At least that's if you believe Matthew Dowd's assertion that "historical analysis suggests John Kerry should have a lead of more than 15 points coming out of his convention". That assertion of Dowd's has been mentioned again and again in recent news stories on the campaign, particularly since Edwards was selected by Kerry and appeared to give the ticket a bounce.

Every time I've seen this Dowd prediction, I've scratched my head and thought: "that's completely nuts--it can't possibly be true that a 15 point lead is normal". But I've been a bit busy and never got around to doing the spadework to show how nuts that prediction is.

And now I don't have to. Republican pollster Tony Fabrizio has helpfully provided the relevant data in a piece on the National Review website, blasting Dowd for, essentially, predicting Bush is going to lose.

This is because:

1. Since 1960, no incumbent president has come back from being 15 points down at the beginning of August to win re-election. In fact, no incumbent's come back from being 10 points down to win.

2. The average bounce from an incumbent president's convention since 1960 is about 6 points. Therefore--contrary to Dowd's assertion that the Republican convention would even things up--a 15 point lead after the Democratic convention would likely leave Bush still 9 points behind after his convention.

Another interesting angle on the Dowd prediction: given that the average bounce from a challenger's convention is about 7 points, does that mean Dowd now believes Kerry is up by 8 points--since otherwise how could Kerry come out of his convention with a 15 point lead based on "historical analysis"?

I suppose it's all about the expectations game: predict some outlandish bad outcome for your candidate and then when it doesn't occur claim things are going your way. But has it come to this: to make their candidate look good, Republican spinners now have to predict that his situation will soon be hopeless?