| Date: | Fri, 14 May 2004 12:16:42 -0400 |
|---|
| Reply-To: | Richard Ristow <wrristow@mindspring.com> |
|---|
| Sender: | "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU> |
|---|
| From: | Richard Ristow <wrristow@mindspring.com> |
|---|
| Subject: | Re: Sample Size question |
|---|
|
| In-Reply-To: | <00a401c439c0$73f5d010$8c00000a@opioff> |
|---|
| Content-Type: | text/plain; charset="us-ascii"; format=flowed |
|---|
At 10:33 AM 5/14/2004, Gerd Weingrill - Opinionis wrote:
>I am doing a analysis for the chamber of commerce. Total population is
>12000 companies I want to know how many companies I have to ask if a
>total population is 12000 in my state and conf. level is 95 % and I
>want to be sure that results are within +/- 3% (6 % interval of
>confidence) to say the result is valid for the total population.
>
>In a former survey the response rate to an email sent survey was 10%
This does not address the power-analysis question you're asking, but:
At a 10% response rate, the worst problem is not small sample size, but
sampling bias. After all, those surveyed chose whether to respond or
not, and probably for systematic reasons: sometimes, the most angry
respond; sometimes, the disaffected feel they aren't listened to and
only the happy respond. (I'm sure those on the list who do surveys can
think of many more possibilities.)
| 
