Study for finals: Psychometrics

My notes read:

When you use the test as a predictor, your calculation for a true positive will change, the rest will remain the same. This is the only cell that will change, the rest will be calculated exactly the same.

Holy shit, um, what?

By adding assessment information, you are going to calculate the Probability of True positive = Base Rate multiplied by the selection rate + the ratio of your test score to the outcome measure multiplied by the square root of the Base rate time 1 minus the base rate times the selection rate times 1 minus the selection rate.

P(TP) = BR x SR + (rXY) square root of [BR(1 – BR) SR (1-SR)]

And this is why I will never work in I/O psychology, I don’t care HOW much money they make.

So the only time you’re going to have a larger shift to a true positive is when you have a really large correlation between your test and the eventual outcome measure.

Kill me.