P-values

Lead Author(s): Peter Bacchetti, PhD

These are usually interpreted as showing the strength of evidence against the possibility that there is truly no association, difference, or effect (often called the Null Hypothesis), with smaller values indicating stronger evidence. By convention, p<0.05 is considered "statistically significant" and a noteworthy finding. Keep in mind, however, that misinterpretation of p-values is rampant in health-related research, and an exclusive focus on p-values will usually lead to confusion or even mistakes when interpreting complex studies. In particular, p-values are irrelevant and should not be used for establishing negative results; see Common Biostatistical Problems, #1. We recommend always examining estimates and the confidence intervals around them, and taking these into account when interpreting results.