Winsteps
Wednesday, November 30, 2005
Winsteps 3.59.0
Suggestions for enhancements pour in from Winsteps users. The challenge is to implement them as software, debug them, and then to check that everything else still works. But modern software is so complex that there is no way that all combinations can be checked, so we all become beta-testers, whether we like it or not!
After 20 years, the pressure to include MCQ distracter (or distractor, if you prefer) point-biserial correlations finally became overwhelming, so these are now implemented. They provide indirectly what the "average measures" for the distracters indicate directly: at what ability level the distracter is selected.
For those who want to explore respondent styles, i.e., the way in which they use your rating scale, the new "transposed" output file option swaps the data rows and columns. Now the race is on to find out who can first publish a paper using this feature!
After 20 years, the pressure to include MCQ distracter (or distractor, if you prefer) point-biserial correlations finally became overwhelming, so these are now implemented. They provide indirectly what the "average measures" for the distracters indicate directly: at what ability level the distracter is selected.
For those who want to explore respondent styles, i.e., the way in which they use your rating scale, the new "transposed" output file option swaps the data rows and columns. Now the race is on to find out who can first publish a paper using this feature!
