Press Briefing -- June 14, 2000
|Thank you for joining us today, which happens to be Flag Day, for those of you who
overlooked that in your morning paper, and also the 225th anniversary of the U.S. Army. And
since I'm the son of a career Army officer, I had to say that.
My name is Steve Jost. I'm with the Communications Office of the Census Bureau.
We're here with one of our standard operational press briefings with some guests who have
joined us today for a special announcement. And I just want to take you though the agenda.
Director Prewitt will have some operational update remarks. Andy Pincus, the general counsel of
the Department of Commerce is with us. He will make an announcement. And former directors
of the Census Bureau, Martha Riche and Dr. Barbara Bryant are with us as well, and they'll have
some remarks at the end of the formal program. Then we'll turn it over to the questions from the
press. We ask, so that we can get good audio pick-up, that you wait just a couple of seconds until
someone with a hand-held mike can get you on the air. You'll identify yourself and affiliation,
and we'll alternate from questions from the room to reporters who are with us by phone. And
with that, I give you Director Prewitt.
DIRECTOR KEN PREWITT: Thank you very much, Steve. I will give a bit of an
operational update. We met just last week for an operational briefing, and I reported to you then
that I had been speaking in various sections of the country and have been, over and over,
impressed with the kind of dedication, commitment, and enthusiasm of the census enumerators
who are out there now doing Non-Response Follow-Up. And indeed, I want to report today that
this particular phase of the census, Non-Response Follow-Up, we are now 98 percent complete -
that is, 98 percent of all of the housing units that did not respond to the census are now accounted
for. That leaves less than one million of our housing units left to visit in this operation. When,
of course, you put that 98 percent together with the mail-back response rate, that means we're
now over 99 percent complete with this particular phase, two phases of the census and we see no
obstacles to keep us from completing this effort between now and July 7th, which is the date on
which we had hoped to be complete.
Indeed, the 10 states of our Denver region are completely finished. That is, all the Denver
region has now completed its Non-Response Follow-Up work, and the island of Puerto Rico is
now at 100 percent with respect to Non-Response Follow-Up, and we expect the entire Los
Angeles region and the Atlanta region to reach 100 percent of their housing units in just a matter
Indeed, we continue, of course, to process data. Our data capture centers have now
scanned in over 121 million questionnaires, 30 million of which were enumerator forms. And
respondent data itself, or themselves, for over 117 million questionnaires have now been
transmitted to headquarters for processing and electronically captured. That is, they have now
become Title 13 data, which is, of course, the final big key issue for us - 117 million
questionnaires have now moved through that process, again, with the same levels of accuracy
that we had been experiencing earlier.
Our Accuracy and Coverage Evaluation program is underway. More than 90,000
telephone interviews have now been completed, and the balance of the 314,000 interviews will
be done in person beginning June 19th. And we expect that, of course, to happen on schedule.
As I said last week, we feel strongly about this being a good census. And what do we
mean when we describe it as a good census? Well, for the career professionals at the Census
Bureau, a good census must have three components. First, it is operationally robust and
successful. Census 2000 meets these tests. Our outreach and promotion efforts were
unprecedented in their scope and the diversity of communities we targeted. We assembled the
largest peacetime workforce in our history - more men and women than were mobilized to land
on the beaches of Normandy. And we've experienced the highest levels of public cooperation in
both the mail-back and the door-to-door enumeration phases than we have witnessed in several
Secondly, a good census is open, transparent, earns the public's trust, and has been
enhanced and improved by critical and public scrutiny. Census 2000 is the first census in our
nation's history that has been planned and implemented, with every detailed scrutinized and
observed, by an unprecedented array of expert scientific and governmental oversight and
advisory panels. The higher public cooperation is also, we believe, a measure of a renewed
public trust and confidence in the census operation.
Third, a good census is dedicated to being fully inclusive. It is a census that has spared
no effort to ensure that every resident is respected, recognized and ensured representation by
being counted. We're, of course, not quite there yet, but we believe that we will achieve that
goal as well.
Now, just a word or two about the undercount. Two presidential administrations, five
Congresses, the scientific community, and the country itself have, over the last decade, focused
on curing the historical undercount that has been the characteristic of every American census
dating back to the first in 1790 conducted by Thomas Jefferson. As you know, and as I believe
will be confirmed by my two predecessors who are here with us today, the Census 2000
operational plan, building upon the work that really started back as early as 1988, and using the
experience of the 1990 census to enhance that plan, continuing through the work that was done
under the directorship of Martha Riche, that this plan is the most innovative, reformed and
modernized census ever, of course, as it should be.
Yet, as I have said on numerous occasions, the census itself is an estimate. That is, there
is a true count of the American population on April 1st 2000, and the processes that we've been
going through, that is the mail-out/mail-back and the Non-Response Follow-Up constitute an
estimate of that true number. And we know that estimate does not include everyone - which is to
say there remains an undercount. We cannot count our way out of the undercount. At the end,
there's a substantially large population in the country which is either resistant to participation or
fearful of being counted.
And let me remind us now as we move into this phase of the census, that roughly half of
the 1990 undercounted were children. A great many of those were simply left off of forms that
we successfully distributed and collected. Taken together, those who refused to be counted or
those who vigorously evade the enumerators, you begin to understand the difficulty of reaching a
100 percent count; that is, reaching the truth, which we don't expect to do. Again, a census is an
estimate. It has to be an estimate. And the question is whether we can get that estimate closer
and closer to what we think is the true, true number.
Even though we have greatly improved the total accuracy of the census over the centuries,
the last 1 or 2 percent become virtually impossible to count, even with the unprecedented
outreach efforts of the new millennium. The Census Bureau has historically been at the forefront
of modern science, helping to improve technology, which then improves the ultimate count. We
led the advance from hand tabulation to machine tabulation. The Census Bureau helped create
the first electronic computer, the UNIVAC, to further speed the tabulation of the 1950 census.
The Census Bureau itself perfected modern survey designs based upon large samples which give
the country reliable and consistent data on major economic activity, including the Consumer
Price Index, unemployment figures, housing data and much more. And today, of course, we're
using the computer-assisted personal interview techniques with portable laptops to speed the
interviewing and provide higher consistency between interviewers. And the other major
innovation, of course, technical innovation, in Census 2000 was our optical scanning system,
intelligent character recognition, which allowed us to process the data, as I just said. Now 117
million forms have been captured as Title 13 data with very, very high levels of accuracy.
In summary, the Census Bureau does bring together more than two centuries of
experience, and countless modern scientific methods, to try to reduce the undercount and to
correct for it where we find it.
Just one final comment on this. The country, especially the press, has discussed the
undercount in great detail - by racial subgroup and at various levels of geography. This
conversation could not take place but for the fact that the Census Bureau conducts the largest
survey of its kind as a quality check on our own work. We then issue a critical report card on our
work, which is the source of the data about the undercount. In the past, this survey was known as
the Post-Enumeration Survey, and for Census 2000, it's called the Accuracy and Coverage
Evaluation. These numbers - this report card on the census, is faithfully reported as facts in the
press, in congressional debates, in GAO studies and CBO reports, in scientific forum, and by
political leaders. It is clear to me that our report card has the confidence of the country with
respect to measuring the size of the undercount or overcount. It is, therefore, somewhat ironic
that there is some debate about whether the numbers are good enough to correct for these errors.
So, on the one hand, we have presented to the country the very best estimate we can for
that part of the population which is missed in the census, and that part of the population which is
doublecounted. And the country has accepted these numbers for purposes of discussion, debate,
argument and budgeting. And yet for reasons that we will have to discuss, not everyone in the
country accepts the use of these data to actually correct the errors that we ourselves tell the
country that exist in the initial count.
Now I want to move to just two announcements. First, for the past few decades, the
Census Bureau has been examining, as you know, the use of statistical methods to statistically
correct the census - that is to do something about the persistent undercount, which we have now
measured for more than half a century. In complete accord with the Supreme Court decision of
last January, and all of our statutory obligations this past, Monday, I forwarded to the Secretary
of Commerce the finding of the Census Bureau that it is both technically and operationally
feasible to use the Accuracy and Coverage Evaluation design to correct for errors in the basic
This document, which is in your press kits, does set forth, in some detail, the nature of
our decision that it is both technically and operationally feasible to use this Accuracy and
Coverage Evaluation and its related statistical procedures to correct for the undercount. It also
makes clear that we have not yet made a decision as to whether we will actually correct. We
have decided that it is technically and operationally feasible, but the actual decision itself will be
made after we've examined the results of the Accuracy and Coverage Evaluation and made
certain that it was sufficiently robust to justify its use.
As we say in the document itself, and I'll just read the very first sentence - "This
document sets forth the rationale for the Census Bureau's preliminary determination that (1) it is
feasible to produce statistically corrected data within the time frame required by law and (2) the
statistically corrected data will be more accurate."
So we have now, as I say, transmitted this to the Secretary. And on the basis of this
document, the Secretary is today publishing a draft regulation for public comment, and I will ask
Andy Pincus, who is the counsel to the Secretary of Commerce, to comment on that document.
MR. ANDY PINCUS: Thank you, Ken. Secretary Daley's highest priority with respect
to the census has been to make clear that there's absolutely no role for nonscientific criteria in
that process. The decision that Ken just mentioned, the decision whether to use statistical
sampling to correct the census data that will be sent to the states for use in legislative
redistricting, obviously will receive significant attention when it's made. And now, well in
advance of that decision, Secretary Daley believes it's important to put in place the procedures
that will enhance public confidence in that decision when it's ultimately made. So we are, today,
issuing a proposed regulation to establish that procedure. And a copy of the proposal is in your
The key elements of the proposal are to leave the decision to the Census Bureau and to
establish an open, transparent process with respect to that decision. A committee of Census
Bureau professionals will make a recommendation to the director regarding whether the numbers
should be corrected using statistical sampling. That recommendation will be placed on the
Bureau's Web site and published in the Federal Register, and the director will make the final
decision on whether to correct the numbers, which will not be subject to review or alteration or
reconsideration, in any way, by the Secretary of Commerce. We believe this ruling assures that
the process will be open, and that it will turn solely on the science and that the public will have
confidence that is the only matter influencing the decision. We will receive comments for 45
days and finalize the rule in the late summer or early fall, so that the procedures are clear, and in
place, well before the decisional process begins.
DR. PREWITT: And I do want to emphasize that this decision, both the feasibility
document that sets forth our rationale for proceeding does rest upon a quarter century of work.
It's indeed under the direction right now of John Thompson, our associate director of the
decennial census, who has been at the Bureau for a quarter century, and, the committee that Andy
just referenced, of course, is our most senior people, which, among them, are several hundred
years of experience in conducting a decennial census.
But, to exemplify the fact that we are building upon our past experience, I'd like to ask
both Marty Riche, Dr. Riche, who, of course, was my immediate predecessor, and Dr. Barbara
Bryant, who was the director of the census during the 1990 decennial, to join me and each say a
word or two.
This is Barbara Bryant. Go ahead, Barbara.
DR. BARBARA BRYANT: After the 1990 census was completed and the evaluation
was completed, we found at the Census Bureau - and by "we" I mean all of the professional staff
there - that there were four very apparent needed changes for the 2000 census, and we began
work in 1991 to move toward these. First, the questionnaire needed to be made more readable
and understandable for the respondents. And we started pretesting questionnaire changes in
Second, the Master Address File, that's the address list, the mailing list, to which the
questionnaires are mailed, could be improved if the Census Bureau and the U.S. Postal Service
were allowed to share mailing lists and to share them with local communities. Now, believe it or
not, there were laws that prevented it, privacy laws of both the Postal Service and the Census
Bureau, and we began working - fortunately, we had a mutual oversight subcommittee in
Congress at the time - we began mutually working to get those laws changed. And in 1994, it
was. And so building this year's Master Address file on the foundation of the computerized
Master Address File from 1990, while the Census Bureau and the Postal Service have exchanged
all the information they have been able to exchange it with local communities who could
compare their tax rolls and all. So, I do believe that 2000 started out with the best possible
mailing address list.
The third decision was that the Census Bureau had to move to paid advertising. We had
clearly gone one census too many depending on public service announcements which were being
given to us at 3:00 a.m. on Sunday morning.
And finally, whatever the quality and level and commitment of the effort, it is not
possible in this diverse society to count everybody. A large follow-up survey is needed to check
the accuracy of the actual count and to enhance it if necessary, and that this enhanced or
corrected count needs to be available before redistricting takes place. In 1991, it was not
available until two months after the redistricting deadlines.
Now, I'm a Republican. I support Republican candidates. But I have spent a 30-year
career in survey research, and I know that the interests of everyone, long-term, and of the
American economy are best served by accurate data. The inclusion in the census of persons that
are missed by the census will not change vote counts in the short-term. About half of those
missed, as Ken Prewitt has already told you, in both the '80 and '90 censuses were children, who
are not eligible to vote, of course. And the adults who are missed tend to be people who are
disconnected from the society. These are nonvoters, for the most part.
Now, any correction of the census will make incremental changes in district lines, but
redistricting, remember, is an activity that is carried out at the state level, and it is greatly
influenced by the party in power in both the governor's office and the legislature in each state.
And these influences are far greater in redistricting than any incremental additions of uncounted
persons can possibly be.
DR. PREWITT: Marty Riche.
DR. MARTY RICHE: Thanks Ken. It's great to be here with Ken and Barbara because
between the three of us, we represent probably the 12-year process of getting this census going.
Barbara was really the person in charge of the planning. I was an advisor to the Census Bureau at
the time. I've actually been following the planning for Census 2000 for almost 20 years now -
that's how long it takes. When I came in, I came in for the design phase, and when I left in 1998,
it was time to turn it over to Ken for the operational phase.
The one thing I did in addition to all the consultations that everyone had done on the
National Academy of Science panel, all of the advisory panels that the Census Bureau has had,
was to also go out and talk to the American people, because, of course, they experience the
Census. And I started with about 30 town meetings and did a number of other activities. And
it's really what they told me then that sort of coalesced the design that the National Academy of
Science panel, chartered by Congress, had come up with.
First and foremost, Americans asked us to build partnerships at every stage of the process.
They said, "You can't do everything alone and do it right." So the Census Bureau has reached
out to build partnerships all across the country with state, local and community groups. It's been
a big success. And I think the Bureau's delivered on the promise that I made four years ago to do
Census 2000, not for Americans, not to Americans, but with Americans.
The second thing Americans asked for was to keep it simple. So we built simplicity into
the process in every way we could. In addition to the paid successful advertising campaign and
outreach, we created multiple opportunities for people to respond, and we broke new design
ground with the form.
I heard people's concerns about intrusiveness, so we removed every question from the
short form that was not required by the Voting Rights Act and every question from the long form
that was not required by law or a government agency to run a program. We went as far as we
could legally go in simplifying the form. And judging by the comments I heard in the last few
months, it's probably a good thing we did.
Our third strategy was to use technology, intelligently. Americans knew that there were
dramatic advances in technology that would make the process simpler and more accurate. This
involved contracting with the private sector on an unprecedented scale. And that, too, has been a
The fourth and final strategy is the one we're really here to talk about today, - to make
better use of statistical methods. Americans are a very common-sensical people, and everyone
we heard from was well aware that some folks have more than one street address, others less than
one, so an address-based process is simply bound to fall short. And they understand that some
folks are just ornery.
I remember talking to a politician in Texas who said to me, "I've got three elderly aunts.
They've never answered the Census because they don't think it's any of the government's
business." He said, "Isn't there some plus or minus technique you can use to make things better?
We use them every place else." And, yes, there is, and the Census test has shown that the
process the Bureau is now embarked on works.
So the bottom line of all of these conversations and meetings was that Americans want a
clear, correct and comprehensive idea of who we are and who we're becoming, because the value
of the Census, as they see it, is not just the snapshot we use to keep our democracy truly
representative, but also that in-depth community portrait that they depend on for making effective
More Americans than ever before will use the data about their communities the Census
Bureau will be delivering next year. I think they're going to be very happy with the innovations
we made for Census 2000.
DR. PREWITT: Could you just stay with me for a second, Barbara and Marty? I just
want to add one final comment to this. I also talked late last evening with Vince Barabba, who,
of course, was the director of the Census in 1980. And I mentioned to Vince the announcement
that Andy Pincus will be making this morning on behalf of the Secretary. And I think I quote
him correctly when I say that he said, "It's about time." Mr. Barabba feels very strongly that the
decision as to whether there should or should not be corrected data used is a decision that
belongs to the Census Bureau and not to the Commerce Department. And I dare say that you will
not find any Census Bureau director, of course, who does not agree to that principle.
It is particularly important, coming from Vince Barabba, because Mr. Barabba, who, as
you know, was the Census director twice, one prior - he had been appointed earlier, a Nixon
appointee, and then he came back under President Carter for the 1980 Census. Vince has
hesitancies about whether we should be statistically adjusting in 2000, for complicated reasons.
And he's an experienced, intelligent man. But that's, for him, not the issue. The issue is where
the decision should be made. And he believes it is a decision that should be made at the Census
Bureau level. And they're nodding to agree with our predecessor, Vince.
DR. RICHE: Yes, and I've heard that from Vince, before, myself.
DR. PREWITT: Good. Thanks very much. Thanks.
So we're now open to your questions; that is to say we, which is to say if they're very difficult,
I'll be able to defer now to lots of help.
MR. JOST: Okay, we'll go in the room here and we'll start with Genaro.
Q Genaro Armas, Associated Press. Dr. Prewitt, the Census director's office is a
political appointee. How does that take away nonscientific criteria from the decision-making
DR. PREWITT: Surely. Well, you're right. The director of the Census Bureau is a
presidential appointee, of course. So is the director of the National Science Foundation, the
director of the National Institutes of Health, the commissioner of the Bureau of Labor Statistics.
I mean, this city is littered with people who are put there because they have the technical
expertise to manage the programs of those agencies.
I believe that's why I was put here. When I was asked in my confirmation whether I
would be capable of making an independent judgment - that is, one not influenced by political
considerations - I said, "Absolutely". And if I, thought that I could not make that decision, I
would resign, and resign publicly. I said that in my conversations with the Secretary and said
that in my conversations leading up to the hearing.
I feel very, very strongly that the Census Bureau is a statistical, scientific agency and that
it is doing what it is doing to try to make a better count. If it did not believe it was going to make
a better count, it would not be doing it. And the fact that I happen to be a presidential appointee
is beside that fact, just like the fact that the head of the National Science Foundation is a
I think if you go ask Dr. Caldwell whether she makes decisions out there about our
national science policies and the degree to which they're independent, she will make it very, very
clear to you that the fact that she's a presidential appointee has nothing to do with the way in
which they conduct their scientific agenda; same with NIH, same with many, many other
agencies in this society.
MR. JOST: Okay, do we have any callers on the phone? We'll remind the reporters on
the phone that if you want to get in the queue, to press 1. And also the handout materials here
today are available at this moment on our Web site. So we'll go back to the room. This is
unprecedented. We feel like we've done a very good job at this moment, folks.
From Dallas, we'll go to Rick Klein with the Dallas Morning News.
Q Thank you. I was wondering if you could explain some of the impact here. I'm
wondering what this means. I haven't seen a copy of the orders that have been promulgated yet.
I'll obtain a copy of it today. But does this change at all what numbers they'll be receiving next
year? Will they still get two sets of numbers?
DR. PREWITT: Yeah, Rick, it's as follows. The budget law - I think it's Public Law 95
- I'm not certain - of 1998, when it was passed, required the Census Bureau to make available
simultaneously data that would not be corrected if we decided to make the corrected data
available. And we have obviously made provisions for that. We've written to all 50 states now
saying that "As we release the data, which are denominated for redistricting, do you also wish to
have the data which, in this case, could be uncorrected? We're not certain yet which the data
will be." And they then respond. If they want it, they send it out. So, yes, we have the
Now, I should say that the reg, as has been put out for public comment by the Department
of Commerce, has a quite critical argument to be made here. The process that is set forth in the
regulation has a 13-member committee of professional statisticians, demographers, survey
researchers; at the Census Bureau deliberating about the A.C.E. results and then making a
recommendation to the director. The director then will make the final decision, according to the
terms of the reg. As I say, it will be out for public comment, but we presume that after the public
comment, we hope that principle still holds.
The reg goes on to point out that if the director should reject the recommendation of that
committee, irrespective of what that recommendation should be, should reject it, but especially if
the committee recommends the corrected numbers and the director rejects that recommendation,
then the Census Bureau would nevertheless release the now-corrected data. So under almost any
circumstances, both sets of data will be released.
Just another word or two on that. The Census Bureau is a scientific organization, and we
routinely make as much of our data available as possible for different kinds of purposes. As I
said, the census is an estimate of the truth. And we have now finished, as I just said, are close to
finish, Non-Response Follow-Up. If we stopped the Census today, or after we finished our 1
million more households, if we stopped the Census at that moment, that would be an estimate of
We have another big field operation to do over the next month or so. That's a coverage
improvement field operation. We are doing that field operation because we think that will move
our estimate closer to the truth. It will improve the number. After that, we have another field
operation. It happens to be called Accuracy and Coverage Evaluation. We will do that one,
because we think that moves the estimate yet closer to the truth.
Every one of our major operations are an attempt to get closer to what we consider to be
the most accurate number that we can give to this country.
MR. JOST: One more chance in the room. We'll go back to the phones. How about
Herb Sample of the Sacramento Bee.
Q Can you hear me?
MR. JOST: Go ahead, Herb.
Q Hello. This is, I think, a two-part question. Will the decision about whether to use
adjusted figures be made before a new administration takes office? And, if not, can that new
administration do anything, by executive order or otherwise, to change this proposed regulation?
DR. PREWITT: Herb, I'll answer the first half of that question. And with respect to the
second half, that's a question we'll have to put to someone who knows more about
administrative law and policies and regs than I know. I don't know if Andy will want to
comment on that or not.
To the first part of your question, the answer is no. Our schedule and procedures will not
allow us to make a decision about whether to correct or not correct prior to January 20th. The
time frame for conducting these operations, continuing to do our work, is simply such that the
decision will be made, our best guess is, right now, toward the end of February.
And with respect to the second half of your question, I'll let Mr. Pincus address it.
MR. PINCUS: This regulation embodies what we think is the appropriate process.
Obviously, regulations can be changed. They can be withdrawn if the proper administrative
procedures are followed, and the law makes provision for that. And if there's a desire to change
the process and those procedures and standards are complied with, then it can be changed. But
what we want to do is put in place the process that we think is the right one.
Q Can I follow up?
MR. JOST: Sure.
Q Is the position of Census director at the pleasure of the president, or is there a fixed
MR. PINCUS: I think it's at the pleasure of the president.
DR. PREWITT: If I could just add a word or two on that, if you all are interested in these
kinds of things. I believe that I'm the only head of a major statistical agency in the world, at least
in the industrial countries, democratic countries, which serves at the pleasure of the chief
political officer. That is, all other heads of major statistical agencies are by term. And indeed,
even within the United States, most of the heads of statistical agencies and other scientific
programs are there by term; National Science Foundation, NIH, and as I said, the commissioner
of Bureau of Labor Statistics. There may be some; I'm not exactly informed on everyone's
nature of appointment. But oddly, the director of the Census Bureau since 1902 has served at the
pleasure of the president.
Q Thank you.
MR. JOST: Thank you, Herb. Staying on the West Coast, up bright and early, is Julie
Sullivan with the Portland Oregonian.
Q I just wondered if you could tell me what the cost of this 1 million households is
going to be, and also if you could talk about the death in the Midwest, this weekend, of a Census
DR. PREWITT: Yes. With respect to the first part of your question, Julie, I don't have a
number to give you today. I can tell you that the unit cost of the last few cases is really always
very, very high. These are the cases we've gone back to three, four, five times. We've sent out a
crew leader. We sometimes send out more than one enumerator to try to convert people, to try to
sort of camp out there, to try to catch them when they're at home.
It's simply the unit cost of the last cases is always just extraordinarily high. And that's
why our overall average seems so high. As we've said so many times, that if everyone simply
mailed it back in, this would be a much less expensive census. But the actual detailed work that
we have to do to estimate the cost of cases that we get at different stages in the census process,
we have not done that yet. So I cannot, unfortunately, give you a number.
You asked secondly, then, about - I'm sorry. Go ahead.
Q About the fatality.
DR. PREWITT: Yes, the tragic death in southern Indiana. This really is a tragic death
and really very unpleasant to even contemplate. It is under active investigation now by local
authorities, and also, I believe, by national authorities. And I have, as yet, no full report on it.
We've obviously expressed our concerns and condolences to the family. We also have sent out
the strongest signal we possibly can that the first concern we have is the safety of our
We have put census-takers in difficult situations, of course, because it's especially now
down to the last cases. And so we don't yet know the facts on this, but it is a tragedy. And there
have obviously been other deaths, but primarily related to traffic fatalities, of census-takers. But
this is one which really brings us very special kind of pains and concerns. And as soon as we
know something about the nature of the attack and to what extent it - well, as soon as we know
more about it, we'll report that.
MR. JOST: Anyone in the room? We'll go back to the phones. Soraya Nelson, still out
there on the West Coast; the Los Angeles Times.
Q Yes. Good morning, Dr. Prewitt.
DR. PREWITT: Good morning.
Q I have a couple of questions here. One concerns the data in terms of the properties
and 99s as they're called. In other words, where you were not able to go and actually talk to the
people who live in a singular household. Do you have a breakdown of the process? How many
of those responses were, in fact, either incomplete or proxies, you know, where you talked to
DR. PREWITT: No, we tabulate that as we go. That is, we keep a very precise record of
how we got each response that's in the Census itself. But we have not yet done the analysis of
that. What a census does is it does the analysis it needs to be doing in order to make the next
operational - take the next operational step. That is, in the middle of a census, the last thing you
have time to do is just additional, interesting though it may be, work. We will be reporting that
proxy rate toward the end of the summer, early fall, we suspect.
I should say that it's the Census Bureau's very strong position - and I suspect that will be
confirmed by Director Bryant as well - that in the last hard stages of a census, we do run into
people who simply will not cooperate. And we indeed, as we reported many times, we have
received checks from people who simply say, "Here's my $100 if you want to fine me. I'm not
going to fill out this thing no matter what you say."
It does make sense, under those circumstances, to talk to people who are knowledgeable
about that household. I gave an example of one the other day in a town in the south, where I
myself was doing some enumerating. And I went, at one point, out with a crew leader, and this
was the sixth visit to that household. And the person had simply answered the door on three of
those occasions and said, "I'm not going to cooperate." And we simply went with other people,
tried, you know, to convert that respondent, and so forth.
It turns out that one of the workers in our office, who works on our administrative work
or payroll work, happens to live right around the corner from that household, knows the
household very well, knows the name of the person, knows her two children, knows her rough
approximate age, knows obviously her gender and her racial and ethnic categories, and knows
that she owns a home; that is, knows everything that is on the short form about that household.
We believe we're much better off getting that information from an informed neighbor
than leaving the household blank. So at this stage in the Census, these last difficult cases, we
simply do go to somebody who can give us better data than no data. It's not more complicated
than that. The actual number of those will be determined not by the Census Bureau, but be
determined by the number of people who are never home or refuse to open the door or opened
the door and said, "I'm not going to cooperate." So the number is a function of the American
public's willingness to cooperate, not a function of a decision made by the Census Bureau. The
Census Bureau's decision is to try to get information from every household on our address file.
Q I have one quick follow-up concerning [inaudible] - by the 10th of June. Can you
speak to us to what the ramifications of that might be, and also what the amounts were that were
paid or who won?
DR. PREWITT: I can certainly speak to that with some force. That is an incorrect
rumor. We have no idea where it got started. We're extremely unhappy with the fact that it has
now been repeated by people who ought to know better. There is no bonus system anywhere in
the census operation. There is no incentive for finishing early, other than the fact that data are
better, the sooner you collect them. That's the incentive. So the answer to your question is that
there is no such procedure in Census 2000 to have any kind of bonus system.
Q Okay, thanks.
DR. PREWITT: Surely.
MR. JOST: We'll go to Tom Ginsberg with the Philadelphia Inquirer.
Q Hi. Back to the tabulating of the correct data. A handful of states have begun
considering bills that would mandate the use of factual data, not corrected data. Do you expect a
situation where, once you do put out corrected data, if you do, that you will have some states
using it through their own political process and others won't, and you may have a situation where
you may have to intervene or come in to lobby on behalf of the corrected data?
DR. PREWITT: Well, the Census Bureau will, of course, not lobby anyone in terms of
how the data are to be used. The end uses of the data are decisions to be made by the political
process, or the legal process, as it may be. The Census Bureau is only concerned about how it
produces the best possible data. And the uses of those data are really up to the user community;
in this case, up to the states that will be using the data for restricting.
So we will simply provide the data. That's our job. That's what a statistical, scientific
agency does. And the uses of the information, whether by individual states, by federal agencies,
by the private sector, become their decision. And so that's all we can say to that issue. We'll put
the data out and then the process will use the data as it wishes.
Q Can I follow up?
DR. PREWITT: Certainly.
Q Doesn't that lead to a possibly inaccurate or unfair situation, even, where you have
one state using correct data, as you determine it, and a neighboring state possibly using incorrect
DR. PREWITT: Just -
Q Some right across the border from each other?
DR. PREWITT: Well, just to back up for a moment. The initial data set that is released
- that is, what's called the apportionment data count released by December 31st - those data do
not include the Accuracy and Coverage Evaluation operation. So in that sense, every state will
be treated similarly with respect to the apportionment count.
By the way, for those who don't know it, that count includes the overseas military and the
overseas diplomatic corps and their dependents. When we then take those data and give them to
the states for other purposes, even if they were uncorrected, we would back out the overseas
American diplomatic employees, military and their dependents. That is, we cannot assign those
particular cases back down to the block level, whereas the redistricting data requires block-level
So there's always been a process where one set of data have been given initially as the
apportionment count, then another set of data have been released for redistricting. In the case
that you now describe, we will provide to each of the 50 states a redistricting data tape that has
the information calculated down to the block level for the purposes of administering civil rights
legislation, Voting Rights Act, among other things. And if one state uses it and another state
doesn't, then the inequities are visited upon the people within that state, not across state
I would think that the state that chooses not to use the corrected data will be making a
decision, in the Census Bureau's judgment, to be using less accurate data than they might have
used; that is, data that we will know will have undercounted certain population groups at certain
rates. We will be reporting that to the states. They will then make up their own mind.
Sorry. It's sort of garbled, but it's a very complicated process. Is that sufficiently
accurate for what you need or sufficiently complete?
Q Yes, I guess so. Thank you.
MR. JOST: Okay, we'll go up I-95 here. Mike Chang of Newsday.
Q Correct me if I'm wrong. But it's my understanding that [inaudible]. So
approximately what percentage of those non-response households are actually counted?
DR. PREWITT: Yes, and that's very important. The Census starts with its control file,
and its control file is its address file. And when we say that we're now 98 percent finished of
Non-Response Follow-Up, that means we have now gone to, and gotten some kind of report back
from, 98 percent of those households. That could either be a completed form or, in some
instances, it could be a proxy form. And in some instances, it's simply an identification of that
household as being vacant.
- And we go back out into the field, which is what we now do in July and August - we
will go back out into the field. In fact, we'll even start that sooner in many instances, because the
Census is so far ahead of schedule right now. We will go back out into the field and
double-check to make certain that a housing unit which enumerators said was vacant was, in fact,
And this is what I said a minute ago. If we finished the Census as of Non-Response
Follow-Up - that is, the next two weeks - and stopped it there, in our judgment, that is an
estimate that's less accurate than the estimate that will be produced after we go back out and
make certain that there are none of those vacant housing units which were incorrectly classified.
That particular operation also includes going into what we call new construction; that is, housing
units which were added. I think about 360,000 housing units were added after we finished our
So you're absolutely correct. What that is - the 98 percent is the 98 percent today of all
of the housing units on our address file. It is not a 98 percent count of the American population.
We simply don't yet know, when we got that form, if everyone that was in that household
responded on the form.
That's the coverage problem. Going to every household does not mean you count every
person who lives in every one of those households. And that's why we do the Accuracy and
Coverage Evaluation. It goes back to make certain that the people who are in those households
were completely and correctly enumerated, and it goes back to make certain that we did not
overlook any housing units. It also has that property that we can find out if we did miss housing
units, and if so, what kinds of neighborhoods, what kinds of people live in those housing units.
And we do that also in the Accuracy and Coverage Evaluation process.
MR. JOST: That new construction number was 630,000.
DR. PREWITT: Sorry. Thank you.
MR. JOST: And last question, Sherry Sylvester, San Antonio Express.
Q Hi. Good morning, Dr. Prewitt. I'm just wondering about the timing of this
announcement. Was the spin too good? I mean, last week you said you had no bad news, that
this was a good census. And this week we get the caveat that Secretary Daley has given you the
power to decide whether or not to adjust the figures. So my first question is, has this
announcement been on the schedule for a while? And second is, don't we anticipate that
decision is going to be made by the courts?
DR. PREWITT: Well, to the first part of your question, the timing of this particular
announcement, as a matter of fact, is largely a result, a consequence, of when we got this work
done. For us, the most - that is, for the Census Bureau, the important part of this task has been
working through whether we believe it is technically and operationally feasible to conduct the
ACE. If we thought it was not technically and operationally feasible, we would not be doing it.
And we have spent a lot of time over the last - really almost since I've been here, almost
18 months now - working on this issue. We happen to have finished that work. And because the
Census Bureau feels very strongly about transparency, when that work was finished, we thought
we should share it publicly. So we have shared it with the National Academy of Sciences, with
all of our committees in Congress, with other interested parties, with the Census Monitoring
Board. We simply have done our work; and, therefore, we are sharing it.
But also, I think the timing is related to the fact that there is going to be heightened
attention on the Accuracy and Coverage Evaluation survey over the next several months, as we
actually do that work. And so it seemed to us that this was an appropriate time to set forth the
logic behind that piece of work, why we are committed to it and believe it will improve the
So the timing was partly just when you get that kind of stuff done and when you can
make it available to the public. We tend not to sit on things as soon as they are available to be
shared. The Secretary could not make a decision about delegating something which we weren't
certain we were going to do. So his decision had to be triggered by the fact that we're where we
are in the census.
With respect to the fact that we reported last week that it was a good census, we believe it
is a good census. We believe that the Accuracy and Coverage Evaluation process will make it a
better census. We're moving along. We're feeling very good. We're ahead of schedule. As I've
said many times before at these operational briefings, every one of our operations, every one of
our major operations, was completed on schedule and at, or under, budget.
We are still getting those last 1 million cases. They're going to be difficult. They're
going to be expensive. But we do not leave the field until we go to those 1 million more
households - actually, it's 908,000 more households - 908,274, I believe is the number. And
every day we look at that number, because we want to get every one of them included so we can
move into yet our other operations which, we think, will improve the census.
Sorry, that was a long answer, and I'm not sure I got to all of what you asked. Did I?
Q I wanted to know about the courts -
DR. PREWITT: Oh, I'm sorry. Yes.
Q - the next step. And I just want to know, since the Secretary has delegated this to
you, what your role will be there.
DR. PREWITT: I suspect my role will be being deposed - (laughs) - for some large
number of hours in my future life. But, no, the Census Bureau, of course, is not a legal entity.
It's a statistical agency. It doesn't have any legal strategy. It simply has a statistical scientific
strategy, and that's what it's doing. I'm not unmindful of the fact that there were court cases in
1980. There were court cases in 1990. There have been court cases leading up to 2000, and I
suspect there will be additional court cases. I think they will come from both sides. There will
be litigation saying that we must use the corrected number and litigation saying that we should
not use it. And the courts will come to their own deliberations.
I'll just conclude by making this point again. It's somewhat redundant. But a scientific
agency, which the Census Bureau is, has to do its science the very best that it can. Then the uses
of that science, especially when they're in the middle of a political process - and there is no
doubt about it - but redistricting and reapportionment and even federal funding are political
decisions. All the science can do is say, "Look, this is what we have discovered. This is what we
know. We have used the best science that we know how to use to provide you this answer."
And then the political process or the legal process will decide whether to use it or not.
That is no different, by the way, than lots and lots of other relationships of scientific
projects to the public policy process. For example, when the country is debating a test ban treaty,
it turns to the scientists to ask, "To what extent can you detect underground detonations in North
Korea?," for example. And the science gives an answer to that question. And then the political
process takes that answer and uses it as it deliberates about whether to sign the test ban treaty or
The census is no different. The census produces science the same way that the biologists
produce science, and the chemists produce science and the physicists produce science. We
produce a scientific product. And as I say, the political process then uses the results of that in
making its own determination. So, our job will be done when we produce the best numbers we
know how to produce.
MR. JOST: Okay, no one in the room? Thank you very much. As we said, every one of
the handouts here is available for downloading from our Web site, and that's
<http://www.census.gov>. And you look for the public information page. And we'll let you
know when we schedule our next operational briefing.
[END OF PRESS BRIEFING.]