Centre for Biostatistics Short Courses

Causality in Health Sciences

Date: 9-12 June 2015

Time: 9:30am-5:30pm

Venue: The University of Manchester, Committee Room D, Whitworth Building, Oxford Road, Manchester M13 9PL

Instructors: Professor Carlo Berzuini and Dr Hui Guo

Fee: (Course fee, lunches and refreshments inclusive)

Students: £150

Academic researchers: £250

Non-academic researchers: £350


This four day course will provide medical and health researchers with cutting-edge statistical formalisations and methods on causality.

We will introduce the concept and the decision-theoretic framework of causality, and address statistical causal inference using the properties of conditional independence in various scenarios. Identification and analysis of causal treatment effect of both randomized trials (non-compliance, intention-to-treat, Mendelian randomization) and observational studies (propensity analysis, doubly robust estimation) will be covered. The course will also cover more advanced topics, such as, mediation analysis, mechanistic interactions, probability of causation and colocalisation. We will illustrate these topics with examples of application in epidemiology, genetics, experimental psychology, neuroscience and other areas.

By the end of the course, participants will have a good understanding of causality, why it is important and how it could be addressed statistically in the health sciences.

Prerequisites: Basic statistical knowledge (equivalent to 2nd year undergraduate level in the UK), for example, normal distribution, binomial distribution, linear regressions, logistic regressions.

Registration: This course is now fully booked, and it is no longer possible to register.


Modern Mediation Analysis in Randomised Trials

Date: 23-25 June 2015

Time: 9:30-17:00

Venue: Room 5.205, University Place, The University of Manchester, Oxford Road, Manchester M13 9PL

Instructors: Professor Sabine Landau, Dr Richard Emsley, Dr Kimberley Goldsmith, Dr Cedric Ginestet


Free - staff and PhD students at The University of Manchester and Institute of Psychiatry, Psychology and Neuroscience.

£100 – external researchers.

Further details are available at the Eventbrite registration form.

This course is now fully booked, and it is no longer possible to register.

Requirements: This workshop will assume that participants have a basic knowledge of the standard analysis of RCTs. Some familiarity with structural equation modelling would be an advantage. Participants will need to bring their own laptop computer with Stata version 12 or higher.

Refreshments: Please note that refreshments are not provided as part of the course.

Handouts: Electronic copies of lecture notes and other course materials will be provided. Please note that hard copies are not provided.

Course aim and content:
DAY 1: The morning session will give an introduction to the terminology of causal inference and mediation analysis, describing both its potential and outlining the major difficulties.  The afternoon session will concentrate on what has been termed “statistical mediation analysis”. We will introduce standard approaches for constructing inferences for mediation parameters in the single mediator model using Stata's -sem- command, and discuss assumptions made by such analyses.  We then consider structural equation modelling approaches that allow us to relax some of these assumptions.

DAY 2: The morning session will introduce the concept of longitudinal mediation analysis and demonstrate how structural equation modelling can be used to investigate respective models. The afternoon session will then extend concepts to what has been termed “causal mediation analysis”. Specifically the aim of such mediation analyses is to further relax assumptions required by methods introduced on day 1 to allow causal effect estimation in the non-linear case and to allow for treatment × mediator interactions. Methods will be demonstrated using Stata commands paramed and mediation.

DAY 3: The morning session will continue to discuss approaches for causal mediation analyses. We will introduce approaches that can deal with measured post-randomisation confounders and hidden confounders in trials. Methods will be demonstrated in Stata using commands gformula, paramed and ivregress. The afternoon will cover more recent mediation methods extensions, e.g. to deal with missing data or measurement error, and sensitivity analysis, and will finish with a question and answer session to allow participants to present their own problems.