3

I have two normally-distributed independent random variables X and Y and I need to calculate its division Z.

As far as I understand the mean of Z is $\mu_Z = \frac{\mu_X}{\mu_Y}$, but I don't know how to calculate the Standard Deviation $\sigma_Z$.

Is $\sigma_Z = \frac{\sigma_X}{\sigma_Y}$?

VGonPa
  • 31
  • 1
  • 1
  • 3

1 Answers1

3

The ratio of two standard normal random variables ($\mu = 0, \sigma = 1$) is a Cauchy distribution. The Cauchy has an undefined variance (and hence undefined standard deviation). For other normals, the distribution is complex, indeed.

See http://en.wikipedia.org/wiki/Ratio_distribution

Of course, for any two samples from random variables, you can compute whatever you like. E.g.,

set.seed(20181)
x1 <- rnorm(100)
x2 <- rnorm(100)

ratio <- x1/x2
sd(x1)
sd(x2)
sd(ratio)

The SD of the ratio is 5.35.

But with a different seed, the SD can be very different. I ran this with 3 different seeds and got values as high as 11.21.

iNyar
  • 147
Peter Flom
  • 119,535
  • 36
  • 175
  • 383
  • @Royi - the issue is that indefinite integral is undefined. (Both its positive and its negative parts integrate to infinity.) See https://stats.stackexchange.com/questions/36027/why-does-the-cauchy-distribution-have-no-mean – James Feb 17 '21 at 12:14