https://svendemaeyer.netlify.app/ Reproducible Stats in Education Sciences Distill Mon, 31 May 2021 00:00:00 +0000 Sven De Maeyer https://svendemaeyer.netlify.app/posts/2021-04-Generalizability Generalizability theory is a very powerful tool for analyzing the strengths and weaknesses of measurements. It uses estimates of different sources of variance in the measurement situation. However, these estimates in themselves are of course not absolute. The uncertainty behind these estimates is usually not included in the research. In this post, I explore the idea of approaching generalizability theory from the Bayesian angle. I am curious to hear your ideas on this! R Bayesian inference Statistics brms Generalizability theory https://svendemaeyer.netlify.app/posts/2021-04-Generalizability Mon, 31 May 2021 00:00:00 +0000 Sven De Maeyer https://svendemaeyer.netlify.app/posts/2021-03-24_Time-to-Switch A controlled randomized experiment is often seen as a gold standard in experimental research. But in many situations ethical considerations make it impossible to conduct such a 'real' experiment. For instance in education it can be considered unethical to deny a group of students an educational intervention of which we assume a high impact on their achievement. The 'switching-replication' design can come to the rescue! In this post we will shortly describe this experimental design and demonstrate how we can model data coming from such a design making use of one of my favourite packages in R: `brms`. So we will approach it in a Bayesian way! R Bayesian inference Statistics brms switching replications https://svendemaeyer.netlify.app/posts/2021-03-24_Time-to-Switch Wed, 24 Mar 2021 00:00:00 +0000 Sven De Maeyer https://svendemaeyer.netlify.app/posts/2021-01-29-modelling-multilevel-models-using-brms Bayesian statistics is on the rise. There is no doubt about that. But the learning curve is quite steep for many researchers. In this post, I give that extra push as you climb that mountain on your route to learning Bayesian statistics. A short demo on how to handle the super package brms to estimate some standard multilevel models can hopefully be inspiring and reassuring. While you're on the road you may also learn some neat plotting skills! Bayesian inference Multilevel modeling brms Stan R TIMSS tidybayes bayesplot https://svendemaeyer.netlify.app/posts/2021-01-29-modelling-multilevel-models-using-brms Mon, 22 Feb 2021 00:00:00 +0000 Sven De Maeyer https://svendemaeyer.netlify.app/posts/2021-01-18-bayesian-analysis-of-comparative-judgement-data Comparative Judgement (CJ) is slowly gaining momentum in the context of educational assessment. Data coming from CJ is typically analysed with the BTL-model, making use of a frequentist approach. Increasing computational power of computers has made Bayesian inferences more accessible, resulting in the rise of Bayesian model estimation. In this post we will demonstrate how the package `pcFactorStan` can be applied to analyse data from CJ within a Bayesian framework, making use of data from CJ judgements on argumentative writing coming from the D-PAC project. Bayesian inference Comparative Judgement Comproved Stan R https://svendemaeyer.netlify.app/posts/2021-01-18-bayesian-analysis-of-comparative-judgement-data Tue, 26 Jan 2021 00:00:00 +0000 Sven De Maeyer https://svendemaeyer.netlify.app/posts/2021-01-20-4-bayesian-concepts In this post I will introduce 4 crucial concepts that will cross your path if you start reading research reporting Bayesian analyses. R Bayesian inference Statistics https://svendemaeyer.netlify.app/posts/2021-01-20-4-bayesian-concepts Sun, 07 Jul 2019 00:00:00 +0000