XClose

Institute of Clinical Trials and Methodology

Home
Menu

Practical use of multiple imputation to handle missing data in Stata

Next course: TBC

Our aim in this course is to provide participants with the ability to analyse their own data using multiple imputation, but also to be aware of the pitfalls and limitations of the technique.

We will give plenty of practical examples from our own experience of analysing data in medical research.

We welcome participants bringing their own data and problems, and one session is dedicated to discussion and possibly "live" analysis of some participants' data.

tutorial based on this course has appeared in Statistics in Medicine.

Course aims

  • Explain the problems of missing data and the need for methods such as multiple imputation
  • Explain how multiple imputation works, with a focus on imputation by chained equations (ICE)
  • Explain how multiply imputed data are analysed
  • Enable participants to analyse data by multiple imputation in Stata using the commands mi impute chained and mi estimate
  • Give participants an awareness of the assumptions underlying multiple imputation and of its limitations.

Target audience

The target audience for this course is researchers needing to analyse incomplete data:

  • Attendees are expected to be familiar with running Stata from the command line (i.e. not using menus) at least to the level of fitting a regression model to complete data and producing simple graph
  • No prior knowledge of multiple imputation is assumed.
  • Participants should bring their own laptop computer with Stata 12 or newer installed. Participants without Stata should contact the course administrator and we will aim to provide a temporary copy.

The course would also be suitable as a refresher for people familiar with multiple imputation using Stata’s ‘ice’ and ‘mim’ commands wanting to learn about the newer ‘mi impute chained’ and ‘mi estimate’.

Software required

All participants will need their own laptop running Stata 12 or newer, as we will be using 'mi impute chained' which was new in Stata 12.

It would save time if you could download and install the datasets for the practical sessions before the course.

Data for practicals

The data sets for the course practicals are in a zip file. We suggest you unzip these to a directory on your hard disk that you create specially for this course.

How to do this may depend on your browser, etc., but you should be able to get the files just by clicking on the link above. When you see the list of files, click Extract. Then select the directory you want the files to go in. Alternatively, save the zip file to disk and double-click the saved file.

Fees

The course fees cover copies of the course lectures and practicals, lunch and tea/coffee breaks. Delegates are advised to arrange their own travel and accommodation.

  • UCL staff and UCL students outside of ICTM: £200
  • Non-UCL students: £200
  • Not-for-profit organisations: £360
  • For-profit organisations: £480

This course is free for staff from Units within ICTM (CRUK-CTC, CCTU, MRC CTU at UCL and PRIMENT), although the number of places for this group is limited.

Facilitators

You may also be interested in

  • Flexible Imputation of Missing Data (2012) by Stef van Buuren
  • Multiple imputation and its application (2012) by James Carpenter and Mike Kenward
  • Other ICTM Short Courses