Swing states, historically

Robert Kosara over at eagereyes has a great post about visualizing which states swing which way, and trying to find historical trends. While his original visualization is nice, I think the ones put forth by derek, at the Information Ocean, bring out really interesting trends. Take a look at these two, in particular: 1, 2. Notice the blue-red reversal after 1964’s democratic landslide. (Looks up on Wikipedia) That was Lyndon Johnson’s election. From this graph alone, I would suspect something fundamental happened on those years that changed US politics. Surely enough, Wikipedia says:

While losing quite badly in the 1964 election, Goldwater laid the foundation for the conservative revolution to follow. Ronald Reagan’s speech on Goldwater’s behalf, grassroots organization, and the conservative takeover of the Republican party would all help to bring about the “Reagan Revolution” of the 1980s. […] Johnson went from his victory in the 1964 election to launch the Great Society program at home, signing the Voting Rights Act of 1965 and starting the War on Poverty. He also escalated the Vietnam War, which corroded his popularity. By 1968, Johnson was so unpopular that he had to withdraw as a candidate. Moreover, his support of civil rights for African-Americans helped split union members and Southerners away from Franklin Roosevelt’s Democratic New Deal Coalition, which would lead to the phenomenon of the “Reagan Democrat”. Of the ten presidential elections that followed, Democrats would win only three times. Columnist George Will had this to say about the lasting effects of the 1964 election: “It took 16 years to count the votes, and Goldwater won.”

If derek writes a post about these plots (they currently live in the comments section of Robert’s post), I’ll let you know. Incidentally, these different visualizations highlight Tufte’s warning that “one plus one equals three or more”. Tableau’s rendition of the data involves large circles of the appropriate colors, which happen to leave this really jarring white background pattern that hurts your eyes as you look around the table. Derek’s version is simpler, but using a full painted square seems much more effective. In fact, I’d even get rid of the fine white outlines.

EDIT: Robert Morton over at Tableau Software points out that Tableau can create just the plot I suggested. Thanks, Robert! One final nitpick. The equiluminance of the blue and red is now sort of annoying – it reminds me of the Vis keynote. My solution would be to do something like this, for example (pardon the quick gimping).

Advertisements

8 responses to “Swing states, historically

  1. Hi Carlos,
    Your critique is valid but nothing inherent in Tableau forced the view that Robert Kosara used. A view like Derek’s is just as easy:

  2. carlosscheidegger

    Beautiful – that was exactly what I was looking for. If you don’t mind, I’m going to use that link on my post. Thanks for setting me straight 🙂

  3. Feel free! I do like Derek’s approaches to clustering states as it does a good job of showing demographic shifts over time. I’ll let you know if I create any new charts.

    My full post on this topic is here: http://www.tableausoftware.com/blog/state-electoral-vote-sorting

  4. I tried the squares, but the whole thing became a bit too filled in for my tastes – so those circles were chosen quite deliberately. I’m not going to argue that they’re better than the filled-in squares, but I’m thinking that there’s something else that’s wrong with the chart. I can’t quite put my finger on it yet, though.

  5. carlosscheidegger

    Robert, one thing I do like about the circles is that it’s easier to follow a particular column or row.

    I’m trying to come up with something nicer as we speak. I don’t have Excel or Tableau (go starving grad student!), so it’s writing code for me :).

  6. carlosscheidegger

    Didn’t finish the state ordering optimization code in time, but here’s my SVG version so far.

  7. What did you have in mind for order optimization?

  8. carlosscheidegger

    I was thinking about writing out an “energy” for each possible ordering. For example, you could penalize states that look different for being adjacent to one another. Or penalize states that are far apart geographically but close in the ordering.

    Obviously, the optimization problem is reducible from TSP. However, reasonable energies for the optimization would probably obey the triangle inequality, which means we can get some good approximation algorithms. Then, just use randomized optimization and let it run for a while.. I’ll try to hack something like this soon.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s