Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

re: Re: st: why is there no weight option in -esize-?


From   "Ariel Linden, DrPH" <ariel.linden@gmail.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   re: Re: st: why is there no weight option in -esize-?
Date   Mon, 29 Jul 2013 15:45:55 -0400

And of course, I completely agree with Nick. It could likely be any of the 3
points, and I would be curious to hear from StataCorp on which it was...

I did look at the underlying code and it doesn't seem to be a limitation of
the code. For example, instead of starting with -ttest- to derive the n,
mean, and sd, for each group (which would not offer a weight option), the
program could use -sum- for each of the two groups (treatment and control),
and merrily go from there (allowing the weight option). 

That said, the weight option would not be available for the immediate
version of -esize-

Ariel

________________________________________
From
  Nick Cox <njcoxstata@gmail.com>
To
  "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject
  Re: st: why is there no weight option in -esize-?
Date
  Mon, 29 Jul 2013 16:16:54 +0100
________________________________________
I am going to disappoint you by writing a different comment.

In my experience there are three main reasons why something is not yet
implemented:

1. Pure oversight, or "I don't want that myself".

2. Impending deadlines.

3. The programmer(s) is/are not (yet) clear how to implement something
_so that it produces dependable results_.

In the case of common or garden users, guess at #1 or #2.

In the case of StataCorp, guess at #3. Often (indeed always) writing
the code is secondary for StataCorp. There is typically a project
based on simulations designed to work out how some method behaves with
the extra something. For example, perhaps weights cause sampling
distributions to be very poor approximations whenever there is some
extreme pattern of weights.

It's not just pushing code out of the door. It's defending or
explaining to users when tech support gets reports of really
strange-looking results.

Ariel will know this, really, but as always a thread started by one
can be of wider interest, hence the broader remarks.

That said, I have complete ignorance of any particular reasons that
may apply with this command.

Nick
njcoxstata@gmail.com


On 29 July 2013 15:57, Ariel Linden, DrPH <ariel.linden@gmail.com> wrote:
> Can anyone think of a reason why the new effect size estimator in version
> 13.0 -esize-, is not programmed to accept weights? I cannot think of a
> statistical limitation, but perhaps I am missing something?
>
> As a general request to StataCorp, I would ask that commands that lend
> themselves to use weights, be programmed with that option.
>
> (I am somewhat expecting a response from Nick saying "feel free to write a
> command, yourself, that accepts weights...")
>
> Ariel



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index