← Blog

Letters of recommendation give a PhD admissions committee something you cannot supply yourself: another researcher’s assessment of your work. Faculty reading your file want to know whether you can do original research, how you compare with other students, and whether someone who knows the field will vouch for you.

You don’t write the letters, but you shape the conditions in which they are written. You choose whom to ask, when to ask, and what material your recommenders have in front of them when they sit down to write.

This post covers both sides: how applicants can arrange the strongest possible letters, and what recommenders should put in them. It is the third guide in my series on PhD applications, following the statement of purpose and the personal statement.

What committees look for in a letter

Political science departments say plainly that letters matter. Northwestern’s department calls them “very important” and says its committee most wants letters from people who can assess an applicant’s academic and research abilities in their field. Dan Nexon, writing in 2012 when he directed admissions in Georgetown’s Government Department, went further: for some committee members, recommendations are the most important part of the file.

A strong letter answers questions your grades cannot:

• Can you do research? Evidence from a thesis, research assistantship, or substantial paper is especially useful.
• How do you compare? Committees want a comparison group: the best student in a class of 60, for example, or one of the three strongest research assistants in ten years. Many forms ask for this directly. Stanford’s form asks recommenders to place applicants in a band for overall ability, ranging from the top 1–2% to the bottom quarter, with a “no basis for judgment” option.
• How well does the writer know you? A letter grounded in years of close work can say much more than one based on a single course.
• Is the writer credible? Researchers who understand doctoral work can explain why your preparation matters for it.

The best letters also tell specific stories. “She is very intelligent” tells the committee little. “She spotted a coding error in our replication data that two co-authors had missed” tells them much more.

A weak and a strong letter

Consider this illustrative letter. It can hurt an applicant without saying anything openly negative:

[Name] was a student in my Comparative Politics course in Spring 2024 and received an A. He is sincere, hardworking, and punctual. I am confident he will do well in graduate school, and I recommend him for admission to your esteemed program.

It is short, confirms only what the transcript already shows, and offers neither a comparison nor evidence of research ability. To a committee used to detailed letters, those omissions can read as a lukewarm endorsement.

Here is an illustrative excerpt from a stronger letter:

I have taught for twelve years and supervised more than forty undergraduate theses. [Name]'s is among the three best. She asked why some city councils publish their budget data and others don't, built a dataset of 210 councils from scratch, and taught herself the panel methods she needed in one semester. When her main data source turned out to have gaps, she found the missing records in a regional archive and rebuilt the analysis in three weeks. She is the strongest student I have taught in the past five years, and I recommend her without reservation.

The second letter gives the writer’s basis for comparison, a ranking, a specific research story, evidence of independence, and an unqualified close.

Choosing your recommenders

Most programs ask for three letters, although you should check each program’s requirements. Look for writers in roughly this order:

1. A faculty member who supervised your research, such as your thesis advisor or a professor you worked for as a research assistant. Harvard’s Joshua Kertzer points out that faculty who have employed you as an RA are well placed to judge your research abilities.
2. A political scientist, ideally in your subfield. Kertzer advises political science applicants to have at least one letter from a political scientist.
3. A professor whose demanding course you did well in, and who knows your work beyond the grade: a seminar paper, office-hour conversations, or a methods project.
4. A professor in a related field, such as economics, statistics, or sociology, who can speak to your quantitative or analytical skills.

For a research PhD, I would generally aim for at least two academic letters. Nexon advised applicants who have been out of school for a while to seek two academic letters and one from a manager who knows them extremely well. A boss from a job with no research component usually has less to say about your readiness for doctoral research.

Don’t choose a writer for their name alone. Daniel Drezner wrote in 2012 that nearly all the letters he had read from very famous scholars were “bland, impersonal boilerplate,” and warned that such a letter could hurt an applicant. A detailed letter from a less famous professor who knows your work is more useful than a vague one from a star.

How and when to ask

Ask early. Six to eight weeks before your first deadline is a reasonable minimum; earlier is better if your recommender writes for many students. Stanford’s admissions office tells applicants to register recommenders in the online application as early as possible.

Ask in person or by video call if you can, then follow up in writing. Give the person an easy way to decline:

Do you feel you know my work well enough to write me a strong letter for PhD programs?

Take a hesitant yes seriously. A lukewarm letter can do more harm than asking someone else.

Once they agree, tell them how many programs you’re applying to and when the first deadline is. Each program usually sends its own request, so ten programs can mean ten separate uploads.

What to send your recommenders

Send the materials together as soon as they agree. Recommenders can only write the specifics they remember, so help them recall the work.

The one-page summary often does the most work. List the courses you took with them and your grades, the papers or projects you completed, your RA tasks, and two or three specific moments they might remember. These could include a question you raised in a seminar or a problem you solved in the data. Keep it factual and leave the adjectives to them.

Portals, waivers, and reminders

Most applications collect letters online. You enter each recommender’s name and email address, and the system sends them an upload link. Enter recommenders as early as each portal allows. Some systems send the request when you add a name; others wait until a later step, so check how each one works.

US applications commonly ask whether you waive your right to inspect the recommendations. Keeping that right does not normally let you read a letter during admissions: access generally concerns letters retained in your educational record after you enroll. A valid waiver removes that right of access. FERPA’s regulations also say that an institution cannot require the waiver as a condition of admission.

I generally recommend waiving access once you have chosen writers you trust to support you. As the University of Michigan’s fellowships office explains, strong letters often contain comparisons that writers would not want circulated. Some writers only agree to submit confidential letters, and committees usually regard confidential letters as more candid. The decision is yours, so read the waiver before selecting an option.

Then keep track and follow up politely:

• About two weeks before each deadline, send a short reminder listing which letters are still pending.
• A few days before the deadline, check each portal and follow up only where a letter is missing.
• Make sure the recommender has everything they need without having to chase you for information.

If a professor asks you to draft your own letter

This happens in the US and elsewhere, but check the rules before agreeing. Some universities forbid it outright. Stanford requires recommenders to be the sole authors of their letters. Applicants violate its application terms if they draft, write, translate, or submit their own reference, even when a recommender asks them to do so.

There are practical problems too. A letter you draft may sound like you, and the committee reads your statement of purpose in the same file. You also cannot supply the experienced scholar’s independent judgment of how you compare with other students.

A useful response is to offer the material a busy professor needs:

I'd be glad to make this easier for you. Instead of a draft, could I send you a one-page summary of my work in your course and on the project, with specific examples and dates, along with my CV and statement of purpose? Some of the programs I'm applying to require that letters be written entirely by the recommender.

If your recommender still asks for a draft, check each program’s policy first. Where applicant assistance is permitted, provide facts and examples while leaving the writing, comparisons, and judgments to the recommender. The letter must represent their assessment.

For international applicants

Letter-writing conventions vary across countries. In some settings, a recommendation is traditionally short and formal, closer to a certificate than an argument. US doctoral recommendations often expect more detail and explicit comparisons; forms such as Stanford’s ask writers to rank applicants directly.

Experienced committees know that styles differ, but a detailed letter gives them more evidence to work with. A few things help:

• If you have a choice, consider recommenders who have written for US programs before or studied abroad themselves.
• Explain politely what the programs request: specific examples, comparisons with other students, and an assessment of research ability. Sharing a short guide can make that conversation easier.
• Make sure each recommender can describe your work in detail. The one-page summary matters even more here.
• If a recommender writes in another language, ask each program how it handles translations. Stanford treats translating your own reference as a violation of its application terms.

For recommenders: what to put in the letter

If you’re writing the recommendation, these are the things a PhD committee needs to understand:

1. How you know the applicant: in what role, for how long, and how closely.
2. Your basis for comparison: how many students or research assistants you have taught or supervised, and where.
3. Evidence of research ability: one or two specific stories about a thesis, paper, or RA project, with details you observed.
4. Relevant skills: writing, methods training, independence, and how the applicant handles setbacks.
5. A clear ranking, where you have a basis for one: explain where the applicant stands among students you have taught, and make the letter consistent with the ratings on the form.
6. A clear close: if you can offer strong support, say so plainly. The strength of the endorsement should reflect your honest judgment.

One to two pages on letterhead, signed, is a useful target unless the program specifies otherwise. Write for a research PhD: research potential comes first. Leave out personal details that do not bear on the application.

Watch for faint praise. Words such as “hardworking,” “diligent,” and “dependable” describe effort rather than demonstrated ability; a letter built around them can read as a cautious endorsement. A University of Arizona Commission on the Status of Women tip sheet recommends emphasizing accomplishments and considering descriptions such as “insightful,” “independent,” and “accomplished,” when the evidence supports them. Drawing on research about academic recommendations, it also warns that effort-focused language appears more often in letters for women. Check that your descriptions give each applicant’s achievements comparable attention.

Common mistakes

• Choosing a famous name over someone who knows your work.
• Asking two weeks before the deadline.
• Sending a request without materials, leaving the writer with little beyond your grades.
• Relying on people who cannot assess research ability, such as an employer in an unrelated job or a family friend.
• Letting your letters and statement of purpose tell different stories about your interests.
• Making a confidentiality-waiver decision without understanding what it means.
• Sending more letters than a program requests without checking its instructions. Extra letters may add little and create more reading.
• Forgetting to say thank you.

A timeline

Working back from a first deadline in early December:

1. Summer: identify your three or four strongest options and keep doing good work with them.
2. Early October, about eight weeks out: ask each recommender.
3. Within a few days of a yes: send the full packet.
4. As each application opens: enter your recommenders in the portal.
5. Two weeks before each deadline: send a short, polite reminder.
6. A few days before: check that every letter has arrived.
7. After you submit: send a thank-you note.
8. In the spring: tell each recommender where you were admitted and where you’re going.

The last step matters. Your recommenders invested in you, and many will write for you again: for fellowships, grants, and eventually jobs.

A last word

The strongest letters come from people who know your research, have the material to describe it in detail, and have enough time to write well. You can do a great deal to make that possible: choose carefully, ask early, send a useful packet, and say thank you.

Good luck with your applications.

Further reading

• Joshua Kertzer, “Information for Prospective Graduate Students.” A Harvard political scientist on the application process, including who should write your letters.
• Dan Nexon, “Applying for a PhD in Political Science” (The Duck of Minerva, 2012). An admissions director’s account of letters and the rest of the file.
• Daniel Drezner, “So you want to get into a political science Ph.D. program… Episode I” (Foreign Policy, 2012). Includes his warning about generic letters from famous professors.
• Stanford Graduate Admissions, “Recommendations.” The rating scale recommenders see and the requirement that they author their own letters.
• University of Arizona Commission on the Status of Women, “Avoiding Gender Bias in Reference Writing.” A one-page checklist for recommenders.