Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: autocorrelation test returning 'invalid subcommand dwatson'

From   Nick Cox <>
Subject   Re: st: autocorrelation test returning 'invalid subcommand dwatson'
Date   Thu, 19 May 2011 10:37:15 +0100

I mean, given what you have just said, use

predict, residual

but presumably there is one set of residuals for each equation, so 14
tests which can hardly be independent of each other?

I think you need advice from people who also do this kind of thing.


On Thu, May 19, 2011 at 10:23 AM, ABDUL ADAM <> wrote:
> Thank you very much Nick for your reply.
> I am estimating a system of expenditure share equations (14 equations altogether). I am using a weekly time series data, and since one usually suspects that there could be serial correlation of the error terms in such data, I wanted to see whether this is true in my case by using D-W test on the nlsur residuals. Moreover, I have already tsset my data (before the estimation). I am not sure, though, if I understand what you mean "calculate explicitly".
> More explanations or other tips are highly appreciated.
> Abdul
> --- On Thu, 5/19/11, Nick Cox <> wrote:
>> From: Nick Cox <>
>> Subject: Re: st: autocorrelation test returning 'invalid subcommand dwatson'
>> To:
>> Date: Thursday, May 19, 2011, 1:50 AM
>> Yes, but which -estat- subcommands
>> are supported depends on the
>> estimation command previously run and those subcommands are
>> not
>> supported in this case.
>> Evidently you have a system of equations that you are
>> estimating with
>> -nlsur-. What precisely were you expecting e.g. the
>> Durbin-Watson
>> command to work on? This is not my field -- and this is the
>> limit of
>> my advice -- but I suspect you would need to calculate that
>> explicitly
>> and also declare your data to be time series.
>> Nick
>> On Thu, May 19, 2011 at 9:38 AM, ABDUL ADAM <>
>> wrote:
>> > Dear Members,
>> > I have run an nlsur programme and after the estimation
>> tried to do an autocorrelation test. I used both estat
>> dwatson, estat dwstat, estat bgodfrey. In each case I get
>> error r(321) 'invalid subcommand'. I know from the Stata
>> reference manual that estat is one the postestimation
>> commands that are available for nlsur. I am using version
>> 10.0. Any idea of what I could do to proceed with my
>> autocorrelation test?

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index