Illustrating Bayesian evaluation of informative hypotheses for regression models

Frontiers in Psychology

Herbert eHoijtink¹  Anouck eKluytmans¹  Rens eVan De Schoot¹ 

[1] Utrecht University;

关键词: effect size;    Bayes factor;    inequality constrained hypotheses;    informative hypotheses;   

DOI  :  10.3389/fpsyg.2012.00002

来源: DOAJ

【 摘 要 】

In the present paper we illustrate the Bayesian evaluation of informative hypotheses for regression models. This approach allows psychologists to more directly test their theories than they would using conventional statis- tical analyses. Throughout this paper, both real-world data and simulated datasets will be introduced and evaluated to investigate the pragmatical as well as the theoretical qualities of the approach. We will pave the way from forming informative hypotheses in the context of regression models to interpreting the Bayes factors that express the support for the hypotheses being evaluated. In doing so, the present approach goes beyond p-values and uninformative null hypothesis testing, moving on to informative testing and quantification of model support in a way that is accessible to everyday psychologists.

【 授权许可】

Unknown