On Independence and Sensitivity of Statistical Randomness Tests

Statistical randomness testing has significant importance in analyzing the quality of random number generators. In this study, we focus on the independence of randomness tests and its effect on the coverage of test suites. We experimentally observe that frequency, overlapping template, longest run of ones, random walk height and maximum order complexity tests are correlated for short sequences. We also proposed the concept of sensitivity, where we analyze the effect of simple transformations on output p-values. We claim that whenever the effect is significant, the composition of the transformation and the test may be included to the suite as a new test