Thanks for your reply, Nick.
Maybe I didn't express myself clear enough. What I expected was that
. set seed 3562
. scatter y x, jitter(2)
. set seed 3562
. scatter y x, jitter(2)
would produce two identical graphs, just as
. set seed 3562
. gen v1=uniform()
. set seed 3562
. gen v2=uniform()
produces two identical variables. This, however, is not the case. Ergo, the jitering does not seem to depend on the seed (unless I miss something here).
Of course, Nick's solution (creating random perturbations outside -scatter-) would work. I just thought it would be nice to have full control over the jitering-option without having to use a backdoor. And, by the way, I am really curious how the jitering is determined if not by the seed. (because I feel commited to the scientific reproducability policy ... although this might be splitting hairs in this case)
ben
> -----Urspr�ngliche Nachricht-----
> Von: Nick Cox [mailto:[email protected]]
> Gesendet: Dienstag, 5. August 2003 17:59
> An: [email protected]
> Betreff: st: RE: how to control jitering in scatter plots
>
>
> Ben Jann
> >
> > Does anyone know how the random noise generator of the
> > jitering option in scatter plots can be controlled (Stata 8
> > and/or 7). I'd like to produce a series of related graphs
> > for didactical purposes. The data points are always the
> > same, however, some other stuff changes from graph to
> > graph. It would be really nice if the data points would be
> > jittered *exactly* the same way in each graph. (Up to now,
> > I just produced one graph with everything in it, and then
> > manually deleted whatever wasn't needed by editing the eps-file.)
> >
> > My first guess was, that the jittering depends on the seed.
> > So I tried:
> >
> > set seed ...
> > scatter ...
> >
> > This doesn't work, however.
>
> I guess that the jittering does depend on
> the seed, but it won't remain constant
> from graph to graph, any more than you
> get the same sample of random numbers
> repeatedly even if the seed remains unchanged.
>
> For your purposes, I think you will need to produce
> one or more sets of random perturbations outside
> -scatter- and then add to your variables. To
> ensure that the perturbed variables have the
> same variable labels, value labels, format,
> etc. as your originals, there are various commands, among which
> -copydesc- on SSC is in the public domain.
>
> Nick
> [email protected]
> *
> * 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/
>
*
* 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/
