The National Bureau of Economic Research (NBER) released this month the results of an experiment designed to measuring the degree of racial bias displayed by various large companies when choosing which job résumés justify further consideration for hiring.
Ninety-seven of the some of the largest companies in the country were sent made-up résumés by fictional job applicants, in nearly identical pairs with equivalent qualifications but bearing names that (the researchers presumed) suggested that the applicants were white or black, and male or female. Latisha and Amy was one pair; Lamar and Adam was another.
This week the NBER released the results, the researchers’ conclusions, and names of the companies. The study seemed to show that, on average, employers contacted the fake white applicants 9.5% more often than the fake black applicants, though this depended on the company. Those logos above represent the companies with the smallest racial gaps in hiring, based on the experiment’s results.
This was the largest such experiment yet, with researchers sending 80,000 résumés applying for 10,000 jobs between 2019 to 2021. The apparent racial bias seemed to spike in food stores, food products, freight and transport, and wholesale enterprises. The New York Times concludes, “The results demonstrate how entrenched employment discrimination is in parts of the U.S. labor market — and the extent to which Black workers start behind in certain industries.”
The Times also quotes Daiquiri Steele, an assistant professor at the University of Alabama School of Law who previously worked for the Department of Labor on employment discrimination as saying, “I am not in the least bit surprised. If you’re having trouble breaking in, the biggest issue is the ripple effect it has. It affects your wages and the economy of your community going forward. The results demonstrate how entrenched employment discrimination is in parts of the U.S. labor market — and the extent to which Black workers start behind in certain industries.” (Gee, what kind of name is “Daiquiri”?)
The jobs were all entry level, not requiring a college degree or substantial experience. The experiment tested for bias against other characteristics protected by law too, besides race and gender, such as age and sexual orientation. (How did the résumés hint at sexual orientation using names?)
A thousand fake applications were delivered to each of 97 companies, applying for up to 125 jobs per company nationwide. The researchers then tracked whether a company contacted the applicant within 30 days.
You can download the study here, and read it if you have more tolerance of Authentic Frontier Gibberish than I do. Sample:
We show that the tradeoff between these notions of information and reliability emerges naturally from a series of pairwise decisions in which an analyst guesses the ordering of parameters for each pair of units. When presented with multiple gambles of this form, the analyst faces an optimization problem subject to logical transitivity constraints
requiring all pairwise comparisons to be consistent with a coherent underlying ranking. A parameter λ trades off the gains of correctly ranking pairs against the costs of misordering them. When λ = 1, it is optimal to assign each unit a unique grade to maximize the expected rank correlation with the true performance levels.
Why do researchers write like that? There’s no excuse for it. My bias is to assume that whenever someone writes in jargon, they are trying to make something seem more impressive and authoritative than it is.
Other observations:
- I don’t doubt the existence of biases that may disadvantage minorities in seeking employment. I do question the assumption that it’s necessarily based in racism. Just as we are constantly hearing from minorities who say they want to be represented by, served by and see in movies “people who look like them,” that bias exists in majority groups as well. It’s a positive bias, not necessarily a negative one.
- Bill James once observed that all huge studies are polluted, often fatally, by the objectives of the researchers in starting the study. I would like to know the pre-study positions and ideologies of the researchers, for I am relatively certain that they didn’t undertake this study to prove Daiquiri wrong. James wrote that nobody wants to spend years on an expensive and time-consuming study only to find that it proves nothing, so researchers tend to make sure, in ways both subconscious and not, that their work isn’t for naught.
- Name bias is a well-established phenomenon: heck, I wrote about it in my college thesis on leadership. I see no indication in the study (though I might have missed it after the horrible jargon made me cross-eyed) that the researchers used unusual names that did not have racial connotations to see if those applications also were discriminated against, like, oh, I don’t know, names like Millard, Grover and Rutherford. If those names were also hired about 10% less often, there goes the narrative!
- “Lamar” is a black name? The only Lamar who immediately comes to my mind is Lamar Alexander, and he is white. So was Lamar Hunt. And I was serious about the “gay” names. I just looked for a list of alleged gay names, and I don’t know what the hell it’s talking about. (#15 is “Jack.”) The list of alleged lesbian names is similarly bizarre. Barbara? Heather” Carol? Jane?
- If, and I missed this if it was mentioned, any of the résumés included the fake applicant’s “preferred pronouns,” dinging them doesn’t indicate any bias except a bias against narcissists and likely office activists inclined to play power games. I share that bias. In fact, I encourage it. I would be very tempted, if presented with applicants with equal credentials but one of them wanted me to call him or her “they,”to choose the other one.
- The experiment itself was unethical: dishonest, a waste of time for HR departments, and proof of an “ends justify the means” orientation by the researchers. Why should we trust them?
Ethics Alarms 
Kids with non-traditional names or bonkers spellings of traditional names have an increased likelihood of having bonkers and non conventional parents, meaning those kids have an increased likelihood of being bonkers or non conventional-
Traits most risk-averse organizations are inclined to select against.
So?
And it also depends what you’re looking for. An unusual name can set an individual apart from the herd, and help push him into leadership roles. Four of our Presidents named Hiram, Stephen, Thomas and David decided to go with their more unusual middle names, and it worked! Lyndon Johnson was the first person ever named Lyndon with a Y: LBJ’s mother said it would look better on a ballot that way.
1. No, Grant dumped Hiram. He was embarrassed that his initials were “HUG” (this is a man who never let his wife see him naked) so he switched his name around so they would be the only slightly better UHG.
2. LBJ’s mother went on record to say she named he son to be a candidate for office.
3. Not sure how Grover’s origins have any relevance to his choice to make “Stephen” his middle name.
4.”Dwight” has never been an especially common male name. Usually a POTUS will elevate the popularity of a name, especially one as popular as Ike. And it certainly isn’t close to “David”—which was the point, I think.
I’d like to reiterate my points 3, 4, 9 and 10.
When it came time for a name to matter during a presidential election, no one voted against Grant because of “Ulysses”. No one voted against LBJ because of “Lyndon”, but because of his politics. No one voted against Ike because of “Dwight”. No one voted against Cleveland because of “Grover”.
We’ll never know, but it’d be interesting to see, when those men were starting out as new men, if they faced obstacles as compared to men equally situated to them in life, if their names were one of those obstacles.
My guess is the answer is no, for all the mitigating reasons listed.
That is to say, if a resume with the name “Lashawndra” or something got randomly trashed as compared to an “Emily” of ‘equal qualifications’… I’m not sure I’d be surprised.
But if the resume of “Emily” got trashed compared to a “Lashawndra” who happened to be the daughter of a local successful businessman who made the social rounds with other successful people…
I’d be even less surprised.
But that’s just it about the “resume name” surveys. They’re meaningless – no two people are the same on paper. It’s just not true. The real question to ask is *how much* does a name affect resume reviews? And if the number is less than 10%, then to me, that’s not so much an overriding consideration that actual, substantive experience and skills wouldn’t override.
Oh and determination – I would not be surprised that a “Lashawndra” that follows up assertively and confidently hits more interviews than does an “Emily” that’s ‘hoping her white name carries her through’ and does nothing else.
“Lamar” is a black name? The only Lamar who immediately comes to my mind is Lamar Alexander, and he is white. So was Lamar Hunt. And I was serious about the “gay” names.
I thought the same thing: Lamar Alexander, or, as his bumper sticker succinctly stated: Lamar!
I assumed Lamar would be more gay than black
Who knows?
However, as Homer Simpson sagely explained about gay men, “They ruined all our best names, like Bruce and Lance and Julian. Those were the toughest names we had!”
But, yeah, aren’t the testers choice of names themselves indicative of their own bias? And I don’t think they can get away with just giving first names. Names Amy Chen and Latisha Lopez present confusion for even the most accomplished bigots.
maybe the methodology puts some of those concerns to rest, if you can make heads or tails of it.
-Jut
Homer Simpson had other pearls of wisdom: ”Donuts. What can’t they do?”
jvb
Lamar Burton of Reading Rainbow and Star Trek must have skewed perception of the name.
I suggested the name Tyrone for one of my children because it’s a traditional Irish name. It was also the name of one of my white friends growing up, and a white character in one of my favorite films. Name was vetoed for several reasons, but a minor one was because it “sounded black”.
Tyrone Power! Sr. and Jr.!
You mean LeVar Burton, I think.
9.5 percent? Is that even significant? Isn’t that within some kind of margin of error? Any economists or statisticians out there? Anyone?
So does the 9.5% delta mean that “fake black applicants were contacted less often than fake white applicants about 1 time in 10?” So…about 9 times out of 10 that scenario didn’t happen?
Is that an accurate way to read those results (I am asking for a friend who got a D in college statistics)?
Thanks Joel. What I think I was trying to get at.
What I’ve seen argued before is that the “white” names used tend to be associated with the middle class, while “black” names are also associated with lower class, and that if the study was performed with lower class white sounding names (Brandi, Heaven, Buck, Eli) you’d see similar bias. People that seem middle class, like Amy or Adam, will be assumed to be more educated, have better manners, and be more presentable, while Brandi and Buck and Latasha and Lamar will be equally avoided.
That sounds extremely likely to me.
Interesting take. It could also be a experiential bias that creeps in. A young couple we know were wrestling with names for their first daughter. The husband would mention a name, and she would respond with, “I worked with two girls named ‘xxx’ and they were self-absorbed narcissists. I don’t want that one.” Does the elimination of that name make the woman racist?…a thousand times no. Her experience has created something of a bias against that name.
The same bias could happen at the HR desk. Maybe a person with that subject name was previously hired with that name and he turned out to be thief who stole from the company. I’m not saying eliminating someone from possible employment based on a name is a good idea, but it’s not immediately racist, either.
Ugh…that last paragraph has an extra “with that name” in it. My apologies. I think my thought is still understandable, but I should always read my work aloud before submitting.
That could be it, but that’s not likely to show up over a whole study unless lots of people are having bad experiences with people with a certain name. That’s probably going to mean a name associated with either cultural differences between ethnicities, class differences, or generational differences (as with the infamous “Karen.”)
lower class white sounding names (Brandi, Heaven, Buck, Eli) you’d see similar bias
Fundamental naming rules for parents who care:
Somewhere in the top 10 is never give your daughter a name that someone would associate with a strip club.
Good idea. Make up fake resumes. In each pair of women, one gets a ‘conventional’ name and one gets the ‘stripper’ name. Let the hilarity ensue. For the men, one gets a ‘conventional’ name and one gets a ‘porno’ name. Hung Lo and Hugh Jazz for the win.
Well, one classmate is a hiring partner at a firm here, and if he sees obvious “African” names or sees that someone was in the Black Students Forum or the Women Lawyers Forum or whatever, that application gets filed. In the circular file.
Racist!!!!!!!!!!
Didn’t the “Freakonomics” authors do a chapter on debilitating names?
Yes!
Now, let’s let AI take a crack at it. When Google tried to use an AI to do employee screening, it was found to be racist and sexist. I couldn’t find much about the racist part, but that was the sexist part. Now remember, the AI was given the employee files of Google employees to ‘learn’ what a successful employee is like. The AI downgraded any applicant that had ‘women’ in a title ( ‘Women’s chess team’, ‘Women’s Business Alliance’, etc). The answer is, ‘why’? Well, the test set is why. The AI figured out that people with ‘Women Group Title’ were less likely to succeed at Google than people without such things on their resume. The AI doesn’t care WHY it is so, it just noticed the connection.
Awhile back I saw an exasperated comment by a younger member of a large corporation’s hiring team who was handed a stack of several hundred resumes. The senior recruiter, his supervisor, told him to randomly divide the stacks into two equal piles – that is, based on no consideration of the content of the resumes. Once the task was complete the senior recruiter picked up one pile and threw it in the trash, commenting, “we can’t have unlucky people working for us”, then took the other pile to his office to begin reviewing in what would now be a much shorter task.
I don’t know if the anecdote is true or not.
That is awful. Funny. But awful.
jvb
I will counter with:
That is funny. Awful. But funny.
I am happy with either result.
jvb
If the level of education is part of the qualification matrix, names suggesting minority status would cause some to see AA considerations as to the actual learning that happened on the road to the degree. Examples are countless where lesser qualified, if even qualified at all, minorities are admitted to schools that their prior achievements in learning do not earn, and then those students are pushed through the system to “prove” that the AA student did not fail. This is the culmination of the “bias of low expectations” harming people. Hiring teams are there to find hires who will contribute to the growth and prosperity of the organization. They may not be willing to participate in the continuation of a social experiment.
Once upon a time, long, long ago, I didn’t even consider an ethnically Indian bloke for an IT job because he had a distinctive Indian accent.
At that time there was particular frustration here in Australia with overseas, particularly Indian, call centers. Frequently these call centers used technically incompetent, flow chart following, tech support people. Quite simply, having a distinctive Indian accent answer the phone would have introduced a concern, particularly for any new customer who didn’t already know my small, regional, business.
If other potential employers thought the same way, he may well have decided that he was dealing with prejudiced Aussies, when he was just dealing with people who already had enough trouble staying afloat! The problem was actually more India and corporate than anything else.
I might add, there was a hilarious comedy skit came out around the same time with a very ’Bogan’ customer calling a Telstra (telecom company) Indian call centre. Wet yourself funny, and would no doubt give many people the vapours today!
Yeah, well, I just got off the phone with a friendly Rooms-to-Go agent.
Here is the problem: My chair has a broker front leg. No amount of Elmer’s Wood Glue, staples, dowels, or other attachments will fix it. Trust me. I tried. So, I called Rooms-to-Go to see if I could get a new front leg/replacement leg. After being thrown around the “For English, oprima número 1 . . .” hell, I finally was able to speak to an agent. It took a few phone numbers to find our account but, lo, there it was. He asked, in perfect English,
“I see the chair. How may I assist you?
Me: ”The front leg is broken. I need a new one/replacement but I didn’t see it on the website.”
Him: ”Let me check your order status. . . . keyboard clackery . . . I see. It is model number 105423rw343, correct?”
Me: ”Yes. That’s the one.”
Him: ”Let’s see . . . more keyboard clackery . . . hmmm , , . Unfortunately, your warranty expired in March 2024. And, because it is our of warranty, we cannot service your chair.”
Me: ”Oh. I don’t need service; I just need the new/replacement leg. I can install it myself.”
Him: ”Again, our service technicians are not able to service your chair. You might want to check out locally what you can do.”
Me: ”What does that mean? It’s your chair. Nobody sells that exact part and it will look weird.”
Him: ”I understand but, unfortunately, I am unable to service the chair.”
Me: ”Well, if the chair were under warranty, what would happen?”
Him: ”Our service tech would come to your house and fix the chair.”
Me: ”Oh. So, if I had warranty coverage, your service tech would bring the leg to my house, unscrew the old one and install (screw in) the new one?”
Him: ”Yes.”
Me: ”Can we pretend that the warranty is still in effect, play like I am the service tech, order the leg – I will pay for it – and ship it to my home?”
Him: ”No, sir. We can’t do that. Your chair is out of warranty.”
Me: ”Don’t you guys have a parts department I can talk to to see if there is a replacement leg?”
Him: ”No, sir. All of our parts are serviced through the warranty department.”
Me: ”Well, that seems troublesome to me. Why can’t I buy the leg?
Him: (Becoming agitated and/or aggravated) “Because your chair is out of warranty.”
Me: ”Well, I seem to be in an out-of-warranty pickle, right?”
Him: ”Unfortunately, yes. Is there anything else I can do to assist you?
Me: (Dejected and sad) “No. You’ve been very helpful.”
jvb
Johnburger2013.
The death of competence.