Pew Research Center's new poll is lighting up the boards. What was once a 51-43 Obama lead among likely voters in September according to Pew, has now reversed dramatically, to a 49-45 Romney advantage.
How did this turnaround occur?
Well, let's look in Pew's documentation (linked above), particularly in a section entitled, "Patters of Voter Support, September-October."
Did Romney's debate performance pluck some Democratic support away from Obama? Nope. The president's lead is virtually unchanged within his own party, from 93-5 in September to 94-5 now.
Did Romney strengthen himself among Republicans? Hardly, these numbers shifted from 92-5 to 91-7.
Did Romney achieve a major breakthrough among Independents? Only if you consider a 2-3 point shift within each candidate major (from Obama leading 45-44 to Romney ahead 46-42).
So where is the answer? Look at these sample sizes:
In other words, we've learned something we should already have known. That if the electorate on Nov. 6 is 3 percentage points more Republican than Democratic, Obama loses. The question is will the turnout be that way?
In 2004, with George W. Bush having the advantage of incumbency and what is considered one of the better turnout operations in recent elections, the electorate ended up 37-all between D and R. I highly doubt we'll see an R+3 electorate next month, but if so, yeah, Obama will lose.
For further coverage of the sample-weighting issue, see my website: