Principles and
practice of the analysis of multivariate data |
|
Classes are Tuesdays
15:15-17:00 and Wednesdays 13:15-15:00 in MD3 in the Mathematics Center, starting on September 12 Content The plan is to introduce general principles by stressing first (1) differences in types of study and importance of planning and protocols (2) discussing causal inquiries as motivation for much empirical research and comparing shortly different attitudes to causal inference (3) using several different research questions and the connected sets of data, which come from my main experience in research in the life sciences. Then the praxis comes after (4) giving some further research questions and sets of data to the participants, together with the research questions that started the data collection and the type of observational plans and description of the measurement devices. (Own data sets may be used instead, if a participant can explain the corresponding research questions to us and knows enough about the study plan and the data collection) Background Participants should analyse these research questions themselves and report results and difficulties, to be discussed in class, i.e. you will benefit most if you have some previous experience via a) a course on regression models including generalized linear models or multivariate analysis, b) having used a statistical software package such as for instance SPSS, STATA, S-Plus or R. Used material Examples of steps in analysis for the research questions and analyses discussed by me are summarized on a web page that is under construction: Graphical models: case studies Checks for nonlinearities: products and squares Part of the course is based on Chapters 6 and 7 of D.R. Cox and N. Wermuth (1996) Multivariate dependencies: models, analysis and interpretation, Chapman & Hall and on parts of Paul R. Rosenbaum (1995), Observational Studies, Springer Other material used in class is given below but password-protected since data analyses are not yet published or original publications are stored as pdf-files which are the property of the journals and therefore not to be made publicly accessible. Type of Examination The course may be taken to give 5 points for graduate students of mathematical statistics. The task is in this case to hand in a written report on the results of having analyzed one of the large data sets, along the lines proposed and discussed during the course. Password protected section (if it does not work please contact webmaster) |