Chapter 8: Generalized Linear Models: Diagnostics
This chapter introduces some of the necessary tools for detecting violations of the assumptions in a glm, and then discusses possible solutions. The assumptions of the glm are first reviewed (Sect. 8.2), then the three basic types of residuals (Pearson, deviance and quantile) are defined (Sect. 8.3). The leverages are then given in the glm context (Sect. 8.4) leading to the development of standardized residuals (Sect. 8.5). The various diagnostic tools for checking the model assumptions are introduced (Sect. 8.7) followed by techniques for identifying unusual and influential observations (Sect. 8.8). Comments about using each type of residual and the nomenclature of residuals are given in Sect. 8.6. We then discuss techniques to remedy or ameliorate any weaknesses in the models (Sect. 8.9), including the introduction of quasi-likelihood (Sect. 8.10). Finally, collinearity is discussed (Sect. 8.11).
- Data Desk: Data and story library (dasl) (2017). URL http://dasl.datadesk.com
- Dunn, P.K., Smyth, G.K.: Randomized quantile residuals. Journal of Computational and Graphical Statistics 5(3), 236–244 (1996)Google Scholar
- McCulloch, C.E.: Generalized linear mixed models. Institute of Mathematical Statistics (2003)Google Scholar
- Singer, J.D., Willett, J.B.: Improving the teaching of applied statistics: Putting the data back into data analysis. The American Statistician 44(3), 223–230 (1990)Google Scholar
- Smyth, G.K.: Australasian data and story library (Ozdasl) (2011). URL http://www.statsci.org/data