# Accounting for a covariate that only exists in one of two groups [duplicate]

I am examining how sex affects the relationship between 2 variables. So, my model looks like: $$y \sim sex * x$$ However, there is a known effect of menstrual cycle on $$y$$ (obviously in females only). Is there a way that I could include this as a covariate in the analysis, while still keeping both groups?

Let's say in females I have factor $$m$$ that is either 0 (follicular phase) or 1 (luteal phase), representing where they are in their cycle. Would my model be as simple as: $$y \sim m + sex * x$$ where I would code all men as 0 for the covariate $$m$$?

Would it matter if the covariate was continuous?

• I'd try recoding sex as a 3 level variable – thomaskeefe May 13 '20 at 0:36
• @thomaskeefe I thought about that. Then would have to construct contrasts for the 'interesting' hypotheses (i.e. not interested if the 2 female groups differ) – tomr May 13 '20 at 18:53
• @kjetilbhalvorsen I think that does. The comments alluded to a continuous case, but got a little muddy for me. Regardless, at this point that is not my issue. Thanks, would never have guessed from that title of that post that it would address my question. – tomr May 13 '20 at 18:54