Replication: Jun and Lee (2022)

Overview

This folder contains files to replicate results from the following paper:

Sung Jae Jun and Sokbae Lee. 2022. Identifying the Effect of Persuasion. https://arxiv.org/abs/1812.02276. Accepted for publication in the Journal of Political Economy.

Replication folder structure

The following diagram summarizes the organization of the replication files.

replications-JunLee-JPE         # Replication files for Jun and Lee (2022)
    ├── data                    # 	Datasets
    |   ├── ChenYang2019                        #	Dataset from Chen and Yang (2019)
    |   ├── DellaVignaKaplan2007                #	Dataset from DellaVigna and Kaplan (2007)
    |   ├── DellaVignaListMalmendier2012        #	Dataset from DellaVigna, List, and Malmendier (2012)
    |   ├── EnikolopovPetrovaZhuravskaya2011	#	Dataset from Enikolopov, Petrova, and Zhuravskaya (2011)
    |   ├── GerberKarlanBergan2009              #	Dataset from Gerber, Karlan, and Bergan (2009)
    ├── scripts                 # 	Replication scripts
    |   ├── libraries           #       	Stata module "persuasio"
    |   ├── logs                #       	Stata log files are saved here
    |   ├── table1.do           #       	Make Table 1
    |   ├── table2.do           #       	Make Table 2
    |   ├── table3.do           #       	Make Table 3
    |   ├── table4.do           #       	Make Table 4
    |   ├── tableD1.do          #       	Make Table D1
    |   |── figureD1input.do    #       	Prepare input for Figure D1
    |   ├── FigureD1.R          #       	Make Figure D1
    |   ├── tableE1.do          #       	Make Table E1
    |   ├── tableE2.do          #       	Make Table E2
    |   ├── tableH1.do          #       	Make Table H1
    |   ├── CY19_data.do        #		Prepare input for Table 2
    ├── results                 #	Replication Figure D1 and all the tables
    └── run.do                  #	Master script that calls the replication scripts (except FigureD1.R)

Instructions

First, download this repository. Then, run the following Stata script: run.do. This master script runs all the scripts required to replicate the main results except Figure D1, which can be generated by running FigureD1.R. Other scripts are contained in the scripts directory and are organized as follows:

• scripts/libraries contains user-built Stata commands used in the analysis.
• To generate the dataset ChenYang2019.dta, it is necessary to run CY19_data.do after downloading the original dataset from Chen and Yang (2019) at https://www.aeaweb.org/articles?id=10.1257/aer.20171765 and storing them at data/ChenYang2019. This dataset is already created and stored at data/ChenYang2019.

Before running run.do, the user must define the global macro Persuasion at the top of that script. Also, it is necessary to install the Stata module outreg by ssc install outreg, replace before running the do file.

Software, memory, and runtime requirements

Running this analysis requires Stata version 16 or higher. Add-on packages are included in scripts/libraries and do not need to be installed by the user. In particular, the package persuasio is a Stata module that estimates the effect of persuasion and conducts inference.

Memory requirements are minimal. Runtime is approximately 3 hours on an iMac (M1, 2021) when running the default specification of nbt = 10000 bootstraps.

Acknowledgment

The structure of the replication files is forked from Illinois Workplace Wellness Study: Public Use Data Repository at https://github.com/reifjulian/illinois-wellness-data.