No abstract

It is the object of this paper to demonstrate that variation in efficiency of a selective or predictive instrument is a direct linear function of variation of its product-moment correlation with the predicted variable and to indicate possible criticisms of E " (*, p. 275) the coefficient of alienation, the coefficient of determination (r 2 ) and other functions of the correlation coefficient which purport to show efficiency of prediction. While the writer feels that most of the argument and interpretation presented is of general significance, caution should possibly advise limiting the discussion to those prediction problems most typical of psychological research, such as the prediction of success in a job, in an assignment, in training for industrial work, or in academic work. It is quite possible that the interpretation emphasized here has been presented elsewhere in the statistical literature. However, if this is the case, repetition would appear worth while in that current journals and statistical textbooks most often used by psychologists and educators usually present interpretations different from or contrary to that here developed/ 1 ' 2 ' 8 -4 -1 " Since a number of the functions employed to interpret the correlation coefficient have their own legitimate meaning it may be well to indicate some typical comments which are in the author's opinion misleading. It is desired in indicating theae comments to emphasize that interpretation of the correlation coefficient as a direct index of efficiency of a predictor definitely differs from interpretations usually advocated, and that the 65 * While the predictor will in general be referred to as singular, all discussion and comment will apply equally to a weighted or unweighted sum of several predictors or, in other words, to multiple and partial regression.

A RATIONALE for the construction of an overall measure of worker effectiveness is proposed and discussed in the light of some of the major problems of criterion construction.As a result of the proposed approach, the criterion evaluations for the validation study may be selected with assurance that variables rejected have minimal effect on the efficiency of operation of the organization and that those retained provide an accurate estimate of the effect of the individual worker on efficiency of operation of the organization.The rationale proposed involves converting production units, errors, time of other personnel consumed, etc., into dollar units. In so doing we give direct face validity to the individual criterion scales; that is, we show that each scale is directly oriented toward the objective of the organization and insure that validation analysis can be interpreted as estimates of the degree to which predictors contribute to this objective.

No abstract

measurement theory, Rasch model, test theory, ability, inverse logistic,