Press Briefing -- April 19, 2000
|[Joined in progress.]
STEVEN JOST (associate director, U.S. Census Bureau Communications): Good
morning. Thank you for joining us today.
My name is Steve Jost. I am with the Communications Office at the Census Bureau.
And today is a special event where we are going to talk about the latest initial mail-back response
rates for the country, in total and for the individual jurisdictions.
I'd like to make one announcement before we proceed. Next Tuesday, we will have one
of our standard operational press briefings, which will talk about the status of the census in all its
glorious detail. And since we have a great deal of information here to go over, we would ask
that, if you have questions about those particular operations that we'll be addressing on Tuesday,
that maybe we can put some of those off until that time and deal with the response-rate issue here
And one other note: On the wall behind me, the map you see reflects initial mail-back
response rates processed by the Bureau through close of business Monday of this week. And
they are the numbers that you should see on the Web site, although we are processing every day
and some of these numbers may change. If they do change, they would go up, although we are at
a volume now where it's doubtful any of these numbers should change. But we just want to
make sure that you double-check, tonight at 6:00, on our Web site. The national number is not
likely to change, but one of the state numbers might change, if they are very close to moving
from one percentile up to the next percentile.
And with that, I'll reiterate our ground rules here. The director will have a few opening
remarks; we'll turn it over to questions; we will alternate between reporters in the room and those
that are on joining us from around the country on telephone. We ask that, when you are
recognized for a question, that you identify yourself and your affiliation.
And with that, I give you Director Kenneth Prewitt.
MR. PREWITT: Well, just to repeat what Steve has said; as has been our practice,
tonight at 6:00 p.m., we will post the mail-back response rate for approximately 39,000
jurisdictions. And these numbers will reflect performance through April 17th, Monday. We
were working hard to see if we can update them slightly -- that is, to post through last night -- but
that's not a promise.
That number, tonight at 6:00, if it reports the response rate through Monday, will report a
national rate of 64 percent. However, the cover page of the Web site will include a statement
from me based on yesterday's returns at the national level; that is, it's easier for us to collect the
information at least at the national level. It's harder for us, of course, to process all the
information for 39,000 jurisdictions.
And that Web page will read as follows, more or less: Sixty-five
percent of the households in America have returned their census forms. This is a serious achievement; it is
news to celebrate. The country has stopped in its tracks a decades-long decline in meeting our
civic responsibilities. The country is saying that democracy is about obligation, as well as
benefits; about responsibility, as well as rights.
In reaching 65 percent, the American public outperformed the expectations of the Census
Bureau, the United States Congress and the Government Accounting Office.
More than 100,000 census partners deserve credit. Congratulations are owed to
thousands of mayors, county commissioners, teachers, community advocates, houses of worship,
and other local government, civic and business leaders. They were determined to ensure the
highest possible mail response rate, and they treated the census as the serious civic event
intended by the founders, when in 1787, they wrote the census into the Constitution. And I
remind you of our major partners, official supporters, of our effort to improve the overall
Now, the Census Bureau and its community partners turn to an even more demanding
task, convincing the millions and millions who did not return their form, nevertheless, to
cooperate when census-takers knock on the door in the coming weeks. A half-million Americans
will be out doing America's task. Census 2000 is poised to be the first successful civic event of
the new century, and we need only sustain the momentum and commitment represented in the
initial phase of the census. And, as I say, that will be up on the Web site later this afternoon.
Let me just go on for a moment about the numbers. The posted numbers tonight will also
show that 15 percent, or nearly 6,000, of America's cities, counties and tribal lands, met the
extraordinarily challenging goal of bettering their 1990 response rate by 5 percent or more. And
obviously, these communities merit special acknowledgment.
Just a random - not random but just selecting a few to read off to you - these are now
parts of the country - counties, cities, townships, tribal lands - which not only met their 1990
goal but exceeded it by 5 percent or more, which, of course, was the extraordinarily difficult
challenge we tried to give to the entire country: Stanislaus County, California; Anchorage,
Alaska; Anaheim, California; Burbank, California; Monument, Colorado; High Springs, Florida;
Decatur, Georgia; Idaho Falls; Lawrence, Kansas; Baton Rouge; Boston, Massachusetts - by the
way, which exceeded its HAKWAN - its 5 percent-plus goal - by 3 percent; that is, exceeded its
overall response rate of '90 by 8 percent - Sandusky, Michigan; International Falls, Minnesota;
Gulfport, Mississippi; Reno, Nevada; Atlantic City; Glasgow, Pennsylvania; Naragansett
American Indian Reservation.
I should also say that Washington, D.C., got very close, got within 2 percent, that is, of
reaching that very difficult goal. I mention that only because I live in Washington, D.C., part of
the time. And I'd like to announce that my own home city, where I live most of the time, if not in
Washington, Manhattan, did exceed by 5 percent its 1990 response rate, as did Putnam County, where I happen to have a weekend home. So, at least wherever the census director is - (laughter)
- we do meet this extraordinarily difficult challenge of exceeding by 5 percent the 1990 rate. So,
as I say, that's 15 percent of the jurisdictions across the country and nearly 6,000 of them have
really, really gone the extra step of not only meeting the 1990 goal, but also exceeding it by at
least 5 percent.
Now, although no state reached this goal, 12 states did exceed their 1990 performance;
that is, they were somewhere between their 1990 performance and the Plus Five goal. The two
strongest states, measured by the differential, are Massachusetts and Rhode Island. That is, those
two states exceeded their 1990 rate by 3 percent. And then a large number of states exceeded it
by 2 percent, including Alaska, California, the District of Columbia, as I have just mentioned,
Nevada, New Hampshire - you'll see all of these on the form - Wyoming - I'm not trying to give
you all of them, but 12 states did exceed their 1990 goal.
Perhaps even more telling, 41 states and the District of Columbia met or exceeded what
the Census Bureau, the GAO and the Congress expected. That is, as you know, we have staffed
and budgeted for a national response rate of only 61 percent. That is, across every state, a drop of
about 4 percent from 1990. Only nine states fell below this expected rate and three of those
missed it by one point. Now, I know this is a lot of numbers to keep in mind, but bear in mind,
the national number today is 65 percent, even though the number we post tonight will be 64
percent, which means there's still movement in these numbers, through yesterday.
So, I only say that three of the states are within 1 percent. I don't know for certain
whether they will reach that goal, but at least of the handful which are below the 1990 number,
some of them could easily catch up with the 1990 number before we're completely finished
processing all of the returns.
Now, I do want to say that there is less than one bright part of this otherwise remarkable
achievement. Although the households that received a short census questionnaire returned at a
rate that easily exceeded 1990, those that received the long census questionnaire fell off
dramatically. That is, in 1990 households which got the short form returned it at a rate of 66
percent and those that got the long form returned it at a rate of 60 percent.
In 2000, thus far, 66.6 percent of those with the short form have returned their
questionnaire and only 54.1 percent of those who received the long form; which is to say, the
current differential between the long and the short form is 12.5 percent.
For reasons that many of you have covered, the census as a civic duty message was
compromised for those who received the long form. A garbled message was sent that somehow
statistical information threatens privacy. On this point, I would like to say just another word or
Information and privacy both matter to the nation, and both can be enhanced. That is,
information and privacy are not in a zero-sum situation, whereas more of one necessarily reduces the other. Information is infrastructure for the new knowledge economy in no less a profound
way than canals and railroads were the infrastructure in the startup years for the industrial
economy. And, of course, privacy is a cherished American value and, for good reasons, we hold
on to it.
But to have carelessly put the Census Bureau in the crossfire between the need for
information in a knowledge economy and the value of privacy misses a simple point. Federal
statistics are not about individuals and thus not about privacy. To have a statistic on how many
of us are elderly and disabled - that is, cannot bathe ourselves - tells no person about any other
person. It is not about privacy, and it's certainly not about marketing soap. But it is information
that puts a veterans' hospital where it will do the most good, that helps motivate charitable giving
and, yes, that reminds the households with bathrooms in excess that millions live differently.
It would help enormously if the press could help explain the difference between a statistic
and private information. For me to announce to the viewers today that there are 75 people in this
room, of whom approximately 30 percent are female, tells no viewer anything about any one of
you. It does not infringe on anyone's privacy. It does not even tell any viewer who is in this
room. It is a statistic that summarizes a characteristic of who happens to be in the room at the
present time. And it really is important, it seems to me, for the country to understand the
difference between private information and a statistical number, because in the absence of
understanding that, we cripple our capacity to get the kind of information that we need if we're
going to sustain a knowledge economy.
And I believe that this elemental distinction got lost in the public discussion and that the
response rate on the long form suffered.
But back to the happier news: I do want to just take a moment to direct your attention to
some survey data. I made reference to some of these survey data earlier - and I think that they're
in your handouts; you don't have to worry about taking those numbers down - but they are quite
revealing, are quite suggestive. As you see, by the time we had finished the major publicity
campaign surrounding the census, we are looking at awareness levels at 98 and 99 percent, which
is to say, essentially, we saturated the public conversation with the census message quite
successfully. And that was in part, of course, related to our major advertising campaign.
As you can see, by the end of the census period, only 5 percent of the American public
were saying that they'd never seen any television commercials about the census, and that's just
TV, that's not all the other material. Indeed, as many as 45, 44 percent, nearly half the
population, had seen the commercials at least 10 times. And that's just the TV, that's not all of
the other kinds of materials that were used in the advertising campaign.
As you can also see, those of you in the media, will note that a very large percentage of
the population did get census stories from news coverage - from TV, from radio and from print.
Between a quarter and a half of the population is making that claim.
It's the last set of these data that I want to draw your attention to. We tracked the reasons
- I shouldn't say "we," but the InterSurvey, which is the survey firm funded, as I have said
before, by private foundations, who have tracked this information during this census experience,
asked: "What would be the reason for returning the census form?" And if you look at the data,
you see that the dominate reason had to do with expenditure of public funds. Now, of course,
since that was the strong message of the advertising campaign, that's what you would expect.
I personally take some gratification in the fact that more than half the population, 60
percent of the population, consistently identified "civic duty/civic responsibility" as one of the
reasons for filling out a census form, and that was constant; that's not something we made
happen, it's just in the body politic. I, personally, am delighted to get that piece of information,
that more than a majority of the American public do see this as a civic responsibility.
The other thing that, of course, increased a lot during the period of the census was the
notion that the census is mandatory by law. As you see, that more than doubled during the
course of the month, from 21 percent to 46 percent. And I'm happy to answer any questions you
might have on that.
Then let me just conclude by returning to the response rate itself. It's Robert Browning
who said that a man's reach should exceed his grasp, or what's Heaven for? Well, so should a
nation, and I am actually very proud that we set before the American people an unusually tough
challenge; that is, to arrest and even reverse the precipitous decline in civic engagement that
we've all witnessed since the 1960s. This decline was stopped in its tracks and, in thousands of
communities across the country, reversed.
We will work hard to sustain this momentum and the tough task yet ahead. A
half-million Americans will be out doing America's job, and it's time to give them our full
support. It is time to urge those who did not mail the form back to cooperate when the census-
taker knocks on the door.
So I'll take your questions.
Q Mr. Prewitt?
MR. PREWITT: Yes.
Q Pablo Sanchez with Univision. What happened with Puerto Rico, and how did the - ?
MR. PREWITT: I'll say a word or two about Puerto Rico. On the map, it shows the
Puerto Rico mail-back response rate at 48 percent. Let me take just a second to say something
about Puerto Rico. In 2000, we conducted the census in Puerto Rico the way we conducted it in
the United States, in the 50 states and the District of Columbia. It's the first time we did a mail-out/mail-back census in Puerto Rico, which means we don't have a 1990 measure.
Now, we put Puerto Rico's goal at 65 percent, which is to say we simply assigned it the
nation's goal. We did that in conversation with the governor, because we didn't know how to
assign it, since they had no 1990 base. So it's a little unfair to Puerto Rico to set as its goal to
reach really what was extraordinarily high the fact that it had not been through this kind of census
before. And I might say that we actually internally budgeted for a mail-back response rate of 50
percent. That's what we thought we would get, and at 48 percent, we are very close. So we're
actually quite pleased with Puerto Rico.
You asked a second question, which has to do with demographic return rates. We simply
haven't done that work yet. We're trying to do a little bit of that work now, and we'll try to
report on that as soon as we have it. But as of now, all we've been able to do is calculate these
response rates by jurisdiction.
MR. JOST: Okay, we'll go to the - we've got someone on the phone, Mae Cheng with
Newsday. Are you there, Mae?
MR. PREWITT: Yes, just quickly, the question has to do with whether we will have
special efforts in those areas in New York City, especially African American areas, where the
response rate is low. As I just said before, we have not yet calculated response rates by
demographic groups. We have only calculated it by geographic areas.
Obviously, the task that's now under way at the Census Bureau is to make our
assignments for the non-response follow-up period. And of course, the operational people are
poring through the response-rate data because that tells them about the workload. And as Steve
said, I will describe the details of that by next Tuesday. We are simply now doing it. But, by
next Tuesday, I'll be able to lay out for you exactly how we're doing it in the various areas.
What happens in a census is that you get this response rate, and that tells you where your
workload is for non-response follow-up. And you based your initial distribution of your response
rate on the 1990 numbers. So anytime you get an area which did better than 1990, then that
reduces your workload; and an area which did less well than 1990, that increases your workload.
And we have to direct the resources accordingly. And that's what we are doing right now, all day
long, every day, working all weekend and so forth.
We are comfortable with our recruitment pool. As you know, we exceeded our goal substantially at the national level. And we are now calling people up and asking them if they want to work for the Census Bureau. We are already training 40,000 crew leaders, even today and tomorrow. So we certainly have in place operational strategies to reach the neighborhoods
and the areas which were unusually low but I will describe that in detail next week.
Now, you asked the more particular question in New York of photo IDs. We have met - I
personally met with representatives from the Mayor's Office, explained the photo ID issue and
they understand why we cannot do that. It's technically extremely difficult; it doesn't solve the
problem of people scamming the census because anybody can go out and get a photo ID.
What we urge the American household to do, those who are concerned about somebody
masquerading as a census-taker, is to follow simple rules: No census-taker will ask to come into
your home unless you invite them in; that is, no one will ask that question. So anyone who asks
that question is already not a census-taker; they are somebody else because they are violating our
policy. That doesn't mean that an individual can't invite them in. But no one who comes to the
door, knocks on the door to take the question, will ask to come into your home. So we say that
strongly to the American people.
Secondly, it's not just the ID that matters because IDs can be fabricated. What's much
more difficult to fabricate, of course, is all the paraphernalia; that is, a census-taker is going to
have a very complicated-looking book. It's fairly large. It has a whole list of materials in it.
Obviously, you don't share the material itself with the respondent because it's confidential
material. But, nevertheless, you see this complicated book. You also have a census bag. You
have the stuff that you need in order to do this task.
So we think that, if the American people simply pay attention to who knocks on their
door, they do not have a problem of being scammed.
MR. JOST: Okay. We're going to go this side of the room, if we can alternate back and
forth, and we'll start with Haya.
Q Yes. Haya el Nasser (sp), USA Today. A couple of questions. One, on the survey,
do you have any new results on the privacy question? And secondly, are there any more regional
concerns in terms of response rates lagging in certain parts of the country?
MR. PREWITT: Quickly, as I had mentioned before, the InterSurvey, which conducted
this ongoing survey during the census experience, did ask a question, "Do you consider census
questions to be an invasion of privacy?" And if we put that on there, it would have showed the
first three weeks at 10 percent, 10 percent, 10 percent. Well, I think it jumped to 18 percent in
the fourth week, and - I could be wrong - I think it was 22 percent in the fifth week. So this is a
strong indicator that the public conversation about invasiveness certainly had an effect.
Whether that led to the decline in the long-form response rate, I'm simply not prepared to
say. I need a direct piece of evidence to know if that is true or not. Lots of things can explain
that, and not just the public expression of concern about privacy.
This is why I go back to the point that I did make. It's extremely important, I think, for the American public to understand that statistical information and privacy are not in a zero-sum situation, because if we allow that to become the understanding in the public, we will no longer be collecting the kind of information that we have actually used in this society for almost 200 years.
I'd just remind you that this century, the 20th century, really is the first measured century
in world history, and that is, the capacity to actually pay attention to what's going on in the
country, systematically, continuously, wasn't developed until the late 19th century. So the 20th
century is really the first time that, in world history, the capacity to actually measure what's going
on in a society systematically, reliably and publicly exists.
I don't believe it's an accident that that century also shows more progress on more fronts
- on economic fronts, on health fronts, on transportation fronts, on public services fronts - than
any other in history. I don't think it's an accident that our capacity to understand who we are,
where we've been as a society - and if that's hard to believe, then compare our experience as a
country to the Soviet Union, which systematically undid its measuring system, or Communist
China, which systematically denied and suppressed information. The countries which have been
strongest over this century are the countries which publicly measured and reported to the public
what we knew about ourselves as a country. So it's very, very risky to undermine that capacity, I
MR. JOST: Okay, we'll go back to the -
MR. PREWITT: I'm sorry. Haya also asked if there are regional patterns that are
There are certainly no big regional patterns. As I say, we are now going to what we call
our hard-to-enumerate areas to see exactly if there's any fall-off.
Early indications are, by the way, if you actually rank the country in terms of
hard-to-enumerate to easiest-to-enumerate, that the gap between 1990 and 2000 is much less in
the hard-to-enumerate areas than in the easier-to-enumerate areas. In other words, in some of our
most difficult parts of country - parts of our cities, rural areas, and so forth - we more than met
our expectations for 2000. And the areas where we either didn't meet it or only barely met it or
fell behind it, turned out to be the easier-to-enumerate areas. And that's actually, across the
country, very good news for us. As I say, by next week, when we reconvene on operational
issues, I hope to be able to lay those data out much more systematically. We just haven't done
that work yet.
MR. JOST: We'll go to the phone. Jack Norman with the Milwaukee Journal Sentinel.
Q Thank you. Jack Norman, Milwaukee Journal Sentinel. Following on that last point, the states which were the highest in response rate in 1990 - Wisconsin, Minnesota, Iowa -
are the states that did fall the farthest this year, even though comparatively they did relatively
well. Do you have anything specific to say about that particular phenomenon?
And the second question, in terms of including in these figures forms returned through the 17th,
does that include any "Be Counted" forms, or is that exclusively forms that were mailed to the
MR. PREWITT: Yes, on the latter part, it excludes all "Be Counted" forms. It
eventually will include - we get forms without ID numbers on them, and that includes the "Be
Counted" forms. We then have to take that form back to our Master Address File and try to see if
we can match the address on it to our Master Address File, and if we can, and we don't have a
form from that address, then we assign it that ID number. And we're doing that work right now.
We're doing that work not just to report on response rate, we're doing that work because that
affects our non-response follow-up workload. So we're taking non-ID'ed questionnaires and
trying to process them, even as we speak, and that will include the "Be Counted" forms.
But as of now, the enumerator and denominators only depend upon the mail-out universe,
that is the universe of addresses to which we actually delivered a questionnaire. So actually, the
overall response rate will be somewhat higher than this eventually, because it will build into the
enumerator the non-ID'ed questionnaires.
Sorry, there were two parts to that question. I forgot the first part.
Q The first one was the phenomenon -
MR. PREWITT: Oh, yes. The, as we once referred to it, Lake Wobegon phenomenon.
(Laughter.) Well, actually something happened in Lake Wobegon world. (Laughs.) You're
quite right. We do not have an explanation for that. We have watched that carefully over the last
two weeks, kept expecting that to close and it simply did not close. That is, the states which
were particularly strong in 1990 are still strong, of course, in 2000, but proportionately speaking,
or relatively speaking, less strong. And indeed, states like California and Massachusetts really
seriously outperformed 1990.
It's just too early for us to try to explain that. Indeed, I no doubt will be reading the press
over the next week or two, and will find excellent explanations there from those of you who go
out and do some field work.
MR. JOST: Right here, behind Ken.
Q I'm Carol Mann with the Alabama Radio Network, and Alabama is one of the states that's pretty low in its response and lower than it was in 1990. You said you were satisfied with your recruitment efforts to get people to do follow-up questioning, and I'm wondering if you are still recruiting, particularly in the states where you've had a low response or if you've closed the
books and you're now training those you've recruited.
MR. PREWITT: Well, first, my records show that Alabama actually did outperform its
1990; that is -- oh, no. You're right. Sorry. Misread this. They are 3 percent below their 1990 response rate. They were 62 in 1990.
To the second part of your question, oh, absolutely. We keep recruiting until we're
finished with the census, in some respects. Obviously, our energies now have shifted to training,
but certainly in areas where we think we still will need to recruit, we are actively doing that. Here
are two things that happen in a recruitment effort.
You first try to get a recruitment pool that's bigger than the number of people you need,
because when you start calling people, they say, "Oh, no, I'm not interested any more," or "I've
got something else," or whatever. "I'm back in school," and so forth. So you start with a larger
pool. We started with a pool that we thought needed to be something like seven to eight times
larger than the number of people we needed. And across the country, that's the target that we hit.
Secondly, we then call people, they come to the training, they get trained for a couple of
days, and say, "This is not for me. I didn't know it was going to be like this. You mean I
actually go out into neighborhoods and knock on doors and ask people these questions and they
don't want me there?" So, they drop out after training. For that reason, we train approximately
twice as many as we think we're going to eventually need. If all of them stay in training, fine.
We have a management system that will allow us to manage them.
But some people will go through training; after two or three days of this, we will think
either they're not performing, we've got a productivity problem or they quit because they don't
like the work. Or we do quality control work on them and we find out that their quality is not
what it ought to be, but maybe they're just simply, as we say, "curb-stoning." We then throw
everything out that they've done and start all over with their workload. So we do systematic
quality checks during the process. And that's why we start with twice as many as we need,
because we need to end the process at a certain level. But we'll start with roughly twice what we
need in the training process, so that at the end of the process, we don't have to re-recruit.
That does not mean in a particular area that where we've had a recruitment deficiency,
where we have high turnover, so forth, we're not going to still need more people. So we will
actually concentrate our recruitment not across the country anymore, but we'll concentrate our
recruitment where we're closer to the margin and where we have an extra workload. And we'll
still have recruitment ads out there. We'll still be taking inquiries through our phone systems
and so forth - so yes.
The good news is that no single operation in the census thus far has not been fielded fully
staffed. But the next big operation is non-response follow-up, where we'd like to have as many as a half a million people start doing this work. And so I don't want to over-generalize
from the success thus far, because the serious challenge is the non-response follow-up period -
long answer, but I need you to understand exactly the pattern. But we will certainly continue to
recruit in those areas where we think we could conceivably need people, until we have finished
MR. JOST: We'll go to the phones next. Mark Skertic of the Chicago Sun Times. Are
you with us, Mark?
Q Good morning, Dr. Prewitt. Is the response rate on the long form - is there going to
be a problem with having a statistically sound sample? Are you going to - are you worried about
not having enough people to make the kind of calculations that you need?
MR. PREWITT: Mark, there are two issues here, one of which is - we're now talking
about the non-response universe - that is, the people who did not send the form in. And we will
go back to every one of those addresses which did not send a long form in, just like we go to
every address that did not send a short form in. So we have to ask ourselves: Will our conversion
rate be less successful for long form nonresponders than short form nonresponders? We don't
know that yet. We may be able to knock on the doors and convince people this information is
important. We may be able to close that gap before we finish non-response follow-up. But as of
now, we don't know. Answer one.
Answer two is, in terms of the statistical validity of long form data, should this gap persist
through the entire census experience, it will depend on two things: one, how geographically
clustered it is. The sample - that is, the long form sample, the 1 out of 6 - is designed in such a
way that we can give statistically reliable information for every community in the country of at
least 15,000 people. That's how this sample works. Below 15,000, we don't have a high enough
sample to make a statement about it. But any community at 15,000 or above, we will be able to
describe its characteristics using the long-form data.
However, if the drop-off is at 12 percent, we won't be at that - I don't know for sure what
the cut-off point will be, but it'll be somewhat higher. If, however, the drop-off in the long form
is geographically maldistributed, there could be some parts of the country for which we have
good data, and other parts of the country where we're really below the threshold for being able to
make reliable estimates.
Then the third level of this, the third problem of this, is what we call item non-response.
It's one thing to get a form in. As you know, there are Internet sites and public announcements
recommending to people that they only put down the number of people in the household. We
cannot take a form which only has that single piece of information on it, whether it's the long or
short form. If somebody sends us a form and says there are 23 people living in this house, and
give us no other information, we can't suddenly put "23" into the official numbers of the United
States, which drive apportionment and redistricting. These are very heavy responsibilities. And
so we have to make sure those 23 people are there, which means we're going to have to go back
and get some basic level of information to make certain they're there.
So, if indeed the instruction that you can see on the Internet, you can read in some of the
press releases of the groups concerned about long and short form were actually followed, then we
would have data deterioration that would be quite serious. We don't know that yet, but we will
be reporting that back as soon as we can.
But, as of now, the only real fact we have is that the overall drop-off in the long form is
twice what it was in 1990. That pulled the entire response rate down for the country, and that's
Q Do you think item nonresponse was raised by comments from people like Speaker
MR. PREWITT: I think that item nonresponse is certainly a result of describing the
census as a pick-and-choose exercise, because basically what you're saying is: Answer those
questions you want to answer. So any public statement that says that the census need not be
completely filled out, simply fill out what you want, will drive item nonresponse up just by
I should say the Census Bureau itself has said that. I have said many times publicly - say
again today - the first important responsibility of the decennial census is to do the count for
apportionment and redistricting. We have to have a basic amount of information to build the
count. After that, it's information that's used for other purposes, of all the public and private
sector investor purposes that we've discussed so many times. Those are the things that are at risk
with item nonresponse, not the count, if we get elementary information.
But if we only get a house count, then we've actually also put at risk the apportionment
number, the redistricting counts. Redistricting is slightly more complicated, of course, because
as you know, those public voices that have said, "You need not pay any attention to the race
item" - short or long form - are basically recommending, I guess - I'm not sure - that we repeal
the Civil Rights Act of 1964 and the Voting Rights Act of 1965, because those acts cannot be
enforced in the absence of high-quality census information about the racial makeup of the
society. Just a fact.
Q I'd like to follow-up, please, on that point. It may put you in hot water.
MR. PREWITT: I've been swimming there for months.
Q Given the fact redistricting is the key issue here, and that by and large it was
Republicans who have tried to undermine the census, were they, politically speaking, trying to
sacrifice the census for political gain?
MR. PREWITT: Oh, I'm not swimming in those waters, hot or cold!
MR. JOST: In the back of the room.
Q Dr. Prewitt, you talked about the effort on the long form being compromised and the
entire process being somewhat undermined. Do you have any sense at this point of how you
respond to that? Is it simply a matter of conversion rates at this point? And can you - do you
have any sense of what you do differently to respond to this compromised situation you find yourself in?
MR. PREWITT: Well, whatever has caused this, it has created a more difficult
operational task over the next 10 weeks. As I say, we're putting about a half million Americans
out there to do America's job. Those half million Americans are not professional survey-takers,
they are part-time people who have been recruited to do this task and many of them are very
committed to doing it well and doing it right. But nevertheless, they're going to get the basic
amount of training that we're able to give them, and we have to train them on lots and lots of
issues - and I'll talk about some of those next week.
To now put the extra burden on them of trying to convince the reluctant, the recalcitrant,
the angry respondent that, yes, the long-form data are important, puts an enormous burden on
people who did not sign up to do that. They signed up to do what they thought was an important
national task, which is to complete the count and get the information as best we can.
We simply don't have any information yet because we're not out on the streets yet. But
three weeks from now, we'll be out on the streets. We'll start getting back information about
how many people are simply being told: "I don't have to return it. I don't have to answer these
questions." And maybe very few will say that. The one thing one wants to keep in mind, of
course, and this goes back to these data, is a very small number of people can make a lot of noise
in this society. A very small number can make a lot of noise because it gets amplified through
talk shows, gets amplified in Internet chat rooms and so forth. We do not yet have systematic
information that a large portion of the American public is going to simply refuse to give this
information. All we know is they did not send it back in. But maybe we knock on the door,
they'll say, "Oh, well, it was just too long. I was waiting for you to come and help me, and I
can't wait to give you my long-form information."
MR. JOST: Okay, we'll go back to the phones. I want to warn you, we'll send a long
form to any reporter who takes a two-part question into a three-part territory.
MR. PREWITT: (Laughs.)
MR. JOST: We've got a long list here. And we'll go on the phone to Ariel Sabar of the
Q Hello. This is, I guess, sort of the flip side of the Lake Wobegon question. Do any
of the demographers there have any sense of why Massachusetts and Rhode Island did as well as
they did this time around, compared to 1990?
MR. PREWITT: No, I'm sorry, we simply haven't been able to do the work yet. As I
say, we were just getting these data ready for this event this morning and we have not done that
kind of work yet. We will be doing that over the next week or two, as best we can. Most of our
senior staff at the Suitland headquarters are very, very busy trying to make sure that the census
goes well. We do still have to go out and knock on about 44.5 million doors. I am not sure that is now the number. By the time we process everything, it could drop a little bit. But
it's somewhere in that 43 (million), 44 million doors; we have got to go out and knock on those
doors and find out why people did not send a form back in.
Getting that organized - getting the kits ready, making sure that these enumerators have
got the right address list and that when they get there and they knock on the door, they know that
that person did not answer and so forth and so on - is not a modest operational effort. So that's
what we're focused on right now.
It would be marvelous, I might say as a social scientist, to be spending my time trying to
explain the patterns in the response rate. But that we simply have to postpone until we've made
sure that all the operations are on schedule.
So again, I turn to the press. You have got about as much information as I have right
MR. JOST: Right in front here.
Q Yes. Did you or your office -
MR. JOST: Can you identify your -
Q Sure. Doug Turner, Buffalo News.
Did you or your office contact Governor Bush or Senator Lott after their remarks in-
(inaudible) - and, if so, what did you say?
MR. PREWITT: No, sir. As some of you will know, there was a Sense of the Senate
resolution that was introduced that reflected some of the sentiments about the fact that no one
should be punished, otherwise, or even harassed, if they failed to complete the census form. We
did explain to the Senate, or we at least sent a message to the Senate, what we thought would be
the implications of that and, I am pleased to say for those of you who followed this detailed
work, that Sense of the Senate resolution was withdrawn in conference committee. So we are
pleased at that.
It exists because, once you pass a Sense of the Senate resolution, it's available for people
to cite and quote. But, nevertheless, the fact that, on reflection, the Senate decided that that was
actually not the message it wanted to send, was a good sign. But, no, we did not do any direct
contacting of any particular political leaders.
Q Would you amplify on what those transactions were with the Senate and who you
MR. PREWITT: Oh, I simply wrote two small paragraphs saying what the Census Bureau's position was on this and simply sent it. I do believe - I think it was Senator Akaka who
read it into the record. So it's in the record.
MR. JOST: Okay, back to the telephones. Andrew Green with the Arkansas Democrat-
Q Hi. Andy Green, Arkansas Democrat-Gazette, two things quickly.
Since collecting long-form data is more time-consuming and it's complicated for
enumerators, particularly once we enter in the possibility of having to do non-response follow-up
for items that were left off, to what extent does the long-form differential mitigate the
better-than-expected mail-back rates at this point?
And, then, second, yes or no, has the bureau considered altering or augmenting the
training in any way, given to the enumerators, based on the heightened concerns for privacy that
have been observed, thus far?
MR. PREWITT: Let me do the second half first.
We actually spent a lot of time trying to decide whether we could modify our training
materials and largely decided we could not. You know, you are talking about 40,000 crew
leaders, who are being trained today, that go out and train the roughly half-million enumerators.
And to start trying to jiggle the materials; you can imagine what it's like to get training kits put
together and making sure they get to the right place at the right time. These 40,000 crew leaders,
by the way, are training in 40,000 different sites, not just in our offices. They're training in
churches. They're training in borrowed spaces. They're training in classrooms. Wherever they
can get the space they need to be doing this task, that's where they're doing it.
So, it was our decision that we could not risk trying to introduce new kinds of instructions
into what is already a very tight system.
We think that as the census non-response follow-up process unfolds, that crew leaders
and the enumerators will have experience, and they'll start talking about their experience among
themselves, in effect, and in that process, they will learn how to try to convert some of the non-responding long-form households.
We obviously thought about it. We care deeply about it. But you've got to balance - in
these operations, you've got to balance off what you might gain with what you might lose.
Then the first half of your question really was back on the issue, as I now recall the first
half of your question -
Q It was on whether the long form differential, given the difficulty - the relative
difficulty of collecting long form information -
MR. PREWITT: Yeah, surely. No, the fact that the country hit 65 percent, as I say, as of
today - it won't be the number that's posted tonight, but that's because the number posted
tonight is through Monday, but including yesterday we actually crept up enough to actually be at
65 percent right now - takes some of the operation workload off of us.
However, the degree to which that operational workload has changed - now going to be
increased because of the long form drop-off, we had planned and budgeted for roughly the same
difference - that is, roughly 6 percent. So the fact that it's twice that does, indeed, impose
operational burdens on us. I can't give you a precise calculus right now of whether those are
going to be a wash or whether, on balance, the overall improved response rate is going to be a
plus - it's a very complicated calculus.
As of now, we're simply proceeding with the system we have in place. We will find out
shortly - that is, when we're in the field - how nonresponsive the nonrespondents are.
We may run into short form problems. We may start knocking on doors and people who
even got the short form - and we say, "Look, it's going to take 10 minutes." They say, "Well, I
don't care. I'm just going to give you how many people live here." And they say, "There are six
of us," and close the door.
So, we simply are not in a position right now to know whether we'll have a different set
of experiences in the field with long-form nonrespondents.
MR. JOST: Okay. We'll take it over here to Chuck.
Q Dr. Prewitt, Chuck Holmes of Cox Newspapers. If the political and rhetorical
controversy hadn't occurred, do you think the 70 percent goal could have been met, and could
you characterize at all that goal in terms of what the Bureau could have done in maybe
responding quicker to this controversy? Is there any second-guessing on your part in terms of
what was done?
MR. PREWITT: Well, it's a fair question, Chuck.
The 70 percent goal was, as we know, an extraordinarily tough goal, because in terms of
our own internal models, it did not move the response rate up 5 percent, it moved the response
rate up 9 percent. Or thought differently, that's 25 percent of the nonresponders we wanted to
talk into behaving different from how we thought they would behave. That's a very, very major
So, I would be very hesitant to suggest that in the absence of any controversy that we
would have reached the 70 percent. I'm fairly confident that in the absence of the controversy we
would have bettered the 65 percent. And indeed, the numbers show that; we would already be at
66, a little over 66 percent, if we didn't have the differential. So there's no doubt that we would
have more strongly bettered it as against just met the 1990 response rate.
In terms of whether we could have done anything differently, it's very, very hard - these
things happen very fast. There was a lot of public attention - I did a lot of press over that three-
or four-day period; other people did as well. There were political leaders, there were certainly
I think the most important thing that happened in that period is that the local leadership
fought back. That is the mayors, the city councilmen, the townships, the tribal leaders - down
where they really knew they wanted that information and wanted a strong count, they simply
pushed back, and I think that helped enormously. I don't think the Census Bureau could have
done much at that stage, but the fact that hundreds and hundreds and hundreds of mayors and city
councilmen and county commissioners and some governors did say, "No, this is wrong" - and
editorial pages. So I don't think that the Bureau itself could have somehow done better than
what the kind of community itself did.
Honestly, the 65 is a serious accomplishment. We have a lot of internal models that show
that 61 percent was already going to be reached. The GAO, as recently as eight weeks ago, was
sending messages - sending a report to the United States Congress that we would not get 61
percent. And the fact that we got there and we kept going and we kept going and we kept going,
and we are still going, is, I think, a very, very positive story.
I would love to have had this census unfold in an environment of a complete commitment
to its importance as a civic moment. And it is, I think, regretful that at the key moment, the
March 27th to April 3rd period, when the nation's attention was on the census, which it was, as
these data show, the nation's attention was on the census, that in part they were hearing
something less than a full endorsement of what this means for our country. So that had to have
hurt; but how much it hurt in terms of percentages, I am very hesitant to try to characterize.
MR. JOST: Okay. I think we have got time for two more. We'll go back to the
telephones to Dilman Warner from the Times Picayune.
Q Hi. That's Coleman Warner --
MR. JOST: Coleman . All right.
Q -- with the Times Picayune in New Orleans.
Did your exact date of the posting of this material to the local jurisdictions, does it correspond now with 1990, since it was pushed back a day? And my bigger question is, does your ongoing survey shed some light on the reasoning that people have for not sending in their form?
MR. PREWITT: Yeah, quickly on that latter; the surveys are not being conducted by the Census Bureau. They are, as I say, being conducted by a private organization. I only have what they have - there are a few more questions than what I have shown here. But these survey data that are being shared with us, do not themselves try to address this question.
There may be other survey data out there in the country. We may pick up the newspaper
tomorrow and find out that some other survey organization, somebody else, has analyzed that
issue. But I don't have access to any information on that.
The first half of your question was - sorry?
Q The actual release date of --
MR. PREWITT: Oh, yes. The calendar in 1990 and 2000 is so different that you cannot
fine-tune that. What the 1990 65 (percent) number is, is really the very final number, after we
had done all of our processing, before we went into non-response follow-up period. We are not
quite there yet; that is, we will hope to be posting a new response-rate number some time in the
next two weeks, when we have finished all of that work, all of the assignments and so forth.
We are certainly experiencing a sharp drop-off. The questionnaires have been out there
now more than a month, and so we do not expect to be getting more in, although as I say, we
certainly did get enough in yesterday to change it from a 64 (percent) to a 65 (percent). We may
get more in today, tomorrow and so forth. And we certainly expect a drop-off this week.
But we will report it at some point. And it's possible that - we will not be less than 65
(percent), of course; and it's possible that we will even exceed that target. And that's why I say
you want to pay particular attention to jurisdictions that are very close to the border; that is,
within 1 percent of either meeting 61 (percent) or meeting 65 (percent) or meeting HAKWAN,
meeting 60 (percent) or whatever their 2000 goal was - be very attentive to those, if they're
within a point or so because that could change over the next week.
MR. JOST: Okay, last question in the back of the room.
Q Yeah, Richard Powelson, Scripps Howard News Service. What do you think of the local communities which have been paying the procrastinators to come in and finally fill out their forms? Five dollars or so is what they're paying.
MR. PREWITT: The local communities did a remarkable thing. And some of you
covered some of those stories very well, I might say. But if you actually looked across the
country and at the way in which there were festivals, there were parades, and there was special
advertising, there was investment of local money, state money - you know, we just saw very
good return rates from California; that happens to be one of the states that put a lot of its own
energy in. Now, I cannot say that's the reason, but certainly it is a pattern. And we will try to do
a bit of that work, that is in terms of those areas which particularly worked hard at boosting their
response rate and seeing if it had an effect.
The particular one that you mention, I believe that was in - I forget - Georgia? - Georgia
- I did read that in the press. I only saw what I saw in the press. Every local community should
do what it feels best it should do. I personally don't want to turn the census, in general, into a
"We'll pay you to do it" exercise.
I want to go back to where I started. I think that democracy is about obligation and not
just benefits; it's about responsibility and not just rights. And the census is a measure of our
commitment to a democracy that respects the importance of responsibilities and obligations.
And I am personally very pleased that as many as 65 percent of the households in the United
States did respond to that, and I think it should be celebrated. Obviously, it would have been
nice to have it higher, but it's not an insignificant accomplishment; that whatever the
environment was, that nearly two-thirds of the households in the United States set aside the time,
made the commitment, and did what, after all, the Constitution says they should do.
The Framers, I remind you, did not say that the census was just to be a head count. The
short form in 1790 - it wasn't the short form because there was no long form - but the census
questionnaires, what the marshals were actually recording in 1790, looks an awful lot like the
short form in 2000. It asked household composition, it asked gender, it asked age, it asked race.
The only questions on the short form in 2000 which were not asked in 1790 were ethnicity and
whether you own or rent. So the Founders, in 1790, wanted the census to be more than just a
head count, they wanted it as an instrument to know something important about the country.
So when people tell me, "Go back to the Constitution and just do a head count," I say to
them: Go back to the Constitution and see what Thomas Jefferson and George Washington put
in place. George Washington was the president, Thomas Jefferson was the Secretary of State;
they were both in Philadelphia in 1787 when they put the census into the Constitution. They
knew what they were doing. Indeed, the person who lost - sorry, I'm now slipping into my
professorial role - (laughter). Nevertheless - forgive me. James Madison, when they were
planning the 1790 Census, the first decennial census ever, James Madison argued very strongly
for collecting more information. He said, this is an opportunity to really measure economic
behavior, manufacturing and so forth. And he did not prevail. They said, no, it's only the
minimum. By 1820, James Madison's argument had prevailed. Starting in 1810 and then again
1820, many, many more questions were put on the census form.
So, the Founders, the people who wrote the Constitution, had in mind that this society
could govern itself better on the basis of quality information.
Just one other thing. Information, high-quality information about government programs
is not just for the government to administer those programs, it's also for the American public to
find out if the programs are being administered well. If we don't have information on reading
scores, if we don't have information on poverty rates, if we don't have information on health, if
we don't have information, broad-gauged information about the quality of what the government
is doing, how in the world do we punish them at election time? The American public constantly
uses social indicators to make a judgment about who is performing well and not performing well.
So, information is a two-way traffic in a democracy. It's to help the government know
what the people need, but it's to help the people to know whether they're getting what they need, because otherwise it's all anecdotes. And anecdotes aren't good enough, as far as I am
concerned, to run a serious democracy.
So, I think at this moment, on the basis of this response rate, we should actually, as a
country, feel good. We should celebrate the fact that we stopped what is, after all, a decline in
civic engagement that starts in the 1960s, and you can see it in all kinds of trend lines, and we
stopped it in its tracks, and that is no modest accomplishment.
MR. JOST: Thank you very much. We'll see you next Tuesday.
[END OF EVENT.]