How to write a clear research hypothesis that your supervisor can work with

A well written hypothesis can make the rest of your project easier: it guides your design, data collection, analysis and discussion. A vague one does the opposite, leaving you unsure what to measure or how to explain your results.

This guide walks through what a hypothesis is, what it is not, and how to draft one that is specific, testable and realistic for a student project. Requirements differ across fields and institutions, so always check your local guidelines and discuss your draft with your supervisor.

What a research hypothesis is (and is not)

A research hypothesis is a clear, focused statement that predicts a relationship between variables in a way that could be supported or not supported by data. It is usually based on prior theory, earlier findings or a strong logical argument.

It is not just a topic, aim or question. For example, “Social media and sleep” is a topic, “To explore social media use” is an aim, and “How does social media use relate to sleep?” is a question. A hypothesis goes one step further and states a specific expected pattern.

Start from a focused research question

Before writing a hypothesis, you need a question narrow enough that you could reasonably answer it with the resources and time you have. If your question is too broad, your hypothesis will also be vague.

Compare these two versions:

• Broad:“How does technology affect learning?”
• Narrower:“Does using spaced practice apps for four weeks improve first year psychology students’ quiz scores compared with their usual revision habits?”

The narrower question already hints at variables (spaced practice app use, quiz scores, first year psychology students) and comparison (app vs usual revision). This makes it much easier to turn into a hypothesis.

Identify your variables clearly

Most simple hypotheses involve two main types of variables: an independent variable (what you change or compare) and a dependent variable (what you measure). Sometimes you also consider control variables that you will try to keep constant or account for.

Using the spaced practice example:

• Independent variable:Use of a spaced practice app vs usual revision
• Dependent variable:Quiz scores at the end of four weeks
• Control ideas:Same course, similar baseline knowledge, same quiz format

If you cannot describe your variables in simple, observable terms, your hypothesis probably needs more work before you move on.

Make the hypothesis testable and measurable

A testable hypothesis can be evaluated with real data. This means your variables must be measurable and your prediction must be specific enough that different observers would agree on whether the result supports it or not.

Compare:

• Vague:“Using spaced practice apps will greatly improve learning.”
• Testable:“Students who use a spaced practice app for four weeks will achieve higher average quiz scores than students who use their usual revision methods.”

The second version avoids the word “greatly” and focuses on “higher average quiz scores”, which can be checked with data. You can later specify details like how you will measure “higher” (for example, statistical tests or a minimum difference) in your methods section.

Choose a suitable level of precision

In some projects, especially in experimental or quantitative work, you might be asked to specify the direction of the effect, such as higher, lower, faster or slower. These are often called directional hypotheses.

Examples:

• Directional:“Participants who sleep at least 8 hours will report lower stress scores than those who sleep less than 6 hours.”
• Non directional:“There will be a difference in stress scores between participants who sleep at least 8 hours and those who sleep less than 6 hours.”

Which type is appropriate depends on your field, prior findings and the expectations of your supervisor or department. If earlier work already suggests a direction, a directional hypothesis is often preferred, but do not invent certainty where there is little prior information.

Align your hypothesis with your design

Your hypothesis should match what your design can realistically test. If you hypothesize a comparison between two groups, your design must include those groups. If you predict a correlation, your design should collect data that can show how two variables move together.

Here are three common patterns:

• Difference between groups:“Group A will have higher scores than Group B.” (Requires at least two groups and comparable measures.)
• Change over time:“Scores after the intervention will be higher than scores before.” (Requires repeated measures on the same people.)
• Association between variables:“Higher values on X will be linked with higher values on Y.” (Requires measurements of both X and Y for each participant or case.)

If your hypothesis mentions something your design does not measure, either adjust the hypothesis or modify the design before you proceed.

Write in clear, simple language

A good hypothesis is easy to read on the first try. Avoid overly technical phrasing, unexplained jargon and long strings of clauses. Someone in your general field should be able to paraphrase your hypothesis without asking what you meant.

Try these practical tips when you draft:

• Use one main sentence for each hypothesis.
• Keep the structure “If X, then Y” or “X will be associated with Y” where possible.
• State who or what your hypothesis is about, not just “participants.”
• Avoid vague qualifiers like “significantly better” in the wording; save statistical details for the methods and analysis sections.

Check for feasibility and ethical sense

Even if a hypothesis is logically sound, it may not be feasible or ethical for your context. Hypothesizing that “daily laboratory blood tests will improve exam performance” might be scientifically interesting, but unrealistic for a small student project and potentially distressing for participants.

Before finalizing your wording, ask yourself:

• Can I collect the data needed within my time and resource limits?
• Is the procedure acceptable to participants and likely to pass ethics review where required?
• Do I have access to the population or materials needed?

If the answer to any of these is “probably not”, refine your hypothesis to fit a more practical design.

Use hypotheses to structure the rest of your project

Once you are satisfied with your hypothesis, use it as a guide rather than just a formal requirement. It can help you decide which prior work to discuss in your background section, what to measure, how to plan your analysis and how to organize your discussion.

You can even list your hypotheses explicitly and then return to each one in your results and conclusion sections. This creates a clear line from question, through method, to interpretation, which supervisors and assessors often appreciate.

Finally, remember that a hypothesis is a starting point, not a promise. It is acceptable for data to fail to support your prediction. What matters is that your hypothesis was well reasoned, clearly written and tested with an appropriate design for your level and field.