what about the Alpha Cronbach bug in leven and giw?

bugtest

None

Cronbach Alpha (option -c) was introduced in leven version 1.48 and giw version 0.12 (both release date 2004/10/27). This code was broken until leven version 1.63 and giw version 0.22 (both release date 2006/01/25). Does that mean that all values for Cronbach Alpha calculated with those older releases are wrong? Perhaps not.

The bug reared its ugly head unexpectedly while hacking the leven program that resulted in much larger dialect differences than usual. A faulty test in the code for Cronbach Alpha was harmless as long as string differences are less than 1, which they usually are because of a normalisation. (Unless you added your own normalisation function.) However, this seemed to be a lucky coincidence with the way the compiler works, and not something to rely on. Are you using a compiler that makes the bug harmless? You can test this by running bugtest. If you are lucky, you get this message:

    Bug did not manifest. You are probably safe.

Probably safe. No guarantees.