I am very pleased and proud to present the Research Access interview of Dr. Robert Groves, Director of the U.S. Census. I conducted the interview on-site at the Census Bureau headquarters on May 23rd. I would like to thank Dr. Groves as well as staff members at the Bureau who made this interview possible.
Here’s the video of the interview:
Here’s the transcript of the interview:
Good morning, Dr. Groves. Thanks for your time today.
It’s great to be with you after all these years.
I think people are very interested to hear the path of how you got to where you are. How your career progressed and became the director of the Census Bureau.
I’m not sure a lot of people are interested in this, but I’ll answer your question. It wasn’t a linear path. I never thought I would be in this job, first of all. To young people watching this, I think I’m not a believer in planning your life out. So it wasn’t one of those kind of progressions.
I went after a bunch of interesting problems, I think. So I had formative experiences as an undergraduate at Dartmouth where I was exposed to measurement of human behavior and attitudes. And I thought that was kind of interesting.
It was at a time when that was being linked with more high speed computational analysis of data. So that kind of came together. And I pursued that throughout my life, finding problem after problem with human measurement in general. So that prepared me in some way that wasn’t clear at the time for the job I have now.
And who influenced you?
I had a wonderful teacher who still lives. His name is James Davis. He’s a sociologist, but at one point was director of the National Opinion Research Center. He was spending some time at Dartmouth when I was a student. And he changed my life as professors often do to individual students.
And what interests you about — you’ve spent much of your academic career studying surveys and survey error. What about that is interesting and exciting to you?
Well, I thought, decades ago, that there was a gap in how we think about measurement of human behavior. And the gap had to do with the statistical analysis of data on one side, and the behaviors and thoughts of human beings that produce the data. So that interface– how people make decisions to participate in a survey in most countries of the world. Once they are participating and they listen to a question, how do they process the meaning of the question and answer the question?
All those things I found fascinating. And what was missing, I think, or what I wanted to plug away at, was the bridge between those social and psychological influences on behavior and the statistical data that came out on the back end. I wanted to try to put those together. Turned out to be a lifelong work.
And you spent most of your career prior to the Census Bureau in academia. What was it like coming into the directorship of a large organization doing practical surveys?
Well, I was director of the Survey Research Center at Michigan, which is, for academic things, is a pretty big place, but nothing like the Census Bureau. So I felt then and continue to feel at home at the Census Bureau intellectually. The basic work of the Census Bureau of conceptualizing measurement, drawing samples, collecting data, processing data, and producing estimates– all of that is something that I love. I love all of those pieces in different ways. So even though this is a very big organization and my position doesn’t allow me to get into the nitty-gritty of everything, it is that process that I feel aligned with.
And you’ve said that later this year you’ll be joining Georgetown University. Congratulations.
As you look on your career so far at the Census Bureau, what are some of the things, the accomplishments that stand out most?
Well, I don’t view any of these as personal accomplishments. The one thing that’s humbling about being the director of the Census Bureau is you quickly realize that there’s someone in the Census Bureau who knows more about every particular topic you’re dealing with than you do.
And I think the accomplishments or accomplishments of the institution– I remind us that in 2008 and 2009, most of the common press on the Census Bureau was that it was headed for a disaster in the 2010 census. That actually energized a whole bunch of people in the country who care about the Census Bureau as an institution. And that was related to my decision to come here, clearly.
But it turned out that all of those predictions were wrong. The team that was assembled late in the decade to run the 2010 census was really much, much better than anyone knew at the time. They had organized things in a way that made for success. So I was the beneficiary of good things that were done before I got there.
And in the 2010 decennial census, what would you say were the main accomplishments in how that was run?
Well, there were a variety of design decisions that were made way before I came. So this movement to a short form census is good for public participation. That’s unambiguous. The use of a replacement form if you didn’t respond initially, although that generated a lot of grumbling, actually helped the return rate. We had a return rate of this mail questionnaire that was higher than in 2000.
You have to be inside the survey biz to understand what an accomplishment that was with response rates falling throughout the decade. We had a bilingual form. That was a good thing. That helped Spanish-speaking folks, and I hope we expand that.
And then, just yesterday we announced the statistical evaluation of the census through the post-enumeration survey. And this is one of the best censuses this country’s ever seen.
Great. And in 2010, you also did a lot of outreach to minority groups. How did that go?
Well personally, this was the most wonderful thing I’ve ever experienced in my life. So I was all over the country meeting with small community groups of every ethnicity and every language group that you could imagine. I learned a lot about the country, first of all. Then this was just a personal gift to me.
But I also realized– in a way that all of my colleagues realize here at Census now too, I’m convinced– that when we do effective outreach to communities, when the communities think that it’s their census, that they’re actually participating, it just works better.
And with a country like ours, very diverse, both in terms of language and culture, we just have to do that. And I think we now believe this so fully in our bones at the Census Bureau that we’ll do better and better going forward.
It was in the news recently that the number of minority births is estimated to have passed the number of white births. And this, in a country that, at its founding, had a very different attitude toward minorities. So have you reflected on how far we’ve come in those?
Oh, yeah. And back to– my reflection on this topic was heavily affected by my experiences in 2010. And one of the things that is happening right now in this country is that new immigrant groups are not just in the big cities on the two coasts. They’re everywhere because of modern transportation and job opportunities and so on.
And so the absorption of these new immigrant groups, the socialization of these groups, isn’t gonna be done in big urban areas as much as it was in prior waves. It’s gonna be done in little villages throughout the country. That fact alone means that the socialization of these groups is gonna be different than in prior decades. And it’s gonna be fascinating to see what the country is like, what the second generation of these new immigrant groups will be like 20 years from now.
What remains to be done at Census?
Oh, there are tons of things to do. The work of the Census Bureau and the work of the survey industry in general is never done, because unlike the hard sciences, the things we study are constantly changing. So if we don’t change our methods and our approaches, we fall behind. And we’re attempting to do that at the Census Bureau in a variety of ways.
One of the things we’re moving to very, very aggressively is to take advantage of other data resources, to take advantages of new modes of data collection. So our vision of the future is that the stock surveys that we were all taught and that I taught to you, I guess, at one time– those are gonna morph into a much more complicated beasts involving multiple modes of data collection, administrative data, or auxiliary data. And the estimates that we make off of those will rely on these multiple data sources simultaneously.
That’s a revolution, in terms all of survey work. It’s going on now throughout the world. The Census Bureau has to be part of that to stay relevant.
Yes, and you’ve announced plans to use web data collection in the 2020 census and other important surveys in the meantime. Tell me about that thought process and what you’ve done and are doing.
I think the first thing to note that isn’t well known is that the Census Bureau’s been using internet surveys, especially in the economic area, for many, many years. We have a lot of experience with it. We’ve solved the IT security problems to the level that we need as a government agency. So we’re just expanding those capabilities.
The decision to do this for 2020 was a really easy decision. That part is trivial. The problem with 2020 is it’s so far out that what the internet will be and what devices we will use to access the internet are really hard to predict. For a government agency this is hard, because we’re used a long planning horizons. And we have to stay more nimble because of this.
So we’re anticipating an internet and the accessing of the internet that’s entirely mobile, wireless, and ubiquitous. And so preparing for that world is what we’re trying to do in 2020.
Will the Census ever be all web-based? When the population chooses to respond all-web, we’ll be there, hopefully. But the findings from around the world on internet response to government agencies surveys are pretty clear. It’s not a magic bullet at this point. It is an auxiliary mode. There’s a lot of substitution behavior going on. That is, if I filled it out on paper, I may choose, if given the option, to go to internet. But there aren’t fantastic jumps in response rate because of internet. It’s another tool in our tool kit, and we need to take it seriously.
The recent conference of AAPOR, the American Association for Public Opinion Research, had a really unprecedented level of focus on new methods like web and even mobile. What’s your feeling about those methodologies and how that’s all evolving?
Well, these are tools. As tools of data collection, we have to be there. There were a lot of good sessions on mobile computing and how we can use handheld devices for measurement. That has to continue. So we’ve developed, here at Census, a Center for Applied Technology, where essentially one wall of the center is devoted to every mobile device on the market. And the purpose of that is to make sure we stay current with new devices coming on the market, to see how our software preforms on those new devices. We just have to make that part of our ongoing business.
AAPOR was also debating sampling issues, that is, how do you– in this mobile world, how do we actually link up data collection mode with sampling modes. And that was part of the controversy there too.
And that’s more challenging, I would imagine.
Yeah, so there were a big debates– it has been going on in the industry now for, I guess it’s maybe nine years or so– about whether volunteer internet panels can be used in some fashion and maybe added or supplemented by other sources of data to produce estimates that are accurate and reliable over time. There were big debates in the sessions.
And with the decennial census– will that ever be conducted using sampling instead of enumeration or in combination with enumeration?
It could be. That would require, I think, either very deliberate action on the part of Congress expressing its will to do that, but also probably some attention in the Supreme Court about a ruling that took that off the table, sampling for estimation in a census. So you can always conceive of this. But right now the legal infrastructure is not there.
Why has it become more difficult across the board for researchers to convince people to participate in surveys?
Well, that’s a huge question. As you know, I care about that question and I spent, I guess, 20 years trying to answer that question. At the end of those 20 years, I think, there are a lot of different forces that are moving in the same direction.
One of the ubiquitous correlates of participation is urbanicity. People in urban areas are less willing, less able to be accessed and less willing to respond than those in other areas. We’re becoming more urbanized as the population grows.
People who live by themselves tend to be less willing to respond. The young people that live by themselves are hard to access. The old people who live by themselves are less willing to cooperate. So there are social and demographic forces going in that direction.
There are a lot of hypotheses that we don’t have good data on, but they’re popular, like we’re over survey. This is something that the survey industry, in its navel-gazing often asks. There’s very little evidence of that. But that might be true. It’s a hard hypothesis to test with real data.
And then there are, as we become increasingly diverse, northern hemisphere developed countries absorbing populations of new immigrants. That too produces issues of whether folks coming in with different cultural backgrounds are reacting to survey requests and in the same way.
So it’s a very, very complicated problem. How it affects the quality of survey estimates is also a very complex thing, as you know.
What would you say to an average American who has questions about the work that the Census Bureau does and in terms of the value that it provides to the American people?
Well, I’m happy. And I have these discussions with people, as you might imagine, from day to day. And I think the first thing to observe is that modern societies run on data. All of the business sector, the private sector, has built more and more management information systems that are quantitative indicators of what’s happening, how things are going.
Countries, too, have this same thirst. So policy-making in most developed countries require measurements on the nature of the thing you’re trying to effect. We run on data. So then the next question is, given that the society runs on data, how do you get these data? How does this work?
Now some data come from natural processes that are ongoing. And government records provide them. But then we’re missing measurements on a lot of important aspects of the country. And we need surveys to do that. So if you want the benefits of surveys in terms of better decision making, both as your community evolves but as your business evolves and so on, then somehow you have to make the connection that when you’re asked to do a survey, that those benefits derive from your personal decision to do that.
So one of the challenges of the Census Bureau is to make that argument to a lot of different publics in a way that’s convincing– that your decision to participate in the survey is linked to the quality of data that the country has to run on.
And how about people that have concerns about privacy-related questions, that are uncertain about some of the questions asked in Census Bureau surveys? What would you say to those people?
Well, we know– I mean, we all share these feelings, right? There’s a part of us that, when any question that is private is raised, we ask of the questioner, why do you need to know that? That’s a legitimate question, first of all. We all need to allow ourselves to feel this way.
Our obligation, I think, as a government statistical agency, is to convince people that the value of their answer, when put together with a bunch of other people’s answers in guiding the answer to the question of how things are going in the country– that’s really important. And that’s why you should do it.
And then we need to say one other thing and we do, I think, well. And that is when you tell us that answer, we have a sacred oath and with legal mandate that we can never reveal that and never associate it with you ever again, that that’s kept confidential.
And the wonderful thing about the Census Bureau is we have a wonderful record of complying with those laws. And that’s something that we’re proud of and can say strongly that, even though this question might seem a little intrusive, you’re telling us your answer will never be revealed in any way and will be used for the good of the country.
And as you go back into academia, what will you bring from your Census Bureau experience to that new role?
Well, I’ve been amazed and disappointed, I guess, with the use of statistical data in real life decisions. So I think that bridge needs help, societally. And the bridge requires the creation of a generation of people who will go into business and go into government and go into the nonprofit sector aware that they need data to make good decisions. And so, returning to academia is part of my quest to try to make that bridge a little stronger.
What are the types of things it’s important for an educated individual to know about social science and surveys and those topics?
I think that basic lessons of statistics need to be more fully dispersed in the country. So it is important, as a resident of any country, to be able to query a number that’s purported to be true, to be critical of it. And the basic lessons of how measurements are done on humans, and what “sampling error” means, and how question wording and measurement tools affect the outcome– the more our society knows that, the more they can absorb and critique different alternative estimates. This is a good and important thing to try to achieve.
Any plans for another book?
Well, yeah I probably have another thing in me. I won’t talk about it now. I like to do book-length things because it gives you a time, enough space, that you can fully develop arguments.
So when people look back on your stewardship of the Census Bureau, what would you like them to conclude or to say?
Well, I would be overjoyed if, 10 to 15 years from now, you can see ingredients of the things that we planted. The seeds that we are planting right now, if they mature and do indeed, as we hope, be part of the process of building a 21st century Census Bureau, that would make me quite happy. But we won’t know this for a while. All the things we put in place, or many of them, I think, will take a while. They’re culture-changing things, and it will take a few years to know whether we’ve succeeded or not.
Dr. Groves, thanks for your time today.
Great to be with you.
In my post announcing this interview, I solicited reader help creating questions to be asked. Our winner was Susan Perkins, who commented on that post, suggesting the following question: “What are the chances that the census will ever be conducted through sampling rather than as an ‘actual enumeration?'” You will see that I asked Dr. Groves Susan’s question, though in the heat of the moment I altered the wording slightly – but I think the basic meaning came through.