How to know when a question is finished, the difference between aims, objectives and questions, and when a hypothesis is required at all.
A research question is finished when it implies the method needed to answer it. If two entirely different designs would both suit your question, it is still too broad — and almost every problem that appears later in a doctorate can be traced back to a question that was never made specific enough. Aims, objectives, questions and hypotheses do different jobs, and using them interchangeably is one of the commonest sources of confusion in proposals.
The most useful check on a research question takes ten seconds. Read it, and ask what you would have to do to answer it. If the answer is obvious, the question is specific enough. If several quite different studies would all be reasonable responses, it is not.
“What are the challenges facing small businesses?” — could be a survey, interviews, document analysis or secondary data, in any sector, any country, any time frame. Not a research question yet.
“How do owners of small manufacturing firms account for their decisions to postpone capital investment during periods of policy uncertainty?” — the method is now close to determined: this needs accounts from people who made those decisions, which means interviews analysed qualitatively.
Narrowing feels like losing something, which is why people resist it. In practice a narrow question produces a study that answers something convincingly, and a broad one produces a study that answers nothing convincingly. Examiners are far more comfortable with the first.
A second check worth applying: could the answer come out other than you expect? A question whose answer you already know is not a research question, and a study that could only have produced one result has not tested anything.
Four terms, four jobs, and proposals routinely muddle them.
| Term | What it does | Typical form |
|---|---|---|
| Aim | The overall purpose — what the study is for | One sentence: “This study aims to…” |
| Objectives | The steps that achieve the aim | A short list of doable actions |
| Research questions | What the study will answer | Actual questions, with question marks |
| Hypotheses | Specific testable predictions | Statements, not questions |
Two things to get right. Objectives are things you do — examine, compare, evaluate, develop — and each should be achievable and checkable. An objective nobody could tell you had completed is not an objective. And questions and hypotheses are alternatives, not a sequence: you do not need both, and having a hypothesis for every research question in an exploratory study is a sign the format was copied rather than chosen.
Keep the number small. Three or four research questions is common; seven suggests either several studies or a set of questions that have not been organised into a hierarchy of one main question with sub-questions beneath it.
The opening word of a question signals what kind of study follows, and a mismatch between wording and design is quickly spotted.
Qualitative questions typically begin with how or what: how do participants understand, experience, negotiate, account for. They are open, they do not presume a relationship, and they do not name variables. Asking “what is the effect of X on Y” and then running interviews is a mismatch, because effect language promises measurement.
Quantitative questions name variables and the relationship being examined: to what extent, is there a difference, does X predict Y. They should specify the population, and where relevant the time frame.
Mixed methods questions need something extra that is routinely missing: a question the integration answers. If you have one quantitative question and one qualitative question and nothing that requires both, you have two studies. A third question — how the accounts explain the measured pattern — is what makes it one.
Some fields use structured formats for framing questions, particularly in health and systematic review work, which prompt you to specify population, comparison, outcome and context. They are useful scaffolding if your field uses them, and unnecessary if it does not.
A hypothesis is a specific, testable prediction derived from theory. It is required when your study is testing something derived from an existing explanation. It is not required — and is often actively wrong — in exploratory or qualitative work, where writing hypotheses signals that a template was followed rather than a design chosen.
What makes a hypothesis defensible:
On the null hypothesis: it is the formal statement of no effect that your statistical test actually evaluates, and conventions on whether to state it explicitly vary by discipline. Follow your field. What matters more is understanding that failing to reject it is not the same as demonstrating no effect — which is why confidence intervals and effect sizes belong in the reporting.
One practice to avoid: writing or adjusting hypotheses after seeing the results and presenting them as predictions. Exploratory analysis is legitimate and should be labelled as exploratory; the same work presented as confirmatory is not.
Common problems and the usual fix:
Read it and ask what you would have to do to answer it. If the method is close to obvious, the question is specific enough; if several quite different studies would all be reasonable responses, it is still too broad. A second check: could the answer come out other than you expect? A question whose answer you already know has not been asked properly.
The aim is the overall purpose, usually one sentence. Objectives are the steps that achieve it, phrased as things you do — examine, compare, develop — and each should be checkable. Research questions are what the study will answer, phrased as actual questions. Hypotheses are specific testable predictions, and they are an alternative to questions rather than an addition to them.
Usually three or four. More than that generally means either several studies packed into one, or a set of questions that has not been organised into one main question with sub-questions beneath it. Each question also has to be answered in your findings, so every additional question is additional work and additional word count.
Only if you are testing predictions derived from existing theory. Quantitative confirmatory work usually needs them; exploratory and qualitative work usually should not have them, and writing hypotheses for an interview study signals that a template was followed rather than a design chosen. Where you do state them, say where each came from — a prediction without a stated basis is a guess.
Usually beginning with how or what, and staying open — how participants understand, experience or account for something. Avoid effect and impact language, which promises measurement your design cannot deliver: asking “what is the effect of X on Y” and then conducting interviews is a mismatch examiners notice immediately. Do not name variables or presume a relationship.
Yes, and most projects do — understanding develops and the study changes with it. What causes trouble is questions written at proposal stage and never revisited while the study moved on, leaving a thesis that answers different questions from the ones it states. Revisit them before submission and make sure what you actually did matches what you claim to have asked.
Almost every later problem in a doctorate traces back to a question that was never made specific enough — and it costs an afternoon to fix now. Send yours and a PhD in your field will push on it.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
