WASHINGTON — There were problems with address lists and some people were counted without proof they were there, but last year’s national census was “well executed in many respects,” the National Research Council said Monday.
Several innovations proved successes in the count, including paid advertising, aggressive recruitment of enumerators for follow-up operations, data operations and the redesigned questionnaire, according to the Council’s report.
The Council is a division of the National Academy of Sciences, an independent organization chartered by Congress to provide scientific guidance to the government.
Post-census studies indicate that fewer people were missed in 2000 than a decade earlier, reducing the undercount from about 4 million in 1990 to 3.3 million last year, the study said.
This reduction was particularly true for groups that have been harder to count in the past, including children, minorities and people who rent their homes, the Council said.
However, the study said a large part of the reduction in people missed resulted from what the statisticians call imputation — listing of people in the census even though they did not return a form and could not be reached for follow-up.
Census 2000 included 5.8 million people who were imputed, the report said, up from 1.8 million in 1990.
In many cases, information on the number and characteristics of a family can be obtained from neighbors.
However, the Council also found “a much more problematic group,” of some 1.2 million people in last year’s census “who were imputed into the census when there was no information about the size of the household or, in some instances, whether it was occupied.”
People who were imputed into the census were disproportionately likely to be minorities, renters and children, the panel noted.
The panel urged the Census Bureau to investigate the factors leading to this type of imputation.
The study also found fault with the effort to develop a master address file. While the concept was good, it said there were problems in execution that may have led to duplicate or erroneous counting.
An initial study of the returns found some 6 million people who may have been duplicated.
After study, 3.6 million of them were deleted and 2.4 million were reinstated in the count.
On the positive side, the mail response rate in 2000 was the same as in 1990, hailed as a success in light of the declines of recent years.
Census officials had launched their first paid advertising campaign in hopes of stemming the declines in mail returns and the report found that effort to be one of the count’s success stories.
The redesigned form and a new mailing strategy also encouraged response, the study said.
It also praised the use of improved technology for collecting data from the census forms and contracting out for data services.
On the Net:
National Academy of Sciences: http://www.nationalacademies.org
Census Bureau: http://www.census.gov