Abstract:
|
In view of recent mounting criticism concerning over-use of p-values, first in the journal Basic and Applied Social Psychology, followed by statements by the American Statistical Association and the American Psychological Association, this paper discusses an alternative approach to hypothesis testing, via the BIC model selection criterion. In particular, those p-values corresponding to some BIC-based tests are computed, with a view toward seeing how much or how little sense the various levels of p-values make. Specifically, one- and two-sample tests in Normal distributions are considered.
|