Re: st: Outlier diagnostics for tobit (postestimation)

[Maarten Buis <maartenlbuis@gmail.com>]

On Fri, Oct 19, 2012 at 12:54 AM, JVerkuilen (Gmail) wrote:
> It's not just the large dataset. There's no really good reason to
> suppose that the usual graphical methods will work at all.

Here I have to disagree. Graphical methods can also work for bounded
variables if you know what kind of pattern to expect if the model is
true. This can deviate considerably from the kind of patterns we would
expect in a "normal" linear regression, so we cannot rely on the rules
of thumb we learned (teach) in our intro-stats courses. However, if
you simulate new dependent variables under the assumption that your
model is true, than the patterns in the simulated variables represent
a reference pattern with which you can compare any pattern in the real
data. An example is given here:
<http://www.stata.com/statalist/archive/2012-10/msg00646.html>.

Hope this helps,
Maarten

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/