Hi all,

I am analyzing a learning experiment, and I am having trouble building a comprehensive analysis that I think readers/reviewers would find accessible. I was hoping for some suggestions or other insight.

The dependent variable for this experiment is the percent duration a correct response was made during a repeated measurement period (trial). The DV was originally scored in seconds, but is easier to interpret in terms of percent correct. I also have a group factor with 1 control group and 8 experimental groups. I am analyzing my data with a repeated measures linear regression through a mixed linear model approach. The regression is PercentCorrect ~ Group + Trial + Group * Trial. These parameters are all important to the theory and questions I am asking. It is simple to include my control group in the intercept, and thus the rest of the regression equation refers to difference from the control group. However, I lack a convincing way to compare the 8 experimental groups to each other. I need to consider both the main effect and the Group * Trial interactions. Generally, there appears to be little differences across group, nor is there much difference in Group * Trial interaction, but the pairwise tests seem to be needed to be comprehensive, and I can already imagine getting that comment from reviewers for this and similar experiments. I would prefer keeping this analysis in a regression framework as this seems to be much easier to interpret and relate to the graphs than a repeated measures ANOVA. I am conducting my analysis in the StatsModels package for Python. I also have access to R and SPSS, but I am less familiar with these tools. 

I appreciate any considerations you may have. Thanks in advance.

More Christopher A Varnon's questions See All
Similar questions and discussions