Uses, Abuses and Misuses of Significance Tests in the Scientific Community: Won't the Bayesian Choice Be Unavoidable?

Abstract

Summary The current context of the “significance test controversy” is first briefly discussed. Then experimental studies about the use of null hypothesis significance tests by scientific researchers and applied statisticians are presented. The misuses of these tests are reconsidered as judgmental adjustments revealing researchers’requirements towards statistical inference. Lastly alternative methods are considered. Consequently we automatically ask ourselves “won't the Bayesian choice be unavoidable?”

The use of significance tests by psychologists, from normative, descriptive and prescriptive viewpoints

“…Par ailleurs, l'examen des publications actuelles, tant en statistique mathématique qu'en statistique appliquée, révèle une évolution considérable vers les méthodes bayésiennes. L'utilisation de ces méthodes devrait donc devenir un défi motivant pour la communauté scientifique [Lecoutre B, Lecoutre M.-P., 2001].…”

“…Nous n'évoquerons ici que les principales en nous limitant aux références les plus anciennes (cf. [Lecoutre B., Lecoutre M.-P., Poitevineau, 2001] ou [Lecoutre B., Poitevineau, 2000] pour une bibliographie plus détaillée).…”

“…La proposition précédente est souvent associée aux méthodes bayésiennes qui sont d'ailleurs souvent préconisées par les mêmes auteurs (voir aussi [Hoc, 1983], [Lecoutre B., 1984], [Lecoutre B., Lecoutre M.-P., Poitevineau, 2001], [Rouanet, Bernard, Bert, Lecoutre B., Lecoutre M.-P., Leroux, 2000], [Rouanet, Lépine, PelnardConsidère, 1976]). …”

Conceptual and methodological issues in comparative neuroscience and psychology: a reassessment

“…Lecoutre, Lecoutre, and Poitevineau 2001), comparative studies face another difficulty, namely that the species that they include in the analyses must be comparable. In this regard, the first analysis of Maclean et al (figure 2 in Maclean et al paper) has a limited biological interest for analysing the neural basis of self-control.…”

Challenges for Metabolomics as a Tool in Safety Assessments

“…Statistical evaluations of the compositional data have typically utilized classical frequentist significance testing. There are, however, several features of significance hypothesis testing that impact its application to compositional comparisons between crops with different agronomic qualities (Lecoutre, et al, 2001). Berger (1985), for example, stated, "We know from the beginning that the point null hypothesis is almost certainly not exactly true, and that this will always be confirmed by a large enough sample.…”