We all know gerrymandering is a thing, but what can we do about it? Can we even quantify how much there is? How can we say what representatives "should have" won? If you put people in different districts and had them vote on different representative options, there's no guarantee of who they would vote for.
I would propose, however, that there should be reasonable alignment between the popular presidential vote and the number of House seats for each party. While looking at each state is fraught with error (some representatives have broad appeal, for example), I would expect the effect to average out. Alternatively: the effect should average out if there's no systemic bias in how districts are drawn.
In the last election, Clinton took 48% of the popular vote, Trump 46%, and other 6%. If we simply scale this out over the 435 house seats, the alignment suggest we should see 209 Democrats, 200 Republicans, and 26 question marks. I would then generally expect that the question marks should also subdivide similarly and would leave us around 222 Democrats vs 213 Republicans.
The actual number is 193 Democrats, 237 Republicans, and 5 empty seats (for example because cabinet appointments who had been representatives have not been replaced yet). That's less Democrats than we'd expect in even the worst case (where the question marks all go Republican). It's certainly far fewer than the slight advantage this model says Democrats should have.
Where does this discrepancy come from? I charted the vote tallies in each state and compared the expected (proportional) representative count to the actual. I rolled the 3rd party votes proportionately into the two main parties. What do I see?
Republicans have an advantage in a lot more states than Democrats
While only 16 Democratic states have more representatives than the model would predict, 31 states have more Republicans than we'd expect. Screening out all the states where the net effect is less than 1 representative (rounding error, essentially), leaves only 7 states with a Democratic advantage compared to 21 going the other way.
Republicans own the majority of the biggest swings
The top 10 swings (positive denotes pro-Republican) are Texas (+5.3), Pennsylvania (+3.9), California (-3.6), Ohio (+3.3), North Carolina (+3.3), Massachusetts (-3.2), Florida (+2.3), Georgia (+2.2), Connecticut (-2.1), Michigan (2.0).
Republicans gain big in close states
It's not surprising that a state leaning heavily one way would have a more extreme representative makeup. However, Republicans make some huge gains in tightly contested states. Filtering those with popular votes between 47% and 53% shows significant wins in Pennsylvania (+3.9), North Carolina (+3.3), Florida (+2.3), Georgia (+2.2), Michigan (+2.0), Virginia (+1.8), Wisconsin (+1.0). In all, Republicans made more than half their gains (14.6 of 26.4) in states with fairly close presidential popular votes. Half again (+7.3) of that gain is made in states where the popular vote was within 51-49.
So what's going on here? There's a common narrative that Republicans got a boost from promising to repeal Obamacare. Even if not for that, perhaps there's just random volatility? Republicans have in fact controlled the House since 2010 (when Obamacare became their rallying cry), but didn't in 2008 or 2006. But, we have to go all the way back to 1992 to see the previous time Democrats had numbers ... though the Democrat win streak then stretches all the way back to 1960 when the party wasn't even the same thing it is today.
What happened in 1994 to turn the tide?
That was when the more heavily partisan rhetoric started to come into play, but I'm not sure how that influenced the outcomes. Perhaps it was more fear-based, playing on people's concerns? Whether through gerrymander, message, or something else, Republicans figured out a new way to play and win.
Republican presidents were winning elections in the 60s, 70s and 80s, proving that the model I've proposed has no predictive power. I still wonder if it should work, if there's some kind of rigging or gamesmanship afoot, or if there are just tides in politics I don't understand.
I would propose, however, that there should be reasonable alignment between the popular presidential vote and the number of House seats for each party. While looking at each state is fraught with error (some representatives have broad appeal, for example), I would expect the effect to average out. Alternatively: the effect should average out if there's no systemic bias in how districts are drawn.
In the last election, Clinton took 48% of the popular vote, Trump 46%, and other 6%. If we simply scale this out over the 435 house seats, the alignment suggest we should see 209 Democrats, 200 Republicans, and 26 question marks. I would then generally expect that the question marks should also subdivide similarly and would leave us around 222 Democrats vs 213 Republicans.
The actual number is 193 Democrats, 237 Republicans, and 5 empty seats (for example because cabinet appointments who had been representatives have not been replaced yet). That's less Democrats than we'd expect in even the worst case (where the question marks all go Republican). It's certainly far fewer than the slight advantage this model says Democrats should have.
Where does this discrepancy come from? I charted the vote tallies in each state and compared the expected (proportional) representative count to the actual. I rolled the 3rd party votes proportionately into the two main parties. What do I see?
Republicans have an advantage in a lot more states than Democrats
While only 16 Democratic states have more representatives than the model would predict, 31 states have more Republicans than we'd expect. Screening out all the states where the net effect is less than 1 representative (rounding error, essentially), leaves only 7 states with a Democratic advantage compared to 21 going the other way.
Republicans own the majority of the biggest swings
The top 10 swings (positive denotes pro-Republican) are Texas (+5.3), Pennsylvania (+3.9), California (-3.6), Ohio (+3.3), North Carolina (+3.3), Massachusetts (-3.2), Florida (+2.3), Georgia (+2.2), Connecticut (-2.1), Michigan (2.0).
Republicans gain big in close states
It's not surprising that a state leaning heavily one way would have a more extreme representative makeup. However, Republicans make some huge gains in tightly contested states. Filtering those with popular votes between 47% and 53% shows significant wins in Pennsylvania (+3.9), North Carolina (+3.3), Florida (+2.3), Georgia (+2.2), Michigan (+2.0), Virginia (+1.8), Wisconsin (+1.0). In all, Republicans made more than half their gains (14.6 of 26.4) in states with fairly close presidential popular votes. Half again (+7.3) of that gain is made in states where the popular vote was within 51-49.
So what's going on here? There's a common narrative that Republicans got a boost from promising to repeal Obamacare. Even if not for that, perhaps there's just random volatility? Republicans have in fact controlled the House since 2010 (when Obamacare became their rallying cry), but didn't in 2008 or 2006. But, we have to go all the way back to 1992 to see the previous time Democrats had numbers ... though the Democrat win streak then stretches all the way back to 1960 when the party wasn't even the same thing it is today.
What happened in 1994 to turn the tide?
That was when the more heavily partisan rhetoric started to come into play, but I'm not sure how that influenced the outcomes. Perhaps it was more fear-based, playing on people's concerns? Whether through gerrymander, message, or something else, Republicans figured out a new way to play and win.
Republican presidents were winning elections in the 60s, 70s and 80s, proving that the model I've proposed has no predictive power. I still wonder if it should work, if there's some kind of rigging or gamesmanship afoot, or if there are just tides in politics I don't understand.
No comments:
Post a Comment