balanced incomplete block design and mixed model
Christian Lerch
I was asked to provide methodological and statistical support for a clinical research project.
The proportion of children able to perform 5 different tasks (outcome: yes/no) will be investigated in 7 age groups. The aim of the study is estimation rather than hypothesis testing. Sample size calculation will be guided by the desired precision of the proportion estimate.
The tasks are rather closely related. It will not be feasible for each child to perform every task, so every child will be randomised to perform 3 of the 5 tasks (I will use algorithms for balanced incomplete block (BIB)designs). I do not think that position, period or task-by-period effects are important (and if, the BIB design will take account of these effects, at least partly).
For the statistical analysis plan, I wonder whether it is appropriate/necessary to split data in single age groups or not.
Moreover, is a mixed model (task as fixed, child as random) appropriate or are there more simplier models. And how can I take the BIB design into account (R or Stata code appreciated)? However, the desired output should look like "67% (95%-CI 60 to 74%) of 4 year old children are able to perform task 1".
Any hints, remarks or differing opinions are appreciated!
Best regards,
Christian
--
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3.5 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/atbrowser