The Los Angeles Times poll on the California recall says that the removal of Gray Davis is favored by 50% (boot him) to 47% (keep him). Since the margin of error on this poll is plus or minus three points, it's a statistical tie. On the second half of the poll, they have Bustamante at 30%, Schwarzenegger at 25% and McClintock at 18%. Here's the whole thing.
Meanwhile, the Stanford University/Knowledge Networks Survey has the recall winning 62% to 38%, Schwarzenegger at 40%, Bustamante at 28% and McClintock at 8%. This poll (here are the details) has a margin of error of plus or minus 4.3 points.
Meanwhile, another poll (the one local station KABC is using) has Davis being dumped by 62% and kept by 37%, Schwarzenegger at 39%, Bustamante at 29% and McClintock at 16%. This one is plus or minus 3.7% and here's the full thing.
And the Field Poll has the recall passing 55% to 40%, Bustamante at 30%, Schwarzengger at 25% and McClintock at 13%. Margin of error: 4.5 points. Here's that one.
So which one is right? No one knows, but if you cruise the political discussion boards, the answer's simple: The one that says that what you want to have happen is likely to happen.
Only two things interest me about these totals. One is that their wide variance ought to remind us how approximate polls are. If two people do the same math problem and come up with two different answers, the logical reaction is that at least one of them is wrong. But we often believe polls in spite of that.
Secondly: The Field Poll says that if McClintock drops out, almost all his votes will go to Arnold. If McClintock is really 32 points behind him (as per the second of the above surveys), then he might as well. If he's seven points behind him, he probably won't. So one of these polls, suspect as its accuracy may be, could influence the election. That's above and beyond the fact that polls have a certain self-fulfilling tendency, energizing this or that group to turn out, donate or stay home.