Anova lectures 11 Flashcards
separate error terms
multivariate approach: the slopes are different. Variability in effect, so maybe the inconsistent outliers caused the interaction etc.
Draw inferences about consistency of the effect. If different people have different effects, outliers drive outcome.
underestimating variability
type 1 error
overestimating variability
type 2 error
test contrast a different way: is it possible?
Yes, Contrasts for two-way within-subject designs can also be tested using a one-sample t-test or paired samples t-test
Take means of contrast scores, preaveraged contrasts. Do a paired samples t test. If you square the t value that you get, it’s the same as the F test.
You can do interaction as well with one sample t test (the mean is the difference score)
what if I want to test the contrast that isn’t repeated measures anova or a paired samples t test with pre averaged means?
Alternatively, analyse the results using a one sample t-test after computing a difference score (aka contrast score) for each participant:
simple effect contrasts
difference between IV1 level a and b only in IV 2 level 1. Free and structered play only in the morning.
WHat stats do you need for within subjects simple effects contrasts?
Just means and F (v1, v2) = …, p>,< .05
what is interaction vs simple effects contrasts?
simple effects is this question vs that question. Interaction contrast is this difference being significantly different from that difference
Three ways to analyse repeated measures contrasts
Repeated measures ANOVA using Yijk
Paired samples t-test using:
- a pair of mean DV for each participant (main effect contrasts), or
- a pair of contrast scores calculated at two different levels of the IV for each participant (simple effect vs. simple effect = an interaction contrast)
One sample t-test using contrast scores (Ycontrast) (can be used to test main effect, interaction effect and simple effect contrasts)
