Democrats could gain as many as 44 seats and emerge from November's midterm elections with control in the House of Representatives, according to a new forecasting tool.
The model predicts a different outcome for the Senate – where Republicans are likely to pick up an additional two Senate seats and preserve their majority.
"The midterm election's outcome will play a major role in policy making and the politics leading up to the presidential election of 2020," says James Campbell, professor of political science at the University at Buffalo and creator of the Seats-in-Trouble model.
Since 2010, Campbell, author of Polarized: Making Sense of a Divided America and an expert in campaigns and elections, has used his model to predict election outcomes with a high degree of accuracy.
"I'm approaching this scientifically in an attempt to get the best forecast," Campbell says. "The model does not explain why there is partisan seat change, but explanation is not its purpose. It's purely predictive."
The Seats-in-Trouble model is distinct from related forecasts which rely mostly on either broad handicapping or national indicators such as the economy or presidential approval ratings.
The model combines individual analysis by experts in each district, information the Cook Political Report provided, with a statistical analysis of historical partisan seat change.
"That's the model's strength," Campbell says. "It's a hybrid that digs into the analysis right at the district level and pairs that with the historical record and how the two have matched up."
Since the mid-1980s, the Cook Political Report has reported pre-election ratings of the competitiveness of congressional elections.
The information has been publicly available on its website since 2008. Cook rates races for the Democrats and the Republicans as solid, likely, leaning toward the incumbent party, or a toss-up.
The model's debut in 2010 predicted a Democratic loss in the House of 51 or 52 seats.
The party actually lost an additional 12 seats, but seat losses that year were greater than at any time since the 1920s and no systematic forecast made before Labor Day that year was more accurate than Campbell's.
The model did well two years later in predicting Democratic gains.
But Campbell, a devout Boston Red Sox fan who once wrote an article on the dead-ball era of congressional elections, began to hone his forecast to improve its batting average.
Small shift, tall order
"When I first used the Cook Report's archive, I counted a seat as 'in trouble' if it was currently held by a party and was considered 'leaning' or 'a toss-up.' That was too generous," he says. "I needed a tighter definition of a seat-in-trouble. I think leaning districts are closer to being safe than they are closer to being toss-ups."
Campbell's predictions for the 2018 midterms align with the history, but he sees magnitude as the critical component when discussing partisan seat change.
"The President's party routinely loses House seats in the midterms. That has been the case in all but three midterm elections since 1900," he says. "Based on that history of seat changes, the real question is not which party will gain seats, but whether Democratic House seat gains this year will be small or large."
When considering the congressional arithmetic for 2018, or what each party needs to hold or win a majority, Campbell's says Democrats face a bigger challenge in the Senate.
"Democrats require only a two seat gain for majority control. This seemingly small shift, however, is a tall order," says Campbell. "Democrats are defending many more seats than Republicans this year. They have to do well just to hold their current overall numbers."
The forecast will appear in an upcoming issue of PS: Political Science & Politics.