Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: Re: st: RE: KSmirnov: testing for first order stochastic dominance


From   n j cox <n.j.cox@durham.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: Re: st: RE: KSmirnov: testing for first order stochastic dominance
Date   Fri, 05 Oct 2007 11:26:58 +0100

I was under the impression that discreteness and Kolmogorov-Smirnov
were not the best of friends. There's been some discussion of this on
the list: search the archives.

Somers' d might be more serviceable. That has the advantage that
you measure, as well as test.

. findit somersd

Nick
n.j.cox@durham.ac.uk

Pedro

The case should be simple: I have a large random sample of individuals,
some of which are black and some of which are white. I am testing whether
the distribution of a (discrete) outcome of whites dominates that of
blacks, i.e. whether one of the conditional distributions dominates the
other.

In principle, this should correspond to doing a Ksmirnov test on the
conditional cdf's in which the alternative hypothesis would be that the cdf
of whites is always above that of blacks. I am however unsure about how to
do this in practice.

>> I need to test for first order stochastic dominance, contrasting two
>> discrete distributions. The KSmirnov command could in
>> principle be useful,
>> but I am unsure about how to make it operational in testing
>> for stochastic
>> dominance.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index