BUFFALO, N.Y. -- James E. Campbell, a University at Buffalo political scientist nationally recognized for his highly accurate election-prediction models, says that this year the Democrats are likely to pick up between three and 14 seats in elections for the U.S. House of Representatives.

This prediction is from the "Seats in Trouble" forecasting models of partisan seat change in U.S. House elections and will be published in the October issue of PS: Political Science and Politics. It will be available online in mid-September at: http://www.apsanet.org/PS/.

The "Seats in Trouble" forecast is based on two forecasting equations published in 2010 by Campbell, UB Distinguished Professor of Political Science and department chair. His models have had a good deal of success in predicting presidential and congressional elections since 1992.

Campbell explains that the core variable in the "Seats in Trouble" model is calculated from competition assessments of district races published by The Cook Political Report in August of the election year.

"The number of seats in trouble for a political party are those considered by the Cook Report to be toss-ups or worse for the party currently holding the seat," he says.

"The difference between the number of each party's seats in trouble is the critical predictor variable in my equations," Campbell says, adding that the equations also take into account the number of seats each party won in the previous election and the president's approval rating in August of the election year.

He says, "The equations proved to be quite accurate in 2010, the first election in which it was used. In late August of 2010, the equations predicted that Democrats would lose about 52 seats, about the magnitude of their 1994 midterm loss and the largest seat change since the Truman-Dewey election of 1948. Though the forecast was short of the actual 64-seat Republican landslide, no other early forecast was more accurate."

This year, according to the handicapping of races in late August by The Cook Political Report, Democrats had 15 seats in trouble and Republicans had 21.

"This six-seat difference favoring the Democrats is close to the middle of the range of past values," Campbell says, "and since 1984, the best seat-in-trouble difference for Democrats was 27 seats in 2008 and the best for Republicans was 44 seats in 2010"

Based on the "Seats-in-Trouble" index, along with a 45 percent approval rating for President Obama from Gallup in late August and the 193 seats Democrats held after the 2010 midterm, Campbell says the seats-in-trouble forecast equations indicate that Democrats are likely to gain between three and 14 seats this year.

"The predicted gain of three seats is derived from the equation that employs the 'Seats-in-Trouble' index along with presidential approval. The predicted gain of 14 seats," he says, "uses the Democratic seat base along with the 'Seats-in-Trouble' index."

Campbell says, "Based on these two equations, I think we should expect Democrats to register small gains in the House this year. The presidential approval version of the equation indicates particularly small Democratic gains since the presidential election is likely to be quite close, as reflected in President Obama's borderline approval ratings.

"The version of the equation that takes the number of seats a party currently holds into account indicates somewhat larger gains," he says.

"This reflects the fact that Republicans made gains in 2010 in areas that historically have been Democratic turf. One might expect the gains of the 2010 Republican wave to recede to restore a few more Democrats to the House, though it will probably not be enough to threaten continued Republican control. Democrats would need to pick up 25 seats for that to occur."

"The country recently saw three unusual wave elections: 2006, when Democrats took 31 seats and control of the House; 2008, when they extended that control, and 2010 when the Republicans picked up 63 seats and regained control.

"I think we may see a return to less volatile House elections," Campbell says, "elections like the 10 national congressional elections held between 1986 and 2004, during which time only one, in 1994, produced a double-digit change in House seats."

"I would not be surprised if the 2012 election restored that pattern of small seat swings."