Institut für Evolution und Ökologie

GLMM workshop 2014

Winter school

Linear mixed effect models and GLMM in R (2ECTS)

24 - 28 February 2014, Mon-Fri 08:30-17:00 (incl. breaks)

Tübingen University, Inst. of Evolution and Ecology, Auf der Morgenstelle 28, E4A20, 72076 Tübingen

Teachers: Dr. Fränzi Korner-Nievergelt, Dr. Stefanie von Felten, Oikostat GmbH Ettiswil, CH.

Organisers: Dr. Nils Anthes, Dr. Katja Heubel, Tübingen

Course form: lectures, worked examples, exercises, own data workshop.

This winter school offers training in cutting-edge statistics using the freely available R environment (www.r-project.org). It focuses on the analysis of complex behavioural and ecological datasets that do not conform to the assumptions of standard linear models. Introductions to statistical theory are always coupled with empirical case studies and individual training. A detailed course program is attached below.

Target audience

The target audience includes postgraduate students as well as senior scientists with a direct incentive to apply the taught statistical approaches to their current research project. Moreover, we invite applications by course advisors who intend to integrate advanced statistics into their own teaching program.

Course aims

Participants of the course will obtain a thorough introduction to linear models (LM), mixed effects models (LME), generalized linear models (GLM), and generalized linear mixed models (GLMM), their implementation in R and their interpretation using classic hypothesis testing as well as Bayesian inference. Worked examples will include:

  • graphical data exploration (various plotting functions)
  • fit of the model to data (R-Functions lm, lmer, glm, glmer)
  • assessment of model fit and model assumptions (diagnostic plots of residuals)
  • visualization and interpretation (summary, anova, predict, sim)

In addition, course participants may apply linear models to their own data during the course.

Course fees and requirements

Thanks to generous financial support through the Institutional Strategy of the University of Tübingen (Deutsche Forschungsgemeinschaft, ZUK 63), this course can be offered without charging individual course fees.

Participants require a profound education in general statistics (data exploration, multiple regression models, generalized linear modelling) coupled with fluency in the basic usage of R statistical software. Make sure you are familiar with the following terms: mean, standard deviation, standard error, and t-test (e.g. chapter 5 in Dalgaard 2008, Introductory Statistics with R, Springer). Basic knowledge in R programming is required (e.g. chapters 1-5 in Crawley 2007, The R-book, Wiley; chapters 1, 2 and 4 in Dalgaard 2008). All participants need to bring their own laptop with the latest version of R installed.

Applications ...

... must be sent by E-mail as a single pdf-file to nils.anthes@uni-tuebingen.de before 31 Oct 2013. In your application, briefly specify how the attendance of this winter school contributes to your current research or teaching program. Applicants from outside Tübingen are invited to specify required support for travel or accommodation.

For confirmed participants, attendance of the course is binding. In case of withdrawal from attending the course later than January 15, 2014, we charge a cancellation fee of EUR 150,-.

Preliminary course schedule

Monday: Data exploration and refreshing linear models

  • Graphical data exploration, ordination, principal component analysis.
  • Refreshing linear models (LM), i.e. multiple regression, ANOVA, ANCOVA.
  • Basics of statistical inference and Bayesian posterior probabilities.

Tuesday: Linear mixed models LME and generalized linear models GLM

  • Assessment of model fit and model selection procedures.
  • Linear mixed models (LME) and how to integrate random effects
  • Generalized linear models (GLM) for count (poisson and negative binomial) and binary data.

Wednesday: Generalized linear mixed models GLMM

  • GLMM with Poisson, negative binomial, or binomial distributions.
  • Overdispersion and its treatment.

Thursday: Advanced topics (depending on time and wishes of the participants)

  • Multinomial models and introduction to WinBUGS.
  • Smoothing using general additive (mixed) models (GAMM).
  • Treating zero-inflated datasets.

Friday: Roundup and own data workshop

  • Guided work on own data and specific additional topics upon request.