BIOL 373 Lecture Notes - Lecture 11: Noncentrality Parameter, Ratio Test, F-Distribution
5 Mar 2018
School
Department
Course
Professor
Document Summary
If larger than critical f value, unlikely due to chance, means are not the same. *use the f ratio test to check if you meet the assumptions in order to use a t test, because t test requires equal variance. Failing to reject the null w/ f ratio test means we can use the t test, because the variances are equal* If null is true, f ratio allows the f distribution characterized by 1 and 2. If null is false, f ratio follows the noncentral f distribution characterized by 1, 2, : power analyses assume null is false and so we must calculate phi and use noncentrality (noncentrality parameter) distribution. Before data are collected: chances of observing a significant relationship given how different the sample means are, essentially the variance of true means between groups. Sample size required: specify s2, k, and , guess n (use equation above, calculate phi, determine if power is < or > theoretical.
