Power analysis and sample size calculation

Reflections on the power and sample size of your study are essential to allow adequate inferential conclusions about your research hypotheses, both when designing your study and when interpreting the results. This course provides a gentle, non-technical introduction to the concept of statistical power and how this is influenced by sample size – among other factors, as well as a practical introduction to running your own power calculations in R and G*Power for common statistical analyses such as t-tests, (repeated measures) ANOVA and linear regression. In addition, we will briefly touch upon sample size calculation for statistical analyses where no formal hypothesis test is conducted, e.g. confidence intervals, factor analysis and cluster analysis. We will also discuss the difference between significance and relevance, how power and effect size come into play during statistical tests and how p-values might be misleading as a standalone criterion for your hypotheses.

The aforementioned topics will be taught during the first two days of the course. The final day will introduce Monte Carlo simulations (in R) as a very flexible tool to conduct power analyses and sample size calculations for more complicated designs and analyses. We will discuss the general idea behind randomly drawing from probability distributions, followed by its application to power analysis/sample size calculation in more complicated settings such as logistic regression models and longitudinal designs. You will learn how to ‘mimic’ your population of interest, sample from this population by means of Monte Carlo simulations and customize these simulation studies to your research questions and analyses

It is possible to take this course as a basic module (first 2 days), advanced module (final day) or both, with customized certificates and (optional) exams. For the advanced topics during the final day, some elementary knowledge of R, logistic regression and longitudinal data analysis is recommended – as well as a basic understanding of power analysis as seen in the basic module or attained by other means