The issue isn't really about percentage of the population. Theoretically, a sample of 2,000 will give you the same level of confidence in your results for a population of 5,000 or 5,000,000 (provided that the sample can be assumed to be normally distributed, etc.). The issue, for most researchers is cost. A random sample of approximately 2,000 will give you a confidence level of about 95%. Larger samples will increase the level of confidence, but it will take many more cases to increase the confidence level even one percent, and that costs more money.

-----Original Message----- From: Michele Atkinson [SMTP:matkinson@DELTATRENDS.COM] Sent: Tuesday, April 06, 1999 8:58 PM To: SPSSX-L@UGA.CC.UGA.EDU Subject: Determining Population Size

Hi all--

I'm pretty sure this isn't quite SPSS related but it definitely has a statistical flavor to it, so bare with me :) I am wondering how to go about determining what sample size I will need in order to fall into a 95% confidence interval. There are approximately 20,000 stores we could survey. We would like to take a sample of those stores to get an approximation of the mean of the entire population. How do I go about deciding what percentage of the total population I should survey in order to be confident that the sample mean is a good estimation of the population mean? Please let me know if I am being unclear or if more information is needed.

Have a great day and thanks! Michele