PPP/DailyKos weekly poll: Obama+5 nationally (likely voters)

[Craigo]

Last week Obama led 50-44% (he had led by that margin since the week of the Republican convention). So this is the first movement in Romney's favor (if you can call it that, w/in MoE) on the Daily Kos tracking poll since the end of August. Not to mention that the party ID of the likely voters is D+7 (the same as 2008 turnout). So if the November electorate looks as Democratic as the 2008 electorate, then Romney trails by about 5 points according to this poll. McCain lost in the 2008 electorate by 7. Romney probably needs to trail in an Obama electorate by 3 or less in order to actually win in November (presuming a less Obama friendly electorate in November).

See rule #7. Or visit unskewedpolls.com, they'll tell you what you obviously want to hear.

Also, a one point shift in a poll with a margin of error of 2.8% is not "movement."

[Craigo]

You asked "if you can call it that." The answer is: No, you can not. Trying to cover one's ass with "It is irresponsible not to speculate" does not make the speculation any less foolish.

You have repeatedly demonstrated that you do not understand polling, particularly party ID.

1) It is not a demographic characteristic
2) It is a dependent variable of voter intent
3) The relationship is weak

There aren't "more" or "less" Democratic of Republican electorates, since party identification is an attitude, not a characteristic. Party ID breakdowns reflect how the sample intends to vote - most respondents decide what party they identify as by deciding what candidates they support, not the other way around. And even those two variables don't correlate very well.

There's a reason why professionals don't worry about party ID.

Bonus question: Which election had a 38D-35R party split?

A. 1984
B. 1988
C. 1992