How much does the American Census lose to the British Census in quality?
In my long life as a sociologist and a pollster, I saw hundreds of questionnaires prepared by my colleagues in Russia and the United States as well as by sociology or media graduate and undergraduate students. I must confess that the questionnaire which the U.S. Census Bureau used for the 2010 Census is one of the sloppiest documents which I ever saw. Meanwhile, according to the U.S. Census Bureau (hereafter referred to as simply CB) official statement, «more than $435 billion in federal funds are distributed every year based on census results.» At the same time, it is said that the «10-year life-cycle costs of Census 2010 are about $14 billion.» These numbers indeed inspire awe.
In order to substantiate my harsh critique of such a respectful organization as the U.S. Census Bureau, let me compare two forms – the one used in the Census of 2000, and the current Census form. The most important question is put first in the 2000 Census, yet, while still at the beginning of this year’s census, it has since become a two-part question (the question #1 and the question #2). The answer to these questions would allow the government to determine the size of the entire U.S. population on April 1, 2000 and 2010, as well as the populations of individual American states, counties, cities, or other types of settlements. The first and most important question in the 2000 Census is the following: “How many people were living or staying in this house, apartment, or mobile home on April 1, 2000?” The respondents were supposed to provide the number of people who live in their homes, including foster children, roomers, or housemates; people living in homes on April 1, 2000 who have no other permanent place to stay; and people living in homes most of the time while working, even if they have another place to live. The respondents were asked not to include in this number college students living away while attending college; people in a correctional facility, nursing home, or mental hospital on April 1, 2000; Armed Forces personnel living somewhere else; and people who live or stay at another place most of the time.
On this year’s census, the two questions aimed to determine this significant figure, that is, the number of people who live in the United States on April 1, 2010, looked a bit different and was the same in both the “short” questionnaire which was to be filled by everybody, and in the «long» questionnaire which was addressed only to a part of the population. The first question asked, “How many people were living or staying in this house, apartment, or mobile home on April 1, 2010?” The second question asked, “Were there any additional people staying here April 1, 2010 that you did not include in Question 1? Mark X on all that apply.” The possible answers included: children, such as newborn babies or foster children; relatives, such as adult children, cousins, or in-laws; nonrelatives, such as roommates or live-in babysitters; people staying temporarily; or, no additional people.
Let me explain why the questionnaire of 2010 is much worse than that of 2000. First, a critical law in statistical and sociological studies which deals with different periods of time is to minimize as much as possible any changes in questionnaires over time in order to make the data comparable. For no reason, the CB violated this rule and in 2010 will receive data not formally comparable with that of 2000. The list of the people which should have not been included in 2000 is absolutely different from the list of the people which have been suggested not to be included in 2010. In 2000, Americans were asked not to include in the number of residents in their homes: college students living away, people in prisons, those in nursing homes or mental hospitals, while these categories were absent in 2010 and replaced with other categories. In 2000, the CB sternly demanded to include in the main number «foster children, roomers, or housemates,» while in 2010, the CB with the same confidence required people to exclude «children, such as newborn babies or foster children.» (Besides, the formulation of the question # 2 is extremely awkward, it is ambiguous and violates the elementary rules known to anyone who did not fail a survey methods course.) Second, the description of the people who should be included in the Census is extremely sloppy. There is no hint in the 2010 form (as well as in 2000 form) about the «fate» of, for example, foreign students or other temporary guests from abroad or Americans who were abroad on April 1, 2010.
I am afraid to present another problem – the danger to omit the significant number of illegal immigrants in the country. Of course, this issue is extremely sensitive. It is serious ground to suppose that many illegal immigrants will try to avoid the census. Still, this is not a reason to pretend, as the CB is doing considering the main census document, that the subject of illegal immigration is not of critical importance for the veracity of the census data. At the very least, the Census Bureau should have made an attempt to do something to counterbalance the negative odds for the census. Why didn’t the CB include at the very beginning of the form a statement that requested for everybody to be included in the census count who was in the territory of the United States on April 1, regardless of their legal status or if they were foreigners? And of course in this section of the census, in addition to the introductory letter, the United States CB would indicate that the data collected in the census form is totally confidential, much like the organizers of the 2001 British census. The declaration of confidentiality would be particularly important for those who hire illegal immigrants (for instance, as a nanny or care taker of elders), and who have to complete the form that arrives at their residence. How can the federal and the local governments plan their expenditures on education, police, and health service without trusting the CB official data of the number of illegal immigrants in the country on April 1? Is here not a case of political sensitivity undermining the veracity of the Census data?
Third, the major flaw of the current Census lies in its inability to produce reliable data about the size of the permanent population in the states, counties and cities. Indeed, in order to achieve this goal it is illogical (as the CB is doing) to ignore the temporarily absent population and temporarily available people on April 1 across the country. First of all, the CB does not ask to include the members of the family and other people who live at the given address permanently but were absent there on April 1. At the same time, the CB asks to exclude the people who live temporarily in another place to be included in the Census! The Census of 2010 repeats the approach of the previous 2000 Census which did not ask respondents to include the absent members of the household but at the same time explicitly required not to include «people who live or stay at another place most of the time.» It is incomprehensible why the CB excludes from the Census results in 2000 and in 2010 millions of people who were absent on April 1 in their permanent place of residence. Question # 10, which asks if a person included in the Census “sometimes lives or stays … at a seasonal or second residence,» does not improve the situation because it does not ask where this residence is located.
Take my example. I am a permanent resident of Michigan, yet at the same time I have an apartment in Florida in which I stayed this year from December to May. As a resident in the condominium in Florida, I got the Census form which (if I fill it out) will not clarify whether I am a permanent resident of Florida or Michigan. As a result, I will diminish the population of Michigan by one. Though I have yet to receive it, there is a chance that I will get a second form from Michigan as my forwarded mail. Let us suppose that I will get this second census form — how will I act? It is clear that I should not fill out both forms, but which one should I choose? If I choose the Michigan form I will deceive the CB because on April 1, 2010 I was not in Lansing but in Lauderdale. If I choose Florida, the choice will «insult» my state of Michigan. By their own initiative, some organizations not at all related to the CB (such as the Ohio Department of Aging) recommend that so-called «snowbirds» — people who spend winter in warm climates — fill out both forms but to not answer any question on the form received in Florida which indicates that the respondent has a seasonal or second residence (the question number 10). The Ohio Department of Aging contends that such were the instructions of the CB, but I when I called the CB, I received different directives. This organization department encourages Ohioans to share these tips with friends or family members who have already left for their seasonal residences. A funny story! Rather than have clear fool-proof order valid for the whole territory of the United States, we are recommended to get advice from an agency which is far removed from the statistical intricacies.
Another problem is that many people will be absent on April 1 and their homes and apartments will be closed. Since these people will not be included anywhere in the United States in the main number (since due to their absence they will be temporary residents everywhere), they will be irreversibly lost to the CB, assuming that they do not get their mail forwarded from their permanent residence. Census takers, whose primary responsibility is to collect Census information from the residents who have not sent back their 2010 Census forms, will not reach people who are absent for a long time.
Such a contrast with the last Census of 2001 in England! At the start, the Registrar General for England and Wales asks in the first question to «include anyone who is temporarily away from home on the night of April 29, 2001 who usually lives at this address.» What more, the Census form includes a special table about «visitors.» The form asks to «list in this table any visitor at this address, on the night of April 29, 2001, who usually lives elsewhere.» Then the form asks to indicate the first name and a surname of each visitors and their «usual address.»
It is necessary to add to my critique also the comments of Robert Goldenkoff, the director of strategic issues for the Government Accountability Office (GAO) who told recently CNSNews.com that «an estimated «50 million housing units out of a mail-out universe of about 120 million» will be non-respondents that will require an in-person follow-up to count».
In my opinion, the American public and the Congress should be aware of the low quality of the current Census. I suggest creating an independent committee to evaluate the situation and elaborate the measures that can, though probably through additional sampling surveys, collect the information which will help to correct the results of the Census. There is an army of brilliant demographers and statisticians in this country who will be able to find the way to solve the problems created by U.S. Bureau of the Census.