How Harvey Mudd Transformed Its CS Program And Nearly Closed Its Gender Gap | Hacker News Reader

How Harvey Mudd Transformed Its CS Program And Nearly Closed Its Gender Gap

techcrunch.com

28 pointsgkuan48 comments

Here are Harvey Mudd's acceptance rates by gender:

Men 14.4%

Women 38%

Source: http://www.parchment.com/c/college/college-545-harvey-mudd-c...

I am perpetually bamboozled how anyone can think anecdotes like this matter. If Harvard wanted too I am sure they could make half their undergraduate population disabled black lesbians. That does not prove that your average state school is horrible and discriminatory just because that is not their class composition.

Elite schools get more applicants of either gender than they have room for and those applicants are of very high quality with very high levels of interest in the subject matter.

Well, the higher female acceptance rate may be because only good female students do CS. If you have a hostile environment, only the most motivated will succeed. It's like a sieve. You are applying selective pressure against a certain group of people succeeding. Expect to get only the most motivated.

Not saying this is the case in reality, but I think the conclusion that they're indiscriminately accepting women is facile.

Citation needed.

It is much more plausible that these engineering colleges have laxer requirements for females in order to get a 50/50 gender balance rather than public school discrimination.

Citation needed.

Anecdotal rebuttal: I know a bunch of female students at Harvey Mudd and that doesn't appear to be the case.

It doesn't need a citation, it is supported by Occam's razor while you are suggesting a complex explanation.

How does Harvey Mudd magically close their gender gap as though they suddenly decided to stop discriminating against females all these years?

If the girls you knew were, say 10% less capable compared to male applicants, would you be able to tell? I doubt it.

> If the girls you knew were, say 10% less capable compared to male applicants, would you be able to tell? I doubt it.

If you can't tell then what difference does it make?

It makes a difference to the male applicants that would've been accepted if there wasn't a bias.

A program erasing the current cultural bias/momentum against females will make a difference to all the females who will not have to face said momentum/bias.

But how do you know that a 50/50 balance is even natural in engineering fields like CS?

Why isn't there a 50/50 balance and other similar initiatives for males for female dominated subjects like nursing and education?

Is there solid evidence of an IQ difference between the genders or a general subject, with a biological understanding of why such a difference exists?

I currently have not come across any such solid evidence and biological understanding. Seems unproductive to operate under an understanding far from 50/50 when I do not have solid evidence that I should be.

I never brought up the argument that males are smarter than females.

Why wouldn't you consider the possibility that most females don't want to sit in front of a screen all day with minimal social interaction often working 70 hour weeks?

Is it because you feel that the field of CS is so superior that no one could dislike it?

> I never brought up the argument that males are smarter than females.

I was not implying you did. IQ is one well know metric but others such as "a general subject", as I asked for before, are welcome as well.

I asked for solid evidence that operating away form 50/50 is more productive. If you have some I am willing to look it over.

I don't know why you assume that 50/50 is the desired ratio when acceptance is supposed to be self-selected and merit based.

This CS professor from MIT explains why fewer females choose engineering studies like CS.

http://philip.greenspun.com/careers/women-in-science

Specifically this quote.

> What about women? Don't they want to impress their peers? Yes, but they are more discriminating about choosing those peers. I've taught a fair number of women students in electrical engineering and computer science classes over the years. I can give you a list of the ones who had the best heads on their shoulders and were the most thoughtful about planning out the rest of their lives. Their names are on files in my "medical school recommendations" directory.

> I don't know why you assume that 50/50 is the desired ratio when acceptance is supposed to be self-selected and merit based.

I was not trying to communicate my desire. Rather what is optimal? I need strong evidence to operate far from 50/50.

> Specifically this quote.

This counts as evidence, but not enough evidence to get me to think the optimum is close to having 14-18% female CS degree recipients(2010).

Nice graph over time: http://blogs.computerworld.com/sites/computerworld.com/files... from article: http://blogs.computerworld.com/it-careers/21993/women-comput...

Summary statistics: http://www.ncwit.org/sites/default/files/legacy/pdf/BytheNum...

You prefer 50/50 but you have no evidence to even suggest that this is the "best" distribution for CS let alone any other major.

Why can't you accept the possibility that females do not like CS as much as males do not like things like nursing or human resources?

If there is a social stigma, should we really force people who normally wouldn't choose it so we can have a 50/50 ratio just because that is the probability to be born as?

Who are you to say that they are happier doing CS instead of what they would've chosen without all this influence?

> You prefer 50/50 but you have no evidence to even suggest that this is the "best" distribution for CS let alone any other major.

50/50 is the default with out additional information. Again I am not trying to statement my preferences.

> Why can't you accept the possibility that females do not like CS as much as males do not like things like nursing or human resources?

I did not say I can not. Like any idea I want good evidence and solid understanding to make strong claims.

> If there is a social stigma, should we really force people who normally wouldn't choose it so we can have a 50/50 ratio just because that is the probability to be born as?

I have not seen anyone make the argument to force people to take CS courses.

> Who are you to say that they are happier doing CS instead of what they would've chosen without all this influence?

Trying is better then not trying. Influence exist weather intentional or not. Unintentional influences are not inherently better then intentional.

With only 660 students and around 3,000 highly qualified applicants, I'm pretty sure Mudd can be very picky about who they admit without introducing a bias.

The 48 percent statistic is for the junior year, the third year. It seems like your argument would require Mudd to grade females more lightly for the at least the first three years. This seems unlikely since it would require some form of collusion between all/most CS professors(maybe TAs too).

I don't follow this argument. You are suggesting that the rate of females leaving the major/school is exactly proportional to the grade they get.

I don't believe anyone would quit a school/major because they scored a little bit lower average. This might be possible if they all happened to be near the threshold for failing but I doubt Mudd is this strict.

Also everyone is well aware of affirmative action, so even if their grades were worse they would not think they would not be able to compete for the same jobs/academic positions.

> I don't follow this argument. You are suggesting that the rate of females leaving the major/school is exactly proportional to the grade they get.

The percentage of females normal drops year by year in the CS programs I am familiar with. This does not seem to be happening at Mudd to the degree it does elsewhere. I have not see you put forward how initial acceptance has changed this trend, or that the trend exists at Mudd and they simply start the first year with a high female ratio say 40%-60% male-female.

> I don't believe anyone would quit a school/major because they scored a little bit lower average. This might be possible if they all happened to be near the threshold for failing but I doubt Mudd is this strict.

People do switch majors when they are below average, not always but I would be surprised if this reason is rare. Your argument now requires that the females are below average compared to the males but not so below average that they want to switch majors. I would count this as a minor complexity penalty to your argument.

> Also everyone is well aware of affirmative action, so even if their grades were worse they would not think they would not be able to compete for the same jobs/academic positions.

This by it self does not work at most universities. So what makes Mudd different that this argument works there?

> The percentage of females normal drops year by year in the CS programs I am familiar with. This does not seem to be happening at Mudd to the degree it does elsewhere. I have not see you put forward how initial acceptance has changed this trend, or that the trend exists at Mudd and they simply start the first year with a high female ratio say 40%-60% male-female.

Extraordinary claims require extraordinary evidence. The burden of proof lies with you, not with me.

> People do switch majors when they are below average, not always but I would be surprised if this reason is rare. Your argument now requires that the females are below average compared to the males but not so below average that they want to switch majors. I would count this as a minor complexity penalty to your argument.

This is not complex at all. Switching majors is major decision with many roadblocks. You will need approval and you will likely need to retake courses. This has dire implications such as wasting credits which means you will need more money to stay in college longer. You might even lose scholarships.

All evidence points that differences between ranked individuals will be difficult to see. In activities that are completely quantifiable such as Olympic racing, the difference between the top people are measured in milliseconds, so it is quite feasible that the difference between male and female CS students at a selective college is similarly small in terms of absolute measurements.

> This by it self does not work at most universities. So what makes Mudd different that this argument works there?

What makes Mudd different is that they accepted more females to start with. This reasoning causes minorities to stay, it doesn't increase the number of acceptances.

> Extraordinary claims require extraordinary evidence. The burden of proof lies with you, not with me.

You will have to explain what claim you find extraordinary and what specific evidence you would find convincing for me to consider doing research to provide some.

> This is not complex at all.

I did not call it complex. You made an argument by occam's razor if details are added to the argument to make it viable it make it more complex and it benefits less from occam's razor. My argument or your added details:

> All evidence points that differences between ranked individuals will be difficult to see. In activities that are completely quantifiable such as Olympic racing, the difference between the top people are measured in milliseconds, so it is quite feasible that the difference between male and female CS students at a selective college is similarly small in terms of absolute measurements.

Weaken an occam's razor argument, some times called applying a complexity penalty.

> What makes Mudd different is that they accepted more females to start with. This reasoning causes minorities to stay, it doesn't increase the number of acceptances.

If I understand you correctly you think that the percentage of females in the 1st through graduation is controlled tightly by the starting percentage. Do you just expect it to hold steady for all 4 years?

I already repeated multiple times what claim I find ordinary: that Mudd is somehow able to retain so many female CS majors simply by accepting more.

It is up to you to prove that they are able to do this otherwise.

> You made an argument by occam's razor if details are added to the argument to make it viable it make it more complex and it benefits less from occam's razor. My argument or your added details

I only explained the common sense reasoning that you somehow keep ignoring. I did not add any details the argument, merely to reinforce the fact that my argument stems from common sense while yours appears to make wild conjectures on how Harvey Mudd is better than every other CS department.

> I already repeated multiple times what claim I find ordinary: that Mudd is somehow able to retain so many female CS majors simply by accepting more. It is up to you to prove that they are able to do this otherwise.

State your prediction for the female population by year dependent on initial enrollment percentage and I will consider do some hard research and bring it back.

> I did not add any details the argument, merely to reinforce the fact that my argument stems from common sense

Occam's razor argument is usually only effective if their is a obvious difference in the complexity of the two arguments. The details are often needed to establish weather or not their is a significant complexity difference.

> yours appears to make wild conjectures on how Harvey Mudd is better than every other CS department.

Is their a particular claim you are referencing? I try to stay away from making wild conjectures.

Please use "women" instead of "girls", unless you are talking about children under ~11.

what if they're under 21? I'd call a male college student a boy in most situations.

Point taken although I'm not sure I agree. I hope "female students" is an acceptable compromise.

How would the students know that they had laxer requirements?

Close to triple through natural occurrence? Come now, you're the one being facile.

The point is the acceptance rate by itself is too murky a statistic to prove the gp's point.

Note that if acceptance rate is a murky statistic, then using underrepresentation to prove discrimination might be even murkier, because underrepresentation is a function of the acceptance rate and the number of applicants.

> Note that if acceptance rate is a murky statistic, then using underrepresentation to prove discrimination might be even murkier,

If it was an isolated incident then yes. The gender difference in engineering/science and in this case CS is fairly broad and it seems unlikely that it does not point to something significant, cultural momentum and bias being a popular choice.

So what?

Under-representation is not used primary to prove discrimination, Under-representation is simply an undesirable state of affairs to many people. Some of the remedies for this involve correcting apparent discrimination, some involve encouraging girls to get involved with CS and some involve making CS more appealing to everyone (with the hope this will include a lot of girls).

The focus or, uh, fixation of a certain breed of poster on "this doesn't prove discrimination, you can't prove I'm a sexist, no way..." seems telling to me but of course what seems clear to an Internet poster isn't evidence sufficient for policy.

Doing the math it looks like there were about 873 female applicants and 2271 male applicants.

It's certainly possible that a higher percentage of the much smaller female group were qualified than the larger male group.

In this case, where the assumption is that high school females are less likely to want to pursue a math/sci/tech degree, I'd guess it's likely that the average female applicant to Harvey Mudd is more qualified than the average male, simply because she's already going against the grain by applying.

But none of that has anything to do with the extremely positive aspect of making the compSci program more fun and accessible to all comers by ditching Java and C++ in the intro classes and focusing on solving real world problems. Bravo to that.

Great point.

So the next question is - for how many years will they admit less qualified women before the policy can be rescinded? Or will it stay in place as long as its required to have a 50-50 incoming class - which basically means quotas forever?

Those statistics are for the entire University. The statistic they gave was 48 percent female in the junior CS class. Most Universities seem to have problems with females transferring out in the first two years. So it does not seem the initial acceptance criterion should weigh agains the accomplishment.

You know, on reflection, this is actually a good thing. Not because of the diversity - screw diversity. Because you'll have more people with aptitude admitted. Programming aptitude is mostly evenly distributed regardless of gender and race, therefore if you keep your number admitted persons constant, but give a good incentive for every kind of person to apply, you'll end up with better applicants. Or to put it another way, the best 1000 white male americans are worse than the best 1000 americans in general. And at least in this case, I'd expect that there would be about as many capable females as males. Things might go skewed if we try to have the same number of heterosexuals and homosexuals or left-handed and right-handed people though.

"Programming aptitude is mostly evenly distributed regardless of gender and race,"

Is it really?

Is there solid evidence of an IQ difference between the genders or a general subject, with a biological understanding of why such a difference exists? I currently have not come across any such solid evidence and biological understanding.

Independently of whether it is or isn't, there is benefit to trying to get as broad a swatch of the population into programming as possible.

My thinking on this has changed over the years. I used to view programming as a pure meritocracy. Then I realized that the same thing could be said about so many other male dominated fields, whose barriers eventually broke down. (Law, Accounting, Teaching...) Now my inclination is to believe that there are other things at work: parental expectations, mentorship, etc. that are bigger drivers.

And how much downside is there if we smartly (key word!) try to be more inclusive in the field? I don't mean lowering standards, but I do mean outreach, identifying mentors, etc.

Harvey Mudd's come a long way. In the 1970's, the few women (less than ten) were joking referred to as "non-males".

Super cool. I hope more departments follow their lead. (Of course, most CS departments are not headed by women.)

There may be more than you think, for instance:

Stanford Computer Science is headed by Jennifer Widom.

Berkeley EECS's Associate Chair is Tsu-Jae King Liu.

It seems a bit odd that they're quoting their Junior class makeup-- makes me think that's the best data point they have, and it exaggerates their success. My understanding is that CS enrollment by gender isn't the issue, it's CS graduation rate.

In my lay observations most universities have a problem with females transferring out in the first two years. So it can across as impressive to me that they had such an even split in the 3rd year of classes. The program has only been going on for 6 years so I would be surprised if all of the upper level curriculum has even been transferred over flexible homework/project structure described so their is probably still room for improvement by continuing to apply the same strategy to the high level classes.

How big is Harvey Mudd's CS program?

about 200 students, iirc.

Great instructors. Zach Dodds is the man!

Reply on news.ycombinator.com