Anova repeated measures repeated many times

Question

I would like to quantify the effect of 5 different visual stimuli on a particular behavior in a set of 10 animals.

Each animal sees each stimulus multiple (8) times. Thus, each animal has received 5x8=40 stimulus presentations. For each presentation, I measure a value between 0 to 1, representing the behavior response.

I want to run ANOVA and post hoc multiple comparisons to understand which stimuli are different from each other in terms of eliciting behavior.

So far, I have averaged the data over the repeated presentations of each stimulus for each animal and performed a one way ANOVA and post hoc Tukey's test.

It occurred to me that repeated measures ANOVA would be more appropriate because each animal sees all stimuli, and there appear to be substantial differences between animals in terms of overall responsiveness. I found a post describing how to do this: cross validated. However, I wonder how this analysis might have to be modified to accommodate specifically the fact that I have multiple presentations of each stimulus to each animal. Any hints? I would like to use R or python to analyze the data.

Answer 1

You don't have to do anything to accommodate that. The assumptions are about distributions and the numbers don't remember where they came from. If your raw data are very skewed or have outliers, you could consider using medians instead of means. In research using response times (usually in humans) it is common to use median response time in each condition for each subject as the basis of the analysis.