With just two states left, I decided to make this map update a little different. One of the major flaws of my past maps has been that it treats all counties equally, when some counties clearly have more voters than others. So, to give a better idea of where the voters are, I´ve darkened areas where voters are sparse and brightened areas where they are dense (in number, not intellectually).
The result better reflects the actual pattern of voters across the country:
However, there are several caveats, as I will detail below.
This is the color scale:
Red is Clinton, green is Edwards, and blue is Obama. Normally, I post color swapped versions, but in the interest in posting this more quickly, I'll just direct you to my Flickr photostream and fix this with an update later.
Update: Color swaps are at the bottom of the diary.
Now the first thing that is noticable about this map compared to past maps is the Alaska is huge. In reality, that's about how big it is compared to the rest of the nation in land area.
In previous maps that haven't been dependent on voter density (and thus area), I thought it was better to keep Alaska smaller to save space and reduce distraction. However, in a map where area is important, I decided to enlarge Alaska to it's proper scale, because while it is the second least populous state in the nation, it is also over twice the size of Texas in land area.
Obviously, there is a greater variability of brightness accross the map, but this did come with a few sacrifices. Since the color of the map is on a linear scale and the intensity is on a logarithmic scale, the monotone map versions are not meaningful (and making a logarithmic scale of an individual candidate's votes would be rather dull and uninformative).
Furthermore, I've sacrificed votes for other candidates in this map. With the exclusion of the Michigan Uncommitted vote (which I've allocated to Obama), I've disenfranchised voters for Kucinich, Richardson, Biden, Dodd, Uncommitted, and Gravel by pretending that they didn't vote in my calculations. In reality, this doesn't affect much. There were only a few sparsely populated counties outside of the early states where this other vote would have had any significant effect.
Now, you may be wondering how exactly these votes are weighted, because I didn't and because of the lack of transparency in some caucus state, I couldn't weight it to the popular voter (which I believe is irrelevant to everything except Clinton campaign talking points for reasons that would require another diary to enumerate), so I weighted it based upon percentage of a vote toward a delegate per square mile, assuming for simplicity's sake that all delegates were determined at the state level at a proportion to the state's vote (they aren't).
Variants
The above map treats Florida and Michigan as if they were legitimate contests that received their original delegate weighting. Obviously, as of now, that's not the case. Instead this is the situation:
Of course, there's likely to be a compromise on May 31 where half of the delegates from these states are seated. This map illustrates this compromise:
Now, this does show one of the weaknesses of using logarithmic weighting. Florida and Michigan appear marginally darker, but essentially unchanged. What this shows is that the density of voters in the brightest areas are actually several orders of magnitude higher than those in the midtone regions, which are in turn several orders of magnitude higher than those in the darkest regions.
Now, it should also be noted that two early states, Iowa and Nevada, held county conventions (I attended on here in Iowa), which change the delegate math, as many of the Iowa Edwards delegates switched to Obama and no Edwards delegates left the county conventions in Nevada. In the end, this lead to a net gain of 10 projected pledged delegates in Iowa. Later at the Nevada state convention, he gained a net of 2, but that wasn't reflected at the county level, which if anything, was a slight gain for Clinton.
So, here's the map modified to reflect those convention results:
Ratio map
To look at the straight ratios in each county without weighting for voter density, here is the original map:
Comparing with the maps above, it becomes pretty clear that Obama has an advantage in urban areas that is not adequately conveyed by a map of straight ratios.
As a side note, in this weighted map, I have figured in both the caucus results and the primary results, but this ratio map shows just the primary results. I'll likely fix this before the final update after Montana and South Dakota.
Oregon
Oregon is classic Obama territory. Sitting on the Northwest coast, has some decent-sized cities in Portland and Eugene and one of the most liberal Democratic bases in the nation (and one of the most conservative Republican bases).
A few days ago, ObamaManiac 2008 made a prediction regarding the result of the Oregon primary that was pretty accurate: within 0.1% of Obama's percentage and 1.1% of Clinton's. Despite the fact that this adds up to 99% in a contest where 100% of votes went to one of those two candidates, an average deviation of 0.6% and a spot on delegate prediction is not bad.
Here's the map that ObamaManiac set as the centerpiece of the diary:
It does pretty well compared to a map of the actual margins:
Kentucky
To start, here's an interesting bit of trivia Kentucky's geographical trivia. It is the only completely landlocked non-contiguous state in the union. The circular region you see in the map above is actually an exclave of Fulton County (the western-most county), separated from Kentucky proper by a southward protrusion of Missouri.
While Obama won Oregon in the urban areas, he lost Kentucky in the rural areas. Unlike Portland and Eugene, Louisville and Lexington barely went Obama, and while the Oregon country-side was split, there was no contest in Kentucky's rural areas. While the overall state result wasn't as lopsided a victory for Clinton as West Virginia thanks to Louisville and Lexington, Eastern Kentucky represents the true heart of Hillary Clinton's Appalachian base, as detailed several times by DHinMI.
Now, winning a state by large margins using a rural base looks a bit different than one using an urban base. Here's the Kentucky margin map on the same scale as the Oregon map above:
Clinton held Obama to a very narrow advantage in the two biggest urban centers while racking up enormous margins is the relatively less populace rural areas. While her final margin percentage in Kentucky was twice his margin in Oregon, The Kentucky map looks an order of magnitude more red than Oregon looks blue, and thats because she had to win by large margins in sparsely populated areas to win the margin she did while his urban margins, weren't as big and are represented by a much smaller area on the map.
Strangely enough, a news outlet that is willing to cover the race problem in Kentucky is none other than the Doha-based Al Jazeera. Apparently foreign outlets are willing to carry information about our country that our own news networks are too squeamish to touch:
via Matthew Yglesias
Looking ahead
The last two states, Montana and South Dakota, vote on June 3 after the Rules and Bylaws Committee meeting on May 31 and the Puerto Rico primary on June 1. If the pattern continues, Obama should win those two states pretty easily. One caveat is that they are two primary states completely surrounded by caucus states (which except for Maine, are a contiguous bloc). If it's really the caucus nature of these contests and not their geographical location and what that entails that is giving Obama wins there (I'm extremely skeptical of this view) or if Reverend Wright was really a major issue outside of Appalachia, then South Dakota and Montana may go to Clinton.
Color swaps
Here are the color swapped versions of the above maps (with the exception of the prediction and margin maps):
Weighted map
Map without Florida and Michigan
Map with Florida and Michigan given half of their original delegates
Map showing the results of the Iowa and Nevada county conventions
Ratio map
Oregon
Kentucky