Using F - Test for the indirect detecting of the autocorrelation of random observations

A proposed test is realized by the separation of N registered observations into k shorter series containing n observations (N = n·k), estimation and comparing the variance of the mean values of these series with an average of their variances. If the ratio of these variance estimators after normalization by the number of degrees of freedom k - 1 and N - k is greater than the critical value F_critic(1 - α, n - 1, N - n) from the F - distribution (α is a significance level) then it testifies to the auto correlation of observations. The effectiveness of the proposed test is investigated by the Monte-Carlo method.