3-Day Multilevel Modelling Course: Modelling the impact of households and geographies in health research
Multilevel models and MCMC methods using the new STAT-JR package
A workshop given by:
Bill Browne, Centre for Multilevel Modelling, University of Bristol
Ian Plewis, Cathie Marsh Centre for Census and Survey Research, University of Manchester
Mark Tranmer, Cathie Marsh Centre for Census and Survey Research, University of Manchester
Wednesday 11 - Friday 13 July 2012
Basement Computer Lab, Humanities Bridgeford Street Building, University of Manchester
This workshop is jointly organised by ESDS Government and e-Stat, a quantitative node for the ESRC Digital Social Research programme which brings together academics working in statistics and computer science.
The focus of this three day workshop is on the use of new statistical software, STAT-JR, for fitting complex statistical models. We will focus on multilevel modelling where the multilevel structure is determined by the sample design, in particular on the way the methods can be applied to the Health Survey for England. This structure will then be used to examine the extent to which variables such as height, weight and BMI are clustered by household and area and whether any of this variability can be accounted for by background variables such as social class and educational qualifications.
The first day will provide a general introduction to the dataset and STAT-JR. On the second day we will give sessions on Bayesian statistics and MCMC methods that the package uses and then describe its application to continuous multilevel models. On the third day we will finish by describing its use on binary response models and give a brief demonstration of an alternative E-book interface to the software. The workshop will be a mixture of lectures and practical sessions using the software on the datasets provided.
What is STAT-JR?
• For the beginner, STAT-JR allows access to many different statistical software packages (such as MLwiN, winBUGS and R) through a simple point and click interface.
• For the advanced user, STAT-JR allows construction of their own functionality through the writing of or customising of the templates that the system is built upon.
• For the software developer, STAT-JR allows incorporation of their packages within the system and easy comparison of their software with other packages that exist.
• STAT-JR offers a web based test interface ('webtest') and an 'ebook' interface that can be used to create and view interactive documents with dynamic execution of the underlying content.
Prerequisites: a working knowledge of multiple regression and the basics of sample design. This course will be ideal for those with limited knowledge of multilevel and MCMC methods who want to explore these in more detail or those with some knowledge of multilevel models who want to learn STAT-JR
.
Cost: the workshop fee is £95. Bookings will not be confirmed until payment is received. Refunds will not be made unless we are notified of your cancellation at least ten working days before the course date. Participants are responsible for making their own travel and accommodation arrangements and expenses cannot be paid. The nearest accommodation to the venue is at Manchester Business School.
Programme
Practicals
| Day one |
| |
|
| 1.30 |
Coffee and registration |
| |
|
| 2.00 |
Session 1: Pierre Walthery, ESDS Government, University of Manchester
1. Accessing the HSE via ESDS
2. The HSE sample design and dataset
3. Descriptive analysis of the HSE |
| |
|
| 3.00 |
Refreshments |
| |
|
| 3.30 |
Session 2: William Browne, School of Veterinary Science, University of Bristol
1. Introducing the STAT-JR package and in particular its interoperability features
2. Descriptive analysis of the HSE using interoperability in STAT-JR to MLwiN, Stata and R. (practical)
3. Fitting single level models in STAT-JR (practical) |
| |
|
| 5.30 |
End of day one |
| |
|
| |
Optional activities on day one
|
| 6.30 - 7.30 |
Guided walk: Engels and Marx in Manchester - meet at the Cornerhouse, Oxford Road |
| 7.30 |
Evening meal - Felicini, Oxford Road |
| |
|
| Day two |
| |
|
| 9.30 |
Session 3: Mark Tranmer, CCSR/Social Statistics, University of Manchester
Introducing multilevel modelling |
| |
|
| 10.15 |
Refreshments |
| |
|
| 10.30 |
Session 4: Mark Tranmer, CCSR/Social Statistics, University of Manchester
Normal response multilevel analysis of the HSE using STAT-JR part 1 (practical) |
| |
|
| 11.30 |
Session 5: William Browne, School of Veterinary Science, University of Bristol
Introduction to MCMC methods and Bayesian Statistics |
| |
|
| 1.00 |
Lunch |
| |
|
| 2.00 |
Session 6: Mark Tranmer, CCSR/Social Statistics, University of Manchester
Normal response multilevel analysis of the HSE using STAT-JR part 2
(Exercises based on normal response multilevel analysis of the HSE using MCMC) |
| |
|
| 3.30 |
Refreshments |
| |
|
| 3.45 |
Session 7: Ian Plewis , CCSR/Social Statistics, University of Manchester
Multilevel models for other response types |
| |
|
| 5.00 |
End of day two |
| |
|
| Day three |
| |
|
| 10.00 |
Session 8: Ian Plewis , CCSR/Social Statistics, University of Manchester
Exercises with binary responses |
| |
|
| 11.30 |
Refreshments |
| |
|
| 11.45 |
Session 9: Bill Browne, School of Veterinary Sciences, University of Bristol
Demonstration of the E-book interface to STAT-JR |
| |
|
| 12.30 |
End |
