Is there a relatively easy way to determine the sample size required to be statistically valid for a marketing test?

Is there a relatively easy way to determine the sample size required to be statistically valid for a marketing test? Can you point me to good resource(s) for this type of information?

A:

Danette McGilvray's Answer: As you know, a sample is a collection of observations that represent only a portion of the total population of interest. In addition to the sample size, you need to ensure that your sampling method is such that your sample population is statistically representative of the total population.

A quick resource to help you select a valid sample size for a targeted test goal can be found in the article, "How to Design a Valid Sampling Test" by Amoy X. Yang. The article was published in DM Direct, August 2003, and can be found on DMReview.com at http://www.dmreview.com/article_sub.cfm?articleId=7230.

Another quick resource is this online statistical sample size calculator. http://www.psychnet-uk.com/experimental_design/statistical_sample_size_calculator.htm.

For those working in the area of testing, it is important to have a basic understanding of confidence levels, confidence intervals, error tolerance levels and other terms pertinent to sampling and testing. If you are new to statistics, a fun and informative beginner's guide can help: The Cartoon Guide to Statistics by Larry Gonick and Woollcott Smith, 1993.

Check with a statistician to ensure your sampling method and sample size meets your needs. There may be someone in your company who can help, as many roles (engineers, actuaries, etc.) have a good background in statistics. Because a marketing test takes time and money, it's worth the time and effort to make certain your test will return valid results.