[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: RE: Population Attributable Fraction (PAF) estimation
It is for StataCorp to comment on future plans, not me.
As often discussed and explained on this list,
StataCorp tend not to go into tremendous detail
on their plans. Sometimes they do say, "This
will be added shortly", especially in respect
of easy small fixes or extras, or "This is not on
our current to do list", meaning if you wait
for this to be done, you might wait for ever.
One way of lobbying for future changes is
through Statalist, so you just did it. Again,
StataCorp tend to listen but not speak. Among
other things, they note the degree of support
(including silences). Of course, there are other
criteria, including the amount of user activity
in an area.
Among many other issues, there are hundreds
and hundreds of methods that are important
to groups of users. It is just not possible
for StataCorp to implement more than a few. On the face of
it, this one is a technique that is user-programmable;
there are just happen to be some problems with
the existing program so far as you are concerned.
In any major area the usual consequence of that
is a flurry of alternative programs and/or
Also, what are especially problematic for StataCorp to
implement are contentious areas in which even
the supposed experts disagree.
These are broad comments, as I am not familiar
at all with this field.
> Thanks for your response. Do you know if there are any plans to
> implement an adjusted PAF estimator in future versions of Stata?
> If not, is there a means of lobbying for implementation of such a
> routine in 'official Stata'?
> Rockhill et al (Am J Public Health 1998; 88: 15-19) comment that
> errors in PAF estimation are common in published papers due to
> confusion over appropriate use of formulae for different types of
> data, especially in relation to valid estimation with covariate
> adjustment. Implementation of a valid and flexible post-regression
> estimator would surely address these problems.
* For searches and help try: