Via Suitably Flip and RCP, the latest poll from James Carville’s Democracy Corps should have the sirens wailing over at Team Obama.  The topline numbers show Barack Obama with a two-point lead over Mitt Romney, 49/47, down from a 4-point lead in July.  The new sample is D+6 at 38/32/28, only a slight change from July’s D+5 at 38/33/25, a margin-of-error change from one poll to the next (which is true of the topline change, from 50/46, as well).  However, there are a couple of big red flags in the data that hint at a much different race than what the toplines represent.

First, take a look at the independent vote for the race, which in this format is the fourth column:

That’s a 15-point lead for Romney among independents, 53.4/38.0.  Fifteen points.  Just to remind everyone, in the D+7 turnout of 2008 in a Democratic wave election, Obama won independents by eight points, 52/44, on his way to a seven-point victory overall.  That’s a 23-point shift in this key demographic, way outside the margin of error, and it’s hard to square that with a two-point Obama lead now.

Nor is that the only problem in the demos.  Take a look at the gender gap, which Obama won by a combined 14 points in 2008.  Romney wins men by 3.5 points now (Obama edged John McCain by one in 2008), and Obama only leads by 6.7 points among women, 51.7/45.0.  That’s a combined gender gap of 3.2 points, a drop of ten points in this measure from 2008.

If Obama loses independents outside of the margin of error and only gets a gender gap within the MoE, he’s toast.

Update: The split with women was 51.7/45.0, not 41.0.  I’ve fixed it above.

Update II: I screwed up the first edit by saying that the gap among women was 9 points; I had that right originally, but rewrote the passage and introduced a new error.  This video clip seems to apply best: