The principal rationale for selecting Hillary Clinton as Barack Obama's running mate is that she would have united Democrats behind their nominee at a time when they have a substantial advantage in party identification. John Kerry received 89 percent of the Democratic vote in 2004; if Barack Obama can get within a couple of points of that, even to 86 or 87 percent, he will be very difficult to defeat.

However, Joe Biden might do nearly as good a job as Clinton of uniting the party, while perhaps paying less of a price among independents.

Rasmussen has fresh approval numbers out for Biden, as well as several other Democratic short-listers. Here, borrowed from Rasmussen's invaluable subscriber service, are their approval scores by party:

Democrats

Candidate Fav-Unfav
Clinton 77-22 (+55)
Biden 65-17 (+48)
Bayh 45-25 (+20)
Sebelius 35-19 (+16)
Kaine 35-29 (+6)


Republicans

Candidate Fav-Unfav
Kaine 29-30 (-1)
Bayh 23-43 (-20)
Sebelius 14-45 (-31)
Biden 22-63 (-39)
Clinton 21-75 (-54)



Indepedents

Candidate Fav-Unfav
Biden 42-29 (+13)
Bayh 31-21 (+10)
Kaine 24-23 (+1)
Sebelius 18-21 (-3)
Clinton 39-57 (-18)















VF = Very Favorable
SF = Somewhat Favorable
VU = Very Unfavorable
SU = Somewhat Unfavorable

... McCain Voters Obama Voters
Candidate VF SF VU SU Net Margin
Biden 4 20 5 10 +2.00
Bayh 4 19 4 16 +0.75
Kaine 8 21 8 20 +0.25
Clinton 11 14 13 14 -0.50
Sebelius 3 13 5 13 -0.50

What's noteworthy is not so much that Biden will turn a lot of McCain voters on -- Tim Kaine and Hillary Clinton would have done a better job of that -- but that he'll turn very few Obama voters off. As a result, this method projects a net swing of 2 points toward Obama, which is better than he'd do with any of the other candidates. Biden also performed quite well in these ratings among undecided (43-22 favorable) and third-party (45-36 favorable) voters, though the sample sizes are probably too small to be worth worrying about.

--Nate Silver