 |
 |
Short Course : Applied Multivariable Regression Modelling Lecturers: Prof. Dr. Willi Sauerbrei (University of Freiburg), Dr A. Perperoglou (University of Essex) Dates: 7 – 11 September 2015 Multivariable regression models are widely used in all areas of science in which empirical data are analysed. In research, data on several variables may be collected to investigate the interrelationships among them, or to determine factors which affect an outcome of interest. An example from medicine is the relationship between the survival time of a patient following treatment for cancer and potentially influential variables (known in this context as prognostic factors such as age, size of the tumour, its aggressiveness, and so on). Often, effects of more than 10 potentially influential variables must be considered simultaneously in a single model. For several reasons a smaller model which includes only a subset of the variables may be preferable. An aim of the analysis, therefore, is the selection of a subset of ‘important’ variables with impact on the outcome. With specific attention to variable selection and determination of the (non-linear) functional form of the effect of continuous variables we discuss statistical methods to develop models which try to best answer specific subject matter questions in the framework of regression models. The aim of the course is to guide practitioners in making multivariable model building simpler, transparent and more effective. The target audience is researchers from medical, physical, social and other sciences where regression models play a central role, as well as graduate students studying regression modelling and professionals in statistics. Several parts of the course are based on the textbook by one of the lecturers ( http://www.imbi.uni-freiburg.de/biom/Royston-Sauerbrei-book/). It provides a readable text giving the rationale of, and practical advice on, a unified approach to multivariable modelling. The course is split into two sessions, a morning lectures session where basic ideas, methods and theory are introduced, and an afternoon hands-on workshop on applying the methods using R software. The topics covered include: - Basic introduction to the foundations of regression models (linear, logistic, Cox regression model, estimation, tests) - Variable selection methods, handling categorical predictors - Modelling continuous predictors (categorization, fractional polynomials (FPs), Multivariable FP procedure) - Interactions (two binary variables, binary-continuous, two continuous, time-varying effects) - Various issues (eg comparison of MFP with splines, model stability, reporting of observational studies, meta-analysis of functions) A basic knowledge of key statistical principles and of the linear regression model is assumed. Registration required, deadline 15 July 2015 Fees: Essex PGR students, alumni, staff : £300, Other academia and not-for-profit organisations: £400, Commercial: £800 This is a post ECDA2015 conference course, but no conference registration or attendance is required. webpage: http://ecda2015.com/post-conference-course/ Register via: https://shortcoursesgateway.essex.ac.uk/Guests/GuestCourse.aspx?CourseRef=AMRM:DMS&dates= *For more information contact * Aris Perperoglou [log in to unmask] tel: +44(0) 1206 87 3036 You may leave the list at any time by sending the command SIGNOFF allstat to [log in to unmask], leaving the subject line blank.