Two election models, like recent polls, tilt toward Obama

September 12, 2012|By James Rainey
  • Before voters hit the polls in November, five things could affect the tight presidential race. Among them: debate gaffes and a September surprise.
Before voters hit the polls in November, five things could affect the tight… (Luis Sinco/Los Angeles…)

There are two competing temptations when looking at the boost in support for President Obama after the Democratic National Convention: to take the surge too seriously or not seriously enough.

With eight weeks to go until the election, plenty of time remains for unforeseen events (a dramatic economic event, a candidate’s stumble, a terrorist attack or some other calamity) to change the dynamic. But a couple of statistical analyses suggest that the recent polling has some significant, if not conclusive, meaning. And the findings all accrue to Obama’s benefit.

Politics Now cited a University of Colorado election model that several weeks ago projected a Mitt Romney win in the presidential election. But that model subsequently came in for a beating from analysts such as New York Times statistical whiz Nate Silver, who found “glaring problems” with the methodology used by the Colorado political scientists. Among other problems, the Colorado model put a huge weight on unemployment rates — but only with a Democrat sitting in the White House.

Silver and others modeling the election, including the financial prognosticator, now find that Obama’s chances of reelection have surged to a high for the year. The Obama bounce has been solid, if not exactly breathtaking, and certainly nothing like what might be known as the “Lehman Bounce” in 2008.

Four years ago, many Americans interpreted the bankruptcy of financial giant Lehman Bros. as a strong sign that the Republican Party should no longer be managing the nation’s economic affairs. Silver recalled in his blog Monday how, after the Lehman collapse, it was clear Republican John McCain would lose, “with Barack Obama’s projected probability of winning the Electoral College increasing by about 25 percent in a period of just 48 hours.” McCain never fully recovered.

Silver’s model, factoring in the candidates’ chances of victory in key swing states, has put the likelihood of an Obama victory at 60% to 70% through much of the summer. But on Tuesday it set the chances of a second Obama term at a high of 79.8%.

NerdWallet, a company that started analyzing credit card rates, extended its data analysis to fields such as education and politics. Joanna Pratt, who heads election analysis for the company, acknowledges that the political sideline was intended to bring more attention to the company’s core business. 

The company looks at polls, particularly in battleground states, and their historical performance in elections. In general, the company notes that polls become more predictive of final outcomes as election day approaches. (No huge surprise there.)

On Tuesday, NerdWallet put Obama’s chance of reelection at 78.4% — just shy of the probability predicted by Silver.  Of 12 states that analysts expect to be most closely contested, NerdWallet projects only two — Missouri and North Carolina — as most likely to go to Romney. Obama has more than a 50% chance of victory in 10 other battleground states: Colorado, Florida, Iowa, Michigan, Nevada, New Hampshire, Ohio, Pennsylvania, Virginia and Wisconsin.

Among the possibilities that could cause the analysis to misfire, according to NerdWallet: “Dramatic events or substantial new information between now and the election....”

Twitter: latimesrainey

Los Angeles Times Articles