Numbers don’t lie, but sometimes they talk out of both sides of their mouth. The latest CBS News/New York Times poll has presumptive Republican nominee Mitt Romney leading Obama by three points. The hair-breadth gap, which falls within the margin for error, doesn’t seem inconsistent with a tight presidential race—yet the poll presents one finding that, if there’s justice in the world, absolutely cannot be right. Forty-six percent of women voters support Romney, while only forty-four endorse the incumbent. If these numbers hold true, then the pro-life, equal pay-ambivalent, anti-contraception GOP can actually stake a claim to the title “party of women,” as Michelle Bachmann and 13 other Congresswomen argued in Politico on Wednesday. Of course, yesterday Fox News unveiled a poll of its own. These results painted a different picture, with Obama racing ahead of the Massachusetts governor by seven points, 46 to 39. (There was no mention of where women were more likely to cast their ballots, however). Like the CBS/New York Times survey, Fox’s census relied on phone interviews with randomized adults across the country. It drew from a pool of 913 people, to CBS/NYT’s 615. It also, somewhat surprisingly, contained a flicker of optimism about Obama’s economic performance, revealing that approval ratings for the POTUS’s financial policies have jumped nine points since August 2011. Meanwhile, though the CBS poll didn’t ask participants explicitly about Obama and the economy, it established that most Americans think the marketplace is in bad shape. That’s good news for the Republicans. What’s so strange about these twinned polls, in fact, is that the one sponsored by two liberal news sources is chock full of good news for the Republicans. (It even registers widespread skepticism about Obama’s sincerity in embracing gay marriage). Meanwhile, the one conducted by conservative Fox News bears glad tidings for the Democrats. Shouldn’t we expect the opposite from our spin-drenched media? As my colleague William Saletan observes, perhaps the takeaway here is that most polling error stems from chance, not bias.