A research question is the single most consequential sentence in a dissertation. It decides your method, your data, your analysis and the shape of your findings chapter — which is why supervisors spend so long on it and why a vague question causes trouble for months afterwards.
Enter your topic and this tool phrases it as six different questions, one for each recognised question type, and names the design each one commits you to. The point is not to hand you a question but to make the choice visible, so you pick the one that matches what you actually want to find out.
Students often choose a method first — "I want to do a survey" — and then bend a question to fit it. That is backwards, and it shows in the results chapter, where the data cannot quite answer what was asked.
The direction runs the other way. A descriptive question needs measurement of one group. A comparative question needs the same measurement in two groups. A causal question needs a comparison across conditions. An exploratory question needs accounts in people's own words. Settle what you want to know, and the method follows almost mechanically.
"How does social media affect students?" cannot be answered — which platforms, which students, which effect, measured how? "How many minutes per day do second-year undergraduates at one university spend on Instagram?" can be answered but is unlikely to interest anyone.
A workable question sits between the two, and it gets there by naming three things: the population, the setting, and the specific variable or experience of interest. Adding a time period sharpens it further. If your question does not name all three, it is not finished.
"What effect does X have on Y?" is the most attractive question and the hardest to support. A defensible causal claim needs a comparison between conditions that differ only in X — which in practice means an experiment with random allocation, or a quasi-experimental design with a strong argument about why the groups are otherwise comparable.
Most student projects can obtain neither. That is not a failure; it is a reason to ask a relational question instead and to write "is associated with" where you were tempted to write "causes". Overclaiming causation from cross-sectional data is one of the most reliable ways to lose marks in a discussion chapter.
Most dissertations carry one main question and two to four sub-questions that decompose it. Each sub-question should be answerable in its own right, and together they should answer the main one with nothing left over and nothing extra.
Your objectives are the same content phrased as purposes, so they must line up one-to-one with the questions. A reader should be able to pair question 2 with objective 2 without guessing. Use the research objective generator once your questions are settled.
Specific: does the question name the population, the setting and the variable? Answerable: can you actually obtain that data, with the access, the time and the ethical approval you have? A perfect question you cannot get data for is worthless.
Arguable: could the answer plausibly go more than one way? A question whose answer is already known is a literature review, not research. Scoped: can you do it in the word count and the months available? Supervisors see far more projects that were too ambitious than too modest.
One sentence. If it needs two, you probably have a main question and a sub-question — separate them and label them as such.
One main question, usually with two to four sub-questions. More than that normally signals a scope too wide to finish.
A question asks; a hypothesis predicts. Quantitative studies often state both — the question frames the enquiry and the hypothesis states the expected result in testable form.
Early on, yes, and most projects do. Once data collection has started, changing it usually means the data no longer answers the question — so settle it before you collect anything.
No. They are built from templates using the words you enter, which is why they contain bracketed placeholders for you to fill in. Nothing about your topic is invented.