Statalist The Stata Listserver


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

Re: st: Question about e(vcetype)


From   jpitblado@stata.com (Jeff Pitblado, StataCorp LP)
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Question about e(vcetype)
Date   Tue, 06 Feb 2007 21:35:35 -0600

Mark Schaffer Mark <M.E.Schaffer@hw.ac.uk> asks about -e(vcetype)- and its
affect on the 'Std. Err.' heading in Stata's estimation table:

> Hi folks.  An obscure little question for you:
> 
> One of the saved macros after estimation with -robust- or -cluster- is
> e(vcetype), which has the contents "Robust".  When this macro is
> non-empty, the estimation output displayed has an extra line and
> "Robust" appears over "Std. Err."
> 
> My question: when writing an estimator, is there any reason to stick to
> "Robust" or some other Stata standard contents, or is this purely for
> display purposes?
> 
> I note, for example, that one can have a bit of fun with Ben Jann's
> eret2.  After an estimation, try
> 
> eret2 local vcetype "Sadly Large", replace
> 
> and then replay the results.
> 
> More seriously, is there any harm, for example, in writing an estimator
> that saves e(vcetype) as "Het.-Rob." if the vce is Eicker-Huber-White
> heteroskedastic-robust, and "Clust.-Rob." if it's cluster-robust?

As Mark points out, Stata's official estimation commands use -e(vcetype)- to
label when alternative methods are used to compute standard errors.  Here is a
list of some of the labels we use:

	Bootstrap
	EIM
	Jackknife
	Linearized	(-svy-'s synonym for Robust)
	Newey-West
	OPG
	Robust
	Unbiased

A few of Stata's commands use -e(vcetype)- to determine which method is valid
for computing tests and other results for reporting in their output.  As a
rule we refuse to report likelihood ratio tests when -e(vcetype)- contains
'Robust'.

It is generally not a good idea to change the contents of elements of -e()-
that already exist, this is particularly so of -e(vcetype)- when it is set to
'Robust'.  So if -e(vcetype)- is set by an estimation command you use to fit a
model, you should probably leave it alone.

Stata programmers who have implemented an alternative method for computing
-e(V)- should feel free to use whatever label they want in -e(vcetype)-; just
be aware that certain labels (like 'Robust') have special meaning.

--Jeff
jpitblado@stata.com
*
*   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