This zip file (or folder) contains the Resume Randomizer program in two parts: the webpage (resume-randomizer-framemaster.html) allows an experimenter to define and input the characteristics of the resume or correspondence general template, while the executable file (resume-randomizer.exe) generates resumes from those templates.  Here we discuss use of the program in creating resumes, but the program is general enough to be used for other text randomization tasks such as creating cover letters.

https://data.nber.org/resume-audit/

Motivation for use of this software is discussed in: Lahey, Joanna and Beasley, Ryan A., Computerizing Audit Studies (July 1, 2007). Available at SSRN: http://ssrn.com/abstract=1001038 or http://dx.doi.org/10.2139/ssrn.1001038

Overview:

The included software is intended to assist researchers in performing large-scale resume audit studies by using computer-generated randomization to assign characteristics to resumes.

The experimenter first uses the webpage interface (i.e., checkboxes, buttons, and text entry areas) to create the template, which is effectively an outline for the resumes to be generated.  Each point in the outline determines the probability that text in its sub-points (i.e., characteristics) will get output to the resume files.  Outline points can be set to repeat a specified number of times, useful for generating work history by repeatedly choosing between possibilities for previous jobs.  Furthermore, the outline can be set up for matched resumes, such as for a matched-pairs audit (though the program is not limited to matching only two resumes at a time).  Specifically, points can be set either so that all of the matched resumes will choose the same sub-point, or so that they will all choose different sub-points.  Combining the repetition and matching settings can improve the above example by forcing all the resume files to choose the same sub-point (a matched characteristic).  That sub-point could contain text listing a specific previous job, or it could have sub-points and force matched files to choose different sub-points all describing functionally equivalent jobs (to keep the resumes from looking like copies).  The program is provided with example templates that demonstrate such uses.  The webpage explains template creation in further detail.

Once the template has been created using the webpage, the executable (resume-randomizer.exe) can be run any number of times.  Each time the executable is run, the experimenter can instruct it to generate any number of resumes, either not matched or matched within groups of any size.  Along with each resume, the program creates a record of the random choices that were made in the creation of that resume, sufficient for exact re-creation of that resume for use in analysis.

Instructions:

1)Unzip all of the files to the same folder.  Before running the program, if one of the files is named "resume-randomizer.ex" (note there are only two letters after the period), it must be renamed "resume-randomizer.exe".  The same goes for "w9xpopen.ex" -> "w9xpopen.exe".  These filenames were truncated to allow the sending of these files through email.  Any template files that are created must be placed in the same folder as the executable.

2)The webpage consists of three html files.  All three are necessary, but only "resume-randomizer-framemaster.html" should be loaded into your web browser.  The webpage has been tested and works in Firefox v23.0.1 and in Internet Explorer v10.0, but it does not work in Chrome v29.0 (due to the way Chrome handles inter-frame communication when in local mode).  The webpage contains further explanation and instructions.

3)Once one or more template files have been generated, run the executable file "resume-randomizer.exe" to generate resumes.  Four sample template files are provided to demonstrate the use of the program: "example_cover_letter_template.rtf", "example_resume_template.rtf", "example_resume_template_with_fragments.rtf", and "example_cyrillic_template.rtf".

Other notes:

A copy of the source code for the executable is included, named "resume-randomizer.py".  It is written in the Python programming language and is open-source under the Apache License 2.0.

The source code is hosted on GitHub at https://github.com/beaslera/resumerandomizer

Related

More from NBER

In addition to working papers, the NBER disseminates affiliates’ latest findings through a range of free periodicals — the NBER Reporter, the NBER Digest, the Bulletin on Retirement and Disability, and the Bulletin on Health — as well as online conference reports, video lectures, and interviews.

Economics of Digitization Figure 1
  • Article
The NBER Economics of Digitization Project, established in 2010 with support from the Alfred P. Sloan Foundation,...
claudiagoldinpromoimagelecture.png
  • Lecture
Claudia Goldin, the Henry Lee Professor of Economics at Harvard University and a past president of the American...
2020 Methods Lecture Promo Image
  • Lecture
The extent to which individual responses to household surveys are protected from discovery by outside parties depends...