Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: st: SUREG with if command.
From
"Schaffer, Mark E" <[email protected]>
To
<[email protected]>
Subject
RE: st: SUREG with if command.
Date
Tue, 25 Sep 2012 18:23:24 +0100
It is indeed possible in principle to use the additional obs. -sureg-
and -reg3- are estimating the error components for a 2-eqn model, so the
estimated covariance matrix is 2x2:
. qui reg3 (mpg rep78) (trunk turn), ols
. mat list e(Sigma)
symmetric e(Sigma)[2,2]
mpg trunk
mpg 29.274269
trunk -5.0326348 12.233795
The above uses OLS appled to 69 obs for both equations. If we use
-regress- to estimate the mpg eqn, where only 69 obs are available, we
get the same error variance:
. qui reg mpg rep78
. di e(rmse)^2
29.274269
But -regress- applied to the trunk equation on its own uses all 74 obs,
and so the error variance is different (and, since it uses more obs,
preferable):
. qui reg trunk turn
. di e(rmse)^2
11.848587
In principle -sureg-/-reg3- should use the additional 5 obs when
estimating the trunk equation. Not to do so is throwing away
information.
--Mark
> -----Original Message-----
> From: [email protected] [mailto:owner-
> [email protected]] On Behalf Of Nick Cox
> Sent: 25 September 2012 17:57
> To: [email protected]
> Subject: Re: st: SUREG with if command.
>
> I just note that the reason 69 are being used is because -rep78- has 5
missing
> observations. If it's OK in principle for -sureg- to use different
subsets of the
> data then your comment has force.
>
> Nick
>
> On Tue, Sep 25, 2012 at 5:44 PM, Schaffer, Mark E
<[email protected]>
> wrote:
> > Nick,
> >
> > I think David is right, and in this case -sureg- is not doing the
best
> > it can.
> >
> > Here's an example with the toy auto dataset. There are 69 obs for
> > rep78 and 74 obs for everything else. In the following example,
> >
> > . sureg (mpg rep78) (trunk turn)
> >
> > Seemingly unrelated regression
> >
----------------------------------------------------------------------
> > Equation Obs Parms RMSE "R-sq" chi2
P
> >
----------------------------------------------------------------------
> > mpg 69 1 5.333491 0.1613 15.86
0.0001
> > trunk 69 1 3.48627 0.3462 26.22
0.0000
> >
----------------------------------------------------------------------
> >
> > -sureg- could be using all 74 observations for the trunk equation,
but
> > it's using only 69.
> >
> > It's even clearer with -reg3- (IIRC, -sureg- is implemented using
> > -reg3-). If you use -reg3- with the ols option,
> >
> > . reg3 (mpg rep78) (trunk turn), ols
> >
> > Multivariate regression
> >
----------------------------------------------------------------------
> > Equation Obs Parms RMSE "R-sq" F-Stat
P
> >
----------------------------------------------------------------------
> > mpg 69 1 5.41057 0.1619 12.94
0.0005
> > trunk 69 1 3.497684 0.3610 37.84
0.0000
> >
----------------------------------------------------------------------
> >
> > -reg3- again uses only 69 obs for the trunk equation even though
there
> > are 74 available and it's being asked to do OLS only.
> >
> > --Mark
> >
> >> -----Original Message-----
> >> From: [email protected] [mailto:owner-
> >> [email protected]] On Behalf Of Nick Cox
> >> Sent: 25 September 2012 17:32
> >> To: [email protected]
> >> Subject: Re: st: SUREG with if command.
> >>
> >> -sureg- will always do the best it can, and there is no extra
> >> trickery
> > except by
> >> imputing missing values.
> >>
> >> For clarity, don't think or write in terms of missing observations.
> >> It's values that are missing, not observations. Remember, for Stata
> >> an observation is a complete row, record, or case in other
terminology.
> >>
> >> Nick
> >>
> >> On Tue, Sep 25, 2012 at 5:21 PM, David Ashcraft
> >> <[email protected]> wrote:
> >> > Thanks Nick, by saving number of observations, I meant e.g. for
> >> > rp1,
> > I have
> >> 120 observations so I want -sureg- to utilize 120 observations not
60
> >> observations. Is there a way, I could utilize all non-missing
> > observations for
> >> each equation in -sureg- model?
> >> >
> >> > edit rp? if dummy==1: two variable rp8 and rp9 have all
> >> > observations
> > as
> >> missing. I dropped these two equations from the model for dummy==1.
I
> >> have got some results and these are inline with my expectations. I
> > also have
> >> checked for observation where dummy==0 and have found several
> > instances
> >> of missing observations.
> >> > Regards
> >> >
> >> > David
> >> >
> >> > ----- Original Message -----
> >> > From: Nick Cox <[email protected]>
> >> > To: [email protected]
> >> > Cc:
> >> > Sent: Tuesday, September 25, 2012 6:34:03 PM
> >> > Subject: Re: st: SUREG with if command.
> >> >
> >> > I've never used -sureg-. It seems to me that it uses or knows
> > nothing
> >> > about panel structure, so I surmise that it is indifferent to
> >> > unbalanced panels as such. But it seems that you do have missing
> >> > values in different observations and will have problems because
> >> > -sureg- can only function with non-missing values on all
variables
> >> > named.
> >> >
> >> > Look at
> >> >
> >> > edit rp? if dummy == 1
> >> >
> >> > I don't know what you mean by "save the number of
observation[s]".
> >> >
> >> > Nick
> >> >
> >> > On Tue, Sep 25, 2012 at 4:21 PM, David Ashcraft
> >> > <[email protected]> wrote:
> >> >> Hello Nick,
> >> >>
> >> >> I think problem is not with the dummy variable. This may be
> >> >> related
> > -
> >> sureg- model. It seems to me -sureg- needs a balanced panel
resulting
> > in
> >> drop of number of observations considerably while implementing
> > -sureg-.
> >> >>
> >> >> Is there any other way where I can save the number of
observation
> > and
> >> still use seemingly unrelated regression model?
> >> >>
> >> >> Below is some descriptive stats for your review.
> >> >>
> >> >>
> >> >> gen dummy=0
> >> >> . replace dummy = 1 if date2>17532
> >> >>
> >> >> (53 real changes made)
> >> >> . tabulate dummy
> >> >>
> >> >>
> >> >>
> >> >> dummy | Freq. Percent Cum.
> >> >> ------------+-----------------------------------
> >> >> 0 | 96 64.43 64.43
> >> >> 1 | 53 35.57 100.00
> >> >> ------------+-----------------------------------
> >> >> Total | 149 100.00
> >> >>
> >> >> . su rp1 rp2 rp3 rp4 rp6 rp7 rp8 rp9
> >> >>
> >> >> Variable | Obs Mean Std. Dev. Min
> > Max
> >> >>
> >
-------------+-------------------------------------------------------
> >> >> -------------+-
> >> >> rp1 | 120 .001517 .0469446 -.1935012
> > .1102614
> >> >> rp2 | 120 .0008538 .0545707 -.212302
> > .1238899
> >> >> rp3 | 120 .0016796 .0565703 -.2283529
> > .1202257
> >> >> rp4 | 120 .0016847 .0588602 -.2037239
> > .1229283
> >> >> rp6 | 120 .0015542 .056026 -.2226954
> > .1190248
> >> >>
> >
-------------+-------------------------------------------------------
> >> >> -------------+-
> >> >> rp7 | 120 .0016078 .0503465 -.2033414
> > .1073936
> >> >> rp8 | 88 .0023456 .0709356 -.2216449
> > .1371796
> >> >> rp9 | 88 .0033193 .0779086 -.2401649
> > .1579783
> >> >>
> >> >>
> >> >>
> >> >>
> >> >>
> >> >> ----- Original Message -----
> >> >> From: Nick Cox <[email protected]>
> >> >> To: [email protected]
> >> >> Cc:
> >> >> Sent: Tuesday, September 25, 2012 2:54:44 PM
> >> >> Subject: Re: st: SUREG with if command.
> >> >>
> >> >> Your results show 60 observations with non-missing values; dummy
> >> >> is
> > 0
> >> >> on 60 of them (all) and so necessarily 1 on 0 (none) of them.
> > Stata's
> >> >> response is reasonable; there are _no_ observations to do your
> >> >> last command.
> >> >>
> >> >> You should perhaps revisit your definition of -dummy-, which
> > doesn't
> >> >> divide the dataset.
> >> >>
> >> >> Nick
> >> >>
> >> >> On Tue, Sep 25, 2012 at 12:42 PM, David Ashcraft
> >> >> <[email protected]> wrote:
> >> >>> Hi Statalist,
> >> >>>
> >> >>> I am trying to run -sureg- with multiple equation as per below.
> > Now I
> >> have divided my data based on dummy variable and I want to see if
the
> >> results are any different based on the dummy variable. Based on my
> > data, I
> >> should get three different results i.e. one for the whole sample
and
> > two
> >> based on the dummy variable. I am getting the same result for the
> > overall
> >> sample and where dummy==0 and getting no result for dummy==1.
> >> >>>
> >> >>> I don't understand why I am getting this result. Can anyone
help
> > me
> >> direct to the solution of this problem. I have looked at some older
> > posts but
> >> there is no answer.
> >> >>> Regards
> >> >>>
> >> >>> David
> >> >>>
> >> >>>
> >> >>>
> >> >>> sureg (rp1 rm1)(rp2 rm2)(rp3 rm3)(rp4 rm4)(rp6 rm6)(rp7
rm7)(rp8
> >> >>> rm8)(rp9 rm9), corr
> >> >>>
> >> >>> Seemingly unrelated regression
> >> >>>
> >
----------------------------------------------------------------------
> >> >>> Equation Obs Parms RMSE "R-sq" chi2
> > P
> >> >>>
> >
----------------------------------------------------------------------
> >> >>> rp1 60 1 .0071761 0.9616 3298.70
> > 0.0000
> >> >>> rp2 60 1 .0092113 0.9465 2534.55
> > 0.0000
> >> >>> rp3 60 1 .0082266 0.9544 2847.07
> > 0.0000
> >> >>> rp4 60 1 .0091633 0.9491 2198.62
> > 0.0000
> >> >>> rp6 60 1 .0084368 0.9515 2677.13
> > 0.0000
> >> >>> rp7 60 1 .0060539 0.9711 3703.34
> > 0.0000
> >> >>> rp8 60 1 .009352 0.9866 5504.52
> > 0.0000
> >> >>> rp9 60 1 .0115533 0.9832 4137.04
> > 0.0000
> >> >>>
> > --------------------------------------------------------------------
> >> >>> --
> >> >>>
> >> >>>
> >
----------------------------------------------------------------------
> > --
> > ------
> >> >>> | Coef. Std. Err. z P>|z| [95%
> > Conf. Interval]
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp1 |
> >> >>> rm1 | .9992066 .0173974 57.43 0.000
.9651084
> > 1.033305
> >> >>> _cons | -.0000674 .0009287 -0.07 0.942
-.0018876
> > .0017527
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp2 |
> >> >>> rm2 | .9733916 .0193347 50.34 0.000
.9354963
> > 1.011287
> >> >>> _cons | -.0013549 .001196 -1.13 0.257
-.003699
> > .0009892
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp3 |
> >> >>> rm3 | .9942406 .0186334 53.36 0.000
.9577198
> > 1.030761
> >> >>> _cons | -.0008887 .0010727 -0.83 0.407
-.0029911
> > .0012136
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp4 |
> >> >>> rm4 | .9618481 .0205131 46.89 0.000
.9216431
> > 1.002053
> >> >>> _cons | .0010966 .0011632 0.94 0.346
-.0011832
> > .0033765
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp6 |
> >> >>> rm6 | .9920405 .0191732 51.74 0.000
.9544617
> > 1.029619
> >> >>> _cons | -.0008398 .001099 -0.76 0.445
-.0029937
> > .0013142
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp7 |
> >> >>> rm7 | .9873097 .016224 60.86 0.000
.9555113
> > 1.019108
> >> >>> _cons | .0001045 .0007886 0.13 0.895
-.0014411
> > .0016502
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp8 |
> >> >>> rm8 | .9554066 .0128774 74.19 0.000
.9301673
> > .9806458
> >> >>> _cons | .0009302 .0012033 0.77 0.439
-.0014282
> > .0032887
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp9 |
> >> >>> rm9 | .9889603 .0153757 64.32 0.000
.9588245
> > 1.019096
> >> >>> _cons | -.0001408 .0014852 -0.09 0.924
-.0030518
> > .0027703
> >> >>>
> > --------------------------------------------------------------------
> >> >>> ----------
> >> >>>
> >> >>> Correlation matrix of residuals:
> >> >>>
> >> >>> rp1 rp2 rp3 rp4 rp6 rp7
rp8
> > rp9
> >> >>> rp1 1.0000
> >> >>> rp2 0.1214 1.0000
> >> >>> rp3 0.2210 0.9609 1.0000
> >> >>> rp4 0.4345 -0.0342 0.0495 1.0000
> >> >>> rp6 0.2268 0.9595 0.9982 0.0447 1.0000
> >> >>> rp7 0.9088 0.2710 0.3749 0.7114 0.3763 1.0000
> >> >>> rp8 0.2240 0.0839 0.0896 -0.0739 0.0987 0.1232
1.0000
> >> >>> rp9 0.2100 0.1163 0.1462 -0.0321 0.1497 0.1600
0.7653
> > 1.0000
> >> >>>
> >> >>> Breusch-Pagan test of independence: chi2(28) = 338.778, Pr =
> > 0.0000
> >> >>>
> >> >>> . sureg (rp1 rm1)(rp2 rm2)(rp3 rm3)(rp4 rm4)(rp6 rm6)(rp7
> >> >>> rm7)(rp8
> >> >>> rm8)(rp9 rm9)if dummy==0, corr
> >> >>>
> >> >>> Seemingly unrelated regression
> >> >>>
> >
----------------------------------------------------------------------
> >> >>> Equation Obs Parms RMSE "R-sq" chi2
> > P
> >> >>>
> >
----------------------------------------------------------------------
> >> >>> rp1 60 1 .0071761 0.9616 3298.70
> > 0.0000
> >> >>> rp2 60 1 .0092113 0.9465 2534.55
> > 0.0000
> >> >>> rp3 60 1 .0082266 0.9544 2847.07
> > 0.0000
> >> >>> rp4 60 1 .0091633 0.9491 2198.62
> > 0.0000
> >> >>> rp6 60 1 .0084368 0.9515 2677.13
> > 0.0000
> >> >>> rp7 60 1 .0060539 0.9711 3703.34
> > 0.0000
> >> >>> rp8 60 1 .009352 0.9866 5504.52
> > 0.0000
> >> >>> rp9 60 1 .0115533 0.9832 4137.04
> > 0.0000
> >> >>>
> > --------------------------------------------------------------------
> >> >>> --
> >> >>>
> >> >>>
> >
----------------------------------------------------------------------
> > --
> > ------
> >> >>> | Coef. Std. Err. z P>|z| [95%
> > Conf. Interval]
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp1 |
> >> >>> rm1 | .9992066 .0173974 57.43 0.000
.9651084
> > 1.033305
> >> >>> _cons | -.0000674 .0009287 -0.07 0.942
-.0018876
> > .0017527
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp2 |
> >> >>> rm2 | .9733916 .0193347 50.34 0.000
.9354963
> > 1.011287
> >> >>> _cons | -.0013549 .001196 -1.13 0.257
-.003699
> > .0009892
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp3 |
> >> >>> rm3 | .9942406 .0186334 53.36 0.000
.9577198
> > 1.030761
> >> >>> _cons | -.0008887 .0010727 -0.83 0.407
-.0029911
> > .0012136
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp4 |
> >> >>> rm4 | .9618481 .0205131 46.89 0.000
.9216431
> > 1.002053
> >> >>> _cons | .0010966 .0011632 0.94 0.346
-.0011832
> > .0033765
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp6 |
> >> >>> rm6 | .9920405 .0191732 51.74 0.000
.9544617
> > 1.029619
> >> >>> _cons | -.0008398 .001099 -0.76 0.445
-.0029937
> > .0013142
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp7 |
> >> >>> rm7 | .9873097 .016224 60.86 0.000
.9555113
> > 1.019108
> >> >>> _cons | .0001045 .0007886 0.13 0.895
-.0014411
> > .0016502
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp8 |
> >> >>> rm8 | .9554066 .0128774 74.19 0.000
.9301673
> > .9806458
> >> >>> _cons | .0009302 .0012033 0.77 0.439
-.0014282
> > .0032887
> >> >>>
> > -------------+------------------------------------------------------
> >> >>> -------------+----------
> >> >>> rp9 |
> >> >>> rm9 | .9889603 .0153757 64.32 0.000
.9588245
> > 1.019096
> >> >>> _cons | -.0001408 .0014852 -0.09 0.924
-.0030518
> > .0027703
> >> >>>
> > --------------------------------------------------------------------
> >> >>> ----------
> >> >>>
> >> >>> Correlation matrix of residuals:
> >> >>>
> >> >>> rp1 rp2 rp3 rp4 rp6 rp7
rp8
> > rp9
> >> >>> rp1 1.0000
> >> >>> rp2 0.1214 1.0000
> >> >>> rp3 0.2210 0.9609 1.0000
> >> >>> rp4 0.4345 -0.0342 0.0495 1.0000
> >> >>> rp6 0.2268 0.9595 0.9982 0.0447 1.0000
> >> >>> rp7 0.9088 0.2710 0.3749 0.7114 0.3763 1.0000
> >> >>> rp8 0.2240 0.0839 0.0896 -0.0739 0.0987 0.1232
1.0000
> >> >>> rp9 0.2100 0.1163 0.1462 -0.0321 0.1497 0.1600
0.7653
> > 1.0000
> >> >>>
> >> >>> Breusch-Pagan test of independence: chi2(28) = 338.778, Pr =
> > 0.0000
> >> >>>
> >> >>> . sureg (rp1 rm1)(rp2 rm2)(rp3 rm3)(rp4 rm4)(rp6 rm6)(rp7
> >> >>> rm7)(rp8
> >> >>> rm8)(rp9 rm9)if dummy==1, corr insufficient observations
r(2001);
> >> >
> >> > *
> >> > * For searches and help try:
> >> > * http://www.stata.com/help.cgi?search
> >> > * http://www.stata.com/support/statalist/faq
> >> > * http://www.ats.ucla.edu/stat/stata/
> >> >
> >> >
> >> > *
> >> > * For searches and help try:
> >> > * http://www.stata.com/help.cgi?search
> >> > * http://www.stata.com/support/statalist/faq
> >> > * http://www.ats.ucla.edu/stat/stata/
> >>
> >> *
> >> * For searches and help try:
> >> * http://www.stata.com/help.cgi?search
> >> * http://www.stata.com/support/statalist/faq
> >> * http://www.ats.ucla.edu/stat/stata/
> >
> >
> > --
> > Heriot-Watt University is the Sunday Times Scottish University of
the
> > Year 2011-2012
> >
> > We invite research leaders and ambitious early career researchers to
> > join us in leading and driving research in key inter-disciplinary
themes.
> > Please see www.hw.ac.uk/researchleaders for further information and
> > how to apply.
> >
> > Heriot-Watt University is a Scottish charity registered under
charity
> > number SC000278.
> >
> >
> > *
> > * For searches and help try:
> > * http://www.stata.com/help.cgi?search
> > * http://www.stata.com/support/statalist/faq
> > * http://www.ats.ucla.edu/stat/stata/
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
--
Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012
We invite research leaders and ambitious early career researchers to
join us in leading and driving research in key inter-disciplinary themes.
Please see www.hw.ac.uk/researchleaders for further information and how
to apply.
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/