Table L.1 shows allocation rates for short-form population items for all household members, blacks, and Hispanics, separately for people receiving the short form ("100%") and people receiving the long form ("sample").1 Rates are also shown for the "worst" 10 percent of census tracts in the country—worst in terms of those having the highest overall rates of allocation.
Tabulating allocation rates for items common to the short and long forms by type of form permits one to understand the impact of census processing and follow-up rules on nonresponse and allocation rates. Census procedures in 1990 called for questionnaires mailed back to processing offices to be reviewed and edited for consistency in reporting of related items and for missing or incomplete responses. Rules were established for determining whether a questionnaire "passed edit" and was moved along for processing, or "failed edit" and was set aside and scheduled for follow-up. For "failed edit" cases, census personnel in the various offices attempted to contact households by telephone, and when this failed or could not be done, they scheduled a field visit by enumerators to obtain the missing or otherwise incomplete information. For cost and other reasons, however, the rules specified that only a sample of "failed edit" short-form cases for which telephone contact was unsuccessful would be scheduled for follow-up. Thus, only 1 in 10 of the short-form questionnaires that failed edit were sent for follow-up, whereas essentially all long forms that failed edit were followed up. As a result, the items on the short form had substantially higher allocation rates than the same items appearing on the long form (see Table L.1).
The impact of this differential treatment of short and long forms (in terms of follow-up) is especially reflected in the allocation rates for Spanish origin overall and for all short-form population items for the worst 10 percent tracts. Thus, the allocation rate for Spanish origin was 10.5 percent of short forms but only 3.4 percent of long forms. The rate for Spanish origin in the worst 10 percent tracts was 19.0 percent of short forms but only 5.4 percent of long forms. Overall, allocation rates for the basic population items—namely age, sex, and race—averaged between 2 and 3 percent of short forms and 1 percent of long forms.
Table L.2 shows 1990 allocation rates for selected short-form and long-form population items for people in households that received the long form, by whether the household mailed back the form ("household respondent") or the form was obtained by an enumerator in the follow-up stage of census operations ("enumerator-filled"). (Enumerated-filled forms also include all forms obtained in the list/enumerate areas—see Appendix B.)
With few exceptions, allocation rates are substantially higher for forms obtained