What will the Electorate look like on Election Day?

This has probably been the biggest bone of contention among conservative critics of polling this cycle.  It certainly has been the biggest issue with this blog.  We blogged a few different thoughts on the debate from the invaluable Jay Cost’s D +3 estimate to The Winston Group’s historical perspective. Chris Palko of Smart Media Group whose awesome Battleground Counties piece from April inspired the below blog post has another sharp write-up on this debate:

When viewed from certain angles and in certain perspectives, it looks like President Obama is a solid favorite to be reelected. Most public polls have him up, and he appears to be running ahead of his national numbers in the key swing states. But viewed from other angles, Mitt Romney can seem stronger than these polls show. You can look at the changes in the overall political landscape since 2008 and conclude that Romney is being underestimated. What is underlying these disparate views of the election is that no one is sure of what the electorate will look like this November.

The Great Party ID debate: Who will show up on election day?

2008 had the most Democratic electorate in a generation. Accordingly, Barack Obama was the first Democratic presidential candidate to earn over 50.5 percent of the popular vote since Lyndon Johnson in 1964. Thirty-nine percent of the 2008 electorate self-identified as Democrats, compared to 32 percent who identified as Republicans. [This is what we refer to as D +7]

Polls

Most public polls are assuming a turnout model roughly similar to the 2008 electorate. If this was the case, Obama would certainly win. One could make the case that Democrats are more genuinely enthusiastic for Obama than Republicans are for Romney. And it does appear that Obama is running ahead of where a generic Democrat would be running.

What if 2008 was not the constant?

There is another way to look at the electorate. What if the composition looked more like the 2004 or 2010 elections? In both of these elections, the Democrat/Republican self-identification was identical. In 2004, Kerry won independents by one point, but Bush was able to win by getting 11 percent of Democrats to vote for him, compared to 6 percent of Republicans voting for Kerry. In 2010, Republicans won a larger victory because they won independents by double digits. If this was true, a Romney win would be hiding in plain sight.

Inaccuracy in polling

Most signs do not point to a 2008 repeat in turnout disparity, but a top to bottom look at races across the country also doesn’t seem to point to a 2010-style Republican wave either. Most of the uncertainty over where the state of the race is and will be on Election Day is over the partisan makeup of the electorate. Around 95 percent of those who identify with a party will vote for that party’s presidential candidate. Because this is such an ingrained preference, approximating the correct partisan breakdown of the electorate is the most critical part of getting an accurate view of the electorate. If pollsters presumed a 2008 turnout model but the actual election revealed a 2004 model, then there would be a large disparity between final predictions and the results.

Racial make-up also skews today’s polls

Besides partisanship, the weighting of racial groups is also very important for determining the outcome. In 2008, 13 percent of the electorate was African-American—an outsized turnout compared to past elections. In 2004, African-Americans were 11 percent of the electorate. Because African-Americans are nearly unanimously in favor of President Obama, a one percent change plus-or-minus in the share of the electorate can make a great difference. Hispanic turnout is also something to watch. Even though the Hispanic share of the population is increasing, the share of the electorate that is Hispanic has remained stuck at 8 to 9 percent since 2004. If Hispanic turnout stays the same or, as some reports suggest, declines then that can be a determining factor in a series of swing states like Florida, Colorado and Nevada.

Palko smartly points out many ways today’s polls can be juiced through many different means.  What we are seeing recently are polls view in the most favorable light possible for President Obama: Democrat turnout advantage at or exceeding the best in a generation performance of 2008, minority enthusiasm and support levels equal to 2008 though out-sized relative to every prior election, a motivated youth population disproportionately excited for one candidate.  If Barack Obama can achieve that in 2012 then maybe these polls are correct.  Every ounce of evidence points to the contrary but just like football teams being favored ahead of the game, you don’t know how it will play out until the game actually starts.  In November, we’ll see if these “light most favoring the plaintiff (and then some)” polls are accurate. But remember there are two sides to that coin and legitimate polls do exist that assume a “light most favoring Romney” and the results are equally staggering.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: