The first ninety days of a new PhD scholar

New doctoral scholars arrive having been excellent at something that is about to stop working. They are good at coursework: bounded problems, known methods, a marker who can tell them whether they got it right. Research has none of those properties, and the transition is not gradual. It happens in the first term, and how it goes tends to set the pattern for the whole degree.

The supervisor's leverage is enormous here and almost entirely front-loaded. A habit established in month two is nearly free to set and expensive to change in year three.

Say the quiet things in week one

Every lab runs on norms that feel obvious from the inside and are invisible from the outside. New scholars spend an astonishing amount of energy inferring them, usually by getting one wrong and feeling terrible about it.

Write them down and hand them over. The list is shorter than you think:

• How quickly you reply to email, and what to do if you have not replied.
• Whether they can drop in, or should always book.
• What you expect them to bring to a supervision meeting.
• Working hours — particularly whether your late-evening emails carry any expectation of a late-evening reply. They almost certainly do not, and almost every new scholar assumes they do.
• Who else to ask when you are unavailable.
• What authorship looks like in your lab and when that conversation happens.

That last one is uncomfortable to raise this early and much worse to raise late. A scholar who learns the norm in year three, after it has already been applied to a paper they worked on, will not experience it as a norm.

Give them something finishable

The most common first-term error is starting a new scholar on their thesis problem. It is the obvious thing to do and it is usually wrong.

A scholar three months in cannot yet estimate effort, does not know when something is good enough, and has never taken a research question through to any kind of end. Starting them on a four-year problem means their first experience of research is an unbounded task with no completion signal — which teaches precisely the wrong lesson about what research feels like.

Give them something with an end. Replicate a result from a paper you both care about. Take a small piece of a senior scholar's dataset and produce one clean figure. Write a short review of five papers and present it to the group. The work should be real and it should be finishable in six to ten weeks.

What this buys is not the output. It is that the scholar has now been through the full cycle once — question, method, obstacle, result, write-up — at a scale where they could hold all of it in their head. Every subsequent piece of work is calibrated against that experience.

Teach reading explicitly

New scholars read badly, and nobody tells them, because reading feels like something a doctoral student should already be able to do.

They read linearly, start to finish, at uniform depth. They finish a paper unable to say what its central claim was, but able to recall the third figure. They do not distinguish between the paper they need to know exists, the paper they need the result from, and the three papers they need to understand line by line.

This is teachable in about twenty minutes and almost nobody teaches it. Take a paper you both have read, and walk through how you actually read it: what you looked at first, which parts you skipped, where you stopped to check a claim, what you wrote down. New scholars are frequently astonished that experienced researchers do not read every word, and relieved to hear it.

Then ask for a written summary of the next thing they read — half a page, what it claims, what supports it, why it matters to us. Half a page is enough to expose whether they extracted the argument or just the surface.

Expect the confidence collapse, and name it in advance

Somewhere between month three and month nine, most doctoral scholars go through a sharp drop in confidence. The bounded problems are gone, the first real obstacle has not yielded, everyone else appears to know what they are doing, and the scholar concludes the admissions committee made a mistake.

This is so nearly universal that the useful move is to describe it before it happens. Tell them, in the first month, that at some point in the next year they will be fairly sure they are not good enough for this, that it happens to almost everyone, and that it is a stage rather than a verdict.

This costs one conversation and it changes the experience substantially. A scholar who was warned can recognise what is happening. A scholar who was not will read it as private evidence about themselves, and is much more likely to go quiet at exactly the point where going quiet is most damaging.

Set the record-keeping habit now

Whatever system your lab uses for meeting records, actions and progress, the new scholar should be in it from the first meeting — not from the point where things get serious.

The reason is not administrative. It is that a scholar who has kept a record from week one has, by month nine, something to look back at when the confidence collapse arrives. The single most effective response to "I have achieved nothing" is a list of what they have actually done, in their own words, dated. Almost nobody can reconstruct that from memory, and the reconstruction attempt tends to confirm the fear rather than dispel it.

It is also the only way you will be able to tell, later, whether a struggling scholar has been struggling for three weeks or for two terms.