You are viewing an obsolete version of the DU website which is no longer supported by the Administrators. Visit The New DU.
Democratic Underground Latest Greatest Lobby Journals Search Options Help Login
Google

Reply #67: True when population size is much larger than sample size. [View All]

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Topic Forums » Election Reform Donate to DU
thanatonautos Donating Member (282 posts) Send PM | Profile | Ignore Sun Jan-16-05 04:00 AM
Response to Reply #59
67. True when population size is much larger than sample size.
This is just nitpicking, but anyway.

If n=1,000, and the polling universe is N=1,000,
then clearly the margin of error is zero, since
all entries in the list have been sampled.

The intuitive belief is actually therefore correct:
when the sample size approaches the size of the
polling universe.

Nevertheless, your approximation for the margin of
error is extremely accurate for the cases you list,
namely n=1000, or n=10,000, with N=10^6, N=10^7, N=10^8.

When n approaches a significant fraction of N, however,
there are finite population corrections to consider.

A conservative estimate of the standard error in
a sample percentage, sampling without replacement
from a list of zeros and ones, for a sample of size n
from a list of size N is that the standard error in
the sample is no greater than:

sigma = f * 50%/(n^1/2), with f = ((N - n)/(N - 1))^(1/2)

For your cases, we have:

(n=1000, N=10^6) f=0.99950037
(n=1000, N=10^7) f=0.99995005
(n=1000, N=10^8) f=0.999995

(n=10000, N=10^6) f=0.9949873
(n=10000, N=10^7) f=0.9994999
(n=10000, N=10^8) f=0.99995

So finite population size corrections to your estimate
of the mean sample error are not very large at all in
any of the cases you mention: the largest being 0.25% for
the case n=10000, N=10^6.

(These are only upper bounds, one can certainly do better
by using the sample itself to estimate the standard
deviation of the whole list.)





Printer Friendly | Permalink |  | Top
 

Home » Discuss » Topic Forums » Election Reform Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002 DCScripts.com
Software has been extensively modified by the DU administrators


Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC