Ph.D. admissions works differently that undergrad or masters admissions. Understanding how it works, is a huge help in understanding how to craft your application.

The Goal of Ph.D. Admissions

The goal of Ph.D. admissions is to admit student who will find an advisor to work with while successfully completing a Ph.D.

There are two criteria that must be met for this to happen:

1. The students application must be strong.
2. A compatible (in terms of research area) faculty member must willing (and able) to accept new students.

The second criteria is at the heart of why Ph.D. admissions differs from undergrad or MS admissions. While those admissions processes are about filtering the incoming students, Ph.D. admissions involves both filtering and matching students to advisors.

Individual Faculty Play a Pivotal Role in Ph.D. Admissions

This is how admissions work at UC San Diego, where I served on Ph.D. admissions for many years:

1. You submit you application.
2. 2-3 faculty from the admissions committee read it and mark it as “admissible”, “inadmissable”.
3. The “admissible” files become available to the entire faculty.
4. Faculty members use whatever (legal) criteria they like to select admissible students to be actually admitted.
5. Students that some faculty member selected for admission are admitted. The others are not.

The details vary from school to school, but the critical point is that it is almost certain that some (and probably most or all) of the schools you are applying to have something like Step 4, where the preferences of individual faculty play a pivotal role in admissions decisions.

How Faculty Decide Who to Admit

Many factors can determine whether a faculty member will admit a particular students:

1. How strong the student’s research background is.
2. Whether the student is a good fit for the advisor’s research plans for the coming years.
3. How well the student did during an phone interview with the faculty member.
4. Whether the student’s file has any “red flags” that the advisor particularly cares about.
5. Whether the faculty member has enough research funding to support the student.
6. Whether the faculty member is looking at applicants at all. She might be on sabbatical and not even looking at applicants.

If you don’t consider the diversity of criteria faculty use (1-4 above) and the fact that faculty may not be looking at all (5-6 above), admissions decisions can appear very irrational and/or inconsistent.

For instance, if the faculty that do security research all happen to be on sabbatical, or short of funding, or already have enough Ph.D. students then it’s possible that no security students – no matter how excellent – will be admitted that year. In the same year, “worse” students might be admitted in another area, because faculty in that area are flush with funding.

What This Means For You

For applicants to a Ph.D. program, this process has important implications:

1. Don’t take rejection personally. It might have literally nothing to do with you.
2. Take admission personally. It is often because a faculty member specifically wants you to come.
3. Apply broadly. There is no guarantee you’ll get into any particular school, no matter how good your application.
4. There is no universal answer to questions like “Is my statement of purpose more important than my letters of recommendation?”
5. All the parts of your applications should be as strong as you can make them.
6. Personal connections can have a huge impact on your being admitted.
7. The competitiveness of admissions varies across areas.

On the last point, consider that UC San Diego’s CSE program received 11 times more applications for Ph.Ds in machine learning than for Ph.D.s in computer architecture. However, UCSD has only 2.6 times more machine learning faculty that architecture faculty. As result, the program is much more competitive for machine learning applicants that computer architecture applicants: The admission rates for ML and architecture were 6% and 15%, respectively.

My Experience of Grad Admissions

The sections above are a summary of what I’ve read, heard, and observed about how Ph.D. admissions works. Here, I describe my own experience and approach to Ph.D. admissions to give you data point about how this actually works.

My program runs Ph.D. admissions as I described above, so the onus of identifying good matches for my lab falls mostly to me.

I start by identifying all the students who either 1) mention keywords relevant to my research area in their application, 2) mention me by name, or 3) are outstanding applicants in adjacent areas.

This year that required me to look at about 150-200 applicants. I’d do this in batches of 10-15, moving very quickly. My first pass looks like this:

1. Check most recent school and prior degree to calibrate GRE and GPAs. (5 seconds)
2. Check GPA and GRE for anomalously bad values. I don’t look at transcripts – it takes too long. (5 seconds)
3. Skim resume for publications, interesting (i.e., non-internship) work experience, and anything unusual. (5-20 seconds)
4. Skim letters of recommendation looking for evidence of success on projects and research experience and anything unusual (1 minute each)
5. Skim statement of purpose looking for evidence research/project experience and anything unusual (1 minute)

Based on the first pass, I’ll narrow the big pile down to maybe 30-40 applicants to look at more carefully. I’ll go through them in a little more detail (maybe another 3 minutes per file) and select maybe 20 to interview.

The interviews are 20 minutes are focused mostly on trying to understand their motivation, enthusiasm, how they approach problems, their creativity, and whether or not I can imagine talking to them for about 1 hour per week for 5-6 years. At the end, I’ll select between 6 and 12 for admission.

Other perspectives

• Reflections CS Graduate Recruiting at CMU.
• Nadia Heninger’s take on admissions at PSU and UCSD.

Video Answers To Related Questions

We did a whole information session about this.

• How do I know where to apply? Where might I get in?
• Should I still apply if I don’t have any research experience?
• ​ How important is it to reach out to potential advisors before applying? Can they vouch for you in the application process later on?
• How does work experience as software engineer factor in? Or does that only count if you’re part of a published research project?
• grades, GMAT/GRE, recommendations or research experience?
• What’s the weight of different parts of the application (test scores, grades, letters, SOP, undergraduate research)?
• What are some of the qualities of an ideal candidate for a PhD? Conversely, what might raise a red flag in a candidate’s application?
• What’s the weight of different parts of the application (test scores, grades, letters, SOP, undergraduate research)?
• How should an undergrad get more exposure to the different research fields if we’re not sure what we want to specifically go into?
• Everyone says machine learning is where it’s at, but I love (another area). Should I do ML or follow my heart?
• How many Ph.D. programs should I apply to? How do I choose?
• How confident should I be on a specific research interest before applying to a Ph.D. program?
• Does working on an undergrad research project without producing any publications look bad?
• How to apply to Ph.D. program after spending time in Industry?
• How important are different parts of applications: Letters, SOP, GRE, grades…?
• Do you have suggestions for Ph.D. application timeline?
• What form should the interaction with faculty prior to applying take?