quality assurance - Statistics checklist Feb2021.pdf
(309.51 KB, PDF)
… as a mandatory tool. The recommendations in this list are based on decades of broad experience in statistical consultancy and analysis, but can be discussed with a member of the Methodology & Statistics dept. in case of doubt (for details of whom to contact then, see https://stat.mumc.maastrichtuniversity.nl/consultancy ). Finally, until further notice, statistical quality control is no part of the audits of research quality in CAPHRI. The checklist below has three parts: design, analysis, sample … as a compromise between a parallel groups trial and cross-over trial. A stepped wedge cluster randomised trial is likewise a compromise between a cluster randomised trial and cluster randomised cross-over trial. Before choosing a stepped wedge design, contact a statistician to discuss the pros and cons, and the analysis and sample size needed. 4) In a parallel groups (incl. multicenter) trial and a cluster randomised trial: choose a randomisation method (simple, block/stratified, minimisation) and … to complex models that give rise to numerical and statistical problems when applied to data. 31) Repeated measures on persons nested within organisations can be analysed with so- called 3-level mixed (multilevel) regression models. For these, always contact a statistician with good expertise in mixed modeling. 32) Exception: if the number of organisations in the study is small, say less than 10, then we treat organisation as fixed instead of random, using dummy coding, since the number of …