Discrimination starts even before grad school, study finds | Hacker News Reader

Discrimination starts even before grad school, study finds

blogs.nature.com

42 pointsnkrumm56 comments

For those interested in what the text of the email was, from the study [1]:

-----

Subject Line: Prospective Doctoral Student (On Campus Today/[Next Monday])

Dear Professor [Surname of Professor Inserted Here],

I am writing you because I am a prospective doctoral student with considerable interest in your research. My plan is to apply to doctoral programs this coming fall, and I am eager to learn as much as I can about research opportunities in the meantime.

I will be on campus today/[next Monday], and although I know it is short notice, I was wondering if you might have 10 minutes when you would be willing to meet with me to briefly talk about your work and any possible opportunities for me to get involved in your research. Any time that would be convenient for you would be fine with me, as meeting with you is my first priority during this campus visit.

Thank you in advance for your consideration.

Sincerely, [Student’s Full Name Inserted Here]

[1] http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2063742

This email lacks any personalisation and looks exactly like spam to me. I am wondering if the study should have been entitled "Generic emails containing minority names are classified as spam at higher frequencies."

That problem is difficult to solve for a controlled experiment; as soon as you personalize the messages then two outreaches to two professors become incomparable to each other.

This is true, but as a former professor if I got an email from a possible candidate requesting to meet with me that did not show that they had at least bother to investigate my research area I would tend to ignore them. If you don't personalise your email mails then they really do look like spam.

The way to solve this study would be to do a delayed crossover. I would send the same personalised email (just changing the name), but send them 3 months apart. No professor is going to notice the two emails were identical after this length of time.

I disagree. You could write personalized letters to each of thousands of professors, as long as each contained a slot for the student's name. If you then randomly assign the different student names to the letters, you still have a perfectly valid controlled experiment. No need to compare two outreaches to each other.

Being said that, the study as it is is already very solid proof of discrimination. So what that the email looks spammy? If it looks just as spammy from each and every group, and not every group is equally rejected, that's textbook discrimination.

Those personalized letters won't be comparable unless they are so generic that they're still spam. After randomizing which name goes to which letter, you're stuck with an infinite number of dimensions of variation, where one dimension is the ethnicity of the name signed at the bottom of the letter, and the other infinity represent the "quality" of the personalized letters. You won't be able to conclude anything about the influence of the name.

I still disagree. If there are only two letters, than, sure, the relative quality of them will dwarf the influence of the name.

But if there are thousands of letters, assigned randomly, the Law of Large Numbers does its magic.

No name gets assigned significantly lower or significantly higher quality letters than the average. (something like that could happen, but the odds are small enough to be ignored).

http://en.wikipedia.org/wiki/Law_of_large_numbers

Very interesting hypothesis. If it's true, I'm not sure if it makes it better or worse. Are we unintentionally programming our computers to automatically filter out messages from minorities? Talk about dystopia...

What kind of spam says 'I want to meet you in real life today/Monday' ? I would guess dating sites, but there's no URL whatsoever in the email and those professors have a lot of experience with a lot of real emails that follow the format.

>What kind of spam says 'I want to meet you in real life today/Monday'?

Student spam :) Actually I used to get around a dozen of these emails a week from potential "students" wanting to study with me. What these people seem to be doing is sending the same email to every professor in the department (could be the whole university for all I know). They seem to be scraping the emails either from the department websites or getting them out of the corresponding author detail of my published papers. If I was not too busy I would send back a reply asking why they wanted to work with me. I have to say that I go very few coherent replies.

Advice. If you actually want to study under a particular professor spend the time to actually learn about their research. Ask them intelligent questions that show that you have read their papers. If you show that you are bright and keen and have decided that this is the area you want to work in then the professor will be much, much more willing to meet with you.

In the main plot in the paper they plot the absolute differences in response rates as opposed to plotting relative differences. 90% vs 80% for a given department seems drastically different from 5% vs 15%, but those would get equal value bars. I also tend to have a knee-jerk negative reaction to the phrase "reverse-discrimination" as it evokes the "you can't be sexist against men!" etc. type of ideas. This sort of feeling is enhanced by the splitting of the data set into white males and non-white/non-males for significant parts of the analysis.

It's interesting that they didn't find that percentage minorities or women on faculty had significant impact on response rates (with the exception of small results for asian-asian support). That seems to suggest that any discriminatory tendency is fairly uniform across individuals, reinforcing that policy changes to reduced discrimination must be collective effort.

Otherwise this seems like a good experiment and a good paper. The email thing just seems really clean and a realistic depicter of reality. Let's see if there are any significant changes after peer review, I suppose.

> Most would acknowledge that women and minorities already face more hurdles in academia than their white, male peers. A lack of mentors, occasionally overt discrimination and the academy’s poor work-life balance, are well-documented issues.

OK, I'm stumped. I can't figure out how poor work-life balance is more of a hurdle for women and minorities than for white males. Anyone know?

That part wasn't intended for minorities. As for women, they need a lot of time off for pregnancies and typically want/need to spend more time with kids. Also, they have a biological clock for fertility unlike men, so it's more important for them to have time to socialize to find a partner unlike men who can procreate well into their nineties.

"poor work-life balance" is a greater issue for women than men due to childbirth and general fact that childcare and elder care burdens generally fall primarily on the woman in a couple.

"lack of mentors" and "occasionally over discrimination" affect both women and minorities more than white males.

I suspect elder / disabled care may be an issue for minorities too, of both sexes, if they come from economically disadvantaged backgrounds.

Women tend to choose professions with better work-life balance more often than men. Not sure this holds for minorities (the stereotype is actually that immigrants work extra hard, but I have no specific data).

The author of the article likely believes that women and minorities are united by facing oppression or discrimination, and consequently groups them together as a single category. This cavalier approach to analysis also enables them to appropriate the oppression of certain groups for the others.

When I was a physics undergrad all 12 physics majors in my graduating class were given a "talk" by a senior white faculty member about the "realities" of the physics pyramid. Basically we were told that our Chinese competitors made much better grad students and we should basically not bother applying to grad school because we couldn't compete against the "better" Chinese.

Really?!

In grade school at least one of my teachers made a comment along the lines of this. For a little backstory, my town was roughly 30% white, 25% asian, 25% black, the rest spanish or other races.

Shoot I'll go one further. When I was in 5th grade I offered to help get their huge donation of 486s ready for use (installing windows on them and the like). Most of the teachers didn't know how to use computers at the time, and they were happy with my work so they tasked me to fix/maintain all of them after class. That was pretty cool! I picked a filipino classmate to help out, and one day at honor roll ceremony the principal accidentally introduced him as the computer guy and me as the assistant -_-.

A physics teacher pulled that stunt once too at my community college. He told us the only people that would get As in his class were "real Asians. Not white guys that think they're Asian." He regularly made borderline racist and sexist comments though.

The reality is (at least in my experience) that no one race or gender is better at science subjects, engineering, or math. Most people in those subjects work their butts off to get where they are. The part I dislike is when someone stumbles, there's a lot of pressure on them to quit when there should be encouragement to try again. Sometimes things can be like pounding out a sword on an anvil, it doesn't happen instantly.

Instead it looks like there's always encouragement to kick people out and discourage people from studying a subject.

I worked at a dept with ~2.5% acceptance rate.

There were two, subtle factors in play: soft ageism and PI's hiring their clones (in thought, gender and race).

So the dept gravitated to predominantly two minorities, at least in terms of staff, faculty, visiting researchers and grad students.

EDIT: With regard to what I pointed out in my original comment (below for reference), I believe the Nature article chart is in error. Reading the actual paper (p.55), the gap in "engineering and computer sciences" (while smaller than some other fields) is actually more statistically significant than some of the other results. I suppose that may be because they were able to trial more professors.

I guess that means, be careful trusting charts, even from Nature's blog. The smaller gap may still count for something compared to a few worse fields, but it's not among the statistically-unclear results of the fields studied.

--original below--

From the graph, "engineering and computer sciences" was one of the smallest measured gaps, and further didn't feature the "*" which indicated a statistically-significant result.

So in all the often-justified criticism, the fact that CS/engineering are better than many other fields-of-practice should count for something.

(Interestingly also per the chart, in 'Fine Arts' the faculty discrimination ran in favor of women/minorities, to a statistically-significant level.)

Why does it matter if one result is more or less statistically significant than another, assuming that they are both sufficiently significant?!

(Note: I don't think that 5% or 1% chance of error is sufficiently statistically significant, but I'm a mathematician/programmer, and I know that most social studies researchers disagree.)

Peculiar why women were bunched together with minorities. I wonder how the results would look if they were decoupled. I also wonder what their reasoning was in attributing the small effect linking different names to differing response rates to 'implicit bias' was.

Maintain unity within the political coalition.

The study does break it down further, and the result actually made feel ill.

http://papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID2427725_code...

Go to Page 55+

Things aren't so bad if you're white female. If you're hispanic female, you even enjoy advantages in field of science.

But the numbers for Chinese and Indian (both genders) look absolutely atrocious. I'm seeing on average 30 percentage point difference, but as much as 60%+ in some fields.

This is just disturbing.

Have you had to deal with the number of Indian or Chinese students trying to push into a program they have no place being in?

I get it's the poverty back home that makes them desperate and nothing to do with race, but the one time my former department accepted a Chinese student from mainland China it was a disaster, he was rude, hardly spoke English, his credentials were most likely fake and made the whole thing incredibly unpleasant to everyone with how pushy he was. Needless to say he was kicked out after a year because he kept failing his classes.

This isn't racism, this is people not wanting to be scammed.

Isn't biased stereotyping exactly what racism is though? A few bad apples can spoil it for everyone else? "Hey watch the black guy, I was mugged by once...", if you find that sickening, this is quite similar to that.

Only if it's based on race. You must be American since you missed the part where I said it's the _poor_ Indians and Chinese that are desperate and most departments want nothing to do with them. That isn't racism, that is wealth discrimination. By contrast the poor Australians aren't applying to post graduate programs in mathematics, they are too busy being drunk at the footy, so they are not a group anyone had to worry about at my university. But if one of them showed up with fake credentials they would be as disruptive to the department as that Chinese student was, and would fly out just as quickly.

I doubt this Chinese student was poor if he could get convincing enough fake qualifications; he was probably middle class with city hukou and resourceful parents, but just didn't apply himself because his parents were always able to solve his problems for him (also, very likely an only child). This is actually quite common in China; the poor Chinese students are the ones that have to work their asses off because their parents can't give them any special advantages.

I have my own biases, but having lived in China for so long, they aren't based on ethnicity anymore (e.g. watch that guy in starbucks wearing farmer clothes, he might try to steal my computer).

He might have been rich by Chinese standards but he was still poor by Western ones. He seemed to be vastly more interested in finding a job outside the university so he could stay in the country than actually trying to study for the classes he was taking.

And again note, I didn't say he was lazy, as far as we could tell he was taking a dozen interviews a week for work. He was however a complete academic deadbeat and terrible investment for the university.

In short, we got scammed, someone came over on a student visa, only to try their hardest to migrate permanently without adding anything of value to the university and group he was meant to be working in.

> He might have been rich by Chinese standards but he was still poor by Western ones.

Ugh. It is possible that his parents were richer than yours; they could be driving black Audi's, have multiple Ayi's (house keepers), and multiple apartments; you really can't tell! It is true that having their kids get a job in a western country is important even to middle class/rich Chinese; even they realize how unbalanced the employment situation is even if it often works in their favor.

> And again note, I didn't say he was lazy, as far as we could tell he was taking a dozen interviews a week for work.

This makes sense. They can lazy academically but, instilled by the habits of their parents, realize that the key to success is networking and career climbing rather than studying.

Yes it is racism. I am not sure what to do though if you are on the receiving end of a lot of spam-like emails from people from certain countries. When I get emails from single women purportedly in eastern Europe wanting to meet me these all head straight to the trash. I would not really consider myself racist for doing this.

On the emails, I agree that there is no choice to filter just based on volume.

However, as for generalizing based on direct personal experiences, this cross a line where the abstract sender of an email becomes a real person. That begins to be a real problem.

Yes. Individuals should always be treated as individuals and not a member of some heterogeneous group.

Racism is a lazy way of thinking - instead of bothering to make judgments based on the individual, a racist just decides that all individuals of group x are the same for some particular characteristic.

> instead of bothering to make judgments based on the individual, a racist just decides that all individuals of group x are the same for some particular characteristic.

I define a racist as someone comfortable making blanket denunciations of arbitrary groups, without it ever crossing his mind that, according to well-established logical rules, his group will eventually become the target.

Racism is not always arbitrary nor negative. A good example would be the casual racism implicit is statements like "Jews are smart" or "Chinese are hardworking". Both statements are highly racist and reflect lazy thinking - while I have met many smart jewish people and lots of hard working people who were born in China, I have also met a number dumber than a box of hammers and lazier than a tree full of sloths.

> Racism is not always arbitrary nor negative.

On the contrary. A deep understanding of these issues leads one to the conclusion that stereotyping, whether positive or negative, is harmful because it reduces the described group to a set of cliches, thereby preventing the group from having equal standing with other groups.

If a group is stereotyped with a positive trait (Jews are smart, Asians are good at math), that stereotype will work against them in open competition with other groups (don't give Abraham extra help in class, he doesn't need it, he's Jewish). Same with negative stereotypes.

Tl;dr: stereotyping, regardless of what kind it is, stands in the way of equality and fair dealing.

>stereotyping, regardless of what kind it is, stands in the way of equality and fair dealing.

I agree, but I don't think I ever said that racism was "good", just that it is not always negative or arbitrary. One thing it always is is lazy. Individuals should always be treated as an individual and not as a member of some "convenient" grouping that enables a person to avoid having to think.

> I agree, but I don't think I ever said that racism was "good", just that it is not always negative or arbitrary.

That doesn't change the meaning. My rejoinder says that racism, regardless of whether it elevates or degrades the subject group, always has negative effects.

I read the grand-parent as a note that "denunciation" is overly narrow because it can also be praise that is overly weighted, overly generalized, or misapplied entirely - a claim you seem to agree with. It's "not negative" in the way that "positive feedback" is "not negative" - which says nothing about the consequences.

"We accepted one Chinese student and he was terrible, therefore all Chinese students are terrible" is exactly racism.

Racism isn't just marching in Klan rallies and voting for David Duke and "negroes need not apply". In the modern western world, that's the weakest, most ineffectual racism, because it's easy to call out and despise. The real, insidious, devastating racism is thoughtful, well-meaning people coming up with carefully-considered justifications to stereotype huge groups of people.

It's superficially very rational--black people are statistically more likely more likely to commit crimes, so doesn't it make sense for the police to keep a closer eye on them? Terrorists tend to be Muslims, so isn't it safer if we don't let Muslims near anything dangerous? But a lot of local, short-term "rational" decisions can add up to a monstrous and oppressive society.

As a graduate student I receive spam from addresses with Indian names a few times each month. I know from speaking to my advisor that the level of spam he receives is much greater, and often contains names that sound foreign to him. Even worse, these emails are somewhat generic in ways that resemble the form letter in this study. Normally, when a prospective grad student is interested in working with you, at least in the field I'm in (psychology), they mention something about your interests when contacting, rather then a generic form-looking email.

I'm not saying that this makes not responding to emails from minority students okay. I wonder to what degree professors might see something that is not personalized, from a name which is "similar" to other names they associate with spam, and decide not to respond.

Andy Gelman has a good example of the kind of spam that comes in: http://andrewgelman.com/2014/04/05/bizarre-academic-spam/

The fact that these names are considered "similar" to such names ("Juan Gonzales, but "Claire Smith" too?) are shady but "Brad Anderson" and "Steven Smith" are worth responding too basically demonstrates the authors' point.

Edit: Before I'm further downflagged by people who haven't at least skimmed the paper, I should point out that I'm referring to the actual names used in the study, I'm not being racist myself.

tldr :)

Any indication for the reasoning why for this difference?

Speaking as someone who deals with this often, it might be that the number of requests from Chinese/Indian students is simply overwhelming compared to everyone else, and the professors start putting up subconscious filters. They are also less likely to distinguish gender by Chinese and to a lesser extent Indian names, so we could probably discount gender bias at that point.

As for myself, 100% of the students I get requests from are Chinese (being in China, that makes sense), and I have to filter according to the university they are enrolled in (Qinghua > Beijing Normal University). We really miss some gems that way, but those are the breaks.

"They are also less likely to distinguish gender by Chinese and to a lesser extent Indian names, so we could probably discount gender bias at that point."

The study made sure that Indian and Chinese names used had 95-100% successful gender ID rate. See Page 55-60.

How the heck did they do that without assigning english first names?

Thanks for telling where the data was, as it was indeed disturbing but for the complete opposite case. The abstract is just bad science, and it was disturbing to see how much the truth was twisted.

Lumping together women and minorities is bad math based on the data they provide. white female has a 1% difference compared to white male, which is within the margin of error given the size of the study. Minority Male compared to minority female seems to show the opposite data, in that minority male is more discriminated against than minority female.

So the research found that A), no discrimination against female students, B) male minorities get more discrimination than female minorities, and C) minorities get discriminated compared to caucasians.

Compare those finding with the title, and even the abstract.

How many American women receive selective service notification?

https://www.sss.gov/

some of the differences aren't statistically significant?

Same thing I'm wondering. Would've been trivial to add error bars to this figure.

No shit?

What's the relative proportion of men in prison to women?

If there are more men then women in prison, it can only be because men commit more crime, right?

Women get laughably light prison sentences compared to men.

Here we go...

Reply on news.ycombinator.com