Hi Dr. Zalesky,
I have 53 participants and I want to use NBS to assess group differences, sex differences, and timepoint differences. I have 2 groups (concussed vs controls), 2 sexes (males vs. females), and 2 timepoints (baseline vs. follow-up).
I initially had wanted to do a mixed-effects ANOVA, however, I realized with repeated measure designs it is not good to have variables that are stable across time like group and sex. Without those two variables I would have only been assessing timepoint, so I switched to a simpler plan first. I did 2 two-sample t-tests to look at group differences and sex differences. The timepoints are where I am getting confused. I want to also assess differences from baseline to follow-up, so I thought I would do a paired t-test.
I have attached the design matrix I created with the subject columns to model the within-subject variability, however, I thought that for the between subjects mean in the last column I would put timepoint but I am not sure if that is right. If it is, I would put 1 and -1 for each participant because they each have a baseline and follow-up timepoint (I put a few examples in the design matrix to show you what I mean). However, I’m unsure if this approach is correct for capturing the time effect, or if there’s a better way to structure the design matrix for this analysis. Thanks in advance!
Hi,
I'm not sure if I follow your ciurrent design matrix. I suggest including a column for group, a column for time and an interaction term (group x time interaction). An additional column would then be included for each subject to model within-subject effects. Note that the group column would need to be removed when modeling within-subject effects.
The contrast for the interaction effect would be [0 1 0 0...] or [0 -1 0 0...] assuming that the group column is removed.
Best wishes,
Andrew
Originally posted by ecastro3:
Hi Dr. Zalesky,
I have 53 participants and I want to use NBS to assess group differences, sex differences, and timepoint differences. I have 2 groups (concussed vs controls), 2 sexes (males vs. females), and 2 timepoints (baseline vs. follow-up).
I initially had wanted to do a mixed-effects ANOVA, however, I realized with repeated measure designs it is not good to have variables that are stable across time like group and sex. Without those two variables I would have only been assessing timepoint, so I switched to a simpler plan first. I did 2 two-sample t-tests to look at group differences and sex differences. The timepoints are where I am getting confused. I want to also assess differences from baseline to follow-up, so I thought I would do a paired t-test.
I have attached the design matrix I created with the subject columns to model the within-subject variability, however, I thought that for the between subjects mean in the last column I would put timepoint but I am not sure if that is right. If it is, I would put 1 and -1 for each participant because they each have a baseline and follow-up timepoint (I put a few examples in the design matrix to show you what I mean). However, I’m unsure if this approach is correct for capturing the time effect, or if there’s a better way to structure the design matrix for this analysis. Thanks in advance!
Hi Dr. Zalesky,
Thanks for your reply, I think I understand better what to do here now.