This repo contains all code used to validate the ART-C procedure, documents explaining the process, and logs of results created along the way.
For an approachable explanation of what we did and how we did it (including all of our analysis and additional analysis not included in the paper), read data_analysis.md.
- generate_data_and_run_contrasts.R: R code that was used to generate all data sets and run all contrasts.
- process_raw_logs.Rmd: Raw R markdown notebook that was used to compute observed Type I error rates and observed statistical power for each design. Took in raw logs (output from generate_data_and_run_contrasts.R) with p values for each trial. Raw logs are in folder raw_logs. Outputs res_ults.csv.
- data_analysis.Rmd: Raw R markdown notebook used to conduct all analysis from res_ults.csv and create data visualizations seen in paper. Also contains additional analyses not included in paper.
- process_raw_logs.md: Rendered version of process_raw_logs.Rmd
- ata_analysis.md: Rendered version of data_analysis.Rmd.
- Type_I_all_designs.csv: csv file containing observed Type I error rates for all "designs".
- power_all_designs.csv: csv containing observed statistical power for all "designs".
- raw_logs: Directory containing all raw logs generated by generate_data_and_run_contrasts.R. Logs contain results of statistical tests run on data drawn from different types of population distributions. Directories are labelled with numbers that correspond to the distributions. This isn't particularly useful since I haven't included a legend explaining what each type is. Also contains res_ults.csv -- the csv file created by process_raw_logs.Rmd. It has the same information as a combination of Type_I_all_designs.csv and power_all_designs.csv just in a different format, which is needed to run data_analysis.Rmd, but is more difficult to read.