CRIM 320 Lecture Notes - Lecture 7: Null Hypothesis
Document Summary
Crim320 lecture 7: measuring association between dummy and interval/ratio variables. Often we are interested in comparing 2 unrelated/distinct groups. But basic logic as that used in t-tests. Aka anticipated level of error between measurements of the 2 sample means. Degree of certainty about how closely our calculated means (calculated from 2 samples) resemble the means of the 2 respective pops. Larger se or smaller diff between means results in a lower z figure. Using small samples (n<120) requires the use of t-distribution instead of z-distribution. If the calculated z- or t-statistics exceed the critical value . As sample size increases, t- and z-distributions resemble each other. Should at least have 10 obs in each group to make meaningful comparisons: the sum of the 2 sample sizes is greater than 120, both samples have at least 30 obs. Involves more complicated calculation of the se of the diff between means. Pu is an estimate of the proportion in the pop.