statistical procedures in which the nature of the data being analyzed is such that certain common assumptions about the distribution of the attribute(s) in the population being tested (e.g., normality, homogeneity of variance) are not necessary or applicable. Compare parametric statistics.