A hypothesis is a specific, testable prediction about the relationship between two or more variables. It’s not the same as a research question, though the two are closely related. A research question asks something broad, like “How does sleep affect test scores?” A hypothesis takes that a step further and makes a prediction: “If students sleep fewer than six hours before an exam, then their test scores will be lower than students who sleep eight hours.” That shift from asking to predicting is what makes a hypothesis distinct.
How a Hypothesis Differs From a Research Question
A research question is open-ended. It identifies a topic you want to explore and expresses a relationship between variables, typically in the form of “How does A affect B?” You use a research question when you don’t yet have enough information to predict an outcome.
A hypothesis comes after. Once you’ve done background reading or observed a pattern, you form a prediction about what you expect to find. While a research question says “I wonder if there’s a connection,” a hypothesis says “I believe the connection works like this, and here’s how I’ll test it.” A hypothesis always implies a test. If there’s no way to gather evidence for or against it, it isn’t a hypothesis in the scientific sense.
Where a Hypothesis Fits in the Scientific Method
The scientific method starts with observation. You notice something, and that leads to a question. The hypothesis is your proposed answer to that question, formed before you run any experiment. It sits between the question and the experiment itself, acting as a bridge: it tells you what to look for and what kind of data to collect.
After testing, you compare your results to your prediction. The hypothesis either holds up or it doesn’t. Either outcome is useful, because the goal isn’t to be right. It’s to learn something concrete about how the variables relate to each other.
The “If-Then” Format
The most common way to write a hypothesis is the “if-then” structure. The “if” portion states the proposed relationship, and the “then” portion predicts what will happen during testing. Both parts are necessary. A statement that only includes a prediction (“Students who sleep less will score lower”) is incomplete without identifying the relationship being tested.
Some examples:
- If skin cancer is related to ultraviolet light, then people with high UV exposure will have a higher frequency of skin cancer.
- If leaf color change is related to temperature, then exposing plants to low temperatures will result in changes in leaf color.
This format works well in science classes and lab reports because it forces you to connect your prediction to a testable cause. But in professional research, hypotheses are often written as declarative statements: “Higher UV exposure is associated with increased rates of skin cancer.” The logic is the same, just phrased differently.
Variables: The Building Blocks
Every hypothesis involves at least two variables. The independent variable is the factor you expect to have an influence. The dependent variable is the outcome you’re measuring. In the UV light example, UV exposure is the independent variable and skin cancer rates are the dependent variable.
Thinking in terms of cause and effect helps: the independent variable is the cause (or suspected cause), and the dependent variable is the effect. Your hypothesis predicts how changing one will change the other.
Types of Hypotheses
Not all hypotheses are structured the same way. The type you use depends on how much you already know about the topic.
- Directional hypothesis: Predicts the specific direction of the relationship. For example, “Increasing study time will improve exam scores.” You use this when existing theory or prior research gives you reason to expect a particular outcome.
- Non-directional hypothesis: Predicts that a relationship exists but doesn’t specify the direction. For example, “There is a difference in exam scores between students who study one hour and those who study three hours.” This is useful when previous findings are contradictory or when you’re exploring new territory.
- Associative hypothesis: Predicts that a change in one variable will coincide with a change in another, without claiming that one directly causes the other.
- Causal hypothesis: Goes further and predicts that manipulating the independent variable will directly produce a change in the dependent variable. Causal hypotheses require controlled experiments to test properly.
Null and Alternative Hypotheses
In statistics, hypotheses come in pairs. The null hypothesis states that there is no relationship or no difference between the variables. It always contains a condition of equality, essentially saying “nothing is going on here.” The alternative hypothesis is the opposite: it claims that a relationship or difference does exist.
For example, if you’re testing whether a new battery lasts longer than the advertised 900 hours, the null hypothesis says the average lifespan equals 900 hours. The alternative says it’s greater than 900 hours. Statistical tests then determine whether your data gives you enough evidence to reject the null hypothesis in favor of the alternative.
This pairing exists because science works by elimination. You don’t prove your hypothesis is true. Instead, you gather evidence that makes the “nothing is happening” explanation unlikely enough to set aside.
What Makes a Hypothesis Valid
The single most important quality of a scientific hypothesis is that it must be falsifiable. This means there has to be some possible observation or result that would prove it wrong. The philosopher Karl Popper argued that falsifiability is what separates scientific claims from non-scientific ones. A useful hypothesis is one that has been tested and hasn’t been disproven yet.
Beyond falsifiability, strong hypotheses share a few traits. They’re specific enough that you know exactly what to measure. They identify clear variables. And they’re grounded in existing knowledge rather than random guessing. The FINER framework, used in academic research, evaluates whether a research question (and by extension, its hypothesis) is feasible, interesting, novel, ethical, and relevant. A hypothesis built on a weak or vague question will be weak and vague itself.
Hypotheses Outside the Lab
Hypotheses aren’t limited to science classrooms. In business, companies use them constantly for A/B testing. A marketing team might hypothesize that changing a headline on a landing page will increase the number of people who sign up. They test two versions, measure the results, and let the data confirm or reject the prediction. Product teams test hypotheses about which features matter most to users. Advertisers test whether different ad copy reduces customer acquisition costs.
The logic is identical to a lab experiment: state what you expect to happen, design a fair test, collect data, and compare results to your prediction. Whether you’re studying cell biology or website conversions, a hypothesis gives your investigation a clear target and a way to measure success.
Turning a Question Into a Hypothesis
If you’re starting from a broad topic, the process works in stages. First, narrow your focus to a specific relationship between two variables. “Does diet affect health?” is too vague. “Does daily sugar intake above 50 grams increase the risk of developing type 2 diabetes in adults over 40?” is specific enough to test.
Next, review what’s already known. Background reading helps you form an informed prediction rather than a guess. Then, write your hypothesis using either the if-then format or a declarative statement, making sure it identifies both variables and predicts a clear, measurable outcome. If you can’t imagine what evidence would prove your hypothesis wrong, it needs to be rewritten until you can.