Dr Chris Metcalfe (course organiser), Prof Tim Peters, Prof Jonathan Sterne.
Five days
13-17 May 2013
£1000
To provide an understanding of the statistical principles behind, and the practical application of, univariable and multivariable linear and logistic regression in medical, epidemiological and health services research. By the end of the course, students will:
have a thorough conceptual understanding of linear and logistic regression;
appreciate the common threads running through these methods, including stratified analysis, different options for handling explanatory variables, and concepts such as confounding and interaction;
have a working knowledge of the Stata commands to run these models, and a thorough understanding of the output generated from such a package;
know the basis on which analytical strategy and model choice is made, and how the results should be interpreted.
Those analysing data from medical, epidemiological, and health services research, who have used simple methods such as t-tests and chi-square tests, but who now wish to use multivariable methods to control confounding, accommodate interaction, and increase statistical power.
Participants must have knowledge of statistical methods to the level of the Introduction to Statistics course. Knowledge of Stata is essential. Participants will find knowledge of the basic study designs helpful: randomised controlled trials, cross-sectional studies, cohort studies, and case-control studies. Chapter 34 of "Essential Medical Statistics" is at the appropriate level (see Recommended reading below).
Revision of basic methods in a statistical modelling framework; basic terminology and concepts in modelling (stratified analysis to control for confounding, interaction, indicator variables); interpretation of model coefficients; simple and multiple regression and its synonyms (ANOVA, ANCOVA); odds ratios and relative risks logistic regression;analytical strategies. Teaching time is thirty hours. Learning will be by short lectures interspersed with practical sessions, including extensive Stata practicals on example datasets. Much of the course will involve discussions of how to interpret output from statistical packages such as Stata.
Course handouts will cover all material. All participants will be expected to have a copy of 'Essential Medical Statistics' with them during the course: Kirkwood BR, Sterne JAC. Essential Medical Statistics, 2nd ed. Oxford: Blackwell Science, 2003. The book can be purchased at a discount at the time of booking on this course.
For further information please contact short-course@bristol.ac.uk