MRes course in Medical Statistics (MMB8028)

This is an introductory course in medical statistics comprising sixteen lectures and practicals which will be held between October and December 2012.  The course will be given by Professor John Matthews from the School of Mathematics and Statistics : a timetable for the course can be found here.

The course will concentrate on methods for the analysis of data.  Topics covered will include simple descriptive measures (means, medians, standard deviations and centiles), the Normal distribution, hypothesis tests for continuous data (t-tests and alternatives for non-Normal data) and binary outcomes (c² tests and Fisher’s Exact test); confidence intervals.  Methods for estimating required sample size will be discussed.  Regression and correlation will also be covered.  Other issues included in the course are : the use of P-values, adjusting analyses for confounding variables and approaches to the analysis of skewed data.  There will be four practicals using the statistical package MINITAB.  In order to do these practicals you must have access to the Common Desktop PC clusters located around the University and run by Information Systems and Services. 

The first part of the course deals with material that is well covered by many books on introductory medical statistics.  Ones that are particularly recommended are:

Campbell, M.J. and Machin, D.,  Medical Statistics : a commonsense approach. Wiley, Chichester, 3rd edition 1999. See also the fourth edition, Campbell, Machin and Walters, 2007, which is a slightly longer text.

Bland, Martin, An introduction to medical statistics.  OUP, Oxford, 3rd edition, 2000

For fuller coverage of the subject, including many topics that are not in this course, a valuable text is:

Altman, D.G., Practical statistics for medical research. Chapman and Hall (now Chapman & Hall/CRC Press), London, 1991.

Supplementary notes for the lectures on this course are available from the links below.  Some are simply copies of lecture slides but others are more extensive notes.  The notes are intended to be usable without attendance at the associated lecture and may be of interest to independent medical research workers not attending the course.  Such people are welcome to use the notes: they are all in PDF format.  The slides shown in the lectures are given in the second link on each line: these will not be available until after the lecture(s).

• Introduction: descriptive statistics and the Normal distribution (corresponds to lecture 1 & 2)
• Standard errors and confidence intervals (corresponds to lecture 3)
• Hypothesis testing using the t-test (corresponds to lecture 4)
• Regression and Method Comparison (lecture slides only) (correspond to lectures 5 & 6)
• Analysis of Binary data, including the c² test (corresponds to lectures 7 & 8)
• Interaction (corresponds to lecture 9)
• Sample size determination (corresponds to lecture 10)

The following are handouts for lectures given in previous versions of the course.  They are not part of the module but may be useful when you analyse your data or if you come across these techniques in your reading.

• Measuring measurement error
• Analysis of variance
• Analysis of covariance
• Skewed data (lecture slides only)
• Survival Analysis
• Analysis of serial measurements (reproduced from BMJ, 1990, 300, 230-235, by kind permission of British Medical Journal)

Question sheets, datasets and solution sheets for some of the practical classes can be found below.

• Descriptive data and Normal distribution practical: question sheet, data and solutions
• t-tests practical: question sheet, data,(for questions 2 and 3) and solutions
• Regression practical: question sheet, data HBHT.MTW, GRAVES.MTW, ANSC.MTW and solutions
• Categorical data practical: question sheet, data and solutions

To start Minitab on a machine in a University cluster, click Start -> Programs -> Minitab -> Minitab 16 Statistical Software

Any questions about the course should be sent to John Matthews.