What Is a Hypothesis? Definition, Types and Examples
A hypothesis is the starting point of any empirical research. Before you collect a single data point, you need a clear, testable prediction about what you expect to find and why. Students often confuse a hypothesis with a research question, a theory, or a general assumption. This guide explains exactly what a hypothesis is, how to write one, the seven main types, and what separates a strong hypothesis from a weak one.
Quick Answer
A hypothesis is a specific, testable prediction about the expected relationship between variables in a study. It is based on prior knowledge, observation, or theory, and it can be either supported or refuted by evidence. A hypothesis is not a question and not a fact. It is an informed prediction stated in declarative form that guides the design of your study and the interpretation of your results.
What Is a Hypothesis?
A hypothesis is a proposed explanation for an observed phenomenon, stated in a form that can be tested through research. As Britannica defines it, a hypothesis helps to connect facts and solve problems by providing a tentative explanation that fits within existing knowledge. It expresses an expected relationship between an independent variable (what you manipulate or observe) and a dependent variable (what you measure).
In academic research, a hypothesis serves several functions:
- It gives the study a clear direction and scope
- It connects the research question to the data collection design
- It allows findings to be evaluated as supporting or contradicting the prediction
- It contributes to building or refining scientific theory over time
A hypothesis is always written before data is collected. In the scientific method, a researcher develops a hypothesis, tests it, and then modifies it based on outcomes, not the other way around. Formulating a hypothesis after seeing the data and then claiming it was your original prediction is a form of research misconduct called HARKing (Hypothesizing After Results are Known).
Hypothesis vs. Research Question
These two are related but not interchangeable.
| Research Question | Hypothesis | |
|---|---|---|
| Form | Question | Declarative statement |
| Purpose | Defines what the study explores | Predicts what the study will find |
| Example | "Does sleep duration affect exam performance?" | "Students who sleep at least 8 hours before an exam will score higher than those who sleep fewer than 6 hours." |
| Position in paper | Introduction | Introduction or after literature review |
A research question opens the inquiry. A hypothesis closes it with a specific, testable answer before data collection begins.
Hypothesis vs. Theory
A theory is a well-established explanation supported by extensive evidence across many studies. A hypothesis is a single untested prediction made before a specific study. As Britannica explains, hypotheses are specific and serve as the main tool for data collection, whereas theories are broad and integrate data from various scientific explorations to form a general explanation. Hypotheses are tested in studies. Theories emerge from many confirmed hypotheses over time.
The Parts of a Hypothesis
A well-formed hypothesis contains three elements:
1. The variables: An independent variable (the cause or condition you manipulate or observe) and a dependent variable (the outcome you measure).
2. The predicted relationship: A statement about how the independent variable is expected to affect the dependent variable. The relationship should be directional (increases, decreases, is higher, is lower) or non-directional (differs, is related to).
3. The population: Who or what the hypothesis applies to.
Example:
Among undergraduate students (population), increased daily exercise (independent variable) is associated with reduced levels of reported academic stress (dependent variable).
The PICOT framework is a useful tool for ensuring your hypothesis includes all necessary elements, particularly in health and social science research:
- Population: Who is the study about?
- Intervention or Interest: What is being studied or manipulated?
- Comparison: What is the alternative or control condition?
- Outcome: What result is being measured?
- Time: Over what period?
PICOT hypothesis example:
Among university students (P) who participate in a structured mindfulness program (I) compared to those who do not (C), self-reported anxiety scores (O) will be significantly lower after 8 weeks (T).
Types of Hypothesis
There are seven main types of hypotheses used in academic research. Understanding which type fits your study is an important methodological decision.
1. Simple Hypothesis
Predicts a relationship between one independent variable and one dependent variable.
Students who eat breakfast perform better in morning exams than students who skip breakfast.
2. Complex Hypothesis
Predicts a relationship involving two or more independent or dependent variables.
Students who exercise daily and maintain consistent sleep schedules will report lower levels of anxiety and higher academic performance than those who do neither.
3. Directional Hypothesis
Specifies not just that a relationship exists, but in which direction. Uses terms like "greater than," "less than," "increases," or "decreases."
Longer daily screen time is associated with lower sleep quality among adolescents.
4. Non-Directional Hypothesis
States that a relationship exists between variables without predicting the direction. Used when there is insufficient prior evidence to justify a directional prediction.
There is a relationship between social media use and academic performance among university students.
5. Null Hypothesis (H₀)
Claims that there is no relationship between the variables or that any observed difference is due to chance. The null hypothesis is what statistical tests aim to reject.
There is no significant difference in exam scores between students who use flashcards and those who use written notes.
6. Alternative Hypothesis (H₁ or Hₐ)
The opposite of the null hypothesis. States that a relationship does exist. If the null hypothesis is rejected, the alternative hypothesis is supported.
Students who use flashcards will achieve significantly higher exam scores than those who use written notes.
7. Associative and Causal Hypotheses
An associative hypothesis predicts that a change in one variable is associated with a change in another, without claiming causation.
Higher physical activity levels are associated with lower body mass index in adults.
A causal hypothesis predicts a direct cause-and-effect relationship between variables, typically supported by experimental design.
A 12-week aerobic exercise program causes a statistically significant reduction in resting blood pressure in adults with hypertension.
How to Write a Hypothesis
Follow these steps to move from a research question to a testable hypothesis.
In quantitative research, hypotheses are tested as a pair. The null hypothesis (H₀) states no effect or no relationship. The alternative hypothesis (H₁) states the opposite. As Simply Psychology explains, all research has an alternative hypothesis (either directional or non-directional) and a corresponding null hypothesis.
Statistical testing tries to determine whether the data provides enough evidence to reject H₀ in favor of H₁. You never prove a hypothesis. You either reject the null hypothesis (evidence supports the alternative) or fail to reject it (insufficient evidence to support the alternative).
| Null (H₀) | Alternative (H₁) | |
|---|---|---|
| Claim | No relationship or difference | Relationship or difference exists |
| What you test | Whether data is consistent with no effect | Whether data shows a significant effect |
| Outcome | Reject or fail to reject | Supported if H₀ is rejected |
Example pair:
H₀: Daily meditation has no effect on reported stress levels in graduate students. H₁: Graduate students who practice daily meditation will report lower stress levels than those who do not.
Characteristics of a Strong Hypothesis
A hypothesis is only useful if it meets certain criteria. Elsevier's guidance on writing a good hypothesis emphasizes that the following characteristics define a well-formed hypothesis:
Testable: It must be possible to collect data that either support or contradict the prediction. A hypothesis about untestable phenomena (the meaning of life, the existence of God) is not a scientific hypothesis.
Falsifiable: There must be a conceivable outcome that would prove the hypothesis wrong. As Britannica notes on scientific hypotheses, two key features of a scientific hypothesis are falsifiability and testability. If no evidence could possibly refute it, it is not a scientific hypothesis.
Specific: The variables, population, and predicted relationship should be clearly defined. Vague hypotheses produce vague studies.
Based on prior knowledge: A hypothesis is not a random guess. It should follow logically from existing theory, prior research, or systematic observation.
Stated in declarative form: A hypothesis is a statement, not a question. "Does stress affect memory?" is a research question. "Acute stress impairs short-term memory recall in adults" is a hypothesis.
Ethical: It should be possible to test the hypothesis without causing harm to participants.
From Research Question to Hypothesis with Academly
The hardest part of writing a hypothesis is often not the writing itself. Having a clear, focused research question to start from is the real challenge. Without a precise question, the hypothesis becomes vague. Without a vague hypothesis, the study has no direction.
Academly's New Thesis Wizard generates focused research questions, a relevance statement, and a methodology recommendation from just a title and optional context. Enter your topic, click "Generate suggestions," and Academly produces three thesis directions, each with research questions, relevance, and a suggested method:
From there, formulating a testable hypothesis becomes a much smaller step. If Academly suggests "This study will employ a systematic review of existing literature and case studies," you know you need an associative or descriptive hypothesis rather than a causal one. If it suggests "semi-structured interviews and surveys," a mixed methods hypothesis fits. The methodology recommendation directly informs which type of hypothesis is appropriate.
Once your hypothesis is clear, Academly's Methods & Approach module helps you build the methodology chapter around it, including your philosophical stance, research design, and data collection strategy.
Step 1: Start with your research question. Your hypothesis is the predicted answer to your research question. If your question is "Does caffeine improve cognitive performance in sleep-deprived students?", your hypothesis predicts the answer.
Step 2: Review the existing literature. What do prior studies suggest? A hypothesis grounded in existing evidence is more credible than one invented from intuition. It also positions your study within the ongoing scholarly conversation.
Step 3: Identify your variables. Decide what you will manipulate or observe (independent variable) and what you will measure (dependent variable). Be specific about how each will be operationalized (measured in practice).
Step 4: State the predicted relationship. Write a declarative sentence predicting how the independent variable affects the dependent variable. Be as specific as the evidence allows. If evidence points to a direction, make a directional hypothesis. If not, make a non-directional one.
Step 5: State the null hypothesis. For quantitative studies, write the corresponding null hypothesis. This is what your statistical test will attempt to reject.
Step 6: Check against the criteria. Is it testable? Falsifiable? Specific? Based on prior knowledge? Stated in declarative form? If yes, the hypothesis is ready.
Hypothesis Examples by Discipline
| Discipline | Example Hypothesis |
|---|---|
| Psychology | Students who engage in 20 minutes of aerobic exercise before an exam will score higher on cognitive performance tests than those who do not exercise. |
| Education | Students taught with project-based learning methods will demonstrate higher levels of content retention after six weeks than those taught with lecture-based methods. |
| Public Health | Adults who consume more than 5 portions of fruit and vegetables daily will have lower reported rates of respiratory illness over a 12-month period. |
| Sociology | Individuals with higher levels of social media use will report lower levels of face-to-face social satisfaction. |
| Business | Employees who receive weekly performance feedback will show higher levels of job engagement than those who receive monthly feedback. |
| Nursing | Patients who receive structured discharge instructions will have lower rates of hospital readmission within 30 days compared to those who receive standard verbal instructions only. |
Common Mistakes When Writing a Hypothesis
1. Writing a question instead of a statement
"Does caffeine improve memory?" is a research question. A hypothesis states the expected answer: "Moderate caffeine consumption (200mg) improves short-term memory recall in sleep-deprived adults."
2. Making the hypothesis too vague
"Stress affects students" is not testable. Which students? What type of stress? What outcome is being measured? A strong hypothesis names specific variables and a specific population.
3. Confusing hypothesis with theory
A theory is a well-established framework supported by decades of evidence. A hypothesis is a single prediction for a single study. Do not write "My theory is that..." when you mean "My hypothesis is that..."
4. HARKing: hypothesizing after results are known
Formulating your hypothesis after seeing the data and presenting it as if it was written beforehand is a form of research misconduct. The hypothesis must be stated before data collection.
5. Ignoring the null hypothesis
In quantitative research, failing to state the null hypothesis means failing to specify what your statistical test is actually testing. Always pair your research hypothesis with the corresponding null.
6. Making the hypothesis unfalsifiable
"Students learn better when they are happy" cannot be properly tested because "happy" and "learn better" are undefined. Every variable in a hypothesis must be operationalized, meaning defined in terms of how it will be measured.
Summary
A hypothesis is a specific, testable, declarative prediction about the relationship between variables in a study, stated before data collection begins. It differs from a research question (which asks) and a theory (which explains based on extensive evidence). The seven main types are simple, complex, directional, non-directional, null, alternative, and associative or causal. A strong hypothesis is testable, falsifiable, specific, grounded in prior evidence, and stated as a declarative statement. In quantitative research, the null and alternative hypotheses always work as a pair. Writing a strong hypothesis before beginning data collection is one of the most important foundations of rigorous academic research.