Power

We did not have a chance to go into the concept of statistical power today, so here are a few notes as you study for the final exam:

-Power = 1-beta; and beta is your Type II error rate

-the definition of power is the probablility that the test will yield significant results/reject the null hypothesis when it is false.

-a larger error df increases power because the MS error will go down as the error df goes up, resulting in a larger F...which then results in a smaller p value, which increases power.

-Also, regarding MCPs and power and Type I error: if you increase alpha, you also increase power. Alpha is your Type I error rate. Performing some MCPs (like multiple t tests) increases alpha, and therefore increases power. However, note that you should not inflate alpha by using multiple t tests, because you then increase your chance of getting a Type I error (rejecting the null and saying you have significance, when in fact, you do not and should have accepted the null hypothesis)

Know for exam:

So, (1) higher error df (coming from a larger sample) = more power

(2)increasing alpha increases power...said another way below

as you increase your chances of making a type I error, you also increase power

(3) as p decreases, power increases (more power in p = .002 than p = .025) note that this p doesn't stand for power however. this p stands for probability.