Yesterday’s Quinnipiac University Poll of Florida voters was criticized by Democratic pollster David Beattie for the partisan composition of its sample. He writes “In short the results raise some concerns because more Republicans than Democrats are interviewed, which is not going to happen on Election Day in Florida. They model that 32% of the turnout will be Republican, 29% Democrat and 32% Independent in the 2012 election, this is NOTHING like what the reality will be. There will be a net 2 percentage point or more Democratic registration advantage on election day.”
One problem with Mr. Beattie’s analysis is that he assumes that we know prior to the election what the exact partisan composition of the electorate will be on Election Day. One only has to look at past Florida exit polls to see that the party distribution of the electorate can change from election to election. In 2008, there were more Democratic voters than Republicans by a 37-34 percent margin. However, in 2004, Republican voters outnumbered Democrats by 4 points 41-37 percent.
We are currently using the same methodology that we used when we accurately predicted the Florida presidential results in 2004 and 2008. Respected New York Times polling analyst Nate Silver found that we were the most accurate poll in predicting the 2010 elections.
It should also be noted that Mr. Beattie is comparing apples to oranges. He cites our party identification numbers and then compares them to party registration figures. Major independent pollsters and political scientists have long recognized that party identification is a better measure of political attitudes than party registration. Some people register with one party and even though their political attitudes have changed, they don’t bother to change their party registration. Party identification measures whether people generally consider themselves a Democrat, a Republican, or an Independent, regardless of party registration. It is important to understand that all of the major polls – Gallup, Pew, Quinnipiac, CBS, ABC, and NBC- do not weight their data by party. Those polls, as does Quinnipiac, weight their data for things such as gender and age to match the Census data because these things don’t change.
There is always going to be some small fluctuation in party identification in polls. It might be due to sampling error or short term events. These small movements are generally not worth mentioning. If however, there is a big change that holds up over time that would be a significant finding. Such a phenomena is very rare in public opinion. It is called a “party realignment” and happens when there is a major event, like the Depression in the 1930’s, which causes a substantial number of people to permanently change their party identification.
In our recent poll, we found that in party identification Republicans outnumbered Democrats 32-29 percent, which is not a statistically significant difference. In our three prior polls the Republican percentage was between 29-31 percent and the Democratic percentage was between 32 and 33 percent. This sample is a little more Republican than those earlier polls. But the difference is so small that it is not meaningful. Even with the slightly more Republican sample, the results of the matchup between Obama and Romney were the same for the last 3 polls. In each survey, Romney was ahead of Obama by a statistically insignificant 3 point margin.
All of these Quinnipiac polls have shown that the party distribution in Florida is about evenly split. Given our poll numbers it is certainly possible that there will be a small Democratic advantage on Election Day or there could be a small Republican advantage, or it could be a tie.
One early indicator of a potential problem for President Obama is the enthusiasm gap in Florida. As we’ve seen in other states, Democrats are less enthusiastic about voting than Republicans. So even though the party breakdown is roughly evenly split, Republicans could outnumber Democrats on Election Day if they are more excited about voting in November.
It should also be pointed out that we are about 10 months from the election. This was a poll of registered voters. It is too early to be talking about likely voters. We don’t even know who the Republican candidate for President will be, although it is obviously looking good for Romney. No pollster can tell you what is going to happen in the election in November in January, not even Mr. Beattie. A poll is a snapshot in time. We will have a better idea of what will happen in the election come September and October, with the most accurate polling right before the election.
Finally, I would suggest that whenever a partisan pollster criticizes the methodology of an independent pollster, it be taken with a big grain of salt. The goals of a campaign pollster are very different from an independent pollster. The job of a campaign pollster is to help their candidate get elected. As an independent poll, Quinnipiac’s goal is to provide accurate polling data to the public. Unlike campaign pollsters, Quinnipiac is completely transparent about its methods and poll results. I would suggest that partisan pollsters also be completely transparent and disclose to the news media their own methodology and poll results. If campaign pollsters are going to go on the record criticizing other polls, shouldn’t they be expected to go on the record explaining how they do their own polls? We welcome any members of the news media to visit our Polling Institute and see how we do our surveys. We wonder whether Mr. Beattie or other critics, partisan or independent, are willing to make the same offer.