---

CHRIS ROBERTS'S UNIVERSITY OF MANCHESTER PERSONAL HOMEPAGE

Biostatistics Homepage

---

Medical Statistics (MATH38071) Course Materials 2015/16 Course

Course materials for MATH38071 Medical Statistics will be made available below prior to and during the course. Past examination papers with solutions for MATH38071 are currently available below.

Course Work Assessment
Link for Course Work Assessment BMJ Paper
Course Work Feedback
Notes (Pre-lecture)	Notes with Annotation (Post-lecture)		Critical Appraisal Exercises		Exercise Sheets	Exercise Sheet Solutions
Notes Part I (Weeks 1-5)	Introduction		Critical Appraisal 1		Sheet 1	Solutions 1
	Basic Analysis Continuous Data		Link for Crit. App. 1 Paper		Sheet 2	Solutions 2
	Analysis of Binary Data				Sheet 3	Solutions 3
Notes Part II (Weeks 6-10)	Power and Sample Size		Critical Appraisal 2		Sheet 4	Solutions 4
	Treatment Allocation Methods		Link for Crit. App. 2 Paper		Sheet 5	Solutions 5
	Analysis using Baseline Measurements				Sheet 6	Solutions 6
	Equivalence & Non-inferiority		Critical Appraisal 3		Sheet 7	Solutions 7
	ITT Analyses		Link for Crit. App. 3 Paper		Sheet 8	Solutions 8
	Crossover Trials				Sheet 9	Solutions 9
	Systematic Review & Meta-analysis				Sheet 10	Solutions 10
School of Mathematics Statistical Tables
Past Examination Papers & Solutions
Year		Questions in 2015/16 syllabus of MATH 38071		Solutions
2003/4		A1,A2,A3, B1,B2,B3(i-ii)		2003/4
2004/5		A1,A2,A3,A4,B7,B8		2004/5
2006/7		A1,A2,A3,B7,B8,B9(i-ii)		2006/7
2007/8		All questions		2007/8
2008/9		All except B7(iii-iv)		2008/9
2009/10		All questions		2009/10
2010/11		All questions		2010/11
2011/12		All questions		2011/12
2012/13		All questions		2012/13
2013/14		All questions except A5		2013/14
2014/15		All questions		2014/15
2015/16				2015/16

---

METHODOLOGICAL RESEARCH

My main research interests are:

The Design & Analysis of Clinical Trials with Clustering.

This includes trials of non-pharmacological treatment such as talking or physical therapies, operator effect such as surgery, or group therapies. In such trials the assumption of statistical independence of patients is violated. This is similar to cluster randomised, also called group randomised, trials, but because the clustering effect depends on treatment between arm heterogeneity may be found. This can make the analysis more complex to analyse than cluster randomised trials. For example a group therapy may be compared with a therapy delivered to individuals. I also have a methodological interest in cluster randomised trials.

Statistical Methods for Assessing The Reliability of Nominal and Ordered Categorical Data.

This includes the development of kappa-type statistics, design power and sample size, and the application of non-linear mixed models to such data. I have written some STATA ado files that implement these methods.

---

Updated 20 April 2017

Top

---