How to Study for Comprehensive Exams (Comps and Quals)

Study for a comprehensive exam by reading your own program regulations first, tiering the reading list into what you must know cold and what you only need to place, then writing full answers to the questions that list implies. Preparation runs for months, and the format is decided locally, so a general revision plan will not fit it.

Almost everything that makes comps hard follows from four facts. The syllabus is a reading list instead of a course. The scope is a field. The examiners are a specific committee of people you know by name. And the rules, including how many times you may sit it, are written by your department. What follows works through each of those, with two real sets of university regulations to show how far apart programs can be.

What is a comprehensive exam actually testing?

Departments call it several things. Comps, quals, the candidacy examination, the preliminary or prelim exam, and the general exam all name roughly the same milestone: the assessment that stands between finishing coursework and being allowed to write a dissertation. It is usually pass or fail, the number of attempts is usually capped, and failing it can end the degree. That gatekeeping function is why it feels different from every exam you have sat before, and it is why the preparation cannot be a revision timetable with the subject changed.

Two universities describe the target in almost the same words. Ohio State says the purpose of its candidacy examination is “to test a student’s comprehension of the field, allied areas of study, capacity to undertake independent research, and ability to think and express ideas clearly”. UC Berkeley says the intent of its qualifying examination is “to ascertain the breadth of the student’s comprehension in at least three subject areas related to the major field of study”, and then to judge whether that student can think incisively and critically about them. Neither sentence mentions recall. Both ask for command of a field plus a judgement about whether you can work in it independently.

There is very little experimental evidence about how to prepare for one of these, and it is better to say so than to dress up something thin. The nearest useful study asks a narrower question, which is whether the score tells the program anything about what happens next. In 2020 Juliann Sergi McBrayer, Steven Tolman and Katherine Fallon published Doctoral Candidacy Examination Scores and Time to Degree Completion (PDF) in the International Journal of Doctoral Studies, setting out “to determine if there was a relationship between doctoral students’ candidacy examination scores and estimated time to degree completion, measured by dissertation progression”. They report that “The findings denoted a weak to moderate relationship between candidacy examination score and dissertation progression (defending pre-prospectus and/or prospectus) within one year from taking the candidacy examination.”

That finding is worth exactly as much as the study behind it, and the study is small. Forty-five students, in three cohorts, in one EdD educational leadership program at a single university in the rural southeastern United States. The design was “ex post facto, correlational”, so it cannot show that scoring higher causes faster progress; readiness at the moment of sitting could easily produce both. Every one of the forty-five passed and went on to the dissertation stage, so the sample contains no failures at all and is silent about what a fail predicts. The authors ask for more work themselves, noting that “this study analyzed progression of only 45 students” and that a longer timeline and a larger sample are needed. What it supports is modest: the exam appears to carry some signal about the research that follows, which is the argument programs make for keeping it. It supports nothing at all about chasing a higher score, and no number in it is a study plan.

Your program regulations are the first reading assignment

There is no universal comprehensive exam, so advice that opens with a format is guessing at yours. Two published sets of rules show the spread, and both belong to large research universities.

The Ohio State Graduate School Handbook treats it as one exam in two halves: “The candidacy examination is a single examination consisting of two portions, written and oral”. The written half has no fixed shape even inside that one university, because “Rules for the form, timing, scheduling, sequence, and conduct of the written portion are determined by the Graduate Studies Committee”, which is the department. Some departments there stack another exam in front of it: “The Graduate Studies Committee of certain programs may require students to take preliminary or qualifying examinations prior to taking the candidacy examination.”

The UC Berkeley Graduate Division draws the line somewhere else entirely. “The Qualifying Examination is an oral exam”, held with the whole committee present for its full length, and its campus guidelines add that “The examination must not be narrowly limited to the dissertation topic.” Written preliminaries, where a department wants them, sit outside it and come first, which is why the student checklist there lists “Take and pass preliminary exams if required” as a separate step.

The instruction that follows is short. Find the regulations that actually govern you, read the section on this exam end to end, and write down four things: the format, the length, who sits on the committee, and how many attempts you get. If your reading list is negotiated with that committee, the list is the second document to secure, because everything below depends on knowing what is on it.

How do you study a comprehensive exam reading list?

A reading list is not a syllabus, and treating it as one is the most common way months get wasted. A syllabus is ordered, weighted, and taught. A list is a flat set of items, often dozens of them, assembled by people who expect you to work out the connections yourself. Three moves turn it into something studyable.

1. Tier it before you read anything. Sort every item into three piles: works you must be able to discuss in detail without a prompt, works you must be able to place and characterize in two or three sentences, and works you only need to recognize and situate. Most lists are far too long to treat evenly, and the tiering is a decision your committee will effectively test, so make it deliberately and check it with your advisor rather than discovering it under exam conditions.
2. Write one page per item, from the shut source. Use a fixed template so the pages stay comparable: the central claim, the method or approach, the evidence it leans on, what it is usually cited to argue, and the one sentence you would say about it if asked. Writing it with the book closed is what makes it a study session instead of a transcription session, and summarizing a chapter without copying it out sets out the compression moves that keep a page to a page.
3. File everything under debates, not under authors. Pick six to ten questions the field genuinely argues about, and put every item on your list under the debates it speaks to. Most items will land under two or three. This is the layer that distinguishes comps preparation from course revision, because an examiner asks about a problem and expects you to arrive with a literature already sorted around it.

That third move is where the exam is won, and it is worth being blunt about why. Anybody can be asked which book made which argument. A committee is likelier to ask how a field moved from one position to another and what about it is still unsettled. Forty one-page item sheets answer the first question and none of the second, while the debate files are already organized the way the second will arrive.

Turn the list into questions and write the answers

The exam wants an answer, so the practice has to be producing answers. Once the debate files exist, draft eight to twelve questions you think plausible, and then write full responses to them under something close to exam conditions. Timed, from memory where the format is closed, and long enough to be a real answer rather than an outline.

Two sources sharpen the question list beyond guesswork. The first is whatever past questions your department will release, together with the shape of the list itself, since the proportions of a reading list carry information about what the committee thinks matters. Reading those signals is its own skill, and how to predict what will be on an exam sets out how to read them. The second signal is peculiar to this exam: the examiners are named people, their recent work is often on your list, and a question comes from a person with current preoccupations. Read what each committee member has published lately, and at minimum you will not be surprised by the framing.

Then rewrite. A first draft exposes what you cannot yet say, and it is nearly always worse than you expected, which is useful information arriving early enough to act on. Getting that information out of an attempt you have just made is its own procedure, and how to review a practice test sets it out. Redraft the weakest three or four across the following weeks instead of polishing the ones that already work.

How long should you prepare, and how do you build the months?

Long enough that the plan has to survive ordinary life, which is the real constraint. The dates come from your regulations. At Berkeley, “Once an application for admission to the Qualifying Examination is approved by the Graduate Division, the program has 18 months to administer the examination.” At Ohio State, the candidacy examination “must be completed at least one semester before a student can defend and graduate”. Those anchor the exam to opposite ends of the degree, and one of them is binding on you.

Work backwards from that date in blocks rather than days, because the unit that matters here is a week of reading, and building a study schedule that holds covers the mechanics of laying blocks against a fixed deadline. Two adjustments are specific to comps. Reserve the final third for writing and rewriting answers, since a plan that reads until the week before arrives with a full head and no practice at producing anything from it. And plan around the teaching, marking, or research assistantship you will still be doing throughout, because the failure mode is a schedule built for a student with nothing else on.

If yours has an oral component, or arrives as a take-home

Two formats carry enough technique of their own to need separate treatment, and your regulations will tell you whether either one is yours.

An oral component is common enough that both sets of regulations above carry one, and at UC Berkeley it is the entire exam. Producing and defending an answer live is a different skill from knowing the reading, and it rewards rehearsal that has nothing to do with how well you understand the material. How to study for an oral exam, viva, or OSCE covers that preparation: building a question bank, rehearsing out loud from memory, handling a follow-up you did not see coming, and what to say when you go blank. Nothing in this guide substitutes for it.

A take-home comp, written over several days with your sources open, is the other. The trap is the familiar one: access to everything is not the same as being able to find anything under time, and a reading list you can search is still a reading list you have to have understood. How to study for an open-book exam covers the navigation and the note structure that make an open source usable while a clock runs.

Can a question generator help with a reading list?

Only in one corner of it, and the arithmetic is what says so. A single GeniusPal generation reads a single source, while a comps reading list is dozens of separate sources. Two surfaces accept a source, an uploaded file and a pasted link to a web page, so anything you hold as a physical volume needs scanning or an electronic copy first, and nothing accepts notes you type or paste in directly.

Against a list of forty items, the only number the arithmetic needs is the ceiling on the free plan. A free GeniusPal account is issued two study-set generations, they last the life of the account, and nothing on the calendar adds a third. Two items out of forty is a sample, and it is worth spending only on the two you most need to know cold. A paid plan lifts that ceiling well past forty and leaves the underlying constraint exactly where it was, because a generation still reads one source and a reading list is not a source.

Now the boundary, because the useful version of this recommendation is a narrow one. Where GeniusPal earns an hour is on a single dense item you have to know cold, a monograph chapter, a review article, a methods paper: questions drawn from that one text tell you quickly whether the reading landed. Where it contributes nothing is the work that defines comps preparation, which is seeing that four items on your list are arguing with one another and being able to say what the argument is. Questions generated from one text cannot see the other thirty-nine, and that seeing is what the committee is examining. Write the cross-cutting questions yourself; the layer beneath them is the only part worth handing over.

The last thing worth holding onto is that the feature which makes this exam frightening is also what makes it tractable. It is enormous, and it is known in advance. The list is written down, the committee has names and publications, and the rules governing your attempts sit in a document you can go and read this afternoon. Very little of that was ever true of the exams you sat before.