Who in their WRONG headed mind would self identify themselves with the self identified party of NO?
You would have to prove, with a probability of say p > 95%, that the poll taken was not a random sample.
In other words, where's your objective statistical test?
Versus your perceived subjective opinion.
Individuals self reporting of political affiliation is not proof of anything other than what side of the political bed they got up from on any particular day.
Voting for a D or an R on election day does not make one forever an R or a D.
I would hazard a guess though.
Most people would self identify themselves with a WINNER than those that would self identify themselves with a LOSER.
The hardcore R's are ~20% to ~30% as has been shown in countless polls of Dubya's approval ratings over the past several YEARS.
The self identified D's in the aforementioned poll was 36% while in the general election (GE) the D's totaled ~53% for POTUS.
Hell, in my subjective opinion the D's were underrepresented by a full 17% (or 32.08% using the 53% from the GE), and based on my subjective opinion, Obama's approval rating should be substantially higher than this poll has reported.
PS - The spread between just the D's and R's for POTUS was actually 7.36% (according to the official FEC reportage), so you are off by 13.23% (using your 6.5% spread figure).