A word of caution about polling and the presidential race going into the weekend: We don’t really know yet whether anything has changed after the conventions. If you thought Barack Obama was a slim favorite before Tampa, that’s really all you should think now; if you thought Mitt Romney was a slim favorite, then there’s no real reason – yet – for you to change your mind.

What we do know is that Obama got a nice healthy bounce out of his convention and that Romney did not. That’s given Obama a six-point lead in the latest Gallup reading, taken over the week after the convention; according to the one average of polls set to emphasize recent polls more, he’s leading by about four points.

But it’s still too early to know whether that convention bounce is going to fade away or permanently change the race. As I’ve said before, we can think of convention bounces as a product of two things: people who “should” be for the candidate tuning in to the campaign for the first time and realizing they will support him, and people who are really undecided who temporarily report that they’ll support the candidate because the last thing they’ve heard is positive. The former should stick; the latter won’t – they’ll float back and forth depending on what the last thing they’ve heard might be, at least for a while. And we just don’t know how much of the polling surge comes from each group.

Moreover, this time the Libya attack showed up just days after the convention, and there’s really no way to know how any (presumably temporary) rally effect from that is showing up in the current polling.

All of that will shake out, but it probably will take another week. Note Nate Silver’s graph he’s been running throughout the conventions: By his calculations, the incumbent party’s bounce should still be at very close to full strength a week out, but by a week from Monday it should just about be gone. That’s the point when we’ll really see whether the contest has changed.

Remember, they don’t give out any prizes for biggest (temporary) convention bounce. Elections aren’t cumulative events the way a baseball season is, where a win in April really does count just as much as a win in September. All that matters is how things stand when people start voting. And even with early voting, except for a handful of unusual cases we’re just not there yet.