Or
gen long dyadid = floor(ID/10)
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Nick Cox
> Sent: 07 February 2006 16:10
> To: [email protected]
> Subject: st: RE: Data set with dyads
>
>
> Focusing on the first question only:
>
> I take it from this that after
>
> sort ID
>
> obs (1,2) (3,4) etc. define dyads.
>
> Thus dyad identifiers could be
>
> gen long dyadid = ceil(_n/2)
>
> So we generate random numbers:
>
> gen rnd = uniform()
>
> and then shuffle within dyads and select:
>
> bysort dyadid (rnd) : gen byte select = _n == 1
>
> Observations with either value of -select-
> are a random sample, one from each dyad.
>
> Nick
> [email protected]
>
> > -----Original Message-----
> > From: [email protected]
> > [mailto:[email protected]]On Behalf Of
> Shyam Kumar
> > Sent: 07 February 2006 15:59
> > To: [email protected]
> > Subject: st: Data set with dyads
> >
> >
> > I have a data set comprising of members in dyads which
> > is structured as follows:
> >
> > ID Member-Income Member-Age Member-Educ. Partner-Age
> > Partner- Educ.
> >
> > My interest lies in understanding how Member-Income is
> > influenced by both Member and Partner characteristics.
> >
> >
> > My first question is Stata related. I would like to
> > form a data set by randomly choosing one member from
> > each dyad. The ID variable tells me the dyad each
> > observation belongs to. This variable takes on values
> > 11,12, 21, 22... 101,102....2441, 2442 etc. Hence
> > observations with ID =101, 102 are two members
> > belonging to the 10th dyad. How do I form a data set
> > by randomly choosing one member from each dyad using
> > the ID variable?
> >
> > My second question is more general. The reason I am
> > randomly choosing one member from each dyad is because
> > error terms are likely to be correlated. My plan is to
> > form two separate data sets and test my hypotheses in
> > both data sets separately and then compare the results
> > as a robustness check. Is there a better way to handle
> > dyads?
> >
> > Any insights would be much appreciated. Thanks!
> >
*
* 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/
