[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Rank regression

From   Austin Nichols <[email protected]>
To   [email protected]
Subject   Re: st: Rank regression
Date   Fri, 25 Sep 2009 15:25:10 -0400

Apostolos Ballas <[email protected]> :
I assume the reviewer means quantile regression.  There is some
squishiness in the definitions of rank and quantile in this area--see
for example the help file for -qreg- which says

quantile(#) specifies the quantile to be estimated and should be a
number between 0 and 1, exclusive.  Numbers larger than 1 are
interpreted as percentages.  The default value of 0.5 corresponds to
the median.

and many people would also call that number between 0 and 1 a rank.
Of course, the quantile is not the number r between 0 and 1, it is the
value at which the CDF equals r, and with N obs, the rank would be
about Nr not r, but the sloppy terminology notwithstanding, use -qreg-
to get a sense of the distribution of y given X rather than just the
mean of y given X.

On Fri, Sep 25, 2009 at 1:49 PM, Apostolos Ballas <[email protected]> wrote:
> A very persistent referee is asking that I carry out an analysis using rank
> regression. Could you please suggest how this is done in Stata since there
> appears no reference in the manuals? Also, if anyone could also help me with
> the statistics, how does it compare with quantile regression?
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index