PSY 303 Lecture Notes - Lecture 12: Null Hypothesis, Statistical Significance, Psych
Document Summary
1/27/17: we want more bgv then wgv. When is the variance between group means not due to only/mostly the variance between group means (error) What is an ideal situation of ratio between bgv and wgv: group means will always be slightly different (that is there will always be between group variance. ) Individual differences, transient states, etc: group means could be very different due to error even if there is no treatment effect. But such differences would be unlikely: say give everyone exact same chocolate you get varying group means---it is just unlikely, anova and statistical significance. The assumed probability that the group means differ based on error variance (ie chance) alone. Different from coefficient alpha (used to measure scale reliability) The amount of risk of making a false positive mistake (cid:449)e(cid:859)re (cid:449)illi(cid:374)g to take. By (cid:272)o(cid:374)(cid:448)e(cid:374)tio(cid:374), s(cid:272)ie(cid:374)(cid:272)e is (cid:858)(cid:272)o(cid:373)forta(cid:271)le(cid:859) (cid:449)ith 5% error alpha.
