Mice in three generations were exposed to powerline EMFs to assess the effect on body weight (1). The F-test was used to compare variances in the exposed and sham-exposed animals. A total of 12 F-tests were performed, corresponding to 2 fields x 3 generations x 2 sexes. The test statistic was that F=s12/s22, where s12 was the largest sample variance in each comparison. The two-tailed rejection region of the F distribution for a single comparison was F > F0.025, which corresponds to a probability for a type-1 error of 0.05. The family-wise error rate was controlled by using the Bonferonni procedure to adjust the comparison-wise error to be 0.05/12=0.0042, which corresponded to a rejection region of F > F0.002. After controlling for the family-wise error in this manner, the occurrence of one or more significant tests with F > F0.002 would have been sufficient justification to accept the hypothesis that the field exposure affected variance. Two such cases were observed. It could be concluded, therefore, that EMF exposure affected variance.
(1) A.A. Marino. Different outcomes in biological experiments involving weak EMFs: Is chaos a possible explanation? Am. J. Physiol. 268 (Regulatory Integrative Comp. Physiol. 37: R1013-R1018, 1995.