Mickey Kaus writes:

It’s worth noting that, in the event, not only did successor Arnold Schwarzenegger get more votes (3,744,132) than Davis (3,562,487), he also got more votes than Davis got in November, 2002 (3,469,025) when Davis won reelection.

But comparing these yes-Schwarzenegger votes to the no-recall votes ignores those who voted not to recall Davis, but also voted for Schwarzenegger as their choice for governer if Davis were recalled (since they couldn’t vote for Davis). That is, the straight-up Schwarzenegger versus Davis comparison counts everyone whose first choice was Davis and whose second choice was Schwarzenegger as a yes-Schwarzenegger vote (as well as a no-recall vote), artificially inflating the Schwarzenegger count. And it counts everyone who preferred a third candidate (say, Bustamante) to Davis and Davis to Schwarzenegger as a vote for neither Davis nor Schwarzenegger, artificially deflating the Davis-versus-Schwarzenegger count.

The appropriate comparison is between voters who voted not to recall Davis, and voters who voted to recall Davis and for Schwarzenegger as the replacement candidate. Presumably this is knowable (if the ballots haven’t been destroyed), but it’s not on the Secretary of State’s summary page, at least.

How could you estimate this? One (crude) estimate is that everyone who voted for Bustamante would have voted for Davis if Bustamante hadn’t been on the ballot. This over-counts (it counts voters who preferred Bustamante to Schwarzenegger and Schwarzenegger to Davis as voting for Davis), and it under-counts too (it fails to count voters who preferred some fourth candidate to Davis and Davis to Schwarzenegger as voting for Davis over Schwarzenegger). It’s not obviously worse than the yes-recall to yes-Schwarzenegger comparison, though. It gives 3,744,132 votes to Schwarzenegger and 6,172,557 votes to Davis.

Even if Davis would have only split the Democratic vote with Bustamante (that is, if Davis and Bustamente were equally popular, and if none of the Schwarzenegger votes were from crossover Democrats), Davis would have got 4,920,316 votes to Schwarzenegger’s 3,744,132 votes.

And as for the rest of the comparison:

Almost a million more people (4,416, 280) voted to recall Davis than voted to reelect him last year.

8,203,005 people voted in 2003. 7,318,618 people voted in 2002. (These figures are from dividing the subtotals by the percentages; the web sites don’t list the totals directly.) If 8,203,005 people had voted in 2002 with the same distribution, 4,314,780 would have voted against Davis, so he actually picked up about 100,000 votes.

Just something to think about.