Despite a large number of studies testing Broken Windows Theory (BWT), the reverse theoretical pathway has never been assessed longitudinally (risk perceptions -> incivilities perceptions). It is estimated here using panel data from Baltimore, Maryland. Results show lagged, multilevel impacts of risk perceptions on shifting incivilities perceptions. Furthermore, impacts of risk perceptions on later shifts in perceived incivilities vary significantly across streetblocks. Findings support Harcourt’s assertion that "disorder" is not a fixed and unambiguous label; rather, it is dependent on a person defining his or her surroundings. People who report a high degree of crime risk are "biased" toward defining neighborhood features as more problematic than those who do not.