Advanced Statistical Analysis Techniques
Full course description
The major objective of this course is to prepare students optimally for the use of statistics in their practical work and the period after. The student is taught to apply the most commonly used statistical analysis techniques in a responsible way. Also should he be better able to judge the statistical facets of research as carried out by others.
The training aims at applying advanced statistical techniques in a responsible way. The emphasis will be on concepts underlying the statistical techniques and on interpreting the results, with the mathematics being kept to a minimum. The course material is primarily based on SPSS software. The use of R and STATA will only be briefly approached.
The following techniques will be treated
Analysis of variance and (co)variance
- Linear regression
- Logistic regression
- Analysis of survival times
- Analysis of repeated measures (linear multilevel models)
For each topic there are two lectures and two tutorials. During the first tutorial, theoretical issues are discussed while emphasis on the interpretation of results obtained with SPSS on real data sets is given in the second tutorial. Concerning the lectures, the first one is more theoretical and involves the presentation of the method and the assumptions behind. In particular, the consequences of violating the assumptions are investigated. The practical interpretation of software outputs is also of great interest. In the second part, we analyze a real dataset together and debate over the best choices to make to analyze the data. Then, we discuss how the results can be summarized to be presented to an audience with minimal statistical knowledge.
After completing this unit the participants should have acquired the knowledge and skills required for the independent use and critical assessment of various (multivariable) statistical analysis concepts, procedures and techniques which are prominent in epidemiological research:
- Analysis of variance and covariance.
- Linear regression analysis techniques
- Logistic regression analysis for binary outcome variables
- Analysis of survival data
- Analysis of repeated measurements (linear multilevel models)
For each of this statistical technique, the participant should be able to deal with confounding, interaction and outliers, be aware of the assumptions underlying the use of the technique, know some advantages and disadvantages of the technique, interpret results and use dummy coding. The participant should also be able to choose an appropriate statistical analysis strategy, given a specific epidemiological research question and study design.
Basic literature - Field A. Discovering statistics using IBM SPSS statistics (4th ed). London: Sage Publications Ltd, 2013.