You didn't take my hint about doing the research, but I found some time this morning before a meeting.

According to a Cornell study,

When the federal government issues a terrorist warning, presidential approval ratings jump, a Cornell University sociologist finds. Interestingly, terrorist warnings also boost support for the president on issues that are largely irrelevant to terrorism, such as his handling of the economy.

Then there's this graph, through early August, of terror alert levels versus Bush's approval ratings:

There is certainly a causal relationship there: Bush's numbers rise with the terror alert levels. That doesn't prove that the terror levels have been manipulated for political reasons, but given the viciousness and take-no-prisoners fanaticism of this last campaign, no reasonable person could rule out the possibility that Rove would take advantage of such a powerful weapon in the fight to reelect his boy. Given Rove's reputation for ruthlessness and brilliance, on both sides of the aisle, the smart money says he did.

Remember where this thread began: The level's raised with great fanfare to orange just before the election, quietly lowered to yellow a week after. And note this result from the Cornell study:

When Willer linked the warnings to presidential ratings from 2001 to 2004, he found that each terror warning prompted, on average, a 2.75 point increase in the president's approval rating the following week.

And Bush had a margin of about 3% on November 2, 3% more than all the polls for months had been predicting. He might've lost had the terror level not been raised. Gee Tom, what do you think happened?