Just How “Special” Are Special Elections?

[Another version of the post was published on the Washington Post’s Monkey Cage blog.  Gibbs Knotts helped write this article.]

On September 25th, 2015, Republican John Boehner shocked the political world when he announced his plans to resign as Speaker of the House and retire from Congress.  In the aftermath of Boehner’s announcement, political commentators debated who would be the new Speaker and the role of the Tea Party in causing Boehner’s early retirement.  Less attention focused on the fact that Boehner’s resignation triggered a special election in Ohio’s 8th congressional district.

In yesterday’s contest to fill Boehner’s seat, Republican Warren Davidson defeated the Democrat Corey Foister.  Local and national outlets predicted it would be a good night for Davidson.  After all, Boehner had held the seat for nearly 25 years and the 8th district is one of the most Republican-leaning districts in Ohio (Obama lost to Mitt Romney in the 8th district 62 to 36 in the district).  Simply put, yesterday’s results were not particularly surprising.

All this begs the question: Are special elections really special?

Answering this question hinges on what one means by “special.”  Special elections certainly generate a lot of media attention.  When a Member of Congress dies, retires because of scandal, or resigns for personal reasons, it is newsworthy.  Special elections may also be special because of their ability to predict future election outcomes.  In one innovative study, political scientists David Smith and Thomas Brunell found that the party that wins the most special elections between elections often gains seats in the subsequent general election.

But “special” is also a synonym for “different.”  And in this context, a key question is whether special elections are somehow unique when compared to regular congressional elections.

We examined this question in a paper that was published earlier this year in the Journal of Elections, Public Opinion, and Parties.  We were particularly interested in whether special election outcomes are structured by the same factors as their closest counterpart: regular open-seat contests.

Why might special elections be different from regular congressional elections?  First, they exist in different political environments.  Regular open-seat races occur in a national environment in tandem with 434 other House contests, about one-third of Senate elections, and, every four years, alongside a presidential election.  In contrast, special elections occur during the “off season” and may be isolated from broader electoral forces.

We were particularly interested in the effect of presidential approval on special election outcomes (a factor widely thought to be a strong determent of congressional election results).  In one of the few studies of special election outcomes, political scientists Keith Gaddie, Charles Bullock, and Scott Buchanan found that the president’s approval rating has no effect on who wins and losses in a special election.  Because their results show that presidential approval does have an effect in open-seat elections, they conclude that special elections are insulated from national forces.  In other words, special elections do seem to be “different” from regular congressional elections in a key way.

Notably, however, Gaddie, Bullock, and Buchanan reached this conclusion with data from 1973-1997.  In our paper we hypothesized that the effect of presidential approval would be different because of the increased polarization in American politics since 2000.  For example, we know that voters are more loyal to one political party and that the national parties have greater organizational resources today.  Research has also found a stronger connection in recent years between presidential election results and House election results.   As Jeffrey Stonecash has shown in his recent book, House elections by the mid-1990s became less candidate-centered and more party-centered.

Our analysis confirmed this hypothesis.  In the modern era, the president’s approval rating does indeed have an effect on special election outcomes.  Candidates from the president’s party perform worse in special elections when the president is unpopular whereas candidates from the opposition party perform better.  Like regular congressional elections, our results show the fate of special election candidates hinges, in part, on national forces.  A secondary result in our paper is that the 2002 midterm seems to be the point in time when presidential approval became a significant predictor of special election outcomes.

As a whole, we conclude that special elections are not that “special” because they are structured by roughly the same set of factors as regular open-seat contests.  Much like Lee Sigelman concluded in 1981, special elections do indeed seem to be national contests despite their unusual timing.  Our research, coupled with Gaddie, Bullock, and Buchanan’s work, reveals that this change in the nature of special elections is the product of significant developments that have occurred from the 1970s to today.

In the context of yesterday’s special election in Ohio’s 8th district, however, presidential approval was probably not the deciding factor.  According to Gallup polling, Obama’s approval rating sits at 51% (not enough to have had an effect in either candidate’s favor).  Instead, what almost certainly matted yesterday was the simple fact that the 8th district contained such a large volume Republican voters.  However we expect presidential approval to be a factor in future contests, particularly in more competitive districts.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s