This paper is from Session 8B: Inference in Times of Crisis, Part 2
which comes under Topic 8: New approaches to research in statistics education
Paper 8B2 (Tuesday 10th, 14:00-15:30)
Using “statcheck” to Detect and Prevent Statistical Reporting Inconsistencies
Presenter
- Michele Nuijten (Tilburg University, The Netherlands)
Abstract
Roughly half of the articles in psychology contain at least one inconsistent NHST result in which the p-value does not match the test statistic and degrees of freedom. In one in eight articles, these inconsistencies may actually lead to a different statistical conclusion. To detect and correct these kind of inconsistencies, we developed the free tool “statcheck”. statcheck is an R package with accompanying web app (http://statcheck.io) that automatically extracts statistics from paper, recalculates the p-values based on the reported test statistic and degrees of freedom, and checks if the result is internally consistent. If we teach our students to incorporate statcheck in their workflow, they can avoid statistical misreporting in their own work and easily scrutinize the statistical validity of conclusions in the literature.