selecting a sample

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

selecting a sample
Message posted by ld on November 8, 1999 at 12:00 AM (ET)

we have hundreds of cancelled memberships. we want to (telephone interview) survey a sample of the last 3 or 4 months of cancellations (i.e., 500). how do i determine a representative sample size?
ld

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: selecting a sample
Message posted by JG on November 9, 1999 at 12:00 AM (ET)

The best that you can do is choose a probabilistic or random sample for which the probability of not being representative can be calculated. There are many types of random samples such as stratified samples, cluster samples, simple random samples, etc.
You could of course use you judgement to choose a 'representative' sample but then there is no way of testing if you are right.
Remember that for a sample to be probabilistic every member of the 'population' has to have a known probability of being chosen. Some random samples may not be representative but the probability of this occuring can be calculated and kept at a low value.
In practice some judgement enters into the sampling process but we assume that the sample is random for computational purposes.

Re: selecting a sample
Message posted by ld on November 17, 1999 at 12:00 AM (ET)

Thank you *very* much!
ld

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.