Tommy Wright, statistician in the US Census Center for Statistical Research and Methodology, joined the Stats+Stories regulars to discuss what the US Census Bureau does and why it is important. (The Census data visualization library is a treat and should be explored by anyone with an interest in viewing characteristics of the US.)
+ Full Transcript
Bob Long: Welcome the stats and stories, I am Bob Long. When you turn on the news, you often hear the story continuing numbers, the latest political poll, the employment rate, or statistics on economic growth. You may even find out 2013 is the international year of statistics. You may say "who cares, numbers are boring, they put me sleep," but the truth is the number are important and generally often struggle to find in the way to make statistics relate to your everyday life. A great example is US Census, you probably recall taking part in the Census, but you may not understand why Census statistics are so vital . We are going to explore this topic with Dr. Tommy Wright of the US Census bureau, on this edition Stats and Stories, it is a program created by Miami University of Statistics Department Chair John Bailer and media journalism and film chair Richard Campbell .They share a mutual interest in helping us understanding the importance of statistics in our lives. I will be hosting this program, but before John Bailer, Richard Campbell, and Tommy Wright join me for this show, we asked our stats and stories reporter Megan Thobe to help us understand the relevance of Census data from people who often rely on those statistics.
Thobe: Originally the Census was used to assign the right amount of taxes and representatives to each state. Today the data gathered by the Census Bureau has come to cover much more. Dr. Suzanne Kunkel, the director of the Scripps Gerontology Center, says the Census data is the basis of majority of her work.
Dr. Suzanne Kunkel: We rely on the Census data to help us track changes in the size and the growth and the speed of growth of aging populations as a demographer, the numbers are really powerful to me and the story is so compelling.
Thobe: Kunkel believes Census data can be used to tell a story about government systems.
Kunkel: In terms of the public system, Census data counts of population, projections of population have everything to do with the viability of the system.
Thobe: According to Dr. Philip Russo, an expert in political science, sometimes data even has the power to change how these systems are run.
Russo: It's become a source of benchmarking now, as more and more public sector enterprises, state and local governments find themselves doing more with less.
Thobe: Political and social systems can change based on Census data. The people who managed these systems have learned how to use the data to better those systems. Research and information from past helps people in many field answer questions about what is to come. Dr. Bill Evens is a professor of Economics.
Dr. Bill Evens: Economists do that all the time where they use historical data to generate relationships between variables and then use those relationships to forecast what will happen in the future.
Thobe: According to Dr. Kunkel, the scope of the reliability in Census data provides enough information to plan, but it does not tell us everything.
Kunkel: To have an accurate representative picture of population is really the only way we can go about planning for the future. We're pretty good at predicting the numbers, what we do about that, what that really means for what life will be like, how we spend our resources, what the tax rate will be, that is completely a political process.
Thobe: Russo says that politicians used the variety formulas to help forecast the wants and the needs of their constituents.
Russo: Census data actually becomes significant input into the formulas and calculations.
Thobe: Politicians are certainly not the only group to use Census data. With all the questions Census data can answer, Kunkel believes there is larger message in mind.
Kunkel: So many implications for policies, for public funded programs, for families, that the takeaway is that we all have a lot of to learn from each other.
Thobe: For stats and stories, I'm Megan Thobe.
Long: Thanks Megan. We also thought, though, that it might be fun to see if educated students and staff at Miami University know what the Census is all about.
Woman on the street #1: Let's see, the US Census…
Woman on the street #2 : It figures out the population and information, demographics and stuff like that.
Man on the street #1: They poll Americans to see income, how many people are living where.
Woman on the street #3: Something about populations and who lives where and what they like to do, or something like that.
Man on the street #2: The US Census exists in order to count people and determine certain demographics in certain areas.
Man on the street #3 : The Census Bureau is in charge with maintaining all kinds of both qualitative and quantitative data on the population of the United States.
Woman on the street #4 : I know that census has to do with counting people and demographics and populations.
Man on the street #4: From what I understand, it's taken every ten years to keep updated on the US citizens, and that's as much as I know.
Long: Joining me today for Stats and Stories: our regular panelists Miami University Statistics Department Chair John Bailer and Media, Journalism, and Film Chair Richard Campbell. Our special guest is Dr. Tommy Wright; he's the chief of the statistical research division at the US Bureau of the Census. He earned his PhD in statistics from Ohio State, and he served both on the research staff of the Oak Ridge National Laboratory and as a professor at the University of Tennessee in Knoxville College. He then joined the Census Bureau. He has also served as a research fellow. John Bailer, I'm going to turn things over to you for our first question for Tommy Wright.
John Bailer: This is great, and Tommy, we just want to say how we are delighted that you could join us today. It's a real treat to have you in the studio with us and have you visiting the campus here. We have a number of questions that we'd like to start with, but first, welcome.
Tommy Wright: Thank you for having me today, I have enjoyed myself so far, and I am looking forward to this conversation and exchange.
Bailer: Well great. As you heard from some of the comments that our person on the street interviews highlighted, there's a rather broad and fuzzy opinion about what the census might be, would you like to talk a little bit about what the census is and in particular what is a federal statistical agency?
Wright: Well, first of all, let me just comment on the Census. I had the fortune to speak with some graduate students earlier at lunch, and I really appreciated that. I was sharing with them that it's significant to note that the very foundation of our democracy, if you look at the Constitution; it is actually Article One, Section Two, Clause Three, so these are very small numbers. So it's right at the beginning that our Founding Fathers thought that this form of democracy was going to be based on representation. And so one key thing is that the way this representation was going to be determined was that every ten years, we were going to do a census and try to count people and assign seats in the US House of Representatives, accordingly. So that is a very key thought that there's some need for data, there at the very beginning of our democracy. I think that's a very essential idea of that, but that also leads to lots of other things we do, but primarily, the Census people that we do every ten years is an important thing that we do. And I might add that the comments, I am encouraged by the comments, I think there pretty close to what we actually do as an agency. So there is a Census Bureau, we also do a census of the economy every five years, so this is not well understood by the general public and we actually do release economic indicators and collect and prepare a lot of data that goes in the Gross Domestic Product. We are part of the US Department of Commerce and while we collect a lot of data, data are released by other agencies, and that is one example of it. So censuses give us counts of people, so I think the people we are right on, but we also do lots of sample surveys that give us estimates of characteristics of those people.
Long: You are listening to Stats and Stories. We are exploring how the statistics we hear in the news impact our everyday lives. We are talking today about the relevance of Census data. I am Bob Long with me are regular panelists Miami University Journalist Director Richard Campbell, Statistics Department Chair John Bailer and our special guest you have just been hearing from, Dr. Tommy Wright, who is the head of the US Census Bureau's statistical research division.
Bailer: You know, one thing that I would be interested in having you remind us about, is when was the first Census taken, and just give us a sense of the magnitude of this task, it was pretty remarkable you made remarks earlier today about how many people are involved in this process.
Wright: Well the first Census was conducted in 1790. And it was a conducted by US Marshals who rode around on the horseback, I understand, and sort of counted people here and there. The most recent Census that we conducted, and I should add that the first Census came away with a number about, I think 3.9 million people and it's important to note that even as Thomas Jefferson, who sort of led that first census, handed the numbers to the George Washington, he was saying, "I am sorry this number didn't quite get to four million." Four million was a significant number because it was felt to be a sizeable statement to the rest of world that there were a quite number people here, so we're serious nation. Our most recent decennial Census came away with a count of 308 million people, so we have grown quite a bit.
Bailer: Oh very good. And you had mentioned in our earlier conversation there was something like a million people we hired the last Census that worked over certain periods of time.
Wright: Right, I think so. So the way the decennial Census works, around 2010, is that we first mail out a questionnaire to each address, and we believe that we sent this questionnaire to about 135 million addresses where we think people live, and we ask them to respond to a couple questions about the people living at that address, and for people who don't return these questionnaires, we think maybe this is 47 million or 50 million addresses did not return those, we actually hire near between 600 and 700 thousand people to actually go and follow-up and knock on the doors over a period of about 8-10 weeks. So this is a very costly the thing, of course the non-response follow-up. But that's quite an undertaking. I think one of our directors says it's the largest peace-time undertaking ever for this country.
Bailer: Yeah, I think when you describe it sounds like the Census Bureau is the largest temp agency in the world for a certain period of the time.
Wright: I think that's quite right.
Bailer: We know about the decennial Census, the commonly understood exercise here, that the Census does this every ten years. You know particularly, it's involved in representation and apportionment of representation for us in the country. What other things, you mentioned the economic indicators, can you talk a little about, we jokingly talked about the idea what does the Census Bureau do the other nine years? Can you talk a little more about that?
Wright: So if you think about it, on the one hand there is a Census to count people, and on the other hand every ten years, there is a Census to sort of count businesses, different types of businesses. But we do get a little bit of information on these people and businesses during the Censuses, but in terms of estimating characteristics of these people, the other nine years, actually all ten years, we do a lot of, what we call probability sample surveys. So this is where we take a small portion of the entire list, and you ask more detailed data, sometimes these can be more complex and more complex issues. And mainly getting at, and I like to say this because I think this could be a good way to think about data, these censuses and sample surveys really capture who we are. They really tell us how many we are, actually these things actually tell us what we do as a nation, they tell us things like where we live, and also how we live. And for the Census bureau, we primarily produce around people in our country, and also around businesses in our country, if you want to be very specific.
Richard Campbell: I have a question as the journalist at the table, I know that a lot of people think just generally that the Census just tries to count everybody, and that's a superior way to do the census, but there's also sampling. I know that you've mentioned both things are done and I know just in teaching my classes, and I think a lot of journalists in general would argue, "Well if you count everybody, that's much more accurate than if you just take a sample, which is just few people." Can you explain the difference? You know, for those of us who aren't statisticians.
Wright: Yes, so let me just say that there is some wording in the Constitution which does call for and actually enumerate, I believe those are the exact words in there, and some people take that very literally, but it is difficult, and that is one thing about the census, a census is very believable, it's very credible if you tell someone I contacted everyone in the country, it's believable. If you say something like, I only take the sample of a thousand, two thousand people and I get a good reading on the country, which, by the way is a fact, you can get a good reading on some things by just taking a sample. We'll design and select the sample of the people. So there has been a problem what is the census, since the very first one of undercounting certain groups. So we are unable to get the cooperation of everyone and reaching everyone and getting them to respond in a timely manner is very difficult; we do move about a lot as a nation. So since about 1960, 1970, there is a thought that perhaps the use of sampling could help improve things a bit and there is a lot of theoretical work on this that's been done, but the Census Bureau, actually around 2000, had lots of plans along these lines, but we did not execute the plan exactly because of timing. So one thing is that methodology that we have calls for going to, as I mentioned earlier, for 2010 it would have meant going to about 130 million addresses, and then we come back and do a sample survey, and for this time, we actually did a sample survey of about 107,000 addresses. And we are able to compare these two lists and look at who is captured in the Census, who is captured in the sample survey, and we are able to come up with estimates of how many people we think might have missed, and how many people we think we might have counted more than one time. So there is a theoretical aspect that sampling can, in a complementary way, improve the counting, but actually executing it is quite a big challenge.
Long: This is Stats and Stories, the show that we hope can help you make sense about many statistics you hear about very frequently. I'm Bob Long, the moderator; our regular panelists today: Miami University Journalism Director Richard Campbell, Statistics Department Chair John Bailer, and we're exploring the importance today of census data in our lives. Our special guest, Dr. Tommy Wright of US Census Bureau. You know we also thought a little bit of fun to find out, since we asked people what the Census was, what do they think the Census Bureau does for the other nine years in between each census?
Woman on the street #5: They spy on us.
Man on the street #5: Evaluate the data, figure out what they're going to do next year, figure out how to improve their collections, stuff along that kind of line.
Woman on the street #6: They don't do anything.
Woman on the street #7: Guesstimate, I do not know.
Man on the street #6: They count all the ridiculous amount of information they got from the tenth year.
Woman on the street #8: They analyze it and come up with, I don't know, do some research with the numbers.
Man on the street #7: Uh, just maybe analyze information the information that they have now, because they do not need to update anything until the next ten years.
Long: Well I think things kind of went downhill from the first question.
Campbell: This sort of gets at, I think, the job of educators to making sure we're representing what it is that's sort of important in our culture in terms of numbers and data. This also reminds me, and this is sort of a more general question for Tommy about when you are reading newspaper articles that are representing the data, numbers, do you have some things that you see journalists get wrong a lot, things they're not explaining well? My job here is to help our journalism students, also speak on behalf of journalists, and how hard this can sometimes be.
Wright: I think journalists are doing a very good job, that's my own perception. In fact, journalists, and I will give credit on one specific example, sometimes particularly in the estimating proportions you will hear a journalist say plus or minus 3 percent, and so this is a wonderful thing because statisticians come up with estimates but statisticians always say, in addition to estimate we are morally obligated to say how good those things are. And journalists have been doing a good job of, I believe, in terms of reporting, that plus and minus 3 percent more general error, which is a big thing in statistics, regarding what we do other the ten years, I was just thinking of a list here, of things, anticipating perhaps, this question , so I did my homework. Actually these data from the Census Bureau, by the way I should add that the Census Bureau is just one statistical agency, there are many others, for various other missions of collecting data, some around health, some around education, some around energy use and various things. But in terms of the Census, some of our data are used to actually distribute and reallocate back to the population, hundreds billions of dollars each year, for various kinds of programs and projects. Data are used to plan economic development and excess the need for the schools, for hospitals, for job training, to help plan communities, to predict future needs. The actual road that we travel on every day is determined by people, city planners looking at data, public facilities, analyzing social economic trends, and just a number of things we would not believe of being affected. I have observed in my life that people make decisions, and I am impressed that politicians like to point to data. And data, we believe the data we collect are objective, and we actually do see use of data in decision-making, not just at the federal, but also local levels. So I am kind of impressed with what I see in the media and in terms of reporting something, and in particular that one example I gave about the plus or minus 3 percent.
Campbell: Very good, and I am comforted that I didn't hear spying on your list.
Wright: And that's one thing, there's an aspect of getting consent, so actually there is this idea when do the Census, we actually do contact people directly and ask those questions directly, and when we're doing the sample surveys, as well, we are contacting people directly and asking for those things as well.
Bailer: What led you to work at the Census? What gives you pleasure, what part of your job is exciting to you?
Wright: Well, since 1980, I was, I actually, I won't go into a lot of detail, but I was invited to discuss some papers at an undercount conference at the Census Bureau, and it was around the topic of not being able to count well, and I am just fascinated at the problem, as I mentioned earlier of something so fundamental as counting in mathematics, why it is so difficult and challenging and many aspects that come into this. I'm fascinated by the particular problem of what is it about counting that makes it so difficult? And can we educate, can we communicate well to the public about its need to actually participate and cooperate in terms of sharing information because information helps us live better, and getting people to see that aspect is a challenge. I like that challenge.
Bailer: One of the things we talked about is the importance evidence-based decision making, and that was a major emphasis and thrust and why we wanted to do this and develop this. If there is one thing you would want people to know about the Census, that you do not think they have a really good handle on, what might that be? By the way, you do not need to be constrained to just one.
Wright: I think it is that the importance of having information, in our democracy that information comes from people. It's timely when we get it from people while we do consider our alternatives of data, I think I like the idea of going to be the sample survey, going to people with censuses, and letting them know, out front, what is being asked, and we can do a better job in communicating what the uses are. That is task for us to work on as well.
Long: We're talking about the US Census today with Dr. Tommy Wright of the US Census Bureau. Our regular panelists are Miami University Statistics Department Chair John Bailer and Journalism Director Richard Campbell. We have time for a little bit more for discussion here to wrap things up.
Bailer: Okay, very good. I think one thing you said earlier, a critical idea, and it's the idea of probability sampling, that is something that, a lot times, we see these studies that are samples of convenience. The samples that a radio station might take or a local newspaper might take. Can you explain a little bit on the idea, what is a probability sample, and why is that generally a more reliable way of getting an opinion than these voluntary response things that we encounter.
Wright: When you do a census, there is this feeling that everyone, or every unit has been contacted, therefore, it must be the truth. When you take just a portion, and say that this represents the entire universe, or entire population, that's a bit more challenging, it is a tool called probability, which allows us to do this. If one selects one sample, you might get one result of twenty people, if you observed twenty different people, you might get a different result, so this is idea of variability comes into the play when you talk about taking a sample ,so probability allows us to say just how good the data are that comes from the sample, and also, it allows us to make an inference about the entire universe, and so sometimes there is a feeling, and I like to point this out, there is a feeling how can one thousand or two thousand people actually represent 308 million people, for example. So it's not so much the size of the sample and the size of the universe the sample is aiming to each, I'll give you an example, suppose you want to estimate the height of in the country, and suppose everyone happened to be the same height. You want to estimate the average height and if everyone is the same height, then you probably need a sample size you would say a sample size one is good enough. So already, your response that tells you that it's variability, and so statisticians have an ability to do things, and have methods that allow us to try to decrease this variability and so the information that we do get from samples, can have little variability and come pretty close to estimating the true characteristics about the populations; that's what we want to do.
Long: Richard Campbell, time for one final question from you today.
Campbell: I think my last question would be on following up with John. We get a lot of students in our classes that do surveys, they say they're surveys, you know they go out. We see this sometimes in newspapers, or you'll see it on cable TV stations, where cable TV hosts will say we're going to take a sample of our listeners. And rather than telling us why that's not statistically valid, they put it out there as something, "90% of our listeners believe this…" can you talk a little bit about what's wrong with that?
Wright: Well, you want to this sample to "represent," you want this sample to really reflect the what is in the real population. And for example, sometimes, if you get volunteers, the issue that's being discussed may only provoke volunteers to respond, so you will not be getting a true representation of what might exist in the population. It could just be the listeners that radio station, so it might be a certain type of jazz music, I don't know what the situation might be, so it may not actually represent well the true population you are aiming for. So that would be the concern there. So statisticians have this idea of randomly selecting, so I'll go back to the idea of the decennial census.It is so important, because once we have that long list of about 130 million addresses, we have tools from probability which allow us to touch that list and come away with a sample. We know how that sample was related to that 130 million addresses, and so we are able to make inferences there. So that's the preferred way of doing it. I think if you stray from that, you're less sure about what you get from the sample.
Long: Tommy Wright of the US Census Bureau thanks for your insights on Stats and Stories. Also our thanks today, panelists, Miami University Statistics Department Chair John Bailer, and Media, Journalism, and Film Chair Richard Campbell. We also want to let you know, your email comments are welcome at firstname.lastname@example.org. Be sure to visit our Stats and Stories website for more background information about today's program and guest, and to listen to previous programs in our archive. We hope you have enjoyed our monthly program, giving you the statistics behind the stories, and the stories behind the statistics