General Linear Models
General linear models (GLMs) form a unified underlying theory that covers simple and multiple linear regression techniques and general analysis of variance procedures for balanced and unbalanced data. An essential feature is the use of a normally-distributed residual or error term.
This course briefly presents the theory of general linear models and discuss their application and interpretation in problems of biological and medical sciences and in pharmaceutical work. Many examples are used to illustrate a wide range of GLMs. Practical sessions based on SAS help participants understand the ideas involved.
Statisticians who have experience with multiple regression and analysis of variance, and have had some previous exposure to the analysis of variance for unbalanced data. Familiarity with SAS software is assumed.
Participants will gain confidence in correctly using SAS for analysing different types of problems and in output produced by GLM fitting. In particular, identifying the correct type of sums of squares to use will be of benefit in analysing real-life problems.
- Review of the theory underlying the general linear model
- Analysis of variance and regression models
- Models with both quantitative and qualitative factors
- Non-orthogonal data structures
- Different types of sums of squares
- SAS facilities for estimating contrasts, and conducting tests of hypotheses
- Model-checking procedures.
Last updated: 13 August, 2014