EEB225H1 Lecture Notes - Lecture 13: Test Statistic, Dependent And Independent Variables, Bonferroni Correction
14 Mar 2017
School
Department
Course
Professor
Document Summary
Homoscedasticity equal variances for all populations. Type 1 error increases with multiple t-tests: probability that null hypothesis is falsely rejected. Multiple t-tests means that there is a cumulative chance of falsely rejecting the null hypothesis. Do this with pairwise comparisons that are protected from inflated alpha. Multiple comparisons would cause the t-tests to reject too many true null hypotheses. Tukey-kramer adjusts for the number of tests it becomes more conservative the more tests that are done. Bonferroni also adjusts for number of tests but the way it is calculated makes the statistic less conservative and more powerful than bonferroni. Test statistic is q (very similar to t) Larger critical q than the critical t for the same comparison. Less conservative than bonferroni but still more conservative than multiple t-tests. With the tk method, the probability of making at least one type 1 error through the course of testing all pairs means is no greater than the significance level .
