[email protected]
>
> I have been asked to implement Chamberlain's Percent
> Positive Agreement (PPA) which is the number of
> occurrences for which both observors report a positive
> result out of the total number of observations for which
> at least one observor reports a positive result (see Szklo
> and Nieto 2000).
>
> My questions to the list are:
>
> 1) Is there an existing Stata procedure to implement
> Chamberlain's PPA?
>
> 2) Do any list members have alternatives approaches to
> Chamberlain's PPA?
>
I have never heard of this, but no matter. Nor
does the reference help, absent a book or journal title.
It may well be lurking somewhere in Stataland under
another name. If not,
Suppose obs1 and obs2 are variables containing
results of the two observers.
Then
qui count if (obs1 > 0 & obs1 < .) | (obs2 > 0 & obs2 < .)
local denom = r(N)
qui count if (obs1 > 0 & obs1 < .) & (obs2 > 0 & obs2 < .)
di `r(N)' / `denom'
Here is some code:
=========== begin ppa.ado
program def ppa, rclass
version 7
syntax varlist(numeric min=2 max=2) [if] [in]
marksample touse
tokenize `varlist'
args o1 o2
qui count if ((`o1' > 0) | (`o2' > 0)) & `touse'
local denom = r(N)
qui count if (`o1' > 0) & (`o2' > 0) & `touse'
local ppa = r(N) / `denom'
di
di as text "Chamberlain's PPA measure " _c
di as res %5.4f `ppa' " (" %5.2f 100 * `ppa' " %)"
end
============== end ppa.ado
Syntax: ppa obs1 obs2 [if] [in]
On measures of agreement in general, it is
difficult to comment without knowing
research objectives, etc.
The program -pairplot- recently posted on SSC
is directly aimed at graphic comparison of
such paired variables.
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/
