Module overview
This module aims to introduce students to a wide range of statistical models grouped by the unifying theory of Generalized Linear Models: Linear, Logistic, Multinomial, Cumulative Ordinal and Poisson regression, as well as Log-linear models are presented, with emphasis on the underpinning theory and practical examples. Students are also exposed to the basic foundations of estimation for GLMs.
Aims and Objectives
Learning Outcomes
Learning Outcomes
Having successfully completed this module you will be able to:
- Summarise data with an appropriate statistical model.
- Use models to describe the relationship between a response and a set of explanatory variables.
- Use a range of popular statistical models for continuous and categorical data.
- Use the statistical software package R to fit statistical models.
- Interpret the results of the modelling.
- Understand the foundation theory of Generalised Linear Models.
Syllabus
The module is divided in 4 sections as explained below:
Section 1. Introduction:
Review of statistical modelling, Linear Regression, Deviance, model checking and regression diagnostics.
Section 2. Foundations of GLMs:
Foundations of Generalised Linear Models, the exponential family of distributions and its properties, Maximum Likelihood estimation, Score functions and Information, the Newton-Raphson and Fisher scoring algorithms.
Section 3. Categorical data and Logistic regression (Binary/Multinomial/Ordinal):
One-way contingency tables, two-way contingency tables, measures of association, odds ratios and properties of odds ratios. Binary logistic regression, probit regression, multinomial logistic regression, ordinal logistic regression, Maximum Likelihood Estimation, latent variable approach, deviance, residual analysis and model selection.
Section 4. Poisson regression and log-linear models:
Models for count data / Poisson regression, Log-linear models for rates, offset terms. Over dispersion and Negative-Binomial regression. Log-linear models for multi-way contingency tables and Simpson’s paradox. Residual analysis, Model selection, Deviance and Likelihood Ratio tests.
Learning and Teaching
Teaching and learning methods
Teaching will be delivered by a mixture of synchronous and asynchronous online methods, which may include lectures, quizzes, discussion boards, workshop activities, exercises, and videos. A range of resources will also be provided for further self-directed study. Face-to-face teaching opportunities will be explored depending on circumstances and feasbility.
Type | Hours |
---|---|
Teaching | 44 |
Independent Study | 106 |
Total study time | 150 |
Resources & Reading list
General Resources
Software requirements. You will require access to R, which is available on the University’s workstations and can be downloaded to your own computer for use with your studies.
Textbooks
Fox, J Sandford, W (2019). An R Companion to Applied Regression. Sage Publications.
Agresti, A (2013). Categorical Data Analysis. Wiley.
Faraway, J,J (2016). Extending the linear model with R: generalized linear, mixed effects and nonparametric regression models. CRC Press.
Fox, J (1997). Applied Regression Analysis, Linear Models, and Related Methods. Sage Publications.
Faraway, J,J (2015). Linear Models with R. CRC Press.
Dobson. A.J (2008). An Introduction to Generalized Linear Modules. Chapman and Hall.
Agresti. A (2007). An Introduction to Categorical Data Analysis. Wiley.
Assessment
Assessment strategy
There will be opportunities to evaluate your progress through formative assessment, with summative assessment based on two online assignments.
Summative
This is how we’ll formally assess what you have learned in this module.
Method | Percentage contribution |
---|---|
Coursework | 50% |
Timed Assignment | 50% |
Referral
This is how we’ll assess you if you don’t meet the criteria to pass this module.
Method | Percentage contribution |
---|---|
Assignment | 50% |
Coursework | 50% |
Repeat Information
Repeat type: Internal & External