0

Is it appropriate to use the coefficient of variation (CV) for non-parametric data?

I'm not sure whether it is appropriate since you use the mean of the data to calculate it together with the SD. While for parametric data the median is more representative for the central tendency. But perhaps I'm overlooking something.

Karolis Koncevičius
  • 4,282
  • 7
  • 30
  • 47
cinclus
  • 1
  • 1
  • 3
    The term "non-parametric" is used to describe statistical procedures, not the data. What do you mean by saying your data is non-parametric? – Karolis Koncevičius Jun 14 '18 at 16:41
  • Yes, you are right. I meant to say data does not display normal (Gaussian) distribution – cinclus Jun 14 '18 at 16:46
  • To clarify the potentially lingering misapprehension - there's really no logical connection at all between "nonparametric" and "non-normal". There's all manner of non-normal parametric procedures and nonparametric procedures are quite suitable for Gaussian variables (were you ever in a situation to know you had one, which is a fairly good approximation of *never*), including, in many cases, readily available nonparametric procedures which are fully efficient in large samples. – Glen_b Jun 14 '18 at 23:18
  • 1
    Okay, thank you for your additional comment Glen_b, I appreciate it very much. I realize I've still much to learn to be able to apply statistics confidently and correctly. My original question has been answered by Karolis (thanks for that). – cinclus Jun 15 '18 at 07:06

0 Answers0