Raw data and results for the paper "Conditional non-parametric bootstrap for non-linear mixed effect models"

*Data* (comets_condBoot_data.zip)

Data was simulated according to an Emax model (scenarios 1 and 2) or a Hill model (scenarios 3 and 4). The archive contains 4 folders with the data simulated in the first 4 scenarios (N=200 simulated datasets in each folder):
- scenario 1 - pdemax.rich
- scenario 2 - pdemax.sparse
- scenario 3 - pdhillhigh.rich
- scenario 4 - pdhillhigh.sparse
The data used in scenarios 5 and 6 was a subset of the datasets simulated in scenarios 3 and 4 respectively. In scenario 5, 20 subjects were taken from each dataset (subjects 1-5, 26-30, 51-55, 76-80) from the datasets in folder pdhillhigh.rich. In scenario 6, the datasets were constituted by the first 20 subjects from each sampling group of the data simulated in pdhillhigh.sparse.

*Results:* (comets_scenarioXXX_results.zip, XXX=1,.. 6)

6 simulation scenarios were assessed in the paper. Each file corresponds to 1 of 6 folders, one for each scenario:
- scenario 1 - pdemax.rich/results
- scenario 2 - pdemax.sparse/results
- scenario 3 - pdhillhigh.rich/results
- scenario 4 - pdhillhigh.sparse/results
- scenario 5 - pdhillhigh.n20rich/results
- scenario 6 - pdhillhigh.n20sparse/results

In each "results" subfolder, the results for each bootstrap method and each dataset are written to a separate file, eg for simulation 1 in the first scenario:
- case bootstrap: scenarioHill1_bootstrapCase_sim1.res  
- non-parametric bootstrap: scenarioHill1_bootstrapNP_sim1.res
- conditional non-parametric bootstrap: scenarioHill1_bootstrapNPc_sim1.res
- parametric bootstrap: scenarioHill1_bootstrapPar_sim1.res
The folder also contains:
- the saemix estimates for the 200 simulations: scenarioHill1_fitOrig.res
- tables with the bias and SE for the different bootstraps over the set of simulations, used to evaluate the methods: rbiasSEboot200.res, rbiasSEboot.res, rbiasWRsampleEstimates.res