Philip Greengard points us to the above-titled news article by Philip Bump.
The article was just fine, a reminder of modern-day political polarization. The only thing that bothered me were the graphs. I redrew them above. Here were the original versions:
I see a few problems with these graphs. First, the information is duplicated because the percentages all add up to 99% or 100%. Second, the patterns are super hard to follow because your eye jumps up and down between the yes and no percentages. Third, the x-axis is multiplexed so it’s hard to compare age groups within parties. Fourth, nothing is really done with the color scheme. I think my redrawn version (it took me about a half hour in R; I guess Hadley could’ve done it better in 5 minutes) fixes these problems. I put age on the x-axis because it seems natural to go from young to old.
Once I did this, I thought it would be good to get some more discrimination on the age scale. An earlier graph in that news article showed age groups of 18-35, 35-49, 50-64, and 65+, so I clicked through to the survey report from Quinnipiac poll. But that left me even more baffled because I didn’t see the party x age breakdown in the report at all. Maybe I didn’t know where to look.