“…Crime rates may be correlated with latent factors and, if we do not correct for them, the coefficients of the variables homicide rate and cocaine confiscations could be reflecting the effects of these factors and, thereby, be biased. For example, poverty, inequality, and lack of education are elements strongly associated with crime (Gimenez, Svitková, Tkacheva, & Barrado, in press). To avoid biased estimations, the model included the SDI, a variable explained in the previous section, that accounted for a wide set of socioeconomic factors in the districts, generally believed to be correlated with severe crimes.…”