Analyzing Virginia Polls

http://usconservatives.about.com/od/campaignselections/a/Analyzing-Virginia-Polls.htm

Four recent presidential polls for the state of Virginia show that Obama has leads of 3, 4, 8, and 5 points. But the data tell a different story. If you are interested in polls, this post offers a chance to look at polls by actually analyzing data, rather than simply looking at the final "result" that the media release. The method used here to calculate is open and honest, and offers a data-driven perspective of what story the polls might actually tell. Using the same data - but actually analyzing it - shows possible Romney leads of 4, 8, 2, and 2 points.
Background
When President Obama won the presidency in 2008, he had a record-breaking year in voter turnout and enthusiasm. In Virginia, Obama outspent McCain by more than 300% having dumped $26M into the state to a paltry $8M for McCain/Palin. With every possible factor in his favor, Democratic turnout bested Republican turnout by 6 points, with 39% of voters being Democrat and 33% identifying as Republican. A repeat of this turnout is highly unlikely given the major switch in voter enthusiasm and the competitive financial situation for the Romney campaign. Yet nearly every poll sample below shows that the turnout situation for Democrats has improved over 2008, while Republican turnout has dropped from the upper 30s to the mid 20s.

While the 2008 turnout was 39D to 33R, it rebounded in post-Obamacare Virginia in 2009 where it was 37R and 33D, a near reversal from a year earlier. In 2010, the Republican trend continued. While there is no exit data for 2010, Republicans dominated the congressional scene in that year. Democrats held a 6-5 advantage in the congressional delegation after 2008, but that turned into an 8-3 advantage for Republicans in 2010, thus suggesting similar turnout for Republicans that was seen in 2009. The turnout for 2009 and 2010 is similar to the turnout in Virginia in 2004, where Republicans also had a 4 point advantage (39R, 35D), and in 2006, where Republican had a turnout advantage of 39% to 36% over Democrats. There is no reliable exit poll data for 2000, but those election results suggest a similar pattern to 2004, 2006, 2009, and 2010. So, the question is, are the polls right and Democrats will outnumber Republicans by an even greater margin than in 2008, or is the analysis below right suggesting that turnout will be closer to every other election before and after Obama's 2008 win?
We Ask America
The We Ask America poll shows the race at 49% for Obama to 46% for Romney. Not a crazy number by any means, but what does the data say? For starters, Romney is leading with Independents by 6 points, 47-41%, which reflects an 8-point swing from 2008. Romney is currently capturing 87% of Republican voters while Obama captures a similar 88% of Democratic voters. If 2012 turnout is similar to 2008 turnout, then yes, Obama is up by 3. But if it is closer to either 2009, 2010, 2000, 2004, or 2006 turnout then Romney has 49% and Obama is back four points at 45%. Given everything that is known about enthusiasm, registrations, and money, which is the more likely scenario?
Quinnipiac/CBS/NY Times
The Quinnipiac/CBS/NY Times (disclosure: About is owned by the NY Times) poll shows Obama up 50-46%. This poll shows both Obama and Romney attracting 95% of their own constituencies, but Romney wins Independents by an even greater margin than in the We Ask America Poll, where he holds a 53-42% advantage. So how does Obama get a 4 point lead? The sample is made up of 35% Democrats and just 24% of Republicans. Remember, the average turnout for Democrats over the last two elections cycles is 36% and for Republicans it is 35%. Average turnout for Republicans over the last 6 election cycles is closer to 38%, while average turnout for Democrats is 35%.
A key data point found in Quinnipiac’s poll is that only 38% of Democrats are more enthusiastic about voting this year while 53% of Republicans say the same thing. With a sample that uses just 24% of Republican respondents, it’s actually remarkable that Romney is only “losing” by 4 points. With an 11 point lead among Independents, Romney would only need to have the average turnout of the last two election cycles to win the state. That would assume a turnout of 36% for Democrats and 35% for Republicans, which would still be depressed for Republicans and better-than-normal for Democrats, who are less excited than normal to vote.

But yes, if we assume that Democratic turnout in Virginia stays the same (despite losses in registration, decreased voter enthusiasm, and no money advantage) and assume that Republican turnout plummets by 13 points since 2009 (despite increases in voter registrations, increases in enthusiasm, and a healthy financial situation) then Obama is “winning” by 4 points. If you use the most recent turnout data (and for every year where it is available) instead, then Romney has 52.4% of the vote and Obama has 45%, an almost 7.5 point lead for Romney.
Washington Post
A Washington Post poll declares that Obama has a “clear” lead over Romney in Virginia, where he leads by an impressive 52-44%. The poll shows that Obama and Romney each hold their own constituencies equally well, and essentially split the Independents (Obama 47%, Romney 45%). But, once again, Democrats make up a reasonable 35% of the respondents, which is probably about what the electorate will wind up being. But Republicans make up just 26% of respondents, an astonishing 11 point drop from 2010 and even a seven point drop from 2008, when Republicans turnout was extremely depressed. So sure, if Republicans completely stay home despite all evidence to the contrary, then Obama is “clearly" winning by 8 points. Again, run the numbers using normal turnout data and it’s Romney 49.2%, Obama 46.9%.
Public Policy Polling
The final firm we look at is a poll from Public Policy Polling. They show that Obama leads in Virginia by 5 points, 51-46%. Their turnout model is a little more reasonable, showing Democrats with 35% (which, again, is about the normal turnout over the last decade) of the total sample. Republicans are still undervalued at 32% (which, also once again, is below every known data point over the last decade, including historically bad year 2008). The details of the data, however, are pretty similar to the rest of the polls. Romney leads with Independents by 2 (47-45%) and attracts 92% of his core supporters. Obama attracts 95% of Democrats, meanwhile. Once again, run the numbers through the most recent election where data is available, and Romney has 49.4% of the vote while Obama has 47.4%.
Analysis
In all of the above samples, we weighted the data to match 2009, 2010, 2004, and 2006 turnout, where Republicans had an advantage turnout of about 4 points over Democrats. There is no guarantee that turnout will match what it was in those years. But does anyone think that turnout will be better for Democrats (and worse than Republicans) than 2008, a historically major outlier in US presidential elections? That is essentially what every poll shows.

There is no evidence to support any justification that there will only be 24% Republican turnout in Virginia. It’s almost a statistical impossibility. But there is a good shot that Republican turnout will beat Democratic turnout, or at worst tie it. Factor in Romney’s coming money advantage, Obama’s decreasing popularity, the unpopularity of Obamacare, the recent Republican sweeps in Virginia, an increase in GOP registrations, a decrease in Democratic registrations, a major enthusiasm gap, and the fact that Virginia is a Republican-leaning state and there is plenty of evidence to suggest that the data shows Romney is winning in Virginia.

There are many ways to look at data. The pollsters above did not weight the data in any form, and ran with unrealistic assumptions of voter turnout. There is no analysis, other than to look at what the final number spits out and declare "Obama is winning!" We weighted the data to the most recent election cycle, which also equals the data from every election outside of 2008. Has the mood changed since 2009 and 2010? The recent elections in Wisconsin would suggest not, as well as Obama's mid-40s approval ratings, and public support to see Obamacare repealed. In the end, the data within the polls is a lot more telling than what makes the headlines.