Post-Pennsylvania map update

Now that Pennsylvania has voted, it's time to release an update to the primary map.  This time, I don't have many variants, because I decided to switch to a different template for my map, using an SVG graphic as the original template, which can be better resized and allowed me to do away with the boundaries which obscure some of the trends between states.

I think it's pretty clear that Pennsylvania voted in a similar fashion to the surrounding states and that it adds even more clarity to the future contests.

Larger versions of the maps are available on Flickr (just click on the image to get to its Flickr page).  In the above version, blue represent Obama, red represents Clinton, and green represents Edwards.  If you don't like these colors, there are color swapped versions below the fold.

In some ways this is a step back, as I haven't yet divided Alaska into state house districts and I haven't yet made the same modified map versions as before.

I will hopefully have a greater array of modifications ready after the May 6 primaries.

•  Tips n/t

  by Meng Bomin on Wed Apr 23, 2008 at 10:59:21 AM PDT

  •  Thank you

    I love these maps - they show us so much more than coloring each state red or blue.

    by CA Pol Junkie on Wed Apr 23, 2008 at 11:28:22 AM PDT

  •  You shoud read

    my diary on color choices in political cartography.  This would be much more clear if you used white rather than purple to represent intermediate results.

    I hope you'll try this out to see if you like it.  Other than that -- good job!

    by Seneca Doane on Wed Apr 23, 2008 at 11:45:50 AM PDT

    •  I may.

      However, white as the "neutral" color doesn't work with three candidates without a serious loss of information in areas where the third candidate is competitive.

      Now, the advantage that white gives to a map is that it increases the contrast.  Personally, I find it to be a bit artificial.  You point out that shades of purple all look pretty similar (though you use overly bright shades of purple as demonstration), and that's true...but that's partially because 49-51 is pretty similar to 51-49 and I actually think that the fact that those look similar is a strength.

      I personally think that the comparison of monotone maps can also be quite effective, because shows percentage vote for a candidate rather than margin between candidates.

      I may do a comparison diary later.

      by Meng Bomin on Wed Apr 23, 2008 at 02:12:57 PM PDT

      •  It's actually better for three candidates

        Edwards victories becomes shades of green from the palest pastel through emerald (or forest, if you want to go darker at some point.)  Clearer than what you have now.  You can even mix the results of the two top candidates out of three to create more hues, with intensity as the different over the third, but I think you'd find it easier just to make a map of the leaders and a new map for other purposes.

        I look forward, quite seriously, to seeing your experiment!  (Include a poll!)

        by Seneca Doane on Wed Apr 23, 2008 at 03:40:34 PM PDT

        •  The problem is that

          Clinton and Obama minority votes disappear.  A Edwards-Obama-Clinton dominant county is a shade of teal while a Edwards-Clinton-Obama is more yellowish.  Both of those would be green using the white neutral map.

          I did make a pair of white maps after you suggested it (I haven't posted it yet), one where I eliminated the Edwards vote completely and the other with the Edwards vote.  Iowa (the only state where Edwards was truly competitive) is significantly whiter when the Edwards vote is figured in than when just Obama and Clinton are figured in.  The problem comes from the fact that a margin map (which is what the white maps are) only shows the margin between the first place and second place candidate, not figuring in the third place candidate.

          by Meng Bomin on Wed Apr 23, 2008 at 04:03:38 PM PDT

          •  I went to your Flickr page

            and looked at the maps in more detail, and honestly I still could not make out what was happening in parts of Iowa and the Florida Panhandle, even at highest resolution.  I know that you have the third-candidate info in there, but I honestly can't perceive it.

            I'd be very interested in seeing those new maps, by the way.  I'm not sure how you did those two other maps.  It sounds like you did one where the map just showed the Obama-Clinton margin, which was not what I was suggesting.  I'm not clear on why the other, which included the Edwards vote, should have been more white.  I would think it would be more green.  The best approach there would probably be to just give the margin of Edwards versus the second-place candidate.  I think I know how to do this and retain all of the information you want to keep.

            First, you would determine the appropriate color balance between the second and third place candidates -- much as you have with your purple maps, except that this could apply whichever two candidates were trailing.  The will give you information in the map about the margin between the second and third place candidates.

            Second, you color in the map showing the margin between the top candidate and the second place candidate.  However, rather than fill that in with intensities of pure red and blue, as I've suggested, you would color it in with the inverse of the color you determined in step one.  So, where Edwards won, the color balance between Obama and Clinton could result in a pure red (Obama gets no votes in third place) to a pure purple (they tie) to a pure blue (Clinton gets no votes in third place.)  These would generate, respectively, a coloring for Edwards of pure cyan, pure green, and pure yellow.  The margin of Edwards over whoever finished second would determine the intensity of that color.  (And this could be the same for Clinton and Obama.)

            The only problem I see with this formula is what happens when the two leading candidates tie.  In other words, distributons of 50-50-0 and 40-40-20 and 33-33-33 all turn out white.  The solution may be something like keying intensity in a three party map to the difference between the leader's percentage and the mean between the second- and third-pace candidates, rather than just the second-place candidate.  I still need to think that one through.

            by Seneca Doane on Wed Apr 23, 2008 at 11:27:00 PM PDT

            •  "fill that in?"

              Interesting discussion - I do graphic design, and this sort of thing fascinates me.  I was assuming the maps were generated using a database and plugging the numbers into a program that then displays them on the map.

              If this is all hand-done, this is about 10 times more impressive than I first thought.  And I'm already impressed. Great visual reference tool, one way or the other!

              by pakaal on Sat Apr 26, 2008 at 12:40:42 PM PDT

              •  I assume that they are computer generated

                and I assumed that "filling them in" could be done by using computer software that used a different equation to derive colors.

                by Seneca Doane on Sat Apr 26, 2008 at 12:45:19 PM PDT

                •  OK, cheers.

                  No expertise in GIS or related mapping stuff - for a moment I had an image of trying to do that all by hand, and it was a frightening one.

                  by pakaal on Sat Apr 26, 2008 at 03:51:02 PM PDT

•  Good color choices this time

  Clinton in red is appropriate.

  by broui on Wed Apr 23, 2008 at 11:02:08 AM PDT

  •  What do the colors mean?

    Maybe you can explain that to the uninitiated.

    by MadScientist on Wed Apr 23, 2008 at 11:07:20 AM PDT

    •  Do you mean for this map or in general?

      For this map it's Obama blue, Clinton red, Edwards green.

      Where they dominated, they get the shading.

      by broui on Wed Apr 23, 2008 at 11:11:42 AM PDT

      •  What type of data?

        What type of data are we talking about here? (I think thats his question, forgive me if I'm wrong MadS)  Is it nominal data(straight numbers)?  or (hopefully not) ratio data(percentage)?

        by Perdurabo on Wed Apr 23, 2008 at 11:34:46 AM PDT

        •  Ratio data.

          The straight numbers (as in exact number of votes) are not available for several caucus states, and if I were to make such a map, the cities would be much lighter than the rural areas (and I would be using a logarhythmic scale).

          I did do a map that was weighted based upon Kerry voters in 2004, which you can find here.

          by Meng Bomin on Wed Apr 23, 2008 at 03:41:55 PM PDT

•  Michigan shows up as the reddest

  state on this map, which is pretty misleading don't you think?

  by Olds88 on Wed Apr 23, 2008 at 11:04:10 AM PDT

  •  Hillary was the only major on the ballot. n/t

    by MT in Austin on Wed Apr 23, 2008 at 11:07:55 AM PDT

    •  I am very aware of that

      That is why it's misleading to show MI heavily favoring Clinton.

      by Olds88 on Wed Apr 23, 2008 at 11:10:03 AM PDT

      •  "misleading" doesn't equal "available data".

        She was the only name on the ballot.  It's not misleading, it accurately shows the data provided.

        Or rather, it's not misleading to folks who have an understanding of what's going on in the Primaries.  You might not like the fact that it's incomplete data, but the alternative (figuring out which percentage to give to Obama from the "undecided" group in MI) would be truly misleading, since it'd be one person's interpretation of what the numbers mean.

        by pakaal on Sat Apr 26, 2008 at 12:35:10 PM PDT

  •  In the next update, I'll have some modified maps

    that deal with Michigan (and Florida) in different ways.

    by Meng Bomin on Wed Apr 23, 2008 at 11:11:49 AM PDT

•  i think we can

  assume from this how the other states will look.

  North Carolina will look like Virginia with most of the eastern and central part of the state backing Obama strongly, while the panhandle, with the possible exception of Buncombe County, going heavily for Clinton. (See, the 1960 GE results map for a look on what I expect)

  Indiana may be all over the place, with Obama stronger in the urban areas (Indianapolis, Evansville, Bloomington, South Bend, Gary) and Clinton stronger in rural areas.

  South Dakota, Montana and Oregon will be more strongly Obama.

  Kentucky and West Virginia are not going to be pretty. I think it's likely every county in West Virginia is go Clinton and Obama will only be strong around Louisville, Lexington and maybe in the west south of Illinois in Kentucky.

  by nrafter530 on Wed Apr 23, 2008 at 11:05:18 AM PDT

  •  I believe KY has at least 7 Obama offices

    set up across the state and I haven't heard of any offices for HRC.  I believe Obama will do well in Louisville, Lexington, Bowling Green and western KY (Paducah, Evansville, etc)...This is really where the population base is for KY...

    by Kyman on Wed Apr 23, 2008 at 11:08:48 AM PDT

•  Great work! A LOT of interesting detail. nt

  by Terra Mystica on Wed Apr 23, 2008 at 11:10:56 AM PDT

•  If I were looking for a jigsaw piece...

  ... to fill in North Carolina, it would be very blue.   Indiana, harder to tell.  Where does it turn from Illinois blue to Ohio red? Misleading to include Michigan, in my opinion.

  Kentucky and West Virginia seem pretty red.  And Montana, Oregon and South Dakota blue.

  by Land of Enchantment on Wed Apr 23, 2008 at 11:51:10 AM PDT