What Name-Bias Studies Can—and Cannot—Tell Parents Choosing a Baby Name

The research is about decision-makers, not defective names

Parents sometimes encounter a frightening claim: a particular kind of name will reduce a child’s chances of getting a job. The claim is usually attached to a résumé study in which researchers send fictitious applications and vary the applicant names. Some of those studies find unequal employer responses. The inequality deserves attention.

The location of the problem matters. An employer who treats otherwise similar applicants differently is producing discrimination. The name did not cause a lack of ability, effort, or suitability. When advice jumps immediately from evidence of bias to “choose a safer name,” it transfers responsibility from the decision-maker to the family exposed to the bias.

This does not mean parents must ignore foreseeable friction. Families routinely consider spelling, pronunciation, forms, initials, and how a name will be read in several languages. The useful question is how much a study can actually tell them about one candidate, and which practical responses preserve the child’s identity instead of organizing it around other people’s prejudice.

Begin with a firm distinction: evidence that discrimination exists is not a ranking of names from safe to unsafe. A name listed as Traditional, Modern, or International can be interpreted differently across listeners. The categories cannot predict a hiring outcome.

What a résumé correspondence experiment measures

In a typical experiment, researchers create applications designed to be similar in relevant qualifications. They assign names intended to signal a social identity, send the applications to real vacancies, and compare employer contact. Random assignment helps isolate the effect of the signal bundle attached to the application.

The outcome is usually an invitation, callback, or other employer contact—not a final job offer, long-term earnings, school treatment, or a whole life. The applications are fictional and controlled. That design is powerful for detecting differences in the tested stage, but it does not follow individual bearers of the names through childhood and adulthood.

The names are research instruments chosen because participants or prior evidence associate them with groups. They are not a random sample of all names used within those groups. A finding about the instrument set cannot automatically be assigned to every culturally related name, every spelling variant, or a candidate that merely sounds similar.

Other résumé details also matter. Researchers may vary address, school, work history, age signals, or job type. A careful article will state which attributes were randomized and which were held constant. A headline that says only “names affect hiring” removes the design choices readers need to interpret the result.

The largest recent US study offers scale—and important variation

A large study published in the Quarterly Journal of Economics sent more than 83,000 fictitious applications to 108 major US employers. Applications with names chosen to signal Black applicants received employer contact 2.1 percentage points less often, on average, than otherwise similar applications with names chosen to signal white applicants.

The average is important, but it is not the entire finding. A public research summary reported that discrimination was concentrated among 23 of the 108 employers. That variation changes the practical interpretation. The study revealed different organizational behavior, not one fixed penalty automatically attached to every child carrying a particular name.

Scale improves the researchers’ ability to estimate patterns among tested employers and applications. It does not transform a group-level average into an individual forecast. A child born today will apply to different organizations under future conditions, with a real résumé, personal networks, chosen forms of address, and identities that no fictitious name can fully encode.

Parents can hold two conclusions together: the study provides serious evidence of systemic discrimination, and it cannot calculate their baby’s career outcome. Rejecting the second claim does not minimize the first. It protects the evidence from being stretched beyond its design.

A name signal can contain more than one assumption

Listeners may infer race, gender, religion, language, generation, social class, nationality, or family background from a name. Those inferences can be wrong. In an experiment, researchers select names that make a signal legible enough to test, but real-world listeners do not all read that signal identically.

A historically common name can sound generational. A spelling associated with one community in one city may be interpreted differently elsewhere. A surname can reinforce or complicate the first-name signal. The same full name may also be read differently on a school roster, a creative portfolio, a medical chart, and a résumé.

This is why parents should resist informal tables claiming that one Classic name earns more trust than one Unique name. Unless a study tested those exact forms in a relevant setting, the ranking is invented. Even when it did, the result belongs to that design, time, and outcome.

Spelling families add another complication. Sophia and Sofia are distinct legal spellings with different associations for some readers, even if many speakers hear a shared name sound. Combining them may help answer classroom familiarity, but it could destroy the signal a discrimination experiment intended to study.

Why parents cannot optimize away discrimination

A family might conclude that the safest choice is a name perceived as racially, culturally, and socially neutral. Such neutrality is unstable. It depends on who is listening, what other details are visible, and which identities society currently treats as unmarked. A surname, address, school, accent, photograph, or in-person interview can change the listener’s assumptions.

Optimization also creates a moral trap. Parents from marginalized groups are told to reduce signals of their own history while families from dominant groups are praised for choosing the same names as adventurous or cosmopolitan. The unequal reception is the bias. Erasing the signal does not make the underlying institution fair.

A child may later value a name that links them to language, relatives, religion, or community more than the adults anticipated. Replacing that connection with a name chosen primarily to appease hypothetical employers asks the child to pay an identity cost before any decision-maker has even appeared.

Names from Honorable, Spiritual, or Literary traditions can carry precise family stories. Those stories are not captured by a callback rate. Parents may still weigh practical burdens, but the equation must include what would be lost.

Practical usability is a different question from prejudice

Some concerns can be tested directly. Can close relatives pronounce the name after instruction? Does the full name create confusing initials? Will a letter be consistently misread under the local spelling system? Is the proposed nickname acceptable? Does the family have a plan for two writing systems?

These questions describe interactions rather than judging a culture or social group. A child with any name may correct spelling. The amount and type of correction can be estimated by testing real listeners, but parents should distinguish inability, unfamiliarity, and refusal. An accent is not disrespect; refusing to try often is.

Consider flexible use without treating one form as a disguise. A complete legal name may support an established nickname, a middle name used professionally, or a language-specific form used with relatives. Document each relationship accurately. Do not assign an unrelated “résumé name” as though the original belongs only in private.

A Short or Unisex name is not guaranteed to reduce bias. Short names can be culturally marked; gender ambiguity can trigger other assumptions. Choose these qualities because the family values them and the full name works, not because a style tag promises institutional protection.

How to read a dramatic name study responsibly

Start with the original paper or a careful research-center summary. Identify the country, cities, employers, occupations, dates, number of applications, name sets, randomized features, and measured outcome. If those details are missing, postpone applying the result to a personal choice.

Then translate the result in plain language. A 2.1-percentage-point average contact gap is not “2.1 percent less successful” unless the calculation actually uses that denominator. An average across employers does not mean every employer behaved that way. Statistical significance does not make a result universal across time and context.

Check whether the article reports limitations and heterogeneity. Did effects vary by employer, job, gender signal, or location? Was the experiment limited to one occupation? Were names selected from an older generation? The boundaries are part of the result, not disclaimers to be skipped after the headline.

Finally, ask whether the decision you face matches the outcome tested. Choosing between two spellings for pronunciation is not the same task as estimating employer contact. Selecting an Elegant middle is not comparable to a race-signal experiment. Evidence earns authority through fit.

A family discussion can focus on support instead of prediction

Parents who remain concerned can discuss what support they can control. They can teach the child the story and pronunciation of the name, correct records promptly, ensure schools use the requested form, keep accurate documents, and respond when an institution treats unfamiliarity as a reason for exclusion.

They can also avoid making the child hypervigilant. Repeatedly warning that a name may close doors can turn identity into a threat long before the child encounters one. Share information in age-appropriate ways and place responsibility clearly on unfair behavior.

Where possible, choose institutions with transparent practices and accountability. Structured hiring, consistent evaluation, documented pronunciation tools, and mechanisms for correcting records address more than one family. Individual nickname strategies may help in a moment; institutional change reduces the demand for those strategies.

Families can also decide in advance how they will respond when a teacher, coach, or administrator repeatedly records the name incorrectly. The first response can be ordinary and specific: provide the spelling, a pronunciation cue, and the correct record. Escalation belongs to a pattern of disregard, not every initial mistake. This protects the child without making every introduction an anticipated conflict.

Adults should be careful with informal experiments of their own. Sending two versions of a real résumé under different names can create record conflicts and does not reproduce a controlled study. Asking friends which candidate sounds most employable mostly measures those friends’ stereotypes. Neither exercise gives a baby a reliable risk score.

Do not make a culturally meaningful middle name serve as an emergency substitute unless the family genuinely likes that everyday use. A hidden alternative can send the message that one identity is presentable and another is private. Flexible naming works best when every form has a positive, documented role rather than a hierarchy of acceptability.

When relatives repeat a frightening headline, return to the study’s unit of analysis. The experiment observed employer behavior across applications. It did not observe the family’s candidate, the child’s future qualifications, or the institutions they will encounter. That sentence can end an argument without denying that discrimination requires action.

Keep the child’s future agency visible. They may use the full first name, a nickname, initials, or a middle in different settings. That flexibility is ordinary and can be empowering when it is chosen. It becomes burdensome when adults imply the authentic name must be hidden to deserve fair treatment.

Do not let a bias study make the final choice

Put research findings beside—not above—the family’s other evidence: connection, accurate meaning, pronunciation, surname fit, honor, language use, sibling context, and the child’s likely access to their own story. No single study measures all of those values.

If two names are otherwise equal and one creates a practical burden the family does not want, that can inform the choice. Say exactly what the burden is. “Our nearest relatives cannot yet distinguish these two sounds” is testable. “This name will hurt a career” is usually a prediction the evidence cannot support.

Choose a Timeless, modern, culturally specific, or uncommon form because it belongs in the complete family decision—not because an average callback result appears to certify it. The child will live with the name across relationships no experiment observed.

Name-bias research should enlarge parents’ understanding of discrimination, not shrink the child’s permitted identity. Read the studies closely, keep their boundaries intact, prepare for ordinary usability, and refuse to make a baby responsible for preventing an institution’s unfair behavior.