# What is a p-value? An expert explains the most misunderstood number in science
The p-value stands as one of science's most frequently misinterpreted tools, with consequences that ripple through research, policy, and education. This single metric has shaped how scientists communicate findings and how the public understands evidence.
A p-value measures the probability of observing results as extreme as those found in a study, assuming the null hypothesis is true. It does not measure the probability that a hypothesis is correct. This distinction matters. A p-value of 0.05, the conventional threshold for "significance," means there is a 5 percent chance of seeing data this extreme if no real effect exists. It does not mean a 95 percent probability that the finding is true.
The misunderstanding runs deep across disciplines. Researchers often interpret a p-value below 0.05 as proof their hypothesis is correct. Journal editors and funding bodies have reinforced this binary thinking by treating p-values as gatekeepers. Either a study "passes" the significance test or it fails. This framework has distorted incentives in research, encouraging scientists to chase statistically significant results rather than pursue meaningful ones.
The consequences for education research have been particularly troubling. Studies showing small improvements in test scores or instructional methods gain visibility and adoption if they cross the p-value threshold. Studies showing null results disappear into desk drawers. Educators implement interventions based on statistically significant findings that may have minimal real-world impact.
Institutions like the American Statistical Association have called for rethinking reliance on p-values. Researchers increasingly favor reporting confidence intervals, effect sizes, and pre-registered study designs. These approaches provide clearer pictures of what data actually show.
The p-value's persistence reflects a deeper problem: scientists and institutions want simple answers. Reducing complex evidence to a single number