The conventional wisdom isn't very helpful when you're trying to predict anti-American riots.
In just 10 days, a crude film caricaturing the Prophet Muhammad and denigrating Islam set off anti-Western protests and riots in 44 countries. Predictably, observers offered up the usual set of ideological, religious, economic, and demographic explanations for this reaction: rising anti-Americanism, poor economic growth, mobs of unemployed youth, radical interpretations of Islam, and lack of experience with free speech. Such explanations seem like reasonable indicators of potential unrest, but do they help us predict which countries run the risk of sudden, possibly violent protests and attacks on embassies?
It turns out they do not. The best predictors of where protests will erupt after an initial galvanizing event -- such as the release of a controversial video -- are actually simple measures relating to the organizational readiness of Islamist movements. Many countries have large numbers of aggrieved citizens, but not all have the organizational and institutional wherewithal to quickly channel popular unrest into action.
A closer look at the data provides striking support for this argument. As part of a broader research project on outrage triggered by "blasphemous" events -- such as the Terry Jones Quran burning in 2010 and the Danish cartoons of 2006 -- we examined how the 113 countries with at least 100,000 Muslims responded to the current controversy. The usual explanatory factors emphasizing ideology, religion, economics, and governance do little to explain where protests erupted in the first 10 days after the film emerged. Specifically, countries' wealth, growth rate, unemployment, age structure, state capacity, civil liberties, democracy level, and the percentage of the population that is Muslim were all utterly unhelpful in predicting where protests would occur. Where available, even measures of religiosity from the Pew Global Attitudes Project, including prayer frequency, the importance of religion, and prevalence of the belief that there is only one correct interpretation of Islam, were not predictive.
So what really matters? Statistical analyses suggest that the ability to quickly field protests and riots is better explained by the organizational capacity of Islamist movements. Accounting for all the variables listed above, we find that protests occur most frequently in countries that had any reported demonstrations during the Arab Spring movement (a measure of recent mobilization), have an Islamist political party, and/or have organized radical militant organizations. In fact, one can very accurately identify which countries had protests in the first 10 days after the video emerged by using only these three factors; of the 70 countries with none of these factors, a mere 11 percent had protests. By comparison, of the 44 countries with just one of these factors, 77 percent had protests. The 25 countries with two or more of these factors all had protests in the first 10 days after the film became public. In other words, measures of organizational capacity are very accurate predictors of rapid protest, far more so than measures of religious ideology, economic conditions, or regime type.
Protests that turned violent such as those in Libya and Yemen show the same pattern: none of the usual explanations are helpful, but organizational features are highly predictive. For violent events, we find that the presence of an Islamist party is by far the most powerful predictor. Among the 95 countries examined with no Islamist party, only one had a violent protest. But among the 18 countries with an Islamist party, eight (44 percent) had a violent protest. Such protests were thus 42 times more likely when an Islamist party was present. It's no surprise then, that the United Arab Emirates, Turkmenistan, and Gambia, where there are no formally organized Islamist parties, did not experience protests while Pakistan, Afghanistan, and Lebanon, where there is a long tradition of Islamist political mobilization, erupted with violent demonstrations in recent weeks.
Indeed, qualitative evidence suggests that the protests themselves, as well as the use of violence during protests, often appears to be the result of deliberate decisions made by organizations rather than the boiling over of an angry public. Journalistic accounts from Pakistan, Afghanistan, Egypt, Libya, Lebanon, and Yemen have highlighted how religious, party, or extremist organization leaders directed, prevented, or halted protests. In Lahore, for example, it was members of Lashkar-e-Taiba -- the group responsible for the 2008 Mumbai attacks -- seen leading the march on the American consulate. Elsewhere, religious organizations helped keep protests peaceful.
We emphasize that the relationship between these organizational factors and the likelihood of early protest is not necessarily causal. It's entirely possible that some third variable causes both of these indicators and the observed effect. One possibility is that unobserved but deep-seated anti-American attitudes explain both Islamist organization and rapid protests. While we cannot rule out this possibility without better data on anti-American attitudes, it seems likely that Islamist parties and militant groups at least exacerbate anti-American attitudes, and more to the point, play a role in rapidly channeling and coordinating that sentiment into visible protest. Whatever the causal relationship, however, Islamist organizational strength is a powerful and potentially useful predictor of sudden protests and riots, particularly as the United States or other Western nations seek to prepare for future protest events.
Our results only speak to the sudden outbursts of protest in the 10 days immediately after the film spread. Protests have in fact spread to other countries in subsequent days. For example, Mauritius, China, Japan, Greece, and Macedonia have all recently experienced peaceful protests. Nevertheless, the lesson we take from the data is that if we want to prepare for rapid eruptions of possibly violent collective action in the future, looking to the usual predictors such as religious fervor, political freedom, or economic frustration may not help. Instead, signs of preexisting organizational capital and recent mobilization appear to be far more telling.