A research question is the one question your thesis sets out to answer. The rest of the thesis depends on it. You can test a research question before you build on it, and you can test it again on a finished draft.
Duke's writing program puts it in one line: “A research question guides and centers your research.” In practice it decides what you read, what you can leave out, and what counts as an answer. Someone who knows your question should be able to predict roughly what your chapters will do.
George Mason University's Writing Center lists what such a question should be. Two of its criteria matter most here. A focused question means “it is narrow enough that it can be answered thoroughly”. An arguable one means “its potential answers are open to debate rather than accepted facts.”
Run these five tests. Each one has a sign that shows up in the draft when the question fails it.
| Test | Ask yourself | Sign that it fails |
|---|---|---|
| Can't be looked up | Could someone answer this with one search or one report? | The chapter that answers the question mostly summarizes one source |
| Clear scope | Who, where, when, and what exactly? | Your literature review keeps growing and never feels finished |
| Evidence exists | What material would count as an answer, and can you get it? | You changed data sources halfway through and kept the same question |
| Method fits | Can your method produce this kind of answer? | The question asks about an effect, and the data come from a single survey at one point in time |
| Answer isn't a given | Could the answer surprise you? | The introduction already states the conclusion as fact |
Here is a hypothetical example. It shows the reasoning and is not a question to copy, since yours has to come from your own topic and needs your advisor's sign-off.
Start with: "How does social media affect teenagers?" Which platform? Which teenagers? Which effect? The word "affect" also promises a cause, and a survey taken once can show an association at most.
One narrower version: "Among high school freshmen at three public schools in Ohio, is daily time on short-video apps associated with self-reported hours of sleep?" You can now name the sample, the two variables, and the method. You can also say what the study will not show: that the apps cause the sleep pattern.
Not every field uses all three. Some programs ask for one main question plus sub-questions, and some theses have no hypothesis. Your handbook or advisor will tell you what your program expects.
| Term | What it is | Example form |
|---|---|---|
| Research question | What you set out to find out | "Is X associated with Y among Z?" |
| Hypothesis | The answer you expect and can test | "Students with more X report less Y." |
| Thesis statement | The answer you argue for, usually once you have it | "X matters less for Y than earlier studies assumed." |
Esmond evaluates every assignment before you submit. It never writes it.
Esmond assumes your advisor has already signed off on your research question. It looks at how your thesis answers the question and comments where a passage clearly doesn't, so you can decide what to change.
George Mason's Writing Center says it should not be, and suggests that questions often begin with how or why. Some fields do test yes-or-no hypotheses, so check what your discipline expects.
There's no set number. Each extra question needs its own evidence and its own space. Your advisor can tell you what fits your project.
Questions can shift once the data come in. Talk to your advisor first, then make the new wording consistent across every chapter.
Your advisor and committee. The tests on this page show you what they may probe; they don't replace that approval.
Read on November 3, 2026.
Esmond is not taking orders on this site yet.