The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
The 2000 Census: Counting Under Adversity
the difference was smaller in the C2SS. For most housing items, imputation rates for enumerator returns were higher than the rates for self returns for both surveys.
At least three factors argue why the C2SS could be expected to achieve lower imputation rates than the long-form sample and, in particular, why the C2SS interviewers could be expected to outperform the census enumerators. First, the C2SS interviewers were more experienced and much better trained than the temporary census staff, which meant that they were better able to obtain responses from reluctant respondents. Second, the C2SS interviews used computer-assisted telephone and in-person interviewing. (By design the C2SS used sampling for households that did not mail back a return and did not respond during the initial telephone-follow-up phase; the personal interview sampling rate was 1 in 3 of nonresponding households.) Third, the goal for the C2SS was to collect all of the information. In contrast, the essential goal for the 2000 census was to obtain a complete count—if household respondents balked at answering the additional long-form-only questions, the enumerators were not as likely to press hard for a response as would the C2SS interviewers. Yet these factors were apparently not sufficient to overcome respondents’ reluctance or inability to answer questions on income and housing costs in the C2SS as well as the census.
7–C.3Missing Data Patterns for Additional Items
No analysis has been conducted to date of patterns of response and nonresponse to the additional long-form items; that is, whether people tended to omit single items or clusters of items or most items. The panel made a limited set of tabulations of nonresponse patterns of cases in the A.C.E. E-sample who fell into the 2000 long-form sample and whose records were augmented at the panel’s request by the additional long-form items. These tables focused on the income and employment items, which exhibited high nonresponse rates.
A tabulation of people age 16 and over in the E-sample long-form records found that 71 percent answered all 9 income items, 11 percent answered 6–8 items, 2 percent answered only 3–5 items, 8 percent answered only 1 item, and 8 percent did not answer any of