What Is a Hypothesis?

Direct Answer

A hypothesis is a tentative, testable statement that proposes an explanation for an observed phenomenon or predicts a relationship between variables. It is the working claim of science: the proposition that observation and experiment are designed to evaluate. The word comes from the Greek hypothesis, literally "a placing under" — the foundation laid under an argument or investigation. A scientific hypothesis is not a guess in the ordinary sense. It is an informed conjecture, built from background knowledge, that satisfies two requirements. It must be precise enough to generate specific, observable predictions, and it must be falsifiable enough that some possible observation could show it to be wrong. "Stress causes headaches" is a hypothesis only when it is sharpened into a testable form: "Workers who report high stress will report more headaches than matched low-stress workers, under controlled conditions."

Historical Context

The concept of the hypothesis has a long history. Aristotle used hypothesis for the premises assumed in demonstration, and the term passed through the medieval logic of inquiry into early modern science. Francis Bacon, in the Novum Organum, warned against leaping too quickly from observations to generalizations and demanded that hypotheses be disciplined by systematic experiment. René Descartes, in the Discourse on Method, treated hypotheses as explanatory models to be tested by their consequences. The nineteenth century brought the logic of hypothesis into focus. Charles Sanders Peirce named the creative inference by which hypotheses are generated abduction: the leap from surprising facts to a plausible explanation. Karl Popper then supplied the modern standard in The Logic of Scientific Discovery: a hypothesis earns scientific status not by being verified but by being falsifiable — by taking the risk of being refuted.

Key Distinctions

Three distinctions matter. Hypothesis versus theory: a hypothesis is a single, focused, tentative claim; a theory is a broad, well-tested explanatory framework supported by many hypotheses and lines of evidence — the theory of evolution, for example, contains countless specific hypotheses. Hypothesis versus law: a law of science is a statement of a regular relationship (such as the ideal gas law); it describes what happens, while a hypothesis explains why it happens, and hypotheses are proposed to explain laws. Everyday guess versus scientific hypothesis: a guess need not be testable; a hypothesis must specify the evidence that would confirm or refute it. In statistics, this takes the formal shape of the null hypothesis — the default assumption of no effect — which a study attempts to reject with data, alongside the alternative hypothesis the researcher actually expects.

Contemporary Debates

Modern debates about hypotheses center on the logic of testing. The Duhem-Quine thesis, associated with Willard Van Orman Quine, holds that a hypothesis is never tested in isolation: it always depends on auxiliary assumptions about instruments, conditions, and background theory. When a prediction fails, the logic alone cannot tell us whether to reject the hypothesis or repair the auxiliary. Popper acknowledged this by distinguishing bold, refutable conjectures from ad hoc adjustments that shield a hypothesis from test — a distinction that remains debated today. A second debate concerns hypothesis generation: since the era of Peirce, philosophers have recognized that there is no algorithm for inventing hypotheses; the creative leap remains irreducible, which is why discovery is sometimes said to have "no logic" even though testing does. Finally, the replication crisis has pushed researchers to pre-register hypotheses, so that confirmations cannot be manufactured after the fact by reinterpreting the data.

Practical Implications

In practice, a good hypothesis does three jobs. It focuses inquiry: "What exactly do I expect, and in which population?" It designs the test: the hypothesis dictates the variables to measure, the controls to include, and the analysis to run. And it disciplines interpretation: because the hypothesis was stated in advance, the data can be read honestly, and the risk of confirmation bias is reduced. In everyday reasoning, the same discipline applies: when you suspect a cause — "this change in routine is why I sleep worse" — you can sharpen it into a testable hypothesis, ask what evidence would falsify it, and look for that evidence rather than only for support. This habit of converting vague suspicions into falsifiable, testable claims is the core habit of critical thinking in any domain.

Related Concepts

• What Is the Scientific Method? — the cycle of inquiry hypotheses belong to
• What Is Falsification? — the standard hypotheses must meet
• What Is a Controlled Experiment? — how hypotheses are tested
• What Is a Theory? — what well-tested hypotheses grow into
• Correlation vs Causation — the pitfall hypotheses must respect

Further Learning

• Read the SEP entry on the Scientific Method
• Explore abduction with the IEP article on Abduction
• See the standard of falsifiability in The Logic of Scientific Discovery
• Visit the Critical Thinking & Logic collection for the full learning path