This is a session of Topic 8: New approaches to research in statistics education

Session 8A (Friday 13th, 11:00-12:30,   Level 2 - AV Study Room)

Inference in Times of Crisis, Part 1

Organizer

• Rink Hoekstra (The Netherlands) : Session chair

Abstract

The current research crisis has, yet again, put the spotlight on the use and usability of null hypothesis significance testing (NHST). It has been shown that a pragmatic use of this technique has most likely lead to a literature that is filled with a much larger proportion of false positives than many expected (see for example the outcomes of the Open Science Collaboration, 2016). Notably, NHST seems strikingly immune from criticism: even in recent papers NHST is still abundantly used (in sharp contrast to its alternatives, like for example confidence intervals and Bayesian techniques), and in most curricula the amount of time spent on NHST outweighs the attention for its alternatives by a lot. The current crisis makes old discussions and questions surrounding NHST and its alternatives prominent again. How do researchers actually report NHST, and do they do that as selectively as is often claimed? Are alternatives superior? What is the responsibility of statistics teachers in this crisis? In this session the role of NHST and its alternatives in the current replication crisis will be discussed, alongside its implications for our teaching of inferential statistics.

Papers