Zero Inflated Models and Generalized Linear Mixed Models with R (2012)

Zuur, Saveliev, Ieno

Chapter 1 provides a basic introduction to Bayesian statistics and Markov Chain Monte Carlo (MCMC), as we will need this for most analyses.

In Chapter 2 we analyse nested zero inflated data of sibling negotiation of barn owl chicks. We explain the application of a Poisson GLMM for 1-way nested data and discuss the observation-level random intercept to allow for overdispersion. We show that the data are zero-inflated and introduce zero-inflated GLMM. 

Data of sandeel otolith presence in seal scat is analysed in Chapter 3. We present a flowchart of steps in selecting the appropriate technique: Poisson GLM, negative binomial GLM, Poisson or negative binomial GAM, or GLMs with zero-inflated distribution.

Chapter 4 is relevant for readers interested in the analysis of (zero inflated) 2-way nested data. The chapter takes us to marmot colonies: multiple colonies with multiple animals sampled repeatedly over time.

Chapters 5 – 7 address GLMs with spatial correlation. Chapter 5 presents an analysis of Common Murre density data and introduces hurdle models using GAM. Random effects are used to model spatial correlation. In Chapter 6 we analyse zero-inflated skate abundance recorded at approximately 250 sites along the coastal and continental shelf waters of Argentina. Chapter 7 also involves spatial correlation (parrotfish abundance) with data collected around islands, which increases the complexity of the analysis. GLMs with residual conditional auto-regressive correlation structures are used.

In Chapter 8 we apply zero-inflated models to click beetle data.

Chapter 9 is relevant for readers interested in GAM, zero inflation, and temporal auto-correlation. We analyse a time series of zero-inflated whale strandings.

 

Keywords

  • Introduction to Bayesian statistics, MCMC and WinBUGS (run from R) using oystercatcher data.
  • Zero-inflated models for 1-way nested owl data. GLMM and overdispersion.
  • A protocol for dealing with overdispersed sandeel otoliths in seal scats.
  • Zero-inflated models and GLMMs for 1-way and 2-way nested marmot data.
  • Hurdle models with random effects for Common Murre density data.
  • Zero-inflated models with spatial correlation for skate abundances.
  • Zero-inflated models with spatial correlation around an island for parrot fish data. GLMMs with spatial correlation.
  • Zero-inflated models for click beetles.
  • Zero-inflated smoothing models with temporal correlation for sperm whale strandings time series. 

 

Table of contents

Table of Contents

 

Reviewer comments

‘I'm seriously impressed with how the authors teach statistics. It is exactly the right approach, and the code included will help learning tremendously.’

Aaron MacNeil, Australian Institute of Marine Science, Australia

 

‘Zero Inflated Models and Generalized Linear Mixed Models is one of those rare volumes that clearly presents new state-of-the-art statistical methodology in a clear and understandable manner. The text is both theoretical and thoroughly applied, offering readers a solid understanding of zero inflated count models from both a Bayesian perspective and as an extension of generalized linear mixed models. This is the book to read on zero inflated count models. Absolutely superb reading.’

Joseph M. Hilbe. Arizona State University and Jet Propulsion Laboratory. Author of Negative Binomial Regression

 

Data sets used in the book

The current (July 2012) errata list of the book can be found here.

All data sets used in the book are provided as *.txt files. Right-mouse click on a data file and click on Save-As.

  • Chapter 1: Oystercatcher data.
  • Chapter 2: Owl data. R code: Owls.zip. This file is password protected. The password is the very last word on page 66.
  • Chapter 3: Awaiting publication
  • Chapter 4: Marmot data
  • Chapter 5: Common Murre data
  • Chapter 6: Skate data. Support files used in this chapter. File 1 and file 2. Use Right-Mouse click, Save-As, or else your browser may add *.txt to the two R files.
  • Chapter 7: Parrot fish data
  • Chapter 8: Beetle data
  • Chapter 9: Sperm whale data

 

Support chapters

To avoid duplication of material that we published in other books, we provide two pdf files:

These pdf files are available to readers of some of our books. Both chapters are password protected. The passwords can be found in the Preface of the book that you bought.