What does statistically significant actually mean?

Statistical significance is a statement about probability under a specific assumption. Researchers begin with a null hypothesis, that there is no real effect, then calculate a p-value: the probability of observing results at least as extreme as theirs if that null hypothesis were true. By convention, a p-value below 0.05 is called significant, meaning results this extreme would occur less than five percent of the time by chance alone if there were genuinely no effect. The 0.05 threshold is an arbitrary convention inherited from early 20th-century statistics, not a law of nature.

The misreadings are consequential. Significant does not mean large or important, since with a big enough sample, a trivially small effect can clear the threshold, which is why effect size and confidence intervals matter more than the p-value for judging whether a finding means anything. It also does not mean the result is definitely real, because running many analyses guarantees some false positives, the practice known as p-hacking, and because a single significant result is far weaker evidence than replication. Nor does a non-significant result prove no effect exists; it may just mean the study lacked the power to detect one. The replication crisis across psychology, medicine, and other fields grew substantially out of these misunderstandings, which is why many journals and statisticians now push for reporting effect sizes, confidence intervals, and preregistered analyses rather than treating 0.05 as a verdict.