Generalized Linear Models
Date: 16th - 17th April 2012
Duration: 1 day (16 Apr: 1.30 – 5.15pm // 17 Apr: 11.30am – 5.30pm; Lunch 1 - 2pm not provided)
Delivered by: Professor Brian Francis and Dr Andrew Titman
From 13:30pm till 17:30pm (each day)
Registration deadline has passed.
Please contact psc@lancaster.ac.uk for more information about this course.
- External from industry/commerce - £255
- External from academic institution/public sector/charity - £220
- External postgraduate student - £150
- Lancaster University staff - £60
- Lancaster University postgraduate student - £30
- Members of Mathematics and Statistics at Lancaster University - £ 0
The aim of this course is to consider generalized linear models as a broad class of statistical models applying the general principles of likelihood inference to a variety of commonly encountered data analysis problems in the social sciences. The course will also introduce the software package R as tool for such statistical analysis.
Concepts introduced in the Statistical Inference course will be reviewed and extended to this broad class of models. The use of factors, covariates and their interactions to build a flexible class of relationships will be considered. Applications to the modelling of categorical data will be described.
- The basics: Linear models, General linear models and Generalized linear models;
- Different GLMs: Simple linear regression, multiple linear regression, regression with binary data, regression with count data;
- Use of continuous and categorical (factor) covariates;
- Using interaction terms;
- Model building and testing (F-test, ANOVA, Likelihood ratio test, AIC);
- Applications of GLMs;
- What to report;
- Using R for GLMs.