The estimable Seneca Doane has already published a diary http://www.dailykos.com/...
about the slanting of the sample toward older white college-educated voters/ Well, one of the numbers within that sample virtually screams that this survey was over-weighted with Romney-lovers:
5. Do you have a favorable or unfavorable opinion of Mitt Romney?
(Post Debate) (Pre-Debate)
Oct. 3 Sept. 28-Oct. 2
Favorable opinion 56% 54%
Unfavorable opinion 42% 42%
No opinion 2% 3%
http://i2.cdn.turner.com/...
So before the debate, 54% of this sample had a favorable opinion of Romney, and that went up to 56% after the debate.
But what did CNN ITSELF find among ALL likely voters, 9/28-9/30?
Mitt Romney's favorables were 49%, unfavorables 50%
http://www.pollingreport.com/...
So the sample was clearly biassed toward people who favor Romney.
QED