Repeated measures ANOVA controlled by co-variable. Why I can not carry out a posthoc in co-variable?
Good afternoon,
I hope everything is going well.
I am writing because I need to carry out some statistical analyses and I am very naive on it. Given that I have a question:
I have carried out a repeated measures ANOVA 2x2x2. I have controled by a co-variable (Autistic Quotient), and I got some significative interactions (p-value < 0,05).
I performed a post-hoc analysis, but it is not possible to do it with AQ, even when I have a significan interacion between the variables "pse_condition * AQ" and "pse_condition * block * AQ". Do you have any clue about why I can not perform that analysis? Is there another way to know where is the significant interaction occurring?
Many thank in advance!! and best wishes for your statistical enterprises,
Mayte
Comments
Post hoc tests are t tests that compare one category-level to another (with respect to mean values of the dependent variable). However, a covariate is an interval/ratio variable and so does not produce categories that can be subject to post-hoc t-testing.
R
Thank you so much for your generous help 😊. I am just wondering what does it means that JASP shows as an output:
pse_condition * AQ -> p= 0,044
pse_condition * block * AQ -> p= 0,012
It is weird to me that shows those interactions because the effect of the co-variable is controlled in all the interactions. Actually, if I remove the AQ covariable from the interface, the p-value from the pse_condition * block, also changes (see the image below).
Thank you a lot in advance,
Mayte Vergara Manríquez
It's not the case that 'effect' of the covariate is controlled in all interactions. It's controlled in all interactions *except*, of course, those interactions that actually include the covariate.
'pse_condition * AQ -> p= 0.044' refers to the interaction between pse_condition and AQ.
'pse_condition * block * AQ -> p= 0.012' refers to the interaction between pse_condition, block, and AQ.
R