Abstract

All surveys are susceptible to two broad classes of errors—sampling error and nonsampling error—both of which can result in false or biased information. Sampling, or random, error refers to the amount by which the selected sample finding deviates from data that would have been obtained if the entire population were surveyed. Nonsampling error is all the other errors that might confound the survey results. Together, the two types constitute total survey error. Because either class of error can significantly reduce the usefulness of the results of a survey, both sampling and nonsampling errors must be controlled. This entry focuses on nonsampling error, of which much less is known.