PSYC 218 Study Guide - Final Guide: Homoscedasticity, Sampling Distribution, Effect Size
Document Summary
Degrees of freedom for t-test for correlated/dependent groups= n-1 (n is number of pairs) Shouldn"t confuse statistical significance w/practical significance x1-x2 (standard error of the diff betw means) decreases. Effect size: in situations which metric of dependent measure is arbitrary, use of effect size, a standardized unit to convey the magnitude of a diff in 2 means is used. Matched-pair designs: when used properly can be more powerful than design where subjects randomly assigned to treatment groups: practical consequences. Sample of pair members from at least 1 of the groups is rarely representative of its respective pop. Likely that the groups differ on host of other var. For ea value of x, distribution of y values (conditional distributions) is normal distribution for ea value of y, dist of x is normal dist: 2. Y mean for ea value of x fall on straight line: 3. Homoscedasticity (sd for y values is constant for all values of x and vice versa)
