UK Causal Inference Meeting 2017
Final Programme now available
The University of Essex is pleased to host the 5th annual UK Causal Inference Meeting (UK-CIM), which will take place from Wednesday 5th to Friday 7th April 2017. The pre-conference workshops will be held on Tuesday 4th April.
The meeting is sponsored by the ESRC Research Centre on Micro-Social Change (MiSoC) and Understanding Society: The UK Longitudinal Household Study based at the Institute for Social & Economic Research.
Elizabeth A. Stuart
- Friday 17th February 2017: Deadline for a) ‘Early Bird’ registration fee and b) submission of abstracts for talks and poster presentation
- Early March 2017: Authors notified whether their abstracts have been accepted.
- 20th March 2017: Registration closes
Details of the meeting (including registration fees, suggestions for accommodation and travel) can be found by clicking the relevant tabs above.
Aims and Objectives
UK-CIM is an initiative to organise a regular UK based meeting on causal inference as a collaborative effort across the methodology research community in the health, economics and social sciences.
UK-CIM aims to:
- Provide a forum for people interested in causal inference to meet informally
- Provide a forum for early career researchers to highlight their work
- Offer opportunities for networking to foster future research opportunities and collaborations
Please note that registration for the meeting is not restricted to people from the UK, and we welcome participation from anyone who would like to attend.
Theme of the Meeting
The theme of the meeting is “Causal Inference in Health, Economic and Social Sciences”. Causal inference is broadly defined, and the focus is on methodology and challenging applications, though presentations relating to interesting applications that highlight necessary methodological extensions are also encouraged.
For questions about the UK-CIM 2017, please contact the local organisers at email@example.com.
For questions about the UK-CIM initiative, please contact the chair Richard Emsley at Richard.Emsley@manchester.ac.uk or any member of the steering group .
The keynote speakers are:
Elizabeth A. Stuart
Meeting and course fees
Note that the registration fee includes the wine reception on 5 April and conference dinner on 6 April.
|Meeting only||Meeting plus workshop|
|Early Bird (until 17 February)||£200||£250|
|Late Bird (18 February until 20 March)||£260||£310|
|PhD students (until 20 March)||£100||£135|
Abstract submission: A discount code will be emailed to those (non-students) who submit an abstract entitling them to “Early Bird” registration fees, provided they register before 20 March.
One-day rate: A special one-day-only rate is available on request. Please contact firstname.lastname@example.org for details.
Below are hotels we recommend in Colchester:
- Wivenhoe House Hotel - from £89.00 per night
- The Rose & Crown - from £77.00 per night
- The George Hotel - from £75.00 per night
- The North Hill Hotel - from £69.50 per night
- Premier Inn Colchester Central – from £65.00 per night
- Premier Inn Colchester Town – from £55.00 per night
How to find your way to and around the University of Essex Colchester campus by car, bus, train and taxi:
If you are travelling by car, you may park free of charge on campus for the duration of the conference by printing off and completing the permit below, and displaying it in your windscreen (see recommended car parks on the map above, cannot be used at the Multi-deck carpark).
The conference dinner will be held on Thursday evening, 6th April, at The Rose & Crown Hotel, Colchester. Attendance at the conference dinner is included in the registration fee.
We will be running the following workshops on 4 April.
- Workshop A: José Zubizarreta (half-day)
- Workshop B: Johan Steen (half-day)
- Workshop C: Rhian Daniel (full-day)
Note: Those who register for a workshop can attend either Workshop C or both half-day workshops A and B.
Workshop A: Optimal matching methods for causal inference
José Zubizarreta (Columbia University)
In observational studies of causal effects, matching methods are often used to approximate the ideal study that would be conducted if it were possible to do it by controlled experimentation. In this workshop, José Zubizarreta will discuss new advancements in matching methods that allow the investigator to overcome three limitations of standard matching approaches, and: (i) directly obtain flexible forms of covariate balance; (ii) produce self-weighting matched samples that are representative by design; and (iii) handle multiple treatment doses. He will also discuss extensions to matching with instrumental variables and in discontinuity designs, and for matching before randomization in experiments. These methods will be illustrated with the R package designmatch for R.
Workshop B: Flexible models for causal mediation analysis: an introduction to the R package medflex
Johan Steen (Ghent University, Belgium)
In most, if not all, of the empirical sciences, mediation analysis has become the applied practitioner’s primary statistical tool to improve one’s understanding of the processes or mechanisms through which a causal effect of interest comes about. Traditional methods for mediation analysis building on the linear structural equation modelling tradition from social sciences, however, often fall short, as they tend to produce estimates for mediated and direct effects that, given adequate adjustment for confounding, can only be assigned a well-defined and causal interpretation in strictly linear settings. In this workshop, I will discuss a novel modelling and estimation framework for mediation analysis that builds on a more formal approach to causal inference and thereby addresses much of the shortcomings of traditional methods that are most pronounced in the presence of interactions or other forms of nonlinearity. This approach has intuitive appeal as it requires fitting so-called generalized linear 'natural effect models' for potential outcomes that directly parameterize the component effects of interest. Using worked examples, I will demonstrate how this approach, by casting mediation analysis within a GLM framework, considerably simplifies reporting and hypothesis testing. A comprehensive assessment of mediating mechanisms becomes particularly challenging whenever more mediators come into play (especially when such mechanisms are interdependent and/or interact in their effect on the outcome). The proposed natural effect modelling approach offers elegant solutions to cope with this increasing complexity in the face of multiple mediators. In this workshop, we will illustrate how progress can be made using the functionalities of the R package medflex.
Steen, J., Loeys, T., Moerkerke, B., & Vansteelandt, S. (2017). Medflex: An R Package for Flexible Mediation Analysis Using Natural Effect Models. Journal of Statistical Software, forthcoming.
Workshop C: Introduction to causal inference
Rhian Daniel (London School of Hygiene & Tropical Medicine)
By popular demand, Rhian Daniel will be running “Workshop C” from UK-CIM 2016. Her workshop is aimed at newcomers to Causal Inference, and will cover the foundational topics needed to follow the majority of the talks.
UK Causal Inference Meeting participants
Bianca De Stavola (right) with the prizewinners for Best Early Career Presentations, Laura Forastiere, Michael Wallace and Simon Newsome
Bianca De Stavola (right) with the prizewinners for Best Posters, Janine Witte and Jessica Rees.
In partnership with: