Category Archives: Census

Brookings map shows MSAs mapped the right way

The previous post discussed the problems with maps of Metropolitan Statistical Areas that showed the extent of the counties included. It included this map from the Census giving the misleading impression that the area of the greatest population increase was in the southwestern United States, the large dark purple area:

Percentage Change in Metropolitan and Micropolitan Statistical Area Population: 2000 to 2010. Source: Metropolitan and Micropolitan Statistical Area Population: 2000 to 2010. Source: U.S. Bureau of the Census. 2011. Population Distribution and Change: 2000 to 2010. 2010 Census Briefs.

Source: U.S. Bureau of the Census. 2011. Population Distribution and Change: 2000 to 2010. 2010 Census Briefs.

In that post I also presented an alternative map of MSAs that avoids this problem and referred to a note with a more extensive discussion of these issues. That note ended with my doubts about whether the way MSAs were mapped would change. I was wrong.

A recent post by William Frey of the Brookings Institution looked at recent population changes from a new census report (US population disperses to suburbs, exurbs, rural areas, and “middle of the country” metros). It includes a map illustrating population changes from 2016 to 2017 for the 100 largest metropolitan areas:

Source: William H. Frey analysis of U.S. Census Population Estimates, released March 22, 2018

Source: William H. Frey analysis of U.S. Census Population Estimates, released March 22, 2018

Hooray! Point symbols are used to map the MSAs and their population changes. The map does not give the misleading impression of the Census map shown above.

One very minor suggestions to Brookings for further improvement: The points appear to be located at the centers of the MSAs. It would be even better if they were located in the vicinity of where most of the populations of the MSAs were located.


The problem with maps of Metropolitan Statistical Areas

Metropolitan Statistical Areas (MSAs) are one of the most important statistical units for the reporting of data by federal agencies. It is common to see maps of showing those areas, such as this map from the census illustrating changes in population from 2000 to 2010:

Percentage Change in Metropolitan and Micropolitan Statistical Area Population: 2000 to 2010. Source: Metropolitan and Micropolitan Statistical Area Population: 2000 to 2010. Source: U.S. Bureau of the Census. 2011. Population Distribution and Change: 2000 to 2010. 2010 Census Briefs.

Source: U.S. Bureau of the Census. 2011. Population Distribution and Change: 2000 to 2010. 2010 Census Briefs.

The MSAs are colored with various shades of purple, darker for greater percentage population change. (The green areas are the Micropolitan Statistical Areas, the smaller analogues of the MSAs.)

A large area of MSAs can be seen in the southwestern United States, extending from eastern Arizona to the Pacific (and up through California). But much of this area is hardly metropolitan in any meaningful sense. It is sparsely settled mountains and desert land. The only reason this is within MSAs is that MSAs are composed of entire counties. And many counties in the West are very large, resulting in such empty areas being included along with the portions of the counties that are truly metropolitan. (See my earlier post on not calculating densities for MSAs for further discussion.)

This not only affects perceptions of the extent of metropolitan areas, it can also be quite misleading when MSAs are used to map data. Consider the map shown above. The darkest purple areas are the MSAs that had the greatest percentage population increase over the decade. One immediately observes the very large dark purple area extending across southern California, Arizona, and even into Nevada and Utah. So is this where large population gains were most dominant? Looking more closely at the map, you might notice that numbers of the MSAs in Texas are also dark purple. In comparison, these areas look small and scattered. But they include all of the largest MSAs in Texas and many smaller ones. And the Texas MSAs in this highest growth category had a total population in 2010 that was over twice that of the highest-growth MSAs in the southwest. Sizes of MSAs can be very misleading.

But MSAs do include entire counties, so what is the alternative for mapping these areas? One possibility is to use point symbols in place of the county areas for showing the presence and location of MSAs. And if one wants to show some characteristic of the MSAs, the sizes of those symbols can be varied. This is an alternative map of MSAs with the symbols graduated to show the population sizes of the areas:

Metropolitan Statistical Areas by Population 2010.

Metropolitan Statistical Areas by Population 2010.

A more extensive discussion is in the note, “The Problems with Maps of Metropolitan Statistical Areas,” which can be downloaded here.

What you can’t say about variation within metropolitan areas

The New York Times recently published an article titled (on the web) “Why Outer Suburbs in the East and Midwest Have Stopped Booming.” It’s a curious piece, based on data by county across the U.S., showing those counties in which deaths have exceeded births. Indeed, the article includes an animated map of the entire country showing such counties from 1991 to 2016 and notes that over 1,200 counties had more deaths than births in 2016. This map of the counties, by the way, provides absolutely no way of identifying the counties in the outer suburbs of metropolitan areas or, for that matter, metropolitan area counties in general.

But the primary focus of the article, from the first paragraph to the last, is on what the title refers to as “outlying suburbs.” The article states that “about one in four outer-ring suburbs were experiencing more deaths than births, including 18 of 30 such counties in New York, New Jersey and Pennsylvania,” equating the outer-ring suburbs with counties. I won’t quibble about the lack of any definition of how a county is considered to be an “outer-ring suburb.” This is, after all, a piece of journalism, not a scholarly article.

The problem with drawing the conclusion about regional variation in births and deaths in outlying counties (claimed to be an issue in the Northeast and Midwest) is that this requires the existence of such “outer-ring suburb” counties. And the reporter’s view does not go that far west of the Hudson River, as is so often the case with the New York Times. Note the reference to the counties in New York, New Jersey, and Pennsylvania.

In order for a metropolitan area to have outer suburban counties, the metropolitan area must have a sufficient number of counties so that these can be distinguished from the inner suburban counties and from the urban area counties. For this to be the case, the metropolitan area must be fairly large and the counties must be quite small. And this is the problem. The Northeast and Midwest have numerous metropolitan areas that meet these criteria, for which one can draw such conclusions about “outer-ring suburbs.” Some metropolitan areas in the South also meet these criteria–the Atlanta metropolitan area would be a prime example, consisting of many counties.

But metropolitan areas in the West, where counties tend to be larger–often much larger–have few, if any counties that could be considered to be “outer-ring.” The San Diego Metropolitan Statistical Area (MSA) consists of one county. The Los Angeles MSA has 2 counties. San Francisco-Oakland MSA has 4 counties, but none could reasonably be considered to be “outer-ring suburbs.” If one took a more expansive view of the metropolitan areas than the MSA definition (which I would advocate), it might be plausible to identify a few counties as being outlying suburban counties. But even then, these would be few and isolated. Nowhere near the number of counties for which you could draw conclusions like the 18 of 30 counties in New York, New Jersey, and Pennsylvania.

Bottom line: It may be the case that “outer-ring” suburban counties in the Northeast and Midwest are now experiencing more deaths than births (though this is necessarily a weak conclusion given the lack of a definition of those counties). But stating this conclusion implies that this is not the case in the South and West. And it is impossible to draw any conclusions about “outer-ring” suburbs in most of the West based on county-level data.

You cannot say anything meaningful about variation within metropolitan areas across the U.S. using county level data. Differences in the sizes of counties and the very large counties in the West make this impossible.

More on population-weighted density

Some time ago I did two posts on population-weighted density, which is the mean of the densities of small areas such as census tracts weighted by their populations. In the first post, About population-weighted density, I described this alternative to traditional, conventional density and discussed some of the issues surrounding the use of this measure. The second post was Population-weighted density and urban sprawl, which argued that both conventional density and population-weighted density were appropriate measures of the extent of urban sprawl, relevant to different consequences associated with sprawl.

Doing these posts got me more interested in population-weighted density and led to my writing a full paper exploring this alternative density measure. In addition to expanding on the topics addressed in those two earlier posts, the paper provides much new information. Two of the highlights are the demonstration of the relationship of population-weighted density to conventional density and the comparison of conventional densities and population-weighted (actually housing-unit-weighted) densities across the 59 large urban areas in my urban patterns research.

In the first post I said that to the extent that people are more heavily concentrated in some tracts with higher densities than in others, those higher density tracts will be given more weight in calculating the population-weighted average. This will cause the population-weighted density to be greater than the simple or conventional density. In the paper, I move beyond this qualitative statement to deriving the mathematical relationship between population-weighted density and conventional density. It is actually quite simple: population-weighted density is equal to conventional density plus the variance in density across the subareas divided by conventional density. To the best of my knowledge, this is the first time this relationship has been demonstrated in this manner.

Looking at conventional and housing-unit-weighted densities for the large urban areas, the distribution of the weighted densities is more highly skewed towards higher values, with New York being an extreme outlier on weighted density. The housing-unit-weighted densities were more strongly related to the size of the urban areas, including size in earlier years, suggesting that the presence of areas of concentrated high densities was established in some urban areas decades ago.

Much more information on can be found in my paper, “On Population-Weighted Density” which can be downloaded here.

Have we had our last good census?

As readers of this blog know, I make extensive use of census data in my research. Also, I am very concerned (some would say obsessed!) with the details of the census and the data I am working with. Given this, I am concerned about the prospects for the 2020 census and am wondering whether 2010 saw our last good census. A combination of the general state of the nation combined with the actions (and inactions) of the current administration is producing this concern.

Obtaining an accurate census counts depends fundamentally on the cooperation of the entire population. And this cooperation is dependent upon people having a basic level of trust in the government. I don’t think I need to elaborate on the general decline in such confidence. I would especially note such things as the the rhetoric against various groups, the travel ban, the cancellation (at least until now) of the program for the dreamers, and many other things. If I were Hispanic, an immigrant, or especially if I were undocumented, I don’t believe I would respond to the census.

Actually, I might respond while taking the risk into account. And I’m sure that this would be the case for others as well. Because of the importance of the census for redistricting and the allocation of funds, perhaps I would choose to respond but not truthfully and with lots of kids to up the count in my area.

The cuts in funding for the census will weaken its preparation and ability to plan and conduct effective outreach. Especially given the shift this census to conducting the bulk of the enumeration online, I am concerned that this could especially affect the count of lower-income, less-educated persons. Those with higher levels of online sophistication will have no problem filling out the forms. Others may fall between the cracks, census follow-up efforts notwithstanding.

Finally, there is the pernicious effort on the part of the current administration to include a question on citizenship on the census. First off, there is no reasonable need for this information from the census. The American Community Survey asks this question and provides all of the detail needed for any conceivable purpose (and in a more timely fashion than the census as well). I can see only two reasons that the administration might be pushing for the inclusion of this question. The more “benign” (!) purpose would be to discourage non-citizens, especially those who are non-documented, from participating in the census. Even more ominous would be an intent to violate the confidentially of the census to use the information for immigration enforcement purposes. I believe that just the proposal to include the question, even if it does not become part of the census, will further erode trust and participation. Inclusion of the question in the census would be a disaster.

Some urban researchers are careless…and wrong

I have read a number of scholarly articles in which the authors were using census Urbanized Area data from 2000 or later in which they described those areas as consisting of territory with a population density of 1,000 or more. And that is incorrect. The density threshold for adding blocks or other small areas to an Urbanized Area (or Urban Cluster) is 500 persons per square mile. I’m not into naming and shaming and won’t. But come on! If you can’t even describe the data you are using accurately, why should anyone trust anything else you are saying?

I know where the error comes from. Starting with the 2000 census, the Census Bureau dramatically changed how they defined the notion of “urban” and Urbanized Areas (for the most part greatly improving the definition). Under the old definition, it was the case that a small area had to have a population density of at least 1,000 persons per square mile to be included in an Urbanized Area. An excellent summary of how the census definition of “urban” has evolved can be found here.

I assume that a researcher making this error had read earlier articles that described Urbanized Areas as consisting of areas with densities of 1,000 or more (either correctly, if referring to pre–2000 Urbanized Areas or incorrectly, if referring to the later areas). I expect this would be the source, not the census definition of the earlier Urbanized Areas, for if these authors were too careless and lazy to look up the definition for their current work, they likely would not have done so in the past either.

The current Urbanized Area density minimum plays a key role in the definition of urban areas for my urban patterns research. And of course I am continuing to read new articles that are published that deal with urban patterns, including those using Urbanized Area data. The first few times I read articles referring to the 1000-person-per-square-mile cutoff for 2000 or 2010 Urbanized Areas, I panicked. Did I make a mistake in understanding the definition and get it wrong? (It is a complex definition.) Each of those times I went back and re-read the formal notices on urban area criteria for 2000 and 2010  in the Federal Register. After having assured myself several times that I was correct, I no longer have to repeat this.

Technical note

The 2000 and 2010 urban area criteria do make use of a population density minimum of 1,000 persons per square mile in the first stage of the delineation process. An urban area core is defined that includes small areas with population densities of 1,000 or more. Then additional areas are added with densities of 500 persons per square mile and above. The existence of an initial urban area core meeting the higher density threshold will not be an issue for Urbanized Areas.

On the choice of Combined Statistical Areas

Last year, I wrote a post discussing why I chose to use the larger Combined Statistical Areas (CSAs) for my urban patterns research rather than the commonly used Metropolitan Statistical Areas (MSAs). I followed this up with a second post giving examples of how the sharing of transportation infrastructure–commuter rail and airports–could be an indicator of the integration of areas that should be considered together as a single, larger metropolitan area.

This decision to use the CSAs is of such fundamental importance to my research that I felt it deserved more extended, formal treatment. I prepared the paper “On the Choice of Combined Statistical Areas” that provides greater background, covers the topics addressed in those blog posts in more detail, and addresses some other implications of the the choice of CSAs over MSAs. It also shows how the CSAs are comparable in extent to MSAs as they had been defined earlier for the 2000 census. This last topic was also addressed in an earlier post.

The paper is posted on the Research page of the website and can also be downloaded here.