My new post at 538 addresses this question. See also this piece by Robert Erikson and Christopher Wlezien, and this post by Thomas Holbrook. My posts summarizes some of their research.
Here is the bottom line, from me:
bq. I would expect only small bumps for either party. Neither candidate is really polling above or below expectations at the moment. The fundamentals going into the race suggest an Obama victory, but by a relatively small margin — and that’s exactly where the polls are right now. Moreover, scheduling the convention late in the summer and back-to-back should mitigate their impact. And there is only a small number of self-described undecided voters, which may help explain why another high-profile event, the naming of Representative Paul D. Ryan as the running mate, has not really moved the national polls.
bq. To be sure, even “small bumps” might be enough to put the race on its head, moving Mr. Romney from a slight underdog to a slight front-runner. If that proves true, the question is whether Mr. Romney will sustain that lead through Election Day.
Holbrook estimates a 3.6-point bump for Romney and a 1.1-point bump for Obama, which would move Romney to the front.
One postscript regarding soon-to-be Hurricane Isaac. If Isaac hits New Orleans, there is some speculation that this will negatively affect the Republican National Convention for this reason:
bq. It is the last thing the Republicans need for the public to be reminded of the Bush administration’s inaction during the Katrina crisis as they are nominating Mitt Romney.
(Update: See also this NY Times piece.)
That’s not the real issue, I think. The bigger problem is that a convention bump derives in part from a bunch of media coverage of the nominee, according to Holbrook’s research. Anything that displaces the convention from the news is a problem for Romney.