Congressional TestimonyTestimony of Bruce K. Chapman
President of Discovery Institute
Before the House of Representatives
Committee on Government Reform and Oversight
February 29, 1996
On the 2000 Census
Good morning, and Happy Leap Year, Mr. Chairman and members of the Committee. I am honored to be asked my opinion about the 2000 Census.
Leap Year, indeed, is a good time to review the Census Bureau’s plans, because a great statistical leap into the 21st century is what the Bureau is proposing. The Congress may wish to look very hard before you take this leap, however, because it is a leap into the unknown.
I am particularly concerned about the plan to use sampling to adjust the hard count for the 2000 Census. Looked at through a green eye shade, it may seem prudent. Obviously, it is financially burdensome to obtain the last 10 percent or so of Census names, so sampling looks like a winning solution. Undoubtedly that thought must have weighed heavily in the recommendations of several bodies that are advocating adjustment. But if money is what matters most, you could save even more by relying on still more sampling. The Census is, indeed, expensive. But consider what you get with it. It holds up the whole statistical system of the Unite States. Having said that, I do acknowledge the legitimacy of the cost issue.
The statistical accuracy argument must also be acknowledged. It is embarrassing to have an undercount after every decennial Census. However, it is an embarrassment the republic has managed to survive each decade for two centuries. I am not sure how well we would survive its alternative. The problem with sampling and adjustment does not lie only with accuracy, which statisticians can debate, but with its admitted core of artificiality.
My argument with adjustment, thus, is not primarily a matter of statistical methodology. That’s not my background, in any case. If anybody could pull off an adjustment in a conscientious manner it is the U.S. Census Bureau. My experiences at the Bureau left me with the greatest respect for the honesty, ability and public spirit of the people who work there, right up to and including the outstanding current director, Dr. Martha Riche. Moreover, many of the statistical and management innovations the Bureau has proposed for the 2000 Census, including new “matching” software to spot double-counting, the digital “capture” of data on forms, more user-friendly forms and the added mail contacts with each address, all have excellent prospects, in my view. The 2000 Census, as with past censuses, will see improvements, regardless of adjustment.
Should an official adjustment take place, however, and you have one national number, the Census will still not have counted everyone, even in theory. For example, the present plan will have done nothing to count the literally hundreds of thousands of Americans who are not employed by the government and who are temporarily living overseas. (Overseas government employees now are counted, of course.) The people I’m talking about are U.S. citizens; some of them own homes and other property here; some are on the voter roles and motor vehicle license lists. I certainly agree with the proposal to explore the use of many other administrative records, such as Social Security and welfare rolls, to seek names of people who may have been missed in the Census head count. I also understand why the Bureau must count illegal immigrants, even though they may lack reliable addresses and their stay in this country may be very temporary, indeed. But shouldn’t we also be trying to identify citizens of the U.S. who happen to be out of the country around the Census time, but whose names can be found in public records and who count the United States as home? Isn’t their stake in the country important enough to be reflected in the numbers that are used, for example, to apportion Congress?
The answer given by the Census Bureau, with OMB no doubt looking over its shoulder, is that such a search would be costly. Well, if so, why isn’t it proposed to sample and adjust for the overseas Americans, too? But, meanwhile, the controversy of this very subject returns us again to the central problem in the 2000 Census plan, and that is the moral or philosophical concept of an official, adjusted enumeration. Sampling is a proven and accepted form of measurement, but there usually has to be some hard, broadly accepted data base from which the sample is drawn. More importantly, I can’t think of a statistical sample that is so closely tied to a civic act as is the Census enumeration, any more than I can think of an election survey that has the moral authority of an actual election.
The Census enumeration is a participatory function of government hallowed by Constitutional mandate. There are not too many comparable civic acts; voting is one, serving on juries is another. Perhaps participation in the Census falls between those two–it is neither as voluntary as voting, nor as demanding as jury service. In any case, the Census is one of the fundamental democratic institutions of our society, and I believe, it is a noble one. It gives us the numbers that make other numbers meaningful. It must not only be trustworthy, but palpably trustworthy.
The current Census director herself has spoken of the three legs of the Census stool that must dictate the process: cost, accuracy and public perception. The first two legs are important, as I have acknowledged, but if the latter breaks, the whole construction comes down. The term “public perception” could also be described as trustworthiness. In a time when public mistrust of government is rife, I question a change that would introduce the invention of statistical persons into the Census–robots constructed of sampled data and intellectual abstractions–to stand in the place of real human beings.
The Census has long represented an honest attempt to count every person in the United States–every real person, not every “virtual” person, not every “statistical analog” for a person, not every “full time equivalent” of a person, but every real person. Once citizens realize that the Census number is partly an invention–however brilliant an invention it is–they will begin to question many things. Suspicions will be aroused. How, some will ask, do we know that the creation of statistical persons does not have some partisan or factional or geographic regional basis? How does this corrupted Census count infect other numbers and political institutions?
Let me stress again my own full confidence in the integrity of the individuals who serve at the U.S. Census Bureau. It is largely because I don’t want to see that integrity even put up to question that I raise this problem of perception–of evident trustworthiness.
Let’s next consider legal problems with the Census; do you think you have them now? In fact, the costly lawsuits challenging the Census invariably are won by the government. Common sense tells a judge that a good faith effort to count every person cannot be faulted for occasional human mistakes, especially when plaintiffs can seldom if ever produce evidence of actual individuals who have not been counted. But under an adjusted Census it will not be difficult at all to produce evidence that the Census erred, and erred intentionally, by imputing fictitious people into neighborhoods where the real neighbors can readily prove that the interlopers do not, in plain truth, exist. In some cases, the fictional, imputed people are going to amount to a difference between drawing a redistricting line one place rather than another, and maybe even change the apportionment of states. (After all, isn’t that part of the whole idea?) In such cases, I suggest, common sense will now become the government’s enemy, because you will need to give all the judges and ordinary citizens an extensive course in statistics to persuade them that it is just and fair to invent people for something as supposedly straightforward as a U.S. Census. How especially do we explain including invented people in determining such matters as apportioning Congress when, as I noted before, we do not count real American citizens living abroad?
Maybe the American public will accept all this with total unconcern or even satisfaction. But maybe, instead, this will be seen as one more way the government is trying to pull a fast one on the people. Won’t any resulting public cynicism make the hard count even more difficult to obtain in the first place? If it contributes to further falloff of response, won’t that and the ever-present financial incentives, lead to still more sampling in subsequent Censuses, in a downward spiral? And won’t it seem perfectly acceptable soon to conduct the whole Census through sampling? And won’t there be an irrestible temptation to try, Census by Census, to improve the content of what constitutes the invented persons created through the sample, and hence the meaning of such invented persons?
To conclude, Mr. Chairman and members of the Committee, I am here to urge you to look at the Census as one of the most venerable civic institutions of the United States, and to do all you can to protect it. Once trust in it is undermined, it will be very hard, if not impossible, to restore.