A hypothesis is a testable and falsifiable proposed explanation or prediction about an observation, question, or relationship between variables. In science, a hypothesis provides a specific idea that researchers investigate using observations, experiments, or other evidence. A good scientific hypothesis is based on existing knowledge, makes predictions that evidence can support or contradict, and helps guide further investigation.
Hypotheses are central to the scientific method, but they are not simply guesses. They connect observations and questions with evidence. A hypothesis may ultimately be supported, rejected, or revised, but even an unsupported hypothesis can advance science by ruling out an explanation or suggesting new questions.
Key Takeaways: Hypothesis
- A hypothesis is a testable and falsifiable proposed explanation or prediction.
- A scientific hypothesis is based on observations, prior knowledge, or existing evidence rather than a random guess.
- A good hypothesis is specific enough to test and makes predictions that evidence can potentially contradict.
- Common types include null, alternative, directional, non-directional, simple, complex, associative, and causal hypotheses.
- Scientists test hypotheses using experiments, observations, models, or existing data.
- Evidence can support a hypothesis, but scientific testing generally does not prove a hypothesis with absolute certainty.
- A hypothesis differs from a scientific theory: a theory is a broad, well-supported explanatory framework.
What Is a Hypothesis?
A hypothesis is a proposed explanation for an observation or a proposed relationship between variables that can be evaluated using evidence.
In experimental science, a hypothesis often predicts how changing the independent variable affects the dependent variable.
For example:
Hypothesis: Increasing the amount of sunlight a bean plant receives each day increases its growth rate.
Here:
- Independent variable: Amount of sunlight
- Dependent variable: Growth rate of the bean plant
- Prediction: Plants receiving more sunlight grow faster, within the conditions being tested.
Not every scientific hypothesis involves manipulating variables. In astronomy, geology, evolutionary biology, and other observational sciences, researchers often test hypotheses by comparing predictions with observations that occur naturally.
For example:
Hypothesis: A particular crater formed through a meteorite impact rather than volcanic activity.
Scientists cannot repeat the original event, but the competing explanations predict different geological features, minerals, and structures that researchers can examine.
Hypothesis Meaning
The word hypothesis comes from the Greek hypóthesis, meaning a foundation, supposition, or something placed beneath an argument. In ordinary language, a hypothesis can mean an assumption or proposed explanation.
Science uses the term more narrowly. A scientific hypothesis must connect with observable evidence in a way that permits meaningful testing.
This distinction is important because not every statement, prediction, or explanation qualifies as a scientific hypothesis.
For example:
Scientific hypothesis: Adding nitrogen fertilizer increases the average growth rate of corn plants under specified growing conditions.
Not a scientific hypothesis: Corn plants grow better when they are happy.
The second statement does not define “happy” in an objective, measurable way, so it is not readily testable as written.
Why Is a Hypothesis Important?
A hypothesis turns a broad scientific question into an idea that researchers can investigate systematically.
For example, consider the question:
Does temperature affect the rate of a chemical reaction?
This is a useful research question, but it does not state an expected relationship.
A hypothesis makes the idea more specific:
Increasing temperature increases the reaction rate because a greater fraction of reactant particles have enough energy to overcome the activation energy barrier.
Researchers can then design an experiment, identify variables, collect measurements, and determine whether the observations agree with the prediction.
Hypotheses are important because they:
- Focus scientific investigations.
- Identify relationships that researchers can test.
- Help determine which variables or observations should be measured.
- Generate predictions.
- Connect existing knowledge with new experiments.
- Provide a framework for analyzing results.
- Allow competing explanations to be compared.
- Lead to new questions and revised explanations.
Importantly, a hypothesis does not need to be correct to be useful. A well-designed experiment that contradicts a hypothesis still provides scientific information.
Hypotheses and the Scientific Method
The scientific method is not a single rigid sequence that every scientist follows, but hypotheses often play a central role in scientific investigation.
A simplified sequence is:
- Make an observation.
- Ask a question.
- Research what is already known.
- Propose a hypothesis.
- Derive a testable prediction.
- Test the prediction using an experiment or observations.
- Analyze the results.
- Support, reject, or revise the hypothesis.
- Communicate the results and investigate further questions.
Real scientific research is usually less linear. Scientists may revise hypotheses while designing experiments, discover unexpected observations, develop several competing hypotheses, or use existing data to test new ideas.
Also, not every scientific investigation begins with a formal hypothesis. Exploratory and descriptive research may first establish patterns that later lead to hypotheses.
History of the Hypothesis
The concept of proposing explanations and evaluating them using reason and observation predates modern science.
Ancient Greek philosophers developed systematic approaches to reasoning about nature. Aristotle distinguished different forms of reasoning and emphasized observation, although his approach differed substantially from modern experimental science.
During the Scientific Revolution of the 16th and 17th centuries, systematic observation, mathematical reasoning, and controlled experimentation became increasingly important. Francis Bacon promoted inductive reasoning based on observations and experiments, while scientists such as Galileo Galilei combined measurement, experimentation, and mathematical prediction.
Isaac Newton was famously cautious about speculative explanations. In the General Scholium added to the second edition of the Principia in 1713, he wrote hypotheses non fingo, commonly translated as “I frame no hypotheses.” In context, Newton was rejecting unsupported speculation about the cause of gravity rather than rejecting testable scientific reasoning.
During the 19th and 20th centuries, philosophers and scientists increasingly examined the role of hypotheses in scientific reasoning. William Whewell emphasized the formation and testing of explanatory ideas, while later thinkers distinguished more clearly between hypotheses that could be scientifically tested and claims insulated from evidence.
In the 20th century, philosopher Karl Popper strongly emphasized falsifiability: a scientific hypothesis should make predictions that could, in principle, turn out to be wrong. Although modern philosophy of science is more nuanced than any single criterion, testability and vulnerability to contrary evidence remain fundamental features of useful scientific hypotheses.
Statistics also transformed hypothesis testing during the 20th century. The work of Ronald Fisher, Jerzy Neyman, and Egon Pearson helped establish formal methods for evaluating statistical hypotheses. Statistical hypothesis testing is related to, but not identical with, the broader scientific concept of proposing and testing explanatory hypotheses.
Characteristics of a Good Hypothesis
A good scientific hypothesis has several important characteristics.
Testable
There must be some practical or logical way to compare the hypothesis with evidence.
Good: Salt concentration affects the germination rate of radish seeds.
Poor: Invisible forces that cannot be detected determine whether seeds germinate.
Falsifiable
There must be some possible observation that would count against the hypothesis.
A statement that is compatible with every imaginable outcome cannot be meaningfully tested.
Specific
The hypothesis should clearly state the proposed relationship or explanation.
Vague: Light affects plants.
Better: Increasing daily light exposure from 6 hours to 12 hours increases the average growth rate of bean seedlings under otherwise identical conditions.
Measurable
Important variables should be observable or operationally defined whenever possible.
Instead of saying that plants become “healthier,” specify measurable quantities such as height, biomass, leaf number, chlorophyll content, or survival rate.
Based on Existing Knowledge
A hypothesis normally grows from observations, previous research, established principles, or a logical argument.
This is why “educated guess” is only a rough description. A scientific hypothesis should have a rational basis.
Relevant to the Question
The hypothesis should actually address the research question being investigated.
Simple Enough to Test
A useful hypothesis avoids unnecessary assumptions and excessive numbers of variables. Simpler hypotheses are often easier to test and interpret.
This does not mean that nature itself must be simple. Some systems genuinely require complex hypotheses.
Predictive
A strong hypothesis implies what researchers should expect to observe if the proposed explanation is correct.
Hypothesis vs Prediction
A hypothesis and a prediction are closely related but are not exactly the same thing.
A hypothesis proposes an explanation or relationship. A prediction states an expected observation under particular conditions.
For example:
Hypothesis: Higher temperatures increase the rate of an enzyme-catalyzed reaction up to the enzyme’s optimum temperature.
Prediction: If identical reaction mixtures are tested at 20 °C and 30 °C, the reaction at 30 °C will proceed faster, assuming both temperatures are below the enzyme’s optimum and other conditions remain constant.
A single hypothesis can generate multiple predictions.
Main Types of Hypotheses
Hypotheses can be classified in several ways. These categories overlap, so a single hypothesis may belong to more than one type.
| Type | Description | Example |
|---|---|---|
| Null hypothesis (H₀) | States that there is no effect, difference, or association of the specified kind | Fertilizer treatment does not change mean plant height compared with the control. |
| Alternative hypothesis (H₁ or Hₐ) | States that an effect, difference, or association exists | Fertilizer treatment changes mean plant height compared with the control. |
| Directional hypothesis | Predicts the direction of an effect or relationship | Plants receiving fertilizer grow taller than untreated plants. |
| Non-directional hypothesis | Predicts a difference or relationship without specifying its direction | Fertilizer treatment affects plant height. |
| Simple hypothesis | Relates one independent variable to one dependent variable | Increasing light intensity increases photosynthetic rate. |
| Complex hypothesis | Involves multiple independent or dependent variables | Light intensity and CO₂ concentration affect plant growth and biomass. |
| Associative hypothesis | Predicts that variables vary together without necessarily claiming causation | Higher air temperature is associated with higher electricity use during summer. |
| Causal hypothesis | Proposes that a change in one variable causes a change in another | Increasing light intensity increases photosynthetic rate under specified conditions. |
Null Hypothesis
The null hypothesis, written H₀, usually represents the absence of a specified effect, difference, or association.
For example:
H₀: Mean germination percentage is the same for seeds exposed to the treatment and seeds in the control group.
The null hypothesis is particularly important in statistical hypothesis testing.
Alternative Hypothesis
The alternative hypothesis, written H₁ or Hₐ, represents an effect, difference, or association considered as an alternative to the null hypothesis.
For example:
Hₐ: Mean germination percentage differs between treated seeds and control seeds.
Depending on the research question, the alternative hypothesis may be directional or non-directional.
Directional Hypothesis
A directional hypothesis predicts not only that variables differ or relate, but also the expected direction.
Example:
Students who sleep eight hours before an exam score higher, on average, than students who sleep four hours.
Non-Directional Hypothesis
A non-directional hypothesis predicts a difference or relationship without specifying which direction it takes.
Example:
Exam scores differ between students who sleep eight hours and students who sleep four hours before an exam.
Simple Hypothesis
A simple hypothesis proposes a relationship between one independent variable and one dependent variable.
Example:
Increasing water temperature increases the dissolving rate of sugar.
Complex Hypothesis
A complex hypothesis involves two or more independent variables, dependent variables, or both.
Example:
Temperature and light intensity affect both the growth rate and flowering time of a plant.
Associative Hypothesis
An associative hypothesis predicts that two variables are related without necessarily proposing that one causes the other.
Example:
Daily temperature is positively associated with ice cream sales.
This statement does not claim that buying ice cream raises the temperature or that temperature alone determines ice cream sales.
Causal Hypothesis
A causal hypothesis proposes a cause-and-effect relationship.
Example:
Increasing the concentration of a reactant increases reaction rate when other relevant conditions are held constant.
Demonstrating causation generally requires stronger evidence than demonstrating an association. Experimental controls, randomization, replication, and consideration of confounding variables may all be important.
Research Hypothesis vs Statistical Hypothesis
It is useful to distinguish a research hypothesis from a statistical hypothesis.
A research hypothesis describes an expected scientific relationship:
Higher nitrogen availability increases plant growth.
A statistical hypothesis expresses a claim about a population or probability distribution in a form suitable for statistical testing.
For example:
H₀: μ₁ = μ₂
Hₐ: μ₁ ≠ μ₂
Here, μ₁ and μ₂ represent population means.
The scientific hypothesis motivates the investigation, while statistical hypotheses provide a mathematical framework for analyzing particular data.
How to Propose a Hypothesis
Writing a good hypothesis starts before writing the sentence itself.
1. Start With an Observation or Question
Suppose a student notices that seedlings near a window appear taller than seedlings farther from the window.
The research question might be:
Does the amount of light affect seedling growth?
2. Research the Topic
Find out what is already known. Prior knowledge about photosynthesis suggests that light availability can affect plant growth.
Research helps avoid proposing hypotheses that have no scientific basis or simply repeat well-established findings without a useful reason.
3. Identify the Variables
Determine what changes and what is measured.
Independent variable: Amount or duration of light exposure
Dependent variable: Seedling growth
Other important conditions, such as plant species, temperature, soil, water, and experimental duration, should be considered as control variables where appropriate.
4. State the Expected Relationship
A basic hypothesis might be:
Increasing daily light exposure increases the growth of bean seedlings.
5. Make the Variables Measurable
Improve the hypothesis by specifying what “growth” means.
Bean seedlings receiving 12 hours of light per day will have a greater mean increase in dry biomass over three weeks than seedlings receiving 6 hours per day, under otherwise identical growing conditions.
This version is easier to test objectively.
6. Check Whether It Can Be Falsified
Ask:
What result would contradict this hypothesis?
If seedlings receiving 12 hours of light do not show greater mean biomass gain under the stated conditions, the predicted relationship is not supported by the experiment.
If no conceivable observation could count against the statement, it needs revision.
The “If, Then, Because” Hypothesis
Students are often taught to write hypotheses using this format:
If [independent variable changes], then [dependent variable will respond], because [scientific reasoning].
For example:
If the temperature of water increases, then sugar will dissolve faster because higher temperature increases molecular motion and the frequency of interactions between water molecules and the sugar surface.
This format is useful for learning because it forces the writer to identify variables, make a prediction, and provide reasoning.
However, a scientific hypothesis does not have to use an “if, then” sentence. Professional scientific hypotheses are often expressed as straightforward declarative statements.
For example:
Sugar dissolves faster in warmer water than in colder water under otherwise equivalent conditions.
The quality of a hypothesis depends on its scientific content, not its grammatical format.
Hypothesis Examples
Here are examples from several branches of science.
Biology
Increasing soil salinity decreases the germination percentage of bean seeds.
Chemistry
Increasing reactant concentration increases the rate of this reaction when temperature and other conditions remain constant.
Physics
Increasing the length of a simple pendulum increases its period when other relevant conditions remain constant.
Earth Science
Streams with greater average flow velocity transport larger sediment particles than slower-moving streams under comparable conditions.
Environmental Science
Ponds receiving greater nutrient inputs have higher average algal biomass than comparable ponds receiving lower nutrient inputs.
Psychology
Participants exposed to acute time pressure make more errors on a specified attention task than participants given sufficient time.
Astronomy
An observed periodic decrease in a star’s brightness is caused by a planet transiting across the star’s disk.
This last example illustrates an important point: scientists can test hypotheses about phenomena they cannot manipulate experimentally. The transit hypothesis predicts specific features of the star’s light curve and potentially other observations, such as periodic radial-velocity changes.
Hypothesis vs Theory vs Law
One of the most common sources of confusion is the relationship between hypotheses, theories, and laws.
| Concept | Meaning |
|---|---|
| Hypothesis | A testable proposed explanation or relationship |
| Theory | A broad explanatory framework strongly supported by multiple lines of evidence |
| Scientific law | A statement or mathematical relationship describing a consistent pattern in nature |
| Prediction | A specific expected outcome derived from a hypothesis, theory, or model |
A hypothesis does not become a theory simply by accumulating enough evidence.
Scientific theories generally have much broader explanatory scope than individual hypotheses. For example, evolutionary theory explains a vast range of biological observations and generates many specific hypotheses that researchers can test.
Similarly, a theory does not “graduate” into a law. Theories and laws serve different purposes: broadly speaking, theories explain patterns and mechanisms, while laws describe regularities or relationships.
What Happens After Testing a Hypothesis?
After collecting and analyzing evidence, researchers compare the results with the predictions of the hypothesis.
Several outcomes are possible.
The results are consistent with the hypothesis.
The evidence supports the hypothesis under the conditions tested. Further testing may strengthen or challenge it.
The results contradict the hypothesis.
Researchers may reject or modify the hypothesis, examine assumptions, or consider alternative explanations.
The results are inconclusive.
The experiment may lack sufficient statistical power, measurements may be too uncertain, or the tested conditions may not distinguish among competing explanations.
Scientific conclusions are therefore better expressed as supported, not supported, rejected under specified conditions, or consistent with the evidence rather than simply “proved.”
Can a Hypothesis Be Proven?
No. Strictly speaking, empirical evidence usually supports a scientific hypothesis rather than proving it with absolute certainty.
Repeated successful tests increase confidence in an explanation, but new observations may reveal exceptions, limitations, or a better explanation.
This does not mean that all scientific conclusions are equally uncertain. Some scientific findings have extraordinarily strong evidential support. Rather, scientific conclusions remain open to revision when compelling new evidence appears.
Statistical hypothesis testing adds another important distinction. A researcher may reject the null hypothesis or fail to reject the null hypothesis at a chosen significance level. Failing to reject H₀ is not the same as proving that H₀ is true.
Common Mistakes When Writing a Hypothesis
Making It Too Vague
Poor: Fertilizer is good for plants.
Better: Under the specified growing conditions, tomato plants receiving the tested nitrogen fertilizer produce greater mean dry biomass after six weeks than untreated plants.
Writing a Question Instead of a Hypothesis
Question: Does fertilizer increase plant growth?
Hypothesis: The tested fertilizer increases mean plant growth compared with an untreated control.
Writing Only a Prediction
Prediction: The plants in Group A will be taller.
This may be useful, but a stronger hypothesis identifies the relevant relationship or proposed explanation.
Using Variables That Cannot Be Measured
Terms such as “better,” “healthier,” or “more successful” require operational definitions.
Making the Hypothesis Impossible to Falsify
A statement that accommodates every possible result cannot provide a meaningful scientific test.
Changing Several Variables Without Accounting for Them
If treatment plants receive more light, water, and fertilizer than control plants, it becomes difficult to determine which variable explains a difference in growth.
Confusing Correlation With Causation
Finding that two variables occur together does not by itself show that one causes the other.
Trying to Prove the Hypothesis
The goal of an experiment is not to produce the desired result. Researchers test predictions and evaluate evidence, including evidence that contradicts their expectations.
Common Misconceptions About Hypotheses
“A hypothesis is just a guess.”
A hypothesis can involve uncertainty, but a scientific hypothesis is an informed, testable proposition grounded in observations, evidence, or scientific reasoning.
“Every hypothesis must use ‘if, then.'”
No. This is a useful classroom format rather than a requirement of science.
“A hypothesis becomes a theory when it is proven.”
No. Hypotheses and theories differ primarily in scope and function, not simply in how certain scientists are about them.
“A theory becomes a law.”
No. Scientific theories and laws are different kinds of scientific knowledge.
“A failed hypothesis means the experiment failed.”
No. Evidence contradicting a hypothesis can be scientifically valuable. An experiment fails when its design or execution prevents it from answering the intended question, not merely when the prediction is wrong.
“Scientists always have to propose a hypothesis before collecting data.”
Not necessarily. Exploratory, observational, and descriptive research can discover patterns and generate hypotheses for later testing.
“Failing to reject the null hypothesis proves there is no effect.”
No. It means the available evidence did not justify rejecting the null hypothesis under the assumptions and criteria of the statistical analysis. An effect may be absent, too small to detect, or obscured by insufficient data or measurement uncertainty.
How to Tell Whether a Hypothesis Is Good
A quick checklist helps evaluate a proposed hypothesis:
- Does it address a scientific question?
- Is it based on observation, evidence, or reasonable scientific principles?
- Can it be tested using observations or experiments?
- Could evidence potentially show that it is wrong?
- Are the important variables clearly identified?
- Can the variables be measured or operationally defined?
- Does it make a meaningful prediction?
- Is it specific enough to guide an investigation?
- Does it avoid unnecessary assumptions?
- Can another researcher understand what is being proposed?
If the answer to several of these questions is “no,” revise the hypothesis before designing the investigation.
Frequently Asked Questions
What is a hypothesis in simple terms?
A hypothesis is a testable proposed explanation or prediction about an observation or relationship. It states what a researcher expects and provides an idea that evidence can evaluate.
What is an example of a hypothesis?
One example is: Increasing the temperature of water increases the rate at which sugar dissolves, under otherwise equivalent conditions.
Is a hypothesis an educated guess?
“Educated guess” is a common introductory definition, but it is incomplete. A scientific hypothesis is an informed and testable proposition based on observations, evidence, or scientific reasoning.
Does a hypothesis have to be testable?
A scientific hypothesis must be testable in principle. It should make predictions that can be compared with observations or experimental results.
Does a hypothesis have to be falsifiable?
A useful scientific hypothesis should permit observations that could count against it. If every possible result is compatible with a claim, evidence cannot meaningfully distinguish whether the claim is correct.
Does a hypothesis have to be an “if, then” statement?
No. “If, then” is a convenient educational format, but hypotheses can be written as ordinary declarative statements.
What is the difference between a hypothesis and a prediction?
A hypothesis proposes an explanation or relationship. A prediction describes a particular result expected if the hypothesis is correct.
What is the difference between a hypothesis and a theory?
A hypothesis is a specific proposition that can be tested. A scientific theory is a broad explanatory framework supported by extensive evidence and capable of generating many hypotheses and predictions.
What is the difference between a hypothesis and a research question?
A research question identifies what a researcher wants to investigate. A hypothesis proposes an expected answer, explanation, or relationship that can be tested.
For example:
Research question: Does salinity affect seed germination?
Hypothesis: Increasing salinity decreases the germination percentage of radish seeds.
What is the null hypothesis?
The null hypothesis, H₀, is a statistical hypothesis typically stating the absence of a specified difference, effect, or association.
What happens if a hypothesis is wrong?
Researchers may reject or revise it and investigate alternative explanations. A hypothesis that turns out to be wrong can still produce valuable scientific knowledge.
Hypothesis at a Glance
Definition: A testable and falsifiable proposed explanation or relationship.
Purpose: Guides investigation and connects scientific questions with evidence.
Based on: Observations, existing knowledge, previous research, or scientific reasoning.
Must be: Testable and capable of being contradicted by evidence.
Often identifies: Independent and dependent variables or other measurable relationships.
Possible outcomes: Supported, not supported, revised, or replaced.
Not the same as: A prediction, theory, scientific law, or research question.
References and Further Reading
- Gauch Jr, Hugh G. (2002). Scientific Method in Practice. Cambridge University Press. doi:10.1017/cbo9780511815034.011. ISBN 978-0-521-81689-2.
- Hinkelmann, Klaus; Kempthorne, Oscar (2008). Design and Analysis of Experiments, Volume I: Introduction to Experimental Design (2nd ed.). Wiley. ISBN 978-0-471-72756-9.
- Popper, Karl R. (1959). “The Logic of Scientific Discovery”. Physics Today. 12 (11): 53. doi:10.1063/1.3060577
- Urbach, Peter (1987). Francis Bacon’s Philosophy of Science: An Account and a Reappraisal. La Salle, IL: Open Court Publishing Co. ISBN 978-0-912050-44-7.