Great uncertainty still lies before us

For months, the narrative has been imprinted on our brains: Weighed down by an economy that still feels mired in recession and a president whose popularity is sagging as a result, Democrats face electoral Armageddon in November. Fueling this plot line has been a relentless focus on Republicans’ rising share of the “generic” vote, and more recently, on state and district surveys, all sifted by Washington’s leading political prognosticators.

Don’t misunderstand: Democrats will lose a good many seats this year. However, answering the ultimate political questions — which party will control the House and Senate — requires far greater precision. The difference between losing 36 House seats and 40 is all the difference in the world. 

{mosads}Some develop such granular assessments from the bottom up — examining the particulars of each House and Senate contest. Others take a mostly top-down approach, forecasting seat changes by analyzing national conditions. This latter group, my focus here, can be further subdivided into those who look at economic and political fundamentals alone and those who add poll data, particularly the generic vote, to their calculations. Careful examination of the varying results produced by these methods adds some air of uncertainty to the prevailing narrative.

Two weeks ago, I considered this ever-bouncing generic vote (though some saw my allusion to Mitch Miller’s bouncing ball evidence of creeping decrepitude). It’s still bouncing, with no clear resting place in sight. Gallup led us from a widely reported 10-point Republican advantage to a mostly ignored 1-point Democratic margin in two weeks. They now report a chilling 18-point GOP landslide in a low-turnout scenario, which dwindles to 3 points among all registered voters. Newsweek compounds the confusion, giving Democrats a five-point margin over Republicans among “definite voters.” GOP-friendly Rasmussen finds Democratic momentum in the generic vote — a 12-point Republican advantage in early September shrinks to 10 in mid-September and to just 3 points as October began.

With Democrats somewhere between 5 points ahead and 18 points behind, and either gaining or losing ground, all we can safely say is that this poll question has not yet found a comfortable equilibrium. 

Digging deeper also exposes a Republican-dominated generic vote wildly out of sync with public perceptions of the two parties. Voters certainly dislike Democrats — according to the CBS/New York Times poll, 45 percent have a favorable view of the Democratic Party, while 48 percent harbor unfavorable views, yielding a net favorability rating of -3. But voters detest Republicans whose net favorability is a gaping -22 — one of the most negative judgments of either party, ever. Usually the generic ballot is strongly correlated with the relative image of the parties. Today, many polls ask us to believe the far more unpopular party will collect by far the most votes. Possible, to be sure, but the discrepancy at least adds further uncertainty to a narrative that admits none.

Adding to the veiled mists, academic models focusing on political and economic fundamentals suggest Democrats will lose fewer seats than those that incorporate the generic vote as a key variable. Forecasting models developed by five sets of academics, built around economic and political data, but not the generic vote, project an average loss of 35 House seats — just under the magic number at which control of the chamber shifts. Three models based at least in part on the generic vote estimate an average loss of 14 additional seats, and control along with them.

So which generic vote, if any, will rule on Election Day? Will voters’ revulsion at Republicans limit their gains? Will fundamentals or the national polls prove more prescient? An accurate narrative requires greater humility in assessing our powers of prediction and greater acknowledgment of the uncertainties yet before us.

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