When I run these two in R, I get different values. I thought that I should get the same values since it just includes their interaction terms.

  • $\begingroup$ Did you check whether you got the same predictions? $\endgroup$ May 26, 2015 at 14:30
  • $\begingroup$ they are the same. I am more interested in analyzing each coefficient to see how well they fit the model. I don't understand why B * C * D is different than C * B * D for the interaction terms, I assumed order doesn't matter $\endgroup$
    – kevin
    May 26, 2015 at 14:33
  • $\begingroup$ I'd suggest checking which main effects made it into each model. $\endgroup$ May 26, 2015 at 14:38
  • $\begingroup$ the main effects all have very similar p-values and slopes. I am interested in the interaction terms. Which have changed drastically. $\endgroup$
    – kevin
    May 26, 2015 at 14:41
  • 1
    $\begingroup$ Not only the slopes have changed, but what the interaction terms represent have also changed. $\endgroup$ May 26, 2015 at 14:53