You may have seen the new Survey USA robo-poll out this morning, which purports to show Wendy Greuel up three points in the race for L.A. mayor. 

We're getting tired of trashing Survey USA, but this poll may get more attention than most, because it shows a violent, 12-point swing in Greuel's direction since the last poll, taken two weeks ago. So let's address this.
Here's the result:

Greuel 45
Garcetti 42
Undecided 13
Compared to two weeks ago:
Garcetti 49
Greuel 40
Undecided 11
What's happened in the last two weeks that could cause such an enormous swing? Well, Greuel started airing her TV ads, and Bill Clinton came to town. Could that tip the balance 12 points in her favor?
Well, let's ask John Shallman, Greuel's top strategist, what he thinks of Survey USA. Here he is in January, responding to a Survey USA robo-poll that showed Garcetti up eight points:

This poll is bogus and contrary to every public and private survey we've seen. Everyone knows that robo call polls are inaccurate. In fact, ABC's polling director Gary Langer was quoted in the Hill Newspaper in 2011 saying, “We don't regard autodialed, pre-recorded polls as valid and reliable survey research.” The Survey USA poll is misleading and projects a 63% turnout which would nearly triple the voter turnout from previous municipal elections.

Today, shockingly, the Greuel campaign feels much better about Survey USA. In a fundraising email, campaign manager Janelle Erickson touted the “great news” in the survey, which she said shows “Wendy has the momentum.”

In a memo to the press, the campaign acknowledges they've had “concerns” about Survey USA before, but now they argue the sample is more representative of the electorate.
However, nothing appears to have changed since January about Survey USA's methodology. The turnout in the most recent poll is projected at 65% — which is about 40 points off. Something has indeed changed, however, with the projected composition of the electorate.
This was Survey USA's estimate two weeks ago:
Latino 41%
white 34%
And today:
white 50%
Latino 27%
Today's sample is much closer to reality. However, the wild fluctuation doesn't generate much confidence in the methodology. Somehow, in the course of two weeks, we're to believe that a third of likely Latino voters have decided to stay home, while white turnout is up by half. Possible, but not likely. 
That by itself would cause a big swing to Greuel, because Garcetti generally does better with Latinos than with whites. But the poll also shows a 26-point swing to Greuel among Latino voters. At the same time, white voters moved to Greuel by just 6 points.
Has anything happened in the last two weeks that would have such a disparate impact? Greuel hasn't gone out of her way to target Latino voters in the last two weeks. Her ads aren't even airing on Spanish-language TV. The Clinton visit might explain some of the difference, but a 20-point gap?
One could go on. The point here is not that Garcetti is way ahead. Who knows, Greuel may be in the lead. The point is that you just can't tell it from this.
Also worth noting: In case you thought that the Greuel campaign had the monopoly on inconsistency, the Garcetti team used to tweet out Survey USA polls that showed him ahead. Today, they're singing a different tune:

“The poll is not real and conflicts with every other public poll in 2013,” said Garcetti spokesman Jeff Millman, “except for one other Survey USA poll the day before the primary, when they were 7 (!) points wrong and Garcetti won by 4 points.”

Survey USA realizes that this is a pretty big shift since the last poll, so in the report the pollster runs through a series of three “sanity checks.” And what do you know, they declare themselves sane. Everyone else will have to reach their own conclusions.

Advertising disclosure: We may receive compensation for some of the links in our stories. Thank you for supporting LA Weekly and our advertisers.