Link to Census 2000 Central
Press Briefing -- June 8, 2000
Director Prewitt

STEVE JOST: ... has been doing for almost a year now on the status of census operations, and we're going to provide you with an update on where we are as of today with our National Non-response Follow-Up efforts and other census operations.

My name is Steve Jost. I'm with the Communications Office at the Census Bureau. I think most of you have been through this experience before. But as you know, we have reporters who will call in on the phone from around the country. We ask, after the director's opening remarks, that you identify yourself and your affiliation if you have a question, and we'll alternate with questions in the room to those on the phone.

And just one announcement of a housekeeping nature. Our next operational press briefing of this kind is scheduled for next Tuesday, the 14th, at 11 A.M. And Maury, we'll be in what? What's our location for that?

Here in Salon F on the 14th. Is that a Wednesday? Wednesday. Thank you.

And with that, I give you Director Prewitt.

DIRECTOR OF THE U.S. CENSUS, DR. KENNETH PREWITT: Thank you and good morning. I will cover a few topics and then, obviously, take whatever questions you may have. I have to start with a warning. I cannot give you any bad news today. I appreciate that sometimes that makes it easier to write. I have, unfortunately, no bad news to report. Indeed, I think it's fair to characterize this census as "the good census," and I will try to demonstrate that a bit.

I am fresh from a trip to half a dozen different areas where I talked to a large number of people engaged in the census, many of our workers, of course, maybe half a dozen different local offices. I also helped enumerate in Section 8 housing. I met with an immigrant group, the Somalis, in Columbus. And I'd like to talk at the conclusion of my opening comments a bit more about that, because it was really quite revealing, for me, a trip at this stage in the census.

But first, just a few highlights of Census 2000.

We do have a month to go in the operation that we call Non-Response Follow-Up, which is the enumerator phase of the census. And, as of today, with yet a month to go, we are 93 percent completed with our Non-Response Follow-Up workload. That means we have 3.1 million of our initial 42 million housing units yet to visit.

When you add that to the mail-back response, that means that we're now ahead of 1990, or slightly more than 97 percent completed as of June 7th of all of our workload for these current operations. We obviously have other operations. I'll mention those in a moment.

To give you some idea of why we are willing to characterize this today as a "good census," if you compare the 1990 number, on June 1st - and we can only compare to June 1st because that's the way the data were collected in 1990 - on June 1st of 1990, we were at 61 percent complete with our Non-Response Follow-Up workload, and on the equivalent day, June 1st, 2000, we were at 82 percent. And there's no reason to think that that differential hasn't carried from June 1st to June 7th, but I don't have equivalent June 7th data for 1990.

Just to repeat that. In 1990 on June 1st, we had completed approximately 60 percent of the workload and, on the same date in 2000, we had completed better than 80 percent of the workload. So the entire system is simply well ahead of what we expected, well ahead of what we planned for, and needless to say, to be 97 percent complete of all of our workload on these operations at this stage is, for us, very good news.

Indeed, about a third of our 520 offices are in their final stages. They're completing work on the Non-Response Follow-Up phase, and, indeed, three of our whole regions are in the last percent or so of the Non-Response Follow-Up workload. Six states are now at 100 percent - Wyoming, North and South Dakota, Nevada, Utah, Montana, with another state, Nebraska, right on the verge of 100 percent completing the Non-Response Follow-Up workload.

In the areas where we're not there yet, we have stepped up our radio and TV ads in selected markets to further motivate the public. And, of course, in certain areas, we continue to recruit - as I've said so often, this is a process where you recruit until you're finished. And I'll talk a bit more about recruitment in a second.

Other major highlights. We have now scanned in our questionnaires for over 100 million addresses, and that includes 20 million enumerator forms. I pause on that because, if you've seen any of the enumerators at work in the field, you appreciate that sometimes they're working under difficult conditions, which is to say they've got hold of their forms in their hands, and they're sort of thumbing through and they're trying to take the answers down, and maybe it's raining or it's snowing in some parts, or it's been muddy or whatever.

So we worried a lot about whether our enumerated forms would scan as well as the ones that were simply mailed back in this nice clean envelope. And our accuracy rates for our enumerator-scanned forms are exactly what they were for our mail-in forms, and that's an enormously important small little item because we simply did not know. We tried to field-test this as best we could, but not until you start getting millions and millions and millions and millions, which you do in the census environment, can you know for certain.

Now, obviously - just one other highlight. I want to say our Accuracy and Coverage Evaluation is one of our next big field operations, many of you know, has already been initiated over the phone, and our personal visits will begin later this month on a flow basis. We do expect to complete that in September, and we've now completed approximately 90,000 cases in our Accuracy and Coverage Evaluation survey.

All of this is what, I think, gives us some pause or some cause to pause and declare this, at least as of now, to be a good census.

Now, I don't want, as always, to ignore the fact that we do have to take corrective action in a number of areas. We've identified a small number of LCOs, where the operations still need attention, where we want to sustain the momentum or build momentum. They're not at a point where we would like them to be at this stage. And in those areas where we're lagging behind the overall national response rate, we have taken very strong steps. We've added more enumerators. We've authorized overtime. We've purchased spot television ads. We have extended, for example, radio ads for four weeks in about 180 of our largest and medium-sized cities, created a public service announcement in English and Spanish, which we will be feeding to 93 markets around the country under the general title "There's Still Time to be Counted."

So even though we're, overall, in extremely good shape, we don't for a moment relent in terms of trying to make certain that we do it in every community across the country.

Indeed, this last week, I was able to visit, as I say, a number of places. I was in Baltimore, where the mayor, Mayor O'Malley, and I had a press conference, along with Congressman Cummings, and where the mayor announced a major June Non-Response Follow-Up campaign, including using sound trucks, public meetings, targeted mailings, promotional literature.

Yesterday, I met with Mayor Daley in Chicago. The city has also pledged money for radio and TV ads and been very supportive in promoting the census, and even having rallies as we are speaking. Congressman Gutierrez went door-to-door in his district encouraging residents to cooperate.

Earlier in the week, this week, I was in Milwaukee, where I met with Mayor Norquist and Congressman Barrett. Again, a major rally event and meeting with the other leaders, including representatives from the governor's office to continue to maintain our momentum in Milwaukee.

Then, last week, I also was in Columbus, Ohio, met again with local leaders and with Ken Blackwell, of the Census Monitoring Board, and together, we went and visited the Somali community to make certain we got a complete count there.

Which is to say, as you well know, we've had this strong partnership with local leaders and community action groups, and we will sustain those partnerships every place we need them in order to finish this in the remaining 30 days.

Rather quickly, on other operations, there is an operation we call Coverage Edit Follow-Up, which is in the field and it stays in the field through all of June and July. And this is an important operation for us because, as you know, sometimes you get forms back in which there's a discrepancy between the number of people who are written on the form as belonging to that household, and the number of individuals identified on the form. And we use this process to correct for those discrepancies.

Also, any form that we get in where there are as many as six people, we go back and double check it because, if it says six, we know you can only identify six persons in a household - that's all the form allows - and so we go back to all of those households to make certain that there aren't other people who they simply left off or ignored because they didn't have a way to give them their complete information.

So this Coverage Edit Follow-Up process is really to make certain that we get everyone we can in all of the households. We have some reason to think it makes sense to do that work. We do that work on the telephone.

We have about 2.2 million households in that category and we finished about 300,000 already. And, as I say, we run that through all of June and July, so we're easily on schedule with that program.

Another major activity is our Coverage Improvement Follow-Up work, which we start on a flow basis. That will start in the field - it's a field operation - on July 6. I've briefly described it before and I'll just be brief today. It's primarily designed to improve coverage of housing units that may have been inaccurately classified as vacant or nonexistent in an earlier census operation. As you know, we start with our Master Address File. That's our control file. And sometimes, the enumerator will come in with a report that says that house is vacant or it's not there anymore.

Well, we go back out and double check that. We double check it both in our quality assurance program, but also as a major field operation. And this operation, by the way, which we call Coverage Improvement Follow-Up, CIFU - forgive me for yet another acronym - and it sort of sweeps up a lot of the stuff, like that vacant/delete issue. That's where we do the new construction.

We just talked to the city of Chicago yesterday. They really believe there's been a lot of illegal conversions in recent months. And so they think buildings which were not inhabited when we did our Master Address File are now inhabited, and obviously, if we can now get those addresses, we will put those addresses in this process. So it's an opportunity to go out and make one more shot at making certain that we do contact every resident, every residential household in the country.

And, as I say, it will be approximately 8.2 million housing units, and we will be in the field starting July 6th, and we will do that on a flow basis, and that will take a good part of July and August to complete.

Finally, we are updating our address list, even as we go. We have our "Be Counted" forms, our Telephone Questionnaire Assistance program, persons reporting usual home elsewhere. And we have to geocode all of those responses back to our Master Address File. There are approximately 2.7 million of those addresses. We've now processed about two thirds of them, and have successfully either electronically matched them or, in some instances, clerically matched them, and so we're well under way to resolving any discrepancies between our address file and these additional new addresses.

And, then, as I said, we do continue to recruit. Indeed, overall, we've tested 3.6 million people, applicants, of whom we've concluded to be about 2.6 million. And indeed, we continue to be at 100 percent of our recruitment goals across all of our regions with one very small exception that's slightly under 100 percent today, but they are still recruiting there. By the time the day is over, they may well be over 100. But basically, across the country, we continue to be 100 percent equipped to finish up the work.

On average, we've maintained payroll at about 492,000 persons per week, and our rich applicant pool allows us to keep going.

Just one other small footnote for those particularly interested in the language program, we have about 1 million foreign-language forms. About 100,000 of those came from our "Be Counted" program and, of course, about 900,000, slightly more than that, came from our language assistance program. Most of them, of course, in Spanish. About 765,000 were Spanish, and then Chinese, Korean, Tagalog and Vietnamese. All of those have now been processed successfully.

Finally, as I said, we have now started the Accuracy and Coverage Evaluation program, about which I'm sure we will talk more in the next several months. And the telephone part of that is under way and we are very pleased with the quality of the work that is currently being done. And there, as I say, we've processed about 90,000 cases, which is about 28 percent of our total workload.

I want to just draw your attention to one other thing. There have been some very nice press stories that we characterize as "census heroes" press stories and, indeed, people helping residents give birth, census workers themselves giving birth and then going back to work, census workers dealing with very difficult circumstances and situations out in the community - the angry-dog phenomenon, etc. I've read about a number of other kinds of animal situations. An enumerator the other day helped - in a farm area - to deliver a foal while making her rounds in Illinois.

And I want to just use that to explain something because I saw it so much out in the field this last week. When you ask me why is it a good census, what explains the fact that, with all of this cynicism in the country, with all of the problems of response rates for all kinds of civic responsibilities and so forth, what do we come up with as an explanation for this?

We're still dealing with real opposition out there. This is a thing that's being put on doors in various parts of the country, or at least I'm not sure where, but they're somewhere. "Census workers are not welcome here. Do not knock." And simply put in your sex and your age right here, and that's all the census director or the census taker has a right to get from you.

So we are still dealing with serious resistance, some of which is, as in this case, organized. And, therefore, I do want to come up with some sort of explanation of how, despite all of this, despite the cynicism, the indifference, the difficulties, we're prepared to say that it is, as of now, a good census. And it is in part heavily attributable to the enormous energy and commitment that's true of these several hundred thousands of temporary workers. You simply cannot go into these offices and not get infected with enthusiasm and commitment.

Now, there's a sociological theory for this. There was a very important book called "The American Soldier," written after the Second World War, and the focus of that book was what explained the way in which fighting units fought. And they went in with a standard kind of hypothesis. These people were very patriotic to their country. These people are fearful of a security threat, our soldiers and so forth. These people are fighting for their loved ones back home. All of those hypotheses failed the test, a serious social science test. And the finding that came back and came back and came back is they fought with each other because they were in this together, and, indeed, the word "peer group" got invented in the analysis of the American soldier data after the Second World War. And it was simply a commitment to each other. That's what you run into in these offices. You simply run into people who have become a team, who are really dedicated to doing this work.

And it's moving. You know, they didn't know each other. They're just randomly collected from the various neighborhoods in the various communities, and they get into those offices. They take on this challenge and, my gosh, they're going to do it.

I was in an office in the Near South of Chicago yesterday, which is very close to being completed. In fact, they hoped to be able to complete it before I got there, and they were within a half a percent and they were working very hard so they could say to the director - I was very pleased that they didn't complete because they completed it the next day and I could take credit for it. The director goes to an office, it finishes its work the next day.

But as the Chicago Tribune wrote it up, it was almost like a revival. They were just on their feet, clapping, hugging each other, shouting. It was really just great. And I think part of what has made it work has been that kind of engagement in this task. It's not just the operations, it turns out to be these enumerators, and why they finally cared for their communities. And I think the fact that we heavily recruited out of the communities so they saw the benefits, understood both the resources and the representation and so forth just made them care about their communities and they knocked on those doors and knocked on those doors and knocked on those doors until they got every one of them to open, one way or the other. Even if they finally couldn't get them to open dealing with these kinds of problems, then they did do whatever else they had to do. They went to the neighbors. They went to the building managers. They simply didn't quit until they got the information as best they could.

So I think my explanation for why this is a good census is, first, the enormous enthusiasm and commitment and dedication. And part of what happens, of course, in something like this - you hire a lot of people. And you know, you've been writing the stories, there is attrition. People leave. But that means what you're left with are the people who stick it out, who want to do it.

And so, by this stage of the census, you're dealing with your most dedicated and your most experienced people. And when I say that we make corrections for areas that are less far along than other areas, that includes moving people around, taking the best people from one part of Chicago and putting them in another. We've moved people even across the country in order to strengthen the operations.

But the other thing I want to say is that we underestimated the lingering effect of the advertising and the partnership program. We made a lot of noise about that, of course, talked about it endlessly during the mail-back period, counted on it to boost the mail-back response rate, and it happened, as we all know. But we simply didn't understand that there was going to be a residue of that. So even today, we are getting questionnaires mailed back in, and we're certainly getting people who are opening the door who are willing to cooperate - not everywhere. We do run into those hard cases at the end of this kind of operation. But if you simply think of these visual in the rooms - you know, you started with the promotional campaign, and the advertising campaign and that's what that marvelous poster of the Indian girl reflects. You know, then you go to the form and then you go to the mail-back period, and you scan the data. You begin to process it, and then you go to Non-Response Follow-Up.

But there's a reason that we did it this way and it's to remind us all that advertising, promotional partnership campaign that was going on so heavily in the country in January, February and March had an echo effect that continues, I believe, to help explain why we're well ahead of where we were in 1990.

And then, just to make sense of this for our country and make sense of this as a civic mobilization effort, we need to keep that in mind. We literally underestimated it. We did not believe that it would have that much of a lingering effect. And that's the only thing I can come up with, that plus, obviously, the enumerator spirit and commitment. And that doesn't surprise me as a social scientist who knows about how groups form and why they get enthusiastic about their tasks and so forth.

But what I think we did underestimate was the degree to which the American people would continue, basically, to cooperate with this effort right down to the wire.

So, sorry for that little homily, but I was so moved by my experience in the offices that I can't resist sharing it with you.

So, I am open to your questions.

QUESTION: Some of the Republicans in Congress, including the chairman of the Census Subcommittee, have expressed a concern that some of the enumerators may have relied too much on what's called proxy data from neighbors and so forth. Are you tracking how much use there is of proxy data, as opposed to actual interviews? And are there any areas where that's a concern?

PREWITT: The answer to whether we're tracking it is no. And this is why. It isn't because we won't track it. We will certainly track it, but we don't track it during the process. That is, when we complete the census, we give a very full report of the number of questionnaires that will have been completed by a proxy interview.

And we simply don't do it in the process because there are lots of reasons. We're actually still doing the census and we wanted to finish the operations and we don't want to stop and do operations that are not germane to the immediate task of making sure we get everyone.

But let me just say a word or two. I certainly have some anecdotal evidence. There's no doubt but what we will use proxies in this census, as we will always use proxies. And I'll give you an anecdote. I was in a Southern city the other day with a crew leader, and we went to a house. This particular house was clearly inhabited and the woman who answered the door when the first enumerator came said, "I'm not going to cooperate." And so we then shifted the case to a different enumerator, hoping that a different enumerator would have better luck.

And the second enumerator went. And she wasn't home. And the second enumerator went back the third time, and she was there again, and she says, "Well, I'm too busy. Sorry." Obviously, that's a softening. "I won't cooperate" to "I'm too busy". We then made our regular three phone calls. And she simply would not be cooperative on the phone. We then sent yet somebody else back. She wasn't home. I then went back with a crew leader. This is now the eighth visit and she either isn't home or won't open the door. We then go to a neighbor and the neighbor says, "I'm sorry. It's not my business to tell you who's in there so I'm not going to help."

So here we are. We know this house is inhabited. And we cannot get the information. Now, what do we do?

It turns out, with respect to this house, a person who is in our administration in the office, who does payroll, lives across the street, and she knows this family very well. She knows that it's a single mother. She knows approximately her age. She knows she has two children because her children go to school with those kids across the street. They play together. She knows their names. She knows their ages. She knows their ethnicity, their race. She knows she owns this house. She knows all of the short form data. Now, that's a proxy interview. But it's much, much better than simply recording that as a vacant house because we know that's not true. We know there's someone there.

And so, there will be many instances of that where we would much prefer to get proxy data - especially good reliable proxy data - than no data.

You recall the exchange at Supreme Court hearing when, I believe it was Justice Breyer, who said, "Well, if the lights go out and you know somebody's there, what do you do?" Well, actually, what you do is you go to proxy. You know someone is there, and if somebody won't answer the door, then you do get the best information you can.

So, I am not at all defensive about the fact that the Census Bureau will, in those cases, use proxies. And I've got dozens and dozens of anecdotes of those kinds of stories where, in those cases, we will go to proxy data because we would much prefer to have some data about the household composition than no data, which is our option.

Sorry, a long answer, but I know we're going to be talking about the proxy data a lot over the next month or so, and we simply are unable to tabulate its current rate.

For one thing, you know, you've got duplicate problems. You've got all kinds of things that are going on in the data file until we've cleaned it up and made certain we know exactly what we're talking about. We will tabulate the proxy file probably by August.

QUESTION: Follow-up. Do the local offices have a hand in this?

PREWITT: No. They - sorry.

QUESTION: So, I guess, what safeguards are there to make sure that there's not overuse of it and that somebody isn't supervising to make sure enumerators aren't taking an easy way out?

PREWITT: Yes. Well, that our quality assurance program does. After all, not to forget, we sample 5 percent of every enumerator's work every day. And if we pick up a pattern of proxies that is, we think, out of line with what it should be - if we go back and find out, my goodness, this person was quite willing to give us the information - don't forget, we redo every bit of that enumerator's work.

So we would do that on an enumerator-by-enumerator basis. We do not believe - I don't want to characterize, to accept quite the premise of your question, which is, at what point is it too much. The point at which it's too much is only if you could have gotten the data without it. Then, we would prefer the resident do the data themselves. But when I compare proxy data to no data, there is no such thing as too much. By definition, you want data on every household, and so we don't have a magic number that says it is too much.

We certainly do a quality assurance process to make certain that any given enumerator is not taking the easy way out, and that we would, of course, put a stop to.

JOST: We'll go to the phones. Phil Lamos (ph) of the Hartford Courant. Phil, are you with us? On the telephone, Phil Lamos of the Hartford Courant.

QUESTION: Hi, you guys there?

JOST: Yes, go ahead, Phil.


In Connecticut, we've seen some complaints from Post Office box holders that they still haven't been contacted by the census. And I understand your policy of not mailing to P.O. boxes because you want to find out where people live as opposed to where they actually get their mail.

JOST: Correct.

QUESTION: I have a two-part question. I wanted to play devil's advocate here for a sec and just ask if you've ever considered just mailing to Post Office boxes and simply asking them to put their home address on the census form and not put a Post Office box or if you felt that would be too problematic?

And the second question is what assurances can you give the Post Office box people that the door-to-door people will reach them?

PREWITT: Yes, it's a fair question. Let me do the first part of it.

When we encountered this condition where we thought these were city-style addresses, we had city-style addresses, but it turns out the local post office did not do any city-style delivery, and indeed put them - they only had the post offices - where we could, we went and retrieved them and hand-delivered them ourselves. We could not do all of them. We did as many as we could in those circumstances.

These are anomalies. Just so everyone understands: what the census has is addresses which it calls "inside the blue line" and "outside the blue line." Inside the blue line are city-style addresses, where we mail the questionnaire. What we encountered around the country was that there were pockets within the blue line which turned out not to have city-style delivery situations. Indeed, it's a community for some reason within the city areas which gets its mail at a local post office. So, when we sent those questionnaires, they could not be delivered to the household because the Post Office didn't deliver them, but we did not know that until they came back.

So, what we did was, where we could, hand-deliver them. Where we couldn't, we simply then put these in the Non-Response Follow-Up process.

The first part of your question is why, then, we did not go ahead and send them to the Post Office box and simply instruct people to make certain that they put their address down very quickly so that we could geocode it. A perfectly fair question. And the real answer to that is one of the reasons this census has been successful is we made a very, very firm decision not to try to invent new operations in the middle of it. And what has gone wrong with previous censuses is, when you run into these anomalies, you suddenly try to create a new operation to fix it. And in creating that new operation, in the enormous pressure of doing this census and doing it right, you run the risk of making a mistake.

And it's been our judgment that we're much better off sticking with our design, even though we recognize these anomalies, than trying to introduce major new operations in the middle of the census. That's the real answer. It's not that that's a bad idea. In fact, it's a perfectly understandable idea.

But in the larger scope of trying to run this operation, every time you try to write new software in a hurry, you try to get new kinds of assignments, you try to get addresses some place that they didn't use to be, every time you do that, you run the risk of creating bigger problems than you're solving.

With respect to the second part of your question, we are actually very confident that we will get to all of those addresses. They are on our Master Address File, and since we do go to every one of the addresses on our Master Address File, we will continue to go there.

I reminded you a minute ago that we still have 3 million households yet to visit, and some of those households will be among that 3 million. And if they're not part of our Master Address File, we still have this subsequent process, this CIFU process, where we're still trying to add any addresses that make sense.

So, I don't want to be 100 percent certain of anything, but I am very confident that these housing units will be visited by a census taker before we're finished.

JOST: Yes, sir, right up front here.

QUESTION: ...with Fox News.

I wonder if you can talk to us a bit about the response rates from upper income communities and the problems of getting enumerators into some gated communities. There was a report yesterday suggesting Malibu, for example, California, 15 percent below. How much of a problem is this? How much of a problem was caused, perhaps, by some Republican concerns about the intrusiveness of the questionnaire and the suggestion from Majority Leader Lott that people simply not return it?

PREWITT: The gated community phenomenon is an interesting phenomenon, and what's happened in this society generally is that people are putting more and more barriers between them - themselves, their households -- and anything else. So the proportion of unlisted phone numbers in the country has simply gone up: the number of people who have mail forwarding and, then, of course, the number of people who have guards or other kinds of gated community phenomena, where they simply are protecting their household from any kind of intrusion. And, so, that sets up a set of barriers for us. And the Census Bureau, therefore, has to figure out a way to get through and around those barriers. They're not different from barriers - they're different in kind from other kinds of barriers, just like the barriers in linguistically isolated communities. You have to figure out a way to get through there.

We knew those were there. We do have special procedures in place. We did expect a somewhat higher mail-back response rate from some of those communities than we got. We got very high mail-back response rates from them, so we're always talking about the difference between what we would have liked to have had and what happened. It's not because - we did not have low mail-back response rates. I don't know the Malibu response rate in my head, but I know some other areas, which are characterized as gated-community areas, and you start with a high mail-back, which means the total number that you have to get in Non-Response Follow-Up is still pretty small. That is the absolute number.

Nevertheless, these have presented exceptional challenges. And we are busily, even as of today, working out particular strategies. We just worked out a nice one yesterday in Chicago with Mayor Daley that we think will work in that community. And that has to do with actually using the offices of the mayor to talk to the building managers and keep reminding them how important this census is.

These building managers in guarded communities are paid to keep people from getting into that building. It's what they're paid to do. And unless somebody can sort of convince them that's not what they're going to do in this particular instance, that's what they are going to do. And, of course, we're census-takers. We're not enforcement officers. All we can do is come and try to plead and so forth. And they say - "My job is to keep you out, sorry." Therefore, what we're doing is working.

We did this also in Baltimore; did this in Chicago, working with the leadership that says, "Look, talk to the building managers. Tell them how important this is." What we're doing in Chicago - very innovative - we have something in our processes called cultural facilitators. Now cultural facilitators are people whom we thought we would be using and have been using in places like the colonies in southwest Texas.

We swear these cultural facilitators in. They become sworn - not employees, we don't pay them - but we actually swear them in terms of Title 13 confidentiality protections - and then they either go with or go ahead of the enumerators and justify our presence. And we did not expect to have, quote-unquote, "cultural facilitators" in very wealthy high-rises in the Near North of Chicago. But lo and behold, if you think of these buildings as their own culture, then we can justify swearing in the building managers as cultural facilitators. They then go to the door or walk the hallways and say, "Look ...," or put notices up that say the census-taker will be coming through tomorrow. "I'm letting you know. It's very safe. It's very secure. We've worked this out. We believe this is important," and so forth. So that's what we're doing.

In other places, we've realized, of course, the people who have the best access to the gated communities are the house workers, the servants, and so we are making some of them enumerators because they're there. And so they become regular enumerators and they then know which households - and they're recognized in the hallways, so no one is worried about strangers in their hallways.

So we do lots of things. All of which is to say it is a challenge, but it's not a challenge that we don't believe that we will easily surmount in the remaining four weeks.


QUESTION: Do you find a lower response rate from people at the upper end of the income scale?

PREWITT: Well, we don't have precise data on that. I mean our response rate data is only being collected right now at the census tract level, and it's actually using census tract information from 1990. We haven't begun to do that kind of analysis. It's only anecdotal. We can't do systematic analysis yet of who's really responding in 2000.

I can tell you that, if you're at 97 percent completed across the country, you don't have a particularly low response rate for any group because you couldn't be at 97 percent if that were the case. So you're really talking about very small differentials. And it's very easy to take a case like maybe Malibu. I haven't read about that case; I did read about some Dallas cases; I was just yesterday in Chicago. It's very easy to take a few small cases and think that this is a generic problem. It's a --

QUESTION: Well, with that amount of money represented in a small percentage of the population, you could have a high non-response rate from 2 percent of the population and still be - have a 97 percent ...

PREWITT: Oh, certainly. Oh, you're absolutely right. But we haven't done that kind of analysis yet. As I say, the only systematic response rate we have is based at the census tract level, which are fairly large units.


JOST: Speaking of Chicago, we'll go to the phones and Evan Osnos of the Chicago Tribune.

QUESTION: I'm interested in staying on the privacy question for a second.

Earlier on, of course, there were conservative lawmakers who raised the issue. And I'm interested in finding out whether their own objections to the census have seemed to have had any effect in their own home districts, in their home territory, whether response rates or general levels of cooperation on their home turfs have been comparable or lower or higher than across the country.

PREWITT: Well, let me remind you that, to my knowledge, no political leader in the country did other than recommend cooperation with the census. Some of the political conversation was about the long form and some of the questions on the long form. And indeed, to my knowledge, every political leader who spoke out on the long form also said, "But we strongly encourage all of our residents to cooperate with the census, to complete the form. If you feel a particular question you shouldn't answer, then we understand that."

I happen to believe, as you know, because I've said so many times publicly, that's not the best message. This is not a pick-and-choose exercise. Every question is put there by law, and so forth. But nevertheless, I have no reason whatsoever to think that particular - I don't want to say suggestion, but recognition of some of the concerns about some of the questions - led to any lower levels of cooperation. I do think some of the other systematic attempts of the sort that I just described - you can't see it, of course, on the phone - I held up this thing earlier and it basically says - big notices - "Census workers are not welcome here. Do not knock." And then it goes on to say, "This is all you need to know or are entitled to ask." And it says, "Person one, sex, age; person two, sex, age." And so forth down to person six, and their sex and their age. And of course, this is false. This is untrue. This is not all that we are expected to get from the American people.

So I think that kind of campaign is problematic.

On the other hand, we are continuing to get cooperation. I am pleased that, in a recent court ruling in Houston, where there was a suit brought against the Census Bureau on the privacy question, the intrusiveness question, the judge ruled strongly in our favor that these questions are not intrusive and that they are supposed to be asked.

JOST: Bob, in the back.

QUESTION: Bob Rosenblatt, L.A. Times.

On the question of the long form and the short, if someone fills out only the first six questions and does not fill out the rest because of a recommendation from a congressman or anyone else, does - do you then give that form to an enumerator and say get the rest of the questions or what do you do with it? I've been getting contradictory answers when I call census people. Some of your folks have told me we give it to an enumerator. It goes out into the field, and other folks said, no. We - I'm not sure what you do if you only have six questions out of 40-something answered. And can you clarify how that's working?

PREWITT: Yes, and I do apologize if we don't sound entirely consistent on this because our policy is very consistent. We cannot - do not have an operation that would allow us to look at a questionnaire, determine that it was incomplete and then send someone back out in the field to complete that questionnaire. We simply are not doing that in the census process at all.

What we do - and these are the processes I described quickly earlier - we do look at a questionnaire and if it comes in and it says there are seven people or three people or 29 people living here, and there's no information on it, we go back to that household to say - well, who are these people. Because we can't just simply take a number and stick it into the census file. We have to be able to determine that there are people in that residence. However, a very small amount of information allows us to make that determination. If we get the surname and the age or the age and the sex and the race, if we have a good address, and we have some information that someone is there - and this goes back even to the proxy issue - if someone can say, I know there to be three people there. I know them to have the following genders or following rough ages or the following race and ethnic characteristics and so forth, then we know someone is there and we put them in the census count.

But at this stage, indeed, at no stage would the Census Bureau be able to do what we would call following up on item non-response. We will simply make the best that we can with whatever the quality of the data are that come in.

I believe, as I said before, what we're most concerned about in terms of the attack on the long form questions - we're most concerned about the quality of the data, not the overall response. We think we will still have good coverage of the population. We will try to count everyone, but we may well end up with less robust data than we believe the country and, indeed, the U.S. Congress wants out of this Census 2000.

But is that clear, Bob? I know there's been some confusion - but absolutely, we do not go back and complete the questionnaire, except we go back and complete enough of it to make certain that our count is correct.

QUESTION: When will you be able to tell us whether the data is, indeed, robust or not?

PREWITT: We will not be able to do a lot of that work until much later this year and early next year. When I say that we've already captured 93 million questionnaires, those include the short-form questions from the long form. We are very concerned about whether we have enough information from each one of those questionnaires that came in to make certain that there are people there. If we don't have enough information, that's when we go back into the field in these other operations.

So that's why we're now capturing the information. We're not even looking at the non-short form questions on the long form. We just won't even know that. It could all be blank. It could all be filled in. It won't make any difference to us at this stage. The only thing that makes a difference to us at this stage is we've got a household, so many people there, is that the right number of people?

JOST: Back to the phones. We'll go to Frank McCoy (ph) at U.S. News & World Report.

QUESTION: To boost their federal funding and congressional representation, some Southern states want visiting snowbirds to switch their residency. Consequently, the states have used marketing campaigns ...


... most of the year. And how are you following up on snowbirds who weren't in their primary residence on April 1?

PREWITT: Well, we actually try to find everyone, of course, where they were on April 1. But the question that we ask them is what was your primary residence on April 1. It need not be where you are. And indeed, many people could be in their second home. They could be snowbirds. But they consider their primary address to be someplace else.

We have to rely - like so many parts of the census - we have to rely on the report that people give to us. And we try to make this instruction quite clear. We want to know what is your primary residence. Now, the snowbird phenomenon is a complicated phenomenon for this society because what some of the governors and the mayors are saying is, "Look, we have them half the time, and we have to provide half of their services. But they're only counted once, and they're counted in Detroit instead of Tallahassee or Tallahassee instead of Detroit, or what have you."

The real solution to that, as a matter of fact, is what we call the American Community Survey. The American Community Survey, which I won't describe in detail now, does take the long-form data and spreads it across a sample, an ongoing sample that we hope to start in 2003, approximately 3 million households a year, which after five years is now 15 million households, which is roughly the number we get on long-form responses now, so it's the same robust sample, but it's on a continuous basis.

That particular questionnaire is designed to, in effect, allocate people to different residences, proportionate to how much time they spend in those residences. So we will actually solve this problem. And it's a much, much more powerful way to collect information about a population that's highly mobile. It won't solve it for redistricting and reapportionment purposes because that person can only be in one place for those purposes. But for federal funding purposes, other social planning purposes, it's a much, much more intelligent way to collect the data. And we do hope the American Community Survey is the next major innovation that the Census Bureau launches, but that does, of course, depend upon congressional authorization.

Is that a fair and adequate enough answer? I know it's a complicated question.

Lost him anyway.

JOST: Yes, right back here.

QUESTION: Thanks. Marina Abonne (ph) with National Public Radio.

Can you tell me, does the better-than-expected success of doing the head count boost the argument of those who say the sampling isn't needed?

PREWITT: It's very, very important to understand that, when I say we're 97 percent completed our workload across the country, that what we are measuring is we start with an address file. That's our control file. And when I say we're 97 percent completed, that means 97 percent of the addresses we now have a form back from or we've determined that it is vacant.

That does not tell us anything about the quality of coverage. And the Accuracy and Coverage Evaluation has to do with coverage, which is to say we may have gotten a perfectly legitimate form back and it said there are two people living here and they simply left their kids off. We know that happened in 1990. We expect some of that is happening today.

The Accuracy and Coverage Evaluation is a mechanism to make sure we have complete coverage, not did we get a form from every household. So they're really quite separate phenomena. So the 97 percent completion has to do with our address file, but not of the population count. We will not know the accuracy of the population count until after we do the Accuracy and Coverage Evaluation because there's no other way to know it.

You know there's an address there. You know you've got a form back from it, but you don't know if that form accurately reflects the number of people who were living at that household on April 1st and that's what we do with this other process.

QUESTION: What's your reaction, then, to the legislation passed this week in the New Jersey House where they decided to use unadjusted census figures in redistricting?

PREWITT: The Census Bureau's obligation is to give the country the most accurate data it's capable of doing. You've been following the census very closely now, many of you. You know that's not an easy task. There are a lot of complicated things going on in this country as we try to get these data. We strongly believe that, by doing this Accuracy and Coverage Evaluation and matching that back to the original census records, and then making the adjustments as appropriate, depending upon what we find, it is more accurate data.

Now, that's the Census Bureau's job. If the legislatures in different states or the courts in different states decide they don't want to use the most accurate data, the Census Bureau says, "Good, we gave it to you, but it's your choice how to use it." So we actually don't have a position on that. Our strong position is it's our job as the statistical agency to conduct a census, design and conduct a census that we will believe to provide the best possible estimate of the population.

Just another word or two on this. Any statistic is an estimate of the truth. You know, none of us will know the true number of people in the United States on April 1st, whether it's 235,311,000 or what have you. The census is an estimate of that count. And we believe that the estimate, which is improved by the Accuracy and Coverage Evaluation, is a closer approximation to the truth than the original estimate that comes out of the enumeration process. And we believe that on very, very sound grounds. And we'll be talking about that more, of course, in the next several weeks.

JOST: Back to the telephones. Amos Brown with the Indianapolis Record.

QUESTION: Yes, good morning, Director Prewitt.

Chicago seems - the Chicago region seems to have made a very strong recovery in Non-Response Follow-Up. Indianapolis is the second largest market and second largest city and we are getting - it's very difficult for us to get information on how Indianapolis has done in Non-Response Follow-Up. Can you help?

PREWITT: Certainly. Not now, but I can help you when I get back to the office. I can say that about 10 percent of our local offices are at 80 percent or lower completed for Non-Response Follow-Up with a month to go. I do know that there are no offices in Indiana that fall into that number, which means I have no reason to think that there's a particular problem in Indiana at all or Indianapolis.

QUESTION: In other words - yeah, because you've visited Milwaukee, where there were problems, Chicago, Columbus, Ohio, just kind of flew over Indiana. And as you're describing the overall census as a good census, and you're at 97 percent of the addresses reached, the people of Indianapolis would like to know where we stand. Are we at 97? Are we close to 100? Are we slightly below that?

Because we would - I think the community would like to know how well they have participated in this good census.

PREWITT: Yes, certainly. As I say, I don't have all 520 local offices in front of me. All I can assert without qualification is that I know you're at 80 percent or higher, but you could be at 95 percent or higher. I simply don't know that number, but when we get back to the office in a bit, I'll be happy to talk to you about it.

QUESTION: We would like to know. If it's good news about a good census, we'd like to report it.

PREWITT: Good. No problem.

JOST: We have to get this down for history.

QUESTION: I'd like to follow up on the question about the short form - the long-form, short-form data you're capturing. Are you finding item non-response on the short-form questions as you're scanning them in? And is that a concern? Is it more than 90? Can you tell us how much non-response and in what questions?

PREWITT: Again, it's a pretty inadequate answer because the degree to which we are scanning it now is to ask ourselves the question, "Do we have to go back and try to get more information from that household?" We're not doing systematic analysis of it. It goes back to what I tried to say before. At this stage in the census, every one of our operations is focused upon the current problem, which is Non-Response Follow-Up, making sure that every household is counted and so forth.

And we have made a determined decision - even though it would be interesting to us as well as the press and the public, the non-response. But we made a very determined decision not to deflect our people from the task at hand. And, therefore, we will turn to that task, as I say, later this summer. So the only level at which we're scanning it is a level at which the questionnaire gets kicked out if it has insufficient information on it to be confident that we've got the right count in that household...going back to Bob's question.

QUESTION: [Off-mike and inaudible.]

PREWITT: Yes. I don't have any reason to think that - I haven't really asked that question. I have no reason to think that proportion is any higher than 1990. Certainly, no alarm bells have been set off.

Now, I also think that we have a much, much better process in 2000 of identifying those questionnaires and going back and doing coverage edits. You see, not everything's equivalent to 1990, and so it may well be that we end up doing more than we did in 1990 just because we have a better process, not because the data were more defective on that score.

JOST: OK. I think we've got time for two more and we'll finish on the phones with Sherry Sylvester of the San Antonio Express.

QUESTION: Good morning, Dr. Prewitt.

PREWITT: Good morning.

QUESTION: Just a quick question. I can't see that sign you're holding up, but does it have a sponsor name on it? You said this was organized.

PREWITT: Yes, the sponsor seems to be the Heartland Institute.

QUESTION: OK. Have you followed those up?

PREWITT: We can certainly have it faxed to you if you're interested.



QUESTION: The other thing I wanted to ask you is the LCOs that are performing under 80 percent - so that would be about like 55 offices, it sounds like.

PREWITT: It's actually 48.

QUESTION: Forty-eight. Is there a list of those somewhere? Is that public?

PREWITT: Not completely. I don't mind trying to answer specific questions.

Part of it is these numbers shift, you know, every day. For example, if you set the threshold a little higher, or a little lower, as the case may be, depending on how you think about this - at 70 percent workload completed, or less than 70 percent completed, we're only talking about nine offices, and part of what's happened - and can I just say a word about this - the horse race phenomenon, which we do understand in American society. We understand the press's interest in this and local mayors' interest in this, and so forth. The Census Bureau does not see this as a race between one community and another community, between one state and another state.

The Census Bureau wants to get these data as quickly as possible because they're better data. That is, we know the closer we get the information to April 1st, the better the data. And our experience tells us that from time immemorial and all of our systematic research tells us that. So our sense of trying to complete this task on schedule has to do with the fact that we end up with better data. But whether, you know, one office or another office closes earlier or completes this particular phase of the census earlier, is not for us the issue. The issue is whether the whole system is moving forward at a steady pace.

And I simply don't want a local office manager, who is slightly lower than some other places, suddenly - and this happens - suddenly, the entire press, not having maybe anything else to do, decides - oh, my goodness, we've now found this and we'll all descend on this particular local manager. And suddenly, he's not doing the census anymore. What he's now doing is explaining to lots and lots of interested people this phenomenon.

So that's why - it's not like it's secret data, and I don't mind trying to specifically answer a particular question, as I just said about Indianapolis. But I simply don't want to create the sense that this is a horse race. Every local office is being measured against where it ought to be at this stage, which is not necessarily marked by the national average. Let me just give you a couple of examples. I'm sorry to go on for this, but it's important to understand, it seems to me.

The Denver region. The Denver region, as I read off those states you can tell, is obviously in very, very good shape today. But now the explanation for that is, in part, because the Denver region had a very large area of list/enumerate and update/leave, which means, when it actually started the Non-Response Follow-Up period, it had in place a trained enumerator staff who had already done these other operations, so they could get started faster. It isn't - you know, from our point of view, perfectly predictable.

In other areas, like the Chicago region, which did not have list/enumerate or update/enumerate, we did not have already in place this group of enumerators ready to go. It took another week or two to get it all organized and move it forward. These were all perfectly understandable phenomena. So that's why we weren't terribly anxious or worried. We knew we were going to get the job done.

So that's why the horse-race phenomenon sometimes misses these more subtle reasons for why some areas are further along right now than other areas.

When the entire system is 20 percent ahead of where it was in 1990, then even a case which is 20 percent below the national average is still where it ought to be. So we're not looking at, right now, more than a handful of offices where we believe we now have to put in extra effort, and that's what we are doing.

A long answer. I'm sorry. I can tell you, as I just said about Indianapolis, because I do have some lists in front of me, that of those offices which are at 80 percent or lower completed with Non-Response Follow-Up, San Antonio is not one of them.

QUESTION: Thank you.


JOST: Final question. Anymore? I think - right up front.

QUESTION: Doug Peterson with Nation's Cities Weekly.

I've got some calls from university campuses counting those students and wanting to make sure that they're on campus, and some concern about - because of the timing of the census and when the enumerators visited and final exams, that the counts that are being produced, the group-quarters counts that are being produced on some university campuses are much lower than the institutional counts and the registration that the universities keep.

What's the process? What can people do to help that?

PREWITT: Well, I'm afraid I'm not going to be very helpful on that because I don't have those reports in front of me.

I do know that, in some instances, where we were dealing with vacations on college campuses and so forth, we did go to administrative records. That is, we did go to the university authorities and try to use that as a way to complete the count. Certainly, we can revisit some of those places. I can't tell you in detail, but if the Census Bureau has reason to believe either there are so many people in that dormitory and our count is short of that, that's just part of the Non-Response Follow-Up process, and we will continue to try to get those cases.

QUESTION: [Inaudible; off-mike.]

PREWITT: Well, fair enough. But there are ways in which we can do that, in effect. As I say, partly going to administrative records, and so forth. And this final sweep, this Coverage Improvement Follow-Up sweep, as I say, there are a lot of - by definition, there are some loose ends in something as complicated as the census, and we, as best we can, take care of all of those loose ends.

So I just don't have any detail on this. I can't believe it's a huge problem or I would know about it is the point. There may be random instances of dormitories and we will certainly dig into those and see what we can do about them.

JOST: Okay. Thank you much. And a reminder, next Wednesday, June 14th, 11:00 A.M. in Salon F. Thank you.


Source: U.S. Census Bureau
[an error occurred while processing this directive]

Created: October 13, 2000
Last Revised: June 14, 2010 at 01:38:51 PM