If you mean a contrast with weights of 1 and -1 applied to the two means, it will be equivalent to the F-test for the main effect of that two-level factor (t2=F). But perhaps you have something else in mind. If so, please elaborate.
If the factor has two levels, the model F-test is equivalent to the Wald-t-test of the the coefficient coding the simple effect and also to the post-hoc test or test of the contrast of the two levels.
No need to do any contrasts. If the factor has two levels, you are in fact doing, as another commentator pointed out, a t-test. If the difference between the two levels is significant according to ANOVA results, then that closes the issue (i.e. the two levels are different). There is nothing a contrast could add...unless you have something else in mind.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude