An educated guess that can be tested is called a hypothesis. In science, a hypothesis is a proposed explanation for something you’ve observed, stated in a way that allows you to design an experiment or study to check whether it’s correct. It’s the starting point of the scientific method, sitting right between asking a question and planning an investigation.
What Makes a Hypothesis More Than a Guess
Calling a hypothesis an “educated guess” is a common shorthand, but it undersells what a hypothesis actually is. A guess can be random. A hypothesis is built on prior knowledge, observation, or existing evidence. You notice something in the world, ask a question about it, and then propose an explanation based on what you already know. That explanation is your hypothesis.
More precisely, a hypothesis is a proposed mechanism for a naturally occurring event or a predicted outcome of an intervention. It states an expected relationship between variables: one thing you change or measure (the independent variable) and another thing you expect to be affected (the dependent variable). Expressed simply: independent variable → dependent variable = hypothesis.
Scientists at UC Berkeley’s Understanding Science project make an even sharper distinction. They argue that hypotheses are not guesses at all, of the wild or educated variety. They’re reasoned explanations for a fairly narrow set of phenomena, grounded in what’s already known and structured so they can be checked against reality.
Why “Testable” Is the Key Word
The defining feature of a scientific hypothesis is that it can be tested. Unless you can design an observation or experiment that could show your hypothesis is wrong, it isn’t really a hypothesis in the scientific sense. This idea, called falsifiability, was developed by philosopher Karl Popper, who argued that what separates science from non-science is precisely this willingness to make claims that future observations might reveal to be false.
A testable hypothesis has to meet a few practical criteria. It needs to be testable with available technology and current scientific understanding. The test also has to be ethically acceptable. If there’s no realistic, ethical way to gather evidence for or against a statement, it doesn’t function as a scientific hypothesis, no matter how reasonable it sounds.
Here’s the difference in practice. “Plants exposed to low temperatures will show changes in leaf color” is testable. You can set up two groups of plants, expose one to cold, and measure what happens. But a statement like “the universe has a purpose” isn’t testable, because there’s no observation that could confirm or rule it out. Similarly, “playing the lottery will make you rich” is a simple prediction, not a hypothesis, because it doesn’t propose a mechanism or a relationship you can investigate systematically.
How to Write a Testable Hypothesis
The most common format for a hypothesis is the “if-then” statement. The “if” portion contains the proposed relationship you want to test, and the “then” portion is your predicted result. This structure forces you to be specific about what you’re changing and what you expect to happen.
A few examples across different fields:
- Biology: “If leaf color change is related to temperature, then exposing plants to low temperatures will result in changes in leaf color.”
- Health science: “If skin cancer is related to ultraviolet light, then people with high UV exposure will have a higher frequency of skin cancer.”
Notice that each example identifies a specific cause, a specific effect, and a way you could measure the outcome. That measurability is what transforms a hunch into something science can work with.
Where a Hypothesis Fits in the Scientific Method
The scientific method follows a sequence, and hypothesis formation comes early. The standard steps are: observe nature, ask questions, develop a hypothesis, plan an investigation, assemble data, analyze the data, construct explanations, communicate conclusions, and then pose new questions. Your hypothesis is essentially the bridge between curiosity and action. It gives your investigation a direction and a specific prediction to evaluate.
Once you have a hypothesis, the next step is designing a test. In formal research, this involves setting up two competing statements: a null hypothesis, which assumes there is no relationship between your variables (nothing is happening), and an alternative hypothesis, which states what you actually expect the data to show. The experiment then tries to determine which one the evidence supports. If your data contradicts the null hypothesis, that’s evidence in favor of your alternative hypothesis. If not, you revise and try again.
How Hypotheses Relate to Theories and Laws
A common misconception is that hypotheses “grow up” into theories, which then become laws. That’s not how it works. Hypotheses, theories, and laws are all scientific explanations, but they differ in scope, not in rank.
A hypothesis explains a fairly narrow set of phenomena. A theory is a broad explanation that often integrates and generalizes many hypotheses. The theory of evolution, for instance, ties together countless individual hypotheses about genetics, natural selection, and adaptation. A scientific law, meanwhile, describes an observed relationship between phenomena, often in mathematical terms, like the law of gravity. A hypothesis doesn’t become a theory, and a theory doesn’t become a law. They’re different tools for different jobs.
What Makes a Hypothesis Fail
Not every proposed explanation survives testing, and that’s the point. A hypothesis that can’t possibly be proven wrong isn’t useful to science. Popper noted that adherents of pseudoscientific ideas routinely adjust their claims to fit whatever is observed, making their ideas permanently immune to disproof. Genuine scientific hypotheses take the opposite approach: they stick their neck out with a specific, risky prediction and let the evidence decide.
A hypothesis can also fail before you even test it if it’s too vague to measure, if it involves variables you can’t isolate, or if testing it would require an unethical experiment. The most productive hypotheses are specific, measurable, and tied to variables you can realistically manipulate or observe. When a hypothesis does get disproven, that’s not a dead end. It’s information. You use the results to refine your explanation, form a new hypothesis, and test again. The scientific method is built as a cycle, with each answer generating the next question.