PSYC 3130 Lecture Notes - Lecture 6: Ronald Fisher, Null Hypothesis, 2Degrees

• School: University of Windsor
• Department: Psychology
• Course: PSYC 3130
• Professor: Ken Cramer

Document Summary

Comparing two or more independent means: anova (analysis of variance) Anova will work for two groups, but is imperative for more than two groups. Independent group (no correlation to predict one score from the other) Ex) could not be used for mothers, daughters, sons. Anova can compare infinite number of means and an infinite number of independent groups. Find the largest variance, put it in a ratio, divide by smallest one. Under null, assume no difference among means where k is number of groups. The alternative is that there is difference among some of the means (some could be equally as ineffective) Calculating an f statistic (ronald fisher), which is ratio of 2 variances trying to estimate the same thing. Top variance takes group means and estimates. Under null hypothesis, treatment effects are negligible, so expected value of f is 1. F statistics cannot be negative, because variance cannot be negative. The mean of all scores, regardless of group.