0

I would like to compare variances of two independent samples (of different sizes) from two unknown distributions. It seems I should use a one-sided nonparametric test. Bartlett's and Fligner's tests are two-sided, am I right? What test in R could I use to determine which sample has greater variance?

For example, consider two samples:

x <- c(1,2,3,4,5,6,7,8,9,10)
y <- c(10,2,3,4,5,6,7,8,9,100,200,56)
whuber
  • 281,159
  • 54
  • 637
  • 1,101
  • 1
    Could you explain why you need a one-sided test? What exactly is your hypothesis? – whuber Jan 13 '19 at 19:06
  • With only two samples your Bartlett should be equivalent to the ratio of variances test and the Levene to a t-test on the absolute deviations from the mean; in both cases they can be done one-tailed – Glen_b Jan 14 '19 at 07:39

0 Answers0