Sample Size: How Big Is Big Enough?

In conducting a survey, how large a sample do you need to feel confident about the accuracy of the results? Are 100 responses enough, or do you need 1,000? Whatever number you decide on, what response rate can you expect? In other words, how many people do you have to survey to get the number of responses you want? Let’s take a look at these questions.

How Confident Do You Want to Feel?
The size of your sample depends on how confident you want to be about your results. They key question is: "Are these results truly representative of a larger population?" If you would like to have a confidence level of 95%, you would need to have a margin of error of 5% or less. This means that there would only be a 5% chance of your results differing from the results you would get if you surveyed the entire population. (The margin of error is also called the confidence interval.)

There is a simple equation you can use to determine margin of error:
The margin of error equals 1 divided by the square root of the sample size. Suppose you had a sample size of 400. Your equation would be 1 divided by 20 (the square root of 400), which equals 5%. So to have a confidence level of 95%, you would need to have a sample size of 400.

Obviously, the higher the confidence level you want, the larger your sample needs to be. A confidence level of 95% is widely considered to be acceptable.

Of course this assumes that you have a truly random sample. As experienced pollsters know, it is easy for sampling bias to sneak into a survey without anyone noticing.

How Many to Ask?
So once you have decided on a confidence level, you can easily determine how large a sample you need. But how many people do you have to survey to get the desired number of responses? That is a trickier question.

Response rates vary greatly from one population to another. One of Mineful’s clients, a medical association, received a 25% response rate to a member satisfaction survey. Another client had a similar response to a survey conducted in connection with a conference. But such high rates are the exception rather than the rule. For a survey conducted through cold email invitations, a response rate of 1% is considered good.

Response rates are affected by a number of variables:
  • How strongly people feel about the survey topic.
  • How much loyalty people feel toward the organization conducting the survey.
  • How easy it is too complete the survey.
  • How confident people feel that someone will pay attention to their responses.
Of course these factors can be difficult to measure, and experience is the best way to predict response rate.

Fortunately, there is an online sample size calculator that can make this whole process much easier (http://www.raosoft.com/samplesize.html). These are the variables you will need to plug in to use this calculator:
  • Acceptable margin of error
  • Acceptable confidence level
  • Population size (for populations smaller than 20,000
  • Response distribution
The calculator explains what each of these variables means.

1 comments:

  • Thank you for all the great posts from last year! I look forward to reading your blog, because they are always full of information that I can put to use. Thank you again, and God bless you in 2010.

Post a Comment











Back to TOP