| Instrumental variable estimation for binary outcomes |
| PAUL CLARKE (CMPO, Uni. Of Bristol) |
| Instrumental variables are used in biostatistics, economics and epidemiology, and the (strong) assumptions for causal inference are well understood in the linear case. We look at the situation for binary outcomes and review the available estimators, highlighting the additional assumptions required for causal inference and making recommendations for practitioners |
| Documents |
| Clarke presentation | PC RSSpresentation upload.pdf |
| |
| Causal criteria for aggregate change in demography: issues and illustrations |
| MAIRE NI BHROLCHAIN (Uni. of Southampton) & TIM DYSON (LSE) |
| A set of causal criteria is proposed for aggregate demographic change, intended to support the process of causal inference, analogous to the Bradford Hill criteria in epidemiology. Several demographic examples illustrate the issues |
| Documents |
| NI BHROLCHAIN & DYSON presentation | Ni Bhrolchain & Dyson causation May09.pdf |
| |
| What was your question again? The perils of using observational data to answer questions other than the one you wish to ask. |
| MIGUEL A. HERNAN (Harvard School of Public Health) |
| Statistical methods are often used to make causal inferences. In the absence of well-defined causal questions, the validity and implications of a particular statistical analysis are difficult to assess. This problem is more severe when estimating effects from complex longitudinal data with time-varying processes. |
| Documents |
| Hernan presentation | Hernan.pdf |
| |
| Path analysis for discrete variables: The role of education in social mobility |
| JOUNI KUHA (LSE) & JOHN GOLDTHORPE (Nuffield College, Oxford) |
| This talk examines the impact of education on Intergenerational social mobility in Britain.The question is examined using a new method of path analysis,which can be used to estimate direct and indirect effects even in systems where some of the variables are categorical. |
| Documents |
| Kuha presentation | Kuha.pdf |
| |
| Causal evaluation of lifestyle intervention trials: The Counterweight Project |
| SARA GENELETTI (Imperial Collge) & GARY FROST (Imperial College) |
| The Counterweight project (CWP) is a Lifestyle intervention trial (LIT) recently completed in the UK aimed at assessing the effectiveness of a weight-loss management programme for obese people. We use the Decision theoretic approach to causal inference (Dawid 2002,2007) to answer the question of interest to the policy maker "given that CWP has worked in the selected sample, would applying it to an average UK practice result in desired weight-loss and subsequent burden to the National Health Service? |
| Documents |
| Geneletti presentation | Geneletti.pdf |
| |
| Propensity score matching and causal inference |
| BARBARA SIANESI (Institute For Fiscal Studies) |
| The use of matching methods to draw inference on average treatment effects has been steadily growing. Focusing on the estimation of the causal effect of education on earnings, the talk will introduce matching methods and consider their relationship to the classical dummy variable regression model, fully-interacted linear model and control function model. From a methodological as well as applied point of view, this exercise will highlight the unique strengths and weaknesses of matching methods in drawing casual inference. |
| Documents |
| Sianesi presentation | Sianesi RSS May 09.pdf |
| |
| Meeting Contact: Antony Fielding (A.Fielding@bham.ac.uk) Jouni Kuha |
| Organising Group(s): Joint meeting of Social Statistics Section and GAS |
Unified Solutions.