Rating Scale Content: II. Effect of Rating on Individual Scales¹

Abstract

Summary and Conclusions T his report, the second in a series on the relationship of scale formats to graphic ratings, is concerned with the relationship among the seven scales employed in the study. These were divided into one group that dealt with personal attributes of the individual and another that was work‐oriented. It was found that job traits showed greater agreement between first‐ and second‐level supervisors than did personal traits. The difference is greater in the less structured of the two formats considered. Intercorrelations among job traits were substantially higher on both formats than they were for personal traits. With the exception of “leadership” ratings, raters seemed to be more lenient when rating personal characteristics than job characteristics. An over‐all scale correlated from .89 to .94 with the optimum sum of the seven individual ratings. The maximum zero‐order correlations for the four populations ranged from .75 to .88. The comparison of standard score regression weights indicated that the purely personal scales contributed practically nothing to the variance of the over‐all scales. The value of multi‐trait scales is questioned as is the impact of human relations training of supervisors.