Scope: In this module we study how to analyse data that are not normally distributed. We look at fractions (logistic regression), counts (Poisson regression, log-linear models), ordinal data (threshold models), and overdispersion. We discuss (quasi-) maximum likelihood estimation and the deviance. In computer practicals participants can familiarize themselves with models and methods. The main statistical software used in this course is R.
On day 3 we look into GLMM’s (Generalized Linear Mixed Models). By introduction of random effects into the ordinary GLM, the assumption of independent observations is relaxed. Some knowledge of Mixed Models is required, e.g. as presented in the PhD course Mixed Models. We will mainly use R package glmmTMB to fit the GLMM’s.
- Target group: The course is aimed at PhD candidates and other academics
- Prior knowledge: Knowledge of Basic Statistics and Linear Models and some experience with the software package R are assumed
- Course duration: 3 days
- Contact: PE&RC Office: office.pe@wur.nl
- Registration of interest: You can register your interest HERE (note: this is not an official registration).
| Day 1 - morning | Introduction GLM’s, logistic regression |
| Day 1 - afternoon | Testing, lack of fit and overdispersion in logistic regression |
| Day 2 - morning | Poisson regression |
| Day 2 - afternoon | Miscellanea GLM (betabinomial, negative binomial, multinomial, gamma, threshold model) |
| Day 3 - morning | Miscellanea continued (Generalized Linear Mixed Models with glmmTMB, DHARMa for model checking) |
| Day 3 - afternoon | If time allows: own cases |