NGO Experience and Employer Interest

The key manipulations in our audit study are applicants' prior work experience. Each applicant will have two prior jobs on their resume. These jobs are randomized to be for either a for-profit or non-profit employer. We identify real for-profit and non-profit employers in the CBSA where the targeted job is located. We identify sets of for-profit and non-profit employers within this CBSA that are relatively large and well-known and others that are small and lesser-known. Based on a set of real resumes, we use LLMs to create sets of skills and past experiences that correspond to the occupations we will be applying for. We randomize these skills such that we can isolate the effect of the applicant's prior employer, as opposed to the effect of inferences regarding skills that hiring firms might make based on the presence of certain employers on an applicant's resume.

Individuals are randomized to have either typically female or typically male names as well as names that are typically Black or White.

Applicants' resumes will also list an undergraduate degree from a college located in the CBSA or state in which the job is located. Where possible, these are large, well-known state universities. Where no such university exists, we list a private college in the area that is ranked below the top-ranked public college in our collection of created resumes. Our motivation here is that highly elite private universities may consistently elicit high callback rates, decreasing our ability to measure the effects of the other resume components that are of more central interest. We randomize applicants' GPAs to be between 3.5 and 4.0 inclusive. Besides the name of the school, we will randomize whether the resume indicates the applicant participated in an activist-related or non-activist-related club in college.

2025-10-13

2026-01-31

Our primary outcome is whether employers respond to job applications. We are interested in how this callback rate changes when a resume includes prior NGO experience.

We will examine how the effect of NGO experience on callback rates varies with hiring employer characteristics (e.g., history of reputationally hazardous events, regulatory exposure, political partisan composition within employer), the nature of the NGO, participation in an activist club in college, and with applicants' gender and race.

We implement randomization and resume construction using a custom application that is similar to the Resume Randomizer created by Lahey and Beasley (2009). Items are randomly combined creating randomized resumes that follow standard resume templates containing: name, contact information, education, college extracurricular activities, and work history.

Once a job is identified, the applicant's profile will be randomly chosen from among the designed profiles. Each job opening will receive two applications: one from an applicant with NGO experience and one from a different applicant without NGO experience. It is possible that multiple locations of the same chain business could use the same hiring managers. To avoid allowing these managers to view similar applicants, we will use the same pair of profiles whenever applications are made to the same company within the same geography. This is equivalent to the same applicant applying to multiple nearby locations, which is common. Applications to locations in different geographies, in contrast, will use different profile pairs.

The randomized assignments of resume information and the combining of this information into a submittable resume will be done using a proprietary application we created that is similar to the Resume Randomizer created by Lahey and Beasley (2009).

As explained above, for each job we create two resumes (with randomized components), where one resume includes prior NGO experience and the other does not.

Yes

We will apply to 6,000 jobs.

We will apply with two resumes to each of the 6,000 jobs for a total of 12,000 resumes.

We will apply with two resumes to each of the 6,000 jobs for a total of 12,000 resumes.

As explained in greater detail in the pre-analysis plan, we have conducted power analysis and calibrated for a minimum detectable effect of 1 percentage point reduction in response rate caused by the inclusion of NGO experience. Based on prior literature, we anticipate the baseline callback rate in the non-NGO condition to be 6 percentage points. This means that a treatment effect of -1 percentage point would represent an approximately 17% reduction. Given this is a binary outcome variable, we do not discuss this effect size in terms of standard deviations.