A PSYCHOLOGY JOURNAL BANS THE USE OF SIGNIFICANCE TESTS. I have posted, including here and here, about how scientists can rig statistical tests to make it easier to show that a result is “statistically significant” (that is, that the result would occur by chance only 5% of the time). Now, a journal, Basic and Applied Social Psychology, has banned the use of the null hypothesis significance testing procedure and related statistical procedures from their journal. This article by Steven Novella at the Science-Based Medicine website notes that this is an important development because the use of the null hypothesis significance testing procedure is “almost-universal”.
Unfortunately, there is no generally accepted alternative to null hypothesis significance testing. I came across the link to Novella’s article in a post by Tyler Cowen at the Marginal Revolution site. The comments on Cowen’s post discuss what can be done to enable the reader to form a judgment about a study. One recommendation, which is supported by the journal, is to increase sample sizes (I posted here that baseball fans are well aware of small sample problems).