Skip links

Crystal Ball 2014: So how’d we do?

On Monday, we offered our final calls in all 507 of the Senate, House, and gubernatorial races.

As of this writing, 490 of those races have been called for one party or the other, and we got 476 correct (97%).

We did best in the category everyone was watching most closely, the battle for the Senate, successfully calling 32 of the 33 called races. Of the three races remaining, it appears that our Leans Republican rating of Alaska will eventually be correct, and we also see Sen. Mary Landrieu (D-LA) as a decided underdog going into her Dec. 6 runoff with Rep. Bill Cassidy (R).

In one of the shockers of the evening, Sen. Mark Warner (D-VA) and former Republican National Committee Chairman Ed Gillespie (R) remain locked in a tight battle, though Warner’s small lead will probably hold up. Still, we tip our cap to Gillespie for running a strong race: This contest was a lot closer than we and a large bipartisan majority of other analysts expected.

Still, when it’s all said and done, it looks like the new Senate is going to be 54-46 Republican, almost exactly as we called it (53-47 Republican).

Our gutsiest Senate call was picking Sen. Pat Roberts (R-KS) to win reelection, and he did so by a surprisingly wide margin. (Most of the home-state analysts and many national observers were convinced that Orman would win.) The lesson perhaps is that betting on a state’s partisanship — Kansas is strongly Republican — seems like a good strategy to pick winners these days.

Additionally, we are proud of a number of the calls we made far in advance. For instance, we had the Kentucky Senate race rated Likely Republican the entire cycle, and we also never had the Michigan Senate — won by Sen.-elect Gary Peters (D) — as a Toss-up. Those picks look easy now but they weren’t when the polls were a dead heat a year ago, and many saw them as a coin flip.

In the House, we projected the Republicans to add to their sizable majority (we thought they would add nine net seats), and it looks like they are going to do a bit better than that, adding a dozen or more seats. Still, despite the lack of accurate polling or other information in these races, we picked many of the most competitive races correctly: Of the 423 seats called, we correctly picked 415 of the races. Of those, we called a couple of the few races to break for the Democrats Tuesday night, the losses by Reps. Steve Southerland (R, FL-2) and Lee Terry (R, NE-2), the only two House Republican incumbents who were defeated in this otherwise sterling GOP year. We also consistently cited Rep. Nick Rahall (D, WV-3) as the most vulnerable Democratic House incumbent this year, and he lost decisively.

Nearly a dozen of the gubernatorial races looked like toss-ups going into Election Day, but as per our tradition we offered picks in all of them. Assuming that Bill Walker, the independent challenger to Gov. Sean Parnell (R-AK), hangs on (he currently leads), and that Gov. Peter Shumlin (D-VT) is reelected by his state’s Democratic-controlled legislature because of his failure to win a majority on Tuesday night (likely), we will have picked 31 of 36 of these very close contests, with those we missed breaking to the Republicans, giving them a very good gubernatorial year.

Our best, most consistent gubernatorial call was listing Gov. Scott Walker (R-WI) as a favorite for the entire cycle in one of the most-watched races. He ended up winning comfortably.

There was a lot of uncertainty going into this election: We now know that much of the polling was flawed, and there were big questions about who would show up to vote and how extensive a drag President Obama would be from state to state. If we understated anything, it was the potency of the anti-Obama, anti-Democratic environment, which led to a slightly better GOP year than we expected in all three of these categories.

That said, we’re pleased with our record this year, and we want to thank the many sources who helped us along the way (you know who you are).

We’ll offer some broader observations about 2014 next week after we get a chance to dive into the data. Then: It’s on to 2016.