# R package All scripts, raw data files used for the simulation are inside the R package provided. Out of it only the minimal files and results are provided. # Relevant figures files and tables ## Files - inst/results/**group_fp.csv**: Main table - inst/figures/**bar_plot.tiff**: Main figure # How to reproduce the analysis # rtsimpack --------- All the functions used in the analysis are uploaded as a R package, **rtsimpack**. They all have their own documentation. (See https://devtools.r-lib.org/ particularly the function devtools::load_all() to learn how to load and use the package in your R session) Drake plan ------- The main file to explore when willing to reproduce the analyses is drake/drake_plan.R This file is a drake pipeline that will reproduce the whole analysis starting from the raw files to the final table and figures. It should be quite self-explanatory. (See the documentation of this type of pipelines on https://github.com/ropensci/drake) Dependencies ------- This package depend on many other, therefore an **renv** file is distributed in the package and the whole dependency can be installed using the renv package particularly the renv::restore(function) (See https://rstudio.github.io/renv/articles/renv.html) Contact us ------- If you still cannot make it run or have any doubts please feel free to send an email to luis.moris.fernandez(at)gmail.com **Caution**: 1. The simulation process is computationally intensive (1 to 4 days) therefore we recommend the user to reduce the number iterations in the simulations process (10000 is reasonable in a common laptop). 2. Although it is mentioned inside the drake plan, the number of iterations does not correspond to ni, but to ni*nchunk. The computation of each dataset is divided in nchunk chunks, so it can be parallelized in an easier way. 3. If you don't want to run the simulation in parallel you have to modify/comment the lines corresponding to the **future** package. (See drake documentation) ## Differences between reported results and OSF files and your own simulation Unfortunately analyses were run without setting a random seed, this means that small variations (irrelevant to interpretation purposes) may appear between: the results reported in the paper, those appearing in OSF and those run by other people. Nonetheless, they are all comparable and lead to the same conclusions.