ACD: Analysis of Categorical Data

Lecturer

Associate Professor M. Silvapulle, La Trobe, Semester 1.

Syllabus

Inference for two-way contingency tables; methods for binary response variables including an overview of generalized linear models, logit models, models for categorical data and logistic regression models; loglinear models for two-, three- and multi-dimensional classifications; analysis of deviance.

This is essentially a course on theoretical statistics. Mathematical statistics up to third-year level will be assumed.

The text book for this subject is

Agresti, Alan (1990) Categorical data analysis, 1st edition, John Wiley: New York.

Every student enrolled in this subject is expected to have access to this book throughout the semester.

Prerequisites

1. Probability Theory


Basic properties of the following distributions; Binomial, Multinomial, Poisson, Normal, chi-square, t, F. A suitable level is Chapters 3 to 4.4 in Hogg and Craig "Introduction to Mathematical Statistics", 4th edition, Collier MacMillan.

2. Inference


Basic concepts of estimation; confidence intervals and hypothesis testing; maximum likelihood estimators; sufficiency; Cramer-Rao lower bounds; Likelihood ratio tests. A suitable level is Chapters 6.1--6.5, 7, 10, 11.1 in Hogg and Craig.

3. Linear Models


The general theory of least squares and its application to ANOVA; ANCOVA and multiple regression models; use of residuals to assess the adequacy of a model. A suitable level is Chapters 5, 6, 10--13 of Wetherill ``Intermediate Statistical Methods'', Chapman and Hall.

4. GLIM

• Agresti, Alan (2002) Categorical data analysis, 2nd edition, John Wiley: New York.
• Christensen, R. (1997) Log-Linear Models and Logistic Regression, 2nd edition, Springer.
• Collett, D. (1991) Modelling Binary Data, Chapman and Hall.
• Lindsey, J.K. (1995) Modelling Frequency and Count Data, Oxford Science Publications.

Last updated: 19 February 2003.