Per-replicate results of the simulation study on analyses of clustered binary outcomes with few clusters and short series

Per-replicate results of the simulation study in the paper Choosing an analysis for clustered binary outcomes when clusters are few and series are short: a simulation study and two clinical applications (Vedovelli et al., submitted to Statistical Methods in Medical Research). MMM_results_full.tar: the original simulation runs, one R data file per chunk of replicates (mechanism, scenario, number of clusters N, number of occasions T, hypothesis: manual = null, mmlb = alternative), 28,800 files. MMM_results_v4_per_replicate.tar: the per-replicate tables read by the table chain of the paper (results_v4/S4/rows_v4.rds and the files of S4, S6, S6cc, S6b/ns, S7d, S7e, S10/degen and S11 listed in README_deposit.txt). The table variants used in the paper (cc, nc, cv) are rebuilt from these files by the scripts of the code release. Code that produced and analyses these files: https://doi.org/10.5281/zenodo.23917381 (https://github.com/lucavd/clustered_paper). R package fewclust: https://doi.org/10.5281/zenodo.23917406. Simulated data only; no patient data.