(WITHDRAWN) When Fear of Victimization Is High: Constrained Behaviors and Their Implications on Health and Social Inequities

Background/Purpose: Fear of victimization, along with increasing violence, is a significant social problem with corrosive effects on physical and behavioral health. Four decades of studies have contributed to the conceptualization and measurement of fear of crime and its various sociodemographic predictors. However, most of these studies have focused on categorical differences in behavior modification based on perceived risk between males and females or across racial/ethnic backgrounds (Schafer et. al., 2006; Callanan & Rosenberger, 2015). Little is known regarding potential heterogeneity that exists among individuals who express fear of victimization which has implications for tailoring social work interventions. Patterns of behavioral responses to perceived risk of victimization are also less understood, especially in areas with climbing crime rates like Mexico, where interpersonal crimes continue to rise. The purpose of this study is to examine patterns of constrained behaviors due to fear of victimization among a representative sample of adults in Mexico and how these behaviors are clustered— which has important implications toward designing family-centered interventions that address violence-related health and social outcomes.

Methods: Data were drawn from the National Survey on Victimization and Perception of Public Security conducted in Mexico in 2017 (N=102,129). Descriptive analysis was conducted to examine sociodemographic characteristics of individuals who are likely to change behaviors due to perceived risk of victimization. Latent Class Analysis (LCA) was run with 15 indicators of fear of victimization,as indicated by behavior change, related to travel (e.g., taking public transportation); financial (e.g., carrying debit card; bringing cash); interpersonal (e.g., visiting family/friends; permitting minors to go out); and recreational (e.g., going for walks; going to the movies) factors. Subsequently, we tested for differences across the resulting latent classes in demographic and social characteristics and whether multilevel factors (e.g., neighborhood characteristics and institutional responses) is associated with class membership.

Results:  Results show fit statistics that indicate a 2-class model best fits the data. Class one includes individuals (“Protective; n=56,178) with none to very low changes in behaviors related to individual safety, but have the highest probability in no longer permitting minors to go out alone; Class two (Self-Protective”; n=35,085) includes individuals who have the highest probability of constraining behaviors related to their own safety (e.g., stopped bringing cash, wearing jewelry, going for walks, and visiting friends and family). Findings from group difference tests provide insights regarding the distribution of gender, sociodemographic, neighborhood and state-level characteristics between latent classes, and highlight variations in patterns of behavior change and health outcomes across four Mexican geographic regions (U.S.-Mexico border states; southern border states near Guatemala/Belize; interior states; and states on the Pacific).

Conclusion/Implications:  LCA results, a person-centered approach, reveal important evidence regarding heterogeneity that is sometimes masked in variable-centered analyses. Findings reveal a sub-group of individuals who are protective of others, especially towards minors who are highly vulnerable to victimization. These individuals, most likely parents and other family members, are potential protective resources for children and youth. Implications regarding family-centered social work interventions that address social and health outcomes related to violence and victimization will be discussed.