Saturday, August 23, 2014

Putnam, trust, and flirting

Everybody else seems to have already read Putnam's paper showing that diversity reduces both out-group and in-group trust. I'm curious whether the startling part--the reduction in in-group trust--was replicated in other cultures as well.

Now that I think of it, does the increase in political activity in diverse cities reflect a loss of trust?

Someone noted that attitudes towards Indians in the US improved the farther away they were. I don't recall who noted this or on what basis, but I notice that Twain and Nye, though on occasion noting good qualities of an individual, tended to express less flattering views of Indians than Rousseau or Pope. The lack of trust of outsiders can be based on experience.

But, since Putnam tried to control for the crime rate, let's assume that this isn't the driving factor.

There were a couple of interesting features in the distributions Putnam showed. One is that the relationship between ethnic diversity and trust was linear. This seems a little surprising to me: insofar as it diminishes I expect it to diminish faster as the environment becomes more complicated. That we see the simpler relationship seems to suggest that the response isn't a matter of being overwhelmed, but is a rational response to some factor.

One simple model of what is going on with trust of other groups is that lack of trust varies with risk. The risk can be physical (see crime rate), but would more usually be "status risk."

For example, is the young lady with the tight blouse flirting with you? Maybe it's obvious, but maybe frequent eye contact just signified confidence back in her home town, and your response just earned you a snicker instead of a smile.

The neighbors in the green house are exceptionally loud. Are they having marital problems? Is he threatening violence? Or is this a benign case of "loudest wins?" This is a live issue in our household: a touch of Asperger's can make it hard to tell from the tone of voice (words are typically not very intelligible) whether the shouter across the street is seriously angry or not.

Risk is obviously proportional to the degree to which you cannot "read" the other person's signals. Only some of the signals are verbal denotation. We all know lots of non-verbal signals, and lots of idioms and local connotations, such as the Brit saying goodnight to an American lady with "I'll knock you up in the morning", or when Arthur Blessitt asked to hold a "rally" when he brought his cross to Liberia "rally" means a church fund raiser there.

We can sometimes tell when someone is a bit "off." (Generally they're harmless, but you have to learn more about them to tell for sure, and about 1-2% of the population will prey on you given the opportunity--more if their culture encourages predation on outsiders.) I'm guessing here, but I'd think it much easier to spot problematic deviations from the norm when you have a fairly mono-cultural norm. I'm not sure whether different races confuse, but I'm pretty sure different cultures do. (But maybe I was just a little dense as a youth.)

So I'm guessing that in a multi-ethnic community people will be slightly more likely to misidentify reliable people as unreliable, or at best of uncertain reliability.

There are also things that seem more like carrier waves or heartbeat signals. This can include what AVI calls tribal markers, but I think they don't just signal "I'm here" but also, from the non-verbal components, let your co-tribesmen know whether you're OK or having some problems. Small talk about the weather (or whatever is the thing for your tribe) allows the conversation to grow if needed. (Cold calls are hard for most of us to learn to do; so similarly is starting a heavy conversation cold.)

You know the frantic search to find something useful to answer when somebody opens the conversation with something completely alien, like BDS or the prospects of the Ukrainian rugby team. Having something to ease conversation into being is important.

Sometimes the tribal markers or tribal courtesies are experienced as micro-aggressions, and some of the unspoken courtesies (what volume you use when arguing in public, whether you nod or hail-fellow, what clothes you wear(*)), will rub you the wrong way by their absence.

So far so good. You clearly run a higher "status risk" when trying to deal with someone from an alien culture. But why should there be a difference in trust between having 10% of your neighbors be different vs having 20% of them?

It may have to do with risk estimate. Suppose the risk of misreading the intentions of a member of an "alien culture" is p, and the trust will be proportional to 1-p). And suppose the risk of misreading the intentions of someone of your own culture is q, smaller than p. If there are 100 in your neighborhood, the chances of misreading anybody is ((1-p)^N)((1-q)^(100-N)). Let q be .1% and p be 1%. For between 1 and 50 members of an "alien culture", the risk of misreading someone looks nearly linear, as shown below. (Sorry, my X-axis runs the other direction from Putnam's.) I pulled these numbers out of the air, of course, but I mostly expect people to be fairly good at figuring out who’s not trustworthy after a trial or two. Of course some of us make a good living off that failure rate.

But wait, why am I looking at the overall trust, and not trust in my group vs trust in the other group?

I have thought of only two simple models that explain the decline in both domains of trust as a rational reaction. This “calculation” is part of the first model—though it doesn’t deal with the reported offset (see Putnam’s Figure 6) between in-group and out-group trust.

  1. The survey respondent feels the whole group to be a community, and is describing the sense of trust he feels with respect to the community as a whole, with an (unexplained) offset for trust in his own group. This is one reason I'd like to see whether this effect is reproducible outside the West, in more explicitly tribal environments. We try(ied) to cultivate a "you're responsible to the community" attitude, which might change how people answer.
  2. The people in the community fall back to a kind of pidgin set of signals, and the probability that you use the pidgin within your group rather than the full set varies with the probability that you have to use it outside the group--which is proportional to the fraction of “alien culture” people in the community.

I’m not persuaded by either model. The first doesn’t explain the approximately constant offset in trust, and the second doesn’t motivate why you should use the pidgin in-group—it is just assumed easier. AVI suggested that the in-group distrust came from “activation” of distrust, but that’s not a rational reaction. Not that people are always rational, but this exercise was to look for a rational reason for the changes.

It is possible to take Putnam's Figure 6 too seriously. The statistics aren’t really good enough to accurately determine a second-order variation. There might be a slope hidden in the variation, or something non-linear—though not a big one. And the loss of trust in-group may not be a rational reaction after all.

(*) You don't need a degree in semiotics to notice that clothes or other adornments are always used for communication.

1 comment:

Texan99 said...

Our neighbors down the road used to have fights so loud that their closest neighbors could hear them from several hundred feet away. Recently came the news that, while he was off on a several-day fishing trip, she moved out, taking all of the household furniture with her, right down to the dishwashing detergent under the sink. It's been a couple of months now, and it appears she's moved back in.

I don't spend much time with this couple, but I've always enjoyed their company in a distant way, without having any real understanding of what makes them tick. Somehow or other, though, they've put out the signals that make me consider them trustworthy. If I were in serious trouble, I wouldn't hesitate to call on them for help. I know they'd arrive like the cavalry, and that we would agree on important issues of right and wrong.

At the same time, we don't socialize with them. We started out with some back-and-forth invitations, but gradually discovered that they aren't willing to eat dinner before about 11pm, apparently because they think you can't drink after dinner, and they aren't willing to stop drinking until quite late in the day. We found this incomprehensible and impossible to incorporate into an ordinary social relationship.