From: Ben Y. Zhao, Professor at Univ. of Chicago, Alumni Interviewer for Yale

OK, so I’m going to go ahead and disagree with the answers already here, which is why I’m writing. I’m writing in the context of PhD applications in computer science, although I think much of the high level arguments generalize.

Yes, students can take GREs. Yes, some of these schools will have graduates from good institutions (although graduates from the very top grad schools tend not to go back to “unheard-of-colleges in the third world”). Yes, there may have been prior applicants from some unknown school X.

Ultimately, none of these factors really address the elephant in the room, which is that if you’re coming from a truly unheard-of-school from a third world country, chances are, you’ve got a serious uphill battle to get into any top graduate program. Yes, you can indeed score really highly on GREs, which of course would help. But that’s not the biggest challenge. The biggest problem is that your letters of recommendation are going to be worth very little, because they will come from people who a) might not know how to write good LoRs for top graduate programs, and b) are unknowns to grad admission committees, and therefore their opinions are completely uncalibrated.

Much of the graduate admissions process is about mitigating risk. The very best students are often hard to identify, and the committee can be more focused on reducing the likelihood of admitting unqualified students, those who will flounder in the system or drop out in 2 years with a MS degree. High test scores and high GPAs help reduce risk, but the best factor for reducing risk in an applicant is a strong letter from someone who is well-calibrated (i.e. their technical qualifications and LoR writing styles are known to the committee). GPAs and uncalibrated letters are all dependent on the quality of the institution (and therefore problematic for unheard-of-schools). Outside of GREs, which aren’t a great predictor of research potential, the only other way to get calibrated indicator of a student’s potential is peer-reviewed publications.

It’s not fair. But top graduate programs have their pick of the best applicants both domestic and international. Given the choice between a “safe” candidate, e.g. 3.9 GPA from Yale with superb GREs and amazing letters from colleagues known to the committee, and a high-GPA, high GRE applicant from unknown school X in a third-world country with uncalibrated letters, the decision is often obvious. I recall being upset many years ago when I learned that Berkeley CS had an unspoken rule where for many years, they did not consider any applicants from China, simply because they had so many near-perfect domestic students in their applicant pool. I think they’ve changed those policies since then (but I’m not sure when). If a school like Berkeley can disregard applicants from Tsinghua, Peking Univ, and the like, imagine how they would regard an unknown school from a third-world country.

This is the unfortunate reality in many of the top departments. If you are an applicant from a little known school with no ties to top graduate departments, your best shot (in addition to scoring extremely high on GRE and TOEFL), is to get published at well-regarded (ideally the top) conferences in your research area. Peer-review serves as a well accepted, global calibration mechanism. I’ve seen how 1 or 2 top papers can help the committee downplay the school’s reputation and uncalibrated letters.

I’m sorry the news is not more positive. I believe it’s in students’ best interests to remain cautiously optimistic after learning the facts. But remember that you don’t need the odds to be good, you just need to be that one exception to the rule.

Good luck!


Source