Sample Sizes and a Statistical Panel

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

Sample Sizes and a Statistical Panel
Message posted by Charles Miller on April 4, 2000 at 12:00 AM (ET)

I have just discovered this site, and it has been very educational. Thank you for providing it. I am mathmatically declined, but I know that I need to have a clearer understanding of the fundementals.

I read the information about sample sizes, but I have heard that a sample size is smaller for a panel (a prescreened group of people with similar specific characteristics).

Is this true? If not, how does one determine the sample size for a panel? For example: if I only wanted to survey people in Kansas City who have IRA accounts, what would be the minimum number of people required in the sample? Do I still need over 400? What information would I need before I begin?

Thank you for this forum to ask questions like this.

Charles Miller

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

Re: Sample Sizes and a Statistical Panel
Message posted by Alex on April 11, 2000 at 12:00 AM (ET)

Ideally, you want sample size (or N) to be as large as possible because this helps to insure attaining statistical significance. If N is large enough, you'll attain significance of almost anything. However, just becauas you have "statistical" significance, this does not mean that you have "substanitive" significance. Substanitive significance refers to the amount of variance in the dependent measure(s) accounted for by the independent measure(s). This can be determined in a variety of ways--depending on the type of statistical analysis you have just performed. If you were looking for a difference between groups, either an eta square or omega square would tell you the substanive significance of your results. If you are looking for a relationship between/among variables, r square (or R square for multiple regression) would tell you the substanive significance.

Typically, I like to use a sample size of 100 or more. Also, you can locate what is known as a power table that tell you the sample size you need, contingent upon the level of power you opt to obtain.

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