Trust and violence

Agnostic has been marshaling a sundry series of posts that trace how shifts in the rate of violent crime are associated with changes in the behaviors and culture of broader society, from the rise and fall of shorts' length (heh) to disappearance and resurrection of drive-ins.

Initially, I figured, understandably enough, that trust levels would tend to rise as violent crime declined, and fall during times of increasing violence. Agnostic argues that this is not the case, however. Instead, increasing levels of trust propel people out of their cells and into the commons where they come into more social contact with other people than they do during low trust times, when people keep to themselves. It is this increase in social mingling that brings on an increase in violent crime, as the perpetrators of those crimes now have greater and more varied opportunities to strike (and it is not always necessarily a consequence of mendacious intent on their part--more social contact inevitably means more confrontations that may result in violent outcomes, even if neither party involved planned on having it end up that way).

Agnostic's trust-influences-violence-and-violence-influences-everything-else theory (my eloquent naming, not his!) is intriguing if not obviously convincing. How does the increasing racial and ethnic diversification of the US, and the consequent "hunkering down" it brings, play into this relationship? Robert Putnam shows that diversity dampens social participation, but it is also associated with higher rates of violent crime. Of course, despite increasing racial and ethnic diversity in the US, crime levels have fallen steadily for two decades now. As more diversity brings more social retreat and lower levels of trust, will violent crime rates continue to decline? Perhaps a continuing drop in violence should be thought of as a silver-lining in the balkanizing skies above.

Anyway, I wondered if any correlation between trust levels and violent crime could be detected in the US from the early seventies to today. Scaling the national rate of violent crime to responses to the GSS question over whether "most people can be trusted" or "you can't be too careful in life", I graphed the relationship:


The correlation is an insignificant .10. When violent crime is compared to trust levels a couple of years prior, the correlation becomes an inverse (which is to be expected in Agnostic's conception of the trust-violence relationship) .16 (p = .47). That's only slightly less easy to dismiss as meaningless.

By no means does this refute any part of Agnostic's theory, as the GSS survey question just gauges a general feeling among respondents, not how they actually behave, and is subject to some statistical noise (though the average annual sample size is over 1,400). Further, trust has been declining for three decades, violent crime for two. The lag might simply be considerably longer than two years, maybe closer to ten, something that is unfortunately not testable using the GSS. But it would've been neat if it clearly supported the theory, which it does not do.

GSS variables used: YEAR, TRUST(1-2)

Trusting for fairness

The role of trust in understanding how societies and individuals function is a fascinating topic, and one I don't feel like I coherently understand. High-trust countries function better than low-trust countries do, and they're more desirable places to live.

What is a high-trust society? Trust for family members and trust for strangers are two distinct things, and may even be inversely correlated. It is the latter--trust for strangers--that is generally referred to when levels of societal trust are discussed or measured, however. In a post where Arnold Kling takes a stab at it, commenter Lars Smith pithily defines it for me:

A high trust society is one where vegetables can be sold unattended on the roadside, placed in boxes together with a price list, a pair of scales, and a cigar box for the money.

And of course this works fantastically if everyone plays by the rules--no need for the deadweight labor loss of guarding the vegetables, wrangling in court with those accused of trying to steal them, and ultimately punishing them for their theft. When the rules are breached though, it doesn't, and there is only so much we can expect a pro bono unassuming local guy to do.

Trustworthiness is a desirable trait at every societal level save for the individual one (at least in certain circumstances, including the obvious ones like being able to successfully steal merchandise from a retailer without being caught). Being trusting of others, however, is not necessarily so--naivete is not self-evidently desirable like trustworthiness is, and being overly trusting allows for free riders and cheats to gain more from their free riding and cheating, thus incentivizing them to do more of it in the future.

There is a chicken-and-egg question to ponder here, too. Are Scandinavians more trusting than Brazilians by natural disposition, or are they more trusting because other Scandinavians are more deserving of trust than other Brazilians are? Both reasons, probably. Is, ceteris paribus, it better to be a trusting person or one who is suspicious of others? Holding all other things equal when asking a question like that is virtually impossible to do.

That said, my working assumption is that a trusting society, irrespective of how trustworthy members of said society are, is preferable to one in which most people are suspicious of others. The US is usually considered a mid-to-high trust society, lower than Northwestern Europe but higher than South America, Asia, Africa, and the Middle East. Tapping the GSS, the following tables offer some insight into how relatively trusting various demographic groups are in the US.

The trust item employed here was introduced in 2010, so all the results have contemporary relevance. The question, set on a five-point scale, asks each respondent if he thinks most people try to be fair with him or try to take advantage of him when they get the chance. The higher the score, the more trusting of others the group is. The standard deviation for the entire respondent pool is 1.37.

By race:

Race	Trust
Whites	3.24
Blacks	2.46
Hispanics	2.93
Asians	2.87

By intelligence via wordsum vocabulary test scores, with respondents broken up into five categories; Really Smarts (wordsum score of 9-10, comprising 13% of the population), Pretty Smarts (7-8, 26%), Normals (6, 22%), Pretty Dumbs (4-5, 27%), and Really Dumbs (0-3, 12%):

IQ	Trust
Really smarts	3.63
Pretty smarts	3.30
Normals	3.04
Pretty dumbs	2.87
Really dumbs	2.59

By religious belief:

Belief	Trust
Atheists	3.32
Agnostics	3.42
Uncertain believers	3.12
Firm believers	3.00

By partisan identification:

Party	Trust
Republicans	3.30
Independents	3.05
Democrats	3.00

By political orientation:

Ideology	Trust
Conservatives	3.13
Moderates	3.06
Liberals	3.10

By educational attainment:

Education	Trust
< High school	2.54
High school grad	3.00
Some college	3.04
Bachelor's degree	3.41
Graduate degree	3.77

By age:

Age	Trust
18-29	2.75
30-44	2.90
45-64	3.18
65+	3.58

Educated, intelligent, white, Republican, older, and agnostic--these are the characteristics of a high-trust individual. Save for agnosticism, these are also all characteristics of the America of the past. Immigration-driven changes in the US' demography are ensuring that this country becomes a less trusting one in the future than it is today.

GSS variables used: FAIR5(1-5), RACECEN1(1)(2)(4-10)(15-16), AGE, EDUC(0-11)(12)(13-15)(16-17)(18-20), GOD(1)(2)(3-5)(6), POLVIEWS(1-3)(4)(5-7), PARTYID(0-1)(2-4)(5-6)

Cold War, (relatively) warm feelings

As someone who sees the future as more clannish and tribalistic than the one he grew up in, I wanted to take a look at whether or not that intuition is borne out by data both within the US and beyond its borders.

First, the home front. The following graph shows the percentage of GSS respondents saying that "most people can be trusted" (as opposed to "you can't be too careful in life"):


The always present year-to-year random fluctuations are present, but the trend is clearly downward. Despite all the happy talk about diversity, multiculturalism, a global world, and the like, as the US increasingly sees its dichotomous white-majority, black-minority demography replaced by a demographic composition that serves as a microcosm of the entire globe, the levels of trust people have for others continues its steady decline.

What about the other 95% of humanity? The World Values Survey is the best place to turn. The WVS began in 1981, but prior to 1990, it would be more aptly named The Free, First-World Values Survey. So it is during the twilight of the Soviet Union where we begin, and the years just in front of the 'Great Recession' where we end.

The following table shows the percentages of respondents, by country among the 27 that participated both times, who said most people can be trusted in 1990 and during the period of 2005-2008. The last column (by which the table is organized) shows the change in societal levels of trust during that nearly two decade period of time:

Country	1990 (%)	2005-8	Change
1. Switzerland	43.2	53.9	10.7
2. Norway	65.1	74.2	9.1
3. Germany	31.8	36.8	5.0
4. Romania	16.1	20.3	4.2
5. Brazil	6.7	9.4	2.7
6. Sweden	66.1	68.0	1.9
7. Slovenia	17.4	18.1	0.7
8. Japan	41.7	39.1	(2.6)
9. Finland	62.7	58.9	(3.8)
10. France	22.8	18.8	(4.0)
11. Italy	34.2	29.2	(5.0)
12. Turkey	10.0	4.9	(5.1)
13. Argentina	23.3	17.6	(5.7)
14. South Korea	34.2	28.2	(6.0)
15. China	60.3	52.3	(8.0)
16. The Netherlands	53.1	45.0	(8.1)
17. Bulgaria	30.4	22.2	(8.2)
18. South Africa	28.3	18.8	(9.5)
19. Canada	52.4	42.8	(9.6)
20. Chile	22.7	12.6	(10.1)
21. Russia	37.5	26.2	(11.3)
22. India	35.4	23.3	(12.1)
23. United States	51.5	39.3	(12.2)
24. Poland	31.8	19.0	(12.8)
25. Great Britain	43.6	30.5	(13.1)
26. Spain	34.3	20.0	(14.3)
27. Mexico	33.5	15.6	(17.9)

A handful of mostly homogeneous, non-Anglophone Western European countries show increases. Everywhere else, declines in trust levels is the norm. I suspect that current trust levels are considerably lower than they were even 3-6 years ago, when the economic outlook conventionally appeared far more favorable than it does today.

I hear Dave Matthews singing, "Don't trust me, trust yourself".

GSS variables used: TRUST, YEAR