How to Write a Hypothesis: Null, Alternative and Directional

Reviewed by Prof. Dr. Alexander Holtermann · Updated 16 October 2026 · 5 min read

A hypothesis is a testable prediction about the relationship between variables. Write a null (H₀) and an alternative hypothesis (H₁) for each test.

Key takeaways

A research hypothesis is a precise, testable statement about the relationship between two or more variables, derived from theory or previous studies.
In statistics you test the null hypothesis (H₀: no effect, no difference, no relationship) against the alternative hypothesis (H₁: there is an effect).
Directional hypotheses predict the direction (“higher”, “positively related”); non-directional ones only predict that there is a difference or relationship.
Good hypotheses name variables and population, are falsifiable and can be tested with the data you will actually collect.
Qualitative studies usually work with open research questions instead of hypotheses.

What is a hypothesis?

A hypothesis is a testable prediction. It states what you expect to find about the relationship between variables before you collect data. It is not a guess: it follows from theory or from previous research, which is why your literature review should lead directly to your hypotheses.

Hypotheses belong to quantitative, deductive research. They make your study falsifiable: the data can contradict them. If you are doing interviews or other qualitative work, you normally formulate open research questions instead, because you want to discover rather than test.

Research question: “Is social media use related to sleep quality among undergraduates?”
Hypothesis: “Undergraduates who spend more time on social media in the evening report poorer sleep quality.”

Null and alternative hypothesis

Statistical tests work with a pair of hypotheses. The null hypothesis (H₀) states that there is no effect, no difference or no relationship in the population. The alternative hypothesis (H₁ or Hₐ) states the opposite: there is an effect. Your test calculates how likely your data would be if H₀ were true (the p-value). If p is below your significance level (usually .05), you reject H₀.

Two points matter: you never “prove” H₁, you only reject or fail to reject H₀. And “not significant” does not mean “no effect”; your sample may simply have been too small.

Type	Example (two groups)	Example (correlation)
H₀	Mean exam anxiety does not differ between first-year and final-year students (μ₁ = μ₂).	There is no correlation between study hours and grade (ρ = 0).
H₁ non-directional	Mean exam anxiety differs between the groups (μ₁ ≠ μ₂).	Study hours and grade are correlated (ρ ≠ 0).
H₁ directional	First-year students report higher exam anxiety (μ₁ > μ₂).	Study hours and grade are positively correlated (ρ > 0).

Directional or non-directional?

A directional (one-tailed) hypothesis predicts the direction of the effect. Use it only if theory or consistent prior findings clearly justify the direction, and decide before you see the data. A non-directional (two-tailed) hypothesis only predicts that there is a difference or relationship.

Many supervisors and journals prefer two-tailed tests even for directional hypotheses, because they are more conservative. A one-tailed test needs fewer participants (for d = 0.5 about 51 instead of 64 per group at 80% power), but switching to one-tailed after seeing the results is a form of p-hacking.

How to write a hypothesis in five steps

Follow these steps from research question to testable hypothesis:

1. Start with a research question that asks about a relationship or difference.
2. Review the literature to find out what theory and previous studies predict.
3. Identify your variables: independent variable (predictor, cause, group) and dependent variable (outcome).
4. Write H₁ as a clear statement including the population, variables and, if justified, the direction.
5. Write H₀ as the statement of no effect, and check that your planned test (t-test, correlation, chi-square) can actually test the pair.

Criteria for a good hypothesis

Check each hypothesis against this list before you finalize your proposal.

Criterion	Weak	Better
Testable	Social media is bad for students.	Evening social media use is negatively related to self-reported sleep quality.
Specific population	People sleep worse with phones.	Undergraduate students at UK universities …
Measurable variables	… are less happy.	… score lower on the Satisfaction with Life Scale.
One relationship per hypothesis	Tutoring improves grades and motivation and reduces anxiety.	Split into three hypotheses (H1a, H1b, H1c).
Grounded in theory	I think older students are more motivated.	Based on self-determination theory, mature students report higher intrinsic motivation.

Hypotheses in your proposal and methods chapter

Number your hypotheses (H1, H2, H3, or H1a/H1b for related ones) and list them at the end of your literature review or at the start of your methods chapter. In your results chapter, report each one explicitly: “H1 was supported, t(126) = 2.41, p = .017, d = 0.43.” In the discussion, explain what the result means and why unsupported hypotheses may have failed (effect smaller than expected, measurement problems, sample).

If you are under time pressure, consider preregistration: writing your hypotheses and analysis plan down before data collection protects you against accusations of changing them later. Many open science platforms offer free preregistration templates.

Common mistakes

Examiners see these errors often:

Hypotheses that are actually questions (“Is there a relationship …?”).
Writing H₀ as the expected result. H₀ is the statement of no effect.
Claiming a hypothesis was “proven”. Say “supported” or “not supported”.
Hypotheses about variables you never measured.
Choosing a directional test after looking at the data.
Hypotheses in a qualitative interview study where they do not belong.

Tools and institutional requirements

Our research question tool helps you sharpen the question your hypotheses rest on, and the power analysis tool tells you how many participants you need to test them. Some departments require formal H₀/H₁ pairs, others only research hypotheses. Your university’s own guidelines take precedence; upload them to Professor Digital and the suggestions are tailored to them.

Frequently asked questions

What is the difference between a null and an alternative hypothesis?

The null hypothesis (H₀) states that there is no effect, difference or relationship. The alternative hypothesis (H₁) states that there is one. Statistical tests decide whether to reject H₀.

How do you write a good hypothesis?

Name the population and the variables, state the expected relationship clearly, make it testable with your planned data and base it on theory or previous research.

What is a directional hypothesis?

A hypothesis that predicts the direction of an effect, e.g. “Group A scores higher than Group B”. It is tested one-tailed and should be justified by theory before data collection.

Do qualitative studies need hypotheses?

Usually not. Qualitative research works with open research questions because it aims to explore and understand rather than test predictions.

Can a hypothesis be proven?

No. Data can support a hypothesis or lead you to reject the null hypothesis, but a single study never proves it. Use “supported” or “not supported”.

Sources and further reading

Creswell, J. W., & Creswell, J. D. (2023). Research Design: Qualitative, Quantitative, and Mixed Methods Approaches (6th ed.). Sage.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). Lawrence Erlbaum.

Your university’s examination regulations and style guide always take precedence.