Trends in presidential voting by state

When you look at presidential election returns by state, they jump all over, but when you subtract the national vote and group by region, some clear patterns jump out:

50trends.png

I’ve ordered the regions roughly from strongest Democrat-trending to strongest Republican-trending, and only cheated slightly by putting Colorado in the southwest and Alaska in with the mountain west. I’m also embarrassed to say that I adjusted the state labels by hand to stop them from overlapping; I really need to write a program to do this. If it were really important to identify which state was which, I’d probably label the states on the left of each plot also. Overall I think it came out pretty well, though.

To the R users out there (not to mention the people who set the R defaults), please also note the relatively few tick marks, the narrow space between the plots, and the internal titles for the subplots.

4 thoughts on “Trends in presidential voting by state

  1. Is it difficult to let R make a spike-lines type graphic? That means no frames, only one x-axis at the bottom, only one y-axis at the left etc

  2. I agree that the traditional R graphics engine puts too much white space in multi-frame layouts. Lattice graphics will give you a better default display. Something like this:

    xyplot(votes ~ year | region, group=state, type="l")

    should do the trick (where region and state are factors).

Comments are closed.