The scientific method provides a systematic framework for testing claims about the world: observation leads to questions, which generate hypotheses, which are tested by experiment, with the resulting data analyzed to draw conclusions. A hypothesis is a testable, falsifiable prediction about the relationship between variables, formulated before experimentation. Karl Popper emphasized falsifiability—the requirement that a statement admit some possible observation or experiment that could prove it false—as the defining feature that separates science from non-science. Across this process, the burden of proof falls on the claimant: the person making an assertion must provide evidence, rather than placing the obligation on others to disprove it.
Experiments gain their power through control. A controlled experiment manipulates one variable (the independent variable) while holding others constant, comparing results against a control group to isolate causal effects. Statistical inference relies on the null hypothesis, the default assumption of no effect or no relationship between variables, with experiments aiming either to reject it or to fail to reject it. Rejecting a true null yields a Type I error, or false positive, while failing to reject a false null yields a Type II error, or false negative. In interpreting results, it is essential to distinguish correlation—two variables tending to occur together—from causation, where one variable directly produces the other. Occam's razor guides scientists toward the simplest explanation that accounts for the evidence, reducing unnecessary assumptions ("Do not multiply entities beyond necessity").
Scientific claims are further strengthened by community practices such as peer review, where independent experts evaluate methodology, accuracy, and significance before publication, and reproducibility, the requirement that experiments be repeatable by others with the same results. Together, these features—falsifiability, controlled experimentation, null hypothesis testing, peer review, reproducibility, and Occam's razor—form a self-correcting system that distinguishes empirical inquiry from mere speculation.