Modeling longitudinal rank-order data for a psychological experiment

data analysis modeling strategy
1

Hi,
I am planning an experimental study where I will ask participants to rank different strategies multiple times during a training intervention. The participants will rank-order the strategies {A,B,C,D}. I want to test whether an experimental group ranks these strategies differently over time than the other experimental groups. The data structure will be that each participant will do the ranking five times during the intervention. I have no experience working with this data, so I would appreciate your help, guidance, and discussion.

From this discussion https://discourse.mc-stan.org/t/rank-ordered-logit-model-in-brms/14481/7., One option is to use Thurstonian IRT model. But I feel this modeling approach is meant for other purposes (my gut feeling). I also seen in the comments that an ordered logit model can prove helpful for this kind of data, but I have yet to experience with this. Can I get some guidance from people on this forum?

2

I think you’ll get a much better answer than this, but you may want to consider an “all-possible pairs of subjects in different experimental groups” U-statistic and correlate mean absolute difference in ranks with time. A method like this is described here.

1 Like