Gubernatorial Predictions, 2006.
Kos (9 pickups):
NY, MA, OH, AR, MD, CO, NV, RI, MN
Leip (5 pickups, note reversed color scheme):
MN, AR, OH, NY, MA
Update. Alaska was one other state Kos identified as a potential pickup, given that the GOP incumbent has a favorability rating of minus fifty. And it looks like there will indeed be an opportunity there.
A couple of weeks ago Chris Bowers of myDD posted his map of political power by county:
This map is an attempt to show (using the typical red-Republican and blue-Democratic colors) how the country now looks in terms of approval ratings. Using it, Bowers can reasonably claim that the country has gone quite "blue" and he cites the latest SurveyUSA poll to show that in only 3 states does the President have favorable ratings (Idaho, Utah, Wyoming, see map).
Looks encouraging for liberals, right? In fact this map points up some interesting issues to do with politics at levels more detailed than the state. First, the map above is speculation, or perhaps at best extrapolation. As most people know, there are over 3,100 counties in the USA and it's difficult to say how they would look. As comments to Bowers' own post pointed out, his method is unclear (and not likely to be accurate).
Why is this? And why are counties so difficult to predict, politically?
1. There is a lack of data collected at that geographic scale
2. Inferring county-level data from available state data runs into the well-known ecological fallacy.
3. Even if every county was polled, there are many sparsely populated counties in America, which renders the result susceptible to sampling error. We'd also need to know the applicable "sampling frame" or the subset of the population the sample was drawn from (eg., registered voters, random telephone calls, etc.) which also shapes the result.
Perhaps we should forget about this level of data then? No: counties (and geographies at even finer levels of detail) are important:
1. Although many races occur statewide (notably Presidential and Senators) the House of Representatives and of course races for state office are based on districts within the state.
2. Getting an idea of support at the detailed level is important for voter registration and GOTV. You'll want to target areas perhaps even as detailed as the local precinct in GOTV or voter registration.
While I don't know the method used in producing the map above, it wouldn't be impossible to produce a map that estimated political support at the county level, if you could combine good local data/knowledge, previous county-level data (eg., percentage voting Kerry-Bush in '04) and some stratified sampling. I'm thinking of the Census Bureau ACS here.
1. David Leip's Electoral Atlas (purchase data at county level)
2. Anthony Robinson has free voting data he created himself
Posted by Ubikcan at 12:24 PM
Following my post about the county level political data Morten Neilsen of "SharpGIS" has created a cool new mapping application to display the whole dataset (available from Anthony Robinson at Penn State).
Morten uses a double-ended color ramp which I'm not sure would pass muster in design textbooks, but that aside, it's a good example of what can be done when raw data are shared. Using his web app you can pull down and instantly display any of the data sets that Penn State makes available. I wasn't previously aware of the application SharpMap he uses but I was impressed by how quickly something can be created with it. Thanks!
Another correspondent also left a note about the FairData site which takes a slightly different approach. They have tons of data and provide an interactive map interface to it (ie they handle the mapping and the data rather than distributing the data for people to play with). This has the advantage of providing users with maps without having to learn map software. They appear to use Maptitude for the Web, which I found a bit clunky. Specifically the Info button doesn't work in Firefox (which they acknowledge).
These kinds of web-distributed political efforts have a real potential to help precision-mapping of get out the vote (GOTV) and voter registration efforts. They may even help people become more interested in politics since these tools (previously only available to precinct captains and activists) are now available to all.
Posted by Ubikcan at 1:29 PM
As a long-time user of the Internet, I'm naturally very concerned when something threatens access to it or threatens to balkanize it into little segments. Unfortunately there is such a threat at the moment. Telecom companies want to regulate content on the web and internet by differentially charging users to see content. Obviously, if access to web content varies by where you live or by how much you're willing to pay, that is, pay to play, this is not good. It will especially affect the little guy, most of us.
The concept of "net neutrality" is that the web should be open to all--as it is now. Major content providers such as Yahoo and Google are in favor of it. Tim Berners-Lee, who invented the web, is in favor of it.
Today a bipartisan bill, HR 5417 is being marked up. It's stated goal is to offer protections to net neutrality. Here's more info.
Update. The Judiciary Committee Dems seemed to have voted the right way in favor of the bill. Some commentators are warning of telcos obfuscation of the issue, so watch out.
Posted by Ubikcan at 2:45 PM
Follow the money:
USA Political donations
Atlanta political donations
I think the ATL map is interesting because everything I've seen about Atlanta suggests it is a tale of two cities ( north and south) rather than just urban/suburban ring. We see that again here. It really is amazing how consistent the pattern is. Mind you, DeKalb county to the East is not really on this map, so check out liberal Decatur:
Source: Fundrace, Matthew Kane
Posted by Ubikcan at 1:38 PM
Usually for voting results you see statewide maps. These are useful in a presidential election because of the electoral college. they are less useful for getting a more detailed view of voters' preferences. Unfortunately data at the county level are not usually free.
However, I recently found some great county level data from Anthony Robinson which has voting data like the following:
I think it makes a very interesting map. You could do the same for a Bush strength map as well.
Posted by Ubikcan at 3:08 PM
Kos has a new prediction map of state governors:
He's predicting Dem pickups in 9 states and no GOP pickups. As he points out, the deep South remains a problem for Dems.
David Leips's polling data are available here (note reversed color scheme).
Posted by Ubikcan at 9:41 AM
It's a common meme that mapping is dead. Cartography has been killed off by digital maps and GIS. Yet the fact is we live in a world populated by more maps than ever. More maps are produced today than at any other time in human history. MapQuest may very well be the world's biggest map publisher with 60 million maps published every day.
How mappy are we? Consider this story* from South Texas, where a small town you've never heard of has implemented an online GIS for their residents. There's nothing particularly different about this system, the point is that places like McAllen Texas are implementing them. It's the sheer everydayness of mapping that is the point here.
*Please excuse the link to a newspaper owned by the yucky Freedom Communications, who enegage in so-called "astroturf" editorials.
Posted by Ubikcan at 9:30 AM
A new DNA analysis of remains have confirmed that at least part of Columbus' body is buried in Spain:
MADRID, Spain - Scientists said Friday they have confirmed that at least some of Christopher Columbus' remains were buried inside a Spanish cathedral, a discovery that could help end a century-old debate over the explorer's final resting place.I suppose this still leaves open the possibility that there are some remains in the Dominican Republic as well but they don't appear to want them to be tested.
Posted by Ubikcan at 9:14 AM
The Times predicts the end of paper maps:
[GPS is] gradually killing off maps, the charts that have revealed the changing contours of our world and minds since the birth of culture. English mapmakers once placed the phrase Hic sunt dracones, “Here be dragons”, on maps to mark the edges of the known world. Charlene [the GPS] has slain what few dragons remained. With a GPS embedded in dashboard, wristwatch or mobile telephone, we will never be lost again.Apart from the fact that cartographers never actually did put "here be dragons" on maps, I don't think our London correspondent is right. There's no way that paper maps will disappear. Did the horse disappear with the invention of the automobile? No, it exists alongside the car, although certainly its purpose changed.
The paper map is just too darned useful; it's light, needs no electricity, and is interoperable across systems and cultures. How many times have you printed out a digital map? I rest my case.
Posted by Ubikcan at 9:30 AM
The latest (5/15/06) opinion polls for President Bush in Georgia show an interesting geographical split between the Atlanta metro area and rural Georgia. But even outside the metro area his unfavorable ratings are high. Georgia as a whole is 43% approve, 55% disapprove.
South and East Georgia
Posted by Ubikcan at 12:52 PM
Crashing the Gate came to Atlanta on Tuesday. Markos of Daily Kos was there and talked a bit about the new political landscape of netroot activism before taking some Q&A. It's podcasted here by the Georgia podcast network. The turnout, in Atlanta's famous Manuel's Tavern, was fantastic! I've only got low-res cellphone pictures, but here they are:
Markos and founder of Air America(?)
My attempt at a crowd scene...
Posted by Ubikcan at 7:55 AM
Block diagrams are out of favor today, but at one time they were a useful and evocative way of representing terrain. One of the acknowledged experts in this field was Armin Lobeck (1886 - 1958). It's a shame that Lobeck's work is not better known (though his book things Maps Don't tell us is still in print and features a selection of his diagrams). As far as I know the only biographical work on him is a Masters thesis by Fred Grandinetti.
During WWI Lobeck was employed by the US government in their preparations for peace and he made a number of detailed studies of the areas likely to be discussed.
Here are some images of these maps (available at the National Archives, College Park, MD).
Posted by Ubikcan at 12:49 PM
That comment was made in response to a story on Eric Boehlert's new book Lapdogs: How the Press Rolled Over for Bush which documents the way the press is not doing it's job anymore: critically investigating the spin emanating from the White House. In fact, the book suggests (not for the first time) that the press's complicity in the "Swift Boat" attacks on Kerry lost him the election and all that that entails.
Colbert's amazing routine at the annual White House Correspondents' Association dinner last Saturday is another example, and if it has been under-reported by the "MSM" this week, it's probably because of zingers like this one:
But, listen, let's review the rules. Here's how it works: the president makes decisions. He's the decider. The press secretary announces those decisions, and you people of the press type those decisions down. Make, announce, type. Just put 'em through a spell check and go home. Get to know your family again. Make love to your wife. Write that novel you got kicking around in your head. You know, the one about the intrepid Washington reporter with the courage to stand up to the administration. You know -- fiction! Journos = typists.
So it's no surprise that people turn to the internet to get other news. Things like google.news.com or the BBC Online and as I've said before investigative blogging (Juan Cole, Atrios, TPMuckraker, etc.).
In this context then the recent moves to threaten "net neutrality" are extremely dangerous. If content is to be differentially moved across the internet because telcos and cable companies want people to pay them based on that content [not format] then we're hosed. I know threats to the internet have been around almost as long as... well the internet itself, but read here and make up your own mind. It's worth noting that the inventor of the web, Tim Berners-Lee, is on record preferring content neutrality.
We should know by now that knowledge is not separate from power.
Posted by Ubikcan at 7:56 AM
Since the NLCD data in the study cited below are a little old as Brian mentions below (not that that invalidates the data of course, particularly for time series analysis), here are some references to more recent studies:
Yang, Xiaojun "Satellite Monitoring of Urban Spatial Growth in the Atlanta Metropolitan Area" Photogrammetric Engineering and Remote Sensing, 2002, 68(7). Uses data from 1973-1999 (Thematic Mapper/Landsat).
Yang, X. and Lo, C.P. "Modelling Urban Growth and Landscape Changes in the Atlanta Metropolitan Area" Int. J. Geog. Inf. Sci. 17(5): 463-488.
Finally, newer NLCD data actually are becoming available here (2002). Comparison of NLCD and Landscan data:
NLCD 2001 on left, Landscan on right. Approx. same scale but not exactly (and on diff. projections I believe).
Detail of downtown Atlanta. Red indicates urban areas I believe. City of Atlanta boundary shown in black. Original resolution is 30m square.
Posted by Ubikcan at 2:24 PM