Screenshot from the New York Times

I thought I’d pass along a data reference tool that I’ve found very useful in understanding our city and region.  A few months ago the New York Times launched a project called Mapping America – Every City, Every Block.  The maps use data from the 2009 American Community Survey to display basic population data (density, race/ethnicity, income, education etc…) but they are especially useful in their use of automatic scaling; the maps adjusts your viewing for either census tract or county depending on your level of zoom.

It’s always nice to have a reliable and easy-to-read data source to use in urban research.  Enjoy!

34 Replies to “Urban Maps for Geeks”

  2. Also, check out both the near-universal statewide decrease in income on the county level and how many of the tracts from Tukwila and Seatac south have suffered significant decreases. And why is there an elevated number of gay couples near Spokane?

    1. Spokane has for years been actively courting gays to revitalize parts of the downtown… I could bother going to Google it, but I won’t.

  3. I wouldn’t put a lot of stock in this “data”. I pulled up my neighborhood and it shows there are 75 white folk and 25 Asians living in Viewpoint Park. I’ve only ever seen one vagrant living there. There’s also hundreds of people living in Bel-Red, apparently in their cars.

    1. The dot maps are generated by figuring out, for each region, how many dots of each color there should be, and then distributing them quasi-uniformly across that section of the map. It’s not supposed to be accurate when zoomed in closely. Note the very large population of Cougar and Tiger Mountains.

      1. “Mapping America – Every City, Every Block.” Pure hyperbola, it’s a random scatter by census tract. Bridle Trails is divided into two census blocks. When you look at the map it shows the same density between 140th and 134th (exclusively R-1) as between 140th and 148th which is half multi family. But to the west of 134th which is primarily R-2.5 it’s deserted. To the north the same thing with Rose Hill which takes all of the data and peppers Bridle Trails State Park with population. It’s so grossly inaccurate that it’s fraudulent to present it as block by block data. The zoning data is all public record. If they’d put just a tiny amount of work into this it could be semi accurate. As it is they shouldn’t have tried to present a zoom level anything greater than regional. If falls apart even at county wide mapping. Another clue is the mapping of the waterways. Census tracts accurately follow the shoreline of Lake Washington and Lake Sammamish but not Lake Union.

      2. I never seriously expected this page to show me block by block data, for much the same reason as I’ve never expected to win a Harrier jump jet by purchasing cans of Pepsi.

  4. I’d wait for the actual census tract subzip breakdown of the longform results.

  6. Personally, I think these census maps about race are basically nonsense (the ones about income, housing size, education, etc. are useful). Race is becoming ever more meaningless with the (rapidly) growing population of multi-racial people, and the way the government counts race can’t take this into account.

    Which is why you get confused data like this. “Most kids are minorities” or something doesn’t take into account a lot of “minorities” are in fact not minorities by cultural definitions (if not by census definitions). It makes no sense to describe my daughter as 100% asian any more than it does to say 100% european, but the government doesn’t agree. And if her children were 25% asian, the government would still count them that way. Kind of silly, and makes these maps increasingly suspect.

      1. But if you select “white” and “asian” you count as 100% asian in their data. The NYT had an article recently about a woman who was 1/4 chinese and 1/4 cuban, and she counts as “Asian” with “Hispanic ethnicity”.

  7. Not to be a downer, but identifying people by location and “race” may not be such a good idea.

    I seem to recall that in the mid nineties in Rwanda, the authorities ordered schoolteachers to start grouping their kids by ethnic group, so as to make certain near-term plans easier to carry out.

    Can somebody tell me any really good use for a map like this?

    Mark Dublin

    1. It can tell you if non-governmental forces are operating to segregate people by race or some other prohibited factor. For decades, the French government prohibited anyone (including private parties) from collecting racial statistics, although the reason given was to do with national unity or some other waffle. That law certainly didn’t stop the decades-long ghettoization of African immigrants, nor the riots that erupted a few years ago. It just meant that when the problem finally exploded into violence, no-one had any real data on it, just ball-park estimates.

      1. Bruce, you mean like Cleveland now? Put in 44120 (Shaker Hts., OH) and zoom out a little. You’ll observe the exact same map you would have observed if you had “decades-long ghettoization of African immigrants.” Unfortunately, the people of east Cleveland are too busy working three jobs to riot.

      2. A fact that we can state with certainty thanks to the data collected and published by our government.

  8. Mark,
    Arguing for *no* data on race runs the dangerous likelihood that racial inequities and oppressions are harder to prove with any empirical data. Arguments for not collecting demographic data on race are typically made by conservative, white folks and their representatives.

    Also- I believe ACS is not actual empirical data; it’s based on a 5 year average. So the dots you’re seeing are estimates of *the tract over a time period*; not a snapshop of individuals (like the decennial census data).

  9. This data shows my neighborhood of Capitol Hill near the freeway as the densest neighborhood in the entire state. Curious to see if this holds up to scrutiny.

    1. It does, think of anywhere else in our state that would have a greater population density?

      1. I believe the Western Slope has been the densest area in Seattle for decades. That area is both very built up and almost entirely residential, unlike the other built up areas of city like Lower Queen Anne (lots of offices), Pioneer Square (lots of offices and unused buildings), Belltown (lots of parking lots) and Downtown (almost entirely offices). Nowhere outside of Seattle even comes close, not even John Bailo’s much loved East Kent Hill.

    2. Yes, ‘ol tract 74 has been our densest for a long time. This time they split it into two tracts, 74.01 and 74.02. One of them is over 50,000 per square mile, by far the highest in the western continental US outside California. The other (guessing the north part) is over 40,000/sm.

      Other neighborhoods don’t have such convenient census tract alignments that allow for easy comparison. Half of Belltown is connected to a SLU-dominanted tract that goes to Mercer & Westlake. Lower Queen Anne’s tract is diluted by Seattle Center. But we have tracts in the 30,000+ range in Belltown and the U-District.

      I haven’t seen 2010 numbers for Portland, but the Pearl might give Seattle’s tracts a run for their money. In 2000, their densest tract was apparently about half 74’s density.

      1. Fairly unlikely since The Pearl is split between Tracts 5 and the 51 and neither tract is residential only. Further, the Pearl had a lot of in foreclosure/vacant condos rather than rentals in early 2010. No doubt the situation has improved but I’d take a developed rental over that at any time.

        Tract 48 and 56 remain the most dense in Portland from what I’ve seen.

  10. This map is somewhat misleading as the U.S. Census does not consider “Hispanic” to be a “race”. “Hispanic” is an “origin” or “characteristic” that gets applied to Whites (George Lopez), Blacks (Luis Tiant), Asians (Filipinos, who could also be classified as Pacific Islanders) and Native American (Indigenous peoples of Latin America), according to the U.S. Census.

    But the U.S. Census does not ever explain how they will interpret data on the census forms lest they “bias” respondants’ reporting.

    Also, this Census allowed for multiple race reporting, and is self-reporting.

  11. An even better map is just out which has actual 2010 census data:

    http://projects.nytimes.com/census/2010/map

    Click on “view more maps” in the top left and it has population density too, though it’s a bit useless as it colors almost all of Seattle with “5000 and higher” people per square mile.

    1. I’m surprised that percentage wise the south sound and Whatcom County outpaced King, Snohomish and Pierce. And of those three King was the lowest growth rate (11.2% for the county vs Seattle’s 8%) and only Snohomish exceeded the State average of 14.1%. But Everett, although growing faster than Seattle was only half the growth rate of the county as a whole. And Tacoma grew at an anemic 2.5%. The only county that grew more than 20% and could be considered to be in a major metro area was Clark (Vancouver, WA). And it grew faster than any of the Oregon counties around Portland. I guess folks like Washington’s income tax and Oregon’s sales tax! The remaining three boomers are east of the mountains.

    2. Oi, the map of change in vacant housing units is scary. At least most of Bellevue is doing OK with the exception of downtown which for now is massively overbuilt.

      1. Hey, the housing units seems to be done by blocks instead of full census tracts, that’s pretty cool. If you look at SLU for example (98109) there are only dots where there’s housing for the most part. The lakefront is an exception… my guess is that all the Eastlake houseboats are being spread across the full shore, including a bunch of dots in Lake Union Park!

      3. The housing units data is much better than the population. They seem to have at over layed current use data from the property records. For example the’re no units shown in Viewpoint Park which is a tiny pocket park. It’s tough to get to fine a representation when the minimum increment seems to be 12 housing units per dot. It’s still skewed by census tract tough and not looking at zoning as they have less density in areas that are R-2.5 than those that are R-1 and the dividing line is the census tract. For downtown Bellevue it’s pretty accurate.

  12. I found this fascinating, whatever techincal/statistical flaws there may be. At first I was focusing on the concentrations, which seem more defined where the density is higher (New York and environs most striking). Then I started looking for integrated areas. I found that Pasadena where I went to college is a good example. Also the decades old sprawl SE of Renton is looking a lot like a melting pot — if the diverse peoples are actually interacting, going to the same churches, having sex, etc. Makes me wonder what Kent is like.

Comments are closed.