Select a sample-size calculator program. There are many free online programs available on research companies' websites.
Choose your confidence level and enter in the appropriate box. This number determines how confident you can be that your results are accurate. Options usually include 90 percent, 95 percent and 99 percent. Most researchers use the 95 percent option.
Determine your confidence interval and input this information. Also known as the margin of error, this factor determines how much inaccuracy you are willing to have in your results. It allows you to extrapolate a smaller sample to a large population. For example, your question asked a group of people if they watched a certain TV program and 45 percent said yes. If your confidence interval was 3 percent, then you would know that between 42 percent and 48 percent (plus or minus 3 percent) of the whole population did watch that TV program. 5 percent is a typical confidence interval.
Indicate your population size. If you have a large population to choose from, enter 20,000, as there is little statistical difference above this number. If you have a limited number of people, i.e., all students in your college, enter as precise a number as possible.
Select your response distribution rate. This is your prediction about the number of people who will answer the survey in a specific way. Not all calculators include this step, but if you are uncertain, the default is 50 percent.
Press the "calculate" button to determine your results (some programs will do this automatically). The answer is the number of people that you need to interview, at minimum, to get a statistically valid sample.