Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Hold-out sample prediction in binary probit

From   Maarten Buis <>
Subject   Re: st: Hold-out sample prediction in binary probit
Date   Tue, 17 May 2011 13:52:28 +0200

On Tue, May 17, 2011 at 1:15 PM, Eleimon Gonis wrote:
> I have estimated a binary probit model with firm year observations between 1995-2006 and I would now like to keep 2006 as a hold-out year to validate my model.
> I am sure it's easy to exclude 2006 from the sample by following the routine sample specification procedure. However, my question is the following:
> Once I estimate the model, is there any way (command-wise) to tell STATA to do the prediction for me rather than do it manually?

probit y x if year < 2006
predict pr
sum pr if year == 2006

Hope this helps,

On Tue, May 17, 2011 at 1:15 PM, Eleimon Gonis wrote:
> Your swift responses will be greatly appreciated.

I do not know if this is intentional, but this sounds to me like you
are asking for quick responses. Such requests are not appreciated on
this list. To quote the Statalist FAQ:

"Urgency is your concern only. Pleas of urgency, desperation, and the
like are deprecated on Statalist. Your urgency, however compelling, is
a private matter and does not translate into urgency for other members
of the list. In fact, labeling your question as urgent is more likely
to lead to your question being ignored by list members, who know that
in most cases urgency arises from disorganization. On Statalist, the
principle of charity is that you answer questions because you are able
and willing to say something about the question, not because you have
pity on the questioner."

An alternative interpretation would be that you noticed that people
often get quick responses and you wanted to express appreciation for
that as well.  Since the Statalist FAQ also asks us to "assume
goodwill even if it is not evident", I'll assume that the latter is
the case.

The easiest way to avoid this problem in the future is to thank after
you have received advise rather than in advance. This is perfectly
acceptable on this list.

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index