The decline in a measure of split-ticket voting. Explanation below.
Last month, Steve Singiser noted the paucity of congressional seats won by a presidential candidate of one party but held by a member of the opposite party following the elections of 2014 in his post,
"Is split-ticket voting dying?" As Steve relates, the consequences of this change are more detrimental to Democrats than Republicans, especially in state legislatures.
In the graph above, we can see this decline in split-ticket voting for US Representatives clearly plotted. In 2000, a Republican running for the House received a vote share that was nearly nine points off (median) from what George Bush had in that same district; by 2012, that number had fallen to a little more than three points off from the vote share of Mitt Romney.
Join me below the fold to see graphs of these results.
In the graphs below, we see the results of House elections for 2000 and 2012. Only contests where the Democrat and Republican together received more than 99 percent of the vote are shown on these graphs:
Click to enlarge.
It's clear from a glance that the data points hew far more closely to the diagonal line in 2012 compared to 2000. Points for Republican House candidates who had the same exact vote share as Romney would fall on this line. Republicans tended to do better than Romney or Bush in Republican districts, where their Democratic opponent was more likely to be a Some Dude; likewise, Republicans tended to do worse in Democratic districts. We can see this more clearly in the next graph:
Click to enlarge.
On average, points are much farther from zero for 2000 compared to 2012. The median distance from zero on this graph, for each year, is what is plotted in the first graph at the top of this post.
The Midterm Shift
But what about midterm elections? Here's what these two types of graphs looked like for 2014:
Click to enlarge.
Clearly, things are a little different. On both graphs, everything is shifted a little higher. There were very few districts in which Republican candidates did worse than Romney. It was obviously a good year for Republicans. Indeed, the median shift toward Republicans compared to Romney's performance was 4.7 points. This is actually similar to both 2010 and 2002:
The graph makes clear the unsettling (to Democrats) but increasingly apparent possibility that 2010 and 2014 were "normal" midterms, due to the regular dropoff of mainly Democratic younger voters (which is not a new phenomenon).
The graph also shows that the median for presidential election years is close to zero—a phenomenon that will come in handy later.
What about third parties?
What if we look at contests where fewer than 99 percent of voters voted for a Democrat or Republican? What happens to the graph?
The medians generally are close to each other for all three cases (99 percent, 97 percent, and 95 percent D+R), especially after 2004. And, with the notable exception of 2006, the more third party votes there are, the further toward
Democrats the median shifts. In other words, on the whole, third party votes tend to take votes away from Republicans in House races.
There is one complicating wrinkle to take into account: there tend to be fewer votes for representative than for president in a given district. This down ballot dropoff is not considered in the analysis here.
Nonetheless, the overall trend is clear: split-ticket voting, between House and presidential contests, has decreased by about two-thirds since 2000. As with most political trends, it could easily be reversed; given the current political polarization, it seems unlikely to do so anytime soon.
