Likely voters voice majority support for Proposition 19 by a margin of 56 percent to 41 percent when presented with an automated questionnaire, according to an internal poll conducted by EMC Research on October 13-14. The same poll, which sampled views of 1,403 respondents, reported less support when voters were asked to state their opinion to live interviewers.
The contrast in voters’ responses on automated surveys versus live calls likely explains the discrepancies in recent poll results. For example, the most recent Survey USA automated poll, release on October 18, shows Prop. 19 leading 48 percent to 44 percent — a margin that is consistent with previous automated polls. The result conflicts the results of a PPIC live poll released one day previously that shows Prop. 19 trailing in voters’ opinion.
New York Times analyst Nate Silver previously wrote on this trend, dubbing it the “Reverse Bradley Effect,” which postulates that voters overall — and certain blocks of voters in particular — are uncomfortable telling strangers how they would vote on controversial policies.
Our allies at Firedog Lake have further analysis here:
I have previously speculated that Prop 19 might be do better in polls conducted without live interviewers. There is still a stigma in many communities attached to marijuana use which could make some voters embarrassed to tell a stranger over the phone they plan to vote for legalization.
PPP and SurveyUSA ,which use automatic interviews, have consistently shown greater support for the initiative. We have seen recently that SurveyUSA, using mostly automated interviews, found the measure winning 48-44 while PPIC, using live interviews, had it losing 44-49.
This internal polling from the campaign confirms not only that interviewees seem to be lying to live pollsters, but also that this effect is quite pronounced among certain groups — particularly young voters. In live interviews, voters under 30 support the measure only 49-37. But in the automatic interviews, young voters support Prop 19 by an enormous 73-22 margin.
In general, ballot measures tend to be very difficult to poll. The social and legal issues associated with marijuana use makes things even more complicated. The ability to do a straight-up comparison of the results of automated versus live interview polling helps explain some of the wild discrepancies we’ve been seeing in Prop 19 polling of late. The results provide very positive news for supporters of the measure, and if they are correct, Prop 19 will likely become law.
Yet the results also a reminder that we should treat all polling on this measure with a healthy dose of skepticism, given how hard it appears to be to get accurate information on how people truly intend to vote come election day.