n. (in statistics) the degree to which an observed relationship between two test groups is unlikely to have occurred by chance alone. An initial assumption is made that there is no meaningful difference between the groups or conditions under investigation (the null hypothesis). This can be tested using various statistical procedures, and a calculation that there is a probability of less than 5% (P value <0.05) is usually considered sufficient to reject the null hypothesis: the observed difference is statistically significant. Some tests are parametric, based on the assumption that observations will be distributed in a normal or Gaussian *distribution, where 95% of observations lie within two *standard deviations of the mean (Student’s t test to compare means). Nonparametric tests (Mann–Whitney U tests) make no assumptions about distribution patterns. See also standard error of the mean.
Health Encyclopedia