Authorised by Academic Registrar, April 1996
Objectives For students to understand the theoretical framework of least squares estimation and likelihood principle based tests for the general linear statistical model; apply the above theory to particular cases of the general linear model; identify outliers and high influence points in data analysis based on the general linear model; analyse data using related models including components of variance models and the logistic regression model; compute rejection regions (both exact and large-sample) for non-parametric tests based on ranks and runs, and apply these tests to appropriate data sets; understand the theoretical framework of basic sampling methods for finite population sampling, including simple and stratified random sampling.
Synopsis This subject continues the study of statistical inference beyond subject GAS2631. In particular, the subject develops inferential techniques for the general linear model and some extensions. Non-parametric inference and inference for finite population models are also covered. Topics include the general linear model, the method of least squares, estimability, the Gauss-Markov Theorem; hypothesis-testing including the likelihood ratio test for the case of normal disturbances, analysis of variance for experimental design models, the analysis of covariance, introduction to components of variance models, and logistic regression models; non-parametric methods including theory and application of simple tests based on ranks and runs; the goodness-of-fit problem; Kolmogorov-Smirnov statistics; sample survey theory including theory of simple and stratified random sampling, brief consideration of other sampling methods.
Assessment Assignments: 40% + Examination: 60%