-
Notes
- Russell Sage Foundation
- Chapter
- Additional Information
131 Notes INTRODUCTION 1. Congressional office of Ron Paul, press release, March 28, 2000. 2. Individuals could request a census form in Spanish, Chinese, Tagalog, Vietnamese, or Korean or get help from language assistance guides printed in forty-nine languages. 3. Although we are primarily concerned with census behaviors rather than census knowledge, research has also found a participation gap following from the same logic (see, for example, Berinsky 2005; Eveland and Scheufele 2000). And even the research focused exclusively on knowledge levels typically assumes a fundamental link between knowledge and behavior. 4. There is considerable debate among academics regarding the extent of the decline in civic engagement and social capital documented by Robert Putnam (2000). For criticisms of this thesis, see Ladd (1999), Levi (1996), Skocpol and Fiorina (1999), and Tarrow (1996). 5. The data collection was funded by a consortium of private foundations, including the Russell Sage Foundation, the Ford Foundation, the William and Flora Hewlett Foundation, the Annie E. Casey Foundation, the John D. and Catherine T. MacArthur Foundation, the Andrew W. Mellon Foundation, and the Carnegie Corporation of New York. Norman Nie and Jane Junn were the principal investigators of the survey collection. 6. Knowledge Networks’ panel recruitment methodology uses the quality standards established by selected RDD surveys conducted for the federal government . All telephone numbers have an equal probability of selection, and the sampling is done without replacement from the entire U.S. telephone population who fall within the Microsoft Web TV network (87 percent of the U.S. population fall within this network). Selected numbers are called, and the head of household is informed that the household has been selected to join the Knowledge Networks panel. The household cooperation rate during this time averaged 56 percent. Detailed information on the Knowledge Networks 132 Notes methodology and comparisons of probability versus volunteer Internet samples can be found on their website at: http://www.knowledgenetworks.com/ ganp/index.html and http://www.knowledgenetworks.com/info/press/ papers/Volunteer%20white%20paper%2011-19-03.pdf. 7. We conducted all analyses using weights provided by Knowledge Networks. Poststratification weights, benchmarking to Current Population Survey (CPS) distributions on demographic benchmarks for age, gender, ethnicity, region, and education, reduce the sampling variation. They include a nonresponse adjustment using the demographic data of those who were initially selected but who did not complete the surveys. 8. The individual completion rates for each of the monitoring surveys: monitor 1 (83 percent), monitor 2 (82 percent), monitor 3 (61 percent), monitor 4 (58 percent ), monitor 5 (64 percent). The differences in completion rates generally reflect differences in the fielding period. CHAPTER ONE 1. The text of this chapter was drafted primarily by Kenneth Prewitt and draws on his experiences as director of the U.S. Census Bureau from 1998 to 2001. 2. We make reference here to two distinct measures of census mail-back cooperation : the mail-back response rate and the mail-back return rate. The response rate refers simply to the percentage of census forms returned out of all mailed out by the Census Bureau. Census forms are mailed or delivered using a master address file, and when that file is compiled, the Census Bureau does not know which houses or apartments are occupied. It is not until a form has been delivered, not returned, or followed up with a personal visit that the bureau learns whether a given residency is occupied or vacant. It takes many months to complete all the checks required to determine the vacancy rate of households on Census Day, which can be as high as 10 percent of the residencies in the United States. After vacancies and other address errors are subtracted, the Census Bureau has a list of valid occupied households . Using that as the denominator, it calculates a mail-back return rate. In 1990 the return rate was 75 percent, or 10 percent higher than the initially announced response rate. See figure 1.1 for an apples-to-apples comparison of response rates, including the 2000 estimated response rate. We discuss the distinction between the response rate and the return rate in greater detail later in the chapter. 3. The Census Bureau has used two methods to estimate net undercount figures: a demographic analysis, and in 2000, a separate post-enumeration survey, termed the Accuracy and Coverage Evaluation (ACE) survey. 4. Erroneous inclusions might include, for example, a child born after April 1, a person who died before April 1, or...