11
$\begingroup$

I need to compare the distribution (unknown) of a set of data to the distribution of another one (unknown). In particular, I want to check for equality of the two distributions.

What are some statistical tests for this?

$\endgroup$
1

1 Answer 1

12
$\begingroup$

A huge subject! The standard all-purpose test is Kolmogorov-Smirnov.

$\endgroup$
5
  • $\begingroup$ There is chi-square also. Most nonparametric tests compare the empirical cdfs. For examp[le the Kolmogorov-Smirnov test looks at the maximum difference of the two ecdfs. $\endgroup$ Jun 18, 2012 at 20:53
  • 1
    $\begingroup$ @André Nicolas I was going to use Kolmogorov-Smirnov, but then I read this: "# Perhaps the most serious limitation is that the distribution must be fully specified. That is, if location, scale, and shape parameters are estimated from the data, the critical region of the K-S test is no longer valid. It typically must be determined by simulation." itl.nist.gov/div898/handbook/eda/section3/eda35g.htm This is a problem for me, right? $\endgroup$ Jun 19, 2012 at 9:46
  • 2
    $\begingroup$ Depends on what you want to determine. If you are interested in determining whether the distributions have the same mean, and don't care about the rest, then K-S is not best. K-S is very good for determining whether two samples in essence come from the same population. $\endgroup$ Jun 19, 2012 at 12:49
  • $\begingroup$ No, I need a non-parametric test. I don't know the distribution of either data set and I want to compare one with the other. What are good solutions for this? Mann whitney u test, for instance? $\endgroup$ Jun 27, 2012 at 11:50
  • $\begingroup$ From your description, it is K-S that seems to fit your description best. If you are interested mainly in comparing locations, then Mann-Whitney is the standard non-parametric test. $\endgroup$ Jun 27, 2012 at 13:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .