Lilianna Jarmakowska-Kostrzanowska
Interdisciplinary Center of Modern Technology, Nicolaus Copernicus University in Toruń
Abstract:
Statistical analyses in a vast majority of psychological research are based on null hypothesis significance testing (NHST) – a several-step procedure which is considered an invalid mixture of two incompatible approaches – one of Fisher, the other of Neyman–Pearson. Despite that many advocates abandoning NHST, a lot of researchers, psychologists among them, still use this flawed routine. The article presents ideas, which NHST is based on, and also controversies around its most famous construct, p-value. It also discusses objections against p-value (e.g., dependence on sample size) and ways of interpreting it (e.g., conditional probability). The main task of the article is to present the basic concepts of NHST in an understandable manner for Polish social science researchers.
Keywords: null hypothesis significance testing (NHST), p-value, hypothesis testing
Cite this article as:
Jarmakowska-Kostrzanowska, L. (2016). In the statistical matrix: Null hypothesis significance testing and p-value controversies. Psychologia Społeczna, 39, 458–473. doi: 10.7366/1896180020163906