It is confusing to determine where we are in the Presidential race. Headlines scream out: Obama up by four, Romney and Obama tied in latest poll. Then there are the confusing state poll results. On the same day a poll shows Romney leading in a national poll, a state poll will show Obama pulling away in a “battleground” state.
This article sets forth a model that can be used to accurately describe the current state of the Presidential race and who is likely to win in November.
The first thing that has to be considered is that since there is an incumbent running for reelection, that fact has a major impact on how the race unwinds. A very accurate indicator of how the incumbent will perform on Election Day is how people perceive his performance on the job. People’s perception of his performance is measured by his job approval numbers. Polling companies regularly take polls asking the question, do you approve or disapprove of the performance of the President. Those numbers are expressed as approval/disapproval numbers. Thus, a 52A/48D result would mean 52% of the sample approved of the President’s job performance and 48% disapproved of the President’s job performance.
Not surprisingly, there is a strong correlation (if not cause and effect) between the job approval numbers and the total vote percentage the incumbent receives in November. Historically, the President will receive between 0% and 1% more than his job approval numbers. For the purposes of this model, we will assume the “extra” percentage is 1%. Thus, if the job approval number for the President is 50%, then the President is very likely to receive 51% of the vote in November.
That takes us to today’s situation. I have taken a 30 day moving average of the job approval numbers for President Obama distributed by the Rasmussen Polling Service. Rasmussen has been the most accurate predictor for the last few election cycles and he uses a likely voter sample which is more accurate than registered voters and much more accurate than an all persons sample. Further, by averaging the last 30 days results, this smoothes out the results and practically removes all traces of sample error.
Using the 30 day moving average ending on July 30, 2012, President Obama has a job approval average of 47.57% (as a point of confirmation, the Real Clear Politics approval numbers are 47.4%). If we use the “extra” percentage described above of 1%, we can estimate that President Obama will receive 48.57% of the vote in November (47.57% + 1%).
However, our President is chosen by the Electoral College which means there are 51 (counting D.C.) state by state elections. How can we predict how those total national votes will be distributed through out the states? In this model, I use the factor of how the states have voted in the Presidential elections compared to the national vote totals.
For example, if a Democratic candidate received 48% of the vote nationally and received 46% in Iowa that would make that state a minus 2% state for the Democrats (48% – 46%). In this model, I compared the state voting patterns for the last three presidential elections versus the national vote total. I then added up the divergences and divided by three to come with whether a state was a plus or minus state for Democrats. In completing this average, I used 1.5 times the year 2008 divergence, 1 times the year 2004 divergence and 0.5 times the year 2000 divergence so the more recent results carry greater weight.
Using this approach, and the above projected vote percentage for President Obama of 48.57%, the Electoral College Votes should be divided 291 for Romney and 247 for Obama. As an example of how this model works, we will look at the battleground state of Ohio. In Ohio, the Democratic incumbent should perform 0.92% worse than his national totals. With Obama’s national vote total projected at 48.57% that would mean Obama should receive 47.65% of the vote in Ohio (48.57% – 0.92%). The State of Pennsylvania is a plus 2.05% Democratic State. The Democratic candidate should perform 2.05% better than his national vote total. Thus, President Obama should receive 50.62% of the vote in Pennsylvania (48.57% + 2.05%).
This model is far more accurate than the national polls where up to 15% of the vote is undecided. It is historically proven that most of the undecided vote goes for the challenger. That result is inline with the other historical fact that the incumbent receives essentially the same vote percentage as his job approval numbers. Those who are satisfied with the job performance would vote for reelection. Those who are dissatisfied with the job performance are split into two camps. The first camp is those who will vote for the challenger. However, there is until late in the election cycle a large percentage of voters who will not vote for the incumbent but have not made up their minds about the challenger. Even Ronald Reagan did not pick up many of the dissatisfied with President Carter voters until very close to the election. Thus, poll results that show Obama up 44% to 40% do not place Obama in the lead. In reality, when undecided voters are attributed to Obama and the challenger based upon historical results, that poll shows Romney with the lead.
As for the battleground states, here is how the model predicts the states to fall with the vote totals for President Obama:
|Obama Vote Total||Obame EC||Romney EC|
Some may argue that states could change their voting pattern this time as shown in state polls. State polls are the least reliable because of the small sample sizes and the fact that only infrequently are state polls taken. Further, though states voting patterns change over time versus the national numbers, they do not change radically in one presidential election as a comparison versus the national numbers. For example, Oklahoma over time has become even more of a Republican State and Vermont has become a stronger and stronger Democratic State. That has no effect, as those states were already safely in their respective columns. Montana has become less hostile to Democrats over the last three cycles going from minus 15% to minus 10% to in 2008 minus 6% versus the national numbers for Democrats. However, that still does not place the state in the battleground area.
Thus, we need to examine the above battleground states to see if any of them have had significant movement that would lead one to believe that even though they may be barely in the Romney column, considering the historical trends, they should be moved to the Obama column and vice versa. Colorado, Nevada and New Hampshire (especially Nevada) have had significant movement in the last three cycles toward the Democratic ticket when compared to the national vote. The movement has been so strong I believe the small lead enjoyed by Romney for Nevada and New Hampshire above is not sustainable with Obama’s current job approval numbers. Colorado may provide Romney with enough of a lead to hold off the historical trend. Further, on the Romney side, Virginia has had a big move toward the Democratic side. But, like Colorado, I believe Romney holds a large enough lead there to hold off the historical tide.
On Obama’s side of the battleground ledger, no state has moved enough over the last three elections to indicate the model should be overridden. Pennsylvania has moved ever slightly toward the Republican fold and New Jersey has made some impressive moves toward the Republican side, but it is too far in the Obama camp to see the historical trend lead New Jersey to the Romney column.
Thus, if we subtract out Nevada and New Hampshire from Romney and add them to Obama, Romney should receive 281 Electoral Votes and Obama 257 in November.
Obviously, this projection is based upon President Obama’s current job approval rating. If that changes significantly, then the projections above using this model will change right along with it. It should be noted that Obama’s 30 day moving average has been stuck between 46.5% and 49% since February of this year, and the vast majority of the time it has been very close to his current approval numbers of 47.57%. Thus, unless big events occur in the economy or in the World, it is unlikely big changes in the job approval rating of Obama will occur.