When you think about it, it’s not all that surprising that most special elections for House seats result in the party that originally held the seat retaining control. After all, the voters are the same; if they wanted a Republican in the previous election, it makes sense that they would want a Republican when voting on some random Tuesday 15 months later. Special elections are special, happening at weird times and generally seeing weird electorates. But they generally turn out the same way.

Smart Politics’s Eric Ostermeier looked at 378 special House elections that have been held since the 77th Congress in 1941. His analysis determined that 20.1 percent of those elections resulted in the seat changing hands between party. The other 79.9 percent of the time, the same party controlled the seat before and after the vote.

Ostermeier also determined that the current stretch of 19 straight special-election wins by the incumbent party is tied for the longest since 1941. (The other 19-election stretch was from November 1953 to May 1959.) If Democrat Jon Ossoff can prevail in the race in Georgia’s 6th Congressional District on Tuesday, it would be the first seat to flip since a special election in November 2012.

The data Ostermeier used for his analysis is proprietary, and historic House special election results are tricky to come by. (The House Clerk didn’t return our request for information by the time this article was published.) Wikipedia has a lengthy list of House special elections, though, allowing us to do some analysis.

Of the 350 races since January 1917 on that list, slightly more seats were held by Democrats than Republicans — and Republicans were slightly more likely to flip the seat to their advantage.

Overall, the seats changed hands at about the same rate as Ostermeier determined: 20.5 percent. There’s no particular pattern to when it happens, though there are clear droughts over time when there were either no special elections or no seats that flipped. In only four years (indicated below with gray bars) were there no special elections.

The overall pattern? Twofold. Parties generally retain seats, but when they flip is fairly scattered.

What does this tell us about Georgia? A Republican seat in a Republican district will usually be won by a Republican in a special election. But one-fifth of the time, it won’t.