Bayesian Hierarchical Modelling using WinBUGS/OpenBUGS
Dates: 23rd - 24th February 2012
Duration: 2 days (9.30am to 5.00pm)
Level: 2/3
Course Leader: Nicky Best
Course Fee: £440 (standard fee); £120 (reduced fee for ESRC funded researchers and staff at UK academic institutions and registered charity organisations; £60 (reduced fee for UK registered postgraduate students)
CCSR offer 5 free places to research staff and students within the Faculty of Humanities at the University of Manchester.
Course requirements: Attendance on introductory WinBUGS/OpenBUGS course (http://www.ccsr.ac.uk/courses/bayesian/index.html) and/ or familiarity with the basic principles of Bayesian inference (prior, likelihood, posterior; simple conjugate analyses) and the use of Markov chain Monte Carlo methods and/or experience of using WinBUGS/OpenBUGS software.
Course summary: Many quantitative researchers in the social and health sciences will be required to analyse data with a hierarchical or multilevel structure, or with missing or mis-measured values, at some point in their careers. Bayesian methods offer a natural approach to handling these types of problems, through their ability to specify distributions both for model parameters and for missing or imprecisely measured data. This two-day course provides researchers who already have some basic understanding of Bayesian methods with a more in-depth treatment of applied Bayesian methods for modelling data with complex structure. There is a large practical component to this course with time for hands-on data analysis using examples drawn mainly from the social and health sciences. This course is organised by the BIAS node (www.bias-project.org.uk) of the ESRC National Centre for Research Methods (www.ncrm.ac.uk).
Course content: Topics covered will include
- Introduction to Bayesian hierarchical models
- More complex Bayesian hierarchical models (e.g. hierarchical regression models, hierarchical models for longitudinal data, hierarchical modelling of variances, hierarchical mixture models)
- Bayesian measurement error models
- Bayesian modelling of missing data
- Model criticism and comparison for Bayesian hierarchical models
- Practical exercises using WinBUGS/OpenBUGS software
Target audience: Statisticians, data analysts and quantitative researchers who are interested in learning more about Bayesian methods for modelling data with complex structure (including hierarchical structure, missing or mis-measured values), and how to implement these methods using the WinBUGS/OpenBUGS software. The course would also be of interest to researchers with experience of multi-level modelling using likelihood-based methods who wish to find out more about fitting and interpreting Bayesian versions of multi-level models. Participants with no previous experience of Bayesian inference or WinBUGS/OpenBUGS are strongly advised to attend the introductory course first (http://www.ccsr.ac.uk/courses/bayesian/index.html).