Introduction to Generalised Linear Models for Ecologists

Session 4 – 01:15:00 – Binary Logistic Regression I
Theoretical background of logistic regression; fitting and interpreting binary outcome models.

Session 5 – 01:15:00 – Binary Logistic Regression II
Predicted probabilities, odds ratios, nested model comparison using likelihood ratio tests, and visualising predictions.

Session 6 – 01:15:00 – Assessing Model Fit
Diagnostics for binary logistic models: residuals, influence, and assessing linearity in the logit.

Session 7- 01:15:00 – Binomial Logistic Regression
Modelling grouped binary/proportion data; understanding weights, logit vs probit vs cloglog links.

Session 8- 01:15:00 – Multinomial Logistic Regression
Polychotomous outcomes and the multinomial logit model, interpreting output and visualising effects.

Session 9- 01:15:00 – Ordinal Logistic Regression
Ordered categorical outcomes, cumulative logit models, assumptions and interpretation.

Session 10 – 01:15:00 – Poisson Regression
Theoretical and practical introduction to Poisson regression; rate-based modelling and interpreting log link coefficients.

Session 11 – 01:15:00 – Poisson Regression in Practice
Offset terms, exposure variables, and visualising count model predictions.

Session 12 – 01:15:00 – Case Study: Poisson Modelling in Ecology
Worked example applying Poisson regression to real-world count data, including plotting and interpretation.

Session 13- 01:15:00 – Overdispersion and Quasi-Likelihood
Identifying and diagnosing overdispersion in Poisson and binomial models; fitting quasi-Poisson and quasi-binomial models.

Session 14 – 01:15:00 – Overdispersion Models
Fitting overdispersion models, when and why to use each approach.

Session 15 – 01:15:00 – Interpreting and Comparing Overdispersion Models
Model selection with AIC, BIC, and cross-validation; visual comparisons and residual diagnostics.

Session 16 – 01:15:00 – Zero-Inflated Models I
Understanding zero-inflation and the structure of ZIP and ZINB models.

Session 17 – 01:15:00 – Zero-Inflated Models II
Interpreting zero-inflated models, model diagnostics, and visualising predictions.

Session 18 – 01:15:00 – Hurdle Models and Alternatives
Distinguishing between hurdle and zero-inflated models; when to use each; fitting in R.

Session 19 – 01:15:00 – Data Wrangling and Preparation for GLMs
Long vs wide formats, dealing with missing data, preparing factors and numeric covariates, centering and scaling data.

Session 20 – 01:15:00 – Model Extensions: Interactions and Nonlinear Terms
Fitting interaction terms, polynomial and spline expansions, interpretation and plotting.

Session 21 – 01:15:00 – Visualising GLMs
Plotting fitted GLM effects, and confidence intervals.

Session 22 – 01:15:00 – Introduction to Bayesian GLMs
Overview of Bayesian inference, the role of prior distributions, and how posterior inference differs from frequentist approaches. Emphasis on the probabilistic interpretation of coefficients and model uncertainty in the context of GLMs.

Session 23 – 01:15:00 – Bayesian GLMs in Practice
Hands-on fitting of Bayesian logistic and Poisson regression models using R (e.g., with brms or rstanarm). Comparison of Bayesian and frequentist estimates, intervals, and interpretations. Introduction to posterior predictive checks and visualising uncertainty in model predictions.

Session 24 – 01:15:00 – Model Evaluation and Interpretation in Bayesian GLMs
Deeper exploration of how to evaluate and interpret Bayesian GLMs. Includes assessing model fit (e.g., using posterior predictive distributions, WAIC), interpreting posterior intervals and probabilities, and presenting Bayesian model results clearly. How Bayesian approaches inform decision-making and communication of uncertainty in applied contexts (e.g., ecology, epidemiology).

Session 25 – 01:15:00 – GLMs for Presence-Absence and Occupancy Data
Applying logistic regression to presence-absence and detection/non-detection data; interpreting coefficients in ecological terms; modelling site-level covariates and detection bias.

Session 26 – 01:15:00 – Count Models in Ecology
Poisson and negative binomial models for abundance and count data; modelling species richness, trap counts, or observation events.

Session 27 – 01:15:00 – Ecological Case Study: Species Distribution and Abundance Modelling
A guided walkthrough of real ecological data: fitting GLMs for species occurrence or abundance, selecting predictors (e.g. habitat, climate), and communicating model results with ecological interpretation and plots.

Session 28 – 01:15:00 – Mixed Effects GLMs (Intro to GLMMs)
Motivation for using random effects in GLMs; fitting random intercept models with lme4::glmer(), interpreting output, and diagnosing convergence.

Session 29 – 01:15:00 – GLMMs Continued: Nested and Crossed Designs
Random slopes and more complex structures; fixed vs random effects; challenges with non-convergence and overfitting.

Session 30 – 01:15:00 – Model Comparison, Best Practices, Wrap-Up and Further Directions
AIC, BIC, and likelihood-based model selection in GLMs and GLMMs; model simplification vs overfitting; common errors and how to avoid them. Course review, recommendations for further learning (e.g. GAMs, zero-inflated Bayesian models, Bayesian GLMMs); curated list of packages, vignettes, and readings.