PhD Advisor Red Flags: What You Can Check

PhD Advisor Red Flags: What You Can Check

Three PhD advisor red flags can be checked from public records before you accept an offer: funding that ends before you would finish, a graduation record worse than the department's own numbers, and a lab page that disagrees with the papers. Department baselines exist. Stanford reports that 59.6% of its Computer Science PhD students finish within six years; MIT's dashboard shows 15.2% attrition for EECS doctoral cohorts entering 1997–2019.

Personality red flags ("toxic", "micromanager", "absent") are hard to check: you cannot observe them in a one-day visit, and current students may not say them to a stranger. Our view is to check the records first, then use conversations with students to explain what the records show. If you are still at the stage of choosing whom to apply to, start with how to find a PhD supervisor instead; this page is for the admit stage.

Every university page and database below was checked on October 29, 2026. Dashboards are updated yearly; recheck the numbers for your department.

On this page: when you can vet · red flag 1: money · red flag 2: completion · red flag 3: lab page vs papers · questions for students · worksheet · checklist

When you can actually vet an advisor

At the three US programs in this table, you cannot vet an advisor before admission, because the program does not settle advising then.

Sources: MIT EECS admissions FAQ, CMU CSD doctoral admissions, Stanford CS PhD admissions FAQ, Oxford DPhil in Computer Science (2027-28 entry; read via a reader service after a direct request returned HTTP 403).

MIT also says that once a student is admitted, "some interviews and student visits will be arranged in order to determine the level of shared interest between the student and faculty." That visit is your vetting window.

Our view: where advisors are matched after admission or after arrival, vet the department's rules first (who funds you, and what happens if you change advisor), then the two or three people you are most likely to work with. The countercase: where you name a proposed supervisor at application, as Oxford expects, or apply to an advertised project, the person is close to being the offer. There, the person-level checks below come first and earlier. Even at Oxford, the department allocates supervision, so ask who your supervisor will be before you accept.

Red flag 1: the money ends before you would finish

First, read the department's own promise. Some departments fund every admitted PhD regardless of which professor's grant pays:

• MIT EECS: "students are fully funded throughout the duration of their graduate program, assuming satisfactory progress in their research and in their academic PhD milestones."
• CMU's Computer Science Department "offers all admitted Ph.D. students comprehensive financial support … while they are in good academic standing in their respective programs."

Under those rules, one professor's grant ending does not, on the published wording, end your stipend. It can still change your project. MIT says "Research assistants are chosen by faculty who are seeking students who share their research interests," so the money follows research topics. Where the department makes no such promise, your stipend may depend on one professor's grants, and this check matters more. Our fully funded PhD guide explains what a funding promise covers.

Then look up the professor's federal awards:

1. NIH RePORTER (reporter.nih.gov) accepts "PI names" in its search. Each project record lists a project start date and end date.
2. NSF Award Search (nsf.gov/awardsearch) lets you search by principal investigator first and last name, and filter active or expired awards. Each award carries a start date and an expiration date.
3. Compare the end dates with your finish year. Stanford CS reports a median time to degree of 5.5 years; MIT's dashboard shows 5.8 years for EECS. If the only award you can find ends in your second year, ask what pays for years three to six.

The limit of this check: these two databases cover NIH and NSF only. A lab can be funded by other agencies, foundations or companies that these searches do not show. An empty result is a question to ask, not a red flag.

Red flag 2: students who do not finish

You need a baseline before a lab's record means anything. Two departments publish one:

Sources: Stanford IR&DS Doctoral Completion & Time-to-degree dashboard; MIT IR Graduate Education Statistics dashboard, department "Electrical Eng & Computer Sci".

Stanford defines "discontinued" as students who "have either left the university without a degree or switched to a non-doctoral degree program (in many cases a master's degree)." That definition is what to look for in a lab's history.

How to use the baseline:

1. List the PhD students who joined the lab at least six years ago. Sources: past versions of the lab page, the department's list of past students or dissertations, and the professor's papers.
2. For each, record whether they finished the PhD, left, or left with a master's.
3. Compare with the department. In Stanford CS, roughly three in four of the 2018-19 and 2019-20 entrants had completed by November 2025. A lab where one of five finished is worth a direct question.

Small numbers mislead. One or two departures can have reasons unrelated to the advisor, such as a job offer or a family move. Treat a pattern across several students as the flag, and ask about it.

Red flag 3: the lab page and the papers disagree

A lab page is written by the lab. The papers are a second record. Compare them over the last five years:

• Students on the lab page who appear on no papers. A first-year student is normal. A fourth-year student is a question.
• Names that disappear. A student listed two years ago and missing now, with no dissertation, is the "left" category above.
• Who is first author. In fields where the student who did the work is listed first, a lab where students never lead a paper is worth asking about.
• Alumni listed with a master's, not a PhD. That matches Stanford's "discontinued" definition.

Use the professor's own publication list, the department's dissertation list and archived copies of the lab page. Note what you found, not what you assume about why.

Questions only current students can answer

The records show what happened. Current students can tell you why. Our recommendation is to ask these privately, during the admitted-student visit, and to ask at least two students:

1. How often do you meet the professor one-on-one, and who sets the agenda?
2. Who chose your current project, and when did it last change?
3. How is author order decided, and when?
4. What paid for your last summer, and was there a gap?
5. What happened to the last student who changed advisors?
6. How long did the last three graduates take, and where did they go?
7. Is the professor planning a sabbatical, a move or a startup leave?
8. When a grant ended, what happened to the students on it?

Each question maps to one of the three records above. If the answers and the records disagree, believe the records and ask a follow-up.

Constructed example: a vetting worksheet for two professors

Constructed example, written by us. Both professors and all details are invented. Both sit in one department that, like CMU's, promises funding to all admitted PhD students in good standing.

How to read it:

• Dr. Menon's single award is not a stipend risk here, because the department promise covers the stipend. It is a project risk: if the 2027 award ends, the work may change in your second or third year.
• The completion record is the real flag. One of five finishing, with three leaving with a master's, is well below a department baseline like Stanford CS's. That is the question to ask Dr. Menon directly: what happened to those students?
• Dr. Okafor's record raises nothing to ask about. It does not prove a good fit; it only means the records show no flags.

Neither row is a verdict. The worksheet tells you which questions to ask before you accept.

Your checklist before you accept

1. Read the department's funding promise and its conditions. Write down whether the stipend depends on one professor.
2. Search NIH RePORTER and NSF Award Search for each likely advisor. Record award end dates.
3. Find the department's completion data, if published. Write down the baseline.
4. List the lab's students from six or more years ago and record outcomes.
5. Compare the lab page with five years of papers.
6. Ask two current students the eight questions above.
7. Check the rule for changing advisors in the department handbook. For CS students working outside their department, see can your CS PhD advisor be outside CS.
8. Get funding terms in writing; nine-month versus twelve-month PhD funding lists what to confirm.

Weighing two offers? Choosing between two PhD offers turns this worksheet into a side-by-side decision.

Questions about PhD advisor red flags

What are the biggest red flags in a PhD advisor?

The ones you can verify: funding that ends before you would finish, a lab whose students finish less often than the department's published rate, and a lab page that does not match the papers. Personality concerns matter too, but check them through several current students.

Can I ask a professor about funding before I am admitted?

At MIT EECS, no: "Faculty members will not discuss financial support or thesis supervision with a student who has not yet been admitted." Other programs differ. Read each program's admissions FAQ.

What if the professor has no NSF or NIH grants?

Ask where the lab's money comes from. NIH RePORTER and NSF Award Search cover only those two agencies; money from other agencies, foundations or companies does not appear in them.

What is a normal PhD dropout rate?

There is no single figure. For Stanford CS, 15.9% of 2018-19 entrants and 7.9% of 2019-20 entrants had discontinued as of November 2025. MIT's dashboard shows 15.2% attrition for EECS cohorts entering 1997–2019. Use your own department's numbers where they exist.

Can I switch advisors if it goes wrong?

Check the department handbook before you accept. At Stanford CS, students align with an advisor after their first few months, so the first choice is made after arrival. Rules after that point vary by department.

Get a free review of your PhD statement of purpose

Get a free Quick Review of your PhD statement of purpose against GradPilot's adviser-fit rubric. It reads whether your statement ties a focused research direction to a person, lab or area in the program with checkable details, and what your own work would contribute. For the whole statement, use the general PhD statement of purpose rubric. Writing to a named supervisor for an advertised position? Use the PhD position cover letter rubric. The PhD SOP review page explains how it works.

More PhD guides: CS PhD acceptance rates at Stanford, MIT and CMU, contacting a PhD supervisor, and the PhD essays hub.

Sources

Retrieved October 29, 2026.

• MIT EECS, Graduate Admissions FAQs (no year stated): supervision and funding discussions, funding promise, research assistants, admitted-student visits.
• Carnegie Mellon Computer Science Department, Doctoral Admissions (fall 2027 cycle): advisor matching, funding promise.
• Stanford Computer Science, PhD Admissions FAQ (no year stated): advisor alignment.
• University of Oxford, DPhil in Computer Science (2027-28 entry): proposed supervisor; department allocates supervision.
• Stanford IR&DS, Doctoral Completion & Time-to-degree (text and dashboard, Computer Science; cohort status updated November 2025): time to degree, six-year rate, cohort status, definitions.
• MIT Institutional Research, Graduate Education Statistics dashboard (Electrical Eng & Computer Sci): time to degree, cohort completion, attrition.
• National Institutes of Health, RePORTER (data as of October 25, 2026): PI search; project start and end dates (checked through the public RePORTER API).
• National Science Foundation, Award Search: PI name search, active and expired filters; award start and expiration dates (checked through the public NSF awards API).