Tuesday, May 27, 2008

Lies, damned lies, and statistics

The fastest way to build trust and generosity in someone is to spike her nasal spray with a dose of the hormone oxytocin. But hurry to the point--money, promotion, sex, whatever it is you can't get without cheating; the drug degrades to half its strength in only three minutes. The science article with this info coincidentally comes with a graph that suggests, among 29 nations, Iranians are the fourth least likely people to resort to this sort of deceitfulness. Norwegians stole the gold medal in trustworthiness, Denmark grabbed the silver, and if it hadn't been for the Chinese, Iranians would have come home with the bronze. Intrigued, I went surfing and found data that I liked even better. According to a World Values Survey of no less than 88 nations, the top scores in the trust games were Denmark: 60.1, Sweden: 62.4, Norway: 63.9, Iran: 65.4! Whatever you do though, don't leave your nasal spray unattended in Brazil. They scored dead last in the survey with a score of 4.8. If honesty were soccer, Iran would clobber Brazil 13.6 to one.

But before we start the doodooridooing, it is fair to ask how the integrity score is calculated. Simple; the score is the percentage of the population sample in each country who say other people can be trusted. To make sure a high score doesn't indicate a nation of trusting fools, these numbers are checked against "wallet drop" tests. The two approaches correlate well. This means if you drop your wallet in Iran, there is a 65.4 percent chance it will be returned to you-or maybe it means if you drop a wallet with $100 in it, you'll get back $65.40; take your pick. [see note 2]

Iran's first place result was so astonishing that one study using the World Values Survey threw out the data for Iran altogether, citing some technical equivocation. I checked the authors expecting to find a list of Brazilian sore losers. Instead, there was only Dr. Christian Bjornskov of Danish nationality (fourth place). His paper "The Determinants of Trust" posits that trust in a society is a kind of capital just like any other kind of economic asset. This economist belongs to a school of thought that says high levels of trust in a country bring about social goodies like economic growth, rule of law, democracy, clean government, good education, and low crime. The IRI is sitting on the world's largest reserve of social capital, yet it is still waiting for prices to come down. This is why our glorious score was thrown out as suspect.

Yet Bjornskov objectivity is also suspect. According to his research many Muslims use the phrase "Inch' Allah" (sic) in their daily life. He concludes, "...which means that only contingent on a number of factors do people feel morally obliged to keep their promises. This God given uncertainty naturally could lead to lower trust in fellow citizens."

Most Iranian Muslims I know refrain from saying "Inch' Allah," or dare think it. A similar sounding phrase, often mumbled, advertises the speaker's belief in an entity that would dip him in molten lead if he doesn't return people's wallets. If the above Iranian mumbled things like, "110% absolutely," then I would know he is clueless about the melting point of lead, and would not trust him. Invoking the will of Allah is a trust builder, not a trustbuster.

Bjornskov's reasoning is more at home in analyzing his native Nordic phrases. He notes that the old Viking saying, "a word is a word," is sometimes followed by "and a man is a man". This shows that, "...if a man was to break his word he would no longer qualify to be treated as such." Assuming every Danish male wants to be treated "as such," Bjornskov's predictor guarantees a minimum integrity score of 50% for his country. Presumably, the balance is contributed by Danish women who happen to be virtuous for unexplained reasons.

Before our dispute descends further into mother/sister name-callings, let me say I was actually impressed by Bjornskov's masculine appeal to "a word is a word, and a man is a man." Despite his low opinion of Islamic peoples, he seems familiar in a Viking way with the Iranian concept of looti gari. This makes me phrase my reaction to him in different terms: "Daashteem Pahlevoon? When did it become the rasm of the best and brightest of Vikingdom to spread misunderstandings as social science?"

As for the misunderstanding about Brazilians, no amount of oxytocin is going to make me believe that communities can be different from each other by trust factors as high as 13.6. Here's what I propose to Brazil: we Iranians are often desperate for a good soccer coach, and you folks obviously need world class coaching in how to respond to surveys. Shall we shake hands on the deal, or would you trust a mustache hair as collateral?



Note 1: The survey scores can be found in the appendix at the end of Bjornskov's paper, "The Determinants Of Trust," linked to in the above the article.

Note 2. In the second paragraph, the technical term "correlate" is used with some artistic license. A score of 65.4% in the survey doesn't necessarily mean a score of 65.4% in the "wallet drop" test.