zelticgar

The Diversity Pipeline

352 posts in this topic

Doing some work and one of my tasks is to collect data on the University Pipeline in the US. Lots of pressure around hiring and making sure we are finding diverse candidates. Some interesting data: 

  • 2015 Total Grads - 2.8 Million Degrees (57% Female / 43% Male)
  • 2015 STEM Grads - 476,000 Degrees (34% Female / 66% Male)
  • 2015 Computer Science Grads - 93,000 Degrees  (22% Female / 78% Male)
    • Bachelor (18% Female / 82% Male)
    • Masters (30% Female / 70% Male)
    • PhD (22.5% Female / 77.5% Male)

55% of Masters and PhD studying computer science are foreign nationals so the 30% of female students studying for masters degrees in CS are likely disproportionally non US citizens. The same trend holds true for STEM Grads. 

When looking at most of the major tech firms they tend to get  the negative press around not doing enough to increase the diversity in STEM and CS but the reality is that the University and High School level education systems are not doing as much as they could either to grow the pool of students going into these majors. It seems like the University level has more incentive to continue to grow their Masters programs which are predominately foreign national students. Nothing wrong with that but it leaves many students behind and if the H1B program were ever taken down there is no backup plan to fill in those lost students with US citizens. I don't have any good answers, I just find the topic interesting and I like pointing these things out because it really drives home the opportunities that exist for students willing to focus on STEM. 

 

Share this post


Link to post
Share on other sites

One issue is the feedback loop. The reputation of STEM companies - their work culture, biases, diversity etc - itself prevents people choosing the subjects. Which limits the pipeline, which propagates the existing cultural issues.

Share this post


Link to post
Share on other sites

What leads to the idea that "willing" is the most important factor at play? Or what schools do? There are personal inclinations and abilities that cannot always be simply trumped by "willing" or persuading otherwise. Especially when talking about advanced degrees. If one believes in anything like some distribution of abilities it seems obvious that one cannot increase these numbers by sheer force of willing or persuasion.

(Do foreign students bring more money to the university? this might be another incentive...)

Share this post


Link to post
Share on other sites
5 hours ago, zelticgar said:

I don't have any good answers, […]

Trust me, neither does anybody else. The problem is that people who (like me) admit that they don’t know how to even frame this problem (let alone solve it), have very little leverage.

But currently, all levels of education (K-12, Higher) are filled with idiots making decisions about this very big, complicated, real problem, who were selected for their ability to advocate snake oil. It is very frustrating.

Share this post


Link to post
Share on other sites
5 minutes ago, Jo498 said:

(Do foreign students bring more money to the university? this might be another incentive...)

One perspective that I find eye-opening and depressing is that the problem becomes larger, the more feminist a society becomes. The ultra-feminist Scandinavian societies have even more of a problem than the US. Societies that put a lot of values on individual self-actualistation and self-expression, and give large individual economic freedom to women, have very few women in CS.

Crudely put, women’s liberation is well measured by their absence from CS educations.

This confuses me.

Share this post


Link to post
Share on other sites

But why is this even a problem?  (Not to be too facetious, but this imbalance seems a major problem for all the male students in these subjects who find it harder to find girlfriends if they mainly hang out with fellow students from their major which is a common thing to do.)

Should people be forced into computer science who'd rather do law, medicine or (God forbid!) nursing, art history or agriculture? Gently tugging or showing opportunities is one thing. But I think we have been doing this for decades now.

I am the first person to concede that "equality of opportunity" can be an empty phrase if outcomes are blatantly different from what one would expect. But I think one can really exaggerate "equality of outcome", especially in cases where it is not at all clear that a personal or common good is served by more female CS students or more male veterinarians (the latter must have been a rather male profession until quite recently but at least in Germany it seems now overwhelmingly female). I think there are already enough bad/false incentives for people picking certain careers against inclinations and abilities or funnelling people into basically parasitic sectors (like FIRE) because there is the most money to be made that we do not have to add more bad incentives to change some percentages by a few points.

(I wonder if the remarkably gender-balanced STEM sector of the former Soviet Union was a result of the socialist education system or of central planning or of incentives different/missing compared with a capitalist society.)

Share this post


Link to post
Share on other sites
2 hours ago, Jo498 said:

Should people be forced into computer science who'd rather do law, medicine or (God forbid!) nursing, art history or agriculture?

Has anyone ever suggested that they should?

2 hours ago, Jo498 said:

Gently tugging or showing opportunities is one thing. But I think we have been doing this for decades now.

Well, we've been making attempts to do this. Some pretty half-hearted, most of them not notably successful, and all undermined by the continuation of negative social influences. 

The implication of your argument is that the lack of success is a factor of genuine free choice by individuals, which invites the question: why would genuine free choice result in these differences?

Share this post


Link to post
Share on other sites

Posted (edited)

4 hours ago, Jo498 said:

But why is this even a problem?  (Not to be too facetious, but this imbalance seems a major problem for all the male students in these subjects who find it harder to find girlfriends if they mainly hang out with fellow students from their major which is a common thing to do.)

Should people be forced into computer science who'd rather do law, medicine or (God forbid!) nursing, art history or agriculture? Gently tugging or showing opportunities is one thing. But I think we have been doing this for decades now.

It is a problem because it is a significant driver of future economic opportunity that is not being utilized by people who might otherwise be set on a path to success. The issue for why US students are not interested in pursuing STEM fields at the level needed to keep up with demand are more complicated than simply assuming computer nerds are just looking for girlfriends.

When you start to look at Ethnicity it gets even more dire. Almost 600,000 degrees are conferred to Black and Hispanic students per year but only about 65,000 are in STEM and only 15,000 in CS. There were 51 PhD's issued to Hispanic CS majors in 2015. 

 

4 hours ago, Jo498 said:

What leads to the idea that "willing" is the most important factor at play? Or what schools do? There are personal inclinations and abilities that cannot always be simply trumped by "willing" or persuading otherwise. Especially when talking about advanced degrees. If one believes in anything like some distribution of abilities it seems obvious that one cannot increase these numbers by sheer force of willing or persuasion.

(Do foreign students bring more money to the university? this might be another incentive...)

The reason for the large number of foreign students is due to the economic incentives tied to coming to the United States. The typical path for Indian nationals is to complete their undergraduate studies, work one to two years at a consulting firm or large US tech company in India and then come to the United Stated and spend two years obtaining a Masters Degree. They typically pay full tuition for the MS and when they graduate they move into the work force with a degree from a top US school, two years of prior work experience and can command anywhere from 80k to 120k to work at large tech firms.

Clearly not all students have the capacity to study CS or other STEM fields but there are plenty of students who opt to pursue majors that provide less long term economic benefits that could otherwise have pursued a STEM field. 

Edited by zelticgar

Share this post


Link to post
Share on other sites
1 hour ago, mormont said:

Has anyone ever suggested that they should?

Well, we've been making attempts to do this. Some pretty half-hearted, most of them not notably successful, and all undermined by the continuation of negative social influences. 

The implication of your argument is that the lack of success is a factor of genuine free choice by individuals, which invites the question: why would genuine free choice result in these differences?

Below is a historical view of CS Undergrads since 1996: 

Something to think about the next time you hear about the struggles of big tech firms with diversity. Somehow the education system has been able to deflect the bad press and they are not being held accountable the same way that industry is. 

 

Year
Bachelor's degrees
Total
Males
Females
Females as a %
Number Annual % Change
1996-97 25,422 3.70% 18,527 6,895 27.10%
1997-98 27,829 9.50% 20,372 7,457 26.80%
1998-99 30,552 9.80% 22,289 8,263 27.00%
1995-96 24,506 -0.09% 17,757 6,749 27.50%
1999-2000 37,788 23.70% 27,185 10,603 28.10%
2000-01 44,142 16.80% 31,923 12,219 27.70%
2001-02 50,365 14.10% 36,462 13,903 27.60%
2002-03 57,433 14.00% 41,950 15,483 27.00%
2003-04 59,488 3.60% 44,585 14,903 25.10%
2004-05 54,111 -9.00% 42,125 11,986 22.20%
2005-06 47,480 -12.30% 37,705 9,775 20.60%
2006-07 42,170 -11.20% 34,342 7,828 18.60%
2007-08 38,476 -8.80% 31,694 6,782 17.60%
2008-09 37,992 -1.30% 31,213 6,779 17.80%
2009-10 39,593 4.20% 32,414 7,179 18.10%
2010-11 43,066 8.80% 35,477 7,589 17.60%
2011-12 47,406 10.10% 38,796 8,610 18.20%
2012-13 50,961 7.50% 41,874 9,087 17.80%
2013-14 55,271 8.50% 45,320 9,951 18.00%
2014-15 59,581 7.80% 48,840 10,741 18.00% 

 

 

Share this post


Link to post
Share on other sites
1 hour ago, mormont said:

Has anyone ever suggested that they should?

Well, we've been making attempts to do this. Some pretty half-hearted, most of them not notably successful, and all undermined by the continuation of negative social influences. 

The implication of your argument is that the lack of success is a factor of genuine free choice by individuals, which invites the question: why would genuine free choice result in these differences?

Your last is a very thorny question.  The assumptions one carries in to any attempt to test it will strongly influence what is found in the course of testing.  

Share this post


Link to post
Share on other sites

So what kind of hypotheses are entertained as explanation for a drop of female CS graduates by about 1/3 both in percentage and absolute numbers since the early 2000s? (Maybe in the mid/late-1990s CS was touted because of the Y2K problem?)

Should long term economic benefits (for the student?) be the main or only reason for picking a certain subject? This is obviously one strong incentive, but it is obviously not decisive, otherwise there would be far more STEM applicants.

I agree that it is bad to let people study something for 4 or 6 years and then drop them into precarious jobs. But from this does not follow that more people could go into STEM. (It might as well follow that fewer people should go to college and especially that fewer people should do subjects that hardly improve their employability.)

Someone who is gifted enough that they would have been brilliant both in e.g. computer science and art history will with a high probability do well enough in art history to get one of the few decent jobs. Maybe one loses a good programmer but gains a good museum director. Nothing wrong with that person following their dreams.

I seriously doubt that there are a considerable number of students who pick art history because of their inclinations but would have done well in CS. Everybody knows that art history has only very few decent jobs to offer for graduates, unlike CS. So the "typical" art history student who mainly wants to follow their inclination for a few years and ends up as barrista would probably have failed CS anyway.

I don't doubt that there is some "collateral damage" (I should know, I am one), i.e. students who would have done decently enough to be employable with e.g. an engineering degree but not gifted or competitive enough to get one of the few decent jobs in some humanities. To minimize this one needs mentoring and other help. But I don't think one can go by numbers and demand certain proportions for e.g. males or females. I don't know about the US and it is hard to compare because the systems are different but in Germany the percentage of college students of a cohort has almost doubled in the last 25 years. I tend to the unpopular opinion that people have not gotten so much smarter in that time period that this development is a laudable achievement.

Share this post


Link to post
Share on other sites
1 hour ago, mormont said:

The implication of your argument is that the lack of success is a factor of genuine free choice by individuals, which invites the question: why would genuine free choice result in these differences?

Why would they not?
A different way to ask the question would be "why would free choice not result in some differences?"

To talk about the gender divide... Of course we all know that living in patriarchal societies has a considerable impact on such statistics, especially if higher education is expensive like in the US or England.
However, one thing to bear in mind is that the statistical differences persist, and are sometimes even worse, even in Western countries where higher education is almost free of charge. In other words, even if you remove the major obstacle to higher education, you still get crushing statistical differences. Or to put it differently, what one would expect to be a major obstacle to free choice actually has little statistical influence.
Of course, one way to interpret such differences is to say they are a problem because they are the result of other undesirable factors. I dunno. I think even if you somehow do away with most patriarchal elements you'll still find significant differences. Or yet a different way to put it is that one should be careful about what statistical differences "ought to be" corrected, and which ones are in fact the result of genuine free choice.

Share this post


Link to post
Share on other sites
8 minutes ago, Rippounet said:

Why would they not?
A different way to ask the question would be "why would free choice not result in some differences?"

To talk about the gender divide... Of course we all know that living in patriarchal societies has a considerable impact on such statistics, especially if higher education is expensive like in the US or England.
However, one thing to bear in mind is that the statistical differences persist, and are sometimes even worse, even in Western countries where higher education is almost free of charge. In other words, even if you remove the major obstacle to higher education, you still get crushing statistical differences. Or to put it differently, what one would expect to be a major obstacle to free choice actually has little statistical influence.
Of course, one way to interpret such differences is to say they are a problem because they are the result of other undesirable factors. I dunno. I think even if you somehow do away with most patriarchal elements you'll still find significant differences. Or yet a different way to put it is that one should be careful about what statistical differences "ought to be" corrected, and which ones are in fact the result of genuine free choice.

Ripp,

Which is exactly why I think the assumptions an individual brings to a this will have tremendous impact on what they find in a given statistical analysis.

Share this post


Link to post
Share on other sites

There are also a vast differences between fields. Many academic/professional fields that were considered "male" in the 1950s have achieved (close to) gender parity or have actually more female graduates by now, e.g. medicine and law. I don't really believe in curiously selective patriarchy that affects high prestige fields like medicine far less than electrical engineering or CS. While I find the drop in the stats above since the early 2000s astonishing, 18% is not that bad from a "social" perspective, i.e. one would not be the only woman in a class most of the time. When I studied physics in the 1990s the anecdotally felt percentage of women majoring in physics was in the low single digits and I believe it was worse in some engineering subjects (but slightly better in maths and certainly better in chemistry; biology probably had already a female majority, not sure about CS).

Share this post


Link to post
Share on other sites

Posted (edited)

8 minutes ago, Jo498 said:

There are also a vast differences between fields. Many academic/professional fields that were considered "male" in the 1950s have achieved (close to) gender parity or have actually more female graduates by now, e.g. medicine and law. I don't really believe in curiously selective patriarchy that affects high prestige fields like medicine far less than electrical engineering or CS. While I find the drop in the stats above since the early 2000s astonishing, 18% is not that bad from a "social" perspective, i.e. one would not be the only woman in a class most of the time. When I studied physics in the 1990s the anecdotally felt percentage of women majoring in physics was in the low single digits and I believe it was worse in some engineering subjects (but slightly better in maths and certainly better in chemistry; biology probably had already a female majority, not sure about CS).

Jo,

I've read somewhere that there are more women going to law school now, on average, than men.  I'll have to look for it.

[ETA]

Here it is:

https://www.nytimes.com/2016/12/16/business/dealbook/women-majority-of-us-law-students-first-time.html

Edited by Ser Scot A Ellison

Share this post


Link to post
Share on other sites
1 minute ago, Ser Scot A Ellison said:

I've read somewhere that there are more women going to law school now, on average, than men.  I'll have to look for it.

Anecdotical experience: I've been teaching in law schools for the past four years in two different French universities and I usually get around 80% female students. The official national proportion is around 65%.
Numbers are comparable for med school (where I also taught, although for a single year).
In engineering, the numbers are reversed (i.e. only around 30% of female students).

Share this post


Link to post
Share on other sites

This is the CS Table for Masters Students. Apologies for not adding the year column but i'm lazy (96 to 2015). You can see that there is a steady increase in female MS students. Almost all of this can be attributed to increased enrollment of foreign students (55% of the graduates are foreign).

Somehow other countries (Primarily India and China) have figured out how to incentivize females to study CS. Imagine what would happen if this section of students were to go away via an H1B policy change? There are very few US citizens to step in to fill this void. 

 

Master's degrees
Total
Males
Females
Females as a %
10,513 7,526 2,987 28.41%
11,765 8,343 3,422 29.09%
12,843 8,866 3,977 30.97%
10,579 7,729 2,850 26.94%
14,990 9,978 5,012 33.44%
16,911 11,195 5,716 33.80%
17,173 11,447 5,726 33.34%
19,509 13,267 6,242 32.00%
20,143 13,868 6,275 31.15%
18,416 13,136 5,280 28.67%
17,055 12,470 4,585 26.88%
16,232 11,985 4,247 26.16%
17,087 12,513 4,574 26.77%
17,907 13,063 4,844 27.05%
17,955 13,019 4,936 27.49%
19,516 14,010 5,506 28.21%
20,925 15,132 5,793 27.68%
22,782 16,539 6,243 27.40%
24,514 17,472 7,042 28.73%
31,474 21,892 9,582 30.44%

Share this post


Link to post
Share on other sites

So what is your main point here? That there should be more female CS students to confirm to some standard of diversity (implying that the rate should be around 50%)? Or the danger that the US would soon not have enough qualified graduates for the field? Should not the invisible hand fix the latter quickly ;) because then wages would rise and people would have a stronger incentive to go into the field?

Share this post


Link to post
Share on other sites

Posted (edited)

12 minutes ago, Rippounet said:

Anecdotical experience: I've been teaching in law schools for the past four years in two different French universities and I usually get around 80% female students. The official national proportion is around 65%.
Numbers are comparable for med school (where I also taught, although for a single year).
In engineering, the numbers are reversed (i.e. only around 30% of female students).

Sorry.  I misread your post.

Edited by Ser Scot A Ellison

Share this post


Link to post
Share on other sites
14 minutes ago, Jo498 said:

So what is your main point here? That there should be more female CS students to confirm to some standard of diversity (implying that the rate should be around 50%)? Or the danger that the US would soon not have enough qualified graduates for the field? Should not the invisible hand fix the latter quickly ;) because then wages would rise and people would have a stronger incentive to go into the field?

Point is wasted opportunity to improve the economic status of a huge population. I also think we need to spend as much time pointing the finger at our education system as we do at industry when discussing problems related to lack of diversity in particular fields. 

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.