Systemic Discrimination Among Large U.S. Employers [pdf]

https://eml.berkeley.edu/~crwalters/papers/randres.pdf

by leephillips • 5 years ago

35 86 5 years ago

eml.berkeley.edu

Is 2.1 percent large or small? For example, is it 50.0 vs 52.1 percent (and thus marginal), or 1.0 vs 3.1 percent (and thus massive)?

chairmanwow1• 5 years ago

from the paper:

> "an effect equal to 9% of the Black mean contact rate"

hbrav• 5 years ago

In the main body of the paper they say "We find that distinctively Black names reduce the likelihood of employer contact relative to distinctively white names by 2.1 percentage points, an effect equal to 9% of the Black mean contact rate".

I think we take that to mean that it's an absolute 2.1% reduction, and if it's also 9% relative reduction, then that would have black applicants getting a response 23.3% of the time, and white applicants 25.4%.

rewq43212• 5 years ago

It's 2.1 percentage points which they say is about a 9% difference.

CheezeIt4• 5 years ago

It’s 25.1% vs. 23.0% response rates. Table 1.

yarky• 5 years ago

Statistically speaking, it could be a due to randomness, but that depends on the test statistics. I'd check the maths if I had any doubt about the conclusion ;)

phnofive2• 5 years ago

Abstract:

"We study the results of a massive nationwide correspondence experiment sending more than 83,000 fictitious applications with randomized characteristics to geographically dispersed jobs posted by 108 of the largest U.S. employers.

Distinctively Black names reduce the probability of employer contact by 2.1 percentage points relative to distinctively white names. The magnitude of this racial gap in contact rates differs substantially across firms, exhibiting a between-company standard deviation of 1.9 percentage points.

Despite an insignificant average gap in contact rates between male and female applicants, we find a between-company standard deviation in gender contact gaps of 2.7 percentage points, revealing that some firms favor male applicants while others favor women.

Company-specific racial contact gaps are temporally and spatially persistent, and negatively correlated with firm profitability, federal contractor status, and a measure of recruiting centralization. Discrimination exhibits little geographical dispersion, but two digit industry explains roughly half of the cross-firm variation in both racial and gender contact gaps. Contact gaps are highly concentrated in particular companies, with firms in the top quintile of racial discrimination responsible for nearly half of lost contacts to Black applicants in the experiment.

Controlling false discovery rates to the 5% level, 23 individual companies are found to discriminate against Black applicants. [See Appendix] Our findings establish that systemic illegal discrimination is concentrated among a select set of large employers, many of which can be identified with high confidence using large scale inference methods."

phnofive2• 5 years ago

Would have been nice to see the stats for the twenty-three clear offenders up top (p. 66) - still trying to figure it out now.

I'm open to being convinced that Census data for racially associated names is common knowledge among recruiters, but I'm not there yet.

nerdponx4• 5 years ago

I suspect that you could predict the "Blackness" of a name using character n-grams, with reasonable accuracy. Not to mention given+family name combos as mentioned in a sibling comment. You don't need to know the actual Census data for each name.

babesh• 5 years ago

I bet that you can figure it out based on the information given: Fortune 500, automotive, service, furniture.

option• 5 years ago

I lived in US for over 10 years working in tech. What is “distinctively white/black names”??

david38• 5 years ago

For black - [Terrell, Malik , Tyrone ] + [Johnson, Jackson, Washington]

For white - [Jack, Conner, Luke Scott] + [Anything Italian, Smith, James, Anything German, etc]

jldugger• 5 years ago

Check out the table B2 in the appendix if you need examples. Basically: German (Schmidt, Stolzfus) => white, French (Francois, Pierre) or English (Witherspoon, Smalls) => black.

Presumably this mapping comes from freedmen taking surnames from their former owners, and german settlers being relatively late to America, and chiefly in the northern states.

endisneigh17• 5 years ago

This isn't really surprising - this sort of thing has been shown for quite a while as the citations showed. I had a friend who happened to be Black and has a very "Black" sounding name tell me that he actually took away his photo on LinkedIn and changed his name and got way more recruiters contacting him (though I've also had similar things said by non-Blacks, so IDK).

I actually did not believe him since he's also in the tech industry and didn't think such things would happen for Software Engineers - alas the table at the end (Figure 9) shows a small difference.

I wonder why the food product industry has such a huge difference.

eplanit3• 5 years ago

Does capitalizing "black" compensate, somehow?

endisneigh17• 5 years ago

what do you mean?

flowerlad• 5 years ago

For college admissions it is different: it is Asians that are discriminated against, not African Americans https://www.latimes.com/local/california/la-me-adv-asian-rac...

And the way it is done is through "holistic review" process, see: https://www.theatlantic.com/education/archive/2013/12/the-fa...

Excerpt: From colleges’ perspective, “holistic” is just shorthand for, we make the decisions we make, and would rather not be asked to spell out each one. It’s a way for schools to discreetly take various sensitive factors—“overrepresented” minorities, or students whose families might donate a gym—into account.

endisneigh17• 5 years ago

how's college admission discrimination relevant to employer discrimination?

denkmoon• 5 years ago

Not many jobs going out there that don't require some kind of degree these days. University admission is a necessary step for job acquisition in anything but unskilled labor positions.

nerdponx4• 5 years ago

I don't think it's anything other than a somewhat-related and interesting situation. People downvoting are reading too far into it.

CheezeIt4• 5 years ago

It’s a good point of comparison, because it’s a much bigger deal. A 9% lower callback rate, concentrated in the most marginal candidates, won’t have as big an effect on your life as college acceptance.

rewq43212• 5 years ago

I'm curious whether it's the same type of discrimination that this study presents. I.e. is it "group A is undesirable" or is it "most applications are from group A, and so when we positively discriminate for diversity/equity reasons, group A is most affected in terms of absolute counts" your excerpt seems to suggest the former, but my previous assumption was that it was the latter.

nerdponx4• 5 years ago

It's pretty clearly the former, otherwise they could just call it affirmative action which is widely accepted and legal.

lugged2• 5 years ago

positively discriminate for diversity/equity reasons

Must be nice to be so privileged you can make up nice names for plain old racism.

You mean discrimiate against Asian and white people for flawed diversity and equity reasons.

Most of the black people who end up in these universities are privilleged children of immigrants.

Those black faces you see around you are from Nigeria, not Chicago.

daenz3• 5 years ago

It seems like they just did "black names" vs "white names." It would be interesting if they tried this with unique "white" names, like "Göran Hägglund" or "Alf Svensson" (picking Swedish names here). It would be interesting to know if the bias is against unique-sounding names or if it is against the inferred race.

Also, anyone know if there a reason why "Black" is capitalized in the paper but "white" is not?

dragonwriter5• 5 years ago

Also, anyone know if there a reason why "Black" is capitalized in the paper but "white" is not?

While its not my personal preferred style, the NY Times has recently adopted this style (from an earlier one which capitalized neither) and explains it reasonably well (and other explanations of users of the style have generally been similar, so its what I take as, in outline, the usual explanation):

https://www.nytimes.com/2020/07/05/insider/capitalized-black...

daenz3• 5 years ago

white doesn’t represent a shared culture and history in the way Black does, and also has long been capitalized by hate groups.

Thanks, and yikes, I had no idea this was a thing. So if I wrote "White" in the same way some might write "Black", people might get the wrong idea? To be honest, this change just seems like more petty ways to divide us.

eplanit3• 5 years ago

So, it's an institutional directive to its writers to treat subjects differently based on race: i.e. systemic racism.

Because, the best cure for perceived systemic racism is _actual_ systemic racism. It's all absurd.

nomdep2• 5 years ago

Because every country in Africa is exactly the same so they share the same culture? What a ridiculous and also incredible racist idea. NYT always the same.

zrail• 5 years ago

AP explains their style choices here. Not that this paper is written to AP style but the authors may share the same view. https://apnews.com/article/archive-race-and-ethnicity-910566...

jdavis703• 5 years ago

Swedes make up 1% of the US population, while Black’s make up 10%. Just statistically speaking, but a Swedish name is going to way more unique than a Black name.

And Black is capitalized for the reason Swedish is capitalized.

mc325• 5 years ago

No it’s not. Capitalization of black is political. It’s like deciding to capitalize “women”. I don’t capitalize black or white, nor caucasian, unless the person is from the Caucasus. I capitalize African and Asian because they are place name demonyms which get capitalized in English.

daenz3• 5 years ago

My point wasn't "use Swedish names", it was to use unique sounding white names, names that are as representative in the population as black names are in the population.

In other words, if a "white name" is as unique as a black name, does it receive the same amount of bias, or different?

dudul• 5 years ago

Is black a nationality?

eplanit3• 5 years ago

Did they try "redneck" names like "Elrod", "Rusty", or "Betty Jo"?

epicureanideal5• 5 years ago

Exactly. See my other comment in this thread. The names they chose are unusual at a glance, and it might be that the 2% variance they see is due to some groups of names unintentionally having higher perceived social status or other attributes.

For example, some of the first names are "Lashonda", "Lakeisha", "Takwanda" in the "Black" list, whereas the "white" list includes names like "Geoffrey", "Kristen", "Jennifer". If the perceived social status of the names isn't matched, that could invalidate their results.

As a follow-up, it would be interesting to do a survey on the perceived social status of each of the names they used to see how that might have influenced the results.

zuminator2• 5 years ago

In the social research world, it's fairly standard practice [0] but not universally accepted to capitalize one and not the other. With an opposing point of view, the National Association of Black Journalists believes that all races should be capitalized.[1]

[0] https://www.cjr.org/analysis/capital-b-black-styleguide.php

[1] https://www.nabj.org/page/styleguide

Aunche5• 5 years ago

This is tangentially related, but does anyone know how "diversity internships" are legal? For example:

https://buildyourfuture.withgoogle.com/programs/step/

The internship program has a focus of providing development opportunities to students from groups historically underrepresented in tech, through technical training and professional development.

This sounds like they blatantly violate the Civil Rights Act. I wonder how much discrimination is caused by people trying to "make up" for affirmative action (not that I think this is ok).

endisneigh17• 5 years ago

your example doesn't seem to be a great one - which group are you thinking is not represented in the photo?

Aunche5• 5 years ago

I was referring to the quote, but the photograph raises a good point. There is not a single white or Asian male, despite them making up the majority of CS classes.

epicureanideal5• 5 years ago

The names they chose seem to be fairly uncommon.

Did they correct for the confounding variable of perceived social class of each of the names? (Similar to what someone else said, I wonder if they checked... not for uniqueness, because some names are unique but familiar to other names, but rather for recency of immigration, which can actually have both positive and negative effects I would imagine. For example, someone with a distinctive first + last name combination for a particular place might seem more "interesting" to interview.)

Even a change in the presentation of the same name can indicate social class... Bill vs Will vs William, with middle initial or without, etc. Also hyphenated last names.

Also, they included so many other things in the resumes... club memberships, etc., it makes me wonder whether the results were influenced by that.

Even so, let's say the response is 23% vs 25% by race as they seem to say. That's obviously not good if it's due to racism for example, but it doesn't sound like an absolute prevention from working in an industry.

pseudalopex4• 5 years ago

They chose last names at least 10,000 people had in the 2010 census. Over 43.6% had a more uncommon last name.[1] And some of the study names were much more common.

Which first names seem uncommon to you?

Black names are perceived as lower class because they're associated with black people. And education and work history are stronger class signals.

Which names do you think sound like immigrants?

You looked at the names they used. So you know they didn't use different forms of the same name. Or hyphenated last names.

They calculated the effects of each characteristic. They found discrimination against men and women balanced out. They found modest but significant age discrimination. And they found club memberships had no significant effect.

Absolute prevention is a straw man.

[1] https://www2.census.gov/topics/genealogy/2010surnames/surnam...

Systemic Discrimination Among Large U.S. Employers [pdf]

Related Stories