This week, a new version of the R package agrmt saw the light of day. I have been contacted because one of the functions in the package didn’t produce the right answers. I really appreciate this (the contacting), because it allowed me to fix the code. It was a matter of mixing up i and j. The first reaction in this case is always the worst: what if I got it all wrong? What if I can’t find the bug? To me, fixing code in my packages is important — not because of the undeniable satisfaction from getting it right — but because it is a small way to give back to the (virtual) community that gave us R and the many packages that come along with it.
The package, by the way, implements measures of agreement (consensus) in ordered rating scales, especially Cees van der Eijk’s (2001) measure of agreement A, but also measures of consensus (dispersion) by Leik, Tatsle and Wierman, Blair and Lacy, Kvalseth, and Berry and Mielke. Moreover, it implements Galtung’s AJUS for R.
Didier Ruedin 