Surveys are subject to two types of error, sampling error and nonsampling error. Because the Census of Governments covers all governments, there is no sampling error. However, the data are subject to nonsampling error, which includes all other sources of survey error such as nonresponse, incorrect reporting, coverage, misclassification of governments, and inaccurate coding of data.
A variety of procedures were applied to keep nonsampling errors to a minimum, including the following:
Review of state legislation for all Government Units Survey (GUS) responses that indicated that they were no longer a government
Review of secondary sources
Reviews of lists of governments supplied by the states, when available
Nonresponse follow-up calls to encourage response
Extensive research on addresses for Undeliverable as Addressed units and others lacking adequate contact information
Use of advance and reminder notice cards and letters for the GUS
The final response rate for the GUS was 75.0 percent with 38.4 percent of the responding units responding via web, 57.5 percent responding via paper, and 4.2 percent responding via some other means.
Back to How the Data are Collected