behavioural economics experiment - non parametric repeated measures manova

Kuba

New Member
#1
hi there, I wonder if sbdy could help me with my dissertation project. I am stuck with the analysis n tried to consult a few people but unfortunately they either have limited experience, knowledge or interest in the subject of statistics/behavioural economics.

Basically, I got a mixed repeated measures design with 2 DV and 2 IV (1 at 2 and 1 at 4 levels). there were 24 participants split between two groups. Each particiapnt performed a task which was measured using two DV between groups. Also, in each group there were four levels of additional IV.

Given that I must use non-parametric tests (non-normal distribution) and i have 2 DV (and so manova) i don;t think i could use the SPSS for this. I am not sure which tests to use and what software (R?). PLease send your comments here, I am also happy to chat over a skype or email.
 

CB

Super Moderator
#2
When you say you have a non-normal distribution, do you mean the marginal distribution of the dependent variable, or the distributions of the dependent variable within each of the 2*4 conditions? (only the latter matters)

If non-normal conditional distributions are the only distributional problem, you don't necessarily need to use a non-parametric analysis. Given that the other RM ANOVA assumptions are met (e.g., independence of errors, sphericity), OLS will still give Best Linear Unbiased Estimates. The sampling distribution of coefficients will not necessarily be normal, which is problematic for significance testing and confidence intervals, but you can use bootstrapping to get around this. I think SPSS allows bootstrapping now, possibly as an add-on module.