This is basically a theoretical question which is based on the following R output. I have the following R output which contains both crossed and nested effects.

enter image description here

The group(3 levels) and phase(2 levels) are crossed factors. The rat_id(8 levels) is nested under group. I need to find the all pairwise comparisons for main effects with family-wise confidence level of 95%.

Since the interaction effect between group and phase is not significant , i can use the Tukey procedure to find all pairwise comparisons for those 2 factors.

My question is , when there is a nested effect(like rat_id) , can i use the Tukey procedure ?

My thinking is that i can use it irrespective of the factor is crossed or nested. But i dont know whether my intuition is correct or not. it is really appreciated if someone help me to figure it out.

  • $\begingroup$ You have 26 regression coefficients. Image if you know their true values, for example, $\beta_0 =1.8, \beta_2 = 3.4 ...$, what is your main effects? $\endgroup$ – user158565 Nov 13 '18 at 4:31
  • $\begingroup$ @a_statistician the main effects are the each regression coefficients. isn’t it? $\endgroup$ – Sam88 Nov 13 '18 at 5:14
  • $\begingroup$ @a_statistician So those regression coefficients are the values of main effects isnt it ? so that means irrespective of whether they are nested or crossed factors , then those effect will be a main effect. Am i correct ? $\endgroup$ – student_R123 Nov 14 '18 at 3:56
  • $\begingroup$ It depends on you coding methods or design matrix. But generally the main effect is the linear combination of the regression coefficients, not just linear coefficient itself. $\endgroup$ – user158565 Nov 14 '18 at 4:05

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.