Examples of Research Questions, With Good and Bad Versions

An example of a research question is the question a paper is built to answer. A useful one names the people or the cases, the comparison or the experience, and the kind of answer the study can give. A question that only names a topic does not do that work.

The examples of research questions in this guide were written for the guide. They are not questions from published studies, dissertations, or writing-center handouts. Copy the shape, then use the people, the setting, and the concepts from your own draft. Where a real question would name a count, a dose, or a measurement from a study, these models leave that detail out. There is no study behind them.

When a question already asks the right thing but sounds stiff, Rephrase AI can rephrase that question. Rephrase AI is designed to preserve the meaning. Compare the new wording with the people, the comparison, and the outcome in your notes before you keep the wording. Your instructor's rules on AI use still apply.

1. What a research question is for

The George Mason University Writing Center calls a research question the question around which you center your research. George Mason says the question should be clear, focused, concise, complex, and arguable. Clear means a reader can see the purpose without a second explanation. Focused means the question is narrow enough to answer thoroughly in the space the writing task allows. Concise means the fewest words that still carry that focus. Complex means the question is not settled by a simple yes or no, because the answer needs synthesis and analysis. Arguable means the possible answers are open to debate rather than accepted facts. The handout was last updated on August 8, 2018.

George Mason also says the question should fit the discipline. A question that fits biology is different from a question that fits political science or sociology. For a course other than first-year composition, the handout suggests discussing the question with the professor.

The same handout gives the question a job in the writing process. The question provides a path through the research and the writing. The specificity helps a writer avoid an all-about paper and work toward a specific, arguable thesis. The opening that carries that thesis has its own shape. See introduction paragraph examples.

Matthew DeCarlo's open textbook adds a plainer test for social work students. The research question is written as a question. A topic name is not a research question. DeCarlo's summary list says a good question is clearly written, is not answered with yes or no, has more than one plausible answer, considers relationships among multiple variables, is specific about the concepts, and includes a target population. Later in the same chapter, a qualitative question may name a single concept instead. The summary list is the stricter checklist. The qualitative exception is described below. The book home for this Pressbooks second edition says the book is still in beta. The question rules in this section follow the chapter.

The Open University course on research questions in international relations was first published on February 24, 2025. The course says a well-designed question defines a focused area of research that is manageable within the project. Later, the course tells students in that field to push the question past description. A descriptive question can help a writer start thinking. The question the course wants for the project should explain, explore, evaluate, or compare, so the answer has to analyze rather than only list what exists.

Ohio State's Quantitative Methodology Center says the nature of the question specifies the purpose of the research. The page names three natures for that purpose. They are descriptive, comparative, and relationship-based. The page also says one primary question is the advisable focus. Several questions can complicate the design, the statistical analysis, and the feasibility of the study.

The question is not the method, and the question is not the finding. DeCarlo says the question informs the design. A survey, an experiment, and an interview study answer different questions. For the section that reports how the study was done, see methods section examples. For the one or two sentences that carry the purpose inside a summary, see abstract examples.

2. How long a research question should be

Use the limit in the assignment or the journal. The guides checked for this article do not share one word count, and they do not share one sentence count.

George Mason's narrowing test is the practical cut. The handout sets an unfocused question about the environmental effect of global warming beside a focused question about one effect, one place, and one animal. Those two questions are not reprinted here. The lesson is the narrowing. Name the people or the cases, the place or the setting, and the outcome or the experience. Then stop adding topics.

Concise does not mean vague. The fewest words that still name those parts will often be one sentence with two clauses. A causal question has to name the comparison. Cutting the comparison to save words turns the question into a topic.

Ohio State's single-question advice is about the primary question. The page says extra questions can complicate the design, the analysis, and the feasibility. For a class paper, smaller questions can stay in the notes. The sentence the paper is built to answer should still be the one primary question.

DeCarlo treats some short questions as unfinished for a different reason. The chapter flags the words factors, causes, effects, outcomes, effective, effectiveness, useful, and efficient. Each of those words asks the reader to guess which concept the writer means. The chapter also flags "and so forth," because a study cannot cover every part of a social problem. Those flags apply when the project is trying to explain and the literature already names the concepts. DeCarlo says the same open wording can fit an exploratory question when the researcher does not yet know which concepts matter. That exception is narrow. The chapter says exploratory research fits only when little scholarly research exists, and a literature review should come before the open wording stays.

The Open University activity makes the same point with a clock and a definition. An evaluative question can name the actor and the event and still be too large if the question has no timeframe. The question also has to say what success means. A comparative question can name two countries and still be too vague if "foreign policy" is the whole object of comparison. The course does not turn those comments into a word count. The course turns them into a scope.

A class assignment may want one sentence. A grant proposal, on the Ohio State account, wants the primary question at the end of the introduction, with the population and the intervention visible. Neither pattern is a universal length. Match the task in front of you.

3. How a research question is structured

Start with the job the question has to do. The guides checked here do not share one list of types. Two guides can use the same word for different jobs. Use the list your assignment names, and do not treat the names as interchangeable.

The naming collision is worth keeping visible. Toronto Metropolitan's third edition says quantitative questions can be explanatory, descriptive, or comparative, and the course defines explanatory as relationship-based questions that aim at cause and effect. That phrase does not match Ohio State's relationship-based example, which asks what the relationship is. Ohio State's later note on a relationship-based hypothesis says the hypothesis does not necessarily mean one variable causes the change in the other. Trochim keeps those jobs apart. Relational work looks at a relationship. Causal work asks whether one variable causes or affects an outcome.

William M.K. Trochim's Research Methods Knowledge Base, on the page hosted by Conjointly, treats the three jobs as cumulative. A relational study assumes the researcher can first describe each variable, by measuring or observing that variable. A causal study assumes the researcher can describe the cause and the effect, can show that they are related, and only then can support the cause. Trochim says causal studies are probably the most demanding of the three.

A public-opinion example on that page shows the stack without a lab. A descriptive question asks what share of people would vote for one party or another. A relational question asks how that share differs for men and for women, which is a relationship between gender and voting preference. A causal question asks whether a recent advertising campaign changed the preference. The campaign is the cause in that example. The preference is the effect. The page does not say a relationship is enough to claim the cause.

Descriptive questions and analytical questions

The Open University section on types still lists descriptive as a real type. The purpose in that table is to describe the features, dynamics, or processes of a phenomenon. The sample stem is a question about the properties of a phenomenon. The next section of the course then tells international relations students not to leave the project there.

That next section says descriptive questions ask what a phenomenon looks like, what the phenomenon contains, and which processes the phenomenon involves. The course says those questions can be useful at the start. The course also says they do not explain how a situation came to be, and they do not say why a process produces a particular outcome. The ambition for the research question, in that course, is analysis. The verbs the course names are explain, explore, evaluate, and compare.

Read those two pages with Trochim, not instead of Trochim. A methods class that asks what exists can still need a descriptive question. A politics project that asks for an argument should not stop at a list of features. The assignment decides which page governs.

Exploratory and evaluative questions

The Open University types table gives exploratory questions the job of understanding a topic better, often where existing research is limited. The sample stem asks which factors might matter for a political problem. The same table gives evaluative questions the job of assessing an intervention or an explanation. The sample stem asks about the impact of a policy.

The course activity on a good research question then shows how an evaluative question can still fail. One question about whether a response succeeded is called fairly good, because the question names an actor and an event. The discussion says the question would be stronger with a timeframe, a definition of success, and a statement of which responses are included. Without those parts, the project can become too large. A comparative prompt that only says to compare two countries' foreign policies is called weak, because the comparison stays descriptive, the reason for the pair is missing, and the object of comparison is too vague.

DeCarlo draws a similar line for exploratory wording. "What factors influence" can be a fair exploratory question when the concepts are still unknown. The same sentence is a working question, not a finished one, once the literature already names the causes. The chapter's watch words are the test. If the literature can replace factors or effective with a named concept, the question should name that concept.

Qualitative and quantitative questions

Toronto Metropolitan's chapter separates a second cut from the type list above. Quantitative questions and qualitative questions are the two general kinds. Quantitative data can be counted, measured, and expressed with numbers. Qualitative research uses data from first-hand observation, interviews, questionnaires on which people write descriptively, focus groups, recordings in natural settings, documents, and artifacts. Those data are generally nonnumerical.

The same chapter says qualitative questions aim at meaning or understanding, and they ask something that cannot be measured with specific numbers and statistics. Words that often appear include lived experience, personal experience, understanding, meaning, and stories. The chapter's key takeaway is that a good question, of either kind, cannot be answered with a simple yes or no.

DeCarlo makes the wording difference more specific. Qualitative questions often seek experiences, understandings, and meanings rather than a general cause that would hold across a population. They are often more general and more openly worded. They may include one concept, though many include more than one. They often ask about lived experience, personal experience, understanding, meaning, and stories. Quantitative questions, in the chapter's contrast, ask how one variable is connected to another and usually name more than one concept.

DeCarlo also says qualitative questions can change during the study. Qualitative work follows what participants say and do. A researcher may start with one rule or one program and learn that participants treat a different problem as the real one. The chapter calls that shift normal. The chapter says the same openness does not fit quantitative hypothesis testing. A researcher who revises the hypothesis to match the finding can no longer be wrong.

The Open University types page adds one limit to this contrast. The course says each type can be asked in more than one way. A comparative question might compare causes across a large dataset, or compare how a kind of political language emerged. The type names the job. The type does not, by itself, choose numbers or interviews.

A clinical question

Ohio State presents PICOT as a practical frame for a more specific question, after the broader FINER check. The page summarizes FINER from Hulley and colleagues in 2007. This guide did not read that book. The items below are the items on the Ohio State page.

PICOT, on the same page, names the population, the intervention for an intervention study, the comparison group, the outcome, and the time used to assess the outcome. The page says a tighter population can strengthen internal validity and can limit how far the findings travel. A broader population may look more like ordinary practice and can increase bias. The page also says an inadequately formulated question can damage the design.

The University of Virginia Claude Moore Health Sciences Library uses PICO, with an optional T for the time period, to build a searchable clinical question before the literature search. The library says a question may not have every piece. The therapy template asks how an intervention, compared with a comparison, affects an outcome within a time period. The meaning template is different. The meaning template asks how a group with a stated condition perceives an experience during a time period. Those templates are shapes. The medical scenarios on that page are not reprinted here.

DeCarlo's feasibility list is the student version of the same worry, and the list is not the same list as FINER. The chapter asks whether the student can reach the data, get consent, avoid direct harm, dual relationships, and confidentiality breaches, do the work competently, and finish in the time and with the money available. A question that fails that list should change, even when the sentence sounds specific.

A question and a hypothesis

Ohio State says that, in quantitative studies, reviewers generally expect hypotheses rather than research questions, and that the two can still be used together because they do different jobs. A hypothesis, on that page, states the researcher's prediction. The page says the question and the hypothesis should be established before the study starts. Building either one from a database already in hand can lead to many comparisons and to findings that are chance. The page also says these are general guidelines, not strict rules, and that a writer who skips the hypothesis should ask a supervisor.

The page is not a single rule about which sentence comes first. One passage says the primary question should originate from the hypothesis rather than from the data. A later passage says hypotheses derived from the question can sit alongside the question. This guide does not flatten those two passages into one order. The shared point is the timing. Set the question and any prediction before the analysis, and do not fit them to a result you already have.

A qualitative project may not have that prediction. DeCarlo's point about a shifting question applies there. Say in the paper that the question changed, and say what the participants turned the inquiry toward. Do not describe a revised quantitative hypothesis as if the revised hypothesis had been the plan.

What a weak question usually does

Read together, the guides point to six recurring problems. The right-hand column is the revision this guide recommends. The column is not a new rule from any one source.

George Mason's own unclear example fails three of these tests at once. The handout says the unclear version does not specify which sites are in view, does not say what kind of harm is meant, and treats the harm as already accepted. The clearer version names the sites, the privacy harm, and the users. That pair is not reprinted here. The pattern is the one to copy.

A question can begin with "does" and still pass the complex test, if the rest of the sentence names a comparison and an outcome the study has to produce. "Does a written reply rule work?" fails because "work" is undefined. "Does a written reply rule reduce one named outcome, compared with no rule, over one term?" is a causal shape. The answer is a result, not a fact a short search already holds. George Mason's warning is about the simple version. Toronto Metropolitan's warning is about a simple yes or no. Neither warning is a ban on the word does.

4. Examples of research questions

Each question below was written for this guide. None of them is a question from a published paper. A real study would name the city, the course, the board, the instrument, or the count from the project. These models leave those names out.

Descriptive questions

• What share of evening bus trips in one city end at a grocery store?
• How often do students in one first-year lab record the hour they measure plant height?
• What percentage of invited students in two course sections finish a written questionnaire about the weekly grocery trip?
• How many daytime buses stop inside neighborhoods that have no supermarket inside their borders?

Comparative questions

• How do grocery-trip times differ between neighborhoods with a supermarket inside the border and neighborhoods without one?
• How do reply times differ between course sections that use a written reply rule and sections that do not?
• How do final-exam scores differ between first-year statistics sections taught with a flipped class and sections taught by lecture, in one department and one term?
• What similarities and differences appear in the way two city councils describe the same cut to evening bus service?

Relational questions

• What is the relationship between daytime bus frequency and grocery-trip time in neighborhoods with no supermarket inside the border?
• Is weekly study time associated with the share of assignments turned in on time in one first-year course?
• What is the relationship between hours of sleep on the night before an exam and the score on that exam, among students in one psychology course?

These relational questions do not say that one variable causes the other. Trochim's stack says a cause is a further claim. A study that only measures the two variables together should keep the relational verb.

Causal questions

• Does a written reply rule reduce the share of non-urgent messages marked urgent, compared with no written rule, in two sections of the same course over one term?
• Does a longer daily light cycle change basil seedling height, compared with a shorter cycle, when the soil, the pots, and the room stay the same?
• Does a phone call within two days of discharge reduce unanswered medication questions, compared with no call, among patients of one clinic over one month?

A causal question names the comparison. Without the comparison, the verb "does" hides the design. The models still omit the counts a real trial would print, because this guide has no trial.

Exploratory questions

• What do staff who answer customer messages treat as urgent when the written rule does not name that case?
• Which parts of a weekly grocery trip do riders in one neighborhood describe as the hard part?
• What might explain why one neighborhood's evening buses are full and the next neighborhood's buses are not, before a cause is chosen?

Use this open shape when the literature does not yet name the concepts, which is DeCarlo's condition. Once the literature names them, replace the open wording.

Evaluative questions

• Does the written reply rule reduce after-hours messages, using the definition of reduction the team wrote down before the term?
• For adults in one clinic who were told to add a daily walk, does a written walking plan increase the number of walks recorded over four weeks, compared with a verbal suggestion only?
• To what extent does one city's evening-bus change meet the access goal the council stated in the year of the vote?

The outcome and the time window are what keep these questions from asking whether something "works."

Qualitative questions

• How do riders in neighborhoods without a supermarket describe the weekly grocery trip?
• What does a written reply rule mean to students who send most of their questions at night?
• How do parents who use one neighborhood food pantry describe the trip to a store when the bus is the only way to get there?

Quantitative questions

• What percentage of invited students in two course sections return the grocery-trip questionnaire?
• How large is the difference in exam scores between the flipped section and the lecture section?
• How many walks do adults record over four weeks under a written plan, and how many do they record under a verbal suggestion?

The quantitative lines ask for a count, a share, or a difference. They do not ask what the experience means. A project can ask both, in that order, only when the design collects both kinds of data. This guide does not treat two questions as one method.

Questions by field

These lines were written for this guide. The field label is the course the shape would fit. The label is not a claim that every paper in the field uses that shape.

The literature question is not a methods question. George Mason's point that the question must fit the discipline is the reason that question sits in the table. The river example is the same made-up essay used in the abstract guide on this site. The example is not a published novel. An interpretive question can still be clear, focused, and arguable. The question names the text's image, the choice, and the span of scenes. A paper would then defend a reading. The question does not pretend to measure a cause.

If a line is accurate but stiff, rephrase these questions in your own words with the public paraphrasing tool. Keep the people, the comparison, the outcome, and any numbers from your notes. The tool changes wording. The tool does not choose a research question.

A weak question and a revision

Weak version

What is the impact of buses on people?

Revision

What is the relationship between daytime bus frequency and grocery-trip time in neighborhoods with no supermarket inside the border?

Why this example works. The weak version names a topic, says impact without a design that could show a cause, and never says which people or which outcome. The revision names two variables, names the neighborhoods, and asks for a relationship. A later study could ask a causal question only after the comparison and the design are real.

Weak version

Does the written reply rule work?

Revision

Does a written reply rule reduce the share of non-urgent messages marked urgent, compared with no written rule, in two sections of the same course over one term?

Why this example works. "Work" is one of DeCarlo's watch words. The revision names the outcome, the comparison, the people, and the term. A reader can see what would count as an answer.

Weak version

Team messages.

Revision

How do students who send most of their questions at night describe a written reply rule?

Why this example works. The weak version is a topic, which DeCarlo says is not a question. The revision is a qualitative question about meaning, with a group and a situation. The revision does not claim that the rule causes a grade change.

A last check before you submit

• The line is a question. A topic name is not finished.
• A reader can see the people, the place, and the outcome without a second explanation.
• The verb matches the design. Relationship stays relationship until a comparison can support a cause.
• Success words are defined. Effective, useful, and "work" name an outcome and a time window.
• Open wording still matches the literature. Factors and effects stay only when the concepts are not yet known.
• The question fits the field. A biology comparison and a literary reading do not share a template.
• Any prediction was set before the results. A quantitative hypothesis does not move to match a finding.
• A rephrased sentence still matches the notes. People, comparisons, outcomes, and numbers survive the wording change.

FAQ

What is an example of a research question?

What makes a research question good or bad?

How long should a research question be?

What is the difference between a descriptive research question and a causal research question?

What is the difference between a qualitative research question and a quantitative research question?

What is the difference between a research question and a hypothesis?

Where should a research question appear in a paper?