Probability of observing data as extreme as what was measured if the null hypothesis were true; used to assess statistical significance.