Link to Census 2000 Central
Press Briefing -- July 26, 2000
Director Prewitt

MR. STEVE JOST: This is another one of our standard periodic operational press briefings where we report to you on the status of census operations that are ongoing at this time. Before we get to the event, I just have a couple of housekeeping matters for members of the press to alert you to some census-related press releases that are coming out in the next few days. And the principal reason we bring this to your attention is, these are some of our standard data products that we release and we just want to caution that they are not Census 2000 data products. They are part of our standard operations and anything you can do in any stories you might report from these three releases to help make that clarification would be appreciated.

Today on our embargo site we are releasing the first data from 21 of the 31 sites from the American Community Survey tests. That data will describe what the American Community Survey, which was in the field last year, collected from the sample in each of those 21 sites. On the embargo site you will find 21 sets of tables for each of those 21 sites. Again, the ACS is something that, if it gains the support of Congress and is fully funded, we hope would replace the long form in 2010. So it's an important story but it is not the Census 2000 results.

In addition, I believe on Monday we will be releasing the voting-age population projections that we do every two years in advance of a national election. Those tables will provide estimates of what we think the voting-age population is by each state and with breakdowns by race and age and other demographics. Again, that's not Census 2000 figures. That's part of our estimates program out of our Population Division. We hope it will be helpful to anyone covering the upcoming elections.

Third, later next week we will be releasing our 1999 population estimates for each of the 50 states at the county level. Again, that's part of our ongoing population estimates program that we do every year at this time. Any help you can give us to reassure your readers and viewers that that's not Census 2000 data would be appreciated.

Today we are going to update you on Census 2000 operations. It's our standard format. I think most of you here are familiar with it. The director will have a few opening remarks. We'll open it to questions. We'll alternate between the room and reporters who are with us by phone. We do have mikes for those of you in the room. When you have a question, if you would, please identify yourself and your affiliation for the benefit of our viewers and our listeners. We'll have our standard procedures. With that, I'll now introduce to you Dr. Kenneth Prewitt.

DR. PREWITT: Thank you, Steve. Thank you for coming. Before I turn to Census 2000 issues I'd like to just pause and say that certainly the Census Bureau is extremely pleased that Secretary Mineta has been confirmed, in record time, of course, as you all noticed. Secretary Mineta is quite familiar with the census. He was a strong advocate in the preparations for 2000. When he was still in the House, a very, very strong advocate for our procedures to turn attention, of course, to the Asian-Pacific Islander, to the Asian populations more generally. Secretary Mineta, for example, was engaged in that level of detail as a member of the House, even with how we formulated the question with respect to collecting data on Asians, and then, of course, very interested in and supportive of the large partnership program that reached into that population group.

I have met with Secretary Mineta, begun to brief him on Census 2000 and have been extremely pleased with the level of knowledge he's already exhibited, not just about the census itself but really about the federal statistical system. So we are very pleased the Senate acted so quickly on his nomination, and that we now have a Secretary of Commerce who is now engaged and will be engaged in Census 2000 issues.

As we finish Census 2000, I turn now to my operational comments. Before doing so, just a word or two. Because many of you have written about our enumerators, many of you have written some very marvelous human interest stories about the obstacles, the tasks, the challenges, the kind of people who are out there doing the job, and we remind ourselves every day that these are the people who are making the census a good census for the society.

One story that caught my attention, I think it was an AP story, ran on July 4th about a 76-year-old Philadelphia woman who said that she'd never been counted, but this time we had found her. Another one who lived on a remote rural road and said in his 40 years that he had lived there he had not been counted, but this time we did find him.

I cite those kinds of cases, obviously not to go just the anecdote route, but there are always people who are missed. It's extremely difficult to find everyone in this country. Even those who want to be counted; it's extremely difficult to find everyone. But what we have, and I'm going to talk about this in some detail, the number of things that we are trying to do this time around to actually go out and knock on at least 100 percent of the doors where we believe there are residents in this country.

I also cite those anecdotes to remind us that we are still doing the census out there. The census is not over. We are still in the field. Roughly 200,000 people are still out there doing various parts of our quality control work and our ongoing counting work. The people who know the census much, much better than I do - that is, who have watched four or five censuses - I met just yesterday, or the day before yesterday, in Los Angeles with all of our regional directors, the first time they had gotten together since we started Non-Response Follow-Up. I think among them they have about 348 years of decennial experience, census experience. They've really been around it. For some of them this is their fourth census. They feel very good about it, and these people know if it's good or not good. They've just had too many experiences.

So when they tell me that they really do think it is a good census thus far and continues to be a strong work then I must say I respect their testimony on this more than the testimony of some people who are simply less experienced with what makes a census hard to do and what makes it above quality thresholds.

As you know, and just very quickly, we have completed the Non-Response Follow-Up part of our work, about 42 million housing units that did not mail back the questionnaire. We completed the list of enumerator field follow-up work in early July. Also in early July we completed our update/enumerate field follow-up work that was in more of the remote areas, the American Indian areas, the colonias down on the border, resort areas with high concentrations of seasonal homes, those with the update/leave areas and so forth. All of those major, big field operations are now completed. But that does not mean that the census is complete.

Of course, many people when they read about this believe the census is complete because that work is done, but what really happens then is that we moved into what we call our 'Quality Counts' phase, which are anywhere from, depending on how you count, as many as nine different major field operations still going on right now. One of those, which I described in an earlier press briefing, was Coverage Edit Follow-Up - we do that primarily on the phone from 13 call centers. The workload for that alone is 2.3 million cases. That's where we do count resolution. That is, if there is a discrepancy on the form between the number of people who actually were listed on the form and the number of persons they said lived in the household, we resolve that discrepancy. I say that's about 2.3 million households, and we're about two-thirds complete with that work.

Another very large field follow-up operation is our Coverage Improvement Follow-Up. Here we're going out to about 8.7 million housing units in a series of four - well, now three waves. We initially thought it might be four but we were able to get it under way quite quickly and so it's only three waves. Past experience tells us that some enumerators do erroneously classify occupied housing units as vacant and, therefore, delete them. What this operation does is make doubly certain that no housing unit is taken off the address file erroneously. That is, no occupied housing unit is taken off. So, as I say, this is about 8.7 million housing units. We send someone back to that door to knock on that door to make certain that it is true that it is a vacant unit.

There are vacant units in, of course, the housing stock of America. Probably, based on our most recent data, not Census 2000 data but earlier data, somewhere between about 9.3 percent of all the housing stock is vacant at any given time. These are housing units that are for sale, or housing units that have been built but have not yet been inhabited, or housing units that are in the process of being demolished, and they are still residential units but they are vacant.

What we want to make absolutely certain is that once we put something on our list as vacant -- therefore we delete it from our address file -- that we make absolutely certain that it really is vacant. That's what this process does. We go back to every one of those housing units which we have reason to believe are vacant to double-check. Of course, most of them will, in the final analysis, be vacant because at any given time, as I say, there are a lot of housing units that are vacant. Seasonal homes also fall into that category and so forth.

When we go into the field for these housing units, our processes are very similar to those in Non-Response Follow-Up. That is, we have multiple callbacks, if we really do believe that it's occupied and we don't find anyone there, then we go to some sort of proxy system.

If you attended yesterday's press conference on the Hill, it was incorrectly suggested that we had moved some deadlines on this operation. There is no single deadline for this operation. There is a series of four waves, and there are completion dates for each of our scheduled waves. The first wave is now completed, and that was - I have that number but not in front of me. I think that was 242 of our LCOs were in wave - no, 342 of our LCOs were in the first wave -- and that work is now completed. We are now starting Wave Two and then we will start Wave Three later on.

There is also an operation called Residual Non-Response Follow-Up. This is for addresses that show up in our address file, but which the data-capture system seems not to have a record. We have about 136,000 such addresses spread across the country. We are going back to every one of those. That is also a process that is on schedule.

We have a field verification process that we will initiate in about a week. That is where we go back to the housing units as best we can on questionnaires that we receive from some source other than from an enumerator or the Post Office; that is, the "Be-Counted" forms, or telephone assistance forms and so forth. About a million of those. These are the housing units we have to make certain really are there because they weren't initially geocoded in our system because of the way in which the responses came in. Somebody simply picked up the phone and said, 'Look, you haven't counted me, my name is, so many people here and so forth,' and they give us an address. We have to go out and make certain that address is actually where they said it was so they can be properly geocoded. So that is a million cases.

We also have an operation, which has not yet been described to any of you because we are just now - we just made the decision that we should do this operation this year. We did not initially think that we would, based on our 1990 experience. This is what we call our "population unknown". It's a supplemental operation that will be going into the field right now and continuing through August.

As the forms came in, we had a somewhat higher than expected number of housing units identified as occupied, but for which there was no estimate of the number of people who lived in that household. So it comes in as an occupied household, but we really did instruct enumerators never to guess. If you don't know how many people live there, don't make a guess. So if they come in that way, they are instructed to write in the number of people living there, into that box, they are instructed to write 99, so we call these cases the cases of the 99s, the 99 being unknown.

As I say, about 742,000, it was a higher number than we had had in 1990 of unknown cases, and so we decided to go back out into the field and check up on those and make certain to see if we can get better data on the population of those housing units because they are occupied, but we simply don't yet know for certain whether three, four, seven people are living in them. So that's another big field operation.

Yet another one: we found 80,000 addresses that were erroneously excluded from our Master Address File that we constructed for our Coverage Improvement Follow-Up operation that I just mentioned. These were addresses that were geocoded somewhere after Non-Response Follow-Up started, but before we started the Coverage Improvement Follow-Up. Both of those are coverage strategies, of course, and we found 80,000 addresses that had been - they had come into the file in such a way and in such a period of time that they had not been captured in either one of those large address file structures. So we are now going back out in the field to get the responses from those 80,000 addresses.

We have a boundary validation process that we're now doing with local governments, where we ask them to validate the boundaries, the jurisdictional boundaries that we are using in the census. Then, of course, we have the Accuracy and Coverage Evaluation, which is obviously already in the field. That's, as you know, 314,000 housing units across the country as our big systematic quality check on our own work. And, in all, about 91 percent of that workload is now completed.

We also, in addition to all of that, do ongoing evaluations of our own work. It will be toward the end of the year before we can report on all of those, but we are doing evaluations of the Internet capture system, block canvassing, Master Address File. Almost every one of our major operations, and even minor operations, has its own evaluation attached to that. During the course of the year we will be reporting the results of all of those evaluations.

The point of all of this is that the census is not finished. We have five months to go before we report our first major set of numbers, the apportionment numbers. Between now and that end date, that big deadline, we will continue to deploy every possible field operation that we already have on the books that we think is appropriate to make certain that we try to reach 100 percent of the housing units in the United States and get the best information we can from each of those housing units. This is the grounds on which I am prepared to say, and continue to be prepared to say, that this is a good census.

As many of you know, yesterday the census was described as a census that had gone awry. The facts simply do not support this characterization of the census. I do believe they support the characterization that the census is a good census, and a substantial number of quality-control procedures in place and being implemented will actually make it a better census.

Just a few of the facts. Concern was expressed, for example, yesterday about offices which got on the job and did it expeditiously. Well, there are two reasons that explain that. First, these are offices that had a higher than expected mail-back response rate, and thus a lower-than- planned-for workload. Indeed, if you think about it, across the country, because of the high mail-back response, higher-than-expected mail-back response, we had a 10 percent reduction in our follow-up workload. That's a substantial reduction in your workload.

The other thing we had across the country was a much better than predicted recruitment pool. So in those areas where we had a good recruitment pool, we could put many people in the field, we had a lower than expected workload - lo and behold, we were able to get this work done expeditiously. Indeed, in one of the offices that concern was expressed about yesterday, I went back and looked at it, we had initially planned to have 222 enumerators in the field in that office. Indeed, when we started Non-Response Follow-Up, we had 700 in the field. So we had triple the number that we had thought we would have. So, big surprise! If you've got three times the amount of workforce and you've got 10 percent less workload, then you are able to do it expeditiously, and I think that is in part what makes it a good census.

As we know, the closer to April 1 the data are collected, the better the data. We have said that repeatedly, that we want to get as much of these data collected as possible close to April 1st because you begin to have memory problems, you have moving problems and all the data are better if you get them closer to the reference date of April 1. So the characterization of the census as a census that has gone awry because some of the offices got their work done expeditiously strikes me as unfounded.

Also, concern was expressed at the other end of the scale. That is, about offices that started slow but finished fast. Well, these were exactly the offices in which we had to make major management changes and brought in the best enumerators that we could from nearby offices. I visited one of these offices. It was understaffed. It was understaffed to begin with. And they struggled for weeks to catch up. Then what happened, of course, is that we brought in new major management. In this particular instance, we replaced the office manager with an area manager, a much more experienced person from our regional staff, and we brought in literally hundreds of enumerators from nearby offices so they could help in this particular office. So, certainly, you're going to get a surge when you suddenly have that kind of recruitment pool available to you.

Another one about which concern was expressed was one of our urban centers, where we had a very hard time getting into the apartment buildings with the guarded doors. We worked something out with the city and the city put a lot of pressure on the real estate industry in those units, and suddenly the doors all opened. So, for three or four weeks you hadn't gotten into any of these buildings and suddenly the doors all opened. We got into them in a hurry, with all of our recruitment and all of our enumerator staff. So, clearly, there can be a surge under those kinds of conditions.

Concern was also expressed about whether we were checking up on our work. We're doing nothing but checking up on our work. Indeed, in addition to all the things I've just mentioned, we have a very special quality control process where we go out and look at 3 million cases across the country, and the entire purpose of that is to simply call the respondent and say, were you enumerated? We were told you were enumerated. Are there the number of people in your house that we've been told are in your house? And so forth. These are just like any production line, you know, there is a row of Coke bottles going down and you pick up every 10th one and you make sure that the quality standards are being measured. We have that big process going on out there right now.

So, there just was, I thought, a characterization of the census which does not really comport with at least the facts as I have tried to explain them and will continue to try to explain them.

Indeed, there was a concern about the number of deletes, as if somehow that meant something special. What 'deletes' mean is that you go out and find out there are some housing units that aren't there that you thought were there, and as a result of the fact that we can reconstruct an address file that was as inclusive as possible, and so that means we put in houses, addresses that we didn't know when we knocked on the door or went to that housing unit if it was actually there.

Housing units get demolished all the time. Housing units get converted from residential units to commercial units. Housing units are built, there is an address, but they are not yet inhabited because it's a new area. Lots of things lead us to put an address on the address file, but then when we went out in the field they weren't all there, so they become a delete. So a delete is not an indicator of a census gone awry. What a delete is is an indicator of a census that's checking and rechecking its work.

The broader suggestion was that this has been a rushed census. I think the facts point in exactly the opposite direction. We are still in the field. We will remain in the field until we've exhausted every procedure that can make this good census a better census. So I guess I just am here to tell you today that the operations are still ongoing. We have five more months before we produce our first big data product; that is, the apportionment numbers. We have something like nine different major field operations that are continuing as we try to find all of the residual cases. We are double-checking the quality of our address file and double-checking the quality of the work, and with that I am happy to take your questions.

Q: Chuck Holmes of Cox newspapers. Dr. Prewitt, do you think yesterday's press conference was a deliberate partisan attempt to question the validity of this census and its credibility?

DR. PREWITT: No. I think that there was some staff work done on some data that we have presented to the staff, explained to the staff that these data are not easily interpretable unless you really do understand them, and we did try to suggest to them that please - look, we knew that they would want to use the data. But we said, please, let us know. We might be able to help make sure you understand it.

So I think what happened is, somebody, you know - I now - let me, since you asked the question, let me give you a slightly different answer. The particular data that were used to bring these charges forth is now data that I wish I'd given to the press three weeks ago, four weeks ago, and tried to explain it. It's quite complicated and it comes from administrative records and a lot of things in it that you need to understand.

For example, one of the items on that data file says, how many of the housing units are what we call Pop 1 units. There is only one person living there. We use that as a signal to us. If we get a higher than expected number of Pop 1 housing units, we go back out in the field and double-check to make sure that that's just not enumerators trying to get the work done too quickly. It does turn out, however, there are very large sections of this country where there are blocks and blocks and blocks full of housing units, which are Pop 1 housing units. Nursing structures, where you have one person per housing unit. They all have their address but they live alone. We have a lot of single apartments, buildings, apartments with a single person living in them.

So we went back out and checked on those Pop 1 housing counts. And, if we found an area we thought where there were more than what the demography would suggest there were, then we double-checked the work. So it was that kind of - you have to do that kind of analysis to make sure you've got it right.

I don't want to characterize this as anything other than an attempt to make a big picture out of what are a lot of very, very small little pieces of this census. Indeed, of the 15 offices that were mentioned, 12 of those offices had a higher than expected mail-back response rate, which means it changed the whole character of what we had to do. But that wasn't taken into account in analyzing these data.

MR. JOST: Okay, we'll go to the telephones. Russell Banale of the Star-Ledger in Newark, New Jersey.

Q: Yes, Dr. Prewitt. I wanted to know. Some of the areas that were mentioned during yesterday's conference, there have been news reports that some discrepancies took place, particularly Chicago. I wanted to ask you locally about Newark and what your research has shown about Newark.

DR. PREWITT: I obviously don't have in front of me all the data for all 520 local offices. I'm more than happy to follow up on the specific question.

Newark did have an office which was one of those characterized as a late-surge office. It was an office which had the characteristic that it had a hard time staffing up, and it was one of those offices where, toward the end of the Non-Response Follow-Up period, we did bring in new management and we did bring in a lot of additional enumerators from around the areas. So in about a two-to three-week period, we did about seven or eight weeks of work. So that's why you get those kinds of patterns.

We do not have, in our judgment in the Newark area, anything that's systematic at all of a problematic sort. But if you have a particular question on a particular, you know, like a delete rate or a pop count rate, or something like that, I can certainly get that to you. But I don't have all of that in front of me.

Q: Okay. Thank you. So in the Newark area, there's not - whereas some people believe that area's questionable, the Census Bureau has not found that so and does not plan a total recount in that area?

DR. PREWITT: Of course, we don't plan a total recount -

Q: Or a recount of the Non-Response Follow-Up.

DR. PREWITT: No, not at all.

Q: Okay.

DR. PREWITT: That's just so far off the mark that there will be no need to recount Newark. We actually, with the help of the mayor, with a Complete Count Committee up there - I visited Newark twice during the Non-Response Follow-Up period. As I say, there were parts of it that were slow getting started, but when the tempo picked up, we had very high quality work. We had extremely good management that we brought in from our regional office in the New York area. So we have absolutely no reason to believe that we have any data quality problems in Newark.

Q: Thank you.

MR. JOST: And I can just say, if you have specific LCO questions, you can call our Decennial Media Relations at 301-457-3691, and we'll try to track down any specific data that we might on any individual LCO.

Back to the room. Herb Sample.

Q: Hi. Herbert Sample, Sacramento Bee.

You mentioned the status of a number of quality checks and field operations. When does that all end and you have all the data you need to start coming up with results?

DR. PREWITT: Sure. All of the work pours into something we call the Census Unedited File, CUF. And the Census Unedited File is the file that gives us our number. We call it unedited because we're still improving the data that's on that file. But we no longer are now changing the count, the basic count that's going to go into the apportionment number. We have just recently - and I want to be very careful about the date I use here, because I won't use a date, and I'll explain why in a second. We have extended the period. We have changed the deadline for the CUF, the Census Unedited File, by about three weeks to make certain that all of these new procedures have time to work their way out. And, so, the answer is in September. And I can give you the exact date. I just don't have it in my head. It's toward the end of September that we will have that.

Now, that doesn't mean we can suddenly just pop all the numbers out. We still have to do a lot of quality work, but we don't expect to be adding new census records after that time. We still do a lot of work with the address file, and so forth. So we don't actually produce, as you know, the state-by-state counts until December. It takes several months of really quite intensive work to make us as comfortable as possible. But after that date, the Census Unedited File date, we're no longer adding new cases.


Q: Before the December 31st release of the apportionment data, do you release any other kind of data, maybe not numbers per se, but - I don't know - increases in, or comparisons to 1990 of how much data came in a particular state or a particular area in a state? Anything like that?

DR. PREWITT: No. Here's what we'll be releasing prior to that date. There're two other big numbers that we're going to share with you. One will probably be in late August, and the other will be about late September, about the time that we'll be doing our Census Unedited File cut. One number that we'll be sharing with you in late August is our final response rate. As you know, we had to calculate our response rate based upon the information that had come into us, the mail-back response rate. We calculated it when we had actually started our field work. And after that a lot of forms continue to come in. Maybe as many as 3 million forms came in after we cut for Non-Response Follow-Up.

Now, the problem - you can't just add that 3 million to the initial base because we're making - we're now doing all the hard work to make sure there's no duplicates. We get duplicate forms, of course. Also, we get blank forms. We have to go back and find out why we got a blank form. And, indeed, some of this follow-up work is going back to a household where we got the form - this is in this residual non-response - we got the form from the address, mailed back in in the envelope, nothing on it. We go back to that household and say, "You know, we'd like to have you in the census," and we try to treat them just like a brand new respondent.

So that work will be going on and then we report to you the response rate. Now, that response rate is still based upon the initial denominator, which was, as you know, an address file that included all of the vacant units. Because we couldn't get them all out until you do the field work. And then about three or four weeks later, we report to you what we call our return rate, our participation rate. And that will have, then, cleaned up the denominator. We'll back out all of the vacant units, all of the duplicates, and that's a much, much more important indicator of the level of civic engagement across this country. I am very optimistic about that number. I can't announce it today because we haven't done the work, but all the bits and pieces that I'm seeing suggest that we're going to have an overall participation rate that's easily going to beat the 1990 comparable rate, and I think strongly so.

And that, I think, is a very, again, a very important statement about - see, what made this a good census was not something the Census Bureau did. It's something the American people did. They simply cooperated at higher levels than we expected or anybody expected. The GAO didn't expect it. The Congress didn't expect it. The Census Bureau didn't expect it. The cynics didn't expect it. They simply cooperated at a higher level than we expected.

And we got a lot of very dedicated people out there. Not only did they respond to the census in terms of the forms, but they responded in terms of coming out and helping us do this job. I can tell you - this is why some of these characterizations of census offices are painful to the ear, because we have 520 offices that were staffed by part-time, new people, just came out of whatever they were doing, out of retirement or whatever, to do these jobs. Is it true across 520 that you had some weak ones that you had to replace? Absolutely. What we'd expect across 520. The big news is that was very, very small. Some of these people are enormously competent. I met 20 or 30 of them going around the country. Ex-IBM executives, ex-military, retired school superintendents, people who simply cared about doing this and wanted to be part of the census.

So, I think, when the big number, that is, the response rate comes out - or the return rate, our participation rate comes out, it's going to be a very strong statement about the way in which the people did connect to the census. And that's a very positive story, I think, for the country, not for the Census Bureau, but for the country.

And there simply are not - we have one census office where we felt like we had to go back and do a lot of recounting. And that was in the Hialeah case and it's a very complicated case, I can discuss it - that's the only single case across 520 where we actually have to go out and do any kind of major recounting.

Sorry, that's a long answer. So long, I almost forgot your question. But those are the two big - so you will get that number. You will not get a population number. The next big population estimate you'll get will be the actual results of the Census 2000 which, as I say, will be in December.

MODERATOR: We remind our reporters on the telephone, if you want to get in the queue, you have to press 1 on your telephone. We'll go back to the room. Genaro?

Q: Genero Armas, AP. In light of the chairman's comments about those fifteen offices yesterday, did you plan to do anything more or differently in terms of quality control procedures there in those places?

DR. PREWITT: Well, I did take the chairman's comments seriously, in that sense. We have already gone to work on all 15 of those offices. We have relooked at the numbers ourselves. We've re-interviewed - we're in the process of re-interviewing the management in all of those offices. We will be reporting fairly quickly, that is, as early as the middle of next week, back to the chairman on what we have found. And based on what I found already, I don't want to overcharacterize it. We still want to do the work, we want to, you know, cross every "t", dot every "i", but based on what we've found thus far, these are very, very explainable kinds of phenomena or patterns. As I say, you get a surge where you changed management and beefed up your enumerator staff. You got an early completion where you had a lower-than-expected workload and a higher-than-predicted workforce. So there's nothing in these patterns that strike us as particularly mysterious or anomalous.

But we're still doing the work. We are certainly - where we have any reason to believe whatsoever from any source, from any source, you know - not to forget, as I said quickly yesterday on the phone to some of you, this census is being done in a fishbowl. We have plenty of people paying attention to it. We have mayors, we have the press, we have the GAO, we have the Inspector General, we have the subcommittee, we have plenty of people who are paying attention to the census. And if there are systematic problems out there and we don't find them, which is extremely unlikely because all of our processes are designed to find them, they're not going to go hidden. They're going to get revealed. And so at this stage of the census, to suddenly find 15 offices that have systematic problems, it's just not plausible. Just not plausible. That doesn't mean we won't go back out and relook at them. And we will do that and then we'll report our findings.

MR. JOST: Okay, anyone else in the room? Ellyn?

Q: Ellyn Ferguson, Gannett News Service. When you make your report to the committee next week, is that something you're going to share with the press or -

DR. PREWITT: I would love to share it with the press. [Laughter] Look, whatever we find, and let me just back up and then I'll let you finish your question in a second. But just to remind the press, because they've been a - they've paid a lot of close attention to this census and it's been a very - I really do welcome that. And you know - look, the Census Bureau's a grown-up agency, you know? We're not afraid to talk about our mistakes. We're not afraid to talk about our flaws. We are the ones who keep telling the country about the undercount. And how we've got to work hard to make this a better census. We're not embarrassed. As I've said many times, we take as much professional pride in telling you if we make a mistake than we did in trying to not make the mistake in the first place. It's just the nature of the agency. And so, we're completely - and if we find problems, we're the ones who decided that we had to recount Hialeah and told the chairman that we were going to do that. And told the Inspector General that we were going to do that because our own investigation told us, well, we just had enough concerns that we'd better do it. So, we did it.

And if one of these 15 turn out to have a problem, we'll go out and fix it. We don't have all these quality control processes to hide something. We have them in place to reveal things. We have hotlines. We invite our enumerators to call if they have any sense whatsoever that there's a problem. If they don't want to call us, call the Inspector General. Call their congressmen.

So, absolutely, we will - there's been nothing about this census which we have not tried to do in the open light of the day, if that metaphor works. Not very well.

Q: And you made reference to the data that was shared with the committee at the beginning of the month that you wish that you had shared it with the press. Are there going to be similar sort of data reports to the committee and are you going to be sharing that with the press?

DR. PREWITT: Yeah, I think if anything else - there's nothing else of that nature, Ellyn, that I know of right now, though I'm always surprised when I suddenly, even in the agency, a big complicated agency, when things turn up, but - but yes. I think we will - anything that we are going to share with the subcommittee or the - any other agency, we will try to get it and try to explain it in a medium like this to make sure that we can interpret it correctly.

As I said at an earlier hearing, you may recall, when it was said that the Census Bureau did not share information very well, that was the hearing that I did remind the subcommittee that we were making available in real time a terabyte of information. A terabyte of information for those of you who don't think that way is not - you take the yellow pages, the Washington, D.C. yellow pages, it's not five of them. It's not 50 of them. It's not 5,000 of them. It's equivalent to 50,000 yellow pages. That's a terabyte of information. And we have made that available to the GAO and the subcommittee on a real-time basis every week. It's our cost and progress data. I don't think the press wants a terabyte of information, but if you ask for it, we'll figure out some way to get it to you.

Q: If you can give us an executive summary of that, that would be appreciated.



Q: And just one last question.

DR. PREWITT: Oh, I've got to correct myself. It's been so long. It's not 50,000, it's 50 million. Really? No. Is it 50 million? Sorry, 50 million.

Q: Then it definitely needs a summary.

DR. PREWITT: So you really want it shorter, yeah, exactly.


Q: And just one last question. On Hialeah, you said that the Census Bureau has been working with the Inspector General. Has the Inspector General given you any sense of when there might be some report on Hialeah?

DR. PREWITT: No, they haven't. The Inspector General is quite comfortable with the action that the Census Bureau has taken. They have an independent investigation for other purposes, of course, to see if there is fraud. We're not recounting because we found fraud. We're recounting because we found misuse of some of our procedures. As we've described - a premature going to close-out procedures for reasons that were explained. And that's why we're redoing it, because we don't think the data - the data are not as good as we want them to be. But we, ourselves, do not have any indication whatsoever that there's any serious fraud. And that's what he's, of course, investigating. But no, he has not told me when he expects to give a report.

Q: Okay.

MR. JOST: Okay, Chuck Holmes?

Q: DR. PREWITT: Several months ago, back during the hubbub about privacy concerns and before the Non-Response Follow-Up, there were - you made a statement that there were concerns that if people didn't return their long form or didn't answer all the questions on the long form, the Bureau might not receive enough data on particular questions to cross the threshold of being valid data for the purposes of the census. Is that still a concern and at what point in the timetable of the process will you know whether a specific question can or can't be used?

DR. PREWITT: Yeah. Chuck reminds us of something very important that did emerge right as we were going into Non-Response Follow-Up and we were concerned about the differential response rate between the long and the short form. But then, in addition to that, because there was so much public talk about the fact that you don't really need to fill this thing out completely, we were worried about item non-response. That is, you get a form in but too many items are left blank and if you - particularly some of those items that were attacked, such as, you know, the handicapped item and so forth - you can get a pattern that really make the data not very useful.

We have not done - I want to state this very carefully and cautiously. We are not yet prepared to say that ... we are not yet prepared to give you a full report on the quality of the long- form data because it's not just been captured and processed yet. However, I am prepared to say that based on all the early indications, because we went back out in the field and began to try to get the long-form data, we did not encounter such resistance that make us - I'll put it this way. My level of concern about that has lowered considerably. Because of our Non-Response Follow-Up experience getting long-form data. A lot of people say, "No, I won't do it." But compared to the number of people who said, "Yes, I will do it" and did do it - we are less concerned about that right now.

MR. JOST: Okay. Anybody else on the phone? Or in the room? Well, unless Ben has a question, I guess we can wrap it up. Thank you much.

DR. PREWITT: Thank you all very much.

MR. JOST: And a reminder also about the ACS data. It's on the embargo site today and the pop estimates and voting age estimates come out next week.


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

Last Revised: June 14, 2010 at 01:38:51 PM