“…Thus, following social disorganization theory (Shaw & McKay, 1942) and collective efficacy theory (Sampson et al, 1997), we collected census tract-level socio-economic data in 2013 from the US Census Bureau and aggregated them to the neighborhood-level to serve as control variables in the statistical models. These control variables include concentrated disadvantage indicators: poverty rate, unemployment rate, as well as the young adult (18–35) rate, and the median household income (Chiricos, 1987; Haberman & Ratcliffe, 2015; Parker et al, 2005; Raphael & Winter-Ebmer, 2001; Sampson et al, 1997), residential instability indicator rental housing rate (Boggess & Hipp, 2010; Sampson et al, 1997), and the ethnic heterogeneity index calculated as 1−(B2 + H2+W2)/(B+H+W+O)2, where B, H, W, and O are the numbers of residents of Black, Hispanic, White, and Other ethnicities in the neighborhood (Bernasco & Block, 2011; Lan et al, 2021; Patsiurko et al, 2012). The total count of crime attractors/generators (public transit stations, ATMs, bank branches, bars, stores, liquor stores, movie theaters, night clubs, schools, hospitals, parks, recreational places, and restaurants) was controlled for as well because they were tested to be closely related to crime opportunities (Brantingham & Brantingham, 2008; Liu et al, 2020a; Stucky & Ottensmann, 2009).…”