RE: Re: st: how to us prais in a situation with comparison group

["Li, Rui (CDC/ONDIEH/NCCDPHP)" <eok8@cdc.gov>]

Thank you all very much for the comments and suggestions. I will try and let you know whether it works. 

Rui

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Austin Nichols
Sent: Thursday, May 03, 2012 12:32 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: Re: st: how to us prais in a situation with comparison group

Ariel Linden, DrPH <ariel.linden@gmail.com>:
-newey- is also for TS not XT data.

webuse grunfeld, clear
newey mvalue kstock, lag(4)
ivreg2 mvalue kstock, bw(auto)


On Thu, May 3, 2012 at 12:17 PM, Ariel Linden, DrPH <ariel.linden@gmail.com> wrote:
> Let me try and help out here...
>
> - prais - will not work on multiple groups, since it is intended for a 
> single group time series analysis. However, there are other available 
> approaches with multiple groups. For example, you could use -newey- 
> which will allows specification of lags and provide appropriate standard errors.
>
> In addition, you can (and should) check for autocorrelation in each 
> group individually, which can be done via -dfuller-, - wntestq- ,  
> -abar- (user written program) or -ivactest- (user written program). 
> You may have to find the correct form/structure for each group 
> separately and then put apply the statistical model to the corrected 
> data
>
> I hope this helps...
>
> Ariel
>
> Date: Wed, 2 May 2012 20:32:56 +0100
> From: Nick Cox <njcoxstata@gmail.com>
> Subject: Re: st: how to us prais in a situation with comparison group
>
> I don't know what autoreg in SAS does, so can't comment much, except 
> for this.
>
> If you can explain how autoreg calculates serial correlation of 
> residuals for your model, then somebody well qualified may be able to 
> comment. You could try betting that someone on this list knows what 
> autoreg in SAS does, but you're less likely to get a good answer.
>
> You can't easily have it both ways. If your data are time series, it 
> is hard to bootstrap them meaningfully, at least in Stata.
>
> Nick

*
*   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/