PSY 303 Lecture Notes - Lecture 13: Froot Loops, Cookie Crisp, Psych
Document Summary
A contrast is an analysis of specific variance. Portion of bgv accounted for by a specific predicted pattern of means. The obtained means are weighted based on what was predicted. To the extent that obtained patterns of means corresponds to the predicted pattern, the contrast will be strongerwhen we are rightstronger test and rewarded for being right. When we are off, punished by it. Estimate predicted group means based on hypothesis. Subtract the prediction g< from each prediction group mean. These contrast weights should sum to 0: ex: gn=20, fl=0, cc=-20. It accounts for a specific portion of the bgv. Now we have sum of squares of contrast. Contrasts are part of between group variance. Contrast degree of freedoms is always 1. Our predictions are better than no predictions. Contrast though, because it is a test of between group variances, this general. In order to use, should come up with predicted means pattern can be accounted for.
