This file serves as a README for the empirical analysis of 
Martin and Marx, "A Robust Test of Prejudice for Discrimination Experiments."
The file includes:

    1. Information on the datasets.
    2. Outline of the scripts used to generate the empirical results. 
    3. Data dictionaries for each dataset.

Any further questions can be directed to the authors at daniel@martinonline.org
and philiplmarx@gmail.com. 

-------------------------------------------------------------------------------------
 
1. Datasets

The empirical analysis makes use of publicly available datasets from two
existing discrimination experiments. These are:


- Reuben, Sapienza, and Zingales (2014), "How stereotypes impair women's careers
  in science." Proceedings of the national Academy of Sciences. The data from the
  experiment is available as an .xls dataset titled "sd01.xls" at:  

    https://www.pnas.org/content/111/12/4403/tab-figures-data

- Bertrand and Mullainathan (2004), "Are Emily and Greg More Employable Than
  Lakisha and Jamal? A Field Experiment on Labor Market Discrimination."
  American Economic Review.  The data from the experiment is available as a .dta
  dataset entiled "lakisha_aer.dta" at: 

    https://www.aeaweb.org/articles?id=10.1257/0002828042002561

In addition, both sets of authors have graciously granted us permission to include their original data in our replication package. 

-------------------------------------------------------------------------------------

2. Script Outline

master.do  

- This serves as the master file and assumes that the aforementioned datasets
  "sd01.xls" and "lakisha_aer.dta" are in the user's current directory. The file
  replicates all empirical results by calling one script file for each
  experiment:  

    -> analysis_RSZ.do
        
        - This script file includes all analysis using the experiment of Reuben,
          Sapienza, and Zingales (2014). This includes Figure 1, Table 1, and in-text
          results in Section 3.1 ("Empirical Application") and Appendix A.1
          ("Empirical Application").    


    -> analysis_BM.do

        - This script file includes all analysis using the experiment of
          Bertrand and Mullainathan (2004). This includes Figure 2, Table 2, and
          in-text results in Section 4.1 ("Empirical Application") and Appendix B
          ("Robustness of Selection Motive Results").

- Notes: The programs use the coefplot command, which can be installed by running "ssc install coefplot". 

-------------------------------------------------------------------------------------

3. Data Dictionaries

The following data dictionaries adopt definitions provided in the respective raw
datasets. They are replicated here for reference. 

Reuben, Sapienza, and Zingales (2014):

- treatment "Treatment" 
- session "Session"
- ownid "Employer's unique ID"
- owngender "Gender of the employer"
- part "Part (sequence of the candidate pair within a session)"
- grouptype "Type of candidate pair"
- pairid "Pair of candidates's unique ID"
- candidatenum "Candidate label"
- id "Candidate's unique ID"
- decision "Employer's decision unique ID"
- gender "Gender of the candidate"
- qguessnoinf "Expected performance with no information"
- qguesscheap "Expected performance with cheap talk information"
- qguesspast "Expected performance with past information"
- picknoinf "Picking choice with no information"
- pickcheap "Picking choice with cheap talk information"
- pickpast "Picking choice with past information"
- qcheap "Candidate's claimed performance"
- qansT "Candidate's performance (trail task)"
- qansI "Candidate's performance (first task)"
- qansII "Candidate's performance (second task)"
- ownqcheap "Employer's claimed performance"
- ownqansT "Employer's performance (trail task)"
- ownqansI "Employer's performance (first task)"
- ownqansII "Employer's performance (second task)"
- winI "Candidate with most sums (first task)"
- winII "Candidate with most sums (last task)"
- iat "Employer's IAT score"
- iat_count "Number of valid IAT trials (out of 120)"
- earnnoinf "Employer's earnings with no information"
- earncheap "Employer's earnings with cheap talk information"
- earnpast "Employer's earnings with past information"
- earnhyp "Employer's earnings if he/she picked this candidate"

Bertrand and Mullainathan (2004):

- id "ID" 
- ad "Ad #"
- education "0=not reported; 1=HSD; 2=HSG; 3=some col; 4=col +"
- ofjobs "number of jobs listed on resume"
- yearsexp "number of years of work experience on the resume"
- honors "1=resume mentions some honors"
- volunteer "1=resume mentions some volunteering experience"
- military "1=applicant has some military experience"
- empholes "1=resume has some employment holes"
- occupspecific "Occup. Specific"
- occupbroad "Occup. Broad"
- workinschool "1=resume mentions some work experience while at school"
- email "1=email address on applicant's resume"
- computerskills "1=resume mentions some computer skills"
- specialskills "1=resume mentions some special skills"
- firstname "applicant's first name"
- sex
- race
- h "1=high quality resume"
- l "1=low quality resume" 
- call "1=applicant was called back"
- city "c=chicago; b=boston"
- kind 
- adid "employment ad identifier"
- fracblack "frac blacks in applicant's zip"
- fracwhite "frac whites in applicant's zip"
- lmedhhinc "log median HH income in applicant's zip"
- fracdropout "frac HS dropouts in applicant's zip"
- fraccolp "frac college educ or more in applicant's zip"
- linc "log per capita income in applicant's zip"
- col "applicant has college degree or more"
- expminreq "min experience required, if any"
- schoolreq "specific education requirement, if any"
- eoe "1=ad mentions employer is EOE"
- parent_sales "sales or parent company"
- parent_emp "employment of parent company"
- branch_sales "sales of branch"
- branch_emp "emp of branch"
- fed "employer is federal contractor"
- fracblack_empzip "fraction blacks in employer's zipcode"
- fracwhite_empzip "fraction whites in employer's zipcode"
- lmedhhinc_empzip "log median HH income in employer's zipcode"
- fracdropout_empzip "fraction HS dropouts in employer's zipcode" 
- fraccolp_empzip "fraction college or more in employer's zipcode"
- linc_empzip "log per capita income in employer's zipcode"
- manager "manager wanted" 
- supervisor "supervisor wanted"
- secretary "secretary wanted"
- offsupport 
- salesrep "sales representative wanted"
- retailsales "retail sales worker wanted"
- req "1=ad mentions any requirement for job"
- expreq "1=ad mentions some experience requirement"
- comreq "1=ad mentions some communication skills requirement"
- educreq "1=ad mentions some educational requirement"
- compreq "1=ad mentions some computer skill requirement"
- orgreq "1=ad mentions some organizational skills requirement"
- manuf "employer industry is manufacturing"
- transcom "employer industry is transport/communication"
- bankreal "employer industry is finance, insurance, real estate"
- trade "employer industry is wholesale or retail trade"
- busservice "employer industry is business and personal services"
- othservice "employer industry is health, educ. and social services"
- missind "employer industry is other/unknown"
- ownership "ownership status of employer"

-------------------------------------------------------------------------------------


