Abstract
Randomness tests versus F-distribution when the measurement scale is discrete. The aim of this research is to examine the influence of sample size on the statistical validity of experimental designs (i.e. assuming random assignment and direct manipulation) when the dependent variable is discrete and measured in an ordinal scale, which is distributed as a rectangular distribution —any point of the sample space is equally likely. The dependent variables in the psychological studies always are not continuous and, even in this case, frequently, it is difficult to guaranteed the normality assumption. In this paper we constructed a sample distribution of the F statistic for the ordinal and discrete scale, and we compare this distribution which the classical distributions which assume that the dependent variable is normally distributed.
