How to Write a Testable Research Hypothesis

A testable hypothesis is a specific prediction about the relationship between variables that data could contradict. In statistical testing it is usually paired with a null hypothesis (no difference or association) and an alternative hypothesis (a difference or association).

Features of a good hypothesis

• Specific: names the population, variables and measure.
• Testable and falsifiable: results could show it is wrong.
• Based on prior knowledge: reasoned from existing evidence.
• Consistent with the objective and analysis.

Forms

• Null (H0): "There is no difference in mean systolic blood pressure between groups."
• Alternative (H1): "There is a difference." (two-sided) or "Intervention lowers pressure." (one-sided, to be justified in advance).
• Directional vs non-directional: choose before seeing the data.

Illustrative example

Invented for teaching. Research question: Does a 12-week walking programme lower systolic blood pressure in adults with hypertension? H0: mean change in systolic blood pressure is the same in the walking and control groups. H1: mean change differs between groups. Primary analysis: difference in mean change at 12 weeks with a 95% confidence interval.

Interpreting results

A small p-value indicates that data would be unusual if the null were true; it does not give the probability that the hypothesis is true, nor the size or importance of an effect. Report effect sizes and confidence intervals. See What Is a P-Value?.

Writing hypotheses for different designs

In a randomised trial, the hypothesis is usually about a difference between groups on the primary outcome. In an observational study, it is about an association, and the wording should reflect that causation is not established by the design. In diagnostic accuracy studies, it may specify minimum sensitivity or specificity. State the hypothesis, the measure and the direction (if any) before analysis, and note the planned significance level.

Common pitfalls

• Writing hypotheses after seeing the data.
• Testing many hypotheses and reporting only those that "work".
• Using a hypothesis in a purely descriptive study.

Key takeaways

• State hypotheses before collecting data.
• Match them to a planned analysis and effect size.
• Not every study needs one.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.