I have content-analysis data that are time series (approx 30 equal-interval time points per case), so the data are temporally dependent. This violates the assumptions for any standard intercoder-reliability index… does anyone know of a way to calculate reliability for non-independent data? Or would you want to collaborate to try to figure out a way?
Right now what I’m doing is sampling every 10th data point to reduce dependencies. I could also calculate difference scores between t and t-1 points to remove some of the autocorrelation. Any better ideas?
Viewing 1 post (of 1 total)
The forum ‘Default Forum’ is closed to new topics and replies.