Download this Paper Open PDF in Browser

Rechtstreeks Verwachtingen Evalueren of de Nul Hypothese Toetsen? Nul Hypothese Toetsing Versus Bayesiaanse Model Selectie (Directly Evaluating Expectations or Testing the Null Hypothesis: Null Hypothesis Testing Versus Bayesian Model Selection)

8 Pages Posted: 29 Aug 2011  

Rens van de Schoot

Utrecht University - Department of Methodology and Statistics

Herbert Hoijtink

Utrecht University - Department of Methodology and Statistics

Date Written: August 29, 2011

Abstract

In veel psychologische artikelen wordt klassieke nulhypothesetoetsing (NHT) gebruikt om onderzoeksvragen te beantwoorden. De resultaten kunnen echter onbevredigend zijn. Rechtstreeks de verwachtingen evalueren zou beter zijn, maar is niet mogelijk met NHT. We laten zien wat de nadelen zijn van NHT en hoe het beter kan, namelijk met Bayesiaanse modelselectie die we introduceren voor nietstatistici.

Researchers in psychology have specific expectations about their theories. These are called informative hypothesis because they contain information about reality. Note that these hypotheses are not necessarily the same as the traditional null and alternative hypothesis. Many researchers use traditional null-hypothesis testing to evaluate informative hypotheses. However, this can be problematic, as will become clear in this article. We offer an innovative solution to evaluate informed hypotheses based on Bayesian Model Selection. The method is introduced in non-statistical terms and its utility is illustrated by applying it to examples from occupational health psychology.

Notes: Downloadable document is in Dutch.

Keywords: nul hypothese toetsing, bayes, model selectie, informatieve hypothese

Suggested Citation

van de Schoot, Rens and Hoijtink, Herbert, Rechtstreeks Verwachtingen Evalueren of de Nul Hypothese Toetsen? Nul Hypothese Toetsing Versus Bayesiaanse Model Selectie (Directly Evaluating Expectations or Testing the Null Hypothesis: Null Hypothesis Testing Versus Bayesian Model Selection) (August 29, 2011). Available at SSRN: https://ssrn.com/abstract=1919023 or http://dx.doi.org/10.2139/ssrn.1919023

Rens Van de Schoot (Contact Author)

Utrecht University - Department of Methodology and Statistics ( email )

P.O.Box 80140
3508 TC Utrecht
Netherlands

HOME PAGE: http://staff.fss.uu.nl/agjvandeschoot

Herbert Hoijtink

Utrecht University - Department of Methodology and Statistics ( email )

P.O.Box 80140
3508 TC Utrecht
Netherlands
030 253 9137/4438 (Phone)

Paper statistics

Downloads
84
Rank
257,411
Abstract Views
573