14 Feb Topics in Meta-Analysis
9:00 am - 4:00 pm
statistics seminar | level: intermediate
registrations and venue via KU Leuven
This colloquium will bring together the Social Sciences, the Economics and Business area and Biomedical Sciences to present an overview of their use of Meta-Analysis. In the morning, there will be two lectures on Meta-Analysis: a general introduction, followed by a R practical session; then, a second contribution from the point of view of marketing research will be presented. The afternoon session will contain a lecture on Meta-analysis and Network Meta-Analysis in medical research.
9:00 – 11:00: ''Meta-analysis: when, why, and how? A first introduction.''
Prof. Van den Noortgate & Belén Fernández Castilla, KU Leuven, Faculty of Psychology and Educational Sciences, Kulak Kortrijk Campus
In this session, we will have a look at reasons for the ‘failure to replicate’ problem, and how meta-analysis can help us in this regard. In addition, an introduction will be given in basic meta-analytic techniques and models, by analyzing together a simple dataset in R.
11:00 – 11:15: Coffee break
Van den Heuvelinstituut, 00.02 (Main hall)
11:15 – 13:15: ''Theory-Testing using Meta-Analysis: Applications in Marketing and Management”
Prof. Van Vaerenbergh, KU Leuven, Research Centre for Marketing, Brussels Campus
Researchers in marketing and management often use meta-analysis not only to synthesize empirical evidence, but also to test specific theories. In this session, we will explore how meta-analysis can be used for such purposes. Participants will be introduced into how nomological frameworks specifying relationships between variables are built. We will also demonstrate how to test these frameworks using path models in R.
13:15 – 14:15: lunch break
14:15– 16:00: ''Meta-analysis and network meta-analysis in medical research. The frequentist versus the Bayesian approach''
Prof. Emmanuel Lesaffre, KU Leuven, L-BioStat.
In the first part of the talk, we compare the frequentist and the Bayesian way to perform a classical meta-analysis. It is shown that the Bayesian approach and software offers some attractive features especially with sparse data.
In the second part of the talk, we introduce network meta-analyses. A classical meta-analysis summarizes the differential effect of two treatments or classes of treatments, say A and B. In contrast, a network meta-analysis compares the differential effect of treatments A, B, C, … . The evidence of each comparison, say A vs B, will then be obtained from direct evidence (studies directly comparing A with B), and indirect evidence (studies comparing A vs C, B vs C, etc.). We first introduce the general concepts of a network meta-analysis and then compare the frequentist with the Bayesian approach. We highlight the advantages and disadvantages of both approaches.
Ample examples from dental research will illustrate the concepts and software implementations will be discussed.
Knowledge of statistical inference, statistical modelling and R is recommended.
Van den Heuvelinstituut (VHI)
PC class H2 - 02.09
The colloquium is organized in a PC Class (30 desktop PC are present in the class). However, participants can bring their own laptop with a recent version of R (RStudio) installed.
PhD's or postdocs of a Flemish university: free of charge
Other academics: 60 €
Non-profit/Public sector: 100 €
Private sector: 200 €
The registrations are CLOSED.
Prof. Van den Noortgate & Belén Fernández Castilla, Prof. Van Vaerenbergh, Prof. Emmanuel Lesaffre