Where doctoral research actually stalls, what genuinely helps at each stage, and how to tell which kind of help you need.
PhD research support is help with the thinking and the technical work of doctoral research: framing a question that can be answered, choosing a methodology that fits it, collecting data that will bear the weight of the claims you want to make, and analysing it in a way you can defend. Good support leaves you able to explain every choice as your own — which is exactly what the viva asks for, and what makes the help worth having in the first place.
Very few doctorates come apart through lack of effort. They come apart at a small number of predictable joints, and almost always earlier than the point where the pain is felt. By the time a chapter is being rejected for the third time, the fault is usually two stages upstream.
These are the joints worth knowing about before you reach them.
The gap that is not a gap. “Nobody has studied this in my country” describes an absence, not a reason. An absence becomes a gap only when you can say what the field cannot currently explain, predict or decide because of it. Examiners press on this harder than almost anything else, because a thesis built on a described absence has no answer to so what?
Method chosen before the question is settled. A researcher decides on structural equation modelling because the department is known for it, or on interviews because the numbers feel intimidating, and then works backwards to a question that will justify the choice. Everything downstream inherits the mismatch, and it cannot be corrected later by better analysis.
Data that cannot answer the question asked of it. A cross-sectional survey asked to demonstrate causation. Twelve interviews asked to be representative of a population. A convenience sample asked to support a claim about a country. The fieldwork is done, the effort is real, and the data still cannot carry the claim.
Analysis that does not match the design. A model is run whose assumptions the data plainly violate; it returns a p-value; the p-value is reported. Nothing in the software objects. The objection arrives in the viva instead, or from a reviewer, when there is no longer time to collect anything else.
A write-up that cannot defend itself. Results are reported without the reasoning that connects them to the claims. Every number is correct and the argument still does not stand up, because the thesis never says why this evidence justifies this conclusion.
Doctoral research is not meant to be done alone, and getting help with it is ordinary. Supervision is help. So are methods courses, statistics clinics, writing centres and the colleague who reads your chapter on a Sunday. Working with someone outside your department is the same kind of thing — usually because you need a particular expertise at a particular moment, and waiting three weeks for a supervision slot is not an option.
What makes support useful is that you come out of it able to run with the work yourself. That is the point of it, and it is also what the viva will ask of you.
The one thing worth keeping in view is that the thesis has to remain yours — your argument, your data, your understanding of the analysis. Help that leaves you fluent in your own work is help you can lean on in the viva. That is the standard the work here is pitched at.
Research support is not one service. What helps at the proposal stage is useless at the analysis stage, and the help that rescues an analysis cannot repair a design. What follows is what typically goes wrong at each stage, and what useful help looks like there.
What goes wrong: a topic broad enough to be a field rather than a project, or a gap stated as an absence in the literature rather than a limit on what the field can currently explain.
What helps: narrowing the topic until it is answerable inside a doctorate, and stating the gap as a question with consequences — what changes, for whom, once it is answered.
What goes wrong: summary in place of synthesis. Fifty studies described one after another, with no position taken and no account of where they disagree or why.
What helps: organising the literature around the arguments in it rather than the studies, making the disagreements explicit, and showing how your question follows from them. For a systematic review, a protocol and a search strategy recorded well enough that somebody else could repeat it.
What goes wrong: questions too vague to imply a method; a theoretical framework named in the proposal and then never used again; a conceptual framework that is a diagram of variables with no theory behind the arrows.
What helps: tightening each question until the method it requires is obvious, and making the framework do actual work — explaining why these constructs, why these relationships, and what the theory predicts.
What goes wrong: a method inherited from a supervisor or a neighbouring thesis rather than derived from the question. Philosophical positions asserted in a paragraph that the rest of the design then contradicts.
What helps: working forwards from the question to the design it demands, and being explicit about what the chosen design can and cannot establish — particularly around causation, generalisation and time.
What goes wrong: a sample size decided by what feels achievable; instruments adopted without checking whether they have been validated in your setting; interview schedules that ask participants to confirm what you already believe.
What helps: justifying the sample against the claim you intend to make, checking instruments before they go into the field, and piloting — the cheapest stage at which a fatal problem can still be found.
What goes wrong: qualitative coding that stays descriptive, producing themes that are only categories of what was said. Quantitative analysis run without checking assumptions, or a model chosen because it is respected rather than because it fits.
What helps: for qualitative work, being pushed from description to interpretation until the themes say something. For quantitative work, testing assumptions first, and being told plainly when a result does not support what you hoped it would.
What goes wrong: findings and discussion that repeat each other; limitations written as an apology; a candidate who has never said their argument out loud until the day it is examined.
What helps: separating what was found from what it means, treating limitations as evidence you understand your own design, and a mock viva — because the first time you defend the thesis should not be the time that counts.
Most researchers arrive describing a symptom. The symptom is often not the problem, and buying help for the symptom wastes both money and the time you have left. This is the translation that comes up most often:
| What it feels like | What it usually is |
|---|---|
| “I don’t know which statistical test to use.” | The question has not been stated precisely enough to imply one. Sharpen the question and the test is usually obvious. |
| “My model won’t fit.” | Often the measurement model, not the structural one — or a structure the data were never going to support. |
| “My interviews aren’t saying anything.” | Coding has stayed at the level of description. Nothing has yet been raised into a concept. |
| “My supervisor keeps rejecting the chapter.” | Usually the argument rather than the writing. Editing the prose will not fix it. |
| “I have data but no findings.” | The analysis is answering a different question from the one the study asked. |
| “I’m too far in to change anything.” | Sometimes true. More often the claim can be narrowed to one the existing data genuinely supports. |
If you are not sure which of these you are looking at, describing the problem to somebody who has examined doctorates will usually locate it faster than another month of reading.
You describe where the research has got stuck, in whatever state it is in. You are matched with a PhD in your own field — not a generalist, because methodological judgement does not transfer cleanly between disciplines. The first conversation costs nothing and is diagnostic: it exists to establish what the actual problem is, which sometimes means being told you do not need the service you asked for.
If confidentiality matters, an NDA can be signed before any detail is discussed. Your data, your ideas and your authorship remain yours throughout, and the work is done so that you finish it understanding it — which is the part that pays off when you are asked to defend it.
Very. Supervision is help, and so are methods courses, statistics clinics, writing centres, proofreaders and the colleagues who read your drafts — all of them ordinary parts of doctoral study. Bringing in outside expertise is the same idea, usually because you need something specific at a particular moment. Universities publish their own guidance on what to acknowledge, and it is worth a glance; declaring the help you had is straightforward and takes the question off the table.
The kind that leaves you able to do the thing afterwards. Being taught a method, having a design reviewed before you collect data, getting an analysis checked and explained, or having a draft read the way an examiner will read it — each of those puts you in a stronger position in the viva than you were before. The test worth applying to any help you are offered is simply whether you understand the work well enough to talk about it comfortably.
Yes, and it is one of the most common reasons researchers get in touch. It works best done with you rather than handed back finished: the method explained, the assumptions and their checks set out, and the output walked through until you can read it yourself. That way the result is defensible and you can answer questions about it — which matters, because examiners ask about analysis more than almost anything else.
It follows from your question rather than from preference. Questions about meaning, process and how people account for their experience call for qualitative designs. Questions about magnitude, frequency, difference and prediction call for quantitative ones. Questions that need both a measured effect and an explanation of it call for a mixed design — provided you can say specifically how the two strands will be integrated, since a mixed-methods study that never integrates is two thin studies stapled together.
Usually yes, though the honest answer sometimes is that the design cannot be repaired after data collection. What can almost always be repaired is the claim: a study that cannot support a causal conclusion may still support a well-defended associational one. Examiners are far more forgiving of a modest claim argued carefully than an ambitious claim the design never justified.
Yes. An NDA can be signed before any detail is discussed, and intellectual property and authorship remain yours. Unpublished data and unsubmitted theses are shared with no one.
Tell us what stage you are at and what is not working. You will hear back from a PhD in your own field, and the first conversation costs nothing — including when the answer is that you do not need us.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
