What Is the Difference Between a Research Question and a Hypothesis?
At a glance
- 1. Ask what you want to know
Write the research question as a focused, answerable question about a topic, population, or relationship.
- 2. Predict only when justified
For a hypothesis-driven study, state a testable expected result or relationship.
- 3. Make terms measurable
Specify the population, variables, and measures so the question and hypothesis can guide a study.
- 4. Check your study type
Exploratory and many qualitative studies may use research questions without stating hypotheses.
A research question and a research hypothesis are related, but they do different jobs. The question sets the study’s direction; a hypothesis states what result you expect, when the study is designed to test a prediction. Some projects use both, while others use only a research question.
The difference in one table
| Research question | Research hypothesis |
|---|---|
| Asks what the study will investigate | Predicts an expected answer or relationship |
| Usually written as a question | Usually written as a declarative statement |
| Can be open-ended or exploratory | Should be specific and testable with evidence |
| Common in qualitative and exploratory work | Common in studies testing predictions or theories |
A question might ask, “Is sleep duration associated with quiz performance among first-year university students?” A corresponding hypothesis could predict that students who sleep longer will have higher quiz scores. The hypothesis narrows the expected finding; it does not guarantee that the finding will occur.
What makes a research question useful?
A good research question identifies what you want to learn and sets reasonable boundaries. Depending on the project, those boundaries may include a population, setting, time period, variables, or experience. The question should be answerable using the evidence and methods available to the researcher.
For instance, “Does sleep affect students?” is too broad. “Among first-year university students, is average nightly sleep duration associated with scores on weekly quizzes during one semester?” is more focused. It specifies a population, a possible relationship, and a context. It also avoids claiming that sleep causes better scores: an observational study can establish an association but may not establish causation.
In qualitative research, the question may ask how people experience or understand something, such as “How do first-year students describe the effect of sleep habits on their studying?” This does not need to predict a particular result. The point may be to explore meanings or patterns that are not yet well understood.
What makes a hypothesis useful?
A hypothesis is a proposed answer that can be evaluated against evidence. In a quantitative study, it usually identifies the expected relationship or difference between measured variables. A hypothesis should be clear enough that the study could produce results that do not support it.
A useful hypothesis is grounded in prior evidence, theory, or a reasoned expectation. It also fits the study design. If you are testing a causal claim, for example, the design must support that kind of inference; simply writing a causal hypothesis does not make an observational study capable of proving causation.
Hypotheses can be directional or non-directional:
- Directional: “First-year students who report more average nightly sleep will have higher weekly quiz scores.” This predicts the direction of the relationship.
- Non-directional: “Average nightly sleep duration is associated with weekly quiz scores.” This predicts a relationship but not its direction.
These are research hypotheses: substantive predictions about the topic. In statistical testing, researchers may also specify a null hypothesis, often written as no association or no difference, and an alternative hypothesis. Those formal statistical statements are part of an analysis plan; they are not simply interchangeable with the research question.
Worked example: from topic to question and hypothesis
Suppose you are interested in whether sleep is related to academic performance. Turn that broad topic into a study plan in stages:
- Choose a population and context: first-year university students in one semester.
- Identify measurable variables: average nightly sleep duration and weekly quiz score.
- Write the research question: “Among first-year university students, is average nightly sleep duration associated with weekly quiz scores during one semester?”
- State a prediction if the design calls for one: “Students reporting more average nightly sleep will tend to have higher weekly quiz scores.”
- Match the claim to the design: if you survey sleep and record scores without assigning students to sleep schedules, the study can examine association, but by itself may not establish that sleep caused a score difference; a causal conclusion would require an appropriate design or additional assumptions and methods.
The question and hypothesis can be paired, but they are not redundant. The question defines what to investigate. The hypothesis makes the expected pattern explicit so the researcher can decide what data to collect and how to assess the prediction.
Do all studies need a hypothesis?
No. A hypothesis is useful when you have a sufficiently specific expectation and a design that can test it. Exploratory studies may instead use one or more research questions, because predicting a result in advance could be premature. Many qualitative studies are organized around open-ended questions rather than hypotheses, though practices vary by field and methodology.
A project can also begin with a broad research question and develop more focused questions as the researcher learns about the topic. If a proposal, supervisor, or journal requires a hypothesis, follow that requirement—but make sure the proposed prediction genuinely fits the study.
Common mistakes
- Treating the two terms as synonyms. A question asks; a hypothesis predicts.
- Making the question too broad. Specify enough context to show what evidence could answer it.
- Writing an untestable hypothesis. “Sleep is important” is a general claim, not a prediction with a clear way to assess it.
- Confusing a hypothesis with a result. A hypothesis is generally proposed before analysis in confirmatory work; findings may support it, fail to support it, or require a more nuanced interpretation.
- Claiming causation from correlation. Use language such as “is associated with” unless the design supports a causal conclusion.
- Assuming every project needs a hypothesis. For exploratory or qualitative work, a well-framed research question may be the better fit.
References
- Research Design: Qualitative, Quantitative, and Mixed Methods Approaches, John W. Creswell and J. David Creswell
- The Craft of Research, Wayne C. Booth, Gregory G. Colomb, Joseph M. Williams, Joseph Bizup, and William T. FitzGerald
Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.