What were some of these pollsters smoking? Prop. 19, the California initiative that would have legalized pot, ultimately failed 46 percent (in favor) to nearly 54 percent (opposed).
But some polls had it much closer: SurveyUSA's data from last week had it at a likely 46 percent (no) to 44 percent (yes). What's more, legalization backers said the SurveyUSA numbers would turn out to be the most-accurate because they were collected via “robo calls” in which users simply had to press a button on their keypads to indicate their choices.
Why would that be more accurate?
Pot promoters said many voters would be too shy to admit to live survey takers that they support legalization. We called it the Broadus Effect in honor of No. 1 stoner Calvin Broadus, a.k.a. Snoop Dogg.
So much for that theory.
SurveyUSA's numbers, as you can see, were the worst of the lot. Well, except for Yes on 19's internal survey data, which had it winning 56 percent (for) to 41 percent (against). Yeah, keep smoking that good stuff.
But even what has, in hindsight, been cited as the most-accurate survey (along side the Field Poll), the USC College/Los Angeles Times poll, was a little off: It had 19 losing 51 percent (against) to 39 percent (for). (Remember, the near final numbers were 54 against, 46 percent for, making this one 7 points off in terms of supporters).
In general, however, the robo-polls were losers, and so was the Fox News-touted Rasmussen data that seemed to favor Republicans, according to our reading of a David Lauter piece at the Los Angeles Times.
Conclusion? Keep smoking that kind stuff. But if you want accurate poll data, the in-person-interview surveys like USC's and Field Poll's are generally more accurate.