Friday, September 26, 2003
Mr. Lawrence: Good morning and welcome to The Business of Government Hour. I'm Paul Lawrence, partner in charge of the IBM Center for The Business of Government. We created the center in 1998 to encourage discussion and research into new approaches to improving government effectiveness. You can find out more by visiting us on the web at www.businessofgovernment.org.
The Business of Government Hour features a conversation about management with a government executive who is changing the way government does business. Our conversation this morning is with Jay Waite, associate director for decennial census, US Census Bureau, in the Department of Commerce. Good morning, Jay.
Mr. Waite: Good morning, Paul. How are you today?
Mr. Lawrence: Great, thanks. And joining us in our conversation is Tom Burlin. Good morning, Tom.
Mr. Burlin: Good morning, Paul. Good morning, Jay.
Mr. Lawrence: Jay, perhaps we could start by finding out more about the Census Bureau. Could you give us a sense of its purpose, mission, and programs, please?
Mr. Waite: Well, the Census Bureau is in the Department of Commerce, as you said. The main purpose of the Census Bureau is the conducting of the census once every 10 years. The census has been in existence since 1790. The main reason for conducting of the census is it's used for apportioning of the House of Representatives. It is the vehicle by which political power is distributed in the United States.
Mr. Lawrence: How do you think about the size of the Census Bureau in terms of its staff and the people? How do you describe it?
Mr. Waite: Well, the Census Bureau during noncensus periods is about 11,000 people. That consists of about 4,500 people in the field throughout the country in our regional offices that conduct interviews and surveys, about another 4,500 people in our headquarters in Suitland, Maryland, and about 1,500 people in a processing center we have in Jeffersonville, Indiana. Now, at census time during the actual time of the census we swell up to somewhere on the order of 800,000 people in actually conducting the census.
Mr. Lawrence: What are the skills of the 11,000 people? I want to think statisticians recording things.
Mr. Waite: Actually, the skills are very diverse. In the field itself the skills are people skills that you would get from a door to door surveying organization where there are people skills, getting into houses, talking to people, interviewing, getting information, managing a field staff, making assignments. Jeffersonville, Indiana, is much more of a processing center. They have clerical skills as well as computer programming skills.
In our headquarters in Suitland we have a diversity. We do have a number of statisticians that count things and write things down. We also have a large IT component because in today's world counting people and counting in the census is a big IT operation. We have maybe 4- or 500 people who would be considered mathematical statisticians and they worry about some of the details of the sampling and the interpretation of the statistics.
Mr. Burlin: Jay, a little bit about your roles and your responsibilities as the associate director of the decennial census and ell us a little bit about the activities of your office.
Mr. Waite: As the associate director of the decennial census I am the chief government person relating to the census. I am responsible for all of the people, the money, and the processes of the census, which is the largest peacetime activity that the federal government produces and conducts.
As the associate director I'm responsible to make sure that the budget is procured, that we've got the right budget, that we have the plan. I endorse the plan and take the plan to the various segments of government to make sure that there's a plan laid out, approved, and funded by the Congress to conduct the census.
Mr. Burlin: I happen to know from our relationship that you've spent your entire career at the census.
Mr. Waite: That's true.
Mr. Burlin: Can you tell us about the positions that you've held through your career and what the census was like when you first joined? What got you interested in joining the Census Bureau?
Mr. Waite: I've always been very interested in mathematics. I've been very interested in statistics. I'm a great sports fan and being a sports fan you get a lot of interest in statistics. What really got me interested in the Census Bureau was the birth of our second child. I was in graduate school and trying to keep from starving to death and my wife announced that we were going to have another child so I started looking very hard for a job.
The Census Bureau offered me a job as a computer programmer in 1971. So I left my home in northern Utah and traveled back to Washington, DC, which was quite a change in culture and in climate, and began working there as a computer programmer. In those days it seemed like the census was a pretty safe place and a good place to work. I have worked in several different venues since then. Although I did work as a computer programmer I had been trained in mathematical statistics so a short time after arriving I got a chance to work in a research program as a mathematical statistician which I did for several years.
I've had several other jobs in the Census Bureau. My current job, I took over as chief operating officer of the census in 2000. I started there in 1997 and I was in charge of all census operations. Then in 2001 I took over as the associate director, which is the senior director on the executive staff at the Census Bureau for the census.
Mr. Burlin: 1971, that's a while.
Mr. Waite: It is a while.
Mr. Burlin: There must have been some pretty dramatic changes in technology and techniques over that period?
Mr. Waite: Huge changes. Since talking to you from IBM one of the big changes I think about is I came as a computer programmer, as I said. Our computer programming at that time, we used to write code on a FORTRAN work sheet. We'd take it to a separate staff who would key it up on a punch card. That punch card would be transferred to a small 2A tape which they would update. Even reassembled programs were all done on a big mainframe with these steel tapes. Maybe two or three days later you would get back your code indicating that you'd left a comma out or you'd left a period out. You compare that with today when everybody has a computer on their desk and we don't think anything at all about on the spot editing.
So things have gone a lot different. They've moved a lot more. Our data dissemination program where we used to print volumes and volumes of books about the census, and that's the only way you could find out about the census, that's what it was like in 1970, perhaps books piled up the height of a human being, now we've turned that in 2000 into instantaneous Internet response of all that data in everybody's home virtually instantly, big changes, particularly, I think, in the IT part of life.
Mr. Lawrence: Let me ask you about your career from a management perspective. You began as a computer programmer but one day you became a manager of other computer programmers or the like. Could you tell us about your step to management and especially dealing with highly skilled people, what that was like?
Mr. Waite: Management was an interesting adjustment. It's one thing to be responsible for yourself and making sure that you're getting your own work done and when you have that job management always seems like an easy job. The manager doesn't have anything to do. He's just sitting there in the office. I've found managing and trying to learn to motivate people to be a big challenging of management, to try to help them figure out how to feel the same enthusiasm for a task that I did.
I didn't find too much of a problem managing highly technical people. I think I've been confident myself about my own abilities. I have a masters degree. I didn't have a PhD. I tend to approach people thing saying I don't really care what your degree is; I'm interested in what you know. I have found, however, that more educated people often times are more knowledgeable but I haven't really had a big issue with trying to motivate them and manage them from a technical or highly educated perspective.
Mr. Lawrence: How about the scale of what you're doing, especially when it grows to the 800,000 people, how do you manage such a large group?
Mr. Waite: It's a massive undertaking. The census itself is a set of relatively simple tasks but it's so huge. It's the massiveness of it. It isn't particularly hard for somebody to get a meeting together of 10 people at 2:00 o'clock in their office but to get a meeting together of 150,000 people at 2:00 o'clock in your office suddenly changes the whole structure of things.
We have a very good management team but clearly managing something as big as the census requires a lot of interaction with a lot of people. I have about 15 direct reports during the census time. But the big thing about the census is we are spending money at just unbelievable rates during the census. Twenty-four hours or forty-eight hours delay in something can cost us huge amounts of money. It's one thing to talk about a census where you say I'll get something from the field. Can we hire an extra person in each one of our offices? One of the first things I had to realize is sure, we can hire another clerk. I found that turns out to be 521 clerks because there are 520 local census offices.
So even relatively insignificant expenditures or what seem like they're insignificant turn into massive things in a hurry. If you decide everybody needs a PC you're talking about 800,000 PCs. It suddenly changes the whole complexion of the day.
Mr. Lawrence: How about just the frequency factor? Your analogy to sports is quite interesting in that you practice every day and in baseball you play a game every day. You only get to do this once every 10 years. How do you get a rhythm?
Mr. Waite: That is a big challenge. You really don't get a rhythm. You spend a lot of time testing and preparing to try to get everything worked out and tested as best you can. We have found that without extensive testing even relatively simple changes in procedures can be disastrous.
As an example in Census 2000 very late in the census process somebody had the great idea that if an individual wanted to have a questionnaire in a different language maybe we should put a little note on our advance letter that we send to every house in the United States in five or six different languages saying if you're interested in a questionnaire in this language please tear off this sticker and send it back to us. That was a change made late in the day. Everything was fine. We tested it and we thought it was great. It turned out we woke up on the morning of March 13, 2000, realizing that every address in the United States with the labeled advance letter on it had a leading 1 in front of the address. In other words if you lived at 51 Oak Avenue you now lived at 151 Oak Avenue. With that big of an operation when you make a little mistake like that you make it really big. Fortunately for us the post office was able to read the barcode which accompanied that address but we came to a very near disaster of having virtually every one of our advance letters undeliverable.
Mr. Lawrence: That's an interesting point especially given the size. What are the other logistics of running the decennial census? We'll ask Jay Waite of the Census Bureau when The Business of Government Hour returns.
Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence and this morning's conversation is with Jay Waite. Jay is the associate director for the decennial census in the US Census Bureau which is part of the Department of Commerce. Joining us in our conversation is Tom Burlin.
Mr. Burlin: Welcome back, Jay. The census has been called the largest peacetime initiative of the US government. Can you describe the logistics required to conduct the census?
Mr. Waite: Well, a little bit. The census is conducted mostly with rented logistics and facilities. Just to put that in a little bit of perspective, during a census we have 12 regional census centers, 520 local census offices. A local census office is about 16,000 square feet of rented space with communications facilities in it, four data capture centers which are basically shopping centers that we've bought and gutted and put data capture equipment in them. We printed over 20,000,000 paper maps in order to assist the enumerators in going out and doing the nonresponse follow-up part of the interview.
We had 10 major IT systems and several secondary systems. About 15,000 personal computers were purchased just for the census, about 13,000 servers, and over 16,000 telephone lines. We hired 860,000 positions for temporary six- to eight-week jobs. So we'd hire them, train them for three or four days, and send them out to do mostly nonresponse follow-up work, 860,000 people who we didn't know the week before, we trained them and who knew at the beginning they had a six- to eight-week job. Managing that staff was a bit of a challenge.
We conducted the census in 49 separate languages which meant we had to have different pieces of paper and different language skills on the part of our enumerators. We had 27,000 local questionnaire assistance centers, John's Barber Shop, Mary's Beauty Parlor, the 7-Eleven, the service station, where we put the questionnaire, we put a little staff there. You can walk in if you needed to and they would help you fill out the questionnaire, give you some answers about how to do the questionnaire.
We answered 5.8 million telephone calls in about nine days from questions that people were calling in about. It came all in one spike as we mailed out all these forms. We captured data from 1.5 billion pieces of paper from March through August. To put that into a bit of a perspective, a billion pieces of paper is 17 stacks of paper the size of the Sears Tower, and we had 1.5 billion pieces of paper that we processed from March through August 2000 in these four big shopping centers.
We printed over 400,000,000 questionnaires and within a year we tabulated data for 9,000,000 census blocks, 51,000 tracks, 39,000 governments. Our data capture centers that I talked about just briefly, they were basically very large shopping center kinds of places where questionnaires would come in by the truckload every day, several tractor trailer loads of questionnaires. They had to be opened, sorted, and captured with optical scanning characters.
We employed about 2,000 people per data capture center. As I've mentioned before, we processed about one and a half billion pieces of paper during summer 2000.
Mr. Burlin: So we're all drawing these mental pictures right now of huge stacks of paper and tractor trailers lined up for miles and miles. Most of us aren't thinking right now about 2010 but I'm sure you're thinking about that every day. What are you plans for the 2010 census and how will it differ from the 2000 census?
Mr. Waite: I'm glad you asked, Tom. The 2000 census was probably the best census ever. As far as coverage was concerned we believe that at the end of the day based on a lot of detailed evaluation we may have actually overcounted the population by less than one-half of 1 percent nationwide. It was a very, very high-quality census from a wide range of avenues.
But it wasn't perfect. The census design is that we count everybody on census day, and approximately one house in six gets what we call the long form. That's where we get the detailed data for journey to work, for education, for income. That data is the only source of very small area data about your block or about the building you live in and it's really valuable for local governments and businesses to plan to understand what does this community look like and what are its characteristics.
That data now comes out only once every 10 years. That's not often enough. If your neighborhood looks exactly the same as it did 10 years ago then once every 10 years is a great thing if you're a planner. But I used to say a lot of times in earlier discussions in 2002 the most recent data that the average mayor in America had about the details of their community was collected in 1990. That's just not cutting it.
So one thing we're talking about in the 2010 census, we're changing that paradigm. We're eliminating the long form from the regular census collection and we're collecting data in what we call the American Community Survey where every month we'll be talking to 250,000 households throughout the country. Rolling that data up over a five-year period will give us 15,000,000 long form sample cases which is a sufficient size to produce small area data every year with a five-year rolling average. People will no longer have to worry about having eight-, nine-, and ten-year-old data to make those kinds of decisions.
Another important thing we're doing is that we have a large database. Virtually every address in the United States is a part of our database for mailing our census. We have accompanying that an electronic map, basically a street center line map, called TIGER. Some of you may have seen or used TIGER/Line files which is the basis for Mapquest and many of the other commercial mapping processes that you see and use all the time.
That file is very useful and very good and we use it to produce these 20,000,000 maps as well as to provide mapping characteristics around our data but that data is not in true GPS alignment. It's proportionally correct, it's relatively accurate to one another, but it was drawn in many cases by hand and digitized. We have a process going on as part of the 2010 census to go through every county in the United States and align that TIGER with true GPS alignment. That's going to mean that the crown jewel is that when we're doing the short form only census this time when you don�t fill out your form and we need to come and do nonresponse follow-up instead of coming with a paper questionnaire we'll be coming with a small handheld PDA with a GPS transponder on it that will allow the enumerator to find your house with your GPS coordinate, conduct the short form interview, transmit the results of that interview back to our headquarters, eliminate the data capture of all the nonresponse part, and basically take care of what I like to say we're replacing electrons for trees.
We're trying to be responsive to figure out some way to save money. The American Community Survey is a wonderful survey but it's expensive. We think it will cost somewhere on the order of $1.7 to $2 billion during the decade. Realigning the map TIGER is a wonderful thing but it costs money. It's going to be on the order of $500- to $550,000,000 during the decade. So we need to save the money in the collection of the short form. Billy Sutton was asked one time why do you rob banks and he says that's where the money is. When I look to saving money I look to the field.
In 2000 the census cost $6.7 billion. About 70 percent of that money was spent in the field. So if you're going to save serious money you have to look to the field. The field spends money on three things, people, gobs and gobs of people, space, and paper. During the 2000 census we printed 400,000 questionnaires and listing sheets and addresses that they're supposed to go to and these 20 million maps. So you get this big volume of paper. Then you have to rent space to put that paper in and you have to hire people to move that paper around. So if you're going to attack the cost in the field you have to attack paper.
We're attacking paper with what we call mobile computing devices, a small PDA, electronically transferring the map as well as the assignment of which particular house your assignment is back and forth, eliminating the use of paper except for the initial mail out and mail back. That is the main thrust of our cost containment activities that we're approaching with the 2010 census.
So we're already testing it. We're beginning the process of 2010 even as we speak. We're doing our first test of these mobile computing devices in the Borough of Queens, New York, and it's a fairly challenging place to take that process, find those addresses, and do that work. So this next spring we'll be having our first census test in Queens, New York, and in three counties in southern Georgia beginning that process.
So we're really excited about the upcoming census. We think it's going to be different in some important ways. We believe that in the future the census will look, act, walk, and talk a lot different. We want to continue our move away from paper and more toward technology and electronics.
Mr. Lawrence: How do you count people who just don't want to be counted? For example, they might have issues with the IRS, they don't want government intrusion, a host of reasons, but they just don't want to be counted?
Mr. Waite: That's one of our big challenges, of course, and we try to do it in a number of ways. We were very successful, we believe, in 2000 with a big outreach and partnership program. For example, people living in those communities that feel threatened in some sense by the government, perhaps they're here illegally or they're not too keen about the government knowing where they are, we tried to work with advocacy groups that knew them getting people that they knew to go into their communities and work with them and help them and help them to understand that we are not interested in whether they're citizens, we are not interested in whether the IRS is after them, we keep our confidential. We use partners throughout the country in a lot of ways, individual advocacy groups or others, to work with and try to encourage people to be counted.
We weren't perfect on that but we made big inroads on that and that was a big reason why we were able to reduce our undercoverage as much as we did and we expect to do that even a lot more. Our interviewers themselves we hire indigenously so that you're talking about people that live in the community are coming to your door. We don't have Jay Waite driving through a neighborhood where he's never been before in a US government car saying please come see me.
But we try to reach down and plus we spend a lot of time and money really on advertising to try and help people understand that a good count in your community is important to you. It will get you the resources or the statistics that you need for schools, for roads, and for other things. We had our advertising campaign for separate languages and for separate markets and for people that we thought were difficult to count.
Mr. Lawrence: How do changes in technology affect the census? We'll ask Jay Waite of the Census Bureau for his perspective when The Business of Government Hour returns.
Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence, and today's conversation is with Jay Waite. Jay is the associate director for decennial census in the US Census Bureau which is part of the Department of Commerce. Joining us in our conversation is Tom Burlin.
Mr. Burlin: Jay, you talked about the enormity of the task, but from a practical perspective it's almost impossible to count every single person. The success of the census is sometimes measured in how close the undercount of the minorities is to the population as a whole and you've already mentioned that 2000 was one of the most successful censuses in history. This undercount was the lowest ever. What does the Census Bureau do to reduce this differential undercount of the minorities?
Mr. Waite: Well, this is a big challenge for us, Tom, of course. When you conduct the census basically the census is a very interesting count of all the addresses and the people living at those addresses. If you have populations that don't live at the addresses the way you might think it makes it more difficult to count. Minorities are more likely than nonminorities to be living more than one family in a building, in an apartment, or in a house and you may or may not catch that. They may be living in several places and they're difficult to catch on any one might where they might be.
We've had a long history since we started measuring this differential undercount in 1940. Coverage has improved in every census but the difference between the minority and the nonminority populations has remained very persistent, about 3 or 4 percentage points difference in that coverage, and we've tried a lot of things. In 2000 we believe we made some really serious inroads. We didn't eliminate that difference but we cut that difference almost in half.
We did it by a large number of different initiatives. Number one, we pushed the language program very heavily. A number of minority people are not counted merely because they see a census questionnaire, they don't really understand what it's asking, they're not sure what is meant, and they don�t get around to answering or responding.
We heavily advertise. We had a fairly big advertisement campaign in 2000 targeted toward the minority populations in their languages by people they knew in their cultures. So we had television, radio, newspaper broadcasts focusing on those minority groups. We also worked with partners throughout the country in those minority communities trying very hard to encourage them to understand the importance of being counted and of trying to figure out how to count them better and how to include them in the census process. In the end we had a combination of these things which we know made a big inroad as we measured it in the count of the minority population.
There is still for a lot of reasons a difficulty in counting everybody and we don't count everybody. The goal of the census is to count everybody once and only once and in the right place. It is not a matter of just counting them but we need to put them in a very small geographic area. We do a very job of counting them but some people we count more than once and one of the big issues, one of the big stories, of the 2000 census is the number of people that were counted twice.
That happens because, for example, a child away at college gets counted at the university. His father who is paying the tuition bills says I don't care where he thinks he lives. As long as I'm paying the bills I'm counting him back here and we sometimes aren't able to link those two up. Children of divorced parents, in many cases both parents are counting those children. The snowbird population, we count people sometimes in Florida and in Michigan because we end up overlapping their trip.
But we have made progress in the minority population. Overall the national undercount or the national count was measured as a slight overcount of 49 hundredths of 1 percent. But some minority populations were still undercounted by as much as a percent and a half. So there is still that difference but we believe we've cut it dramatically. We expect to spend more time and more money and more effort in 2000 on this partnership, advertising, and pushing down to those communities to help them to understand that counting them is really in their best interest.
Mr. Lawrence: Let me shift to the people on your team and especially concerns with the federal retirement wave. I'm curious as to how you're worrying about that especially given the loss of experience and expertise.
Mr. Waite: This is a big problem, I think, in all federal government and a particularly big problem at the Census Bureau because we talked about some sports examples but we only play once every 10 years. It's very important that we have some kind of institutional knowledge from the previous time. That's what we always worry about. I'm one of them they're worrying about now, long in the tooth. Am I going to be around long enough to do the next census?
We have an active plan at the Census Bureau, succession planning, where we're lining up younger people to get training and to work. My immediate staff, I have eight people that work for me as immediate reports. Four of them are certain to be here in the next census and I'm working closely with them to make sure that they get in positions of experience and try and train them but this is a big issue.
Another piece of our aging work force that we have is that our enumerators that we hire during census time, five decades ago there was a large pool of very well-educated, reasonably young housewives that didn't work very much out of the home that a six- to eight-week job was something they could do and do very well and they were easily trained and they did a wonderful job for us. More and more we are now looking for that work force in the retired population and in the very young population because many of the people in the middle-age working groups are already working and really don't have the time to be able to get off to help us.
We're concerned a little bit as we look at this new technology that we're employing with the human factors about a PDA or a Palm Pilot or a program like that. Will that be a problem for older people that would be helping us with their finger dexterity and with their eyes and those things. So one of the issues. We're quite comfortable I think that the technology side of that will work but we're very interested in will the human factors become an issue. They do become an issue sometimes even with pencils and paper and if we put a small computer in front of them with a small screen and little buttons that require nimble fingers will that become an issue for our nonresponse follow-up activities? It's something that we're worried about even at that level as well.
Mr. Burlin: Jay, the Census 2000 was noted for its use of contractors in the operation of the census. What functions did they perform, was it successful, and do you plan changes to your 2000 practices in 2010 based on those experiences? Is the bureau addressing the president's management agenda in competitive sourcing requirements?
Mr. Waite: Yes, we are, Tom. You're right, 2000 was a big paradigm shift for the Census Bureau. We have for many years believed that we were the only people living or dead that actually knew how to do a lot of these things. We had an awful lot of in-house systems, we did everything ourselves, and we felt that the private sector could not possibly be able to figure out how to do this complicated and honorable job that we were doing.
For a lot of reasons we were encouraged and almost forced into a model that said we needed to contract out in some cases some of what we had considered before to be our core competencies. So we took a bit of a leap of faith. We hired contractors to do our data capture. For example, we invented the Fosdick machine, the early cameras, as so we had built all that inside and we felt surely this newfangled thing called optical scanning, newfangled for a government agency, they could never figure out how to read those 400,000 pieces of paper or 400,000 questionnaires in a few days. We contracted out for data capture and found that to be very, very successful.
We contracted out for our big telephone banks where we send out 110 million questionnaires one day and if 1 percent of the people receiving them have a question you've got a big phone line issue. We had tried to manage that in the past ourselves. We turned that over to a private contractor and to the call scheduling business and basically took over that business for a few weeks but, again, that was done by contractors.
In our data dissemination process we also had contractor help. Where before we had attempted to program and code a lot of tables and to produce a lot of data first in books and then in CD-ROMs we got contractor help in order to do that. We're able to do things that we had never done before. The census produces huge amounts of data and our eyes are always bigger than our stomachs. We always have hoped to produce all this data and never quite got around to it. In 2000 for the first time in the history of the census we produced all of the census data, short form, long form, broken out by the different race groups and by the Indian-nonIndian tribal groups. It's all completed as of today. The official census of 2000 ends September 30th. This is the first time in the history of the Census Bureau that we got all that done. We would not have gotten that done for sure had it not been for contractor support.
So it was really a good combination. I think that we learned, surprising to us, that it is possible for a contractor to hire people to do the work that we thought only someone with 80 years of government experience could possibly think about doing. I think some of our contractor friends learned that there was a little bit more to doing a census than they might have thought.
Based on that we're expecting to continue to move forward and probably do much more contracting in 2010 than we did in 2000. I spoke earlier about exchanging electrons for trees. We're almost certainly going to go outside of the Census Bureau to seek the expertise to develop those systems which will be passing the data back and forth electronically rather than to try to hire in-house expertise ourselves.
The cyclical nature of the census makes it very good to think about contracting. For a period of two or three years we need huge amounts of energy, expertise, and people. If you hire them permanently through the federal government you've got a big issue with RIFs or with long-term people and not knowing what to do with them. One of the happiest days of my life is when we finished the data capture centers, we closed down the data capture, and I went to one of the data capture centers we had in Baltimore and I told the contracting personnel get these people out of here. I didn't have to fire anybody, I didn't have to pay any payroll, I didn't have to move any desks. I just drove home smiling all the way home and somebody else took care of that problem.
But I think there's a lot more things that we'll be doing in contracting as we look forward. It was something that we went into with a little bit of hesitation. I think we came out of it not perfect, we're not great contract managers, and everything didn't work well and we learned from each other, but I believe in the end we had a very positive experience.
Mr. Lawrence: That's a fascinating point. Surely with all that data privacy is a concern. How is the census data protected? We'll ask Jay Waite of the Census Bureau for his insights when The Business of Government Hour returns.
Mr. Lawrence: Welcome back to The Business of Government Hour. I'm Paul Lawrence and this morning's conversation is with Jay Waite. Jay is the associate director for decennial census in the US Census Bureau in the Department of Commerce. Joining us in our conversation is Tom Burlin.
Mr. Burlin: Jay, the scope of the census has expanded beyond just the enumeration of people for the sole purpose of apportioning seats in Congress identified in the first article of the Constitution. What do you think about this trend?
Mr. Waite: Well, it's a great trend as far as information. The census becomes a great treasure trove of information about a lot of different subsets of our population. It does complicate the task of doing the census. For example, it's not sufficient any more that you just count a person once and only once and in the right place. You have to determine what their ethnicity is, what their racial background is, where their parents came from. You have to worry about whether they were housed in rented units or owned units, whether they have a house or they don't have a house, what their educational attainment is. It has become a form of identity. A lot of people use the census to establish who they are.
Just a couple of examples about that, it's very important to large segments of the Hispanic community that they are able to identify the origin from which they came. For example, you might be from South America but people in the Dominican community in New York would like to know how many Dominicans there are. Often times as in many groups the young people have long since lost any particular identification with the home country but their parents want them to be part of that. That complicates the process because it turns out that it's not one census. It's seven, eight, or nine censuses because you're trying to do different things.
Language is another big issue. There are clearly parts of the population who do not speak English well enough to do the census and they need language help. There are also groups in the population who really like the idea that a federal agency would have a questionnaire printed in their language. So that gives the census a great mosaic and a great picture of the United States but it does complicate the process. Counting people is difficult but counting them and labeling them and identifying them in the various groups that people want to be identified in adds to our challenge on a day to day basis.
Mr. Lawrence: Jay, no conversation about the census would be complete without talking about sampling. So I'm curious. In 2000 sampling was proposed rather than actual or in addition to actual counting and there was a belief that the sample would yield a more precise result but the Supreme Court got involved and struck this down. I'm curious. How should we think about sampling being used to determine the accuracy?
Mr. Waite: Well, actually to be clear there were really two separate issues with sampling. One was the sampling for nonresponse follow-up. In other words if I've sent you a form and you didn't answer rather than to go to every door that didn't answer maybe I could go to a sample of those doors and get enough information for the count. That use of sampling was primarily a cost savings. I spent a lot of money with people driving to your house and trying to find you home and making two or three trips there, finally getting you and getting the interview. If I only had to go to a sample of those places it would save money.
That's the part that the Supreme Court struck down, that the Census Act did not allow sampling for purposes of apportionment to decide how many congresspersons went to New Jersey or how many went to Ohio or how many went to Georgia. We were hoping to save money on that. The other piece of sampling which was much more contentious is a coverage measurement evaluation. What that really meant was that for about 300,000 housing units we went back after the census was conducted. Independent of the census we went back to those housing units and redid the census and then compared those results with the census results. When they differed we went back yet again to those houses and adjudicated those differences. From that we could estimate the proportion of white males between 18 and 29 that were missed or double-counted in the census.
So the idea was and has been for two or three decades now if you could estimate those proportions at the national level or even at some level smaller than the national level maybe you could then use those estimates of undercount to adjust the census count. In other words if you estimated you'd missed more black males between 20 and 39 than you had white males you might increase black males wherever you'd actually counted them, increase that count to make it compensate for that undercount.
That was not prohibited by the Supreme Court but it was very contentious and the Census Bureau spent a lot of time, money, and effort working on that. In the end we chose not to do that adjustment because we could not convince ourselves that the adjusted numbers were actually better. There is some error in the survey itself. The one big problem I talked a little about earlier about duplication, if you're the child of divorced parents and both parents count you the theory behind this follow-on survey is that if I come back to your house after the census and I explain real clearly who you should count and who you shouldn't that if you've made an error in the census you will see the error of your ways and you will tell me the truth and that difference then will allow me to find out that there was an undercount or an overcount.
What we found all too often is when we went back the second time they told us exactly the same thing they told us the first time. When the survey and the census agreed the survey was blind to the fact that there was duplication. So part of our evaluation of that survey we came to the conclusion that we could not in fact convince ourselves that adjusting was going to be better than not adjusting. That was a long and turbulent process and a lot of the issues associated with the census were colored by that as the different political parties were concerned about whether we did that survey or how we did it and what we used it for. But in the end we did not use it.
We used sampling extensively in statistics and we used sampling extensively in the Census Bureau. But in the end something that is so important as to say I need to count you ideally once and only once and in the right place and I'm going to use that for the distribution of political power, even the smallest amount of sampling error and uncertainty for these small levels just turned out in the end we were not able to compensate for it and we did not use it for those small areas.
Mr. Burlin: Jay, what do you say to the citizenry, the people who feel the census is just too intrusive? You've talked about the long form questions, asking everything about your income, your ancestry, all the way down to whether you have an indoor or outdoor bathroom facility, perhaps. How do you answer those folks about it being too intrusive to their lives?
Mr. Waite: I believe that the census clearly asks for a lot of information. We try very hard to make sure that we don't ask for information that is not really useful, number one, for government planning and, number two, we have an absolute fixed prohibition against disclosing any of that information about you to anyone for any reason for any purpose. Probably our number one article of faith in the Census Bureau is our confidentiality. Your data will never be shared with anybody, government or private, but we do need that data, what we collect, in order to determine where to put roads, where to put schools. That data, and people talk about the famous do I have indoor bathrooms or outdoor bathrooms, those statistics are the major source of data for the government on the quality of the housing stock, where in my county or where in my city do I have serious problems with housing. I don't believe the mayor cares whether we have plumbing, indoor or outdoor at 401 Elm Street but if there is a section of the community that has serious problems with housing conditions, has serious problems with transportation, if there are large numbers of people that are commuting to work and there's no bus line over there, those statistics can be used to your advantage as a citizen to be able to help make the case.
We don't believe it's overly intrusive. We don't ask everyone those long form questions, and we're now reducing ourselves down to 250,000 households per month, 3,000,000 households per year, out of a population of about 125- to 130,000,000 households. From that we believe we can get the information that local governments more than federal need for small area decisions and for small area data.
Mr. Burlin: You've talked about the focus on privacy and privacy protection. In today's era of focus on privacy is there increased pressure on the census to protect this data as it relates to such things as immigrants and other definitive data on folks?
Mr. Waite: I don't know if there's increased pressure. We have always considered that. In fact we collect data for the census under a law called Title 13 of the US Code which forbids by law from sharing that data with anyone for any reason. We've always taken that very, very seriously. People worry in the day of homeland security and other things whether there have been attempts at incursion in trying to find out about our data. I can tell you that nobody has asked us for any information so I don't see that as a big issue.
I believe that most places in the federal government realize that our ability to collect data is so critical that you don't want to compromise that by some short-term advantage. So we've not had any issues. We've not been approached and have not had any challenges at all to our ability to keep your census questionnaire and census data totally and completely confidential.
Mr. Lawrence: Jay, as we established in the first segment, you've had a long career as a public servant. I'm curious as to what advice you'd give to someone interested in a career in public service.
Mr. Waite: Well, public service is a very interesting and exciting field. It has been very good to me. I would say if you're a young person and you have talents and you have interest and you want to work on something that makes a difference you should consider public service as a career whether it be a career of two years, ten years. I'm an old codger. I've been around for over thirty years. There is a lot of interesting work. It's working with people. It's working on real things. The federal government is, I think, a great place for a young person to find out what they can do and for some of us old codgers we've made it a career out of our whole lives.
Mr. Lawrence: I'm afraid that will have to be the last question, Jay. Tom and I want to thank you for joining us this morning.
Mr. Waite: Thank you. The data from the 2000 Census is available. You can get on the website www.census.gov and click on the American FactFinder and that has all the data summarized that were collected from the census about your town, big or small, and you can learn a lot about where you live and the neighborhood in which you are and compare it with the rest of the country. I'd encourage you to look it up. It's there, it's free, thank you.
Mr. Lawrence: This has been The Business of Government Hour featuring a conversation with Jay Waite, associate director for decennial census in the US Census Bureau of the Department of Commerce.
Be sure and visit us on the web at businessofgovernment.org. There you can learn more about our programs and get a transcript of this very interesting conversation. Again, that's businessofgovernment.org. This is Paul Lawrence. Thank you for listening.