Press Briefing -- May 16, 2000
|MR. STEVE JOST: -- We are here today for another in our series of operational press
briefings to talk about the status of Census 2000. And today we're going to talk to you about the
status of where we are in non-response follow-up, and also elaborate a bit on the quality control
measures we have in place to ensure the count is as accurate as possible.
The ground rules, as most of you know, are that we ask if you have a question after the
director's opening remarks, that you identify yourself and your affiliation. And we'll alternate
between the room and reporters on the telephone from around the country.
And with that, I give you Director Kenneth Prewitt.
DR. PREWITT: Thank you. I will begin with just a very few operational updates and
then turn to the major issue; that is, quality control. We have now concluded our list/enumerate
efforts, in which we targeted 343,000 housing units in remote, sparsely populated regions. About
5,000 enumerators were assigned to this task, and that's now complete.
We've completed 99.9 percent of our group quarters enumeration. Eleven thousand
enumerators visited about 200,000 college dorms, hospitals, nursing homes and prisons. And that
is less than 100 percent. It will probably be less than 100 percent throughout the entire process
of the census because we continually find new group quarters. Once in a while one of our
enumerators will go to an address which we think is just in our non-response universe, knock on
the door, find out that it's a group quarter, and then we treat it that way. So in effect, this is a
situation where the denominator keeps growing and that's why we're always going to be slightly
under 100 percent until we're completely finished.
A few other numbers. More than 2-.25 million language forms were mailed out, of which
about 1.9 million were in Spanish and the remainder in, of course, the other four languages --
Chinese, Korean, Tagalog and Vietnamese.
In our recruitment efforts to date, we have hired 26,000 non-citizens. And another number that we're especially proud of, more than 27,000 welfare recipients who are part of the national Welfare-to-Work program are now employed by the Census Bureau. And we will be working closely, of course, with Goodwill Industries, our major partner in this effort, and we expect a very high proportion of these 27,000 to move into the permanent workforce at the conclusion of their work with the Census Bureau.
This, of course, is way over target. We had hoped to have about 10,000 and so, at 27,000
-- and we will still add some in the final versions -- we could easily triple our goal in that
Now let me then turn quickly to the Non-Response Follow-Up situation. And I'll refer at
times to this chart, and it's also in a condensed version in your handout. As you can see, of
course, it's a useful chart to get some idea of where we are right now operationally. The census
starts with the original households in the universe, 120 million. And, of course, we move then 78
million of those into our housing unit mail-back respondents -- that is, the 66 percent, and still
growing slightly, of our mail-back response rate. Then, the remaining 42 million households, of
course, move into our Non-Response Follow-Up universe. And I want to speak primarily about
that today, and then some of the other things, as well, that are going on.
This is the most labor-intensive phase of the census. It began two weeks ago and will run
through several distinct stages, really, until the end of the summer in terms of all of the processes
that have to be completed. We have successfully recruited and trained our local census workers,
who are now, of course, doing this task. And as you know, last Friday we reported that we had
crossed the 30 percent mark in completing the basic Non-Response Follow-Up, door-to-door
operation. Today with still seven weeks to go, I can report that we're at 39 percent.
Now, 39 percent is for us a good number. On the other hand, as we well know, the earlier
weeks are the easier weeks. The rate of productivity will slow down as we reach the harder
cases. Nevertheless, it is very reassuring that we're getting the level of cooperation right now
with the American public that allows us to be already at 39 percent of our non-response
Quickly, on some of the quality assurance issues, the first stage in this is to carve out of
the universe of housing units -- that is, the non-response housing units -- and load this, take the
non-response out of the overall universe of housing units and load the data files for each one of
the appropriate 520 local census offices. Census field maps were printed with boundaries drawn
down to the level of crew leader districts, address registers, assignment directories and envelopes
were printed, individual labels were printed for the predesignated short and long forms. That is,
all 42 million of the non-respondents in the -- what we call NRFU universe - had to be specially
treated at the level of the local offices. Each address was identified according to the appropriate
crew leader district, about 40,000 of those crew leader districts, and then loaded onto our
automated operations control system.
This system, stationed in our 12 regional census centers and our local offices, is the nerve
center of our quality assurance program. It tracks the status of every one of the 42 million
non-response addresses. The assignments given to the enumerators in all cases selected for
quality control reinterview, which I'll speak more about in a moment.
As you know, enumerators must make up to six attempts to contact someone at each of
these addresses and complete the census questionnaire. Enumerators must perform their
assignments correctly, and we must be certain that they are doing so. This is why there are
multiple safeguards for quality control that have been established within our local census office
structure. This quality control starts with a daily review by the crew leaders of enumerator
questionnaires that are submitted each day. If a questionnaire fails this review, it is returned to
the enumerator for correction. At the same time the crew leader determines that the address
registers and maps have been correctly annotated and updated. Each day the crew leader then
delivers the completed questionnaires that pass his or her check to the field operations
supervisor. And, at that level, they are rechecked before they are sent on to the local office.
At the local office, there's a quality control operation with a clerical operation, control
clerks who key information from each form -- that is, not the actual data from the form, but the
identification numbers on the form -- into the master operations control system database. And
this automated check-in system that did begin on schedule on May 1st, then reviews whether the
questionnaire meets our quality standards. If not, they're passed back down to the crew leader
But, if they do pass, as most of them of course do, they not only go through this
automated edit, but that automated edit also has a quality control process, where we key things
more than once -- to make sure that that there are no discrepancies. That is, are there data for
the population count at this address, are there data for the status of the housing unit, the key
things that are part of our control file? If unoccupied, there must be an ID that identifies it as
vacant. And, if nonexistent, it must be deleted.
Just quickly on that: so you're now in this world, of course, and a lot of the forms come in
and they say, "This is a vacant unit," or a unit that's been burnt down or "it's no longer there" or
"we can't find it." It then goes into the "vacant/delete" box, which then, in turn, of course, goes
into a different process, which I will describe in a moment.
Any of those that fail this test at the LCO level are then returned to the enumerator. The
address binders are also checked in at the LCO level. And then, of course, we also have a
number of processes to make sure we check on the quality of the work that the enumerators did.
That is, every enumerator, when they are trained -- some of you have visited the training
sessions; you know this -- have been told that we will spot-check the quality of their work by
assigning 5 percent of their workload for reinterview by separate staff, who then conduct
telephone and personal visits for those housing units. These are selected at random, of course.
So there's a constant daily 5 percent quality check of all of the interviewer work.
In addition, every week each enumerator's work record is compared, using performance
indicators, with those of other enumerators. What this is, in effect, is sort of an algorithm that
says if you get a certain pattern in the data that seems inconsistent with what you're getting from
the other enumerators in that region, then that person's work sticks out, and we take a hard look
at it clerically, to make sure that it's not any kind of unexpected pattern that would signal to us
that we have enumerator problems. And then, of course, we also have quality control checks to
-- for any suspicion of falsification -- what we call "curb-stoning." And those processes then
also are both done by algorithms, but they're also done by simply supervisors looking at the
material and trying to get a sense of whether we might have falsification. Of course, anyone who
is found falsifying data -- all of their work is redone and, of course, they're immediately
Obviously, after the material moves into the local office and it's moved through these
early checks, quite extensive checks, it then gets boxed and sent to our data capture centers. And
there we, of course, have another set of quality assurance processes; they've been described
before, because they were, of course, in effect during our mail-back process.
I should say, about our quality control processes at data capture, they continue to run at a
very, very high level. We're at 99.5 percent quality-control-check accuracy.
All of our mail-back census has now been scanned. There is no backlog in our mail
census forms at the data-capture centers. And about 3.5 million of our enumerated forms have
already been scanned. And this is an important test for us because we are moving from one set of
interviewer forms to a different set of interviewer forms because the enumerated forms are
slightly different from the mail-back forms. And the fact that 3.5 million have already moved
through our data-scanning very, very smoothly without a hitch suggests that we will continue to
experience the same high levels of accuracy at the data capture centers as we have already
One small operation that isn't so small but still important -- it's not even on this chart --
is a Coverage Edit Follow-Up; that is, in this universe of respondents, that is, those who mailed it
back, we have a process by which we check for missing, incomplete or inconsistent data.
This we do on the telephone. We call back any form in which, say, they said that there
were five people at this household but they only listed four; we then call back. We call back all
households which say there are six people in this household because we are concerned there is
only room, as you know on the short form or the long form, for six individuals. And if they say
there are six, we always call back to make sure that there aren't some others that they simply
didn't give us the information for. So we do all those checkouts.
And then we also, of course, look at most of the households who have a large number of
unrelated individuals, nonrelatives, in that household. And we do that again as a double-check.
So that coverage -- we call that Coverage Edit Follow-Up. That's undergoing -- that is, being
conducted right now by telephone from one of our 13 contractor-operated call centers. Those of
course are also randomly monitored for quality control, not only for what the agents say on the
phone, but also for what they enter into their computer screens. We expect this work to be done
by mid to late June. It is proceeding now on schedule.
Then, very quickly, because we'll talk more about this as we get closer to it, but there is a
box that you haven't seen yet. It's on the far left here -- my left, your right -- "coverage
improvement follow-up." That's a process; that's a field process. It starts in late July, and we
expect to complete it by mid-August. It's really a crucial field step in our overall quality
assurance program. Its purpose is to ensure that essential information was collected during Non-Response Follow-Up and that address binders or maps were completed accurately.
But it also provides follow-up to verify the status of housing units -- that previously
identified housing units in earlier operations as either "vacant" or "deleted." And as you see, all
of the "vacant/deletes" move from here into there, which means they get another field visit to
make sure that the enumerator hasn't inadvertently described something as either "vacant" or
"deleted." And we subject all vacancies or deletes to what we call the "double kill." We don't
just take the testimony of one person; we make sure that at least two different sources have
identified that as "vacant." It has to do with our intensive effort to make sure we do count
everyone, we do expand our coverage effort as fully as possible.
Also, in that box -- it's called CIFU -- Coverage Improvement Follow-Up -- the other
thing that goes into this is our New Construction Program and the update/leave adds. During
update/leave, we sometimes picked up some addresses; they go in there. And also we continue
to get, from the Post Office, any new addresses. So we put that box together, and then that gets
treated just like it was a Non-Response Follow-Up universe. The only difference is that it really
has to do with a housing unit that, for one reason or another, was not part of our initial universe.
We expect there to be, all told, about 7 million housing units, as we see here in that box,
subjected to that kind of field effort. And, of course, that field effort is then, in turn, subjected to
all the quality control processes that I've already identified for the Non-Response Follow-Up
And then finally, and again not to say very much about it, because we'll have an
opportunity to talk about it in greater detail, the largest, of course, quality control program is our
Accuracy and Coverage Evaluation, which is our A.C.E., which goes back and, in a way,
operates as a quality envelope across the entire census operation.
And, as you can see from this chart, it's possible to miss persons at various stages of the
census. And, indeed, the most intelligent way to think about the Accuracy and Coverage
Evaluation is that it's looking for people who fall into any one of these peach-colored boxes or
into the peach-colored housing unit. That is, you can miss a housing unit completely. And we
would expect the Accuracy and Coverage Evaluation to pick that up. After all, we're now in a
much more intensive survey mode as against a census mode. Or we can miss persons. We can
get a respondent who sent back household information but it left some people off. We've talked
before about the undercount of the children. Many of those are children who are left off the form
even though the response comes back in the mail. So they would have appeared in this box. But,
then, in all of our processes, in our Non-Response Follow-Up work, we can also find that people
have been missed and, of course, in our CIFU work people who have been missed.
So, there are four different categories where people get missed or potentially missed in
the census operation, and our Accuracy and Coverage Evaluation survey is designed, of course,
to go find them. And that's what creates our measure, finally, of the undercount and the
overcount. I say not only missed them, but also erroneously enumerated (them). These are the
people who are counted more than once or miscounted in some other fashion.
So we can talk in detail about this. We have a number of quality control operations at
every stage of the operation. The ones that were operating during our original mail-back process
and our data capture all worked very, very smoothly and accurately. We're quite pleased with
that. And we expect the same result now as we're doing the more complicated quality-control
process of a major field operation, which is our Non-Response Follow-Up. The fact that we
already have 40 percent, or nearly 40 percent, of our Non-Response Follow-Up workload
completed is also a sign to us that we've now field-tested and run our quality control procedures
long enough or hard enough in our process that we're now convinced that they are working very
So, with that, I will turn to your questions.
STEVE JOST: Okay. Just a reminder for the reporters with us by telephone, if you want
to get into the queue, you need to push "1" on your telephone. We'll start here in the room, in
the back. Pam Fessler.
Q Hi. Dr. Prewitt, have you been able to determine yet how many of the mailed-back
forms were people who only put their names down and did not put all the detailed information?
That was something that you were going to try and determine.
DR. PREWITT: Yeah. That's what this Coverage Edit Follow-Up process is doing right
now. That is, some of those questionnaires go into that box, in effect. We don't have enough
information to know whether this is a legitimate questionnaire or not. So we're doing that work.
And I don't have a number on it yet. If someone simply put down a name, we're now scanning
that. That is going to fall out. It's not going to have enough information for us to put it into our
Q Do you have any sense if it was bigger than, you know, it had been in the past?
DR. PREWITT: Pam, I'm still hesitant because there's so much detail in that question
the way you're asking. And I appreciate what you're trying to learn. And I don't think we're
secure enough in wanting to say that we have a higher-than-expected rate of basically blank
forms, is what you're asking. They came back in. We scanned them in as forms, or we scanned
the envelope in, so they come into this category, and then we will pull them out, begin to look at
them. And I don't know how many have gone into our coverage. We're still doing that, is the
point. See, we just finished this scanning work just -- actually last Friday, when I say we
completed that work. So I don't know how many have been now moved into the next process.
STAFF: Okay. With us by telephone is Mark Skertic of the Chicago Sun-Times.
Q Good morning, Dr. Prewitt. Has any effort been made yet to assess the effectiveness
of the different regional offices? I'm wondering if low response rates or problems in the field
have been traced back to any problems in particular regional offices.
DR. PREWITT: Well, when I said that we had finished 39 percent of our workload thus
far, that's a national number, of course, and there's variation around that. We have very strong
regional offices right now in terms of their productivity level, and then weaker ones.
The weaker ones, however, are not out of bounds. That is, none of our regional offices
are we -- Chicago, for example, happens to be the lowest of our regional offices right now, 31
percent. Our highest is Denver, at 53 percent. But we're doing extra work right now in Chicago.
We don't see that 31 percent as...it's a strong signal to us to do some additional work there, but
it's not yet in the danger zone.
There are certainly local offices, a small number of local offices where they're below
where we would like to have them in productivity. That's almost always an issue of local
management, not really -- thus far, not really of issues of where we really think we're getting
deep resistance across the population.
So, yeah, there's a lot of variation around a national number, but, as of now, we are
adequately staffed to continue this work. We're getting levels of cooperation we expect from the
American people. So we're nervous about areas which are on the low end of that, but we're not
in an emergency mode.
Q: Thank you.
MR. JOST: Okay. Back of the room here. Hi.
Q: Just to be clear on the blank forms that may be coming, or at least partial blanks and
the long form. So you would be starting to pick up some of this through the current quality
control check. That would be the first attempt to try to get people to answer the blanks?
DR. PREWITT: That is true. What I quickly tried to described, called our Coverage Edit
Follow-Up, which is taking cases out of our mail-back respondent universe. The cases that we
follow up on, the problem is they're composed of about three or four different subsets. And
that's why I didn't have a direct answer to Pam's question.
One of the subsets is any form that came in and said that there were six people here,
because we just automatically go back into the field on that one to make sure that we didn't miss
people in that household because the form itself didn't allow them to put down the seventh and
eighth or ninth person. Any form which has a large number of unrelated people in the household,
because previous experience tells us those tend to be the kind of households that don't count
everyone. So we go back and try to make sure that we get anyone else.
Also, any forms which say there are six people here, but they only give us three pieces of
information, or they say there're three, but they give us six. Any inconsistency between the
number of people in the household and the number of people who appear on the form goes into
that universe. But also, then, anyone who simply comes in and has insufficient information to
classify that as a respondent. And what I was trying to fumble with Pam's question and realized
that I didn't even have it is what proportion of this total universe right in the Coverage Edit
Follow-Up consists of those cases. But those would be cases where people have simply given us
a number and no other information, or just given us a name and no other information. And I'm
sorry I don't have -- I think I don't have it because we did not have it, or they would have
prepared it for me, knowing this would have been a question. And so it may be another week or
two before we've really processed all of the respondents to the point at which we can answer that
That would be our first clue. You're right. That would be our first strong clue of any
serious problem out there in the field with respect to the mail-back questionnaires.
Q: So just to follow up, how many actual contacts could there be, maximum number of
contacts with the household if you're lucky, or lucky enough to follow every one of these quality
DR. PREWITT: It could be as many as 11. That would be an extraordinarily unusual
occurrence. But if someone gave us partial information and we went back and got information if
they were -- or if they were one of those cases, which I've already described before, whose form
came in too late to go into the non-response, and we knock on their door. Then they got quality
checked. Or we thought the enumerator was falsifying, so they fell into that universe. It turns
out, if you actually add every possibility up, a household could be contacted -- I could be slightly
off, but not way off on that number. As I say that's an extraordinarily unusual occurrence.
On the other hand, it does tell you that, you know, it would be a reasonable number of
people that we contacted two or three times.
Now, we don't see that as a negative. Obviously, the household who is contacted that
many times will, unless they're very lonely people, will begin to feel bothered and will express
the fact that they're a bit bothered. "Why can't these people get it right?" will be their question.
Well, the only way we can get the entire system right is to do a lot of quality control. And quality
control necessarily means you've got to go back and check on work. And as you go back and
check on work, the same people are going to be bothered again.
These quality control checks are very, very short in duration. That is, "Were you
interviewed? Did you send your form in? Somebody said that you were interviewed. Were you
actually interviewed there? And did you actually say there were four people here?" I mean it
takes a minute, a minute and a half for most of these phone calls, we're not redoing the whole
form. We're simply making sure that they were interviewed.
But I have to say that the Census Bureau did two things in 2000 that made it different
from other censuses. One, it simply expanded its effort to get everybody counted, which means
that there are duplicates. That means we're going to double-count; not double-count, but
double-visit. We've done everything we can to make certain that everybody has a chance of
being included in this census. And when you do that, by definition, at the edges there's going to
be 2 to 3 percent of the population which is going to be swept in more than once. There's no
other way to do it. And since we started out with that as our goal, we now have to recognize that
we're going to bother some people.
Secondly, the census wants to make certain that every form that we have in a census
record is the right person, who actually is the person who does live there, and so forth. Which
means we put in lots of quality-control processes. So you put those two things together, so a
particular household that falls both into the universe of people who got swept in more than once,
and also happens to be a household that's getting a series of the quality processes, that's how you
would get up to that extraordinarily high number. So, as I say, it won't happen very often.
But I would like to say and put on the record that as many as 2, 2.5, 3 percent of the
American public is very likely to be contacted, you know, more than once. A.C.E. itself is
one-half of 1 percent, 300,000 households. So that immediately says that. Quality control is
another 5 percent of all of the NRFU respondent universe. And so you can do the arithmetic and
it's hard to find out that literally 2, 2.5 million, maybe 3 million households will get more than
We hope the American people will understand why that is necessary in order to make
certain that we have a high-quality census.
MR. JOST: Okay. We'll go back to the phones. Paul Johnson of the Bergen County
Record News. Paul, are you with us?
Q: Yes. First, I want to know what you guys' response has been from areas that were
undercounted in 1990? Have you guys had a good response rate from there? And do you expect
the work to be completed early or on time, the non-response follow-up?
DR. PREWITT: With respect to your first question, the overall pattern across the country
suggested that our mail-back response in some of the more undercounted areas from 1990 did
better. That is, they came closer to meeting either the 1990 or exceeding the 1990 rate than some
of the areas which had higher mail-back responses, higher counts in 2000. So, if anything, we
have somewhat leveled the playing field with respect to the undercounted areas as a general trend
across the country.
It's much too early to say whether we will finish the work, on what schedule we will
finish our Non-Response Follow-Up. Certainly to date, we are on schedule. That does not mean
that starting tomorrow we won't have serious problems. But the serious question that has to
constantly be asked is, at what level do we get the cooperation of the American public? And if
the American public continues to cooperate with us in the next seven weeks at the rate they have
in the first two weeks, we will certainly complete on schedule. If they don't, then we will be
struggling longer than we had hoped. But the important news is that we have the work staff to do
the work. We're not now experiencing any sort of slowdown in schedule because we don't have
workers. In some instances, we have to change our management to make sure that people
understand the urgency of getting this done at the pace at which we have scheduled it.
Q: Okay. Thanks.
DR. PREWITT: Sure.
MR. JOST: In the room. Is your hand up there, Ira?
Q: Ira Teinowitz from Advertising Age.
Give us sort of an overall view of where you are right now compared with 10 years ago. Are you ahead of where you were 10 years ago both in the general population and the specific population that you were trying to reach or are you behind at this point?
DR. PREWITT: We're seriously ahead of where we were in 1990. At this stage in 1990,
our field work suffered from an inadequate workforce. We were losing people at a rate higher
than we were hiring, and so we were simply not as well equipped to be doing our major field
work. We obviously were -- we were at the same place in 2000 as we were in 1990 with respect
to the percentage of the households revisited, since we ended up very close to the same number
of mail back of 65, slightly better in 2000 than that. But since we had anticipated it only being
61, we're certainly ahead of our target in that sense.
Right now this census, operationally, is undoubtedly in better shape than the 1990 census.
That is true of our mail-back process. That's true of our launching of our Non-Response Follow-Up. That's true of our quality control processes. That's certainly true of our data capture. That
is, right now this is operationally a much stronger census. However, as we've said so many times
before, we are dealing with attitudes of cooperation. We're dealing with patterns of resistance.
We're dealing with people who are very busy, who are on the go, who are highly mobile. We're
dealing with a demography and set of public attitudes which makes it more difficult to do a
census in 2000 than in 1990.
So, at the end of the day, I can't say the measure of success will not be measured by our
operations. Our operations are doing the task that we've assigned them to do. The measure of
success will be the degree to which the American public will cooperate with this enterprise.
Q: Is that across-the-board, or is it both in the general population and separately in
the -- ?
DR. PREWITT: Well, at the very end of the day, we'll have an answer to that question.
But as of now, we are getting higher levels of cooperation in some of the more -- as I said earlier
- in some of the more difficult-to-count populations than we experienced in 1990. That was
why I say the slight leveling of the response rate. That is, we had an extremely active effort by
many of the community groups that work with minorities, that work with new citizens, work with
immigrants, an extraordinary kind of partnership effort in those communities, which certainly
made a difference in the mail-back process. Whether it will continue to make a difference during
this process, it's simply, again, too early to know.
I would say with respect to advertising, that as we've reported before, there is no doubt
but what the advertising campaign was a success measured by its capacity to reach deeply into
the entire American population, levels of awareness, understanding of the census. It was a very
serious success. It certainly motivated some people who otherwise might not have filled out the
form. Whether it will have that effect right through to the end point, we don't know. We're
knocking on doors every day. It depends on what happens when we knock on those doors. As I
say, 39 percent have been answered thus far and/or have been vacant or delete, I should say. That
is, they've moved from this universe into a different universe. They're no longer non-respondent
households. They're not respondent in some respects. On the other hand, there's 51 percent to
go. No, 61 percent to go. Sorry. Please don't quote me.
MR. JOST: We'll go back to the telephone. Butch John of the Courier-Journal.
Q: Yeah. I apologize. We had a bad connection earlier. You're mentioning 39 percent,
and I'm just trying to put this into some sort of context. Can you help me a little bit with just the
basic figures and where you are compared to where you should be, or want to be, or plan to be?
DR. PREWITT: Surely. The 39 percent is the percent of what we call our Non-Response Follow-Up universe, which were 42 million households after we finished the mailback
process. So it's 39 percent of that population group. In terms of where it is, it is basically on
schedule. I can't be too enthusiastic about it because I know the hard work is yet in front of us. I
do know that if it were much below that, we would be concerned. But it certainly is at a level
where we would like to be at this stage.
In terms of where we would like to get, we obviously want to count 100 percent of the
American population, which means we want to find someone at home at each one of those 61
percent of the non-responding households, find them, get them to cooperate with us and so forth.
That's our goal. But indeed our goal in the mail-back response rate was to get 100 percent back
too, but it wasn't a goal that we thought we would reach. But nevertheless, we would love to
have had no Non-Response Follow-Up effort at all in this census. But, as you all know, 65
percent of the questionnaires did come back in. That left 42 million, and that's what the chart
So, to put this in context, this is a strong indicator of a robust operation that is now
functioning the way it was designed to function. And this is very reassuring to the people
who've been concerned about whether this Census Bureau will be able to pull off this census. At
every stage, we have pulled off the major operation we were expected to have completed, and we
did it always on schedule and we did it at budget, and that continues to be our experience. I
cannot promise that will be the experience right through it. But, as of now, our experience is that
our operations are working successfully.
MR. JOST: We'll go to Dee (Cohn) up front, and then the gentleman in the back, but
after the phone call.
Q: Dee Cohn, Washington Post.
Can you say anything about turnover among enumerators, and also about threats or
injuries involving census-takers?
DR. PREWITT: Right. Quickly on turnover, it actually turns out to be a difficult number
to get, because some of the turnover are people that we're letting go because we're not
comfortable with their productivity. As of now, turnover is not an issue for us, which is to say
any of the places where our productivity is lower than we would hope it to be is not because of
absence of staff. We are continuing to be fully staffed for the workload that we have designed
for ourselves. So it's nothing out of balance right now. We do continue to do replacement
training, and the replacement training, where necessary, well, then we'll fill in behind any
unexpected higher levels than expected of turnover attrition.
More specifically, I guess, Dee, I would say that the point is -- can we maintain the
productivity schedule that we have set for ourselves with our current workforce and anticipated
workforce? And the answer to that is, absolutely yes.
With respect to threats, it's still anecdotal. It's going to be anecdotal. It's just very, very
hard to get this information. There's no doubt that there have been assaults on some of our
census workers. You can read this in the press, of course. We do not believe that it's
extraordinary. That is, if you actually look at the proportion of workplace assaults across all
kinds of different workplace settings, it does not look that our rates are exceptionally high. For
example, the most recent data from the crime victimization study suggests that something like
1.5 percent of the work force experiences some kind of workplace-related assault. Well, that, for
us, would mean 1.5 percent, say, of our 500,000. It would be 75 cases of assault. We have not
had 75 cases of assault. We may have them. But we certainly haven't had them.
So right now we're underneath the norm for the overall workplace. Indeed, the Postal
Service last year investigated about 172 assaults on postal workers. Well, we're not at that rate
yet. We take every one of them very seriously and immediately work with the local authorities,
and we have had instances. But the pattern does not appear to be anything outside of the bounds
of what one would expect in American society today with its levels of criminal assault.
We take each one of them carefully. We collect as much information as we can. They
come in -- you know, they're alleged. We don't have the facts. We turn them over
immediately to the local police. They then do the investigation. But the good news is that our
big first experience is the American people are cooperating, and that we just hope that will
sustain itself as we get into the more difficult-to-count households.
Q: I'd like to follow up on that. You don't have 75, but roughly how many?
DR. PREWITT: Well, again, it's alleged, and that's why I'm nervous about putting too
firm of a number on it. But it's under a third of that. I'll put it that way. It's less than a third of
that number. And, as I say, it depends on what you call an assault, what you call an act, whether
it's verbal, whether it turns into physical. But the number of cases that I actually could tick off if
I listed in my mind, the ones that have been reported to me where I'm fairly clear that it's
happened, is closer to a dozen. But that doesn't mean there're not another dozen in the system.
I'm only saying that it's not --
MR. JOST: Well, there is a definitional issue, and that is the National Crime
Victimization Study is about crimes reported to the local police in 1996. And filing a complaint
with the local police is at the enumerator's discretion. I mean if they want to file a claim, it's not
a judgment we make, although we support them when they do want to make that. So we want to
be careful we're talking about the same kind of thing here.
Q: Well, how many have been filed with police?
MR. JOST: I don't think we know the answer to that either. We don't -- here's an issue.
I mean the Postal Service tracks this as a matter of course. And they've informed us that they
had 176 cases last year of threats against postal employees, assaults against postal employees.
Excuse me. But they have a permanent workforce, and they regularly report this year-round,
whereas we have a temporary workforce of part-timers who are working for a few weeks at a
time, and it's very difficult for us to have in place a massive administrative system to keep track
of it. But you know, anecdotally, we could tell you the numbers that the director reported.
Okay. With that, we'll go to Rick Klein of the Dallas Morning News, and then the
gentleman in the back of the room. Are you with us, Rick?
Q: Yeah, I'm here. I'm going in and out. My question has actually been answered. Thank you.
MR. JOST: Okay. The gentleman in the back.
Q: Yes, Frank James with the Chicago Tribune.
Director Prewitt, you alluded to management changes that had been made in places where
you didn't feel there was a sufficient sense of urgency to get the work done. I was wondering if
you could give us some more details on that. And, being that I represent the Chicago Tribune,
have any of those changes taken place in Chicago?
DR. PREWITT: You know, Frank, I appreciate your question. I was afraid you were
going to ask me a different question altogether from the last time we met. Right, Frank?
Certainly some of those have occurred in the Chicago region, yes. And I cannot give you
many more facts on that. I think I'd rather make sure you talk directly to Stan Moore, because
when you're looking at 520 instances across the country, I just don't have in my mind exactly
what the pattern is. But certainly in the Chicago area, [it's] been one of the areas where we've
made some management changes. But on the other hand, I think we've made management
changes. When I say "management changes," that could well have been a recruitment supervisor;
it could well have been an area manager; it could well have been an LCO manager. There's lots
of different people. There're about four or five recruitment roles per LCO, so the total number is
about 2,000. Many of these people were brought on, of course, without prior experience with the
census. We have an experience with them. They find it wasn't quite what they wanted. So some
of these people have left voluntarily. Some of them actually have left for other jobs, just like in
So we don't think this is extraordinary. The important thing is we actually take very
quick action, and it creates a problem for us, because we then get a lot of other kinds of
complaints. But the census is on a very fast pace. And if you've got an LCO that's not doing the
job that we think has to be done and they say, "Well, give me another chance, and I'll clean this
up in the next week or 10 days; I'll do better, blah, blah, blah," we can't sit around often and wait
for people like you can in other work situations. It's a very fast-moving thing. We make
decisions every day. We get these data back every day. And so we have no choice but to be
fairly determined to create the kind of management structure that can get the job done.
So that's the policy in terms of, actually, the distribution of cases by region. I don't have
it partly because, again, you're moving -- your denominator's moving fast. And many of our
partnership specialists, for example, have already been now released from work because most of
that partnership work was part of our mail-back period and not part of our Non-Response
Follow-Up. So there's a constant flow through our management. Sometimes we're moving
people from one job to another because they weren't doing the right job in this (one), but we
really still think they're a good worker, so we moved them into a different task.
I appreciate that's not very responsive. But, for the details, I would really have to prefer
that you go to the regional director.
MR. JOST: We have a question back here.
Q: Yes. Doug Peterson, Nation's Cities Weekly. What's the split in the 120 million and
the 42 million long form-short form? And of the 39 percent that you got back, is there a
difference there? And on the calendar, is the March 30th date, 2001, the first time cities would
see any counts for their jurisdictions?
DR. PREWITT: With respect to the last question, yes. March 31st, or slightly earlier, we
will release the data by city level and by district level, all the way down to the block level, as a
matter of fact, on a flow basis. But that will start in March. So by March 31st, some states will
have already received it. But I can't sit here today and tell you what percentage. But certainly
our statutory deadline is March 31st for all 50 states and Puerto Rico.
Q: The split of --
DR. PREWITT: Oh, yes. I do not know right now with respect to the 39 percent whether
we're getting a higher level of long form and short form. Enumerators, like any other person,
they've got this work list. And, oh, my gosh, today half of them are long forms. I think I'll go do
my short forms first. But, nevertheless, they're not finished until they get the long form data, as
well. But, no, we haven't broken it down on that. I think I would know if someone said, my
gosh, we don't have any long-form data coming in.
The first part of your question really had to do with --
Q: The split. Of 120 million, how many of those were long form and short form? And
then of the 42 million?
DR. PREWITT: Yeah, right. Of the initial universe, the long form is basically one-sixth,
approximately 17 percent of the population. Then as we reported earlier, the long form return
rate in this box lagged the short form return rate by 12 percent. And I know the exact numbers,
but. Every time I try to do math instantaneously. Like, for a moment, I said that 1.5 percent of
our workforce of 500,000 was 75. Of course, it isn't. It's 7,500. And so when you're doing this
in your head, you sometimes add a zero. I do want to correct that 1.5 percent of our workforce
right now would be 7,500 cases. Of course, the number from the crime victimization study is an
annualized number, and we're talking about a workforce which is only going to be here for about
eight to 10 weeks. So you've got to do the arithmetic to see what the right proportion of that
I'm sorry for all of that.
We can actually give you the exact numbers on the long and short form in the Non-Response Follow-Up, and we can provide that to you. I just want to answer it quickly.
MR. JOST: Okay. One more on the phone. Evan Osnos of the Chicago Tribune.
Q: -- to staffing problems, because there aren't staffing problems. What other sorts of
factors might contribute to regional differences? What sort of other issues should we be looking
DR. PREWITT: Well, one of them I did allude to. Sometimes you do have not regions,
of course, but you might have LCOs that do need some management attention. And if those tend
to be clustered in one region, then you're going to have a slower start-up process there. We've
only been in the field two weeks. So a lot of this is really just start-up time. Some regions
simply got off to a more rapid, immediate start and some got off to a somewhat slower start. So,
what we're looking at now is whether we converge; that is, whether the two or three regions
which are now at the low end of our distribution will more rapidly catch up. One that was at the
low end last week, for example, has been catching up very, very fast because, there, the regional
director did some special work with his LCO managers and area manager and said, "This is not a
pace at which we can operate anymore." So you begin to see their curve, their productivity
shooting way up.
We expect that in -- we certainly expect that in Chicago. We're now putting extra
resources in the Chicago area. Stan, Stan Moore, the regional director, is beginning to reallocate
some of his resources. Central headquarters is providing more resources. So we expect that to
converge rather quickly.
But, no, it's not that we don't have people to do the work. It's that we're not necessarily
assigning the work to them or pushing them as hard as we need to be pushing them. That's the
issue. And it's not necessarily that we're hitting regions where we think we're going to have
higher levels of noncooperation with the public. We may hit that, but we're not hitting that yet.
As I say, we're still basically dealing with what turn out to be the easiest cases. The first ones
you get are always the easiest. And we won't really know if we have regional differentials in
terms of public cooperation until we get into the last 10 percent of this effort.
Q: Genaro Armas, AP. There're been a couple cases of people posing as census-takers
either at the phone or at the door. Any sense of how serious a problem this is right now?
DR. PREWITT: Well, the scam issue is a serious issue. I don't mean serious in terms of
how widespread it is. But every time it occurs, it can damage the census, because, you know, as
a respondent, as a household member, if I feel like there's the probability that this person
knocking on the door is not benign, but has some other kind of criminal intent or scam intent,
they're trying to get my Social Security number; they're trying to get my credit card number.
Every time one of those stories emerges, it obviously creates a dampening effect upon the
willingness of people to cooperate.
So, we take every one of those cases very seriously. We work very actively with the local
authorities or the federal authorities in some instances, where that's appropriate, to try to find
those cases and get them off the street. But there's not much we can do about it. It happens
outside of our processes.
Again to respond to the same kind of question that Dee asked, I basically know what I'm
reading in the press. You're finding those. They're not our people. You're finding those cases
as quickly as we are, or some resident reports it to the local newspaper and it comes up in a story.
And, again, there seems to be, at any given time, there seems to be about, again, somewhere
between a half a dozen to a dozen of those in the country right now, as best we can tell.
But again, we're at the mercy of what we can -- of our own -- we don't have a
surveillance system in place to try to monitor everything that's happening outside the census
Q: Yes. Hi. Steve Lash of the Houston Chronicle, with a parochial question. And I'm
wondering about the region, the Texas region in general, and the Houston area in particular, if
there have been problems there, where it ranks among the regions.
DR. PREWITT: The region that Houston falls in is the Dallas region. It's now -- it's
not on the high side, but it's -- well, it's actually -- it's 38 percent and the national is 39
percent, so it's almost exactly at the national level. But I don't have it broken down. I only have
the regional number. I don't have Houston.
Q: As you may know, my name is Bill Reed. I am the media specialist for the Census
here in the Washington area. My question is, you were talking earlier about the success among
minority groups and urban groups in getting the numbers up. There was an op-ed piece in
Sunday's Washington Post, not written by Dee Cohn, but it was an interesting article, which I
was wondering if you read it. And the reason why, is our community partnership specialists have
made a major part of their presentation about the federal funds allocation, the $185 billion a year.
So I was wondering if you could talk -- it was an interesting article, and it took me back. Maybe
you could talk about it a little bit and explain your point of view on it.
DR. PREWITT: Well, I have to apologize, especially in front of Dee Cohn, that I did not
read Sunday's Washington Post. I was actually out of town on Sunday, and I read a competitor
newspaper where I was. So I did not read the Sunday Post. Someone mentioned this op-ed to
me and I actually have not seen it. I believe it was written by Peter Skerry associated with the
Brookings Institute. I've actually read his book, and I presume this comes from this book,
though I don't know what particular part of what is in his book he put in this op-ed. I only know
that he had one. I know that I've had strong differences with some of his interpretations, very
strong differences with some of his interpretations in the book and indeed have been on a panel
with him and discussed these issues in a public forum.
But I simply -- I'm sorry. I don't want to characterize or try to respond to a particular
editorial, which I haven't read. And I've now lost so much credit with The Washington Post that
who knows what the treatment of me will be from henceforth. It's the last time it'll ever happen,
MR. JOST: Thank you very much.
[END OF PRESS BRIEFING.]