The French Window Into Police Misconduct
September 20, 2016 (Fault Lines) — Police misconduct is a difficult concept. Like Justice Stewart’s obscenity, we “know it when we see it,” but we have trouble nailing it down with any precision. And making matters worse, we’re largely in the dark when it comes to what causes it.
To be sure, people float theories. Racism. Not enough training. An unhealthy culture. But we just don’t know, and we’re short on empirical evidence. In his Cross, police officer Nick Selby argued we don’t even have much to prove the existence of widespread misconduct. Even if we did, there is reason to believe some of our explanations are lacking; both Selby and Seth Stoughton, a former cop turned law professor, argued that the concept of a common police culture is at best flawed, at worst totally mistaken.
Being stuck in the dark is no fun. So what are we going to do about it? One possible answer comes from the world of counterterrorism research.
Recently, political scientists Will McCants and Chris Meserole tried to answer two pressing questions: where is the problem of Sunni Muslim radicalization most acute, and why? They came up with a creative solution that, to put it mildly, produced surprising results.
First, McCants and Meserole had to settle on a metric for radicalization. They chose going to fight in Syria, which they agreed was the best proxy: the war was the biggest jihadist cause between 2011 and 2014, and reaching the country was easy. To that end, they found an I.C.S.R. dataset that estimated, for each of fifty nations, how many Sunnis from that country left for Syria between 2011 and 2014.
The next step was figuring out where radicalization is an especially big problem. To that end, McCants and Meserole calculated each country’s radicalization rate and foreign fighter share. The radicalization rate tells us what percentage of a country’s Sunni population went to fight in Syria. The foreign fighter share, on the other hand, measures a country’s contribution to the total jihadist population.
By itself, neither score is all that helpful. For example, a country with ten Sunnis, nine of whom went to Syria, would have a very high radicalization rate, but it’d be wrong to conclude it’s a major driver of radicalization. Similarly, a country with tens of millions of Sunnis that contributed a couple thousand fighters to the war would have a large foreign fighter share, but the low percentage of the population that went to fight implies radicalization is mostly under control.
But when you multiply the radicalization rate by the foreign fighter share (and standardize the results,) you get the standardized foreign fighter score (SFFS), which is extremely useful. A country with a high SFFS contributed lots of fighters and a significant percentage of its Sunni population to the war, making it that much more likely to have a big radicalization problem.
McCants and Meserole calculated the SFFS for each of the fifty countries in the I.C.S.R. dataset. Five countries – Tunisia, Jordan, Lebanon, Belgium and France – scored way above the rest. Now that they had their likely suspects, the question became: what is it that makes these countries such hotbeds of radicalism? To answer this, McCants and Meserole plugged by-country statistics on a bunch of things suggested as explanations for radicalization (like wealth or education) into an algorithm, in the hope of finding variables that matched up well with and had a big effect on the SFFS rankings.*
It turns out three variables do a pretty decent job of predicting radicalization. Urbanization is good, and youth unemployment is a little better. But the best by far – it’s not even close – is whether a country has French as an official language. On average, speaking French raises a country’s SFFS by over half a standard deviation.
So what is it about French that drives people to kill? Probably not much. Rather, McCants and Meserole believe French-as-an-official-language is a proxy for French civic culture. Specifically, French-speaking countries have an unusually aggressive approach to secularism: their version, called laïcité, goes way beyond the official agnosticism of the Establishment Clause and is often used to justify restrictions on the private exercise of religion.
Notably, in 2011, France and Belgium began to enforce bans on face veils. (Tunisia’s ban was vigorously enforced during the run-up to the Arab Spring.) Of course, thanks in large part to the Sunni Muslim Brotherhood, face veils were commonplace in Europe by that time, and even Sunnis who don’t use them may see the bans and the rhetoric or enforcement practices that accompany them as an assault on their faith. Thus, the hypothesis goes, French-speaking societies are especially likely to present Sunnis with a choice between being Westerners and being Muslims, making some embrace anti-Western extremism.
The fascinating thing is that we could use this model to ask similar questions about police misconduct. As with radicalization, the first step is finding a proxy for what we want to investigate, which is easier said than done: unlike McCants and Meserole, we don’t have a smoking gun like the war in Syria, and many of the metrics we could use have obvious flaws. For instance, if we chose § 1983 lawsuits as our metric, the results would be skewed by the pro-cop legal environment of Graham v. Connor and the doctrine of qualified immunity. Official complaints? They’re never sustained, and a decent chunk may be frivolous. Criminal convictions? Don’t make me laugh.
But if we could find a metric, and if we had a workable dataset – two big ifs – we could calculate “standardized police misconduct scores,” find out which departments are the worst offenders and test theories of police misconduct to see which variables are the best predictors.
Imagine the possibilities! Wouldn’t it be interesting if officers’ educational level turned out to be a weak predictor of misconduct? If urbanization had a similar effect on misconduct as on radicalization? If anti-police demonstrations alienated cops like enforced secularism may alienate Muslims?
We don’t know, and right now, we can’t know. But if we want to get beyond mutual distrust and marinating in confirmation bias, people on both sides of the police reform debate should strive for more data. In the meantime, beware of French-speaking cops.
*For much more on this, and the BART algorithm they used, see here.