The PollyVote survey of elections experts for April, forecasting the presidential election, predicts that President Obama will garner 51.2% of the major-party vote.  This forecast for Obama is 0.4% lower than the experts' estimate in March, though identical to their February prediction.

The survey was received by panelists on April 11, the day after Senator Santorum's withdrawal from the race.  Though the link between distribution of the survey and the Santorum announcement was unplanned, it did enable the experts to factor that news into their estimates. 

The experts' forecast of 51.2% for Obama is somewhat lower than Obama's support in campaign polls that pit him against Gov. Romney, the presumed Republican nominee.  Polls conducted during the expert survey period, April 11 – 20, give Obama 51.9% of the combined Obama and Romney "vote".

Experts' estimates also are lower than predictions of the Iowa Electronic Market.  During the survey period the value of the Democratic vote-share contract on the Iowa market averaged $0.54.  Taken as a share of the combined Democratic and Republican contract prices, the adjusted value of the contract for the Democratic nominee is $0.523.  This implies that the Democrat would receive 52.3% of the vote, 1.1% higher than the experts' 51.2% estimate for Democrat Obama. 

This most recent survey continues the pattern of previous assessments by the expert panelists, that President Obama is likely to win reelection.  The forecast margins favoring Obama, however, have been slim, ranging from 50.7% to 51.6% of the major-party vote.

From today on, Polly's combined poll component forecast draws upon polls that ask voters for whom they would vote in a likely showdown between Obama and Romney.

There are two reasons for this decision. First, in the past two months, there have been almost no polls that ask voters about their intention to vote for a generic Republican candidate. Second, it appears highly likely that Mitt Romney will be the Republican nominee.

The PollyVote team has completed its third monthly survey of elections experts to forecast the 2012 presidential election.  Experts were surveyed March 9-13 and predicted that President Obama will receive 51.6% of the two-party vote against the Republican candidate.  This result is the second slight increase in Obama’s projected vote share since the first survey was conducted in December and early January.  At that time the panel forecast a 50.7% vote for Obama.  In the February survey the experts’ prediction rose half a point to 51.2% for Obama, and has now increased an additional 0.4% in the March survey, to 51.6%.

While the March expert survey was underway, the Iowa Electronic Market’s vote-share contract for the Democratic candidate traded at an average price of 52.9 cents a share, implying a forecast that Obama will receive 52.9% of the vote.  Thus traders on the Iowa market anticipate a larger vote for Obama than do the expert panelists, by a bit more than one percent.    

Polls taken during the March survey period also were more favorable to the president than were the experts’ assessments.  When paired against Romney, Obama was preferred by 52.5% of respondents, 0.9% higher than the experts’ estimate for Obama.  Against Santorum, Obama was the choice of 54.7%; against Gingrich, Obama was favored by 57.4%.    

In the March survey, though the expert panelists have again nudged Obama’s predicted vote share slightly higher, their estimates are somewhat more cautious than predictions based on polls and trading prices in the Iowa futures market.  The experts continue to anticipate a close election.

The research paper that explains the combining approach used in the PollyVote has been updated. The latest version can be downloaded for free from the SSRN website.

Combining Forecasts: An Application to Elections

Graefe, A.; Armstrong, J. S.; Jones, R. J. Jr. & Cuzán, A. G.

Abstract. Combining forecasts is simple, inexpensive, and effective, yet few organizations use it. This occurs because the benefits of combining are counter-intuitive and because people are unaware of the empirical evidence. Here we summarize the literature on the effectiveness of combining by assessing the conditions under which it is most valuable. Using data on U.S. Presidential elections from 1992 through 2008, we then examine the error reduction obtained by averaging forecasts within and across four groups of methods (polls, the Iowa Electronic Markets, quantitative models, and expert judgment). The gains in accuracy from combining increased with the number of component forecasts used, especially when these forecasts were based on different methods and different data. Combining yielded error reductions ranging from 10% to 61% compared to the average errors of the uncombined individual forecasts; this improvement is substantially greater than the 12% that had been previously reported for combining. We also show combining is especially effective in situations involving high uncertainty.

The answer to this question is provided by the big-issue model. This model predicts the election outcome based on information about which candidate voters favor when it comes to dealing with the most important issue facing the country. The research paper in which this model is presented has now been published in the January 2012 issue of the Journal of Behavioral Decision Making.

Predicting elections from the most important issue: A test of the take-the-best heuristic

Graefe, Andreas: Armstrong, J. Scott, Journal of Behavioral Decision Making, 25(1), 41-48.

Abstract. We used the take-the-best heuristic to develop a model to forecast the popular two-party vote shares in U.S. presidential elections. The model draws upon information about how voters expect the candidates to deal with the most important issue facing the country. We used cross-validation to calculate a total of 1000 out-of-sample forecasts, one for each of the last 100 days of the ten U.S. presidential elections from 1972 to 2008. Ninety-seven per cent of forecasts correctly predicted the winner of the popular vote. The model forecasts were competitive compared to forecasts from methods that incorporate substantially more information (e.g., econometric models and the Iowa Electronic Markets). The purpose of the model is to provide fast advice on which issues candidates should stress in their campaign.

Full text available here.

The PollyVote team has now completed its second survey of elections experts to forecast the 2012 presidential election.  Experts were surveyed from February 3-11 and, at this point, are predicting that President Obama will receive 51.2% of the two-party vote against the Republican candidate.  During a comparable survey in December and early January, the experts forecast a 50.7% vote share for Obama. Thus the panelists, 16 scholars in American politics, believe that the outlook for Obama has improved slightly during recent weeks.    

During the survey period the Iowa Electronic Market’s vote-share contract for the Democratic candidate traded at an average price of 50.9 cents a share, implying a forecast that Obama will receive 50.9% of the vote.  Thus traders on the Iowa market are anticipating a virtually identical election outcome as are the expert panelists.

Also during the survey period several campaign polls were conducted, pitting Obama against the four remaining Republican candidates.  For purposes of comparison let us assume that undecided interviewees vote for the candidates in the same proportion as those who have chosen a candidate.  In this instance Obama performs better in the polls than in either the expert surveys or the Iowa markets.  When paired against Romney, Obama's preference share in the polls is 53.3%.  Against Santorum, Obama is the choice of 55.4%; against Gingrich, 55.4%; and against Paul, 56.3%.

In this early February snapshot we find a consensus among these three forecasting methods that Obama is leading the opposition Republican field.  Of the forecasts made this early in past election cycles, elections experts and traders on the Iowa market both have been more accurate in predicting the election outcome than have polls.  The experts and market traders appear to be telling us that, while Obama currently has the edge, the election is likely to be very close.

A note on methods and data:  The 51.2 forecast is the trimmed mean of the individual forecasts in which the high estimate and the low estimate are omitted before the mean is calculated.  The objective of this procedure is to reduce the influence of outliers.  The mean probability estimate of the included forecasts is .87 for a +/-5% confidence interval.  Names of participating experts will be released after the election, with their consent at that time.  Poll data are from realclearpolitics.com, and market data are from the IEM. 

A new ABC News / Washington Post poll, conducted from February 1 to 4, provides information on how voters expect President Obama and likely GOP nominee Mitt Romney to handle some important problems facing the country. Voters favor Obama for issues related to the middle class, terrorism, international affairs, and taxes. Mitt Romney is favored for handling the federal budget deficit and dealing with the economy, which is the most important problem seen by voters. On the issue of jobs, voters are tied.

This information is included in two index models for forecasting the election outcome:

• The Issue-index model predicts the election based on how voters expect the candidates to deal with all issues. The latest model forecast predicts Obama to gain 55.2% of the two-party popular vote.
• The BigIssue model predicts the election based on how voters expect the candidates to deal with the most important problem. This model currently predicts Obama to gain 52.2%.