Systemic Discrimination Among Large U.S. Employers [pdf] | Hacker News Reader

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

eml.berkeley.edu

35 pointsleephillips88 comments

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.

how's college admission discrimination relevant to employer discrimination?

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.

yeah, but I'm not sure how discrimination at extremely elite schools is relevant to broad sector industry discrimination.

don't get me wrong, discrimination is bad, but let's not act as if discrimination in getting into Harvard is in any way comparable to not being able to get a callback at McDonalds because you're black (the food industry had the highest white/black callback discrimination according to Figure 9 in the paper).

Supply of eligible candidates comes from universities. Any discrimination at that stage certainly has an impact on the pipeline.

i'm not following you.

You don't want to.

sure I do - that's why I said I didn't follow.

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

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.

I'd agree if the alternative is not going to college at all.

White Americans attend college at a lower rate than African-Americans when they have the same IQ. I suspect that’s good for whites and bad for blacks, for the average college student.

> White Americans attend college at a lower rate than African-Americans when they have the same IQ. I suspect that’s good for whites and bad for blacks, for the average college student.

That's an interesting claim - you gotta source for that?

I’m not really a source memorizer, sorry. I think at 117 IQ it was 50% and 67% respectively (in some past year). It’s probably tough to google too. But you could look at overall post-secondary attendance rates and compare against 12th grade literacy proficiency rates to see the same picture.

I'm not sure if that exact statistic is listed in his recent book, but a very similar one is... IQ by occupation by race.

Charles Murray's "Facing Reality". It includes many tables of statistics that may interest you.

Note that despite a lot of anger toward him, I've read several of his books, and he goes to great lengths to state that he doesn't believe we should discriminate against people based on their race. I highly recommend reading his books rather than listening to commentary about them.

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.

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

> 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.

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?

> 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...

>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.

Some on the progressive left have actually started capitalizing "white" as well. Traditionally though that was a sign of Klan sympathy. Personally, I will continue to resist capitalization of American racial categories.

People want to eat their cake and also have it. On the one hand “we’re individuals, don’t look at us as a stereotype” (agreed), on the other hand, we share the same culture and history like no one else does, we’re one identity! (This is categorically wrong given the diversity in Africa).

> on the other hand, we share the same culture and history like no one else does

There is no “like no one else does”; the argument is that Black identity (particularly in America) is essentially an ethnic identity produced by shared experience, the same as other ethnic identities. The distinctive part is that the shared experience centrally includes American White supremacy which, among other things, actively erased the preexisting ethnic identities of the black people whom it made Black.

I think one could argue much of the Jewish diaspora also have a "shared" experience in a similar way but I don't see the same tension there, though maybe I'm just not seeing it. There are also African immigrants who would have a different experience, even people from the Caribbean have a totally different experience.

If you look at it from a people perspective, it doesn't make sense because people want to be their own individual, not lumped into a category because everyone experiences things differently. I see this as an artificial construction that is usually argued against.

> I think one could argue much of the Jewish diaspora also have a "shared" experience in a similar way

There are certainly similarities, but then, no one denies that “Jewish” is a national/ethnic/cultural identifier that should be capitalized, even if racial or physically descriptive identifiers generally shouldn't and even if “Jewish” might also be considered a racial identifier.

There are arguably even closer similiarities with the various peoples with roots in the pre-Columbian population of the Americas. But, again, whatever label is used for various collections of those people, one rarely sees arguments against capitalizing the label chosen.

Most black Americans aren't immigrants. And even fewer were when the culture we're talking about started forming.

Ethnic identities are not "produced by shared culture and experiences", they're essentially arbitrary. The modern "White American" ethnic identity was pretty much patched together by early 20th-century Progressives (certainly diminishing the "preexisting ethnic identities" from Europe, though of course not "actively erasing" the bulk of them as the slave trade did to Black identities!), but it would be clearly wrong to suggest that many Americans don't identify as a result with being "white", at least in a subculture-related sense that one might very well call a source of "shared experience" and continuing ethnogenesis.

> Ethnic identities are not "produced by shared culture and experiences", they're essentially arbitrary

Identities are arbitrary, but they aren't ethnic unless they are rooted in an ethnicity, which is a shared culture.

I think the "shared history" part has as much to do with slavery and racism in America (and probably also Britain and Western Europe), which is/was definitely a shared common experience.

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.

I'm not really seeing how that in itself is racist. Could you clarify your point? E.g. I happen to be black and I also happen to usually not capitalize either black or white. Not that I care either way. I do however, generally capitalize "Asian" because it's the adjectival form of a place name, like French, African, American, etc. So am I being racist against Asian people or being racist in their favor, in your estimation?

It is better addressed in other posts, but fundamentally it is about affirming the concept of 'black' as a cultural identification while denying 'white' as one. Some people who identify as a cultural amalgam of generally white European ancestry are bothered that this is not viewed as a legitimate in comparison.

Why can't we treat people both as individuals, as groups, and acknowledge that groups are just generalizations that do not hold for the individual? We can't really deny either.

What is sad is when people hide in the shelter of group identity, instead of being responsible for their own speech and actions. It's often not in anyone's best interests, and distracts from clarity.

Most white Americans identify as their particular amalgam of European ancestry in my experience. Most black Americans can't do that. Slavery erased their particular African ancestry.

I mostly agree. I don't think that means "black" is a valid cultural identity and "white" is not.

Slavery, oppression, and loss are not the only ways in which a shared culture can be formed.

No one said they were. Most white Americans just don't consider themselves culturally White.

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.

That's not the argument. They claim it refers specifically to the African diaspora in the U.S. The presumption being that they share the legacy of slavery and discriminatory laws afterwards. Which many do.

Then many “whites” share the legacy of slave owning and the same laws afterwards. Except for everyone that doesn’t. It is racist, no matter how much you sugarcoat it

> The presumption being that they share the legacy of slavery and discriminatory laws afterwards.

No, that they share an identity forged in the experience of slavery and the discriminatory laws afterward, and in many cases that's all the ethnic identity they have because their ancestors preexisting cultural heritage and identity was forcefully prevented from being preserved.

There are rubbles in Middle East where they cling to the past too. It's not a living identity.

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...

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.

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.

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?

Is black a nationality?

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

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.

I don't see how what you're proposing is even possible. The names themselves are tied with the race due to the incidence of people of certain races being named certain things. If you accept racism as a social construct then it would necessarily impact social status, thus making the entire exercise to "control" for it moot.

I think what they are implying is that of there were socially distinct class markers for black names as there are for white names, and if you looked at both, there would be a spread between upper class blacks and whites and lower class blacks and whites. That may be neither here nor there with regard to this discussion.

I understand that.

What I'm saying is that even if that were true, unless you're saying only "objectively" Black and White people are to judge the social status of Black and White names (such that each race judges their own name) - it would end up being no different than the exercise in the paper.

For example:

Let's say Denise is an "high status" Black name and LeShawn is a "low status" Black name. "Billy" is low status White and "Alexander" is high status.

Even if you knew this you'd need to know how the other race perceives the name since that's ultimately what's being judged here, hence it would be moot.

> Even if you knew this you'd need to know how the other race perceives the name since that's ultimately what's being judged here, hence it would be moot.

And that could be done through research surveys to account for that variable in this research.

I'm still not understanding what magical survey could separate the class aspect of a name from the race aspect. Since Black people can discriminate against other Black people and Black names both racism and classism can be "double counted", thus the exercise is pointless as there's no way to separate the two things.

Perhaps if Black people and White people had a "pure" view of other Blacks and Whites, unaffected by history you could do what you're saying.

Well, if we're trying to split racism from classism, you could try to create names that are equally perceived as being from a particular race, but differently perceived in social class. For example "Lashonda Brown" vs "Shonda Brown" vs "Shawna Brown" vs "Shawna E. Brown". Another confounding issue is that names perceived as "higher status" are not necessarily used by one particular race. For example, "Charlotte", "Elizabeth", "Caroline", are not obviously white or black.

Another example, to split racism from classism but with the confounding variable of national origin: "Lashonda Brown" vs "Abena Nwachukwu" (this is an actual last name that I Googled). On the resume, both could be given membership in a "Women's Society of Some University" to indicate gender, and same for other characteristics.

Edit: Note that even the SPELLING of "Geoffrey" vs "Jeffrey" vs "Jeff" may indicate something about social class or background.

Random source, I didn't keep all my tabs open: https://www.thedailybeast.com/the-elites-top-50-baby-names

I mentioned this in the sibling comment, but it still doesn't seem possible. Who's determining these social class tiers? Black people? White people? Either way, there's no way to control for the status as the names themselves are already indicators irrespective of the social status.

In other words, you're trying to separate racism from classism, but race and class are already inexorably linked to begin with. Any perception of class to the name would already be influenced by historical racism, for better or for worse.

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

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."

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.

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.

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

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

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

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

I’ve known both white and black Tyrones, so that one is less clear cut.

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.

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.

Does capitalizing "black" compensate, somehow?

what do you mean?

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)?

from the paper:

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

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%.

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

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

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 ;)

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.

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...

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).

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

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.

The image is cropped, open it up in a new tab and you'll see the whole gamut.

Also, I'm not sure why you think focusing on "historically underrepresented" groups would be illegal lol.

> Also, I'm not sure why you think focusing on "historically underrepresented" groups would be illegal lol.

Because by "underrepresented groups" they mean race and sex. You'll see far more children of African immigrants than people who graduated from inner city school districts.

do you have a source for this claim? also, certain white and asian people can also be underrepresented.

what exactly is your qualm?

> do you have a source for this claim?

Yes from their own job application.

> Student Training in Engineering Program (STEP) is a development program committed to addressing issues surrounding diversity in our company and in the technology industry, and as such students who are a member of a group that is historically underrepresented in the technology industry will receive priority in the selection process. This group includes women, ethnic minorities and students with disabilities

> what exactly is your qualm?

That this violates the Civil Rights Act?

your quote does not prove that African immigrant children are getting more than inner city Blacks, nor does it violate the Civil Rights act. "Students with disabilities" can be of any race, thus, the program could hypothetically include any person of any background, and so there's no issue.

Not to mention that their list was not exhaustive. There are other groups that are unrepresented other than the ones they mentioned.

Responding to your last comment, which I can't reply to yet.

I was proving my first claim about what they mean by "underrepresented groups." They literally said that they'll give priority to underrepresented minorities or sexes. That is a clear case of discrimination, which is outlawed by the Civil Rights Act.

No, they said they'd give underrepresented groups priority, which include minorities. Favoring "underrepresented groups" does not inherently discriminate against any particular race or sex, thus there is no issue.

But hey, maybe I'm wrong. You should consult a lawyer, you might have a huge suit on your hands.

Reply on news.ycombinator.com