st: RE: Does xtreg (or xtivreg) assume equally spaced time points?

[Lisa Marie Yarnell <lisayarnell@yahoo.com>]

Thanks for your input, Mark and Nick.

OK, yes, I am familiar with hierarchical linear modeling (HLM) in which observations are drawn from larger units such as classrooms or states, and I was not aware that without the FE or RE option the xtreg model in Stata is parallel to what might be called an HLM outside of Stata.  It sounds from what you both are saying that xtivreg and panel models more generally are not necessarily longitudinal models, but simply have multiple observations per panel.

So, if we specify either a FE or RE option after the xtivreg (e.g., "xtivreg, fe"), then dummies for Wave2, Wave3 and Wave4 will help us get at the effects of time in the model.  Are these dummies necessary?  (My guess is no.)  But in using them, would this account for the waves being unequally spaced?

Thanks,
Lisa



----- Original Message -----
From: "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
To: statalist@hsphsun2.harvard.edu
Cc: 
Sent: Tuesday, August 7, 2012 11:11 AM
Subject: st: RE: RE: RE: Does xtreg (or xtivreg) assume equally spaced time points?

Lisa,

> -----Original Message-----
> From: Lisa Marie Yarnell [mailto:lisayarnell@yahoo.com]
> Sent: 07 August 2012 18:35
> To: statalist@hsphsun2.harvard.edu; Schaffer, Mark E
> Subject: st: RE: RE: Does xtreg (or xtivreg) assume equally spaced time
> points?
> 
> Hi Mark,
> 
> Thank you for your response.  I graciously disagree with your view.  I believe
> that the xtreg and xtivreg models DO have time dimensions because these
> are models for panel data, in which variables are measured multiple times, at
> multiple time points.

Sorry, but not true.  xtivreg and xtivreg2 (which we were discussing, not xtreg, though the point applies to it as well as to another estimation command, areg) with the FE or RE options can estimate panel data models stricto sensu, which is what you are describing.  But they can ALSO estimate models fixed and random effects models in which there is no time dimension.

Your model is a static model, i.e., in

> > > Our model is: xtivreg depper (bmi = momobes)

all 3 variables have the same date.  Static models are what these commands are typically used to estimate.  You can think of an observation as a random draw on a panel unit.  In your dataset, the draws are on the same unit (I'm guessing individual persons) but separated over time.  The fixed or random effects capture the time-invariant characteristics of the panel units.

But the same static model, and the same estimation commands, could be applied to a dataset in which the draws are separated over, say, space or some other dimensions.  For example, the panel units could be US states and the draws are of US counties from those states.  Or the panel units could be households and the draws could be of individual household members.  Or the panel units could be classrooms and the draws could be of individual students.  Etc.

Thus you will get exactly the same fixed effects estimation results whether you use xtivreg,fe (or xtivreg2,fe) or an IV estimator with explicit dummies for the fixed effects.

The first-differences estimator is the exception to the irrelevance of exactly what the second dimension is.  In this case you need to know the dates of observations in order to calculate the FDs.  This is why the xtivreg help file says that only the panel variable is needed for the FE and RE estimators, but the time variables is also required for the FD estimator:

    "A panel variable must be specified. For xtivreg, fd, a time variable must also be specified.
      Use xtset."

--Mark

> The model is for panel data.
> 
> I believe that we should add ", i(id)" at the end of our model.  Then time
> points will be nested in persons and the model will have a time dimension.
> 
> My concern is that the time points are not equally spaced.
> 
> Lisa
> 
> 
> ----- Original Message -----
> From: "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
> To: statalist@hsphsun2.harvard.edu
> Sent: Tuesday, August 7, 2012 9:49 AM
> Subject: st: RE: RE: Does xtreg (or xtivreg) assume equally spaced time
> points?
> 
> Lisa,
> 
> Your data have a time dimension, but the model apparently doesn't:
> 
> > > Our model is: xtivreg depper (bmi = momobes)
> 
> So irregular spacing over time is a non-issue - there's nothing you need to
> do.  Just estimate as is.
> 
> --Mark
> 
> > ----- Original Message -----
> > From: "Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>
> > To: statalist@hsphsun2.harvard.edu
> > Cc:
> > Sent: Tuesday, August 7, 2012 8:27 AM
> > Subject: st: RE: Does xtreg (or xtivreg) assume equally spaced time points?
> >
> > Lisa,
> >
> > The model you specify below,
> >
> > > Our model is: xtivreg depper (bmi = momobes)
> >
> > has no apparent time dimension, i.e., it seems to be a static model.
> > If so, the spacing between consecutive periods is irrelevant and you
> > don't need to do anything.
> >
> > HTH,
> > Mark
> >
> > > -----Original Message-----
> > > From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> > > statalist@hsphsun2.harvard.edu] On Behalf Of Lisa Marie Yarnell
> > > Sent: 07 August 2012 15:32
> > > To: statalist@hsphsun2.harvard.edu
> > > Subject: st: Does xtreg (or xtivreg) assume equally spaced time
> > points?
> > >
> > > Hi Statalisters,
> > >
> > > Does xtreg (or really, what we are using, xtivreg--the extension of
> > xtreg that
> > > allows for an instrument) asusme equally-spaced time points in the
> > panel
> > > model?
> > >
> > > We have four waves of data.
> > > Wave I and Wave II are 1.5 years apart.
> > > Wave II and Wave III are 5.5 years apart.
> > > Wave III and Wave IV are 6 years apart.
> > >
> > > Is it wrong to put data from these time points into the xtivreg model?
> > Is
> > > there a way to adjust the model for unequally spaced time points?
> > >
> > > Our model is: xtivreg depper (bmi = momobes)
> > >
> > > How can we alter the model to account for the unequal spacing?
> > >
> > > Thanks,
> > > Lisa
> > >
> > > *
> > > *  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/
> 
> 
> 
> --
> 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/



-- 
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/