[Congressional Record Volume 144, Number 59 (Tuesday, May 12, 1998)]
[House]
[Pages H3025-H3026]
From the Congressional Record Online through the Government Publishing Office [www.gpo.gov]




              CONCERNS ABOUT A FAILED CENSUS IN YEAR 2000

  Mr. MILLER of Florida. Mr. Speaker, I rise today to raise concerns 
that we are moving toward a failed census in year 2000. For over 200 
years this country has conducted a decennial census, starting back with 
Thomas Jefferson in 1790, to count all Americans. The purpose of this 
census is fundamental to our democracy in this country because it is 
the one man/one vote belief. The only way you know you have the one 
man/one vote philosophy is you have to count people every 10 years.
  This is the basis of elected representatives, whether it is the 
school board or Members of the House of Representatives, so it is so 
critical that we do that. Also, billions and billions of dollars that 
flow out of Washington or out of State capitols are based upon census 
information, so it is absolutely critical that we have a census that is 
conducted in year 2000 as one that is the most accurate possible, and 
as one that is trusted and believed in by the American people.
  However, for the year 2000 census, the Clinton administration has 
proposed a radical new idea. Without the approval of Congress, they do 
not want to count everybody now. They have all these smart people here 
in Washington with all these big computers. They say we are going to 
use sampling and we are going to estimate the population. So for the 
first time in history, they are going to count less than the full 
population of this country, and this is where the risk is so great.
  The General Accounting Office, which is the auditor for the Federal 
Government, a nonpartisan organization here in Washington, D.C., has 
said we are moving toward a failed census. Every report they have 
issued, they have said--the most recent one being in March--that the 
risk of failure has increased because they have developed this complex 
scheme that many of us believe cannot be completed. Even if it is 
completed, it will not be trusted by the American people.
  We believe that the President is trying to use more political science 
than empirical science in developing this plan. Last week we had a 
hearing on the subcommittee with oversight of the census. There were 
two fact points I think we learned at that hearing. First was the fact 
that the 1990 census was not that bad of a census. It was the second 
most accurate census in history. But the second part of that census, 
which was dealing with sampling and adjustment, was a failure.
  Let me explain that in a little more detail. The way they conducted 
the 1990 census is they went out and did an enumeration of the entire 
population of this country and counted 98.4 percent of the people; 
again, not a bad count, the second most accurate in history. Then they 
conducted a sample of 150,000 households. They were going to use that 
to adjust the total population they have just counted.
  The attempt at sampling was a failure. Fortunately they did not use 
it, because if they had used it, for example, the original 
recommendation from the Census Bureau was to take a congressional seat 
away from the State of Pennsylvania. They find out 2 years later there 
was a computer mix-up that gave them the erroneous information, so they 
would have taken representation away from a State, Pennsylvania, 
falsely, because of computer error.
  They also found it was less accurate when we deal with populations 
under 100,000. So for communities under 100,000, cities and towns for 
census blocks, census tracts, which is the fundamental building stone 
that we use to build up our congressional district as such, it is less 
accurate, these are the Census Bureau people telling us, in their 
analysis of the attempted use of sampling.
  So sampling was a failure in 1990, even though the census was not 
bad. So what does the Clinton administration propose now? They want to 
totally rely on sampling. Instead of starting off counting everybody, 
they only want to count 90 percent of the people, so they are going to 
say 1 in 10 of the people we are not going to count. We are going to 
have 90 percent of the people.
  That is starting off the sampling, and you have nothing to fall back 
on, because when they come up with this adjustment sample, which is 
going to be on 750,000 households, larger than 1990, five times as 
large, they plan to do it in half the amount of time. Unrealistic.

[[Page H3026]]

 They are going to totally rely on it. If sampling fails like it did in 
1990, for the year 2000 they have nothing to fall back on. They run the 
risk of a total failure there.
  One of the things they did in 1990 is they released information on 
what the total census was. They showed that different parts of this 
country had populations deleted. For example, Bucks County up in 
Pennsylvania, a suburb of Philadelphia, had 3,000 people deleted from 
their county by the Census Bureau computers because the Census Bureau 
computers said, on average, they didn't deserve 3,000 people. So even 
though they were counted, they were subtracted. That is what upsets the 
people. That is the reason people say we can't trust a census where you 
start deleting people after they are counted.
  One thing we find out now, one reason they only want to start with 90 
percent of the population, is they can justify not releasing that 
information and showing the deletions. It is a very risky plan. It is 
moving towards failure. We need to share with the American people 
exactly the details, and we must have a census that is trusted by the 
American people, not the plan that has been proposed by the President.

                          ____________________