This 4-day course discusses the evaluation of theory-based hypotheses using p-values, the Bayes factor, and information criteria. There will be attention for contemporary phenomena, like publication bias, questionable research practices, the replication crisis, the statistical evaluation of replication studies, and studies in which multiple data sets are used to evaluate the same research question. The course will be non-technical in nature, that is, it is targeted at students and researchers who want to use the approaches presented for the evaluation of their own data.
The evaluation of hypotheses is a core feature of research in the behavioural, social, and biomedical sciences. In the last decade, there has been a lot of attention for inappropriate use of hypotheses testing by journals (publication bias) and authors (questionable research practices) as the main causes of the replication crisis (see, for example, Open Science Collaboration, 2015). This course will use different perspectives (classical, Bayesian, information theoretic) to teach participants in a non-technical manner the theory underlying hypothesis evaluation and the appropriate application of hypothesis evaluation. The course is targeted at students and researchers who want to learn how to evaluate theory-based hypotheses.
During the first day of the course, null-hypothesis significance testing (Kuiper and Hoijtink, 2010), questionable research practices, publication bias, and the replication crisis will be discussed. This day will also be used to give a hand-on introduction to statistical analyses using R.
During the second day, hypothesis evaluation using the Bayes factor will be discussed (Hoijtink, Mulder, Van Lissa, and Gu, unpublished). There will among others be attention for Bayesian error probabilities, Bayesian updating, the use of Bayesian hypothesis evaluation for the evaluation of replication studies, and combining evidence from multiple studies addressing the same research question (Kuiper, Buskens, Raub, and Hoijtink, 2013).
The third day of the course will start with an introduction of information criteria (i.e., AIC and its generalization called the GORIC) and a brief repetition of theory-based hypotheses. Subsequently, it will be elaborated how the GORIC can be used to evaluate null, alternative, and theory-based hypotheses. There will be attention for GORIC weights, which have a nice interpretation. This day will, like Day 2, be concluded with examples of replication research and combining evidence from multiple studies in which informative hypotheses play an important role.
During the morning of the fourth day, a debate between three groups (classicists, Bayesians, and information theorists) will be organised in which the (dis)advantages of the approaches presented in this course will be discussed. In the afternoon, there is a lab meeting in which you have the opportunity to analyse your own data (or data provided by the lecturers) with the approaches taught in this course.
Among our Methodology and Statistics postgraduate courses, there are two other courses that address Bayesian statistics:
- Applied Bayesian Statistics, to learn more about Bayes’ theorem, Gibbs sampling and the Metropolis-Hastings algorithm, Bayes factors, the evaluation of informative hypotheses, and Bayesian methods for linear regression, moderation, and mediation with observed variables. Unlike our course, this course pays attention to the statistical theory using formulas.
- A gentle introduction to Bayesian Statistics, for material on Bayesian Structural Equation Modeling. Unlike our course, this course does not cover hypothesis testing nor Bayes factors, and it focusses on SEM.
For an overview of all our summer school courses offered by the Department of Methodology and Statistics please click here.
Prof. Dr. Herbert Hoijtink and Dr. Rebecca Kuiper
Housing through Utrecht Summer School
Tuition fee for PhD candidates from the Faculty of Social and Behavioural Sciences from Utrecht University will be funded by the Graduate School of Social and Behavioural Sciences.
Irma Reyersen - firstname.lastname@example.org