adding covariates for repeated measures ANOVA
Hello,
I'm trying to do repeated measures ANOVA with 2 covariate measures. I can see that there is a covariate box I can use, but that doesn't really seem to have any effect on the results so I'm starting to question if I'm doing things right.
To explain a bit what I'm doing, I have a set of participants who wrote a text twice. Once it was in a timed condition, while the second time it was not timed. I am trying to see if there is a statistical difference between different aspects of their writing, such as for example the number of connectives that they use. I want to use the wordcount of each text as covariates, as text length is known affect these measures. So I have two wordcount measures, one for each condition. If I just simply add both to the covariates box, this seems to change the p values in the "within subject effects" table, but not in the post-hoc table (which the comparison that interests me I believe). So I wonder if I am doing the analysis correctly. Is there a way to make the analysis take into account that each covariate is relevant to one of the conditions, not both? Or perhaps that does not matter?
I am including a screenshot of my analysis. I would appreciate any guidance. As you can tell, I'm not very familiar with any of this. Thank you so much in advance!
Comments
Hi.
(1) I think it would help if you include the estimated marginal means in your post.
(2) The post hoc test result doesn't make sense to me: It should be redundant with the main effect of "conditions" in the tests of within subjects effects.
(3) I have a hard time following the logic of your ANCOVA: When, for example, you include "R Not Timed WordCount," Your asking the question, "What's the best estimate of what the results would be, had the R condition and the T condition been equal with respect to the R word count?" But that's logically impossible since only the R condition (not the T condition) has an R word count. Thus, covariates that vary between-subjects only, make sense. But covariates that somehow also vary within-subjects don't make sense.
R