Arlington, Virginia

Tuesday, October 24, 2000

Mr. Lawrence: Good evening, and welcome to The Business of Government Hour: Conversations with government leaders. I'm Paul Lawrence, a partner at PricewaterhouseCoopers, and the co-chair of the PricewaterhouseCoopers' Endowment for the Business of Government. The Endowment was created in 1998 to encourage discussion and research into new approaches to improving government effectiveness. To find out more about the Endowment, visit us on the Web at endowment.pwcglobal.com.

The Business of Government Hour focuses on outstanding government executives who are changing the way government does business.

Our special guest tonight is Dr. Kenneth Prewitt, director of the Bureau of the Census. Welcome, Kenneth.

Mr. Prewitt: Thanks very much. Glad to be here, Paul.

Mr. Lawrence: Let's start out by finding out about the U.S. Census Bureau. Most of us know that the bureau conducts a census every 10 years. What else does the bureau do?

Mr. Prewitt: The bureau conducts a number of major surveys over the course of its 10 years between the decennials. For example, every 10 years we do a big census of governments. So anytime you're reading information about how many governments there are in the country, what their budgets are, what their staff is, what their program is, you're reading data the Census Bureau collected.

We also, every five years, on '02 and '07, conduct an economic census, which is a census of economic firms. And in the other years we're doing things like what we call the Consumer Price Index. When you see an inflation number, you're seeing a number that the Census Bureau helped create. When you're seeing an unemployment number, market forces behavior, labor market behavior, you're seeing data that the Census Bureau collects. When you see data on housing starts, when you see data on trade statistics, all of those come from surveys that the Census Bureau produces.

Mr. Lawrence: In preparing for our conversation tonight I noticed that throughout your career you've worked in higher education, private philanthropy, and government, and I was wondering if you could tell us about your career and the different positions you've had.

Mr. Prewitt: Well, I was a regular academic for a long time. That is, I taught at the University of Chicago for about 15 years with shorter stints at Stanford and Columbia University. Toward the end of that teaching career I began to do more and more administration. I became a departmental chairman, and then I ran a fairly large research institute at the University of Chicago; not large by Census Bureau standards, but large by academic standards.

After about a 15-year career at the University of Chicago, I went to private foundation work, primarily with the Rockefeller Foundation. Also with an organization called the Social Science Research Council which was like a philanthropy, not exactly, but similar to.

Those management jobs were quite different from the government management jobs on a couple of dimensions. Managing in the research institutes, scientific institutes, on a campus or in philanthropy, it's less exposed. That is, you obviously have constituents, you have persons who care about your performance and the performance of the institution, but it's a somewhat more, how can I put this best, kind of encapsulated world: it's just the world of higher education or just the world of philanthropy.

Government, I've found, was a management job that was completely exposed. That is, I was in a fishbowl or have been in a fishbowl in the Census Bureau in a completely different way than I ever had been before. There is obviously press attention, but also the whole oversight apparatus of government, the Government Accounting Office, the Inspector General's office, the subcommittees and committees of Congress that were paying a lot of attention to the census, as well as the public, advocacy groups and so forth. So it was for me a very major shift to move from management in kind of the philanthropy/academic world, to the government world.

Mr. Lawrence: In thinking back, what positions or management challenges provided you with the best opportunity to develop as a leader?

Mr. Prewitt: Well, certainly the jobs I did have in the academic world at the University of Chicago in particular, but also at the Rockefeller Foundation and the Social Science Research Council, I primarily managed professionals, that is, scientists, and one of the things about managing scientists, you have to give them a lot of space. You can't micromanage them, they don't do their most creative work, and that turned out to be extremely important for a Census Bureau management job because there you also are managing scientists and professionals who are partly responsive to their own sense of what is the accurate or the most intelligent way to go about doing a task.

So you've got to give a lot of space, appropriate space, to a professional scientific community. So it's a somewhat different management job. It's not a line management job. You don't sort of send an order down, the deputy director then tells the associate director, then tells somebody else how to do something. You really manage through conversations. You manage, and I don't mean to overstate this, but you manage by bringing people together and talking through scientific strategies, what does the evidence say, where should we go next? Those kinds of properties of management are somewhat different from what we normally think of as line management. So I would say my experience in scientific management is what I really brought to my government job.

Mr. Lawrence: Many might describe the scientific management you're talking about much more like managing knowledge workers, and I think your successes would seem to be in a smaller environment than a larger environment in terms of number of people. Can that be replicated in a larger environment?

Mr. Prewitt: Well, basically, what happened with respect to the census management job is the things that I thought I understood about how to manage a scientific institute I was able to do, I then had to add new things to it, and that was the challenge. It was the new things that -- then you had to sort of coordinate kind of the new things you did with the old practices that you had in mind.

For example, I had never been in a management job where I had so many bosses. I've always had one boss, if any. In fact, really, when you're managing scientific professional organizations you tend not to have bosses. You may report to someone, but there is an awful lot of discretion.

But suddenly when I got to the U.S. government, I found that I literally had multiple bosses in the sense that the U.S. Congress is an authority figure over the Census Bureau that controls the purse strings, has the oversight responsibility. Obviously, I worked in the administration and reported through the Department of Commerce to the administration. These were different authority -- I don't want to call them bosses -- or I don't mean to trivialize it that way, but these were authority relationships which were themselves in conflict about the next step, for example, in the census.

So to go back to your first question, essentially the way I had earlier managed a scientific institute I found quite replicable, but what I then had to add to it was how do you manage in an environment that's political, and in a serious way political, but it's also multiple in terms of multiple authority, rankings, and relationships, and as I say, sometimes conflict among them. As I quipped after I got to Washington, everyone when I first got to Washington about two years ago, said, "My goodness, you've taken on a major hard, difficult task." After I'd been here about a month I then said, "Well, I guess not because I haven't met anyone yet who can't do it better than I can."

Mr. Lawrence: Well, that raises my next question, which is, how did you manage the conflict? How do you deal with the situation where you have multiple people who want to tell you things, and they are different?

Mr. Prewitt: I think one of the things that I tried to adopt when I got to Washington, I came down here and said, I have no enemies. I may have challengers, I may have opponents, but I don't have enemies, and I decided to conduct myself that way. That is, no one did not like me personally. So you have to really depersonalize in government management. You have no ego in this, and even when you're being beaten up a bit in the press or you're being challenged in a congressional hearing or what have you, you have to believe that it's not about you, it is really about a contest in effect about how things should go forward.

So the first thing I tried to adopt in that sort of conflictual situation was that these were opponents or challengers, but they were not enemies, and so that made it much more comfortable personally. Secondly, I had to adopt the position that everyone had a legitimate claim on me as a person, but also on the Census Bureau. That is, nobody had an illegitimate complaint. They may be more or less sound, they may be more or less rational, maybe more or less based on evidence, but they were all legitimate, and that helped me a lot.

As to really answering your question about how you do it, you simply try to constantly be honest. You simply say, look, this is why I think this way, this is how I do think about this, this is what the Census Bureau should do. And you try always to tell the same story everywhere. You simply don't try to tell one story in one situation to kind of curry favor or move something forward. Then you're going to have to tell a different story some place else. It does not work. At least I did not find it to work.

So I try to tell a single, consistent story right through the entire process.

Mr. Lawrence: It's time for a break. We'll be right back with more of The Business of Government Hour.

(Intermission)

Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence, a partner at PricewaterhouseCoopers, and tonight's conversation is with Dr. Kenneth Prewitt, director of the Bureau of the Census.

Census 2000 is often described as the largest peacetime mobilization in history. Can you tell us more about this effort?

Mr. Prewitt: Yes, you're correct. It is called the largest peacetime mobilization in history. That's slightly inflated in the sense that every census is bigger than the one before, so every decade we can call it the largest peacetime mobilization in history.

It's actually not as large, for example, as counting the people of China, but we do count in a more complicated way. A census is really two things. It's both a count, and it's an assignment of everyone to a residency, to a location, because a census has to do that. So the enormous challenge of a census is to get those two things happening simultaneously. That is, your address file work has to be every bit as rigorous and quality controlled as your counting work because finally, at the end of the day, you have to put people some place because that's how the reapportionment and redistricting and so forth functions.

So, yes, it is huge. You have to prepare a very long time. Indeed, I don't think people understand completely, we started preparing for Census 2000 in about 1988. I wasn't here then, of course, but I can tell you already we have a 2010 planning process partly because when you want to innovate, you want to do something new in a census, you want to test it in a census environment and you only get a shot at that every 10 years. So you can't wait, for example, until 2007 and say, wouldn't it be interesting to do something, so forth and so on. And if you'd only thought about doing that seven years ago and test it out in a census environment, you have a much better sense of it.

A census bureau has to be fairly cautious because it's so big, so complex, if you do something wrong, and it moves very, very fast of course, you don't have time to recover. So you've got to be fairly prudent about how you move into new directions. So that's why if you can test things in an earlier census environment, and that's why we're doing so much work right now that's really aimed at 2010, not at 2000, we're running maybe as many as 40 or 50 different experiments embedded in the census.

The American public hardly notices those. From the American public's point of view they're very tiny little things. But we might for example change the way we write our advance letter just to see if that affected the way in which we got responses. Maybe only 5,000 people get that. Also, not in the census itself but in the census environment, we sent out a questionnaire asking information that we didn't actually have on the actual census questionnaire, and that was a separate process in order to sort of test that in the census environment. So, yes, the planning for a census for anything of this magnitude has a long, running start.

The biggest, biggest task -- well, not the biggest task because it's hard to say what is the biggest task, but certainly the personnel task in Census 2000 was enormous. All together during our major field operations we ran about 920,000 people through payroll. Now, some of them worked three or four weeks, some of them worked three or four months, some of them worked two or three years, but all together it was 920,000 temporary people that were brought in to work on the census. Simply recruiting them, training them, and paying them was itself a management challenge.

Mr. Lawrence: Census 2000 is the first fully computerized census, as I understand it, from collecting data to releasing the final results on the Internet. Can you tell us about the role of technology in the census and how it perhaps differs from the 1990 census?

Mr. Prewitt: Just for fun, let me go all the way back to 1880. We took the census in 1880 and it took about a year and a half to produce the first results. A young engineer working for the Census Bureau said my goodness, this isn't good enough. So he began to play with something. It was a simple tabulating machine, card punch for the 1890 census. His name was Herman Hollerith. Mr. Hollerith devised this little machine, and we used it in the 1890 census and used it again in the 1900 census. It really speeded things up. Mr. Hollerith then left the Census Bureau, taking his key punch machine with him, and formed a partnership with somebody making tabulating equipment, and that became IBM. So the origin of the International Business Machines and then the whole computer industry really started with a Census Bureau innovation.

I tell that story because the Census Bureau has to constantly innovate technologically because the job gets bigger and bigger. One of the things that hasn't changed really for almost 100 years is the time frame within which we have to present the numbers. We collect the data as you know on April 1st, and then we have to give the major apportionment counts, that is, how many seats in the House go to the various states, by December 31 � nine months. That nine months was also the time limit we had, for example, in 1930. The population was half the size, much easier to count. So the only way you can get productivity gains in that short of a time frame, when you're simply doing a more complicated job, is with technological innovation.

A long preface to your question, but by the time we got to 2000, we had put in some really major innovations compared to 1990. One small one, and then I'll talk about a bigger one. Not really small, but perhaps less consequential. One of the things we had to innovate was our payroll system. Hiring, paying, recruiting, training, doing the whole benefits programs and everything else for 920,000 employees, some of whom as I say came in and out in 3 or 4 days, but you had to treat them like a full employee and so forth, we had to have a very automated payroll system, administration system, that had to be sufficiently sophisticated and flexible that it could be managed at the local level by people who themselves were temporary. We had to train them to use the system to train the temporary people. So our payroll administration system worked extremely well, we were very pleased with it, but that was a big innovation in itself.

Certainly in 2000 the more important, in a sense, technological innovation was in our data capture system. If you go back to 1990, you may recall that the census form had little circles called fosdik circles which you filled in, and you could data scan those little marked in circles with your pencil. That, by the way, the fosdik circle technology was also a Census Bureau innovation of a couple censuses earlier.

The problem of course with that technology is that it's a bit more demanding on the respondent, and it's more complicated to write the questionnaire. So what we decided in 2000, and we did this primarily with some important big contracts with Lockheed Martin and TRW, we decided that we would try to go to optical character recognition. That is, a data scanning that could actually read handwriting. It could read your name, your address, your Indian tribe, your ethnicity, your ancestry, what have you. And we designed a questionnaire that would then depend upon optical character recognition, intelligent character recognition, that had to count huge amounts of questionnaires very fast. It worked. We are extremely pleased with this innovation.

Mr. Lawrence: I know you want to tell us more. Hold that point. It's time for a break. We'll be right back with more of The Business of Government Hour.

(Intermission)

Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence, a partner at PricewaterhouseCoopers, and tonight's conversation is with Dr. Kenneth Prewitt, director of the U.S. Bureau of the Census.

You were just talking about some of the challenges and how technology has been used. Do you wish to continue?

Mr. Prewitt: Yes. As I was saying, the technology itself was fairly sophisticated, and our data capture system was extremely important. Very, very pleased with its performance, by the way. We got about a 99.7 percent accuracy out of the technology, which means it had to decide that that was a coffee smudge or not a coffee smudge, and that was a crinkle and not a crinkle. You're after all collecting the census under lots of different circumstances.

I myself went up and did some enumerations in Unalakleet, Alaska, on January 18th which is where we started the census. Well, I'm carrying those forms around. It's 15 degrees below, and you suddenly realize there's some machine down in Phoenix that's got to optically scan this questionnaire. It suddenly hit me: hundreds of thousands of enumerators are going to be out there with these forms in all kinds of wind and sleet, floods and everything else, and yet the data have got to be read.

So to get back quickly to your management question, part of the management challenge is to standardize your operations so that for example a single machine down in Phoenix, we had four machines around the country doing this data scanning -- well, more machines, but four central data processing places -- could do all the very same tasks when the flow of the material is coming in from hundreds and hundreds and hundreds of different circumstances. So the management challenge is how do you create a standardized product and at the same time allow for a lot of local flexibility or to recognize a lot of local conditions.

So we had to create a training system which at the one hand trained everybody the same way, but simultaneously said, well, but if you run into such and such a situation in Unalakleet, or you're running into it in Puerto Rico, or you're running into it some place else, you've got to allow for some kind of capacity to make small adjustments but at the same time have a standardized product. That's simply a challenge, a management challenge, and you have to do that through a training system of course that itself is training brand new employees who are only going to be working for you for four or six months.

The way you do that from a management perspective is it goes back to what I was saying earlier, you try not to do too many new things that you haven't done in early censuses, and training is one of the things that we've learned a lot about over the years in terms of making sure we can get the right kind of materials. The other thing, as much as you can, we do run some so-called small practice census, dress rehearsals kind of in the middle of the decade, we did about three of them before Census 2000, and we work out some of the kinks at that moment.

Mr. Lawrence: One of the main challenges facing managers in both the public and private sector is the need to hire and retain high-performing employees, and you've already described there's been almost a million people involved in the survey. How do you track the employees you need, but on a permanent basis and even the part-time?

Mr. Prewitt: Well, the decennial census is largely part time, so let me speak to that first and then get to the permanent staff. The real challenge is to get high-caliber temporary and/or part time, and some of them don't want to work full time. We did a lot of work with respect to our wage rates and set wage rates higher in 2000 than we did in 1990. It's a very labor market. We knew we were going to have a hard time.

The other thing we did is we made it much easier for people from different walks of life. For example, we made it easier for noncitizens to get these jobs if they were suitably qualified. We had to hire a lot of bilingual people, of course, so we had special training programs and recruitment for them.

One of the things that we did, if you don't mind me saying this, we depended upon people's sense of civic obligation and civic responsibility. I met a lot of people, especially those who managed our offices, ex-military, ex-government workers, ex- school teachers, who kind of came out of retirement or semiretirement because they actually believed in the census. A lot of people work for a decennial census and then don't work again for 7 or 8 years and then show up again and say, I had so much fun last time. It's an adrenaline shot because it happens very fast, it's very big, there are a lot of little challenges and big challenges.

So one of the things that we tried to do is create a work environment in which people who simply found it fun, if you will, to be engaged in something this big, this mammoth, would find an opportunity to work. That's primarily for the temporary.

Hiring the full-time regular employees at the Census Bureau, we have about 10,000 employees on a regular basis, about half of them in headquarters and half of them in our regional offices. Many of those are enumerators, but many, many are also statisticians, demographers, survey managers and so forth. It's very difficult because we're competing in an environment where we can't pay the same kind of salaries that the private industries can pay. We can't give as good working conditions of course. I might say so here right now we have a working condition out in Suitland that's atrocious. We're in a building that's 60 years old that was built as a temporary veteran's hospital. We have health problems. We can't drink the water. You've got to be careful about washing your hands. Pigeon infestation. It's very difficult to hire people to the government where the salary scales are what they are, and sometimes the working conditions are what they are.

What makes it attractive, however, and we use this of course in our recruitment efforts for young mathematical statisticians or demographers, is the challenge of public service and the excitement of public service, and the sheer magnitude of it. If you're a mathematical statistician or a survey researcher or methodologist and you want to work on a big problem, you want to work for the Census Bureau. It's the only place where it goes to full scale.

We have a lot of intern programs. We bring young people in for summer internships for other kinds of programs from the universities and try to excite them about working for the government on a task of this magnitude.

Mr. Lawrence: I imagine that counting every single citizen in the United States is a costly undertaking. What kind of things does the bureau do to keep costs down?

Mr. Prewitt: It's very costly. In fact, one of the frightening things about doing a decennial is the unit cost of the first 60 or 70 percent of counting the American people is fairly low. We mail the questionnaire out, they fill it out. They send it back in, we data scan it and we've got it. When we have to go out and start knocking on the door to get that next 35 to 40 percent, the unit costs begin to go up.

Then we get to the really hard cases. See, the thing about a census, a decennial census, is you've got to count everyone, or at least try to count everyone. It's like going to the moon, you can't decide, well, you're just going to go 97 percent of the way and then stop. You got to get there. The census is like that.

The really expensive cases are those last two or three percent, people who did not want to cooperate or are hard to find or are on the move and so forth, so the unit costs go way up.

To answer your question then, we spend a lot of time and effort, public paid advertising of course and other things, to try to get as many of the questionnaires in, in what is the cheaper end of doing the census, which is to say, mail it back in. Indeed, we're very, very pleased that we managed to get more people this year to mail it in than had done in 1990, reversing what had been a three-decade- long trend. That saved the American taxpayer money.

So the first big thing to do is to sort of figure out what the unit costs are of different ways of getting the information and seeing if you can't get the proportions rearranged, and we spent a lot of time doing that.

The other thing we did -- money is time of course in any big management job -- and we kept the pressure on to try to get out of the field. You also get better quality data. The closer you are to census day, April 1st, the better the data are. But you also save money. If you let it drag out and drag out, your offices are open, managers are sitting there and so forth, so we get a lot of what we called we would have a successful office. We had 520 local offices. We'd have a successful office, it would get finished. We would then take the best people from that office and move them to an office which was slower, and at that office get rid of the people who hadn't been as productive. So constantly we're moving the more productive employees around. That saved us money. Indeed, we're ahead of schedule in terms of closing our offices, and every time we close an office 2 weeks earlier, we're just saving the government that much more money.

Mr. Lawrence: How do we ensure the data is accurate?

Mr. Prewitt: The Census Bureau has a huge number of quality assurance programs. Even after we had basically finished the major field work in the census, we'd gotten forms back from nearly all 120 million households, we went back out and did quality assurance work on 112 million households. Almost 10 percent of the housing units got revisited. Sometimes that was because we got a form that seemed to have some inconsistencies in it. On the front of the form they said there were five people living here, but we opened the form up and only found three answers. So we went back out to say did you overlook two people, did you mean to put five? So we have a large number of what we call quality assurance processes.

During the census itself, using the telephone, we would call back to the respondents and say were you enumerated, we got your form. It says somebody came and knocked on your door and found so many people. So we do that kind of quality assurance, so we find an enumerator that maybe isn't doing it right. We get rid of them very, very fast.

A census is an environment in which you have to make decisions by the hour. We would hire people at 9:00 in the morning and we'd train them that day. We put them to work the following day, and maybe by the end of the second or third day we decide they're not up to it and they're gone and then we hire somebody else.

We never stopped training. So you actually maintain a system in which you are never without an employment pool available to you. That's what we try to do.

Mr. Lawrence: It's time for a break. We'll be right back with more of The Business of Government Hour.

(Intermission)

Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence, a partner at PricewaterhouseCoopers, and tonight's conversation is with Dr. Kenneth Prewitt, director of the U.S. Bureau of the Census.

One of the things I've always found fascinating is the census is a scientific undertaking conducted in a political environment. I'm curious about how you balance those challenges as what the implications for other public sector managers are.

Mr. Prewitt: That's a very important question, of course, and Census 2000 has unusually been politicized in a highly partisan way. I think the simplest answer is that from a management point of view you have to keep insisting that the way you collect the data is a scientific exercise, the way the data are used can be a political set of considerations.

For example, there is really no Republican or Democratic way to do a war on cancer. The Democrats and Republicans may decide to differ on how much money they want to put on the war on cancer, they may decide who they want to make the benefits available to, but the actual science that goes into that by definition doesn't have any ideology connected to it, or should never had any ideology connected to it. And of course the scientists that are doing it at NIH and so forth are doing it and are completely indifferent to the partisan implications.

It's the same thing with the census. The actual collection of the data has to be done according to the most advanced, rigorous scientific methods, and that's what we have tried to do. If the political process decides to use the data this way instead of that way, we're completely relaxed with that. Our challenge is to get the data right and the let them be used in the way that the political process decides they are most effectively used.

From a management point of view then, I had to constantly try to remind our political overseers that our job is to give you the data. Your job is to decide how you want to use it, not my job. When I was asked those kinds of questions I simply said, no, I'm turning it over to you, but here's how we believe the data can be collected in the most accurate fashion.

It's a long, complicated question, and I don't want to get too deep into it, but I can only say that the boundaries between science and politics are very important boundaries for this government to consider to make certain that when you're doing science, do science, and then when you're making a decision about the applications of science, whether it's a test ban treaty, whether it's the census numbers, whether it's how to make an investment in public health, then let politics come in. But when you're actually doing the science, just do the science.

Mr. Lawrence: How do you think the census in 2010 will be different from Census 2000?

Mr. Prewitt: Quickly, I think 2010 may be radically different on some important dimensions. First, we would like by the time we get to 2010 to have taken the long form and instead of asking it in the decennial environment, we would like to ask that as a survey every month on a continuous basis. We'd call that the American Community Service Survey or continuous measurement so that by the time we get to 2010 we won't even have a long form. We will already have those data for the American people, and we've already put in place what we need to do in order to make that happen.

That means we will just be collecting the short-form data, the small number of questions and the basic count itself. We think if that comes to pass that we can use technology much, much more aggressively than we did in 2000. For example, we would expect to have many, many more Internet filers. We're an innovator. We tried out the Internet this year. About 70,000 people including myself did use the Internet to file. But we were fairly prudent. As I said earlier, you have to be prudent in the census environment to be making sure that your systems work. But we would expect to have a much higher percentage of Internet filers.

As I said earlier, a census is two things. It's both a count, and it's an assignment of you to a location. Well, we would like to use geographic positioning systems of course. We would like to have a much more refined way to identify where every structure is in the United States so it's much easier to make certain that we get people to the right address because that's really critical for the success of the census.

So I would imagine the census almost being a postcard/Internet census, a very simple census, using geographic positioning as a way to get the address file up to date, and then having already collected the long-form data on an ongoing survey basis.

Mr. Lawrence: One of the most interesting trends in government is the increased use of public private partnerships. What type of partnerships have been developed in the Census Bureau?

Mr. Prewitt: We obviously had a number of small partners, especially doing evaluation work for us. We contract with other survey professionals. But in 2000 we used the contracting partnership work much more extensively. Data collection, as I said, Lockheed Martin and TRW, that was very large. That was almost a billion dollar bit of work. All of the paid advertising was of course contracted out.

I would imagine by the time we get to 2010 that we will maintain certainly those kinds of contracts. We'll be adding other kinds of contracts. For example, if you're doing something very complicated in geographic positioning, you may well be working with other private contractors.

So I would say that the government cannot itself have all of the technological capacity or knowledge that it needs to manage large technical enterprises like the census is, and therefore it will do much, much more work with the private sector.

The census is complicated because the data are confidential. So the one thing you can't contract out is the data collection because we have such a deep responsibility to maintain the confidentiality of the data. But the processing, the address work, a lot of other features of it can be done with private industry.

Mr. Lawrence: What do you think are the major challenges that the bureau will face in the decade ahead?

Mr. Prewitt: I think one of the important things is to get the American people to understand really how important census data are to them because we do depend finally on the cooperation of the American people. If they won't answer the questions, we don't have good data. If you don't have good data, it's very difficult to do economic planning, difficult to do government.

We're at a point of history where we're almost hitting, how can I put it, sort of an impasse between on the one hand our insatiable need for information, and on the other our desire for privacy, confidentiality, and the Census Bureau is kind of caught in the middle of those two pressures. I think we have to convince the American people that we hold their data confidential. We don't infringe on their privacy, but their answers really are critical to how well this economy works.

We have just finished what could be called the first measured century, and I think there's a relationship between how well we have measured ourselves, our economic activities, our demographic and population characteristics, and how well we've built this economy, and that's going to continue into the 21st century.

Mr. Lawrence: One of the things that's come up through our conversation today is the role of technology and its importance, and how it's even helped reduce costs. How do you keep the employees trained on this up-to-date technology?

Mr. Prewitt: That's a very, very serious challenge, and we do have in-house special programs of course, but it's a fast-moving system out there. I think the Census Bureau, because it's very large, it has to be, as I said earlier, cautious about bringing in innovations before they've been tested. But we simply do it with in-house training programs and intern programs and sending people out when we can.

I do think as an agency we need a good working partnership with the places where the technology is being innovated. We do bring in people. We have seminars. We bring in people from outside to explain what they're doing. But I think we're going to have to do more of that to make certain that the Census Bureau maintains its current state with the new technologies that are going to be available.

Mr. Lawrence: Any big surprises in terms of thinking about it three or four years ago and one day being in charge and now being in charge and seeing how it all worked out?

Mr. Prewitt: I think the biggest surprise for me, and certainly the most exciting and rewarding surprise was the way in which the American people participated in this. The big difference in the 2000 census and 1990, and I think this continues in 2010, is we actually tried to design a census where we shared the responsibility with the American people, with local mayors and governors, with community leaders, with advocacy groups. It may be the first big federal task which was really done in active partnership with hundreds and hundreds and hundreds and hundreds of partners -- we had 120,000 different partner organizations with us.

As a management challenge that was tough because you had to maintain quality control, you had to maintain standardization, but you're now working with lots and lots of people who think they own the census, and so that became one of the really interesting management challenges. I think it was a very major part of our success, and I would hope that this will be just as true in 2010.

Mr. Lawrence: I want to thank you, Kenneth, for spending time with us tonight. I've enjoyed our conversation very much. Thank you.

Mr. Prewitt: Sure. My pleasure.

Mr. Lawrence: This has been The Business of Government Hour, conversations with government leaders. I'm Paul Lawrence, a partner at PricewaterhouseCoopers, and the co-chair of the PricewaterhouseCoopers Endowment for The Business of Government. To learn more about the Endowment's programs and research into new approaches to improving government effectiveness, visit us on the Web at endowment.pwcglobal.com. See you next week.