News organizations love - and I mean love - polls. You will see hundreds more from now until election day. But what you will also find is very little real analysis. My suggestion is that if you are interested in polls, do your own analysis. For the most part, polls can show just about prove any point anyone wants to make.
For instance, Public Policy Polling released a poll this week showing Obama up 5 in Ohio. a 50-45 lead isn't necessarily huge, it is with the Margin of Error, and is roughly what Obama won the state by in 2008. But is it true? I ran the numbers a different way and find a 48-48 tie, but more on that at the end of the post. It depends on how you look and analyze the data. Here's a few data points from the PPP poll that might make some people go "Hmmmm."
* Democrats made up 41% of the poll, while they only made up 36% actual turnout in 2010 and 39% in 2008. Since 2008, swing states have bled Democratic registrations while there has been a major uptick in independent and minor uptick in Republican registrations.
* Independents made up just 22% of the survey. This is important because that is roughly 1/3 below what the actual turnout was in 2008 and 2010. It's significant because the poll shows that Romney is actually leading by 2 points with Independents in Ohio, while Obama won them by 8 four years ago. (So it's difficult to see how Obama goes from winning Independents to losing them, yet is doing better in this poll than he did in 2008.) Clearly, the uptick in Democrats polled came at the expense of Independents, where Romney fares better.
* The gap between women and men surveyed is 8 points. Women tend to favor Obama while men favor Romney. The actual turnout in the last 2 election favored women, but only by 4 points. Again, this will artificially drop a point or two Obama's way.
For fun, I used the data gathered by PPP and ignored their findings that use 41% Democrats and just 22% Independents. I substitute the most recent election figures that actually had the turnout as 37% Republican, 36% Democrat, and 28% Independent. Using PPP's data, but actual turnout from the most recent election, finds that the race in Ohio is actually tied, 48%-48%. Again, on the heels of the Democratic convention, that would not be a comforting number for President Obama.
Ohio will probably be the ultimate swing state in 2012. President Obama has already blown past his 2008 spending in the state, and we are two months away from the election. This suggests that the campaign has some internal polling of their own...
Comments
I saw the Rasmussen polls recently have defied the conventional wisdom of having undecided voters or “leaners” breaking for the challenger. Today, Romney was ahead 47-46, but with leaners, he trailed 49-48. This means that he gained 1 point to Obama’s 3! I saw the same phenomenon the past few days.
What can explain this unconventional outcome. I particularly fail to understand it with Q3 and the Middle East’s recent events.