Reproducibility and performance in random number generation.

We investigated the reproducibility of random number generation task (RNG) in 18 healthy subjects. They were tested using RNG on separate days, and the degree of randomness were measured according to RNG Index, Null score and Phase. The correlation coefficient between two repeated scores (test and retest) was significantly high for all randomness measures, and RNG Index and Null score remained unchanged. Next, 10 of 18 subjects were tested again after they were provided feedback on one statistical analysis of randomness (RNG Index). The correlations between the three measures became insignificant, and both RNG Index and Null score decreased significantly after the feedback. In conclusion, RNG is a useful neuropsychological test to evaluate cognitive dysfunction which is prone to change, since we can employ it repeatedly at short intervals.