Introduction to Applied Multilevel Analysis - HEA00039M
- Department: Health Sciences
- Credit value: 10 credits
- Credit level: M
- Academic year of delivery: 2022-23
Related modules
Module will run
| Occurrence | Teaching period |
|---|---|
| A | Autumn Term 2022-23 |
Module aims
To equip students with the necessary skills and knowledge to allow analysis of multilevel data. By means of lectures and hands-on analysis of data from real studies using the statistical software package STATA. The student is guided through a range of statistical techniques that can be used based on the nature of the outcome when the data follow a multilevel structure.
Module learning outcomes
At the end of the module, the student should be able to:
- Identify when to use a multilevel regression approach versus ordinary regression approach.
- Identify different types of multilevel structures.
- Distinguish between levels and variables and between fixed and random effects.
- Distinguish between different outcomes and correspondingly carry out the appropriate multilevel regression.
- Demonstrate, by example, how to formulate a multilevel model and how to interpret the results obtained from fitting the model.
- Use multilevel modelling for the analysis of cluster randomised trials.
- Use a statistical package to carry out analyses.
Academic and graduate skills
Students will be able to carry out multilevel data analysis on a number of outcomes using a statistical package and critically read research papers that use multilevel analysis.
Module content
|
Prior to commencing the module, students are expected to demonstrate knowledge of ordinary linear, logistic, and survival regression methods and ability to carry these regressions.
Subject content Revision of Generalised Linear Models Topics to include multiple linear regression, logistic regression, Poisson regression, and Survival Analysis
What is Multilevel modelling? Introduce a range of multilevel structures, e.g nesting and cross classification, with examples from real studies, introduce how to represent multilevel structures using subscripts, distinguish between levels and variables, and fixed and random effects.
Multilevel modelling for a continuous outcome Introduce the random intercept and the random slope models pointing out why standard linear regression does not work in the case of multilevel structures together with the assumptions underlying these models and sample size issues.
Multilevel modelling for a binary outcome and count data Extend the random intercept and the random slope models to the case of a binary response and count data focusing on how to interpret the models and introduce the latent variable approach in the case of binary outcome.
Multilevel modelling for time-to-event data Introduce the discrete time approach to analyse time-to-event data and extend the model to the multilevel case.
Multilevel modelling in cluster randomised trials A brief introduction to cluster randomised trials and how to analyse them using multilevel modelling also hinting to other methods used to analyse cluster randomised trials. |
Indicative assessment
| Task | % of module mark |
|---|---|
| Online Exam - 24 hrs (Centrally scheduled) | 100 |
Special assessment rules
None
Indicative reassessment
| Task | % of module mark |
|---|---|
| Online Exam - 24 hrs (Centrally scheduled) | 100 |
Module feedback
Students are provided with collective exam feedback relating to their cohort, within the timescale specified in the programme handbook.
Indicative reading
- Gelman, A. and Hill, J. (2007). Data Analysis Using Regression and Multilevel/Hierarchical Models: Analytical Methods for Social Research. New York, Cambridge University Press.
- Goldstein, H. (2010). Multilevel Statistical Models. 4th edn. Singapore: Wiley Series in Probability and Statistics.
- Klar, N. and Donner, A. (2000). Design and Analysis of Cluster Randomisation Trials in Health Research. Wiley-Blackwell.
- Leyland, A.H. and Goldstein, H. (2001). Multilevel modelling of health statistics. Chichester: Wiley Series in Probability and Statistics.
- Rabe-Hesketh, S. and Skrondal, A. (2008). Multilevel and Longitudinal Modeling Using Stata. 2nd edn. USA: Stata Press.