A2-JeremyHuddleston

From CS294-10 Visualization Fa08

Jump to: navigation, search

Contents

Assignment 2

P1: Data Set

I'm pretty interested in politics and have been following the 2008 presidential election fairly closely. I've chosen the data set available from electoral-vote.com to study in this homework.

P2: Questions

  • Have state opinions polarized or migrated towards "dead-even"?

P3: Notes

I first tried importing the data into Tableau. Data mangling wasn't necessary since it was already in a clean csv format. I dragged around the handles I was interested in and created an initial visualization:

Now, I decided to give Spotfire a try before I got too invested in Tableau. I noticed that Spotfire was a bit more intuitive when it came to creating line graphs, so that's where I started:

Image:02-spitfire.png

Back to Tableau. After messing around with the various buttons and options, I got nowhere. I think the best feature of Tableau is the "Show Me!" button. I pretty much used this button and undo-redo over and over to figure out how to get Tableau to behave like I wanted it to. I figured out out to use the Measures to place the bars side by side and colorized them based on traditional values:

I then found the "Marks" tool in Tableau and was able to change this into a line graph. After some rearranging, I got a pretty good visualization to answer my initial questions:

I tried to figure out a way to get the date marks in each cell rather than just at the bottom, since I think that would be beneficial, but I couldn't figure out how to do that in Tableau.

Next, I decided that this was just too much information. Most people are concerned with just the "swing" states. Rather than filter based on the name, I wanted to come up with a formula for which states to consider "swing." My first attempt was MIN(ABS([GOP]-[Dem])) < 5, but this generated 32 hits. This certainly reveals that we're in a close race, but we're interested in states that are always close, so I further changed this to AVG(ABS([GOP]-[Dem])) < 7 which reveals the states with the closest races.

Getting back to the question, though, we don't care about the absolute results, we care about polarization versus "dead even". So I added a calculated field for "ABS([GOP] - [DEM])" and plotted that rather than the direct polling results for Democrats and Republicans:

P4: The Answers

Have state opinions polarized or migrated towards "dead-even"?

Across all 50 states, there is no trend towards polarization towards a particular candidate nor towards being dead even. Some states have polarized while others have gotten closer.



[add comment]
Personal tools