As a preparation for the stats course that Daniel Lakens and I will run later this year, I tried to create some Stata do-files related to p-curves and p-hacking. Below my attempts (note: unchecked, comments welcome):
- pcurves.do Illustrates that p-values closer to 0.05 are more likely when H0 is not true and rejected.
- phack.do Illustrates the effects of one case of p-hacking: adding participants until significance is reached (or N up to its maximum). This is a low-tech Stata version of Ryne Sherman's post.
Note that the links refer to files in my Dropbox - they may change over time (hopefully for the better).