You’ll remember that I recently returned from a trip to Sweden. While I was there, by a curious coincidence, I happened to read a Swedish study from the Public Library of Science: “Lifting the Veil of Morality: Choice Blindness and Attitude Reversals on a Self-Transforming Survey,” by Hall, Johansson, and Strandberg.
Their results are, shall we say, provocative.
Apparently, illegal immigration is a hot-button issue in Sweden. Hall et al decided to push those buttons during a series of person-on-the-street interviews. You know the type: someone hands you a questionnaire, asks you to rate a series of statements along a 9-point scale where 1 means completely disagree, 9 means completely agree, and 5 means couldn’t give a shit. There were subtleties — some questions were phrased in terms of general principles, others in terms of specific examples — but basically the survey came down to things like It is morally deplorable to harbor illegal immigrants (agree or no?)
Here’s the tricky bit: after a respondent had checked off the numbered boxes, the researchers surreptitiously swapped out that part of the form with a lookalike that reversed the questions (so that morally deplorable became morally honorable, for example). People who hated the harboring of illegals were now on record as loving it, and vice-versa. The questionnaire was then returned to the hapless subject, who was asked to write a brief argument supporting their position.
Fully a third of those who’d registered the most extreme positions — 1 or 9 on the scale — didn’t even notice that their opinions had just magically changed 180°. Not only that, but they went on to construct arguments supporting the exact opposite of what they’d just claimed to strongly believe, effectively enough to convince a blind panel of judges as to their sincerity. And that was only the extremists; if you included those with more moderate leanings a full 69% of the respondents whole-heartedly accepted the turnabout without missing a beat1.
Really? It’s that easy?
I could go to an Operation Rescue meeting in the USA, hand out trick questionnaires to a roomful of rabid anti-abortion activists— and a third of them would spin on that dime and claim to be pro-choice, that they’d always been pro-choice?
Sheer gut-level incredulity makes you want to be really skeptical of these results. Even I don’t want to believe that people can be that stupid. So we could start by criticizing the sample size, a modest 160. Or we could talk about cultural constraints: the study was limited to Sweden, and the Swedes are (in my limited experience) far less prone to the kind of religion-fueled batshit insanity that ignites so many hot-button issues in the US. One can accept that these results apply to one group while still doubting that they’d generalize to a population of delusional fanatics. (Or even among those who are just more actively engaged. Respondents who described themselves as “politically active” were especially likely to detect the reversal— to be “rejecters” instead of “accepters”, in the parlance of the study.)
But you know, N=160 is nothing to sneeze at; if the size of Hall et al‘s sample had been wanting, well, that’s why we have statistics, and according to the stats these findings are significant. And the researchers did choose a politically-contentious issue to survey, so those filling out the form were likely to both be familiar with the issue and to have firmly-established opinions about it. Hall et al‘s’ measure of “acceptance” for a reversed result was very forgiving, too; respondents were encouraged to express doubts or reservations (So, nothing about this study felt kind of odd to you? Nothing at all?), right up to being explicitly shown how they’d been tricked. Even then, if they claimed to remember feeling even the teensiest bit doubtful during the survey, they were let off the hook and reclassified as “rejectors”.
Finally, while political activism may have been correlated with detecting the reversal, simple extremism was not: absolutist 1 and 9 respondents were no less likely to be fooled than the more moderate fours and sixes. You might infer from this that while informed opinion is relatively hard to shake, people whose opinions are both extreme and ignorant are especially easy to reprogram — and that demographic is huge enough to affect the outcome of pretty much any election.
Let’s put aside, for the moment, the fact that this runs counter to everything I’ve ever experienced when dealing with ignorant extremists in North America. Let’s play devil’s advocate, and ask: why shouldn’t these results be valid? They’re utterly consistent with the haphazard way our brains parse reality. Distracted for a moment, we don’t notice that the person we were just talking to has been replaced by someone else. Watching a circle of people throwing balls back and forth, we don’t notice a guy in a gorilla suit waving at us from the center of our visual field. Buildings pop into and out of existence during any number of perceptual experiments; most of the time, we just don’t see it happen. Why would the perception of our inner thoughts be any more reliable than that of the physical world outside our eyeballs?
Hall et al worry about the implications. They fear that the flimsiness of so-called “strong beliefs” might compromise the very idea of Public Opinion, that without taking such effects into account the whole polling-industrial complex might be founded on shifting sand. My own fears are more proximal. I’m not worried about whether people could correct for these effects; I’m worried about people using them. What state or corporation could resist the urge to take such an easy shortcut? Instead of trying to sell the public on your product, just make them believe they’re already sold on it. Instead of campaigning to win over the hearts and minds of the electorate, why not just commission a quick phone survey to make a good chunk of them realize they’ve already been won?
Scary doesn’t begin to cover the proposition.
I mean, I don’t know what I’d do if someone tried to subvert my deep and abiding love for Jesus Christ, my personal Lord and Savior.
1Even those who did notice the difference didn’t realize that the questionnaire had been altered. They just assumed that they’d misread the question the first time around.