This project includes several analyses which examined root causes of terrorism in terms of regime structure and regime stability.
One analysis examined the relationship between democracy and terrorism and found that most of the risk lies in those nations that are "nearly" democracies. Compared to all other regime types, at their peak those countries that are close to full democracies have experienced almost three times as many terrorist attacks. Fully democratic countries generally experienced a moderate but persistent level of attacks throughout the series. In contrast, mostly autocratic countries generally had the lowest average number of attacks. In a second analysis that examines components of democracy, we found that the more veto players in a country (or persons who can block policy change), the more likely that country is to experience terrorism (both in probability of attack and frequency of attack). This finding suggests that nations provide access to the policy process for a wide range of diverse viewpoints to reduce frustration that could lead to terrorist violence. Other factors that were related to terrorism probability and frequency include war and a recent regime change. Finally, in a third analysis, we compared the 1970 to 1997 terrorist trends for nations with various experiences as failed states. Overall, those states that had ever experienced a failure experienced less terrorism in the 1970s but more in later years. By 1997, the "ever-failed" states were experiencing three times more terrorism than other nations. Evidence also strongly suggests that terrorist attacks against the ever-failed states are much more lethal than attacks against other nations. When we change our measure of state failure to only reflect those countries that are in failure during the current year (allowing the composition of failed states to change from year to year), the patterns described above are even stronger. If we consider only the period after 1979, the average number of attacks for the "out of failure" states only once rises above 40 (to 42), whereas the average for the "in failure" states approaches 100 several times. The difference in fatalities by terrorists for nations in and out of failure is even more profound. Nations out of failure average almost 20 fatalities each year, whereas nations in failure average more than 115 deaths each year.
All analyses required that we compile other datasets and merge them with the Global Terrorism Database (GTD). We matched measures of democracies for countries and years from the Polity IV dataset to the GTD. Furthermore, the state failure data came from the Political Instability Task Force (PITF), a multi-disciplinary group originally assembled in 1994. Data on veto players came from the Database of Political Institutions and the Political Constraints Dataset. In order to merge these datasets with the GTD, we had to resolve differences in the sovereignty of countries over time. Statistical methods include descriptive analyses and multivariate analyses to examine the relationship between various possible root causes and terrorism. The multivariate analyses include negative binomial, zero order negative binomial, logit, rare events logit, and Tobit models.