The good news? This is right in line with yesterday’s Fox News poll of likely voters. There Romney led by one, here The One leads by two. Doesn’t matter; both are within the margin of error. Bottom line is that we have a toss-up on the eve of the election, with Romney just three days away from finally being able to tap his huge treasury of general-election funds. Even better, according to CNN’s data, Obama’s attacks ads haven’t done much to hurt him. Mitt’s net favorable rating is now +4 among likely voters while Obama stands at just +5, the second time today that a major poll has shown the “likability gap” between the two shrinking. My read on that is the same as Ace’s — as more people tilt towards Romney on the merits, it’s only natural that they’ll start to view Obama less favorably and the likability gap will close. The reverse is also possible, that O’s attack ads are backfiring on him and his decline in likability is driving Romney’s improvement in the polls. But if that’s the case, how to explain the non-decline among registered voters?

Among registereds, O’s numbers are almost identical now to what they were five months ago, before all the negative campaigning began. The real explanation for the closing of the likability gap among likely voters, it seems, is simply that that sample is more Republican than the sample of registered voters. His job approval lately among registereds looks pretty good too compared to where it used to be not long ago:

Put it all together and you find that Obama’s actually holding up pretty well with registered voters, enough so to keep him on top head to head against Mitt notwithstanding a strong Republican sample among likely voters:

As for Akin-gate, this trendline on abortion is intriguing but I’m not sure what it proves given that the most recent poll before this was taken almost a year ago. Did these numbers surge just in the past week or did they surge over the last several months, possibly in response to the Democrats’ “war on women” messaging?

Usually there’s not much of a gender gap in abortion polls, but there’s one here:

Fully 83 percent say abortion should be legal in cases of rape or incest, which explains why Debbie Wasserman-Schultz was being even more robotic than usual last night on CNN in trying to impute Todd Akin’s position to Mitt.

Exit question for poll watchers: At the top of page five of the crosstabs, preceding certain questions about voter enthusiasm, CNN says it sampled 404 registered Democrats and … 473 registered Republicans. They never specify the partisan breakdown in the total sample, but they do say on the cover sheet that the total sample is based on interviews with 1,055 adults and 924 registered voters. Which means, unless I’m misunderstanding something, that Republicans were a majority of their sample of registered voters. Er, isn’t that highly unusual in a national poll?