Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Using -if- when running panel-corrected OLS pooled regressions


From   "Clive Nicholas" <Clive.Nicholas@newcastle.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   st: Using -if- when running panel-corrected OLS pooled regressions
Date   Mon, 24 Nov 2003 06:24:56 -0000 (GMT)

Statapeeps,

Well, I didn't *quite* admit defeat in trying -xtpcse- the other night:
I've got it to work, but only by replacing missings with zeros. I'm not
happy with this, because as far as my data is concerned, missings indicate
(correctly) that party X didn't stand a candidate in district Y, whereas
replacing it with a 0 says the party *did* stand but got no votes. Are you
as unhappy with this approach I am?

I've tried to get round this by using -if- to take out cases with missings
on key variables using OLS-PCSE models. I tested this using just two Xs,
and it ran. But when trying the full model:

-xtpcse [depvar] [50 indvars], if [13 indvars]!=.-,

it returns an error saying that -if- isn't allowed! Well that cannot be
right: can it?

Also, some of my variables change to quite absurd values from the FGLS
model to OLS-PCSE models. An example using the effects of regional
variables on the nationalist vote (%), 1979-2001 (AR1 specifications):

FGLS     |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
---------+----------------------------------------------------------------
strathcl |   -.036985   1.214225    -0.03   0.976    -2.416823    2.342853
eastscot |   -.199779   1.720757    -0.12   0.908    -3.572401    3.172843
rurlscot |   3.199318   2.483519     1.29   0.198    -1.668289    8.066925
rurwales |   2.503039   .3860939     6.48   0.000     1.746309    3.259769
indswles |  (dropped)

OLS-PCSE
strathcl |  -3337.817   1262.826    -2.64   0.008    -5812.912    -862.723
eastscot |  -3337.568   1262.719    -2.64   0.008     -5812.45   -862.6848
rurlscot |  -3332.719   1262.377    -2.64   0.008    -5806.932   -858.5065
rurwales |  -3334.875   1262.974    -2.64   0.008     -5810.26    -859.491
indswles |  -3339.153   1263.132    -2.64   0.008    -5814.845   -863.4608

This cannot be right, not least because, if I'm not wrong, the OLS
estimates are not meant to change too much (if at all).

What's going on? Any help, once again, would be greatly appreciated.
Thanks in advance. :-)

Yours,
CLIVE NICHOLAS,
Politics Building,
School of Geography, Politics and Sociology,
University of Newcastle-upon-Tyne,
Newcastle-upon-Tyne,
NE1 7RU,
United Kingdom.
*
*   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