0

I have an experiment in which I apply a treatment (g), and measure variables X and Y; the response variable (density; Z) is simply X/Y. If I wanted to know if X or Y, when shifted under the treatment, has a bigger influence on shifting Z under the treatment, could I use an ANCOVA treating X and Y as covariates? If so, how would I write the model for this in R?

James Fischer
  • 1
  • It sounds like you're expecting a treatment to change the definition of density to something other than mass/volume. – Dave Mar 12 '20 at 13:48
  • The treatment changes Z (density), through changes in both X (number of objects) and Y (area of study). I would like a statistical representation of saying "hey Z changes and it's because of a change in X more so than Y", if that makes sense. – James Fischer Mar 12 '20 at 13:58
  • That's just the algebra of $Z=X/Y$. There is nothing statistical going on. There is nothing different about how $X$ and $Y$ influence $Z$ in the treatment group compared to the control group. $Z=X/Y$ is simply the way you define density. – Dave Mar 12 '20 at 14:26

0 Answers0