instructions taking up a full page. (Those listings do suggest different choices and priorities between the two countries, such as the prominent treatment of children in joint custody arrangements and refugees on the Canadian form.) The 2001 United Kingdom census form devotes a full page to instructions and rostering of household members (and, separately, short-term visitors). Though it is still a rather large presentation, the United Kingdom form is a remarkably concise distillation of formal residence definitions and rules that are—if anything—more elaborate than the U.S. residence rules (see Appendix B).
It is not unusual for multiple residence questions to be asked and additional address information collected on the census form. The Australia and New Zealand de facto censuses also ask for usual residence information, and the Swiss form asks for two addresses (and includes a check box to indicate which is the place where “you usually reside (4 or more days a week)” in a small amount of physical space on the page. (Again, the Swiss form is completed by each person, not each household.)
Some of the forms include reminders or cues to respondents, in addition to the standard instructions. The most prominent example of these is the New Zealand census form that—despite being light on formal instructions—makes repeated entreaties to respondents to be sure to include babies in the report. The Canadian census form for 2001 includes the reminder that the respondent be sure to include himself/herself in the count.
The 1980, 1990, and 2000 decennial censuses each included an Alternative Questionnaire Experiment (AQE) as part of their programs of testing and experimentation. Samples of census households (between 42,000 and 50,000 households in 1990 and 2000) received forms that differed in various ways from those received in all other households. The 1980 experiment focused mainly on variations on matrix-style forms most conducive to the electronic data-capture technology used at the time. The 1990 AQE was the most ambitious of the three; its experimental panels included completely redesigned versions of the census long form as well as radical structures, such as a kit of personal-response census forms rather than a single household form (DeMaio and Bates, 1992).
The 2000 AQE included experimental groups that assessed the impact of graphical and narrative instructions to guide the flow through the census long form; it also contained a group that repeated the census race and Hispanic origin questions in their 1990 form (for contrast with the 2000 version, which introduced the option of multiple-race reporting and restructured the ques-