The good news: With four days to go, she’s only down 10. The bad news: With four days to go, she’s still down 10.

In the past two weeks, Republican Christine O’Donnell has narrowed Democrat Chris Coons’ lead in Delaware’s U.S. Senate race from 19 points to 10 points. The latest Monmouth University Poll finds Coons has the support of 51% of likely voters to 41% for O’Donnell. Two weeks ago, this race stood at 57% to 38%…

O’Donnell has also made gains among independent voters, now leading Coons 47% to 42% among this voting bloc. Two weeks ago, she trailed in the independent vote by 51% to 41%…

The poll found that just 35% of likely voters in Delaware feel that Christine O’Donnell is qualified to be a U.S. Senator, while 56% say she is unqualified. That contrasts with their opinion of Chris Coons, who 65% say is qualified for the U.S. Senate to 25% unqualified. These qualification results for O’Donnell and Coons are basically identical to the Monmouth University Poll results from two weeks ago.

How to square this poll with yesterday’s Fairleigh Dickinson survey showing Coons cruising to victory with a 21-point lead? I … honestly don’t know. Both polls use nice big samples of likely voters, so there’s no threshold reason to think either is sketchy. I’ll give you three possibilities for the discrepancy. One: The timing. The FDU poll was taken on Oct. 20 to 26 while the Monmouth poll was taken on Oct. 25 to 27. It could be that the race is tightening at the end and only Monmouth is picking it up. Two: They could be defining “likely voters” differently, the dangers of which I noted in yesterday’s post about that shady Murkowski/Miller poll in Alaska. Monmouth defines likelies as “voters who cast ballots in at least two of the last four general elections and … who say they are either ‘certain’ or ‘likely’ to vote in this November’s election,” which seems like a reasonable screen. FDU doesn’t say how they defined them, which makes me that much more skeptical of their result. Three: Their partisan samples are different. Monmouth’s sample splits 43D/37R/20I by registration and 36D/30R/34I by self-identification; the difference between those numbers tells you a lot about how disgusted people are with both parties, even in Delaware. FDU’s sample splits 40D/24R/37I or, if you force indies to declare which way they’re leaning, 53D/36R/12I. As a point of reference, in the Democratic wave year of 2008 (with Obama at the top of the ticket, natch), the people who voted in the Senate race identified as 48D/31R/21I. I won’t even pretend to guess what this year’s split will be — I doubt even Nate Silver and his supercomputer would feel confident estimating — but obviously Republican turnout will be proportionally greater. Which means Monmouth’s probably closer to the mark than FDU is, especially if they’re right about indies now breaking for O’Donnell.

All in all, yeah, I think it’s probably more likely a 10-point game right now than a 21-point blowout. Which only stands to reason: As our friend Mr. Frank up in Massachusetts could tell you, not even heavily Democratic jurisdictions are completely out of play for the GOP this year. And let’s face it, Harry Reid’s “pet” isn’t exactly setting the world on fire as a candidate. The big worry is that number in the blockquote about which candidate is more qualified for the Senate. There’s nothing O’Donnell can do at this point to turn that around, yet that’s something that’s bound to influence undecideds and leaners in the booth. She needs every last vote she can scrounge to pull the upset and that’s an awfully high hurdle. As for the possibility of a sympathy vote after yesterday’s Gawker piece, crazier things have happened — and since women voters have been O’Donnell’s big problem all along, it could be especially helpful with them — but from what I can tell, the big Delaware paper has stayed away from the story, so who knows how many voters in-state even know what’s going on. For what it’s worth, Nate Silver’s model has the odds of a Coons win at, er, 100 percent, but we’ll see what happens this weekend when the final election polls start dropping from Rasmussen and crew. Cross those fingers!