The Monkey Cage

With John McCain apparently having sewn up the Republican presidential nomination, attention is turning to his choice of a running mate. In previous posts, I’ve generated some way-too-early-to-take-literally projections of the probability of being selected, based on a statistical model that Paul Wahlbeck and I developed a decade ago to analyze vice presidential nominee selection, here, gated. A key to the procedure is the identification of a “pool” of “finalists” who are considered. As of now, we can’t know who McCain’s finalists may be, but speculation is rife. For present purposes, I’ll put those named in Chris Cilliza’s February 22 piece in the washingtonpost.com Politics Blog as the finalists. They are:

Charlie Crist, governor of Florida.
John Huntsman, Jr., governor or Utah.
Tim Pawlenty, governor of Minnesota.
Mark Sanford, governor of South Carolina.
John Thune, junior senator from South Dakota.

(Not exactly household names nationwide.)

This set of five is so homogeneous in terms of their age relative to McCain’s (everybody is at least ten years younger than he is) and campaign experience that the favor in that model that’s capable of distinguishing one from the other in terms of the chance that he’ll wind up on the ticket is the number of Electoral College votes in the potential running-mate’s state. On that scenario, here’s what the model generates as each candidate’s prospects:

55% Crist
10% Huntsman
15% Pawlenty
12% Sanford
08% Thune

Pawlenty, in second place but far behind Crist in my projections, seems to be the media pundits’ top choice for the second spot as of now. Let’s see how it plays out in the coming months. Who knows? Crist may not even end up on McCain’s shortlist, which will obviously be more authoritative than Chris Cillizza’s February guesses. And even if he is one of the finalists, the model, perish the thought, may simply be wrong in this instance. So don’t bet the mortage (if the bank hasn’t already foreclosed on it) on this result, but keep it in mind as an early baseline.