How to Write a Hypothesis: Steps and Examples

A hypothesis is a testable, falsifiable statement about the expected relationship between two or more variables. It is not a guess. It is a prediction derived from theory and prior evidence that specifies what you expect to find and under what conditions.

The most common error in writing hypotheses is stating a prediction that cannot be proven wrong. A hypothesis that is always true regardless of results is not a hypothesis; it is a tautology.

Types of Hypotheses

Null and Alternative Hypothesis

Every statistical test involves both a null and an alternative hypothesis, even if only one is stated explicitly. Understanding the pair is essential for interpreting results correctly.

Null hypothesis (H0): Assumes no effect, no difference, or no relationship. The null is the "default" assumption that the study is trying to test against. If the data provide strong enough evidence, the null is rejected.

Alternative hypothesis (H1): The researcher's actual prediction. If the null is rejected, the alternative is supported (not proven).

Example pair:

• H0: Students who receive peer tutoring do not differ in final exam scores from students who receive standard instruction.
• H1: Students who receive peer tutoring score higher on final exams than students who receive standard instruction.

How to Write a Hypothesis: Five Steps

Step 1: Identify your research question. A hypothesis answers a research question. If you do not have a clear, specific research question, you cannot write a hypothesis. The question should identify the specific variables and population of interest.

Research question: "Does sleep duration predict academic performance in college students?"

Step 2: Review the existing evidence. Your hypothesis should be grounded in existing theory or prior empirical findings. A hypothesis that contradicts everything previously found is not impossible but requires stronger justification. A hypothesis consistent with prior research is more credible before data is collected.

Step 3: Identify your variables. State which variable is the cause or predictor (independent variable) and which is the outcome (dependent variable). If the relationship is more complex, identify any mediating or moderating variables at this stage.

Step 4: State the relationship as a prediction. Write a sentence that specifies the relationship between the variables. If prior evidence supports a direction, write a directional hypothesis. If direction is uncertain, write a non-directional hypothesis.

Step 5: Check that your hypothesis is testable and falsifiable. Ask: what result would prove this hypothesis wrong? If you cannot answer that question, the hypothesis is not falsifiable and cannot be tested. A good hypothesis has a clear answer to this question.

Hypothesis Examples by Type

Simple Directional Hypothesis

"Adults who engage in aerobic exercise three or more times per week report lower levels of perceived stress than adults who do not engage in regular aerobic exercise."

What makes this strong:

• Identifies the independent variable (aerobic exercise frequency)
• Identifies the dependent variable (perceived stress)
• Specifies direction (lower stress for exercisers)
• Identifies a population (adults)

Simple Non-Directional Hypothesis

"There is a significant relationship between sleep duration and GPA among first-year college students."

Use this form when: prior studies have shown a relationship but have found mixed results on direction, or when the direction cannot be predicted from theory.

Complex Hypothesis (with moderator)

"The positive relationship between peer tutoring and academic performance is stronger for first-generation college students than for continuing-generation college students."

This is a moderation hypothesis. It predicts not just a main effect but that the effect varies in strength depending on a third variable (generation status).

Null and Alternative Pair

H0: There is no significant difference in 30-day hospital readmission rates between patients who receive a nurse navigator discharge program and patients who receive standard discharge planning.

H1: Patients who receive a nurse navigator discharge program have significantly lower 30-day hospital readmission rates than patients who receive standard discharge planning.

Hypothesis with Mediation

"The relationship between childhood food insecurity and adult depression is mediated by adverse childhood experiences (ACEs): food insecurity increases exposure to ACEs, which in turn increases depression risk in adulthood."

This is a mediation hypothesis. It predicts not just that X relates to Y, but that this relationship operates through a specific mechanism (Z).

Hypothesis vs Research Question vs Thesis Statement

These three terms are often confused:

A research question asks. A hypothesis predicts. A thesis statement argues. Research questions become hypotheses when enough prior evidence exists to support a specific prediction. Thesis statements are for argumentative writing, not empirical research.

Characteristics of a Strong Hypothesis

Testable: The hypothesis can be evaluated with observable data. "Students who study more learn more" is vague. "Students who study for more than two hours per day achieve higher test scores" is testable.

Falsifiable: There must be a result that would prove the hypothesis wrong. If the hypothesis is equally consistent with all possible outcomes, it is not a hypothesis.

Specific: Vague predictions produce vague results. The hypothesis should specify what variables are being compared, in what population, under what conditions.

Grounded: The hypothesis should be derivable from existing theory or prior evidence. State explicitly what evidence or theory supports the prediction.

Concise: A well-formed hypothesis can be stated in one or two sentences. If you need a paragraph to state it, it is probably more than one hypothesis.

Common Mistakes

Stating a fact instead of a prediction. "Students with higher GPAs perform better on standardized tests" is a finding, not a hypothesis. A hypothesis is stated before data collection.

Using "I think" or "I believe." Hypotheses are scientific statements, not personal opinions. Remove first-person hedges.

Non-falsifiable hypothesis. "People's experiences of education are shaped by many factors" cannot be tested or proven wrong. It describes everything and predicts nothing.

Vague variables. "Exercise improves health" has variables too vague to measure. What kind of exercise? What health outcome? Measured how? When?

Hypothesis as question. "Does exercise reduce stress?" is a research question, not a hypothesis. Restate it as a prediction: "Adults who exercise three or more times per week report lower perceived stress than those who do not."

Direction stated without evidence. If you write a directional hypothesis ("X increases Y"), you should cite the prior evidence or theory that supports that direction. A directional prediction without theoretical basis is arbitrary.

Frequently Asked Questions

What is an example of a hypothesis? "Elementary school students who participate in 20 minutes of structured outdoor recess before math class will score higher on in-class math assessments than students who have no recess before math class." This hypothesis identifies the population (elementary students), the manipulation (outdoor recess timing), the outcome (math assessment scores), and the direction of the expected effect.

What is the difference between a hypothesis and a prediction? A prediction is any statement about a future outcome. A hypothesis is a specific type of prediction that is testable, falsifiable, and grounded in theory or prior evidence. All hypotheses are predictions, but not all predictions qualify as scientific hypotheses. "I predict it will rain tomorrow" is a prediction, not a hypothesis.

How do you write a null hypothesis? A null hypothesis states that there is no relationship between the variables being studied. Write it as the direct negation of your alternative hypothesis. If H1 is "Students who receive feedback perform better than those who do not," H0 is "There is no significant difference in performance between students who receive feedback and those who do not." The null hypothesis is what statistical tests are designed to evaluate.

Should a hypothesis be a statement or a question? A hypothesis is always a statement, not a question. The question form is the research question. The hypothesis answers the research question with a specific prediction. "Does caffeine improve memory?" is a research question. "Caffeine consumption of 200mg improves short-term memory recall compared to placebo" is a hypothesis.

How is a hypothesis different from a theory? A hypothesis is a specific, testable prediction about a single study. A theory is a broader, well-supported framework that explains a pattern of findings across many studies. Hypotheses are tested in individual studies. Theories are built up from many replicated findings. A good hypothesis is derived from theory, but a single confirmed hypothesis does not validate a theory.

For related research writing guidance, see research question examples and how to write a research paper introduction.