What Is a Hypothesis? Definition & Types, with Examples
By Dr. Robertson Prime, Research Fellow
Updated October 21, 2026
13 min read
About 2,600 words
What is a hypothesis? A hypothesis is a testable statement that predicts the relationship between two or more variables, based on existing theory or observation. It is a tentative answer to your research question that your study will support or reject with evidence.
This guide gives you a clear definition, the main types, a 5-step method to formulate a good hypothesis and the steps of statistical hypothesis testing.
- Definition and explanation: what a hypothesis is, and how it differs from a theory, a prediction and a research question.
- Writing it: the features of a testable hypothesis and a 5-step flowchart to write a hypothesis.
- Testing it: null vs alternative hypotheses, p-values and how to reject the null, with weak vs strong examples.
On this page
Key takeaways
- A hypothesis is a specific, testable prediction about variables.
- A good hypothesis is clear, falsifiable and grounded in existing research.
- Statistical tests compare a null hypothesis with an alternative hypothesis.
- You either reject the null or fail to reject it; you never “prove” a hypothesis.
What Is a Hypothesis? Definition and Explanation
So, what is a hypothesis in research? In simple terms, it is a proposed explanation for a phenomenon, stated precisely enough that data can test it.
- A working definition: a hypothesis is a statement that predicts how variables relate, such as how an independent variable affects a dependent variable.
- A tentative answer: it is your best, evidence-based guess before data collection, not a conclusion.
- Testable by design: it must be possible to collect data that could support or contradict it.
- Grounded in evidence: it comes from observation, theory or a review of existing literature, not from a hunch alone.
What is a hypothesis in scientific research?
In scientific research, the hypothesis is the hinge of the research process. It links what you already know to what you will test in your study design.
- Before data collection: you hypothesize a relationship based on theory, observation and prior findings.
- During the study: an experiment, survey or other design collects evidence that could support or contradict it.
- After analysis: results either support the hypothesis or send you back to revise your assumptions.
- Across studies: repeated tests of the same hypothesis in different settings build the evidence behind a scientific theory.
Every hypothesis also rests on an assumption, such as the assumption that your measures capture the concepts accurately. State key assumptions openly so readers can judge your test.
Hypothesis vs theory, prediction and research question
Students often mix these terms up when they first ask what is a hypothesis. Knowing the difference makes your study design clearer.
| Term | What it is | Example |
|---|---|---|
| Research question | The question your study asks | How does sleep relate to grades? |
| Hypothesis | A testable statement answering the question in advance | More sleep is associated with higher grades. |
| Prediction | The specific result you expect in this data set | Group A’s mean grade will exceed Group B’s. |
| Scientific theory | A broad, well-tested explanation supported by many studies | Theories of sleep and memory consolidation |
Where hypotheses come from: observation and existing knowledge
Hypotheses rarely come from nowhere. Researchers generate them from three main sources, each tied to a specific research context.
- Observation: noticing a pattern in practice, such as tired students performing worse in morning exams.
- Existing knowledge: findings from existing research that suggest a relationship worth testing in a new population.
- Theory: a theoretical framework that predicts how variables should relate; see our guide to the theoretical framework.
What Makes a Good Hypothesis: Testable and Falsifiable
Knowing what is a hypothesis is only half the job; you also need to know what makes one strong. A good hypothesis can be tested, and a test could show it to be wrong.
Features of a testable hypothesis
Check every draft against these features before you build your study design around it.
- Testable: you can measure the variables with available methods and data.
- Falsifiable: some possible result would contradict it; the philosopher Karl Popper made falsifiability a core test of scientific claims.
- Specific: it names the variables, the population and, where possible, the direction of the effect.
- Grounded: it builds on existing research rather than personal opinion.
- Simple: it focuses on one relationship between variables per statement.
How to write a hypothesis in one sentence
A well-formed hypothesis is a statement that fits into one clear sentence. Two common patterns help.
- If–then form: “If students use spaced practice, then their test scores will be higher than those of students who cram.”
- Relationship form: “Daily social media use is negatively associated with body image scores among girls aged 13–17.”
- Difference form: “Nurses on 12-hour shifts report higher fatigue scores than nurses on 8-hour shifts.”
Types of Hypothesis in Research
Once you know what is a hypothesis, the next step is choosing the right kind. Different research questions need different types of hypothesis. The most important distinction in statistics is between the null and alternative hypothesis.
Null hypothesis vs alternative hypothesis
Every statistical test works with a pair of hypotheses. The null hypothesis (H₀) states that there is no effect or no difference; the alternative hypothesis (H₁ or Hₐ) states the effect you expect.
Null vs alternative: one study, two statements
Null (H₀)
“There is no difference in mean exam scores between students who sleep seven or more hours and those who sleep fewer than six.”
The default position your test tries to reject.
Alternative (H₁)
“Students who sleep seven or more hours have higher mean exam scores than those who sleep fewer than six.”
The prediction your study is designed to support.
For a deeper comparison, read our guide to the null vs alternative hypothesis.
Other types of hypothesis
- Directional: predicts the direction of an effect (“higher”, “lower”); see directional vs non-directional hypotheses.
- Non-directional: predicts a difference or relationship without saying which way.
- Simple vs complex: a simple hypothesis links one independent and one dependent variable; a complex one involves several.
- Associative vs causal: associative hypotheses predict that variables move together; causal ones predict that one variable causes change in another, which usually needs an experimental design.
- Research vs statistical: a research hypothesis is written in words for readers, while a statistical hypothesis restates it as H₀ and H₁ for testing.
How to Formulate a Good Hypothesis in 5 Steps
Understanding what is a hypothesis makes the next step easier. You can formulate a hypothesis by moving from a question to a precise, testable statement. Follow these five steps to write a hypothesis for a thesis, dissertation or research paper.
- Ask a questionStart from your topic
- Review the literatureWhat is already known
- Identify variablesIndependent and dependent
- Write the statementIf–then or relationship
- Refine and pair itAdd the null hypothesis
- Ask a focused research question. Start with a clear, specific question about your topic, such as how sleep relates to academic performance. Our guide on how to write a research question shows how to narrow it.
- Review existing literature. Read recent studies to see what is already known and where results disagree. This tells you whether to predict a direction and which variables matter.
- Identify your variables. Name the independent variable (what you change or compare), the dependent variable (what you measure) and the population you will study.
- Write the hypothesis as a statement. Use an if–then or relationship sentence that predicts how the variables relate. Keep it to one sentence and one relationship.
- Refine it and write the null. Check that it is testable and falsifiable, then write the matching null hypothesis so your statistical test has a clear starting point.
For more worked advice, see our guide on how to create a strong hypothesis for a dissertation.
Hypothesis Testing: How to Test a Hypothesis
Knowing what is a hypothesis test helps you report results correctly. Hypothesis testing is the statistical method for deciding whether your data provide enough evidence against the null hypothesis. It turns a hypothesis into a decision you can report.
The steps of statistical hypothesis testing
Penn State’s statistics review of hypothesis testing describes the logic as assuming the null is true, collecting evidence and then deciding. In practice, the process has five steps.
- State the null and alternative hypotheses. Write H₀ and H₁ in words and, where useful, in symbols.
- Choose a significance level. Set alpha (α) before collecting data; 0.05 is a common convention.
- Collect data and run the test. Choose a test that fits your data and design, such as a t-test, chi-square test or regression.
- Compare the p-value with alpha. The p-value is the probability of results at least as extreme as yours if the null hypothesis were true.
- Make a decision. If p is less than α, reject the null; otherwise, fail to reject it, and report the result either way.
Reject the null or fail to reject it
Statistics never proves a hypothesis true. The wording of your decision matters.
- Reject the null: the data are unlikely under H₀, so they support the alternative hypothesis.
- Fail to reject the null: the data do not provide enough evidence against H₀; this is not proof that H₀ is true.
- Type I error: rejecting a null hypothesis that is actually true (a false positive).
- Type II error: failing to reject a null hypothesis that is actually false (a false negative).
Testing hypotheses in SPSS
Most dissertations run these tests in software such as SPSS. Choose the test that matches your specific research design and data.
- Comparing one group with a known value: a one-sample t-test in SPSS.
- Comparing two independent groups: a two-sample t-test, after checking equal variances.
- Comparing the same people twice: a paired t-test, such as before and after an intervention.
Weak vs Strong Hypothesis Examples
Comparing weak and strong versions is the fastest way to see what is a hypothesis that examiners will accept. Each pair fixes one common problem.
Too vague to test
Weak
“Social media affects teenagers.”
No variables, no direction and no population; no result could disprove it.
Strong
“Among girls aged 13–17, more daily hours on image-based social media are associated with lower body image scores.”
Names the population, both variables and the direction of the relationship.
Not falsifiable
Weak
“Some teaching methods may or may not help some students.”
Every possible result fits it, so it cannot be tested.
Strong
“First-generation college students in courses using culturally responsive pedagogy will report higher academic engagement than students in standard courses.”
A clear result could support or contradict it.
A value judgment, not a hypothesis
Weak
“Minimum wage increases are unfair to small businesses.”
“Unfair” is an opinion that data cannot test.
Strong
“Small businesses in metropolitan areas reduced new hires in the year after the minimum wage increase, compared with the year before.”
Measurable outcome, clear comparison and a defined time frame.
Hypothesis examples by field
Psychology and education
H₁: “Early-career teachers who use cognitive reappraisal more often report lower burnout scores.” H₀: “There is no relationship between cognitive reappraisal use and burnout scores.”
This pair could be tested with survey data and correlation or regression analysis.
Health and nursing
H₁: “Adults with diabetes who receive monthly visits from community health workers will show lower HbA1c levels after six months than adults receiving usual care.” H₀: “There is no difference in HbA1c levels between the two groups after six months.”
This pair suits an experimental or quasi-experimental study design.
Hypothesis template
Copy this template to draft a hypothesis pair for a quantitative study.
Template
H₁: Among [population], [independent variable] will be [positively/negatively] associated with [dependent variable]. H₀: Among [population], there is no relationship between [independent variable] and [dependent variable].
Hypotheses sit near the end of the introduction chapter, after your research questions; our dissertation introduction guide shows where they fit, and how to write a dissertation covers the whole process.
What Is a Hypothesis? FAQs
What is a hypothesis in simple words?
A hypothesis is an educated guess that you can test. It is a statement that predicts how two or more things are related, based on what you already know from theory, observation or earlier studies.
For example, “students who sleep more get higher grades” is a hypothesis because you can collect data on sleep and grades to check it. What makes it different from an ordinary guess is that it is specific, grounded in evidence and stated so that a clear result could show it to be wrong.
What are the three main parts of a hypothesis?
Most hypotheses contain three parts: the independent variable, the dependent variable and the predicted relationship between them. Many also name the population being studied.
In the hypothesis “students who sleep at least seven hours have higher grades than students who sleep fewer than six hours”, sleep is the independent variable, grades are the dependent variable, and “higher” states the predicted direction. Including all three parts makes your hypothesis clear and testable, and it helps you choose the right data collection methods and statistical test later in your study.
What is the difference between a hypothesis and a prediction?
A hypothesis is a general, testable statement about how variables relate, usually drawn from theory or earlier research. A prediction is the specific result you expect to see in your own data if the hypothesis is correct.
For instance, the hypothesis might be that more sleep improves academic performance, while the prediction is that students in your high-sleep group will have a higher mean exam score than your low-sleep group. The hypothesis explains the relationship; the prediction states what your particular study should observe if that explanation holds.
Do qualitative studies need a hypothesis?
Usually not. Qualitative research explores meanings, experiences and processes, so it is guided by open research questions rather than hypotheses that predict relationships between variables. A qualitative study might ask how nurses experience therapeutic communication, instead of predicting a measurable effect.
Some qualitative researchers develop hypotheses at the end of a study, as tentative explanations for later testing. Mixed-methods research may include hypotheses for its quantitative strand only. If your program’s template asks for hypotheses in a qualitative project, check with your supervisor about what is expected.
Can a hypothesis be proven true?
No. Statistical testing can only show whether your data provide enough evidence to reject the null hypothesis. When you reject the null, your results support the alternative hypothesis, but another study with different data could still find otherwise.
That is why researchers say a hypothesis is “supported” rather than “proven”. When you fail to reject the null, you have not proven the null either; you simply lack enough evidence against it. Careful wording in your results and discussion chapters shows examiners that you understand this logic.
What is a hypothesis statement example?
A clear hypothesis statement example is: “Among early-career teachers, more frequent use of cognitive reappraisal is associated with lower burnout scores.” It names the population (early-career teachers), the independent variable (use of cognitive reappraisal), the dependent variable (burnout scores) and the expected direction (lower).
Its matching null hypothesis would state that there is no relationship between the two variables. When you ask what is a hypothesis that examiners will accept, this is the pattern to follow: one population, one relationship and a direction you can test.
Best Dissertation Writers
Need help writing your dissertation introduction?
Our expert writers can help you draft your background, problem statement, purpose and research questions to your brief.
