[Joined in progress.]
MR. STEVE JOST: We are here to update you on recent developments in the Census
Bureau operations and, if you're new to the process, we want to just give you a quick rundown on
how things work. The director will have a few opening remarks; we'll take questions from the
reporters. We'll alternate between the room and reporters who are online with us on the
telephone. We ask that when you are asked for a question that you identify yourself and your
affiliation, and we have folks with microphones so we can pick up your questions on the audio
for the radio and TV folks.
With that, I give you Director Kenneth Prewitt.
CENSUS DIRECTOR KENNETH PREWITT: Good morning. Thank you.
The Census is well under way. All systems are on go. Our current mail response rate, as
you know, stands at 55 percent, and as you also will recall our operational and budget planning
was based upon a response rate of 61 percent. That was a drop from our earlier experiences with
the census. In 1970, 78 percent returned their forms; it dropped to 75 percent in 1980; and then
to 65 percent in 1990. And, of course, as you know, our goal is to boost the 1990 number by at
least 5 percentage points. That was the basis of our '90 Plus Five challenge, to boost the response
rate, that is to reverse the decline in civic engagement in our society, and to save taxpayers a
significant amount of money by reducing the number of households we would have to contact to
obtain the census information.
There are some hopeful signs: 748 localities have already reached their '90 Plus Five
targets, that is, they're 5 percent above their 1990 base. We also, at the state level, have some
very strong states: Ohio, as you see from this map, is at 62 percent, and Nebraska and Iowa, at
61, Michigan and Pennsylvania, at 60. While these are heartening figures, we still have a lot of
work to do. You'll note that Alaska is currently reporting 42 percent, and Puerto Rico, at about
33 percent. Those of you who have access to, or still have the initial mail response rate for 1990
that we distributed many times before and it's available on our Web site, will recognize that
there's a southern factor. That is, the states that lag the general national response rate are heavily
clustered in the South and Southwest; and of course, that's the pattern you see repeating itself on
this chart today. We suggest that the very same sort of demographic and attitudinal
characteristics in the American population that led to lower than average response rates in 1990,
and indeed also in 1980 -- the same pattern -- that those persist through 2000.
We continue to work very, very hard at getting the response rate to the target that we
would like to have for the American people; and beginning tomorrow and lasting through Sunday
we will have some new radio advertising going on the air. They (the ads) start with the words,
"This is an urgent message, your state" - listing it by name - "needs everyone who has received a
census form to fill it out and mail it back now, today." And that theme, which then concludes
with "This is your future, don't leave it blank," our standard tagline, will be targeted in 11 states
and in the District. Those states were selected because they had both low response rates now or
had low response rates in 1990, or both. Florida, Georgia, North and South Carolina, Alabama,
Mississippi, Louisiana, Texas, Oklahoma, New Mexico, West Virginia and, as I said, the
District.
Spots will air elsewhere as needed, and television ads, including those in Spanish, are
now being edited with new endings to stress "mail it back."
A few other operational things. As of now, our four data capture centers are humming
along day and night, and we are hoping that millions of Americans will put their forms in the
mail, over the weekend, have done so over last weekend, Census Day, and, of course, will
continue to do so. We have scanned in more than 25-1/2 million forms very successfully. I'm
happy to report that our optical character recognition technology continues to function at more
than 99 percent accuracy.
By April 2, more than 58,000 people had taken advantage of being able to fill out their
short form over the Internet. As of yesterday, we received more than 5 million calls on our
toll-free number, and about 13 percent of those asking for Spanish-language assistance. And our
27,000 odd questionnaire assistance centers report a steady stream of people coming to get help
in filling out the forms. If people still need a form, they can call our telephone questionnaire
assistance number, 1-800-471-9424, or they can go to one of our more than 15,000 "Be Counted"
sites that are located in areas with population groups that have historically been undercounted.
These locations opened last Friday, and will remain open through April 12.
Also, operationally, last week we completed our operations to count people with no usual
residence, going to shelters, food services and outdoor locations. I received a report this morning
that we have not experienced or are not having any major operational problems in any of our
many other operations, which includes update/leave, remote Alaska enumeration, service-based
enumeration, military enumeration, maritime enumeration, list/enumerate, update/enumerate and
group quarters. I do remind you that although we have a core census that does depend upon the
mail-out/mail-back, we have many, many special populations, and we do, of course, give them
special treatment. And these are the kind of operations that have to get to those special
populations, and all of them are functioning as designed. We're also counting people with
another special operation with mobile lifestyles, as we say, living in marinas, carnivals, and the
like. And, of course, we started our group quarters effort, college dormitories, nursing homes,
and prisons, migrant workers, people on military bases and U.S. flagships at sea.
Even as we draw to a close this first phase of the census, our outreach program continues
and, our road tour vans are continuing to generate public awareness. We've made over 1,500
visits, and have drawn about 1.5 million people to those activities. Our news outreach, the radio
and television efforts, over and above our paid advertising campaign, has now reached nearly 99
million people, many in hard-to-count areas, and areas with low response rates. Our next milestone comes next week, that is after, April 11 and 12, we have to start drawing up the list of
those households who did not mail back their form. We move into our non-response follow-up
phase. There we will be mobilizing nearly a half million workers to call and go door-to-door to
complete the census. With regard to recruiting this massive number, we have already reached
our goal. In fact, we are now at 104 percent of our national goal, which was to have 2.4 million
qualified applicants by April 19th. As I say, we exceeded that goal well ahead of schedule.
That doesn't mean that we're completely out of the woods. There are about a dozen of our
520 local offices that have some recruiting problems. And, of course, we're putting special
measures in place to solve these. We're emphasizing, for example, recruitment in New England
as well as in Florida and North Carolina. And we will continue recruitment through the
non-response follow-up period to ensure a fresh labor pool. I've said on numerous occasions, and
repeat today, that the success of the census is in the hands of the American people. I mean this in
terms both of its overall success as a census, as an operation, and also in terms of its final cost.
And let me return to something that I have mentioned before. For each percentage point in the
response rate, we have to visit 1.2 million fewer households. And since it costs the taxpayer
about $3 when a household mails back in its form, but up to $35 when enumerators have to make
repeated visits to a single household, we save an enormous amount of taxpayer dollars to get the
forms back in the mail.
Therefore, we are at a key moment in terms of our mail-out operation. The programs are
in place and working well. Initial mail-back response rates are strong but not as strong as we
would like them to be. We will know by the middle of next week whether the American people
have decided to reverse the long tide of civic disengagement, and we will also know the size of
the remaining job we face to obtain a complete and accurate count. My basic message today, as
it has been for the Census Bureau throughout, this is your future, don't leave it blank. It is not too
late to mail back your form.
So, with that, I will take your questions.
MR. JOST: Right over here, if you can identify yourself. And while we're waiting for
that, if you're on the phone you have to punch "1" to get in the queue for a question.
QUESTION: Sir, my name is Lynn Sweet from the Chicago Sun Times. Can you tell us
what impact, if any, on the response rate did the comments of Senate Majority Leader Trent Lott
and Governor Bush have in urging people not to respond fully to the census? Give us your
evaluation.
MR. PREWITT: The Census Bureau has going on right now some unexpected research
work to try to anticipate or to determine that. There are two issues to the response rate that we
have to be concerned with. First, is whether the long forms are coming in at a lower rate than we
would have otherwise anticipated or expected. And, second, is whether we have what we call
item non-response. That is, as you know, some of the comments have suggested that the long form should be sent in, but only filling out, as we say the 100 percent numbers, which are the
short form numbers, or the short form questions. That's the 100 percent items, and that would
create, of course, a huge problem of item non-response. That is, we would not have good
long-form data if we had a large number of long forms which only answered the 100 percent
items.
I cannot, unfortunately, today tell you, because we did not fully expect this massive of an
expression of concern about the long form, so we did not have in place ongoing research
operations that would allow us to do this. We actually want to do the operations of the census
right now, not try to be doing that kind of work. So we will not be doing item non-response
studies sooner, probably, than the next month or so. I hope by the end of this month we'll have
some idea of that.
QUESTION: Can you determine if the broadside will hurt you, but you just can't do it
yet?
MR. PREWITT: Well, we can't on item non-response. On the response of the long form
itself, whether the rate of return of the long form is lower, I will be able to say something - I can't
today, and I'll explain why. I don't want to be evasive about this. We are expecting a large April
1st effect. We know from historical experience that the long form tends to be a form which
people set aside and fill out on weekends. You get the short form, it takes you five minutes, you
send it back in. You see the long form, you think it's going to take you longer, you put it in the
stack of things to do over the weekend.
And two things happen in this census. One, we have both a weekend effect and an April
1st effect coming at the same time, since April 1st was a Saturday. Those forms will be coming
in to our data capture centers today and tomorrow. I am hopeful by tomorrow or even by
Thursday, when I have congressional testimony, that I will be able to give some sort of
preliminary assessment about whether the long-form return rate is lower than what we
anticipated.
QUESTION: Have you done any remedial steps in light of the comments of Bush and
Lott? Have you done anything remedial without even waiting to do an assessment?
MR. PREWITT: Well, our remedial effort, of course, as I just said, is to do special
advertising in areas where we're having response rate problems. We have, of course, issued press
releases. I, myself, have been in the press many times over the last three or four days explaining
the importance of long-form questions, explaining that all long-form questions are there by law,
that they were all reviewed by the United States Congress, that there's not a single question that
doesn't drive some important program, not a single question. So we have done everything we can
to get that message back out to the American people.
More than that, we cannot do. We can simply re-explain time and time again that the Census Bureau is a statistical agency that carries out the laws of the land. The laws of the land
said that we shall ask questions on the following topics. We presented those questions to the
U.S. Congress two years ago. They were all authorized. They were all paid for. We obviously
have a budget to do the long form. And, therefore, we're simply doing the operations as we've
been directed to by law.
Our only remedial action at this stage is to make certain that if we have unusually low
responses to the long form we have in our non-response follow-up effort an additional task,
because it takes longer when you're knocking on the door to get long-form questions than short-form questions.
So, we would try to -- and that's why I say we're working on this right now. We'll
identify those areas where we have a lower-than-expected long-form return rate, and we'll simply
have to beef up our staff.
QUESTION: Andy Sullivan with the Durham Herald Sun in Durham, North Carolina.
You mentioned earlier about attitudinal reasons for the South having a lower return rate than
other regions. Can you tell me what some of those attitudinal reasons are?
MR. PREWITT: Well, in our own best studies that we can do on return-rate issues, you
have two sets of issues. You have demographic issues, which are people who are hard to locate,
hard to get a form to and, of course, the more rural areas are some of the areas which are hardest
to get a form to. But you also have that problem, of course, in certain dense cities where you
have multiple families living in the same household where you may have one address. That
family sends it back in, but there are other families living there. We have to then figure out a
way to find them, get them the form and so forth. So, it's not totally a rural/city breakdown. But
there certainly are demographic characteristics which contribute to low response rates.
Then there's also, as we would put it, attitudinal characteristics. That is, these would be
mistrust of government, concerns about government intrusiveness, an attitude that the federal
government itself is too large, not a mistrust of government, but a mistrust of the federal
government and presuming that the federal government programs should shrink. So those
attitudes -- they're not uniformly distributed across the country -- are all I'm suggesting.
You also have, of course, in the South and Southwest, higher concentrations, certainly did
in 1990, higher concentrations of recent immigrant population groups who don't understand the
census. They've come from countries where a census is done in a different way or not done at all,
or is done in a way that's been misused against them. And you have concerns with, of course,
documentation, whether the census data could ever be used in such a way as to get you in trouble
with the INS. So, you have population groups which are concentrated in the South and
Southwest which have serious and understandable, I should say, ill-founded, but nevertheless
understandable concerns about whether filling out the census form could somehow be used in a
way that could harm them. We have repeatedly said, of course, the data are confidential. The data are confidential. And we also can only get the message out through our advertising
campaign and partnership program the best we can. So, that's some of the attitudinal issues that I
would bring to your attention.
QUESTION: Jim Myers of the Tulsa World. Congressman Coburn from Oklahoma was
one of those that attacked your agency, in addition to urging his constituents not to fill out the
long form. He also said when the census takers come to the door to politely refuse to give them
any additional information. Can your agency do anything about that and are you?
MR. PREWITT: Well, as I said, one of the issues with a lower-than-expected response
rate to the long form means that we will have to send more people back to those households from
which we didn't get the form. If the same households which did not send it in now say, I won't
answer your questions, then we do train our people to try to get that information. The Census
Bureau, of course, has a large number of very professional interviewers. And they know how to
get information, they know how to be persuasive, they know how to urge, they know how to
explain the importance of questions and so forth.
However, the census is not using that labor pool. That's the labor pool we use for our
CPS work, for our economic statistical work and so forth. The census, because of its magnitude,
of course, has to hire a very, very large number of voluntary workers -- excuse me, of part-time
temporary workers is the word I wanted to use, temporary workers, they're not voluntary, we're
certainly paying them. And temporary workers who have been trained, but it's very hard in two
or three days of training to get someone up to the level of professional competence to know how
to, as we say, convert, to try to get refusals to go ahead and give you the information.
So, I have to express some concern that if we are using this labor pool for that purpose,
that is, having to convert a higher than expected number of refusals, it will have some damaging
implications for the quality of the data. That's about all we can do. We can get the advertisement
out there, we can explain the message, we can train our people to try to anticipate those issues. I
can say to any member of Congress, or any other political leader, that these are all data that you
have asked us to collect, that they are used for program purposes. And they are the basis which
brings back services and benefits to the local community. We can repeat that message and hope
it gets through. And we will certainly try to do some of that as we move into the non-response
follow-up period.
QUESTION: Lorna Virgili of Univision. One of the questions we've been encountering
among the Hispanic community is basically the fact that they have requested the form in Spanish,
and either it has taken too long, or they have not received it to this point, to this date. Why are
the forms in foreign languages taking so long for the people to actually receive it?
MR. PREWITT: I do know that as of now, there is no backlog. There was a backlog for
a period of time, but as of last week, as of last Thursday, there was no backlog. That doesn't
mean that they get the form the same day they sent the letter. There has to be a process that goes on. There was a short-term backlog. Over 2-1/2 million language forms have been requested, the large majority of those, of course, in Spanish. All of those that have been requested, unless
we just got it yesterday, or the day before, have now been sent a form. So I would be very
interested if people are still saying to you that they -- if they mailed in their request as much as a
week ago and don't have a form, I would be surprised, though I'd like to know that fact. If we
could identify that, we will try to correct that particular problem.
The reason for some delay, of course, if you think about it, is that a letter comes in, that
letter has to go to our central headquarters where we have our master address file, and it has to be
coded against our master address file, and then a label has to be produced to get on the form and
mailed back out. That is, there are a series of steps to make sure that that language form goes
back to the correct household so that we don't have any mistakes in the census. So that's the
reason for the time lag, and that's why it took as long as it did.
There was no way to shorten that. That is, there are a series of steps that had to take place
to make certain. In our own reports, we certainly had anecdotal evidence, as well, that some
people were concerned about not getting their language form. Our own current reports are those
have dropped off completely. Now, you may have some independent information. I'd like to
have it. But we do not believe right now that there are people who have ordered the language
form as recently as even three for four or five days ago who do not now have it.
QUESTION: Sean McLaughlin, with the New York Times regional papers. You had,
yesterday, a group at the Southeastern Legal Foundation calling for the agency to be abolished
and replaced by two different agencies. One dealing with strictly enumeration, the other dealing
with these questions about income to be used for the federal programs. What do you think about
that idea?
MR. PREWITT: Well, I haven't had a chance to pay much attention to that yet, quite
honestly. We're in the middle of the largest peacetime mobilization in history. We've got
150,000 people out there now trying to do this job. As I said, we've got about 20 different major
operations going on. And so people who are, at this stage, saying that the Census Bureau should
be abolished, I'll worry about that when the census is over.
QUESTION: If I can follow-up, this is the same group that had, to some degree, a
successful challenge to you before, so don't you --
MR. PREWITT: I'm sorry, say that again?
QUESTION: They had, to some degree, a successful challenge in terms of taking to the court the idea of using the statistical sampling versus a strict head count. I mean if you have another group raising concerns about the census, a group that has raised concerns in the past, what does that do to the atmosphere?
MR. PREWITT: Surely I think that at this stage, in the middle of a census, any negative
publicity about the census is harmful by definition. It is what one wants, what one wanted and
still wants is to describe Census 2000 as a marvelous civic moment in American history when we
can only, as a society, do something good for ourselves, we can only get the basic information
that drives programs, private investment, understanding ourselves as a society. And so anything
that subtracts from that moment of a civic responsibility and civic commitment I can only regret.
I can only wish that everyone who cared about this country used this opportunity not to talk about
dismantling the Census Bureau, but to talk about how to make certain we have the best possible
census in 2000.
MR. JOST: Okay. We've got to go to the phone. Our first caller is Rick Kline from the
Dallas Morning News.
QUESTION: Did you hear me?
MR. JOST: Go ahead, Rick.
QUESTION: I'm sorry. Yes, all in the same vein, I was wondering if you would
comment on the lawsuit filed in Houston by Mark Brewer calling the threat of punishment illegal
and unconstitutional.
MR. PREWITT: I'm sorry, say that again.
QUESTION: The lawsuit in Houston, Mark Brewer.
MR. PREWITT: Yes.
QUESTION: I was wondering if you could comment on that.
MR. PREWITT: Well, as of now, every question on the census has been put there for a
legal/programmatic/policy purpose, and the census itself is mandatory. Those are the laws of the
land. The Census Bureau is simply executing the laws of the land. We are an executive agency,
not a lawmaking body. We're not a law enforcement agency. We're an executive agency. We're
executing the laws of the land.
So my comment is that, once again, in the middle of a census one would wish that all of
our leadership, political leadership, community leadership, public leadership, was talking about
"let's make this census the best we can." If we then have arguments with it, let's do this when the
census is over. We are still looking at a response rate that is not a good indicator of civic
responsibility in this country. And that's what I wish our political leaders would be talking about.
QUESTION: Senator Lott in clarifying his comments said that he suggested that perhaps
people who find questions objectionable not answer those questions, yet send in the form. That it's preferable to have a partial form sent in than no form at all. How do you respond to that, and is it better to count people, even if they refuse to answer certain questions on the long form?
MR. PREWITT: Absolutely, we would much prefer a census form with some good
information than no information, by definition. We would prefer a census form with complete
information, and I don't mean we would prefer it - I think the country would prefer it, and the
U.S. government would prefer it and society would prefer it. But, certainly we do want as
complete a count as possible, and we want as full a set of responses as possible. But, certainly if
we get some information, we will make the best that we can with it. We're a statistical agency,
we have an obligation to the country to give them the best information we can.
I do have to say that if responses to certain items fall below our quality threshold, we
would not be able to report that item back to the country. We have high standards, very high
standards. And if we have reason to believe that the information we're providing is below those
quality standards, then we would have to say we really would be very uncomfortable providing
that to the American public, the American business community, the American government.
That's the kind of work we will be doing over the next several months.
MR. JOST: We'll go back to the phones, and it's Jack Norman of the Milwaukee
Sentinel.
QUESTION: Good morning. Yes, are you planning any changes in the training of the
enumerators, given the environment that they would now have to go out in, and among other
things I'm speaking about supervisors in the census, in the Census 2000 project, who are
expressing concern that the atmosphere now of fear and some hostility and resistance has been
fanned by the critics, creating a potentially really hostile and occasionally perhaps dangerous
environment for these part-time, temporary workers to be going into. So are you planning any
changes in the way in which you would be training them to account for the new environment?
MR. PREWITT: Well, I guess I would say two things about that. Obviously, we would
try to train our enumerators to respond to whatever set of conditions that are out there. We
already have in place a large number of contingency plans. If we run into a certain sort of
circumstances in the enumeration situation, or any other kind of situation, we have a large
number of contingency plans in place. And we will certainly train people to try to be prepared
for anything that they may encounter when they're knocking on doors and trying to get
information.
However, the second thing I would like to say is that millions and millions and millions
of American people are filling out this form, accurately, completely, to the best of their
knowledge. We're getting a flood of letters and e-mail who are concerned about the attack on the
census. So there is more than one atmosphere out there. And part of that atmosphere is the
American society doing what needs to be done -- the people who are going to be knocking on
the doors, these are not federal bureaucrats, these are Americans, neighbors from the local
community. These are people who are trying to do America's job. And I hope that when they try to do America's job, that the response will be a positive one. That's what I can hope.
And I'll just give you one anecdote. I read a letter the other day, addressed to me, dear Dr.
Prewitt, in a very spindly writing, it said, I'm a very elderly woman, I want to cooperate with this
census, it's really important. However, my eyesight is so bad I can't read the form. And Mr.
Prewitt, if you don't believe me, here's the name of my doctor. Please call so you'll know that I'm
not just trying to avoid my duty. That's the America that I hope will respond when we're
knocking on the doors. This lady will respond. She will be glad that we sent someone there to
count her. So I'm hoping that will be the real atmosphere when we get into the non-response
follow-up period.
QUESTION: I'm curious to know whether you expect any problems with an overcount.
And it's my understanding that these "Be Counted" forms are new, and how will you tie those to
an address and keep there from being duplication from that?
MR. PREWITT: Yes - We have put the "Be Counted" forms in place as a safety net for
people who do not get their forms at their address or where there's more than one family at an
address, and only one family responded, the other family says, look, we rent here and we only
have one address, but we weren't included on that form and we want to be included. That's what
the "Be Counted" form is there to do, to try to provide legitimate responses from families and
individuals in the society who feel like they were not counted, somehow missed, and want to be
counted. We, as you know, have established a large number of "Be Counted" centers.
Now, the issue with the "Be Counted" form is for it to be legitimate it has to be tied back
to an address. To remind everyone, the census is two things. It is a count, and it is the
assignment of every respondent to a geographic location. You cannot use this count for purposes
of reapportionment or redistricting or expenditure of federal funds unless you can say this is
where someone lives, this is the part of the town, or the block on which they are residing. So we
have to take that "Be Counted" form and get it back to an address. If it comes in addressed, and
we have a special question, I should say, on the "Be Counted" form that says, "Are you
responding from a household which has already been counted on the form?," it comes back in,
and the address is there, no problem.
We have to worry about the fraudulent use of the "Be Counted" form, obviously. We
have as much responsibility to make sure that we count everyone, as to make certain that we
don't count anyone inappropriately or fraudulently. So we have in place a set of procedures,
depending upon where the form comes from, how many of them come in and so forth, to try to
detect the fraudulent responses. I should say, again, we also have in place the A.C.E., the
Accuracy and Coverage Evaluation.
Do not forget that in 1990 the Accuracy and Coverage Evaluation, then called the PES,
Post Enumeration Survey, was as good at picking up duplicates and undercounts as it was double
counts. We frequently talk about the fact that there was an undercount of 4 million people in
1990, actually the undercount was about 8 million people, but the overcount was about 4 million
people. So the net was 4. So we will use the Accuracy and Coverage Evaluation to try to detect
any use of the "Be Counted" forms inappropriately where we got a multiple count, a double
count.
Go ahead?
QUESTION: [Inaudible.]
MR. PREWITT: If people get two forms, because they actually have a weekend home
and a home, we hope they only send one in. For our purposes, it's their judgement, which is their
primary residence. And we will be happy to let them make the decision. But, yes, we will have
to get back in touch with them and say, we have two forms from you; we can only use one of
them, as best we can. Now, finding all of those is not easy. I don't mean to make this a simple
task. Operationally, it's an enormously complex task. But certainly the Census Bureau is as
fixed on an accurate count. And two things contribute to an inaccurate count, counting wrong
people or putting them in the wrong place or not counting them at all. So we have procedures in
place to try to tackle both of those issues.
MR. JOST: We go back to the phones. Adam Bell of the Charlotte Observer, please.
QUESTION: Hi, Dr. Prewitt, can you tell me of the level of concern you might have for
Southern states in the mail-back response rate, specifically in the Carolinas, right now they're at
about 49 and 47 percent?
MR. PREWITT: Certainly, we do have a concern, and we've been expressing that
concern. We've been saying that it's not too late, we still want them to mail forms back in. We're
hoping, as I say, for a major surge over the next two or three days, based upon the weekend
effect, and the April first effect. We have a map that shows the response rate in 1980 across
every state, and then the response rate in 1990 across every state. And the interesting thing, those
two curves are identically shaped, which is to say, across the country the drop from 1980 to 1990
hit every state roughly equally, roughly 10 percent with some variation around that number.
What we would be most concerned about is if a state did not track its 1990 performance.
That is, even if it were a low state in 1990, but it were even a lower state in 2000, then we would
be concerned. What we are expecting, as a matter of fact, is that as 1980 was a very strong
predictor of 1990, we're expecting 1990 to be a very strong predictor of 2000. But certainly now
what I'm really actually focused on from an operational point of view is whether we get a big gap
between what our model would have expected to find in 2000, in terms of the distance between
the 1990 and the 2000 number, because we have planned our staffing, and our budget and our
operations based upon our 1990 experience, of course. And so if we have an area of the country
which is disproportionately lower, we will then have to beef up. And that's what we are doing
right now.
We are starting right now. We are having meetings this Thursday and Friday, to take a
very hard look at these numbers from an operational point of view, to make certain that we have
the number of staff we need in place, the advertising campaign we need in place, the number of
offices, the number of materials, all the kinds of things we have to have in place when we go into
non-response follow-up. We start training for that on April 24th, go into the field on April 27th.
So, yes, my concern from an operational point of view would only be if we've got a state
which is quite a bit lower than what we had assumed we would get on the basis of our 1980 and
1990 experience. And we're not there yet. We won't know if we're going to be there until the
end of this week.
MR. JOST: We're going to take the folks in the right field bleachers over here.
QUESTION: Bob Rosenblatt, Los Angeles Times. If someone sends back a partially
completed long form, let's say there are two questions not filled in, or 22, do you have a standard
at which you say if you have 80 percent of the questions filled in, it's fine, we'll take it, and if you
have less than 80 we'll go back to your house. How does that work?
MR. PREWITT: We certainly do, but it doesn't work quite as mechanically as you've just
suggested. You've got to take into account a couple of things. You've got to take into account
whether we have the addresses right. We start with the quality of the address file, and make
certain that the address is correct, because that is so critical to us. That's our quality control
procedure, to make sure we can get each one of these respondents attached to a legitimate
address. So we would start there, and then we look at the quality of the data. But, we look at a
number of things. For example, if someone says there are 11 people in this household and they
only give us information for one person, that's one set of decision rules. If they say there's one
person here, but they give us information for 11 people, list 11 names, that's a different piece of
information.
We also look at the pattern of item non-response. Obviously, the responses to some items
are a lot more critical to us, that is, the 100 percent items are critical for administering the
Department of Justice obligation under the Voting Rights Act. So we will look actually at
different patterns in the data, so it's not an automatic decision rule. And I can share all that with
you, I don't have all that in front of me today. We will be doing a lot of work on that, as I say,
over the next two or three weeks. And we can share that at a later session. But, the answer is
certainly, yes, we have decision rules, but they're differentially applied depending upon the
patterns that we're getting.
MR. JOST: We'll go back to the phones. It's Andrea Robinson of the Miami Herald.
QUESTION: I was wondering. I noticed you mentioned that Florida is one of those
Southern states that has a low response rate so far, which is surprising, considering the number of
older people in the state, and they're generally more reliable about turning in forms. I'm wondering, are you going to have to increase the number of enumerators in Florida? How is that
going to work?
MR. PREWITT: Yes, well as I was trying to suggest, we already know that we're going
to put more enumerators in certain places than in other places, that is our long-term planning was
based upon that. If the Florida numbers come out to be quite a bit lower than what we presumed
on the basis of their 1980 and 1990 return rate, then we will be putting more people on the street
than we might have thought we would. We're not there yet. We actually expect every state to
roughly track its 1980-1990 pattern. It may be lower, as you know. We expected -- we
budgeted for 61 percent, which is 4 percent below the 1990, but we would expect every state to
be roughly 4 percent below the 1990. If we have a state which is 8 or 10 percent below the 1990,
then we have to put more enumerators there, and we will do that. So that's the kind of work
we're doing right now, to be ready for our operational task.
MR. JOST: We'll go right here in the second row.
QUESTION: Monique Conrad, Blade Communications. Ohio has the highest initial
response rate so far, can you explain why that is, and if you have a high rate of response for
suburban, urban and rural communities?
MR. PREWITT: No, I can't do it. I'm sorry. I mean, that information exists, if you want
to go to our Web site we have every community in Ohio listed, and you could do that work. But,
I haven't actually looked carefully at the state of Ohio. And I can't give you an explanation right
now for why Ohio is unusually high. Somebody has to be the highest. We wish it were a tie for
first place across 50 states and Puerto Rico, but it's not a tie for first place right now.
There are odd things that go on in terms of this response rate. That's why we've said all
along, to remind us all, we are publicizing this response rate in order to encourage the American
people to, themselves, write the census story. We can't write the census story, only the American
people can write the census story. And we're deliberately, for the first time in history, doing this
in order to let the American people know how well we are doing, even as the census is unfolding,
so they can write the last chapter.
Now, we will not know until really April 18th. That's the last day in which we will be
reporting back to you on the response rate. We will be continuing to get a scattering of responses
even after we close down for mail-out period, that is April 11th when we stop this to get ready
for the non-response follow-up period. But we will continue to get a few more in, we hope a lot
more in the ensuing week. On April 18th we're going to come back to you publicly and give you
our full assessment of what we've learned from this experiment, what it means about civic
responsibility, what it means about our operations, what it means about advertising.
Organizations are collecting survey data even as the census unfolds.
I think by April 18th we're going to have a very powerful picture of what this response
tells us about American society. But, at this stage I'm sort of reluctant to pick and choose a given
state. A lot of things can go on in a given state that could either be boosting its numbers or
lowering its numbers. And we need it to converge. As I said, we expect it to kind of converge.
We hope that it's consistent with the 1990 pattern. We would expect roughly every state to be
consistent with this 1990 pattern. We hope that it does not level out at 61. We obviously still
want 70 percent, we actually set the target at 5 percent above our 1990 level as a statement to the
country about the kind of place we want to live in.
And it's in that context that I go back to the point that this is an awkward time to be
dealing with a lot of negative commentary about the census, because we would love to have all
the commentary about the census right now being that this is a marvelous opportunity for the
American society to prove that they can reverse the decline in civic disengagement. And if we
end up at 61, that means we did not do it, that means we continued to decline. If we end up at
65, that means we arrest it, we at least stopped it. That's no modest accomplishment. And if we
get higher than 65, we reversed it. That's even more of an accomplishment. To get it all the way
to 70 would be a stunning reversal, but even above 65 is, I think, a serious victory, both for the
Census Bureau and for the American people.
MR. JOST: Back to the phones, Ben Hammer of the Industry Standard.
QUESTION: In light of all these attempts to raise the response rate, I'm wondering why
the agency has downplayed the ability of recipients of the short form to answer their
questionnaires online?
MR. PREWITT: Did you say why we haven't? Downplayed it, I'm sorry, I didn't quite
hear you.
Planning a census does take about 10 or 12 years. We already have major committees at
work planning the 2010 census. This census started to get planned in 1990. The operations are
actually very big and complicated. And the Internet, of course, in 1989 was not a prominent part
of American society. And we had to put our operations out there. As we got closer to this
census, and realized that the Internet was an opportunity that we should be using, we made it
available. That is, we decided to design some software, make it available for people to do the
short form. We then asked ourselves exactly the question you're now asking, to what extent
should we publicize it and promote it?
The decision that we made is that because the Internet is, at this stage, very inequitably
distributed across the American society, and since our focus was on especially getting people we
have a hard time reaching, we would spend all of our advertising dollars trying to reach the
population groups which we had reason to believe would be undercounted. Those we do not
believe are going to be the Internet responders. We think that the Internet responders, whether
that number is 60,000 or 6 million, are people from whom we would have gotten a paper form.
Therefore, we're going to get their answers, we ought to spend our time and effort trying to get the answers from the people we might not get an answer from. So we made that simple decision.
We're a reasonably cautious organization. We do not like to put new, big operations in
place where we haven't had a chance to test them. And at that stage, this is in 1998-99 when this
conversation came up, we had not yet had the opportunity to test Internet responses in a census
environment. So we thought the most prudent thing to do was to allow it, to test it, to see how
well it worked, and then to decide on the basis of that how major a push to put by 2010, and I'm
sure we'll put a major push on it in 2010.
So it's a long, convoluted answer, and I apologize for that. But it's not because we didn't
want them, it's because we wanted to make sure they would work well. We had to worry about
encryption, of course. We have now a heightened conversation about privacy in the country.
And we wanted to make certain that we would have nobody who would say, oh my goodness,
somebody else got my answer, because I filed by Internet. So we had to be extremely cautious.
And that's why we downplayed it slightly for 2000.
MR. JOST: In the back there?
QUESTION: Andy Sullivan, Durham Herald-Sun. You mentioned earlier that you were having trouble finding enough people to go door to door in North Carolina. Given that state's lower response rate, are you especially concerned about the results from North Carolina?
[TAPE CHANGE.]
MR. PREWITT: -- to be traveling the byways and be knocking on the doors in the city
streets and city addresses. I do want to remind us all that our recruitment target was set for April
19th, which is to say approximately three weeks ahead of it we met it. I would remind us all,
there were many, many, many skeptics in this country. Go back and read the press, as much as
six months ago or 12 months ago, saying, we will never reach our recruitment goal, how are they
going to do this census? Unemployment levels at 4 percent, they will never find the people.
Well, lo and behold, we found them.
Now, that doesn't mean that there aren't certain locations where we still have work to do,
and we're working on that. But imagine standing here today and saying, my goodness, we've got
a response rate issue, and we don't have anyone to go out and knock on the doors. Then there
would be a big story. We don't have that story to write today. Imagine that we are sitting here
today saying, our data capture system isn't working. That's a big important part of our operation.
You would have a big story to write. You don't have any -- sorry, let me phrase this more
carefully. As of now, I'm certain that the operations story that you can write if you choose to
write an operational story, will be a story which says, all operations are functioning the way the
Census Bureau had hoped that they would, and we're very pleased with that. I'll stop with that.
MR. JOST: We have time for two more. We'll go to the phones first. Doris Chandler of
the Gainesville Sun News. How about Maureen Fan of the San Jose Mercury News.
QUESTION: Can you hear me?
MR. JOST: Yes, Maureen. Thanks.
QUESTION: My understanding is that for people who haven't gotten a form, it's too late
for you to send out a new form, and instead they are to wait for an enumerator to knock on their
door, or they can go to a QAC site and get a "Be Counted" form. Which is preferable and how
can you be sure that these people will definitely get a visit?
MR. PREWITT: Well, neither of those are preferable. Actually, it is not too late to get a
form. We would actually prefer that they use the telephone questionnaire assistance system.
That is, to call the 800 number, to remind you, 471-9424. That's the preferred, that gives us the
most amount of quality control over that questionnaire, because then we have already put into our
records to whom we sent it, to what address we sent it. Therefore, it's an easier questionnaire,
when it does come back in, to code to the right address.
The second tier then would be to use a "Be Counted" form. All of our "Be Counted"
forms are now fully functioning. They exist on our Web site. You can also call your local office
to get the addresses of our "Be Counted" sites. But that is a less preferable way to get a form
than to use the telephone questionnaire assistance service. Our telephone questionnaire
assistance services are functioning, they're handling all calls without any difficulty, and we would
urge people to use that.
What I have to say is that there will be people who get a form late in the process, either
through the telephone system or using a "Be Counted" form, and it will be very, very difficult for
us, especially "Be Counted" forms, to code them back to the address in time to keep their address
off of the non-response follow-up list. It's a big undertaking to create this non-response
follow-up list. And we have to start that immediately. That is on April 11th and we will start it
then.
Questionnaires, including "Be Counted" forms, will come in after that period. They will
already be on the non-response follow-up list. Someone will be knocking on their door. There
will, therefore, be people saying, "Look, I sent it in, why are you bothering me?" And we will
have to try to explain that. We would love the help of the press in explaining that. But
operationally, you cannot fine tune this down to getting a "Be Counted" form and suddenly
immediately take that off of a non-response follow-up workload. So there will be some overlap
between the households in that respect.
Back to your question, Maureen, they can still get a form. We would urge them to use the
800 number.
MR. JOST: And right in the back with the headphones.
QUESTION: Bill Small, Bloomberg, Dr. Prewitt. I'm just a little bit curious, in the
questions that have been, by one politician or talk show host or any other, have been declared
intrusive, do you see an impact on any particular demographic group, political group, whatever it
may be, from not answering those questions. And if they're not answered, is it possible to do
something at some point later on to still utilize the information, perhaps some sort of sampling?
MR. PREWITT: Certainly there are groups who are going to be harmed if we have
unexpectedly damaged long-form data. As I've used examples before, let me use a couple of
examples again, if you're an elderly veteran, and you're living in an area of the country with other
elderly veterans, that's where you want the veterans hospital to be. And the long-form data
identify the fact that you're a veteran, the fact that you're disabled, and the fact that you're elderly.
And so, if you're someone who fought in the Second World War, or the Korean War, you fall
into that population group. I think this country owes you the service it's promised you. It's hard
to deliver that service if we don't know where you are. So, yes, that's a population group that will
be harmed if we get bad data on age, bad data on veteran status and bad data on disabilities.
And since one of the -- I choose my words carefully -- cheap shots at the Census Bureau
questionnaire is on the disabled question, on can you bathe yourself, that's a very powerful
indicator of whether people are disabled, whether they need help. That's what the question is
there for. And so, yes, a disabled population is going to be harmed. An elderly veterans
population is going to be harmed.
Head start, if we have bad data, inadequate data on income, the income data create the
Title I schools, and the Title I schools then, in turn, are used for creating programs that provide
services to school children under the poverty level. So, you can just go through the long form
and ask yourself, if you have bad information on that particular item, what are the groups that are
going to be hurt? The people who are complaining to the talk show hosts driving their cars
through congestion every morning and every afternoon, are going to be in more congestion if we
don't have good data on how people get to work, because those data are used to provide planning
and finances for reducing traffic congestion.
People concerned about public health are going to be concerned if we don't get good data
on indoor plumbing, because that tells us something about ground contamination, and that is
used in public health planning.
There is not a single question on the long form which, if it is below our quality standards
and we can't use it, won't hurt someone. So the answer is, yes.
Now, to the second half of your question, you can see I'm a little wound up on this. To
your second question, fortunately, the Census Bureau has been planning for the last several years
something called the American Community Survey. We have been saying for several years, and
certainly will now say more forcefully at congressional hearings and other settings, that we believe this should be the last census that collects the long-form data the way we're collecting it
this year. We've already been saying that. And that's why we have designed the American
Community Survey. And the American Community Survey takes the long-form questions and
spreads them across a 10-year period on a small sample each year. By Census Bureau standards,
this is a small sample, by anyone else's this is a very large sample, roughly 3 million households
a year.
Three million households in, let us say, 2002, would get the long-form questions, another
3 million in 2003, another 3 million in 2004. After five years, we would have the equivalent of
what we now get from the decennial, which is to say the country would have more timely data, a
database of information that would put less of a burden on the American people, and we think
would give us much, much higher quality data.
So, yes, we would urge the U.S. Congress that if we have bad long-form results in 2000
to accelerate the planning for, and accelerate the funding for the American Community Survey,
and use that as the mechanism to create the data that would make sure these programs get to the
right places and the right population groups. We're prepared to do that.
QUESTION: If I could follow-up?
MR. JOST: Just one quick follow-up.
QUESTION: A list of groups that you named in response to the first part of my question,
is my perception incorrect that the majority of those would seem to be primarily lower income,
possibly primarily minority groups?
MR. PREWITT: Well, I don't think of the World War II veterans and the Korean War
veterans necessarily as minorities or nonminorities. I think of those as Americans who deserve
veterans benefits. I don't think of people in traffic congestion or breathing poor quality air as
necessarily minorities or not minorities. It think that cuts across the entire population group.
Certainly, Head Start is targeted. Most of our school reform effort that's being discussed
is targeted on groups in poverty and minorities. But if you actually take a look at the programs
that are being driven by long-form data, you will not find it as focused upon just minorities and
the impoverished as some of the commentary would suggest at all. Education reform is not just,
after all, for minorities. Education reform is for everyone in the public school system, at least
that's the way it's being discussed.
So, no, I think it's a little misleading to try to characterize the concern about the long form
as a concern about particular kinds of social programs. The long-form data are not just used for
government programs. They are used by private businesses, heavily, all over this country in
order to make decisions about plant locations, mall locations, where to build a real estate
development. So those are not specific to any given population group. Therefore, the entire
American society.
Finally, and I will quit lecturing, when we built the industrial economy, we built it on an
infrastructure of highways, of railroads, of canals, of ports. We're now building a knowledge
society, a knowledge economy. The knowledge economy is going to be built on the
infrastructure of information. And, the decennial census data are a critical platform to quality
information in this society. And I believe very strongly that we harm our society if we don't put
in place a basic information infrastructure.
Your Social Security is indexed to the consumer price index. The consumer price index
is, in turn, based upon the use of census data. That's for all of us, every one of us, who have a
pension plan which is indexed to the CPI, which is all of our pension plans, which is in need of
good census data because you cannot run the surveys that produce the CPI in the absence of that
foundation, which is produced by the decennial data.
QUESTION: August Gribbin of the Washington Times. I wonder if you could help me
understand how the American Community Survey would address the question of the
intrusiveness of the questions. It does spread it to a fewer number.
MR. PREWITT: No. An important question, Gus. It doesn't change the questionnaire.
It's still exactly the same long-form questions for the most part, it just spreads it across about 10
years. I think it addresses the intrusiveness issue as follows: We would be able, I think, to better
explain because we're now explaining over a continuous period of time, not every 10 years, over
a continuous period of time, the purposes of this information.
Secondly, quite honestly, the cheap shots, I'll put it that way, about the long form, is
harder to sustain. You know, a lot of people are taking advantage of the census environment
because there's a lot of attention to the census to run with other agendas and, in the absence of
kind of the census environment, it's very hard to run with those agendas. So, I would hope that
we would end up with better data. We would end up with better, deeper public understanding of
that data. And, we would also end up in a political environment in which it would be easier to
provide the country the data that it does, indeed, need.
[END OF EVENT.]
