Electoral college

[ciphergoth]

After watching far, far too much West Wing (ie all of it - finally reached the end which is both bad and good) I created this image to visualize the results of the 2004 election. I want something like this updated live for 2008, with states not yet called ranked based on predictions.



You'll need to view it full size to properly see what's going on.

(updated image with new algorithm for placing labels)

Comments

Re: Nice

[nikolasco.livejournal.com]

The writeup of your suggestion seems very confusing ...
1) You said "number of seats" which suggests you're talking about congressional seats (house and senate), not electoral votes. Are you proposing visualizing a different set of elections or is this an error in terminology?
(I'm currently assuming the latter. I don't think that affects the rest of this comment, but maybe it does and I don't realize it)

2) Right now, we can compare the the influence to the margin of victory by comparing lengths on the x- and y-axis. We can see who won and by how much by comparing the length of red versus the length of blue (both on the x axis).

Following your suggestion, as I read it, would collapse this data onto the x axis, making it a representation of "how crucial" a given state was. While interesting, I think the current use is better. It also seems like the y-axis would be kind of silly; you might as well just plot on the y axis and have the bars be of equal length on the x axis.

3) You said "influence of a change," which might indicate comparing "before and after," which again would be visualizing a different set of data.

I could assume that you mean "influence of a change, if one had occurred" but that doesn't seem at all useful because electoral votes are binary values; it doesn't matter how much a party wins by as long as it wins.

In the current visualization, you can see both which states would have been easiest to change (due to the ordering) and how many you'd need to change (just look at the length). For a potential strategy, it would be more useful to have the states ordered along the x axis by number-of-votes/margin. This is my preferred interpretation+alteration.

Re: Nice

[phantas.livejournal.com]

it was indeed the electoral college votes, a distraction from my part at the time I wrote the post.

Oh, I am not saying it is better (for starters, it's not a precise graph because there wouldn't be a fixed line to compare against), I am just throwing another possibly interesting view of the data.

What I meant by influence of change is that, since electoral votes are binary, how much a change would affect the election (either during counting or with respect to predictions) depends on two things: the closest it is to 0 and the number of electoral votes affected, isn't it? Or am I just confusing things at the moment?

If that's the case, if you order them along the x-axis according to their margin, the width of the rectangle would illustrate that influence.

Still, ciphergoth's suggestion is a much simpler view targeted at who's winning and how close is the race and, thus, better. This other suggestion is more targeted to how likely is there to be a big or small shift, though there are better ways of plotting that.