Mark Mellman: Finding the forecast

Mark Mellman: Finding the forecast
© Greg Nash

With control of the Senate hanging in the balance, all the drama appears to be in that body, with far less focus on the House, which is important in its own right as a test bed for the varieties of forecasts now widely available. 

History suggests Democrats will lose House seats, and so does everyone else. 


But the number remains quite uncertain. 

Only three times in the last century has the party in power in the White House failed to lose seats in the midterm elections, and those three were truly exceptional circumstances. On average, the White House party has lost 30 seats.

It’s a tribute to Leader Nancy Pelosi, Chairman Steve Israel and the staff of the Democratic Congressional Campaign Committee that under such circumstances they have been able to outraise the parallel Republican committee by some $40 million. 

Republicans have set a goal of picking up 11 seats, quite modest by historical standards.

One analytic approach, adopted by several D.C.-based handicappers, incorporates analysis of individual districts with an examination of the overall political environment. Veteran analyst Stuart Rothenberg is suggesting GOP gains will range between two and 10 seats, while House specialist David Wasserman of the Cook Political Report projects a similar two- to 12-seat increase for Republicans. Kyle Kondik and Larry Sabato expect gains in tighter range, of five to eight seats. 

Kondik recently analyzed trends in the generic vote question as part of his case for the range he is predicting: “For the sake of argument, let’s assume that Republicans continue to gain in the generic ballot average so that they reach a national House polling lead of 9.4 points (the Election Day generic average on RealClearPolitics in 2010) or 7.7 points (HuffPost Pollster). And let’s also assume that, on Election Day, Republican House candidates win the overall national House vote ... by 6.8 points, matching their 2010 performance. That performance produced a House with 242 Republican members”— an eight-seat gain in today’s House.

Then there are the modelers who rely on detailed statistical analyses that usually combine the generic vote with a variety of historical district- and national-level data into algorithms less straightforward than Kondik’s formula.  

A model developed by The Washington Post and Monkey Cage projects a 10-seat GOP gain. 

A quartet of academics recently offered forecasts derived from their models, some of which predict more change. Emory’s Alan Abramowitz gave Democrats their best showing, projecting a four- to eight-seat GOP gain, in line with race handicappers. 

James King of the University of Wyoming produced the best forecast for Republicans, foreseeing a 39-seat gain for the majority party. Robert Erikson of Columbia was in-between, with a 14-seat GOP gain, and the State University of New York, Buffalo’s James Campbell was very close to that, projecting a 16-seat GOP gain. 

Average those academic forecasts and you get a 13-seat shift toward the Republican Party. Leave out the best and worst forecast for each party and it’s a 15-seat shift. 

Four things are clear for this run-down:

First, no one yet knows just how many seats Democrats will lose in this very challenging environment. 

Second, Democratic losses are likely to beat the historical average, making it hard for Republicans to claim real bragging rights.

Third, for the most part, academic modelers foresee bigger changes than do the D.C.-based handicappers. By the handicappers’ lights, Republicans are unlikely to achieve their goal; the modelers say it’s quite likely they will meet and even exceed their modest objective. 

Finally, with forecasts ranging from Democratic losses of two to 39 seats, this election sets up an interesting test of alternative approaches to forecasting.

In the end though, unlike with the probabilistic models of Senate control, we will know who was more and less right.

Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the majority leader of the Senate and the Democratic whip in the House.