Privacy Concerns and the Census Long Form: Some Evidence from Census 2000
KEY WORDS: privacy, confidentiality, decennial census, census long form, cooperation, question sensitivity, respondent reluctance, data uses, income
Public concerns about privacy predictably increase during decennial censuses, according to survey evidence from the 1990 and 1980 censuses. This paper examines trends in privacy and confidentiality concerns during the census in 2000, examines reasons for respondents’ concerns, and explores what information might allay them.
During March and April of 2000, a series of five cross-sectional surveys was conducted by InterSurvey, Inc. (later called Knowledge Networks). A random digit dialed sample was recruited, and households that agreed to participate were provided free hardware and internet access, allowing surveys to be administered using a Web browser. In May, a sample of households was recontacted for a survey focused on privacy and confidentiality concerns.
Log linear analyses of these data confirm that concerns about privacy and confidentiality rose significantly during the course of the 2000 census, especially among long form recipients. Moreover, people who received a long form, or who were concerned about privacy, were more likely to return an incomplete census form, or to fail to return it at all, by their own reports. In general, the most common reasons respondents gave for their reluctance to answer questions were that “it’s none of the government’s business” and “there is no purpose for the question,” but different questions aroused different concerns. The most sensitive long form question (income) was also the most likely to arouse concerns about confidentiality.
Most respondents expressed the need for more information to better understand the census. Most commonly, they wanted to know the reasons why particular questions were asked, followed by uses of the data. Long form recipients, and respondents who were concerned about privacy, were more likely to want information about the reasons for particular questions; the latter also wanted to know more about procedures for protecting individual data.
Finally, the survey included questions to test whether respondents’ reluctance would be reduced by providing the information they seemed to be asking for. Reluctant respondents were given brief explanations of reasons for the question, then asked if they felt more or less willing to answer it. Most respondents were unmoved by an explanation, but for two questions (income and disability) there were significant improvements in professed willingness to answer.
CITATION: Elizabeth Martin. 2001. “Privacy Concerns and the Census Long Form: Some Evidence from Census 2000.” Proceedings of the American Statistical Association (Survey Research Methods Section).