[Congressional Record Volume 144, Number 73 (Tuesday, June 9, 1998)]
[House]
[Pages H4244-H4245]
From the Congressional Record Online through the Government Publishing Office [www.gpo.gov]
THE 2000 CENSUS: POLLING VERSUS AN ACTUAL COUNT
The SPEAKER pro tempore. Under the Speaker's announced policy of
January 21, 1997, the gentleman from Florida (Mr. Miller) is recognized
during morning hour debates for 5 minutes.
Mr. MILLER of Florida. Mr. Speaker, we are less than 2 years from the
beginning of the decennial census. The decennial census is a
requirement of our Constitution where we count everybody living in
America every 10 years. Since 1970 we have been doing it, and we are
gearing up now for the 2000 census. It is one of the most important and
controversial issues faced in public policy today.
It is controversial because, for the first time in history, the
Clinton administration has proposed a radically different approach to
be conducting the census. They have proposed this radical change
without the approval of Congress. For the first time in history since
1790, for the first time, they do not want to count everybody. They
only want to count some of the people and guess at the rest of them.
They want to use science to come up with estimates of a population,
rather than actually counting people, the hard work of counting people.
From the days of Jefferson and Washington, we have been counting the
population. Now they have come up with this radical idea.
It is a very important issue because it is fundamental to our
democratic system of government, because most elected officials in this
country are dependent upon an accurate census, and hundreds of billions
of dollars flow out of Washington and out of State capitals on the
census, so it is a critical issue.
The problem we are facing is we are moving towards a failed census.
The General Accounting Office, who is the independent auditor of the
Federal Government, has reported time and again that we are moving
towards a failed census. The Inspector General for the Commerce
Department has also warned us. So we have a serious problem.
Last week the President flew to Houston, mainly to raise money, but
also to talk about the census. I am glad the President has entered this
debate personally. His arguments in Houston were exactly why we should
not use his plan.
What the President talked about was polling versus sampling. Polling
is something we are all very familiar with. It is used in politics and
actually in business and for a wide variety of areas. What the
President was saying, and there is an interesting analogy, is that
polling, and let me quote the President, ``Most people understand that
a poll taken before an election is a statistical sample. Sometimes it
is wrong, but more often than not, it is right.'' That is what the
President said. ``Sometimes it is wrong, but more often than not, it is
right.''
Well, let us look at what really happens with polling. We will see
the problems with it and why it is so dangerous and risky to try to use
polling on the upcoming decennial census. One of the best ways to
evaluate whether a poll is accurate is looking at election results. Let
us look back at the last Presidential election in 1996, less than 2
short years ago.
Right before the election in November, all the major polls came up
with the results that weekend before the Tuesday election. As we all
know, President Clinton won and beat Bob Dole by 8 percentage points.
That is a factor, what do you win by, and what is the difference
between the winner and loser. Bill Clinton won and got 8.4 percent
higher votes, percentage points, than Bob Dole.
CBS/New York Times, on the weekend before the election, the polling
said the victory is going to be 18 points, not 8 percent, 18 percentage
points. ABC said 12 percentage points. The Harris poll said 12
percentage. The Wall Street Journal/NBC, said it was going to be a 12-
point election. CNN/USA Today, conducted by the Gallup poll, said a 13-
[[Page H4245]]
point spread, not 8 points. That was a 50 percent mistake.
How can we rely on polling? We cannot just say it is close enough for
government work. We are going to spend $4 billion on a poll that is not
going to be close, if it is based on the polling ideas, the President
wants us to risk that, and especially since it is something that is so
important and that is fundamental to our democratic system. It is just
wrong.
The President did not mention that back in 1990 we attempted to use
sampling. It failed in 1990. When they tried to use sampling to adjust
the population enumeration, it was a failure. It was a failure because
it would have, for example, taken a congressional seat away from
Pennsylvania and given it away without justification, because it turned
out 2 years later it was a computer error and never should have been
recommended.
It also says that adjusting, based on sampling, is less accurate when
you have populations of less than 100,000 people. I am sure big-city
mayors may like this, but we have to work with census tracts, we have
to work with smaller communities. How do we show this is going to be
trustworthy?
There is another thing I was concerned about in President Clinton's
comments. I do not think President Clinton means to divide America. He
said that Texas would have gained $1 billion if we had used sampling.
We are talking about a zero sum game. A zero sum game means if you give
$1 billion to Texas, you are going to take away $1 billion from
somewhere else. We only have a fixed amount of money when we get to
block grants. When we take money from one area to another area, we had
better explain to people why we are taking the money away.
For example, when we start adjusting the census and subtracting
people from the population, which they tried to do in 1990, that is
when we start making people upset and not trusting our system. We
cannot use this. This is not close enough for government work. It is
wrong. We need to do an actual enumeration.
____________________