Assume a confidence level that you will use in your measurements. For surveys of human interaction, a less precise 95 percent confidence level may be appropriate. For scientific studies of chemical reactions, a more exact 99 percent confidence level is appropriate.
Choose a confidence interval that you will require in the sampling as well. Typical intervals are several percent plus or minus the confidence level.
Convert your confidence level to a statistical z-score. You may use the second reference site to conduct this conversion.
Estimate the response rate of your sample. So if you are conducting a phone survey and 60 percent answer the questions, then your response rate (p) would correspondingly be .6. For an unknown estimated response, you may use .5.
Fill in the formula to complete the sampling scheme and return your sample size.
Sample size = ((z score)^2 X (p) X (1 - p))/ (confidence interval)^2