A relationship between two groups of observations indicating that the difference between them (e.g. between the percentages of smokers and nonsmokers respectively who die from lung cancer) is unlikely to have occurred by chance alone. An assumption is made that there is no difference between the two populations from which the two groups come {null hypothesis). This is tested, and a calculation indicating that there is a probability of less than 5% (P<0.05) that the observed difference or a larger one could have arisen by chance is regarded as being statistically significant and the null hypothesis is rejected. Some tests are parametric, based on the assumption that the range of observations are distributed by chance in a normal or Gaussian distribution, with 95% 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.