Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

Re: st: Pseudo r2 of oprobit

From   Nick Cox <>
Subject   Re: st: Pseudo r2 of oprobit
Date   Thu, 28 Feb 2013 19:56:19 +0000

Thanks for the correction: I distorted Ben's suggestion in my reply.
Sorry about that.

I think this is a "looks like a duck, walks like a duck, talks like a
duck" situation. If the coefficients are the same, the predictions are
the same, and any figures of merit are equally valid across both.

On Thu, Feb 28, 2013 at 7:50 PM, Richard Williams
<> wrote:

> He is using pweights with both approaches though, right? And the
> coefficients will come out the same, I believe. Actually, I have
> wondered much the same thing. When using pw with regular commands, is
> Stata giving you some statistics it shouldn't? Or conversely, when
> using svy, is it sometimes holding back on reporting statistics that
> are legitimate? I am not sure why pseudo r2 would be legitimate with
> one of the approaches but not the other.
> Sent from my iPad
> On Feb 28, 2013, at 2:34 PM, Nick Cox <> wrote:
>> It's the other way round.
>> If you can explain to an audience why ignoring pweights makes no
>> difference to a measure, then the same measure will do equally well in
>> situations with and without pweights.
>> But if predictions change when you take account of pweights, you can't
>> expect pseudo-R2 to remain unchanged.
>> Better to summarize lack of fit in the same way for both models, e.g.
>> by looking at residuals or predictions directly.

On Thu, Feb 28, 2013 at 7:16 PM, Ben Dandi <> wrote:

>>> Can I use the pseudo R2 that I get from oprobit [pw=xyz] to be the pseudo R2 of svy linearized : oprobit (when sample weights are set to xyz)?
*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index