Today is likely to be Poll Day, as we try to analyze the potential outcomes the day before ten states hold their nominating contests in the Republican presidential race.  Let’s start on the lighter side with today’s NBC/WSJ poll, which is a great tool — for having a few laughs.  Neil King reports that Mitt Romney has taken a six-point national lead among Republicans, and that Barack Obama has again hit 50% in job-approval ratings:

Mitt Romney has regained the lead in the Republican presidential contest thanks to new support from conservatives, while evidence emerges that the bitter nomination fight has damaged the GOP candidates’ standing among the wider public, a new Wall Street Journal/NBC News poll finds.

The resurgence of Mr. Romney, who hadn’t led the Journal poll since November, lays the path for a potential matchup against a president whose own position is strengthening. President Barack Obama’s approval rate hit 50% in the poll, its highest since last May, as more voters expressed confidence in the economy.

Why is this laughable?  The poll was taken among 800 adults, not registered voters or likely voters, the usual and more predictive models for political surveys.  In the case of Republicans, the sample data states that they needed an “oversample” of 185 voters (additional interviews above the 800 conducted in the main polling) in order to get to 400 interviews with GOP primary voters.  That would put the original sample at about 27% Republican, which is a fairly significant undersample.

But we don’t need forensic math to determine that.  Question Qf1b/c asks respondents whether they voted for Obama or McCain in the 2008 election.  The split from that question is 43/31 Obama, with 18% who didn’t vote at all (Obama won the national popular vote by seven points in 2008).  The next question on midterm election voting has 32% not voting at all in 2010.

Clearly, this is skewed toward Democrats and useless as a predictive model for voting in 2012.  One has to wonder why they bothered to publish the results.

Update: I missed Q13; Jazz Shaw pointed it out to me:

Strong Democrat ………………………   18
Not very strong Democrat ………….   9
Independent/lean Democrat ……….   16
Strictly Independent …………………..   18
Independent/lean Republican ……..   15
Not very strong Republican ………..   7
Strong Republican …………………….   13
Other (VOL) ……………………………..   3
Not sure …………………………………   1

So the D/R/I in this poll of adults is 27/20/49, which is even more risible.  If you include leaners, it’s 43/35/18.  Either way, it’s nonsense.  My apologies for missing it in the first place, though — usually this question gets asked with the rest of the demographic questions at the end of the survey.