Marijuana poll raises questions about question wording
Posted April 7, 2014
Raleigh, N.C. — A majority of Tar Heel State voters support legalizing marijuana for medical purposes, according to a recent Time Warner Cable News survey. The poll found that supporters outnumber opponents by a healthy 2:1 ratio – 62 percent to 30 percent. Younger and wealthier respondents were the most supportive, while Republicans were the least.
This result raised several questions for me. One, is polling in North Carolina on this question consistent? Second, does question wording matter? Last, how does North Carolina compare with other states or nationally?
I found just two other polls conducted in North Carolina since 2013 that asked about marijuana. Their findings indicate both consistent support and the importance of question wording.
The first survey I found is an Elon University poll from February 2013. Elon reported a significantly higher level of support – 76 percent – for medical marijuana. However, Elon worded its question differently, asking, "Should North Carolina allow doctors to prescribe medical marijuana for reasons such as cancer?" Specifically citing cancer as the reason and naming a doctor as prescribing it, I'm guessing, bolstered support for medical marijuana overall.
More in line with the Time Warner Poll, a Public Policy Polling survey from January that used the same question wording found 63 percent were in support of medical marijuana.
All three surveys suggest there is significant support for pot, right? Not so fast.
There are several policy options, ranging from legalizing marijuana for medicinal purposes to more broadly legalizing for personal recreational use. The same PPP poll found that just 42 percent support legalization of marijuana in general, absent a medical justification. I could not find another recent poll of North Carolina that asked residents about their views on the recreational use of marijuana.
It also looks to me like North Carolina could be more conservative than most states on this issue, although question wording might explain this. In New York, for example, 88 percent of those surveyed support legalizing marijuana for medical reasons – just 9 percent opposed – and 57 percent support legalizing it for personal recreational use. In Florida, 82 percent support marijuana for medical use when prescribed by a doctor.
Even in Indiana, 52 percent support regulating marijuana like alcohol and tobacco. To our north, 84 percent of Virginians support medical marijuana, although support for recreational use of it lags behind opposition, 46 percent to 48 percent.
All of these polls, however, mention a doctor prescribing marijuana, while in North Carolina, it was only the Elon poll that included this information in its question. Since the Elon poll finds higher levels of support, more similar to other states, referencing a doctor prescribing it is likely adding to the levels of support.
Nevertheless, compared to national views, North Carolina is definitely more conservative than trends found in national polls.
A CNN poll in January found national support for legalizing marijuana had reached 55 percent. Other national polls this year conducted by the Pew and CBS/New York Times report similar majorities of Americans support legalization. Gallup reported the highest level of support, at 58 percent.
I have a couple of concluding observations based on some question wording experiments that my students and I undertook.
In a Pack Poll survey this January, we randomly assigned North Carolina State students to different versions of questions about marijuana. Some were asked about legalizing marijuana, while others were asked about legalizing marijuana "for recreational use." The change in wording made no difference to students, supporters outnumbers opponents nearly 2-1 in both versions, but we did find that one in five students had no opinion at all.
That "no opinion" result raises a question about the prevalence of non-attitudes in polling about marijuana. Non-attitudes occur where respondents do their best to answer a question but had not formed an opinion about the topic prior to the survey.
Future polls should investigate this possibility by allowing respondents to explicitly say they have no opinion. They should also measure the intensity of support/opposition, since outnumbered opponents could be more firm in their views.