Did you vote for Mitt Romney in 2012 and a Democratic House candidate this year? If you did, you're pretty unusual. According to our analysis, only five congressional districts that backed Romney voted Democratic in the House this year -- thanks in part to heavily party-line voting over the last two cycles.

That's when you look at split districts. Which is a term we sort of made up, so a definition is in order: A "split district" is a congressional district in which the voters backed a candidate of one party for president and a candidate of the other party for the House. Over the last two cycles, such districts have become in rather short supply.

In theory, you'd think this would happen a lot. Over the past two decades, though, it's been a mix. In some elections, voters have been all over the place. And in others -- and particularly in 2012 -- voters have been more likely to vote for the same party to represent them in the House and the White House.

We took data from a variety of sources to figure out how this trend has evolved. The challenge is that compiling presidential votes by congressional district is tricky, since districts overlap with and split across county lines. Polidata compiles this data and makes some of it available online; for recent elections, users at the liberal site Daily Kos have pulled vote margins. Margins on House races were compiled from the AP.

The result: The charts below. The horizontal axis is how likely a congressional district was to vote for a party's presidential candidate, from a 100-percent margin for the Democrat at the left to a 100-percent margin for the Republican at right. The vertical axis is the same calculation for the House candidate that year, from 100-percent support for the Democrat at the bottom to 100-percent support for the Republican at the top. Dots in the blue or red areas are those in which a district voted for candidates of the same party. Dots in the white area are those in which the district split.

And the closer a dot is to that diagonal line, the fewer crossover voters there were.

With that in mind, here's each election since 1992, animated ...


... And static, for your perusal. The yellow line we will call the "party line," for fun. (Missing data wasn't included on the graphs.)

In 1992, there's a cluster around the center.

It shifts up in 1994, as Republicans swept the House.

In 2000, the cloud shifts right.

In 2006, there's more of a line now. It shifts down, as the Democrats take over.

In 2008 and 2010, it scatters again.

Then, in 2012, it's tightly clustered -- with far fewer districts splitting.

Or, to visualize it another way:


Data from several House races isn't included in that list, including close races this year, which could end up being split districts. But at least five districts this year went from splits to single-party after voters that backed Mitt Romney switched to Republican members of the House: GA-12, NC-7, TX-23, UT-4, and WV-3.

What's most interesting: There are now only five that backed Romney but supported Democrats this year: AZ-1, FL-2, FL-18, MN-7, and NE-2. In contrast, over 50 districts backed George W. Bush in 2004 but a Democrat in 2006, despite that year's Democratic wave.

And indeed, if we compare midterms over time, you can see that 2014 was much more tightly clustered around the party line than past elections.


There's no way to tell if the adherence to the party line that we've seen over the past two cycles is a short- or long-term trend. But one would think that it might correlate to partisanship: As the country gets more polarized, it would make sense that people would be less likely to split their votes between parties.

Time will tell.