The Delaware conundrum, Part Two

In Delaware, conservative Christine O’Donnell is challenging centrist Rep. Mike Castle for the Republican Senate nomination. As I argued here, Castle (with his 52.5 percent lifetime ACU rating) would, if elected, represent half of a conservative vote. O’Donnell would represent a full conservative vote. Thus, if one treats this as a math problem, voting for O’Donnell over Castle makes sense if she has at least half as good a chance of defeating Democrat Chris Coons as Castle does.
When I wrote my original post, O’Donnell led Coon 41-39 in the most current Rasmussen poll. Thus, it was plausible to think that ODonnell’s prospects were at least half as good as Castle’s (he led Coons 47-36 at the time).
But the latest Rasmussen poll tells a different story. O’Donnell now trails Coons by 10 points, 36 to 46. By contrast, Castle leads Coons by a margin of 49-37.
The numbers could swing back in O’Donnell’s favor, of course, but the latest poll contains other information that suggests O’Donnell would have little chance of winning in November. Rasmussen found that 55 percent of Delaware voters approve of President Obama’s performance, with 40 percent of them approving strongly.
I fear that a state liberal enough to approve to this extent of Obama isn’t at all likely to elect a Tea Party style conservative like O’Donnell. It has, however, been electing Castle to its at-large congressional seat every two years since 1992 (usually by margins of 20 points or more) and has twice elected him governor.