Votes and Seats: What Made the 2012 Election Different

by John Sides on November 12, 2012 · 6 comments

in Campaigns and elections


Apropos of Dan’s earlier post, political scientist Matthew Green sends the graph above and writes:
If the numbers reported by Aaron Blake in the Washington Post are correct, it would be highly unusual, historically speaking.

The chart shows, for every election since 1962, the percent of the two-party vote that all Democratic House candidates received ( horizontal axis) compared with the percent of House seats that Democratic candidates actually won (vertical axis).  Leaving aside 2012, for which the data are still tentative, in only one election during this 50 year period (1996) did a party that received a majority of the two-party popular vote fail to win a majority of seats in the House.

What is true, however, is that the winning party has almost always gotten an extra boost in House seats.  With two exceptions—the 1994 and 2006 elections—the party that has received a majority of the popular vote won a larger share of House seats.

In fact, that extra boost used to be much bigger than it is today.  Between 1962 and 1992, the majority party (always the Democratic Party) won an average of seven percentage points more House seats than their share of the two-party vote.  But from 1994 to 2010, that difference was a mere 1.1 percentage points.

Of course, within individual states there may be large discrepancies between the popular vote and seat share (see e.g. Pennsylvania in this election cycle).  But despite—or perhaps because of?—the rise of sophisticated gerrymandering by both parties over the past two decades, the House has become more, not less, representative of the popular vote.

{ 6 comments }

Anthony McGann November 12, 2012 at 4:20 pm

The conclusion that the House has become more representative of the popular vote confuses bias with responsiveness. It may be true that results have become closer to proportionality over time — the number of seats are less responsive to changes in the vote. However, in 2012 there was clearly bias — the party with fewer votes got more seats. These are quite different issues. What is problematic about 2012 is that the results are almost certainly not symmetric — if the Republicans had won a narrow majority of the vote instead of the Democrats, they would not have been reduced to 45% of the seats! Of course, 2012 is only one data point, but when we consider the low number of close races, it is hard to see future results being much different with a plausible Democratic swing.

2012 is the first election under the new districts drawn after Vieth v. Jubelirer (2004) in which the Supreme Court decided that partisan gerrymandering — however egregious — was not a ground for challenging a districting plan. It is therefore dangerous to talk about the House becoming more representative over time — 2012 may well mark a reversal of this trend.

Matt Green November 12, 2012 at 4:39 pm

Good points, thanks. The bias has almost always been present, albeit smaller post-1994. (I found it interesting that the tranformational elections of 1994 were the only ones in which seat distribution mirrored vote distribution almost perfectly.) And maybe the GOP has found a way to guarantee themselves an extra “cushion” of House seats until 2022…if so, it raises some very troubling questions about democratic accountability in the chamber.

Nicholas Goedert November 12, 2012 at 8:33 pm

I’ve looked at this data quite a bit in the past. It seems like what you’re claiming is that the seats-votes curve has flattened in the past twenty years, and I don’t really think that’s true. Rather, we should never expect the “bonus” that party gets to their popular vote share to be constant. Rather, the “bonus” will increase as the popular vote margin increases, until it reaches an inflection point, resembling a probit or logit curve.

I think what you are observing here is a combination of two things:
- (1) There were a lot more close popular vote elections post-1990…prior to this, is was very unusually for the Democrats NOT to win the popular vote by at least 5%. And we should expect a smaller “bonus” when elections are close.
- (2) Prior to 1990, the seats-votes curve is actually biased in favor of the Democrats, largely due to uniform control of state governments (and thus the gerrymanders) in the South. This changed after 1990 when the VRA amendments constrained these gerrymanders.

Matt Green November 12, 2012 at 11:25 pm

Your points are well-taken. The GOP has been winning the popular vote by relatively small margins since 1994. 2010 is an important exception, and it suggests you’re right — the bonus grows as the vote margin grows, regardless of party. Thanks!

Matthew Shugart November 14, 2012 at 1:22 am

A little comparative perspective on spurious legislative majorities:
http://fruitsandvotes.com/?p=6513

Matt Green November 14, 2012 at 10:48 am

Great post, thanks. It’s quite interesting that the US House has produced more spurious majorities over time than legislative institutions in other countries. Great chart on vote distribution, too.

Comments on this entry are closed.

Previous post:

Next post: