Margolis: In polls, sometimes ‘sometimes’ is the easiest answer

Creative Commons photo by asgw via Flickr.

Creative Commons photo by asgw via Flickr.

Editor’s note: Jon Margolis is VTDigger’s political columnist.

Want to know what people think?

Read a poll. It can be helpful.

But be careful. It can also be misleading.

Especially when the topic is complicated. Polls are most reliable when they measure how many folks are going to vote for Smith and how many for Jones. When polls ask about opinions on issues, or what people would do in certain situations, the results are as likely to perplex – perhaps to bamboozle – as to enlighten.

That’s because polls often measure a public opinion that does not exist. Many people – including quite a few who keep up with what’s going on in the world – simply don’t have an opinion on every issue. In many cases, they haven’t given the matter a bit of thought. But not wanting to disappoint the person on the other end of the line awaiting an answer, many a respondent will provide one.

Maybe the one he or she would have provided 10 minutes later after thinking about it. Maybe not.

This week, two polls in the news in Vermont – one about physician-assisted suicide, the other about what men think about violence against women – should be viewed with these caveats in mind.

Not that they are in any other way comparable. The survey of 341 Vermont men taken by the Castleton Polling Institute on behalf of the Governor’s Prevention of Domestic and Sexual Violence Task Force was useful if only because it can help “establish a baseline (so that) in five years we can run essentially the same poll to try to see if there are any changes,” in the words of the task force’s Chris Dinnan.

But the poll of 500 Vermonters taken by Smith-Johnson Research for the Vermont Alliance for Ethical Healthcare does not meet the laugh test for credibility. It starts off by asking respondents to choose among three options: the present law; a “law that allows a doctor to give a terminally ill person a prescription for lethal drugs to take their own life;” or a “compromise.”

Forget the grammatical error for a minute (it should be “take his or her own life”) and concentrate on “compromise.” If there is one thing many Americans – and perhaps especially many Vermonters – love, it’s compromise. It seems so reasonable, so even-handed, so moderate. Especially for those without a strong opinion or much knowledge of the issue, compromise is a comforting middle ground.

Then the poll lists six objections some have raised to a doctor-assisted suicide system – but not, significantly, any refutation of those objections – after which it repeats the first question. And it still couldn’t get more than 36 percent of the respondents to favor leaving the law as it is.

Sorry, Vermont Alliance. You’re going to have to do better than that to refute the finding that most Vermonters (68 percent in a reliable Castleton Institute Poll of last February) favor physician-assisted suicide.

Some guy is sitting in his living room when the phone rings and he’s asked, “How often do you think that accusations of rape are false?” He hasn’t the foggiest idea. And “sometimes” like “compromise” must seem like as good an answer as any. It’s so reassuringly moderate.

The men’s survey was a straightforward effort to measure male attitudes about domestic violence. But there is a big difference between attitudes and opinions, and in examining this poll, skepticism should center not on the results but on their interpretation.

At last week’s press conference announcing the results, for instance, Dinnan said it was “concerning” that almost half the men surveyed thought rape accusations were “sometimes” false, instead of “rarely” false, because national research shows that only 2 to 8 percent of accusations are false.

No, it does not. The research is murky and imprecise, and the 2 percent figure has been largely discredited by scholars. A better estimate, according to an article by Emily Bazelon, a lawyer and writer whose feminist credentials are impeccable, and Rachel Larimore of Slate Magazine, is that between 8 and 10 percent of rape accusations are false.

As Dinnan acknowledged in an interview, the difference between “rarely” and “sometimes” can be “nebulous” and “open to interpretation.”

Besides, the percentage of false accusations is not a matter of opinion. It’s a matter of knowledge (however imprecise in this case), and of knowledge that nobody can be expected to possess. It isn’t as though the topic has been in the news of late. There are no rape-reporting bills before the Legislature. Some guy is sitting in his living room when the phone rings and he’s asked, “How often do you think that accusations of rape are false?”

He hasn’t the foggiest idea. And “sometimes” like “compromise” must seem like as good an answer as any. It’s so reassuringly moderate.

Nor is there any reason to assume that a man who answers “sometimes” is any less concerned about domestic violence than one who answers “rarely.”

At the press conference, task force officials also said they were concerned that while 88 percent think violence against women is common nationwide, only (only?) 77 percent thought it common in Vermont.

This result makes perfect sense. Go back to that guy in his living room who’s asked a question he knows nothing about. One thing he might know that crime in general is less common in Vermont. So it’s a reasonable guess that sexual violence is less common, too.

Task force officials were also unhappy about the results of another hypothetical, one directing respondents to “pretend a man and a woman are in a public area when an argument begins,” and asking at what point (yelling at a woman, humiliating her, threatening her, actually hitting her) the respondent “would intervene.”

And the result was? That only imminent threat or physical violence would prompt most men to get involved, a finding which should not have surprised or displeased the task force. First of all, to that guy answering his phone and having this dilemma sprung on him, “intervene” quite likely meant getting up and trading (or at least absorbing) blows. Not an inviting prospect. Of course, “intervene” also might have meant using his cell phone to call 911, or alerting (depending on location) the head waiter or the lifeguard. But the respondent might not have thought of that right away.

Then there’s that old – and generally admirable – American ethic of minding your own business. Like any rule, it can be taken too far; letting a woman get beaten to a pulp in front of one’s eyes is probably not praiseworthy. But husbands and wives sometimes yell at each other. Staying out of their squabble is not a sign that a fellow is indifferent to sexual violence.

All those questions might be useful for establishing that “benchmark” mentioned by Chris Dinnan (one of three men on the task force). But using any one of them to draw conclusions about the views of the male not-quite-half of Vermont is ill-advised. To repeat: Polls can be useful. If in early November, Smith is seven points ahead of Jones, you pretty much know that Smith is going to get elected. If 42 percent of Vermont men think that rape accusations are sometimes false, you know … nothing at all.

Jon Margolis

Comment Policy requires that all commenters identify themselves by their authentic first and last names. Initials, pseudonyms or screen names are not permissible.

No personal harrassment, abuse, or hate speech is permitted. Be succinct and to the point. Comments should be 1000 characters or fewer. If your comment is over 500 words, consider sending a commentary instead.

We personally review and moderate every comment that is posted here. This takes a lot of time; please consider donating to keep the conversation productive and informative.

The purpose of this policy is to encourage a civil discourse among readers who are willing to stand behind their identities and their comments. VTDigger has created a safe zone for readers who wish to engage in a thoughtful discussion on a range of subjects. We hope you join the conversation. If you have questions or concerns about our commenting platform, please review our Commenting FAQ.

Privacy policy
Thanks for reporting an error with the story, "Margolis: In polls, sometimes ‘sometimes’ is the easiest ..."