Statalist


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

Re: st: ttest for data where observations have standard errors?


From   Austin Nichols <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: ttest for data where observations have standard errors?
Date   Tue, 25 Aug 2009 17:30:58 -0400

Rikhil Bhavnani<rikhil_bhavnani@yahoo.com> :
Maybe a difference in proportions, not means, with aweights?  See
http://www.stata.com/support/faqs/stat/logit.html or try
glm p byvar [aw=w], family(binomial) link(logit) robust
test byvar
where byvar is the dummy marking which group an obs comes from.
See also help weight
...but probably something semiparametric is better.  Where do these
data come from?

On Tue, Aug 25, 2009 at 5:07 PM, Rikhil
Bhavnani<rikhil_bhavnani@yahoo.com> wrote:
> Hi,
>
> I need to conduct a difference of means ttest using data where each observation (which is a probability, incidentally) has a standard error.  Usually, from what I can gather, the ttest command assumes that the standard error for each observation is 0.  Is there any way that I can tell STATA, instead, that each observation has a different standard error, which I specify in a second variable?
>
> Thanks very much.  I appreciate your help!
> Rikhil

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index