How to evaluate undergrad CS programs when eventually aiming for PhD?

Question

My daughter has a strong interest in theoretical computer science, and she has just been admitted to several different universities as a freshman. She is just a bright student, not a genius, with a bit of research experience in cryptography and randomness stuff, and a semifinalist in a couple of national competitions. She intends to continue graduate study in theoretical CS, or a closely related subfield, after finishing her undergraduate degree.

Given her strong interest in eventual graduate study, what criteria should we take into account in choosing where to enroll as an undergraduate? How can we tell which undergrad programs best prepare their students for graduate school?

Answer 1

Although I'm sure you already realize this, you shouldn't let a laser focus on your daughter's professional development control her choice of which school to attend. (Influence, sure, but not control. And yes: HER choice.) Given your daughter's background and experience, she will likely be very successful no matter where she enrolls. I've seen very successful theoretical computer scientists come from a huge variety of undergraduate programs, not all of which were associated with strong graduate CS programs. It is at least as important for her to find a place where she feels (and not just is) supported and challenged, both by her faculty and her peers. Otherwise, no matter how strong the department is on average, it won't be a good fit for her personally.

There is absolutely no substitute for a personal visit to each campus, if you can manage it. Let her drop into the comptuer science department, get a feel for the community vibe, sit in on a class, talk to a professor or two if she can, hang out in the dorms. Meanwhile, you go somewhere else. Go'way. Shoo.

But if you really want hard data, I recommend that you—or better yet, your daughter—directly contact the computer science departments at the schools where she's been admitted, describe her experience and her goals (which should make them very eager to attract her), and then ask questions like the following:

• How many students graduated from your program in the last five years?
• How many / what fraction of those students are currently enrolled in strong computer science PhD programs?
• How many / what fraction of your current undergraduates are involved in research with your faculty?
• How many of your faculty do research in theoretical computer science?
• [The Hail Mary:] Would you mind asking one of them to contact me? I'd like to ask some more questions.

Beware the answer "I don't know", or qualitative waffle like like "quite a lot" that really mean "I don't know". Ask for numbers. Ask for examples. Channel your inner Cuba Gooding Jr: Show me the data!

Your first contact will almost certainly be answered by someone on the administrative staff, who may not have the data you want at hand. Be prepared make an appointment to speak over the phone (or Skype or whatever) with the director of undergraduate programs or another faculty member. If you do speak over the phone, try to keep your conversation short and to the point; these people are very very busy.

Don't bother asking admissions officers these questions; Chances are very high that they just don't know. You really need to contact the departments directly.

Another good place to look is the list of current and past winners and runners-up for the Computing Research Association's Outstanding Undergraduate award, which specifically recognizes undergraduate research. A significant fraction of winners did their undergraduate research in theoretical computer science. Also, a significant fraction of winners were not enrolled in departments with a top-10 graduate program.

But as Anonymous Mathematician notes in a comment, you shouldn't take the CRA list too seriously; some departments are more interested in awards like this and push hard for their students to be selected. The fact that some prominent departments have few or no students on this list may reflect a disinterest in this particular award, not in undergraduate research in general.

[I'm a theoretical computer scientist at a top-10 CS department; I serve on my department's graduate admissions committee.]

Answer 2

Best researchers go to Universities where they get the best students to teach, and the best students go to Universities where they get good education and enough attention from the faculty. So, things balance out. Departmental reputation and standard University ranking tables are perfectly fine to use as a guide to decide where to go. Pretty much all the top departments will have good strength in Theoretical Computer Science. You will need to look closely only if you are looking at lower-ranking Universities, which might have strength in some areas but not others.

There is also no conflict between good undergraduate education and good grad school, because the grad students do contribute to teaching. They run tutorials/discussions and help students with questions and exercises. So, a department that is good at grad school is also likely to be good in undergraduate teaching.

There can be a bit of difference between private Universities and State Universities. Because private Universities are more dependent on the tuition fees and alumni donations, they are likely to spend more effort on undergraduate teaching. State Universities also tend to have larger class sizes, something to consider if you are looking for "personalized attention". You can't ask for too much personalized attention because, if that is what the faculty are engaged in, they are not going to have much time for research and, as a result, the quality of the department would go down in the long run.

If you are really after personalized attention, then you might consider Liberal Arts Colleges rather than Universities. Undergraduate teaching is their No. 1 focus. They don't normally have a grad school. While the faculty would engage in some research, research is not considered the top priority. Teaching is.

Answer 3

It is commonly said that the only university that counts is the one you've graduated from last. If your daughter wants to eventually earn a PhD, the graduate school she attends will most likely figure more importantly to her future employability than her undergraduate school.

With that in mind, it is also true that many undergraduate schools with some of the top graduate schools in CS also produce excellent candidates for graduate school, primarily because the number of undergraduate research opportunities are greater, and also because it can be assumed that there is a good deal of correlation between the quality of the graduate program and the quality of the undergraduate program.

However, it would be wise to also consider the strength of the undergraduate program from the standpoint of all the normal factors that go into choosing an undergraduate school, e.g., (1) the kind of student your daughter is, and (2) whether the undergraduate program will give her what she needs as a student holistically. For example, some phenomenal CS programs are at schools where class sizes can be in the hundreds (e.g., UC Berkeley), but if your daughter thinks she would get swallowed up in a giant class, that probably isn't the place for her. A smaller school with smaller classes might be more appropriate, even if the CS program is weaker as a whole.

Finally, the top two (in my opinion) factors that will affect your daughter's success in getting into graduate school are:

1. Her grades (despite what some prominent and highly successful members of this community have achieved). Getting a 4.0 from any decent school in a particular major is better than a 2.5 from the best school in the country.

2. Undergraduate research, with published results. Just doing research is important, but having published results will help her case significantly.

Bottom line: I suggest choosing a school where your daughter feels comfortable, and with enough of a research program that she could get involved with research as an undergraduate.

Answer 4

Selecting a school is a very challenging thing to do. Just because a school has top-quality research faculty doesn't necessarily mean they have top-quality teaching faculty (brilliant researchers are not always the best communicators, unfortunately). The best way to get one-on-one time with a professor is usually to corner them during their office hours and keep asking questions until they insist they need to leave.

Wherever your daughters ends up, encourage her to start making a connection with individual faculty members as early as possible. Faculty are almost always happy to bring new students into their labs, and there is funding available for undergraduates who are interested in research. Many universities offer REUs (research experience for undergraduates). The primary issue I've come across is that these programs have difficulty finding undergraduates to give money to.

As far as grad school goes: tell her to make good grades and get involved with a lab. Everything else will fall into place (and that's a long way off, anyways).

Answer 5

Have her go to the school where she would be least distracted. Princeton is better than NYU, as it is farther from Broadway (1 hr vs. half-hour). Yale is even better (3 hours). Urbana-Champaign is impossible to beat in that regard -- cornfields for hours to drive (and you can still deny her of a car), she would have nothing else to do but study.

I guess I would also want to make sure that the university's regular math program is good. If she can get a credit for non-math AP courses, that's great, as it will clear some time to do real work in college; math and CS will have to be retaken, or taken at a higher level of expectations, and/or in the honors college. Some good quantitative bridge courses like biology or psychology may be something else to pique her interest to do something CS-related that is funded by another substantive field (many of the great discoveries are being made at the disciplinary boundaries). Knowing what you are going to do for the rest of your life at 18 is incredibly boring -- sounds more like a lawyer than an academician!

Answer 6

here are some standard rankings/surveys that might be helpful. for very elite schools, they are all highly rigorous, and maybe unique nonacademic features particularly appealing to your daughter (eg "culture/atmosphere", which is difficult to objectively characterize) might shift the balance.

the last 2 links are based on Shanghai Jiao Tong University's graduate school of education which recently ranked the world's best universities in computer science, based according to their academic or research performance in each subject field. it ranks 9 of the top 10 in the US. good luck!

• Worlds best universities in computer science, USNews

• best grad schools / computer science, USNews

• Computer Science Colleges' Best Programs, As Ranked By 'Academic Ranking Of World Universities

• The Best Universities: Computer Science