Questions Supervisors Ask

The split is usually written down somewhere, and almost nobody is shown it. In most programs the Chair owns the scholarship — whether the argument holds, whether the method answers the question — and the committee owns specialist judgment in their own areas, while an institutional review stage owns compliance with the university's own standards. Those are three different tests and they are not applied at the same time. What each stage is actually looking for.

All of it, at least once, before it goes forward — and this is the thing nobody says out loud. Committee members read chapters. Reviewers read against checklists. It is entirely possible for a manuscript to reach institutional review having never been read start to finish by one person, which is precisely how a document that passes every stage individually fails as a whole. You are the Chair, and you are the only person who will read all of it. Before the dissertation leaves the hands of you and your committee, it is a task you must complete.

Assess it. Fixing it is somebody else's job, and it is legitimate for it to be somebody else's job. Universities say so themselves. A good example: Capella's current Doctoral Publications Guidebook has a section telling candidates how to hire and vet an editor. A Chair who spends their time on comma placement is spending it on the one thing that can be bought. What we do, and what we will not do.

The line is whether the thinking is theirs and whether they can explain what is on the page. Restructuring, telling somebody an argument does not hold and what would fix it, overall direction on building an outline — all of that is supervision and none of it is doing it for them. Writing the whole outline down to paragraph level, or writing the sentences, is where it stops. A candidate who can defend it understood it. Where the line actually sits.

It reflects on the document, not on your judgment — because you were answering a different question. You assessed the scholarship. Review assesses conformity to institutional standards. Both can be right at once, and the second is where most late rejections come from. Knowing what that stage checks is the strongest protection available for the least effort. What School Review is actually checking.

Some institutions train doctoral supervisors. The unspoken presumption is that an experienced researcher is by definition an effective supervisor — that having been supervised, and having done research, is qualification enough. Unfortunately, those are different skills, and the people who appear to be managing are mostly managing the way you are: from memory, by feel, and privately hoping.

Yes — several, and almost nobody is handed them. There is a genuine literature on doctoral supervision as a craft. Better? The University of Michigan publishes a free guide, and there are seven or eight standard books — listed and annotated below. Any university library will have most of them.

They are performing competence for the same reason you are: because everybody else appears to have it. This is the same dynamic that runs through doctoral candidature itself, one rung up. The difference is that candidates eventually talk to each other about it and Chairs and supervisors generally do not.

Yes, particularly when the work is not progressing and neither of you wants to say so. Dread is usually a signal about an unspoken problem rather than about the person. Meetings become dreadful when they have no agenda except progress that is not happening. A meeting with a specific question in it is a different event from a meeting about how things are going.

Tell them what is wrong with the document, not how it compares to a standard they cannot see. "This is not good enough" gives a candidate nothing to act on and everything to feel. "The argument in chapter two does not support the claim in chapter five, and here is where it breaks" is the same judgment made useful. The kindest feedback is the most specific.

Separate the document from the person out loud, because they will not do it themselves. Candidates who have been through two rounds of revision routinely read every comment as evidence about their intelligence. Saying explicitly that this is a fixable fault in a draft — and that you have seen it before — costs you one sentence and changes what they hear. How rejection actually lands, and what it usually means.

It is more common than anyone admits and it is rarely about the comment that triggered it. Doctoral work runs for years under sustained uncertainty; the meeting is where the pressure becomes visible, not where it originates. You do not have to fix it, and you are not equipped to. Acknowledging it, pausing, and continuing is enough — and if it persists, your institution has support services and pointing at them is a legitimate act of supervision. What is normal in doctoral work, and what is not.

"Depth" is a word candidates cannot act on — name the missing move instead. Most often what it means is that they have reported the literature rather than argued with it: summarizing twenty studies instead of showing what they collectively establish, where they conflict, and what is left unresolved. Say that, and the note becomes a task. What a literature review is actually for.

Sometimes it is defensiveness. Often it is a candidate who does not understand the note and is arguing rather than admit it. The test takes a moment: ask them to state the objection back in their own words before responding to it. A candidate who can restate it and still disagrees is doing scholarship. A candidate who cannot restate it never understood the comment, and the argument was never about the point.

No, and by the end you should not. A completed doctorate means the candidate is the world expert in something quite narrow — the arrangement is working as designed when they know their method better than you. Your job is whether the method answers the question asked and whether the claims match what it can support. That judgment does not require matching their technical depth.

Say so, and ask them to explain it — that is a legitimate examination in itself. A candidate who cannot explain their own analysis in plain terms has a problem worth finding now rather than at the defense. If it is genuinely outside your competence, that is what committee composition exists for, and asking for a methodologist is a normal request rather than an admission.

You probably cannot prove it, and proof is the wrong target. Detectors do not detect AI use; they measure how statistically predictable text is, which is why disciplined academic prose scores badly. What you can examine is understanding. A candidate who cannot explain their own argument, sources, or method has a problem regardless of how the text was produced — and that is examinable. Why detector scores do not mean what they appear to.

Your instinct is correct. A 2023 Stanford study found seven GPT detectors misclassified over 61 percent of TOEFL essays written by non-native English speakers as AI-generated. A high score against a non-native speaker is close to uninformative, and acting on one has ended candidatures unjustly. What the research actually shows.

Ask about the work, not about the tool. Have them walk you through how a section was built, why a source was chosen, what they would change. This is unimpeachable because it is what supervision consists of anyway, and it produces a real answer whichever way it goes. Questions worth asking.

No, and treating it as policing is what makes it exhausting. The examinable question has not changed: does this candidate understand and own the work. That question was always the job, and it answers the AI question as a side effect. What AI is and is not useful for in doctoral work.

Yes, and your university most likely already tells them how. Capella's current guidebook names three categories of editorial help and instructs candidates to ask for client referrals and to check the person has handled complex academic material. Many bricks-and-mortar institutions maintain standing lists of editors. Recommending help is not lowering a standard — it is directing effort to the part of the work that is actually yours to assess. Where the line sits between help and substitution.