PSY-B - Psychology PSY-B 311 Lecture Notes - Lecture 13: John Tukey, Emor, Null Hypothesis
Document Summary
Simplest version: two-groups (2 conditions or levels of iv) e. g. , placebo, low alcohol dose. But experiments can have more than 2 levels of iv e. g. , placebo, low, medium, high alcohol dose (4 levels) One t-test with alpha set at 0. 05 gives 5% (1/20) chance of type i error. The overall chances of making a type i error increases with performing a greater number of tests. e. g. , perform 10 t-tests, the probability of making a type i error becomes 0. 40 (40% chance) Probability of type i error with multiple tests. = 1-(1-alpha)c (where c equals the # of tests performed) Make alpha level for each test more stringent than 0. 05. Bonferroni adjustment: divide desired alpha level (e. g. , 0. 05) by the number of tests conducted (e. g. experiment with 4 comparisons; 0. 05 /4 = 0. 0125; p has to be <0. 0125 for significant effect) But as alphas become more stringent, it increases likelihood of type ii error.