| Description |
As discussed in the paper, there are three main survey waves used in the analysis: baseline, endline, and follow-up.
Programs 11-15 create datasets that turn the raw baseline data files into intermediate baseline files, with datasets for the women, child, household, and family level.
Programs 21-25 do the same thing for the endline, while programs 31-35 do the same thing for the follow-up. These latter programs also merge in the necessary ID code (q00_001a) that gives the baseline ID code to facilitate merging across survey waves.
Programs 41-44 build the key analysis datasets for women and children at the endline and follow-up. This includes creating all necessary constructs.
Finally, the primary analysis programs are run in programs 60-62, which create all the tables in the main paper. Program 61 relies on the Excel file “control_file_all_regs.xlsx” to specify the precise regressions and the tables – each row runs a separate regression, with the outcome, additional control variables, and required “if” statement to subset the data.
Programs 81-85 and 91 use the Midline survey data to estimate Lee bounds impacts (Appendix Table A1). Finally, program 92 runs subgroup impact results, creating Appendix Tables A3 and A4. Appendix Table A2 is created as part of program 60. |
