Tuesday, October 20, 2015

Meta-perception and plausible deniability

Given the readership of this blog, I may be starting a fight I can't win by criticizing social science research in fairly general terms, but I have long had the suspicion that researchers in fields like political science aren't paying enough attention to certain aspects of the data, both in terms of what they gather and what they focus upon. Specifically, I would like to see more time spent studying and discussing what I'm calling meta-perception. The use of the term appears to vary somewhat from user to user. My definition is perceptions of perceptions. We spend a great deal of time asking people "what do you think?" when the operative question might be "what do you think other people think?"

This question is particularly relevant when trying to figure out what is going on in primaries. I previously argued that a great deal of the violent fluctuations we saw in the 2012 Republican race could be explained by voters who were unhappy with Romney trying to decide behind whom the other voters who were unhappy with Romney would coalesce. Of course, those other voters were also engaged in the same activity. Opinions of candidates shouldn't change all that rapidly, but opinions of opinions of opinions certainly might.

One of the reasons that meta-perceptions get pushed to the side may be because they very often track with plain old perceptions. The causality behind this relationship goes both ways. Because of social norming, we instinctively tend to align our views with what we perceive as being the consensus view and we also tends to project our views on to other people. Both of these factors mean that the questions "what do you think?" and "what do you think they think?" will tend to get similar answers, but as with so many situations, it is when variables that normally correlate veer away from each other that things get interesting.

Check out this excerpt from a recent Paul Krugman post:

But the odd thing about these revelations is that they weren’t at all revelatory. We shouldn’t have needed McCarthy blurting out the obvious for the press to acknowledge that the Benghazi investigations have utterly failed to find any wrongdoing; and Clinton has been in public life a long time, so that her strengths were or should have been well known.
Let's phrase this in a different way. We would expect a strong correlation between new information and changes in perceptions. The more information, the bigger the changes in my worldview. That's not at all the case here. At the risk of belaboring the obvious, McCarthy's comments were in no way informative to anyone who has closely followed the story, let alone to members of the Washington press corps or editors of national newspapers, and yet the statements had a powerful and immediate effect on the narrative. That's the odd part.

Why should the conversation change so radically just because a senator refers to something that everyone knows, something that everyone has always known? Because the change came, not in perception but in meta-perception. When McCarthy forgot not to use his outdoor voice, he created an I-know-you-know-I-know situation. The journalists covering the story and the pundits discussing it lost their plausible deniability.

1 comment:

  1. I've used metaperception in some small surveys I've done around views of teachers on technology to get around exactly the same issue.

    But the broader point about metaperception is even better made. Of course, you see echos of this puzzle across a whole range of disciplines (not calling it metaperception). For instance, Havel's Power of the Powerless deals just with these little acts of acknowledgement of metaperception.

    Linguistics/pragmatics deals with these issues under the label of 'performativity' which in anthropology, they are reflected in 'liminality'.

    Sociology often talks about people challenging or subverting the status quo through various small acts.

    All of the above take a slightly different perspective on the same broad issue. But I think it would be very interesting to bring them together in the same way I've tried to do for framing (which is also related). I've been looking at performativity for a long time but I like the metaperception take on it.