# RE: st: autoregression on dummy varialbe

 From "Nick Cox" To Subject RE: st: autoregression on dummy varialbe Date Thu, 9 Feb 2006 15:05:54 -0000

This is really the same question, and a question posed
to you earlier about how why you think this would make sense
has not been answered.

However, we see here some (or perhaps all) of the data.
x takes one of two values, so your data are from a
discrete distribution. I think it difficult to see why an AR model would
be appropriate here. Possibly a Markov chain, which is loosely equivalent,
would make more sense.

There are 4 values of 0 and 4 values of 1 in this sample. Thus
the data are symmetrically distributed. No one-to-one
transformation that I could think of appears either necessary
or useful in this circumstance.

If these really are all the data, you are also challenged on
the score of sample size.

Nick
n.j.cox@durham.ac.uk

tang5@purdue.edu

> thanks for response,and sorry for ambiguity
> here is the sample data:
> yearid  x
> 1       1
> 2       1
> 3       0
> 4       1
> 5       0
> 6       0
> 7       0
> 8       1
>
> then, I want to apply AR for x. is that allowed? since x is
> variable only with
> value 1 and 0. if not, is there any transformation on x that
> we can do?
> Is that clear enough? thanks

Marcello Pagano <pagano@hsph.harvard.edu>:

> > The reason, Tang, why there was not much of a response to your
> > question is that it is not possible to understand what your question
> > is. Of course, one can do AR on any variable, but the interpretation
> > might be a little challenging!
> > Please give us more details about your problem, otherwise don't post
> > again, please.
> >

tang5@purdue.edu

> > >>can i do AR on a dummy variable? what do i need to do to
> transform the
> > >>variable.

*
*   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/