BIAS: Research Programme

Social science data are notoriously full of missing values, non-responses, selection biases and other idiosyncrasies. Simple analyses are usually very misleading; instead a comprehensive set of inter-dependent sub-models are needed to model the data complexities and core processes that social scientists want to understand. It is also invariably the case that a single dataset fails to provide all the necessary information, and many of the complex research questions require the combination of datasets from multiple sources.

Bayesian graphical and hierarchical models offer a natural tool for linking together many different sub-models and data sources.

The BIAS I research programme consists of three methodological components:

In BIAS II, our aims are to address new methodological challenges in the modelling of observational data, in particular surveys, longitudinal studies and small area data; to extend and apply our BIAS I methods to new social science problems; and to foster a number of new collaborations with social scientists, both nationally and internationally. The methodological components of the BIAS II research programme are

We are also collaborating with government and social scientists to apply our methods to in four substantive areas

Published papers and books

N.T. (Jassy) Molitor, N. Best, C. Jackson and S. Richardson. Using Bayesian graphical models to model biases in observational studies and to combine multiple datasources: Application to low birth-weight and water disinfection by-products. PDF. A revised version of this article has been accepted by Journal of the Royal Statistical Society Series A (2008).

C. Jackson, N. Best and S. Richardson. Bayesian graphical models for regression on multiple datasets with different variables Biostatistics (2008) doi: 10.1093/biostatistics/kxn041. Published online in advance of print. PDF (open access).

C. Jackson, S. Richardson and N. Best. Studying place effects on health by synthesising individual and area-level outcomes, Social Science and Medicine (2008) 67:1995-2006. Available online at ScienceDirect.

S. Geneletti, N. Best and S. Richardson. Adjusting for selection bias in retrospective case control studies PDF. To appear in Biostatistics.

C. Jackson, N. Best and S. Richardson. Hierarchical related regression for combining aggregate and individual data in studies of socio-economic disease risk factors. . Blackwell Synergy Society Series A: Statistics in Society 171(1):159-178. (January 2008)

S. Richardson, C. Jackson and N. Best. Bayesian hierarchical models in ecological studies. PDF. Abstract of invited paper, International Workshop for Statistical Modelling, Sydney, July 2005.

C. Jackson, N. Best and S. Richardson. Improving ecological inference using individual-level data (Statistics in Medicine, 2006, 25(12):2136-2159, June 30). Available from Wiley InterScience.

N. Best, C. Jackson and S. Richardson. Modelling complexity in health and social sciences: Bayesian graphical models as a tool for combining multiple sources of information. DOC. In: Proceedings of the 3rd ASC International Conference on Survey Research Methods, eds. Banks, R., Cornelius, R., Evans, S. and Manners, T

Working papers

D. Lunn, N. Best, D. Spiegelhalter, G. Graham and B. Neuenschwander. Combining MCMC with 'sequential' PKPD modelling PDF. Submitted.

V. Gómez Rubio, N. Best, S. Richardson and P. Clarke. Bayesian Statistics Small Area Estimation. PDF. Submitted.

V. Gómez Rubio, N. Best and S. Richardson. A comparison of different methods for small area estimation. PDF. In preparation.

Lawrence McCandless, Sylvia Richardson and Nicky G. Best. Propensity Score Adjustment for Unmeasured Confounding in Observational Studies. PDF. In preparation.

Presentations

V. Gómez-Rubio, N. Best, S. Richardson and P. Clarke. Bayesian statistics for Small Area Estimation. PDF. Presented at 'GSS Methodology Conference 2008'. London, June 2008.

Sylvia Richardson, Lawrence McCandless, Jassy Molitor and Nicky Best. Bayesian Approaches to Adjustment for Unmeasured Confounders. PDF.

Nuoo-Ting (Jassy) Molitor, Chris Jackson, Nicky Best and Sylvia Richardson. Bayesian graphical models for combining mismatched administrative and survey data: application to low birth weight and water disinfection by-products. PPT. Presented at 'Recent Advances in Multilevel Modelling Methodology and Applications', Joint Meeting of the RSS Social Statistics and General Applications Sections 5 December 2007.

S. Geneletti and L. McCandless. Bayesian methods for combining multiple Individual and Aggregate data Sources in observational studies. PDF.

S. Geneletti, S. Richardson and N. Best. Adjusting for selection bias in case control studies. PDF.

V. Gómez-Rubio, N. Best, S. Richardson and P. Clarke. Bayesian Statistics, Small Area Estimation and why no one is poor in Sweden. PDF. Presented at the Royal Statistical Society Conference. York, U.K. 16-19 July 2007.

J. Molitor, S. Richardson and N. Best. Low birthweight and water disinfection byproducts: a multiple-bias modelling approach. Powerpoint. Presented at the Taipei International Statistical Symposium and ICSA International Conference. Academia Sinica, Taipei. 25-27 June, 2007.

N. Best, C. Jackson. Bayesian graphical models for inference from combinations of data. PowerpointPresented at the ESRC NCeSS workshop on combining and enhancing data, Manchester, Jan 2007

C. Jackson. Combining administrative and survey data in a study of low birth weight and air pollution. Powerpoint. Presented at the annual meeting of the ESRC National Centre for Research Methods, Manchester, Jan 2007.

V. Gómez Rubio. Bayesian methods for small area estimation using spatio-temporal models. PDF. Presented at the International Workshop on Spatio-Temporal Modelling (METMA3). Pamplona, Spain, 27-29 September 2006.

C. Jackson. Hierarchical models for combining multiple data sources measured at individual and small area levels. Powerpoint. Presented at the ESRC Research Methods Festival, Oxford, 17-20 July 2006.

N. Best. Introduction to Bayesian inference and computation for social science data analysis. Powerpoint. Presented at the ESRC Research Methods Festival, Oxford, 17-20 July 2006.

V. Gómez Rubio. A comparison of different methods for small area estimation. PDF. Presented at the 3rd Joint Meeting GUDO 3 of the Spanish Region of the IBS and the Spanish Society of Epidemiology, Valencia, Spain, 22 - 23 June 2006. Also PDF, presented at the Spatial Epidemiology Conference, London, May 2006.

C. Jackson. Hierarchical modelling of traffic-related benzene exposure and childhood leukaemia. Powerpoint. Presented at the 8th Valencia Meeting on Bayesian Statistics in Benidorm, Spain, 1-7 June 2006. Also Powerpoint, presented at the Spatial Epidemiology Conference, London, May 2006.

S. Geneletti. DAGs for selection bias in case-control studies. PDF. Presented at the 8th Valencia Meeting on Bayesian Statistics in Benidorm, Spain, 1-7 June 2006.

N. Best. Graphical models for combining multiple sources of information in observational studies. Powerpoint. Internal seminar, Imperial College, November 2005.

C. Jackson. Graphical models for combining multiple sources of information in observational studies. ii) Case study of socioeconomic risk factors for cardiovascular disease . Powerpoint. Internal seminar, Imperial College, November 2005.

N. Best. Graphical models for combining multiple data sources. Powerpoint. Given at the launch event of the National Centre for Research Methods, Oxford, June 2005.

Press articles

Putting ill-health in its place. ESRC The Edge, November 2005.


Presentations by collaborators

Alexina Mason (with thanks to Nicky Best, Ian Plewis and Sylvia Richardson). Methodological developments for combining data. PDF. Presented at the 'ESRC Research Methods Festival'. Oxford, July 2008.

Jon Forster (University of Southampton, UK). Bayesian Methods for Multivariate Categorical Data. PDF. Presented at the ESRC Research Methods Festival, Oxford, 17-20 July 2006.

Juanjo Abellan (Imperial College, London). Bayesian methods for small area estimation and spatial analysis. PDF. Presented at the ESRC Research Methods Festival, Oxford, 17-20 July 2006.