I just got an email from the press office of one of the pro-19 groups saying that their support is running higher on robopolls than on live caller polls. Luckily, our TPM Polltracker graphs allow us to separate out the two kinds of polls and see how they differ.
If you look at just live phone polls you get oppose 4 points ahead ...
But look at this, if you look at only robopolls you get support running 2.7 points ahead.
Normally I have a clear rule of thumb for people who have arguments about how the polls aren't right: they're losing. But a spread of almost seven points between different polling methodologies is actually a pretty big deal, especially when you consider that robopolls this year have tended to lean a bit more to the right in their samples.
Clearly the movement is toward opposition in both samples. And it might just be random chance. 5 live call pollsters have surveyed the race, compared to 3 robopoll firms. And the former group includes the most respected polling outfits in the state -- the Field Poll, PPIC and LA Times. But, again, the folks alleging some sort of 'reverse Bradley effect' may be on to something.