This repository contains the main code for analyzing data during the Next Generation Social Science (NGS2) program conducted by DARPA.
R is primary analysis tool used on WITNESS. This repository contains all the analytic scripts used to prepare and analyze data coming from the WITNESS team's data collection efforts.
WITNESS empanels members of Gallup's nationally-representative panel. These data are collected via an external program and are inputs into the NGS2_WITNESS_empanelment_prep.R
script. The input data are CSV file(s) and the output is a cleaned CSV file.
The NGS2_WITNESS_Cycle1_experiment1_prep.R
and NGS2_WITNESS_Cycle1_experiment2_prep.R
scripts prepare the data (CSV files) from the experimental platform for analysis. The output -- given many inputs CSVs -- are two CSV datasets. In addition, the NGS2_WITNESS_Cycle1_difi_prep_exp2.R
prepares an additional set of analysis files.
The following R scripts are all used for analysis:
NGS2_WITNESS_Cycle1_exploratory_exp1.R
NGS2_WITNESS_Cycle1_exploratory_exp2.R
NGS2_WITNESS_Cycle1_confirmatory_exp1.R
NGS2_WITNESS_Cycle1_confirmatory_exp2.R
NGS2_WITNESS_Cycle1_bayesian_exp1.R
NGS2_WITNESS_Cycle1_bayesian_exp2.R
NGS2_WITNESS_Cycle1_difi_exp2.R
There are two R markdown notebooks that are used for presentation of results: NGS2_WITNESS_Cycle1_Exp1_Notebook.Rmd
and NGS2_WITNESS_Cycle1_Exp2_Notebook.Rmd
.
The NGS2_WITNESS_Cycle1_MASTER.R
will run the entire pipeline of scripts to produce output.
In addition to the R scripts for analysis, there is a Python file to aid in constructing payment ledgers to know how much participants have earned through their experiments. This file -- payout_calculations.py
calculates the monies earned by participants for a given cycle of experimentation. It requires a configuration file that contains local directory paths and other information.