led by Professor Alan Marshall, University of Edinburgh
This workshop will teach you how natural experiments can be used in the social sciences as a tool to provide stronger evidence for causal connections between variables than in observational studies.
The workshop covers:
- Definition of a Natural Experiment in the social sciences and relation to Observational Studies and Randomised Controlled Trials
- The strengths and weaknesses of Natural Experiments
- Examples of Natural Experiments in the Social Sciences
- Models that can be used to analyse Natural Experiments
- Experience of fitting a difference in differences model
- Experience of fitting an interrupted time series model
By the end of the workshop participants will:
- Be able to identify situations where a Natural Experiment might be used to provide stronger evidence for a causal link between variables compared to other study designs such as observational studies
- Understand the key terminology of natural experiments such as an ‘exposure’, ‘counter-factual’ and ‘as if random’
- Be able to fit and interpret a difference-in-differences model
- Be able to fit and interpret an interrupted time series model
- Have an understanding/awareness of alternative models that can be used to evaluate natural experiments
- Identify limitations of a natural experiments and strategies for how such limitations might be minimised
Requirements
Participants should be competent in the use of statistical software such as Stata, and in running analyses using code. Participants should be confident in the use of inferential statistics including regression. You will need to bring your laptop to the workshop to participate in the computer lab elements. You should ensure Stata is installed on your laptop before attending the workshop.