I'm using the fwdmsa package to identify deviant cases in a Mokken scale analysis. I've run into a problem., separate from the one I posted previously. The problem comes with items that are "easy" by IRT standards. A good scale should include a range of difficulties; yet when I include "easy" items in a forward search I continuously run into the problem that these items demonstrate no variability in the smaller subsamples selected for forward search. I can resolve the problem by only running the analysis without the easier items, but this is not ideal for for the analysis.
The data are 150 responses to a 37 item test. For the first 15 items, they look like this (the variable names are cumbersome and so have been removed) : > head(by364.data) 1 1 1 1 0 1 0 1 1 0 1 0 1 1 1 1 2 1 1 1 0 1 1 1 1 1 1 0 0 1 1 1 3 0 1 1 1 0 0 1 0 1 1 1 0 1 1 0 4 1 1 1 0 0 1 1 1 1 1 1 1 0 1 1 5 1 1 1 1 0 1 0 1 1 1 1 0 1 1 0 6 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 You can see that items 1, 2, 9 and 13 demonstrate no variance in this small sample and thus muck up the analysis. This happens even when I have items with plenty of variance, but seeing as they are easy items, they might have a .8 - .9 probability of endorsement, so even in samples of 25-50, still run into this variance problem. I have seeded the forward search with specific cases that create the necessary variance, but these artificial subsamples tend to loaded with deviant observations and thus are excluded after a couple of steps, creating the lack of variance problem. Any suggestions? Robert -- View this message in context: http://r.789695.n4.nabble.com/fwdmsa-package-Error-in-search-normal-X-samp-verbose-FALSE-At-least-one-item-has-no-variance-tp4493531p4493531.html Sent from the R help mailing list archive at Nabble.com. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.