Surveys can tell a business what to sell, advise politicians how to vote and show schools what to improve. Surveys inform decisions and manage conclusions — if they're done properly. Think of surveying as an art: It requires delicacy, calculation and forethought to record results that will help the client, or the person who ordered the survey. "The proper purpose of a survey is not to prove a point but to gather and report information that will help [clients] make good decisions," wrote Bill Coplin, a Maxwell School of Citizenship and Public Affairs professor. To assess a survey's results, examine the client's original instructions, the target population, the sample, the sample size and sample frame and the presence of sample bias.
Things You'll Need
- Pre-selected questions and/or topics
- Voice recorder
Evaluate the client's original instructions. If the survey deviated from the client's wishes, then it won't help the client. If it followed the client's instructions, then it will help him make informed decisions.
Examine the survey's target population. A target population "is the entire group of individuals about whom you gather information," according to Coplin. For example, if a politician wishes to find out how his constituency feels about a new policy, then the constituency is the target population.
Assess the sample — the amount of the target population selected for the survey. The larger the sample, the more accurate the survey will be. With more people interviewed, the survey will have a greater chance of predicting the feelings of the entire target population.
Examine the sample frame (how much of the target population you survey) and the sample size (how much of the target population responded to the survey). If the survey was face to face, then 75 to 95 percent of the sample frame should have responded. With a phone survey, expect between 40 and 75 percent; with mail or email surveys, anticipate 5 to 50 percent. If the response rates differ from these expectations, the survey was inaccurate.
Check for sample bias. For example, if a politician wishes to survey his constituency, he needs to ensure the survey covers the entire constituency. Surveys will sometimes interview more men than women, more rich than poor or more whites than blacks, making the survey useless.