Research Seminar
Examples of Bayesian hypothesis testing in psychologyTom LodewyckxKU Leuven | |
| Abstract: | A key to progress in psychology is the ability to evaluate theoretical ideas quantitatively against empirical observations. The Bayes factor is a model selection tool that formalizes the support for one statistical model when comparing it to another model. This tool allows the researcher to test hypotheses about parameters in general statistical models, or to compare specific models representing competing theoretical frameworks. Defined as the ratio of the marginal model likelihoods, the Bayes factor is conceptually similar to the likelihood ratio test. Analytically intractable integrals make it impossible to derive exact solutions for the Bayes factor for most model comparisons. Therefore, various estimation methods have been developed. We will give a short overview of some existing Bayes factor estimation methods and illustrate them with real data examples. In a first example, we search for the optimal time lag in a time series model for a psychophysiological process. In the second example, non-nested models are compared using the “Mass At Chance” model to test for visual discriminability (Rouder, Morey, Speckman & Pratte, 2007). In the final example, we illustrate how experimental effects in visual discrimination can be tested for in a hierarchical model (Zeelenberg, Wagenmakers & Raaijmakers, 2002). |
| Date: | Tue May 25, 12:00 pm - 1:00 pm |
| Place: | room 02.51 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |