Since the US Presidential election last week a few reasons have been posited for the re-election of President Obama. Today we will look through the vast array of exit poll demographic numbers to determine which explanations of result hold weight. In this post we are using the data from the exit polls conducted for the National Election Pool in 2012 and 2008. The National Election Pool is a consortium of US Television Networks who use the exit poll figures on their broadcasts and websites.You can find the full figures as published by CNN.com (2012, 2008) by clicking the links.
Minority voters won the election for Obama
A common explanation of the result was that Obama continued to attract the support of minority voters; Black, Hispanic and Asian voters, this advantage was compounded by the increasing shares of the electorate which these racial minorities attracted. But was the increased support from racial minorities vital in tippling the election to Obama?
To determine this we shall use a hypothetical. Let’s assume that last week’s election took place with identical turnout to 2008, with each racial group maintaining their 2008 share of voters. Let us further assume that every racial group had the exact change in voting shares that occurred with White voters. What we are trying to do here is remove any change in the vote specific to minority voters to see is the absence of this change alters the outcome of the election. Once we take these assumptions into account, the swing, or change in the major party vote shares in any given state will equate to the swing in the white vote exhibited by the exit polls.
One caveat before we begin is that exit polls and the subsamples within have a margin of error. Furthermore as exit polls are conducted by cluster sampling, the margin of error is almost double than that of regular polls. Once again we are using the exit poll data as it is the only data we can use. But keeping in mind margins of error (and rounding) this exercise is indicative but not conclusive.
First let us turn to the national vote. Please see the table below.
As you can see there was a 4% swing to the Republicans with White voters which was ameliorated for Obama with a lesser swing to the Republicans of 2% with Blacks and a swing of 4% to Obama amongst Hispanics.
If the whole nation swung to the Republicans by 4% then this would have given Romney a narrow popular vote victory 49.7-48.9. However due to rounding of the exit poll figures for both years this swing could be anywhere between 3-5% which gives us range of possibilities between 49.9-48.7 Obama and 50.7-47.9 Romney.
How would the Electoral College pan out? Here is table of the swing states with the 2012 White exit poll swings applied to the 2008 statewide votes of the states which were won by Obama last week by 10% or less.
We have left states with a final margin of 2% of less under this scenario as undecided given that due to rounding of exit poll figures, the swing amongst Whites could be +/1%. Below is a map assigning the states based on our scenario? We have used the map at 270towin.com which is an excellent site for playing around with Electoral College scenarios present and past.
As you can see this race is too close to call. When we set aside the close states , Wisconsin, Michigan and Pennsylvania no candidate has an electoral college majority, however this map does favour the Republicans 262-230. If we assign the undecided states to the leaders we get a 282-256 GOP majority in the Electoral College.
Conclusion: Obama’s increasing advantage with Hispanic voters and the increase in turnout of minority voters definitively won him the states of Florida, Ohio and Colorado. The same was probably the case in Pennsylvania and the national popular vote and is also possibly the case in Wisconsin and Michigan. Obama probably won the election due to his increasing advantage with Latino voters and the increased minority turnout and at the very least these effects can be certain of comprising the margin of victory in the Electoral College.
Another theory is the advent of Hurricane Sandy late in the campaign gave President Obama a few days to act “presidential” and improve his popularity as the incumbent in a time of crisis. An alternative theory is that the dominance of Hurricane Sandy in the news prevented the challenger from having any chance to impact the race in the last week. We can attempt to assess the former but not the latter. We cannot “replay” reality in the last week to see if a shift toward Governor Romney would have occurred without the Hurricane.
We can assess the effect of Hurricane Sandy by assessing the shift in the poll average in the last week and a half of the of the campaign from October 26, which was the last day news of the Hurricane did not dominate the news.. We will then add this shift in the polls to the current margins in each state to determine which states have moved in the Obama column in the last week and a half. The table is below.
We estimate that Florida shifted to Obama in the Hurricane Sandy period. We also note that the margins in Virginia and Ohio would be very tight but still have Obama in the lead. This scenario would garner a 303-235 electoral vote majority for President Obama. Even if one were to shift Virginia and Ohio into the Romney column, this is still a 272 -266 electoral vote victory for the President.
Conclusion: The campaign effect of Hurricane Sandy was to increase President Obama’s electoral college margin, but it did not flip the race. However we cannot say what would have occurred if Hurricane Sandy never happened.
The Religious Right political influence is on the wane
In 2004, the victory of President George W. Bush was attributed to a large turnout of evangelical or born again Christian voters. The influence of evangelical Christian churches and organisations has been evident in US Presidential politics at least since the election of Jimmy Carter in 1976.
However in this election, the Republican loss in the presidential election combined with the victory of pro-gay marriage ballot initiatives in Maryland, Maine and Washington state, the defeat of an anti- gay marriage ballot initiative in Minnesota and the victory of marijuana legalisation ballot initiatives in Colorado and Washington has driven the suggestion that the influence of the Christian Right is waning in US politics.
Lets turn to the exit poll numbers to examine this explanation. Below is a table of presidential preference of self described White Evangelical or born-again Christian voters and the share of all voters they comprised. This is the metric which best measures the influence of the Christian Right.
As you can see White Evangelical voters made up a larger share of voters than in 2004 and voted for the Republican candidate in the same numbers as in 2004. The difference is that non-white evangelical voters lean slightly harder to the Democrats. Whether this is artifact of an increasing propensity of racial minority voters to vote Democratic we cannot say for sure but it is a possible explanation.
Conclusion: White Evangelical voters are still a significant share of the electorate and to claim that they are on the decline is premature, when the share of White Evangelical voters is increasing. Therefore the influence of the Religious Right is not on the decline if their share of voters are increasing.
In any single member election the difference between the winner on a low 50’s share of the vote and lose on a high 40’s share of the vote is often reported as a profound difference between the parties and something that will be hard to overturn for the foreseeable future. However such reporting is an artifact of “winner take all” elections. While the small difference between the two voting coalitions is vital for victory such small differences are not impossible for the losing side in one election to make up in the next cycle.
Consequently, reasons for one party’s victory take hold that exaggerate the factors behind the victory. This is just short term “after the event” myopia, a sober analysis will reveal more nuanced story.