Two Identical Resumes, One Name Changes

A candidate named Greg sends out an average of ten applications before getting a first callback. A candidate with a strictly identical profile, but named Jamal, has to send nearly fifteen. This isn't an anecdote — it's the result of a large-scale field experiment that remains one of the most cited studies in labor economics.

A Protocol of Unusual Rigor

Marianne Bertrand and Sendhil Mullainathan published this finding in 2004 in the American Economic Review, based on a protocol of unusual rigor: nearly 5,000 fictitious resumes, of rigorously comparable quality, sent in response to job listings published in the Boston and Chicago press, varying only a single parameter — the racial connotation of the candidate's first name (Bertrand & Mullainathan, 2004, American Economic Review, 94(4), 991-1013).

The measured callback gap, around 50% in favor of White-sounding names, remained statistically stable regardless of industry, job type, or employer size.

Why This Study Holds a Lasting Place in the Literature

It's precisely this methodological rigor — varying only one variable, on otherwise identical resumes — that isolates an effect classic observational data can't demonstrate as cleanly. This is also what explains the study's lasting place in the literature on labor market discrimination, still cited in recent methodological work on evaluating the fairness of selection processes.

What This Means for a Resume-Screening Tool

The lesson for designing a resume-screening tool is direct: consistency in the criteria applied from one file to the next isn't a nice-to-have — it's a condition for a fair process. But consistency alone isn't enough to guarantee the absence of bias — a poorly designed system can just as easily reproduce, or even amplify, biases already present in its training data.

Clarentia's Position

That's why traceability of the criteria used, and genuine human oversight of every decision, remain a fundamental requirement in this domain, not a nice-to-have — it's the principle we aim to apply at Clarentia at every screening stage, without claiming that any system alone can guarantee total neutrality.

Cited Source

Bertrand, M., & Mullainathan, S. (2004). Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. American Economic Review, 94(4), 991-1013.