The 1940 census was the first U.S. census to include a statistical sample.* Five percent of people -- one of every 20 respondents -- were asked an additional 16 questions, beyond those asked of the general population. In past censuses, for example, the birthplace of one's mother and father was asked of everyone, but in 1940, that question was asked only of the sample group. This sampling allowed analysts to ask more questions than would have been practical otherwise and to publish results ahead of the general census, which required individual tabulation. Understanding more about the sampling method can help researchers better determine if their ancestor participated and then look for their responses.
To achieve a random sampling, each census questionnaire had lines for 40 names, with two lines designated for the sampling. Individuals whose names appeared on one of those two lines would then be asked the additional questions. It is important to keep in mind that responses to census questions are only as accurate as the person providing the information, who may have been someone else in the household. The good news is we now have more insight than in censuses past: the 1940 census was the first to identify the informant in every household.
* Editor Note: Probability sampling, simplified, is a method of measuring a part to reflect the whole. Several factors apply, including random sampling to avoid bias, and a "statistically significant" sampling to control for a chance occurrence -- five percent of every one-hundred is considered statistically significant.
Help us improve this frequently asked questions area. Please send us feedback or additional questions.