After Statistics Reform: Should We Still Teach Significance Testing?

In the longer term null hypothesis significance testing (NHST) will disappear because p-values are not informative and not replicable. As with any reform, the question can be asked whether we should continue to teach the procedures of abolished routines (i.e., NHST), not as a commendable practice but as a means of understanding what our (more or less statistically ignorant) predecessors did. Three arguments are discussed for not teaching NHST in (introductory) undergraduate courses in inferential statistics: experience shows that NHST is too difficult for introductory courses; dichotomous thinking inherent to NHST is a cognitive obstacle for interpretation; and students can find relevant information in research reports without knowing NHST.