Figure 6

Forecasted probabilities for edges that experience attacks and those that do not experience attacks. Forecasted probabilities for edges that experience attacks and those that do not experience attacks.$\bar{\Pi }$ is the mean forecast probability in the respective year. The probability of an attack from one location to another in year t is computed with the TERGM estimates from the model that, in terms of predictive log score, most accurately predicted N^t−1 from$N^{t_0-1,t-2}$. In the interest of clarity, the y-axis is depicted on the log scale. Box plots depict the distribution of Π among those directed dyads that experienced an attack in t.