Re: st: RE: spineplot and survey weights

[Austin Nichols <austinnichols@gmail.com>]

Steve--
No need; -spineplot- supports aweights which give identical point
estimates already.

On Wed, Nov 2, 2011 at 5:21 PM, Steven Samuels <sjsamuels@gmail.com> wrote:
>
>
> You can can construct frequency weights that will give proportions equivalent to those obtained from probability weights.   See Austin Nichols's post at http://stata.com/statalist/archive/2007-10/msg00327.html
>
>
> Steve
>
> On Nov 2, 2011, at 1:28 PM, Nick Cox wrote:
>
> -spineplot- (SJ) is user-written. You are asked to explain where user-written programs you refer to come from.
>
> That is correct: pweights are not supported. The help tells you so and the program will refuse to accept them. You could clone the program and implement pweights according to how you want to be handled. It is not a long program and the handling of weights is only a small part of the program.
>
> The author's position appears to be that he doesn't usually implement pweights because he fears that whatever non-experts do about survey problems is usually wrong from some expert's point of view. At least that's my impression of what the author thinks.
>
> Nick
> n.j.cox@durham.ac.uk
>
> ann montgomery
>
> I have survey sampled data with one categorical variable and one
> binary, and the sample weighting variable.
>
> I would like to compare the distribution of the categorical variable
> (1-6) by the binary (0,1).
>
> -spineplot is exactly what I want to use but it doesn't appear to
> support pweight
>
> Can anyone think of a work around - where I generate new variables for
> the categorical and binary, weighting them, then graphing them using
> -spineplot?
>

*
*   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/