PSYC 51a Lecture Notes - Lecture 12: Analysis Of Variance, Type I And Type Ii Errors, Sampling Error
Document Summary
Anova tests for differences among means but uses variance (variances between treatments relative to variances within treatment) to do this. You can use anova when you have more than two samples (between subjects) or two conditions in one sample (within subjects) Conceptually, anova is a generalization of t test. Any time you use a t test, you can use an anova (in fact t2 = f) 2 = ss = ms = mean square n-1 df. Mean square is short for mean squared deviation from the mean. T compares difference between two means (in between subjects design) T = m1-m2 = treatment + sampling error (chance) Anova compares variance due to difference among group means (=variances between treatments relative to within treatment) F = msbetween = mstreatment = treatment + sampling error (chance) F = differences (variance) including any treatment effects. = differences with treatment effects (treatment) + no treatment effects (error)