It’s after Labor Day and the 2018 Midterms are in just two months. It’s political season, so I made a dashboard of political data. Morning Consult periodically publishes approval ratings data, so I grabbed that data for governors and senators, added in the elected official’s party, and designed this. Click here for the interactive version.
The dashboard seems pretty self-explanatory – the least popular governors are in Oklahoma and Connecticut. The most popular governors are in Massachusetts, Maryland, and Alabama. The most popular senators are in South Dakota and Minnesota. The least popular senators are in Arizona and Kentucky. Boom. Question answered. Note that Senator Cindy Hyde-Smith of Mississippi did not have data, so there are 99 senators with data. Having said that, I wanted to go through a few pieces of the dashboard.
First, notice the color choices. Because blue, yellow, and red are spoken for by the political parties, I was left with fewer choices to show the approval rating margin. Unfortunately, I settled on orange = negative and green = positive. For red-green colorblind folks, this may make it hard to glean understanding from popular republicans. I could have used shapes as well to encode political party, but I didn’t want to overload users with multiple uses of color AND shape. Hopefully the contrast with blue and yellow + the legend will be sufficient to help red-green colorblind folks see what’s going on.
The biggest decision I had to make was how to visualize the approval margin on a state-by-state map for senators. I had two approval margins for each state. I can’t split the state in half to show the approval margin for each senator. I also wanted to match the governor map, which was much easier to build without this issue. I didn’t want to color the slices of the mini pie chart. I didn’t want to use size. I didn’t want to average the two approval ratings – that would obscure what was happening within a state. My options were limited.
Here’s what I settled on – if the two senators within a state both had negative or both had positive approval margins, I picked the most negative or the most positive, respectively. If the two senators within a state had opposite signs, I picked the approval rating with the highest absolute value. If the goal was to identify the most and least popular senators, the highest absolute value corresponds to that goal.
However, my calculation to produce that output is not currently correct. In the opposite sign situation, the calculation will take the highest absolute value just fine but it will display the number as positive.
For example, in New Jersey Bob Menendez has a -8% approval margin. Cory Booker has a +17% approval margin. The dashboard appropriately displays +17%. However, imagine the opposite scenario: +8% and -17%. My calculation would identify the -17% as the number to display, but the calculation would display +17% and color the state that way instead of -17%. This situation, in which the negative approval margin has a higher absolute value than the positive approval margin, does not actually occur in the data. Thus, I saw no need to fix it. It works appropriately when the positive approval margin has a higher absolute value than the negative approval margin. If you download the dashboard to play around with, keep this in mind.
One of my big takeaways from this dashboard was that senators seem to be more popular than governors (in general). Senators Mitch McConnell of Kentucky and Jeff Flake of Arizona had the most negative approval margins among senators, but they couldn’t hold a candle to Governors Mary Fallin of Oklahoma and Dan Malloy of Connecticut. Senators are at least not as disliked as much as governors can be. This could be just the current state in July 2018, not a broader conclusion that is true over time.