Hogan is the former chief demographer of the U.S. Census Bureau and studied at Princeton’s Office of Population Research, and its School of Public Affairs. He then spent two years teaching at the University of Dar es Salaam and working on the Tanzanian census. He joined the Census in 1979. He worked on household surveys, business surveys, and the population census. He led the statistical design of the 2000 Census. He served as an expert witness in Utah v Evans, in which the Supreme Court considered the use of imputation in the Census 2000. He served as Associate Director for Demographic Programs and later as Census Bureau’s Chief Demographer. He taught as an Adjunct Professor at the Department of Statistics of George Washington University. He is an Honorary Fellow of the American Statistical Association. He was awarded the 2018 Jeanne E. Griffith Mentoring Award. He retired from federal service in 2018.
Episode Description
Demographers study the way populations change. The things they might focus on include births and deaths, living conditions, and age distributions. In the United States, population changes are tracked nationally by the Census Bureau. A conversation with the retired chief demographer of the U.S Census Bureau with Howard Hogan is the focus of this episode of Stats+Stories.
+Full Transcript
Rosemary Pennington
Demographers study the way populations change the things they might focus on include births and deaths, living conditions, and age distributions. In the United States, population change is tracked nationally by the Census Bureau. A Conversation with retired Chief demographer of the US Census Bureau Howard Hogan is the focus of this episode of Stats and Stories, where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington. Stats and Stories is a production of Miami University's departments of statistics and media journalism and film, as well as the American Statistical Association. Joining me is regular panelist John Bailer, emeritus professor of statistics at Miami University. Our guest today, as I mentioned, is Howard Hogan, retired Chief demographer at the United States Census Bureau. He joined the census in 1979, where he worked on household surveys, business surveys and the population census, and where he led the statistical design of the 2000 census, Hogan also served as an expert witness and Utah V. Evans, in which the Supreme Court considered the use of imputation. In the 2000 census, he taught as an adjunct professor at the Department of Statistics of George Washington University, and is an Honorary Fellow of the American Statistical Association. Chance recently featured an interview with Hogan, and we're happy to have him joining us here on Stats and Stories today. Thanks so much for joining us today, Howard.
Howard Hogan
Pleased to be here.
Rosemary Pennington
Could you just describe what the job of the chief demographer of the census is?
Howard Hogan
Yeah, it's a senior research position that basically you get to research anything you want and work with whoever wants to work with you. It was a chance at pretty much the last few years of my career to mentor all the younger generation and teach them what I'd learned about census taking and survey taking and demography over my career. So it was a chance to collaborate with lots of people. It's one of the best jobs the federal government has to offer, maybe the best.
John Bailer
How did you get there? You know, can you give us sort of a synopsis of your career journey that led to this one position?
Howard Hogan
Yeah, well, after I got my degree in demography, I spent two years in East Africa working on the surveys and in the Tanzanian census. And then, after a brief time down the University of North Carolina Chapel Hill, was hired by the Census Bureau as part of the 1980 census, to use demographic methods to measure the undercount. And specifically to see if there was a way to determine the uncertainty some sort of range of various however you want to say it around those demographic methods of measuring the undercount. As part of the 1980 census, the Census Bureau was sued by Detroit, and a number of other cities about the undercount. And I got involved in that litigation at a very low level, writing some responses to their queries and whatever else. Soon after that, I was put in charge of determining a method to measure the undercount for the 1990 census that was spent the entire 1980s perfecting, well, that perfecting is maybe a bit strong, but certainly improving and the methods that were being used, and essentially, the methods we developed in the 1980s, with with improvements, obviously, are the methods that they they've been using to this date. Then after 1990, I moved over to the economic area and worked on economic censuses and surveys for a while. And then move back to the decennial area where I was, as mentioned in the introduction, the head of statistical design, not the operational design, this statistical design for the 2000 census, then back over to the economic area for a few years. And then back to being the Associate Director for Demographic Programs, which is in charge of all the household surveys, including the unemployment survey, and the crime survey and all of those, as well as the demographic estimates, the projections and the estimates. And then after a few years of that, I was offered the position to get out of management, out of supervising got to doing budgets and taking a senior research position as chief demographer, and that's pretty much what I had had was very happy to take that and that's, that's where I finished my career.
Rosemary Pennington
Howard, you mentioned the undercount a few times when you're talking about your work with the census, and I wonder if you could talk about what and what the undercount is and why it's something that we have to pay so much attention to?
Howard Hogan
Well, the census is used for a number of things, the three most important is dividing the 435 congressmen between the states, then it's used by the states to draw their congressional districts within the states. The first is called apportionment. And the second is called redistricting. And then it's used by the federal government to distribute billions and billions of dollars. So when a city, mainly cities, determined that they might have been undercounted, they felt they were cheated out of both representation and out of money in the 70s. As you may know, a lot of big central cities lost population. And so suddenly, the undercount became very important to them, because they were losing and they didn't want to lose more than that, then they actually had, in addition, in this in the slight 60s, but certainly by South by 90, at the one person, one vote was well established. And they realize the cities and states the importance of having a good census to measure that one person, one vote. So between the two, it became politically really important. The flip side of that is demographers starting with as they call but also Jay Siegel, at the Census Bureau, had been developing methods to measure the undercount. Before that, there were some sketchy ideas about how many people might have been missed. And some of the census directors pretty much implied we, we I still use we a lot, we hardly miss anybody. And it might be a few, you know, hermits out living in the desert, but really, we get everybody. And demographers pretty much prove that that was not true. Following up on that statistic using demographic methods, as you know, births minus deaths plus immigration minus emigration tells you how many people should be counted. In addition, statisticians, led by Eli Marx and others, began to use follow-on surveys, what's called a post enumeration survey, where you go out and Census Bureau goes out and does a second survey. But then they do a one to one match, to see how many people on that second survey were also counted in the census. So, the proportion of people in the second survey who were missed by the first survey gives you a measure of the undercount. Then to get it, measure the net undercount, of course, then you have to go to the Census and verify how many of those people how many of those records refer to a unique person who should have been counted and had been counted only once. So we've had these two parallel methods of measuring the undercount for many years, and I started to have one foot in each camp.
John Bailer
It seems like such a natural and intuitive way to quantify when that occurs by using this post enumeration survey. But you know, so but if it was consistent, I mean, if you had this the same degree of undercounting in urban versus rural, then maybe it's not an issue but one of the key issues was this differential right there was absolutely correct.
Howard Hogan
I mean, if it was a uniform undercount, almost all the there's a few exceptions, but almost all the uses are proportional, so it would make no difference. But historically, and census after census, African Americans are undercounted much more than white Americans, the undercount back in the 70s and 80s, the undercount of black adult males was really high, I mean, over 10%, as I recall, and then once we were able to measure the undercount of Hispanics, it's also high. Measuring the undercount of Native Americans is kind of difficult, given a number of technical reasons, but that seems to be also high. Renters are much more likely to meet us than people who live in, you know, owned homes. So there's the flip side of that, of course, college students tend to be disproportionately counted twice, once at college, where they should have been counted. And once at home, where their parents are paying the tuition.
John Bailer
You know, as you were talking about your trajectory there. And in your last piece, you mentioned that you love this work as chief demographer. What were some of the coolest projects that you had a chance to work on when you had that last position?
Howard Hogan
Well, I think the most final project was a few years ago. There were all these news articles about how this one lady born on January 1 1946, was the first baby boomer to, you know, qualify for Social Security and that seemed wrong to me. So I gathered the time series for right after World War Two, and then I collaborated with Bill Bell, who is just a fantastically good time series analyst. And between the two of us, he did the hard mathematical work. We showed him pretty precisely when the baby boom actually began, it began in July 1946. And we did this all statistically. It turns out, I love this. It turns out that's almost exactly nine months after, after the end of the war. Wait, wait, wait, didn't know that's how it turned out. But it's how it turned out
Rosemary Pennington
What propelled your interest in demography?
Howard Hogan
I had the really fantastic opportunity to study for a year at the University of Stockholm, thanks to the Rotary Club. At that time, I was interested in regional planning and city planning. And realize that if you're going to do any regional planning or city planning, you need to know some, some demography. It's pretty, pretty simple. So when I went to Princeton University at what was then called the Woodrow Wilson School, I was doing City and Regional Planning. But mediately decided, well, there was a course in demography that might as well take that. And after one semester, I had fallen in love with demography and fallen out of love with city planning. So I continued on that tract. And by the way, did you know that the University of Miami was the home of demography in the United States? I'm In Miami University Oxford, Ohio, is the home of demography in the United States as it had the first program.
John Bailer
Yeah, there's still the Wellington lecture is still hosted annually. So yeah, okay. So that, so yes, I did. I didn't know that I had some affiliation with the folks in the Scripps Gerontology center now. But yeah, that's a pretty cool story that it said, I'm delighted that you know that too. That's, that's neat.
Howard Hogan
Well, one of my professors graduated from there. So he told me, he actually taught there. So he told me–
John Bailer
You know, what was this? The thing that you mentioned, also, as part of the different responsibilities you had within the census? It really highlights something that we've talked about before on previous episodes, where we've had some of the census colleagues of yours was just the diversity of products that the Census produces. You know, I think there's the sense of the decennial census being such a well known land, and the impact of it as well. But some of these other surveys that are routinely and regularly conducted, it's not clear that they're well appreciated. Okay. Can you just give a kind of a summary of some of those the types of things the census is contributing?
Howard Hogan
Absolutely. Well, we've mentioned that every 10 years decennial census, and then we have a number of household surveys, the biggest and right now is the American Community Survey where we're out literally almost every day of the year, collecting data on veteran status, fertility, unemployment, household housing, almost every demographic topic, you can think of immigration status, mobility, and that that that is it's such a large survey, we're able to publish pretty local data. In addition, the other very famous survey is the CPS, the, the measures where we measure the under unemployment, that's a collaboration with the Bureau of Labor Statistics, but the current population survey, we go out every month and collect data on unemployment. And together with BLS, we publish it. We lot of these surveys are in collaboration with other federal agencies, they they they funded and we do it, the crime survey, community computer Expenditure Survey, a number of those, then as long as we're still on surveys, over on the academic side, we do one of my favorite surveys, which is the monthly retail trade survey. And that's really literally the very first indicator of how the economy is doing. So without 10 working days, after the end of the month, we publish economic data about what the economy is doing. And you will get to work on that survey. It was a lot of fun. When the data were released on the dot I mean, we checked with the National clock at 830 On the day it was released, and the stock market and the bond market in the futures market would have reacted within nanoseconds to our results. So it was fun to work on a survey where you just know right away everybody was eager to see what you had to say. Additionally, there's there's a number of economic surveys but additionally we do the population estimates that where we fill in, you know, the population between decennial census is working again working with our state partners to get the data and, and a lot of federal programs So we're actually based off of the pop estimates program. And as a demographer, one of the fun things to do is the population projections program, where we project what the populations look like. And that's where you will read about how for what is worth the press loves to report on, you know, and at such and such a date, the United States will become a, a minority majority or majority minority country, and this is going to be a great turning point. Those news articles are all based on our projections. And that's a lot of fun to do as a demographer.
Rosemary Pennington
You're listening to Stats and Stories. And today we're talking to Howard Hogan, retired Chief demographer of the United States Census Bureau. One of the things that I saw a lot leading up to the 2020 census was a discussion about mistrust of the census, of concern about whether people will participate, and sort of what that means for the undercount, and so many other things that rely on census data. And I wonder sort of what your take is on the mistrust of census and what can be done to sort of help alleviate that?
Howard Hogan
Yes, the mistrust is there, and especially prevalent in 2020, the voluntary participant participation in the census has been declining. The Census Bureau has an almost perfect record of protecting the confidentiality of its data, people have to go back to what happened in 1942. And world war two to find anything to point to the say the Census Bureau cannot be trusted, you know, the last 70 years has been virtually once a we never made a mistake. But it was where the Census Bureau staff was fanatical about protecting confidentiality, and how to get that message across to the American public. And it's a challenge, because in the final analysis, we're part of the federal government. And if people don't trust the federal government, don't trust the press the promises of the federal government, and they don't say, well, but the Census Bureau is completely different, even though in many ways it is completely different. We just it's a matter of outreach, I can't remember the exact numbers, but you know, the way the census has been done since 1970, is we mailed out a questionnaire and asked people to respond voluntarily before we go knocking on doors. By 2020. That was done a lot on the internet. But the voluntary initial participation has been falling decade by decade by decade, as it has in almost all household surveys. So it's a general societal change. And I don't, I wished I had a good solution. But we partner with community organizations, tribes, cities, about states, we partner with everybody we can partner with to spread the word that you know, we're good guys. And you can trust us. And I'm no longer retired. I want to be clear I'm I'm retired. So when I say we, it's my love. I am no longer speaking for the Commerce Department, the Census Bureau or, or any official capacity?
John Bailer
Well, if it's hard to imagine something more important than things like apportionment and redistricting and distributing, you know, so much money to these various districts. So, it's important work. So it's important to be heard, and people clearly care about it. I mean, you mentioned over the course of both your article and previous comments, you know, the 1980 census lawsuit, 1990 census lawsuit, 2000 lawsuit, and there were different aspects that kind of whether it was early on, you were mentioning the undercount, as being a trigger for these kinds of concerns and legal action, but later it was there were aspects of of sampling and imputation. And it was really fascinating, you know, as someone who was teaching statistics at that time to be tracking this and talking about this in classes, could you just give a little bit of a summary of, of what was the what were some of the issues that surfaced in the 1990 and 2000 Census?
Howard Hogan
Certainly, there are two sort of issues. One, there's a strictly statistical issue: how precisely can one measure it? Of course any survey, including an enumeration survey has survey area errors in it itself and whether those survey errors would swamp what you're trying to measure or not is a statistical issue that was greatly discussed among statisticians. In addition, there are two legal constitutional issues, one of which is that the Constitution talks about counting talks about an actual enumeration and counting the whole number of people. My personal understanding is When they said actual enumeration, they meant something that's not a political deal. That first apportionment was a political deal. But the courts and the Supreme Court have interpreted both of those pretty much to say, you can't you pretty much you have to cram come pretty close to a nose count. And I'll come back to that in a moment. Then when sampling was introduced in the census, the first census to use sampling was the 1940. When that was introduced, the Congress added a provision in the law saying it cannot be used for apportionment. Now, I think what they were thinking is, you know, you can't just count every other county or every other household, but that's what the law says. So there were a number of lawsuits. The final one, when it reached the Supreme Court, they basically said, we cannot use the post enumeration survey, which is based on sampling, to correct the apportionment. They didn't really address whether it can be used for other other uses. But the politics of it were such that every attempt, the Census Bureau had to use the posting ratio survey for other uses and got shut down for political and or statistical reasons. But anyway, the court was pretty clear that sampling could not be used for for apportionment, that that was that those cases were really nailed down in 1990, around 9090, then around the 2000 census, let me step back a second, the way your portion is done is a really cool algorithm, where the 435 congressmen are sequentially distributed to the to the 50 states. And so in 90 in 2000, the 430/5 Congressman, went to North Carolina, and F has been 436 that would have gone would have gone to Utah, Utah was not happy. And, they still pursued, saying, well, if you'd counted our overseas missionaries, clearly, we would have had enough to beat out North Carolina. The head of the census, J Wait, was actually a fairly senior member of that church. And, and he said, You guys probably counted those children at home anyway. But anyway, they lost that pretty court pretty quickly in court. So then they came back and said, Well, you, the Census Bureau has used a whole person. The whole person imputation is strictly speaking, count imputation. To come up with a numbers count imputation is when you knock on a door, nobody answers. And the interviewer after trying the best that she can can't determine whether it's occupied or not. Or if it's occupied, whether anybody knows how many people live there. And so there's sort of three levels of counted imputation, they can't even find the address that happens sometimes, or not sure whether it's a living thing or business, they find that they can't determine whether it's occupied or not. They or they're pretty sure that it's occupied. But they can't determine the number of people. So in each of those cases since 1960, the Census Bureau has imputed the number of people living there. Well, Utah's argument was basically that sampling, you know, it wasn't really their argument, it doesn't really matter whether you count 1% and infer 99 Or you count 99 and infer one. They're both, you know, mathematically, you're inferring the hall from the part and therefore sampling the Census Bureau said, well, no, it's not sampling, uh, for a number of reasons, very technical. And this case went to the Supreme Court. And this is really one of the high points of my career because I was the chief witness for the Census Bureau, and I got to actually sit in the Supreme Court when it was argued and work with the Solicitor General, the United States and the former solicitor general to prepare the case. And it's a case it's now taught in law school as when you take the course in argumentation, because what we came up with was just brilliant. I wish I could say I came up with it all by myself, but I didn't. In here. Here's the hairy, here's how we sold it. If you went into the Supreme Court's library, you brought it home to the Supreme Court's library and you wanted to know how many books there were and you counted the books on every other shelf and multiplied them to your honors. That would be sampling but what would you do if you were going down the shelves, and you're going, you know, volume 11, volume 12, and there's a gap there, volume 14, volume 15, volume 16, why would you infer that, you know, volume 13 was probably checked out. And you would count it that your honor says imputation. And that's the difference. And we won the case, just barely it was a very fractured decision. But looking back, if we had lost that case, at the courts had said, even this very small, and it really is, you know, let around 1% They're very small use of statistics is not allowed, then that would have, you know, really prescribed hardly any use of statistical inference for almost anything in the census, it would, it really would have handcuffed us for the use of entire administered records in 2020. So it's a very important victory and in terms of it allows us at least some wiggle room to use statistical inference to come up with the counts.
John Bailer
You know, our podcast, we called it Stats and Stories. And it was purposeful, you know, in part because the statistics behind the story and the story behind the statistics was part of what we were hoping to explore as part of this podcast. And I was , I'm delighted that you brought that up. In fact, my next question was going to be that because, you know, I love that you said two things. Well, you said lots of things I've really enjoyed. But you talked about, if no one's criticizing you, it means you're not working on anything important. That was one of the quotes from your chance piece. And certainly the work that the Census has been doing and kind of the attention it's gotten from many different constituencies is a clear demonstration of that. And you also said that some advice that you would give is no data without a story. And I that very much resonated with something that was key for us. So I want to thank you for bringing that out. And also reinforcing that.
Howard Hogan
Yeah, it's what I try to teach. And it's not always easy to come up with a story. But when I met with the Solicitor General, General Olson, that was literally his first question. We sat down at the table, his first question was, how can we put this in the form of a story or an analogy, and he was a pretty good lawyer.
Rosemary Pennington
Well, that's all the time we have for this episode of Stats and Stories. Howard, thank you so much for joining us today.
Howard Hogan My pleasure.
Rosemary Pennington
Stats and Stories is a partnership between Miami University’s Departments of Statistics, and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter, Apple podcasts, or other places you can find podcasts. If you’d like to share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net, and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.