Search form

Ask an Advisor Answers

Question:
I have resources to interview 50 people in my community of 50,000. How can I make this survey statistically valid?
Answer:

Excellent question. And in response, a key point is that statistical validity is a continuum, rather than a yes-no matter. There are degrees of validity; you want to make your survey as valid as possible, within the resources you have available. A second key point is that your sample should most accurately represent the population or group you wish to generalize to. What group would that be in your case? In other words, if you would like to generalize your findings to the entire community, then your sample should be drawn from that entire community, here of 50,000 people. If, however, you want your sample to be representative of (and generalize to) a smaller unit, then your sample should be drawn from that smaller population. In your situation, if you are interviewing 50 people, a commonly accepted procedure is to draw a random sample of 50 from all the members in your population ? which could be your entire community, or possibly a smaller group. (A random sample means that each member of the population has an equal chance of being included in the sample.) You would then need to obtain, or approximate, a list of all the members of your population. If your intended population is the full community, this could involve a town census list (if that is available), or a random selection of housing units. If your intended population is smaller, your task may be somewhat easier. Most good social science research methods textbooks will provide further details on survey mechanics. You may also wish to consult Chapter 3, Sections 12 and 13 of the Tool Box for some more general information on conducting interviews and surveys. If, though, a truly random sample is for some reason not possible, then you should normally make your best possible approximation to it, while being aware that this will usually diminish the statistical validity or representativeness of your findings. One more suggestion is to review your starting assumption that you have resources to interview only 50 people. Can this number be expanded? You might expand it either through obtaining additional resources, or through the use of volunteer trained interviewers, or using a smaller number of questions, or through group interviews, or an e-mailed or Web survey, or through staggering the survey dates, or through other efficiencies? On further consideration, it may be possible to include more than 50 people in your sample after all. We hope some of these thoughts may be helpful to you, and hope your survey proves to be very successful. Thank you for writing to us at the Community Tool Box; and all best wishes for success in your important work.

Question Date: Thu, 03/29/2007