CRA study: How long does it take to get a PhD in CS? Longer for women and minorities

April 30, 2014 at 9:12 am 6 comments

Really interesting new study out of Computing Research Association (CRA). How long does it take to get a PhD in CS? How does that compare to other STEM disciplines? How does it differ based on gender or minority status?

Table 3 and Figure 1 show the median time to complete a Ph.D. since first beginning a graduate program, for each subgroup, for each cohort.

Gender . Women take longer than men. This is true in both cohorts; there is a larger difference (almost a year) in the second cohort.

Citizenship status. In the earlier cohort, students on temporary visas take less time than citizen or permanent resident students. In the later cohort, the median times of the two groups are exactly the same.

Minority status. Students from underrepresented minorities (URM) – that is, racial and ethnic groups underrepresented in computing – take longer than majority students to complete a Ph.D.. In the first cohort, the difference is almost two years; in the second cohort it is close to one year.

Carnegie Class. Eighty percent of doctorates in computing are granted by “Very high research activity” institutions; students at those institutions take noticeably less time to complete their degrees than those at the less-research-intensive institutions.

via Computing Research News – Online – Computing Research Association.

Entry filed under: Uncategorized. Tags: , , .

Summer is for Writing A really nice MediaComp Image Blending lesson

6 Comments Add your own

  • 1. Bijan Parsia  |  April 30, 2014 at 2:01 pm

    Ok, so my xteen years is an outlier…I knew that! 🙂

    I’m still struck by the fact that we can, in the UK, pretty reliably deliver ≈4.5 (to viva) and ≈4 years (to submission) PhDs. I don’t feel the thesis quality is lower. They aren’t shorter. People aren’t starting later (a good chunk of the time). It’s highly expected that your first job (in academia) will be a post doc, but that’s normal, right?

    Has anyone studied the content of theses to determine e.g., substance, inherent time-to-conduct, etc.?

    • 2. Bri Morrison  |  May 1, 2014 at 9:31 am

      I’m no expert on the UK system, but in the US I believe we have additional requirements for the PhD than the UK system. We have a coursework requirement (9 classes), I have to TA two classes, plus qualifying exams, proposal, and then defense. Additionally, most students do internships during their summers (rather than research on their dissertation).

      • 3. Bijan Parsia  |  May 1, 2014 at 9:38 am

        Bri, that is all correct (I’ve been in both systems).

        My question isn’t whether we can do all that faster (probably not, though I believe that the proposal to thesis defence is usually unduly elongated and definitely the classes to proposal), but whether we should replace the current system with something more UKish. One downside (possibly) is making the overproduction of PhDs worse. But, on the flip side, I’m unclear that the extra schooling is truly of benefit, in aggregate and esp. vs. the costs to students. Some person power could be transfered from PhD courses downward which might relieve the lost of (exploited) student labor. We might also have rapid PhD then more postdocs (though in all cases, I think it should be MUCH better funded).

        I think “lack of PhD” traps people more than “I’m a post doc and want to be a prof”. It’s psychologically harder to jump ship. So I think even if stipends stayed comparable, that people would be better off.

        • 4. Bijan Parsia  |  May 1, 2014 at 9:48 am

          Obviously, a 3-4/4.5 year PhD a la UK isn’t a panacea, but I’d really like to know the value add we get from doubling the time. My suspicion is that the main value add is cheap labor for Universities and, with the hollowing out of the tenure class + overproduction, the usual expected benefit is gone.

  • 5. Alan Fekete  |  May 2, 2014 at 5:53 pm

    I would say that the key to understanding this issue is the difference in undergraduate education between UK and USA.

    Basically, the UK undergraduate degree is specialized: a typical British undergraduate education is 3 years long, and cosists entirely of subjects that are targeted to their field, so for example in UK students do a Bachelors in Computing (or in Maths, or in Physics, but let’s use Computing as the example). That means that about 2.5 years are computing subjects, and the rest is relevant maths or digital stuff. No breadth requirements or general education reuqirements. Also, the British degrees typically have much less flexibility: all the students take mostly the same subjects in the same order (and so each following subject can assume a lot of prior knowledge, and start further along).

    In contrast, the USA Bachelors is a Bachelor of Arts, or sometimes of Sciences, with 4 years in which students accumulate credits from a smorgasbord of choices. The requirements for a major in computer science are typically 8-10 subjects (the lower value seems common at the large number of small colleges), so between 2 to 2.5 years, and each subject typically has to assume a wide variety of backgrounds in the students [the AI subject may assume programming and discrete maths, but it probably can’t assume logic programming; even if logic programming is part of the major, it is likely that some students do AI first then LP, and others do LP then AI] so each subject typically takes a somewhat broad approach, with quick introductions to relevant sidetopics that are needed or useful. The rest of the 4 year degree in USA is filled with electives which may (and often many must) be outside computing: a foreign language, a history subject, expository writing, etc etc. I believe many/most US colleges require about 2 years of such breadth in the 4 years of the degree (though they vary in whether the breath is fixed core subjects, or just choices that must tick various categories).

    So an American PhD has to be designed for students who have a wide variety of backgrounds, and a significant number of them arrive in the PhD with only a light knowledge of eg operating systems or algorithms, let alone formal methods or database internals or HCI. Look at the ACM Computing Curriculum to see just how little is actually required and thus common to all US undergrad majors. That is why the US doctoral program has lots of coursework (which includes redoing topics from the undergraduate curriculum, but at a much more thorough and deep way). This gives every PhD graduate a really good breadth of detailed expertise, which I would say was missing in the typical British PhD (who got the breadth in their ugrad education, but it is not such deep breadth!)

    By the way, the Australian undergraduate degree is between the British and American models: it is 3 years, but mostly only partially specialized (no humanities subjects, but rtaher more flexibility among the computing and related science/engineering subjects). But then before heading for a PhD, a student must do a 4th year called Honours (unlike the UK or USA where Honours are awarded within the usual degree duration); this is like a cutdown Masters; it has a major research thesis (at least one quarter of the year, and usually half the year in computing; in Physics/Chemistry/Bio it is typically about 75% of the year) and some advanced/specialized coursework. After this, the Australian PhD lasts 3 to 4 years, and has no coursework (or almost none). Then there is the European model: 3 year specialized undergraduate {like in UK), followed by 2 years Masters with advanced coursework and a preliminary thesis or project; follwed by 2-3 years of PhD as an employee in a professors group, doing teaching and research.

    The British model gets one to PhD fastest (say 6 years after high school) but I don’t think it produces a graduate with the same overall skillset as most Americans and Europeans have. To me the difference isn’t in the research speciality, but rather in the breadth: how many totally different topics one knows about in a fairly deep way. I see American PhDs as very broadly skilled, able to teach lots of things, move across research fields, etc. Whether this breadth and the joy of exploring widely during both ugrad and the PhD, is worth the lost time/income (common description in USA: doing a PhD in computing costs you a house) is a point that each potential student must decide for themselves.

    • 6. Bijan Parsia  |  May 2, 2014 at 6:34 pm

      Hi Alan,

      Up front I don’t find your account remotely plausible. First, it is very flattering to the US model in tone but ignores internal tension (eg even if correct in outline it doesn’t explain why people with 4 years of university education including 10+ courses in computer science, take as much or more time than people straight out of secondary school; or consider that UK phd programs train people from outside the UK system with wildly different backgrounds)

      I have been in both the UK and US systems (which I mentioned earlier, but perhaps you missed that? Your comments are rather off if not) and your observations don’t ring true at all (and could easily be generated as a just so story from just looking at the structure of the program).

      I don’t see a fundamental difference between the phd students I’ve trained in the us or in the uk. Yes, us students have had broader exposure to topics, but I don’t see that that translates into greater overall success in teaching or research.

      Finally, you just skipped the part where I suggest that short phd + postdoc might be a better overall model than long expensive phd.

      Finally, I find it shockingly irresponsible and cruel to pretend that the us system is offering a choice to pursue an idyllic period of exploration. There’s a lot of great things about a good phd experience, but, in point of fact, it is, as currently constituted, brutally explorative as well as disrespectful and often tormenting. Even if you think a long phd has virtues, then you should champion eg fair wages.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Trackback this post  |  Subscribe to the comments via RSS Feed

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 9,045 other followers


Recent Posts

Blog Stats

  • 2,024,519 hits
April 2014

CS Teaching Tips

%d bloggers like this: