This paper examines the dynamics of short-run changes in public perceptions of privacy and confidentiality during the course of the 1990 and 2000 censuses, as well as long term changes from one census to the next. Data collected by the Census Bureau before and after the 1990 census and by InterSurvey, Inc. during Census 2000 are used to fit log-linear models to address the following questions:

• How do the public’s concerns about privacy and confidentiality shift during a census?
• Are intra-censal changes in privacy concerns constant from one census to the next?
• Are there apparent trends in public concerns about privacy and confidentiality between 1990 and 2000?
• Are there changes in the structure of public opinion on privacy and confidentiality?
• What factors influenced shifts in privacy and confidentiality concerns during Census 2000?

Analysis provides evidence of both constancy and change in public reactions to the two censuses. The public’s perception of the census as an invasion of privacy rose during the course of each census; however, there is evidence that the privacy reaction was greater during Census 2000 than it had been in 1990. In addition, the data suggest that the public may no longer distinguish privacy and confidentiality concerns in the way it has in they past. The paper examines factors which influenced the privacy reaction to Census 2000, including receipt of the long form and awareness of the controversy about it.