PSYCH339 Lecture 11: Lecture 11
Document Summary
Rx"y" = rxy / sqrt rxx ryy (but we don"t correct for rxx cause we will overstate for validity, to get rid of that error is to design a better test) Trick to high validity: predictors highly correlated with criterion and not highly correlated with each other. Using multiple regression you find r = 0. 88 in this sample. You would choose test b to test along with gma testing because it has the lowest correlation together (0. 01) Key to high utility is high validity and low selection ratio. Dilemma: don"t know hot good they are until we hire them. Solution: try to predict how good they will be. Information available at the time of hire: individul differences. Knowledge and skills are dependent on abilities/personality traits which all lead to how well you perform on the job. I know it"s the right tool to use. Professions with a great deal of training (graduate school, professional school)