Principal speaker
Dr Stephen Occhipinti
This workshop will cover contemporary approaches to the analysis of data from: a) survey designs based on cluster samples; and b) longitudinal designs (e.g., experiments with many repetitions, surveys again, RCTs with follow-up) based on the use of multilevel modelling (MLM). There are many benefits associated with MLM when used correctly and with appropriate designs. These extend to dealing with attrition and missing data, variable spacing of assessment points in repeated measures designs, and incorporating hierarchical data structures much more flexibly and efficiently than in the case of the ANOVA family. Most importantly, MLM techniques provide a far better fit to the predictions that researchers often seek to test than do many traditional analyses. However, in the no free lunch tradition, there are a number of challenges to designers of multilevel studies and to those who would analyse them. Statistics never make up for the poor design and execution of studies. Further, many available multilevel courses are very general and pay little attention to longitudinal cases. Accordingly, the aim of the workshop is to introduce people who have an intermediate knowledge of ordinary linear regression to the use of MLM especially as it relates to longitudinal designs. As well, we will discuss aspects of the optimal design of studies that are to be analysed via MLM.
Event categories
RSVP
RSVP on or before Thursday 5 June 2014 , by email gsbrc@griffith.edu.au , or by phone 0737351186