Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Finite population correction when testing equality of two proportions


From   Fredrik Nyberg <fredrik.nyberg@imm.ki.se>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Finite population correction when testing equality of two proportions
Date   Tue, 12 Apr 2005 09:08:35 +0200

I think what you describe is appropriate if we assume that the whole
population (overall sample) is a random sample of a larger population and is
composed of two groups (eg male/female) that are then likewise random
samples of the larger population.

This is not the same test that I seek.

Consider an overall population of 1000 subjects, with 50% males. If I take
one subsample of 10 subjects with 9 males (90% males) and compare with the
remaining 990 (still ~50%), the proportions are significantly different, but
obviously my second subsample of 990 is not significantly different from the full 1000
(both ~50%).

Any other suggestions on how to perform this latter test (the 990 to the full
1000 = sample 2 to sample 1 in the terminology of my initial question) in a simple/convenient way?

Fredrik

At 2005-04-01 22:06, Richard Williams wrote:

At 09:41 PM 4/1/2005 +0200, Fredrik Nyberg wrote:
Hi all Statalisters

I am wondering about testing two proportions in the following setting:
Infinite overall population size (for all practical purposes)
Sample 1 taken from overall population
Sample 2 subsampled from sample 1

Since these are not two independent samples from the same overall population, I think I need to adjust the test by making a finite population correction somehow for sampling sample 2 from sample 1. Only I can't find any details on this nor a test that appears suitable in Stata.

Can anyone help?
What is the correct test?
Is there Stata-code available?
My initial inclination is to go to sample 1 and code each case as 1 if not in sample 2, 2 if in sample 2. Then, use the -prtest- command to compare the 2 groups. But, my initial inclinations have been wrong before. The -prtest- command is discussed on the last page of this handout:

http://www.nd.edu/~rwilliam/xsoc592/lectures/TwoSample-Stata.pdf

*
* 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/

*
*   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/



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