PSYCH339 Study Guide - Midterm Guide: Criterion Validity, Job Performance
10 Aug 2016
School
Department
Course
Professor
Document Summary
Gma, conscientiousness, integrity = all predictors of job performance. Inter-correlation of predictors (gma, conscientiousness, integrity) should be small because you don"t want them to be redundant with each other. When given a case like this, look for: How correlated it is with job performance (criterion validity) How correlated with it with each other. Predicted performance = b (gma) + b (conscientiousness) + b (integrity) Regression tells us how much to weigh each predictor. So if you run regression, it could tell you that gma = 2, conscientiousness = 1 and integrity = 0. 5. R = more than one predictor and it"s a stronger than each predictor by itself because. Shrinkage if we apply these weights (from above) to other predictors validity will decrease. R (from cross validation) =< r (derivation original sample) This happens because weights are too specified. If samples are large, better off using regression because the regression derivation is more likely able to hold up.
