average Model or WAM that was published in 1991 in Psychological Review. In 2004 in a Personality and Social Psychology Review paper, I published the PERSON Model paper, which essentially re parameterizes WAM, changing correlations to variances. In the 2020 book, I make several modification, which are described below. On page 98, the history of the model is reviewed, including something about a couple of gin and tonics.
A perceiver is assumed to view a target who enacts a series of discrete acts. To each act, the perceiver evaluates that target and assigns a scale value. Moreover, the perceiver attaches a scale value to the target's appearance. As the perceiver views more acts, the influence of the appearance declines. Nonetheless, the impression might well still be correlated with the appearance scale value if there is a positive kernel of truth.
Terms & Standard Values
P or Personality
E or Error
R or Residual
S or Stereotype
O or Opinion
N or Norm
In addition, n is the number of acts viewed by a perceiver and q is the portion of those n acts that are viewed in common. by two perceivers who each view n acts. There are two kernel-of-truth parameters. One is the correlation of P with S and is denoted as m and the other is the correlation of O with R and is denoted as m. The last parameter in the PERSON Model is a, which measures the mutual influence between two perceivers who communicate with each other about a common target.
P is defined by the shared variance between two different acts' scale values viewed by two different perceivers. O is defined by the shared variance between two different acts viewed by the same perceiver, excluding P. N is defined by the shared variance between the same act, excluding P. E is defined by the unique variance of act. These four PERSON components are assumed to be independent and the variance of any given act is P + N + O + E. The S term is defined by the shared variance between the appearance scale values of two different perceivers. R is the unshared variance in appearance scale values. The total variance due to appearance is S + R. Note that consensus is created by P, N times q, and S.
The standard values of the PERSON variances, which are different from the 2004 paper, are as follows:
The rationale for these values are given on pages 343-344. Note also that m and b are defined as correlations, whereas in the 2004 paper they were defined as paths.
At zero acquaintance, consensus equals S/(S + R) = .20. As n gets very large, consensus equals P/(P + O) = .40. Thus, consensus goes from .20 to .40. However, most of that increase is quick, hitting .35 after just 26 acts. The graph of consensus as a function of acquaintance in on page 104, Figure 4.3.
PERSON uses P as the measure of the truth. The accuracy of judgment generally increases with greater acquaintance. However, this is modified by the kernel-of-truth parameter m. As m gets smaller, acquaintance has a greater effect on this increase. It is possible, but perhaps unlikely, if m is very positive, accuracy could decrease as a function of acquaintance. On page 135, the estimate of the kernel-of-truth correlation or m is .36.
The target effect is highly stable, with correlations in the mid to high 90s. The combination of relationship and perceiver effects is lower, but still very high, being in the mid 90s. If we allow for change in O over time, the correlation can be lowered considerably.
Kenny, D. A. (1991). A general model of consensus and accuracy in interpersonal perception. Psychological Review, 98, 155-163.
Kenny, D. A. (2004). PERSON: A general model of interpersonal perception. Personality and Social Psychology Review, 8, 265-280.