Study of Validity of Ratings

As part of research on validation of the Edwards Personal Preference Schedule, this study indicates that in addition to the traditional well-known constant errors in ratings explored by numerous investigators, there is an error, which along with many other possible errors, may influence rating systems. This specific error, which may be called “projection error,” is a tendency to rate others high (or low) on the traits on which raters themselves are high (or low). It was hypothesized that the 20 male raters' own traits as measured by personality scales are more predictive of the ratings they assign to others than of their own self-ratings on the same traits. Analysis indicated that the relationship between the Edwards scales and the mean rater's ratings could not be considered a function of chance. Most variables of the Edwards schedule were more predictive of the mean rater's ratings than of the self-ratings and of the mean peer-ratings, without denying the predictability of the latter two for some traits. The study also showed that neither self-ratings nor mean peer-ratings were superior with respect to the number and the size of the correlation coefficients.