NEIGHBORHOOD EFFECTS ON CHILD MALTREATMENT AMONG IMMIGRANT FAMILIES By Jisuk Seon A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Social Work ÑDoctor of Philosophy 2019 ABSTRACT NEIGHBORHOOD EFFECTS ON CHILD MALTREATMENT AMONG IMMIGRANT FAMILIES By Jisuk Seon Child maltreatment is a significant public health concern . Previous research on factors that influence child maltreatment has focus ed largely on child and familial characteristics. Emerging research on the etiology of child maltreatment has been examining the way in which neighborhood affects child maltreatment, such as neighborhood structural characteristics and neighborhood social processes . One of the major gap s in the research on the impact of neighborhood on child abuse and neglect is the examination of immigrant families. Following social disorganiza tion theory, this study explored the relationship s between neighborhood structural characteristics, neighborhood social processes , and child mal treatment among immigrant families. This study used the Fragile Families and Child Wellbeing Study Wave 1 and Wave 4, and examined a total of 372 foreign -born mothers and their children living in 325 neighborhoods in 20 large U.S. cities. A m ultilevel struct ural equation model demonstrated that foreign -born mothers living in neighborhoods with higher levels of negative structural characteristics reported higher levels of physical assault, higher psychological aggression, and higher neglect. Conversely, foreign -born mothers having higher levels of positive social processes reported lower levels o f physical assault, lower psychological aggression, and lower neglect. However, neighborhood social proce sses mediate d the relationships between neighborhood structural characteristics and physical assault and psychological aggression among immigrant famil ies, respectively. These results highlight the positive role of neighborhood social processes in reducing child maltreatment among immigrant families , even in disadvantaged neighborhoods. They also suggest that ecological multitier social work interventions that consider immigrantsÕ cultural context s may deserve attention as possibly effective in reducing maltreatment of children in immigrant families. Copyright by JISUK SEON 2019 !!!!v TABLE OF CONTENTS LIST OF TABLES .........................................................................................................................v LIST OF FIGURES ......................................................................................................................vii !CHAPTER 1: INTRODUCTION ...................................................................................................1 Significa nce of the Study ..........................................................................................................3 Organization of the Dissertation ...............................................................................................4 CHAPTER 2: LITERATURE REVIEW ........................................................................................6 Theoretical Background ............................................................................................................6 Ecological framework .........................................................................................................7 Social disorganization theory ..............................................................................................9 Prevalence of Child Maltreatment Among I mmigrant Families .............................................15 Neighborhood Effects on Child Maltreatment ........................................................................18 Neighbor hood structural characteri stics ...........................................................................18 Neighborhood social processes .........................................................................................27 Neighborhood and Child Maltreatment Among Immigrant Families ....................................31 Research Questions and Hypotheses ......................................................................................34 CHAPTER 3: METHOD ..............................................................................................................37 Data and Study Sample ...........................................................................................................37 Measures .................................................................................................................................38 Child maltreatment ...........................................................................................................38 Neighborhood structural characteristics ...........................................................................40 Neighborhood social processes .........................................................................................41 Control variables ...............................................................................................................43 Data Analyses .........................................................................................................................47 Descriptive analyses ..........................................................................................................47 Confirmatory factor analyses ............................................................................................47 Multilevel structural equation modeling ...........................................................................50 CHAPTER 4: RESULTS ..............................................................................................................55 Descriptive S tatistics ...............................................................................................................55 Confirmatory Factor Analyses ................................................................................................59 Multilevel Structural Equation Modeling ...............................................................................63 CHAPTER 5: DISCUSSION ........................................................................................................70 Evaluation and Interpretation of Research Hypotheses ..........................................................71 Limitations ..............................................................................................................................79 Implications .............................................................................................................................80 Social work practice and policy ........................................................................................80 Social work research .........................................................................................................83 !!!!vi REFERENCES .............................................................................................................................87 !!!!vii LIST OF TABLES Table 1. Study Variables, Types, and Data Sources .....................................................................46 Table 2. Variable Summary of Confirmatory Factor Analyses ....................................................49 Table 3. Descriptive Statistics of Study S ample ...........................................................................57 Table 4. Confirmatory Factor Analysis for Neighborhood Structural Characte ristics .................59 Table 5. Multilevel Confirmatory Factor Analysis for Neighborhood Social Processes .............60 Table 6. Multilevel Confirmatory Factor Analysis for Child Maltreatment .................................62 Table 7. Intraclass Correlations for Three T ypes of Child Maltreatment .....................................63 Table 8. Results from Mediation Model in M ultilev el Structural Equation M odeling ................68 !!!!viii LIST OF FIGURES Figure 1. BronfenbrennerÕs Ecologi cal F ramework .......................................................................8 Figure 2. Social Disorganization Theory ......................................................................................10 Figure 3. Proposed Study Model ..................................................................................................36 Figure 4. Proposed M ultilevel Structural Equation Model ...........................................................54 Figure 5. Multilevel Structural Equa tion Model ...........................................................................65 Figure 6. Results from Mediation Model in Multilevel Structural Equation M odeling ...............69 !!!!1 CHAPTER 1: IN TRODUCTION Child maltreatment is a widespread social problem in the United States. According to the most recent administrative child welfare data, 676 ,000 cases of maltreatment were substantiated around the country in 2016 (U.S. Department of Health and Human Services [USDHHS] , 2018). This means that approximately 1,852 children are abused or neglected each day in the United States. However, this number accounts only for maltreatment that is investigated and substantiated by child protecti ve services (CPS) auth orities. Responding to evidence of the large number s of maltreated children in the United States and elsewhere, researchers have invested substantial effort s into assessing pathways to child maltreatment. Common theories regarding the etiology of child maltreatment have evolved from an initial focus on clinically defined pathological child -parent relationships (Kempe, Silverman, Steele, Droegemueller, & Silver , 1962) to, more recently, an ecological framework of child maltreatment that highlight s the roles of environment characteristics and interactions among children, families, neighborhoods , and larger environmental contexts (Belsky , 1980). Over the past 40 years, child maltreatment scholars have directed considerable attention to the neighborhood as an extra familial so urce of child maltreatment and child welfare intervention (Coulton, Crampton, Irwin, Spilsbu ry, & Korbin, 2 007; Freisthler, Merr itt, & LaScala, 2006). According to major tenets of social disorganization theory, neighborhood characteristics associated with child maltreatment include structural characteristics in defined geographic areas (e.g., neighborhood rates of poverty and unemployment ) and social processes that refer to interactions between residents within the neighborhoods (e.g., social support and social network ) (Sampson & Grove, 1989 ; Shaw & McKay, 1942) . However , research investigating the relationship between neig hborhoods and child !!!!2 maltreatment ha s predominantly examined neighborhood structural characteristics to date. A key factor to understanding why neighborhoods that are similar in socioeconomic status have different rates of child maltreatment is likely to lie in the neighborhood sÕ social processes (Deccio, Horner, & Wilson, 1994; Freisthler & Maguire -Jack, 2015 ; Korbin, Coulton, Chard, Platt-Houston, & Su, 1998; McLeigh, McDonell, & Lavenda, 2018; Molnar et al., 2016 ). Neighborhood s ocial processes beyond s tructural characteristics may provide protective benefits against neighborhood residentsÕ maltrea ting behaviors. However, in the small amount of research on neighborhood social processes, research on the relationship between social processes and child maltreatment that also considers structural characteristics is limited. The focus of this study , therefore, is to simultaneously investigate both neighborhood s tructural characteristics and social processes to illustrate the possible pathways by which neighborhood structural characteristics affect social processes and these , in turn , influence child maltreatment. Research demonstrating the impacts of neighborho od on child maltreatment among immigrant families is limited. Studies on the roles of neighborhood in child maltreatment have been conducted primarily for nonimmigrant families, despite the significant number of immigrant population s in the United States. Generalizing the findings from nonimmigrants may not be appropriate for immigrants because they do not take into account the specific characteristics of immigrant families , such as acculturation experienced during migration and settlement (LeBrun, Hassan, Boivin, Fraser, & Dufour, 2015). While a few studies have examined how neighborhood affect s maltreatment of children in immigrant families (Dettlaff & Earner, 2012; Dettlaff, Earner, & Phillips, 2009 ; Tajima, & Harachi, 2010) , research simultaneously examining both neighborhood structural characteristics and social processes for child maltreatment among immigrant families is nonexistent , to the best of this authorÕs !!!!3 knowledge . Therefore, the mechanisms by which neighborhood str uctural characteristics and social processes influence maltreatment of children in immigrant families are currently unknown. !Based on the background, t his study aims to examine the following. 1. The relationship between neighborhood structural characteristics and child maltreatment among immigrant families . 2. The relationship between neighborhood social processes and child maltreatment among immigrant families. 3. The mediating role of neighborhood social processes in the relationship between neighborhood structural characteristics and maltreatment of children in immigrant families . Specific research questions will appear after the review of the theoretical background and literature. Significance of the Study This study will examine how neighborhood structural characteristics and social processes influence child maltreatment among immigrant families. This study is significant in several ways. First, although research has examined the contexts of neighborhoods a ssociated with child maltreatment over 40 years, it has focused mainly on neighborhood structural characteristics and less is known about social processes. This study will focus on the potential protective role of social processes in child maltreatment by examining its mediation effect on the relationship between neighborhood structural characteristics and maltreatment in immigrant families. Second, research describing the roles of neighborhoods in child maltreatment often has been limited to either neighbo rhood structural characteristics or social processes at one level without considering reciprocal influence s from both, which could be misestimated if they are not includ ed in the !!!!4 same analytic model. This study will simultaneously analyze both neighborhood structural characteristics and social processes , using multilevel models to better understand comprehensive pathways to child maltreatment among immigrant families. Third, despite significant work demonstrating what causes child maltreatment, the majority of these studies have examined children and families from the general population. Less attention has been paid to immigrant families , partly because information about the immigrant status of children often is unavailable in child welfare system s. As a res ult, the majority of studies using the official records of CPS often have failed to detect maltreatment of children in immigrant families. This study will build on the limited studies on maltreatment of children in immigrant families by examining foreign -born mothers and their children in 20 large U.S. cities , focusing on the roles of neighborhood contexts. Last, although social disorganization theory has been tested widely to figure out differences in child maltreatment between communities, it has primarily targeted the general population and rarely been applied to immigrant families. This study will exten d social disorganization theory to immigrant families by examining the neighborhood social processes that might , through the effects of neighborh ood structural characteristics on child maltreatment , serve as a mediator in the ways suggested by the theory. Organization of the D issertation This chapter sketched the background and s ignificance of the study. The next chapter will describe the theories that guide this study: the ecological framework and social disorganization theory. It will also provide the prevalence of child maltreatment among immigrant families, including a review of empirical literature that summarizes the ecological correlates of child abuse and neglect, focusing on the neighborhood context s that increase or decrease child maltreatment. Additional ly, the research questions will be out lined. The study design, methods, !!!!5 and data analysis techniques will be detailed in chapter three , and the results will be reported in chapter four . Finally , chapter five will discuss interpreta tion of t he studyÕs results, its limitations, and the implication s for social work practice, policy, and research. !!!!6 CHAPTER 2: LITERATURE REVIEW The aims of this chapter are four fold: (1) explore theories used to guide this study; (2) review the empirical literature on neighborhood contexts affecting child maltreatment , including structural characteristics and neighborhood social process es; (3) identify literature on neighbor hood and child maltreatment among immigrant families ; and (4) specify research questions and hypotheses. This study adopts the definition of child maltreatment given in the reauthorized Child Abuse Prevention and Treatment Act (CAPTA, P.L. 111 -320) of 2010 , which described a minimum set of acts or behaviors that define d child abuse and neglect . According to CAPTA, child maltreatment is: At a minimum, any recent act or failure to act of a partner or caretaker, which results in death, serious physical or emotional harm, sex ual abuse or exploitation, or an act or failure to act which present an imminent risk of serious harm to child less than 18 years old (USDHHS, 2010, p. 6) . This study includes data regarding four forms of child maltreatment: physical abuse , psychologica l abuse, sexual abuse, and neglect. Theoretical Background While the majority of research on the etiology of child maltreatment has focuse d on the psychosocial characteristics of parents and children, a growing number of studies suggest that characteristics of the environment in which these families live may also contribute to child maltreatment. Beginning with BronfenbrennerÕs ecological fra mework, which acknowledges that both individual and environmental factors influence child development, this study will introduce social disorganization theory that specifies how and why neighborhoods influence child !!!!7 maltreatment. Ecological framework. Fou ndational knowledge of BronfenbrennerÕs ecological framework is necessary to understand the neighborhood -level study of child maltreatment , since BronfenbrennerÕs focus extend ed the etiology of child maltreatment from the caregiver -child dyad to the interp lay among child, parent , and neighborhood , highlighting the impact of larger environments on an individualÕs behavior. Bronfenbrenner (1979) identified four system levels of environment s affect ing child devel opment: microsystem, mesosystem, exosystem, and macrosystem. As shown in Figure 1, this conceptual framework consists of several concentric circles . The innermost circle represents individual characteristics; the outer circle represents environmental characteristics. All systems, at various levels, can exercise varying degrees of influence on an individualÕs behavior (Bronfenbrenner, 1979) . The microsystem , that system in which children daily participate, includes their immediate relationships wit h parents, family members , and school . For example, interactions within the family may influence how parents deal with childrenÕs behaviors. The mesosystem is defined as the interaction s between microsystems. For parents, this may include interactions with co-parents and co -workers, while for children, relationships with other children and peer groups may influence their functioning. The exosystem includes people and places with which child ren may seldom interact , but wh ich are still influential. Such influences include parentsÕ workplaces, government , and neighborhood. For example, if parents are laid off from work, children may experience negative effects if the parents are, as a result, unable to afford food . Moreover, Belsky (1980) noted that lack of formal or informal support in neighborhoods during times of family crisis may increase parenting stress and the possibility of parents committing ch ild maltreatment . The macrosystem , the last and most remote system of the ecological framework , !!!!8 includ es social value s, culture, and economy, as well as broader polic ies. For example, social norms that accept corporal punishment may influence parentsÕ use of excessive force when disciplining children. Figure 1. BronfenbrennerÕs Ecological Framework Garbarino (1976) was the first to apply BronfenbrennerÕs ecological framework to a study of child maltreatment, focusing on the role of the exosystem . He investigated child maltreatment rates in 62 New York counties and found that lack of economi c and educational resources was strongly related to higher rates of child maltreatment. In this study, Garbarino (1976) highlighted the complex etiology of chi ld maltreatment Ña confluence of the microsystem, mesosystem, exosystem, and macrosystem . Since BelskyÕs (1980) article framing maltreatment as the result of challenges at multiple levels of the social ecology, studies on child maltreatment have advanced to include aspects of the child, parent, family unit, and neighborhood that !!!!9 researchers believed contributed to child maltreatment. Since that time, child maltreatment research has examined neighborhood contexts in the ecological framework for 40 years. Although the ecological framework considers the larger environmental contexts of child development and suggests ways that the environment may influence the interactions between children and parents and between famil ies and neighborhoods, it does not explain h ow and why neighborhoods specifically influence these interactions , and how and why the se neighborhood conditions occur. Despite this limitation, the ecological framework guided development of this studyÕs research questions and literature review, and will affect analysis and interpretation of the results , because it provides a basis for examining a variety of contexts at multiple ecological systems , such as children, parents, and neighborhoods. Specifically, this study will focus on the role of nei ghborhoods (exosystem) in relationships between children and parents (microsystem) among immigrant families , considering their cultural contexts (macrosystem) such as acculturation. Social disorganization theory. The majority of neighborhood studies of c hild maltreatment depend on social disorganization theory as a theoretical foundation. A sociological theory was first developed to explicate why certain communities are more vulnerable to crime or delinquency (Shaw & McKay, 1942) . It posits that Òsocially disorganizedÓ communities lack the structure to maintain social controls that allow residents to realize shared values. Specifically, neighborhood residents who share common goals may be more likely to work together to achieve those goals. However, if the re is disagreement on neighborhood goals, social disorganization will arise. Shaw and McKay (1942) argued that three community -level factors lead to a socially disorganized community: concentrated poverty, residential instability, and ethnic heterogeneity . These, in turn , escalate crime and delinquency. Sampson and Grove (1989) added new concepts !!!!10 by testing intervening mechanisms (mediating variables) between the three traditional factors affecting social disorganization and crime rates. They found that the level of social disorganization is influenced by three factors: collective efficacy, support networks, and community participation (Sampson & Grove, 1989) . These three constructs help neighborhood residents establish common goals even in culturally or soc ioeconomically heterogeneous neighborhoods (Sampson, 2001) . Figure 2. Social Disorganization Theory Research on neighborhood and child maltreatment has applied this theory to understand community variations in child maltreatment (Coulton et al., 2007). An e arlier focus was to examine the effects of neighborhood structural characteristics. Researchers hypothesized that socially disorganized neighborhoods put parents at risk for child maltreatment due to the additional stressors experienced in disorganiz ed neighborhoods and the lack of social norms that could provide a supportive environment for positive parenting. Findings for structural characteristics generally have suggest ed that higher level s of poverty (Coulton, Korbin, Su, & !!!!11 Chow, 1995; Ernst, 2001; Freisthler, 2004 ; Merritt, 2009 ), residential instability (Coulton et al., 1995; Freisthler, Bruce, & Needell, 2007; Korbin et al., 1998), unemployment (Deccio et al., 1994; Freisthler, 2004; Zuravin, 1989 ), child care burden (Coulton, Korbin, & Su, 1999; Klein, 2011; Klein & Merritt , 2014; Merritt, 2009 ), alcohol and drug availability (Freisthler et al., 2007; Freisthler, Gruenewald, Remer, Lery, & Needell, 2007 ; Freisthler, Needell, & Gruenewald, 2005; Morton, 2013 ), and lack of social service availability (Freisthler, 2013; Klein, 2011, Maguire -Jack & Klein, 2015; Maguire -Jack & Negash, 2016; Morton, 2013; Morton, Simmel, & Peterson, 2014; Negash & Maguire -Jack, 2015) increase neighborhood rates of child maltreatment, thus providing support for social disorganization theory. However, findings for ethnic heterogeneity are mixed (Ernst, 2000; Freisthler, Midanik, & Gruenewald , 2004; Klein, 2011; Klein & Merrit t, 2014; Lery, 2009; Molnar et al, 2003 ; Morton, 2013 ). Recent ly, researchers have expanded the focus of neighborhood studies of child maltreatment to include the effects of process -oriented neighborhood contexts to elucidate the pathways to child maltreatment (Coulton et al., 2007; Freisthler et al., 2006). Researchers have hypothesized that social processes promote the willingness of neighborhood residents to effectively organize and intervene for common goals (e.g., child well -being), which , in turn , create s a supportive environment that reduces parental stress and encourages positive parenting (Cao & Maguire -Jack, 2016; Freisthler & Maguire -Jack, 2015; Guterman, Lee, Taylor, & Rathouz, 2009; Maguire -Jack & Wang, 2016 ). Below is a discuss ion outlining the three constructs of neighborhood social processes: collective efficacy, social network, and community participation. Collective efficacy. Collective efficacy is the ability of neighborhood residents to realize common goals and provide social control over behaviors (Sampson & Grove, 1989) . NeighborsÕ !!!!12 willingness to exert social control over undesirable actions prevents people from committing these behaviors. It also can establish a norm for the neighborhood, such that residents intervene when they see undesirable behaviors (Sampson, Rauden bush , & Earls, 1997). Collective efficacy also has been conceptualized as individualsÕ social ties to a neighborhood. Some scholars posit that these social ties promote shared expectation s for desirable behaviors and willingness to intervene in the neighb orhood (Sampson et al., 1997 ). Sampson et al. (1997) interviewed 343 residents of Chicago to examine the relationship between collective efficacy and violence. They found that neighborhood poverty, immigrant concentration, and residential instability predicted lower levels of collective efficacy. They also found that collective efficac y played a mediating role in the relationships between poverty and residential instability for vari ous types of violence (Sampson et al., 1997) . In a related study that analyzed the responses of 8,782 Chicago residents, S ampson, Morenoff, and Earls (1999) found that neighborhoods with concentrated poverty predicted lower levels of collective efficacy for children . It was predicted that residents of a neighborhood with higher levels of collective efficacy, or a stronger willingness to intervene to promote a common goal (i.e., child well -being), would be more likely report abusive or neglectful behavior toward children (Sampson, Morenoff, & Earls, 1999 ). Neighborhood residents may be more willing to support at -risk families who may be likely to engage in child maltreatment. Social network . Social disorganization theory ass umes that a strong social network prevent s crime and delinquency (Sampson & Grove, 1989) . Specifically, it predicts that the presence of a social network enables residents to identify strangers in the neighborhood , so that they can easily prevent strangers from committ ing crimes (Skogan, 1986) . This social network can be considere d as a resource for at -risk families; however, in stressful situations, families cut !!!!13 off from support systems may resort to inappropriate parenting practices (Belsky, 1980) . Such inappropriate practices may include leaving children at home alone for a long time while looking for a job, if no nearby supporters are available to care for the children. In s uch an instance, a social network provide s parenting and emotional support, whil e isolation from the social network may hinder family functioning. Traditiona lly, r esidential instability and immigrant concentration are generally known to affect the strength of the social network. Residential instability decreases the level of friendship networks. Immigrant concentration also decreases friendship networks, and increases neighborhood crime (Sampson & Grove, 1989) . Earlier l iterature has suggest ed that the presence of social network s lowers child maltreatment. For example, Garbarino and Sherman (1980) interviewed 46 mothers and key informants from two impoverished neighborhoods with different levels of child maltreatment risk (high vs. low) and found that the high -risk neighborhood had small er numbers of social networks (i.e., the number of people who ha d an interest in child well -being). Community participation . Community participation enhances social ties and promotes social control of deviant behaviors by strengthening social cohesion (Shaw & McKay, 1942) . It is predicted that, if residents more actively participate in community activities and organizations, they will be more connected to available resources and build up social networks by developing a sense of community in the neighborho od. According to Sampson (1988), residential instability decreases the level of community participation and urbanization decreases community attachment. These factors suggest ways in which social networks are not established, which indicates the presence o f social disorganization. Community p articipation , which has been known to decrease child maltreatment, has only recently been examined in neighborhoo d studies of child maltreatment . (Cao & Maguire - !!!!14 Jack, 2016; Gracia & Musitu , 2003; Kim & Maguire -Jack, 2015). Despite this potential bu ffering role of neighborhood social processes, very few studies have examined the mediating effect of neighborhood social processes on the relationship between neighborhood structural characteristics and child malt reatment (Deccio et al., 1994; Fromm, 2004 ; Korbin et al., 1998; McLeigh et al., 2018). These studies have found that negative impacts of neighborhood poverty on child maltreatment were decreased in neighborhoods with positive social processes , such as social inte gration (Deccio et al., 1994) , social connectedness (Korbin et al., 1998) , intergenerational relationship s between adults and children (Fromm, 2004) , and social cohesion (McLeigh et al., 2018) . This finding has important implication s for social work practice and policy , because it implies that implementation of interventions to build positive social processes between neighborhood residents might decrease child maltreatment , even in impoverished neighborhoods. Not all parents living in neighborhoods with higher poverty rates maltreat their children. Living in a neighborhood with concentrated disadvantages might increase connections among residents due to a higher need for mutual support, which may , in turn , decrease parentsÕ maltreati ng behaviors. Conversely, living in a neighborhood with concentrated disadvantages might decrease these connections because of the higher level s of stress experienced by individual residents, which may , in turn , increase maltreating behaviors. These underl ying mechanisms of child maltreatment can be better understood by examining the role of social processes as a mediator in the relationship be tween neighborhood structural characteristics and child maltreatment. Therefore, t his study will analyze neighborh ood structural and social processes factors suggested by social disorganization theory . First, factors for neighborhood structural !!!!15 characteristics, drawn from census trac t-level information, will include neighborhood rates of poverty (e. g, percent age of families below poverty level) , residential instability (e.g., percent age of housing units vacant) , and racial/ ethnic heterogeneity (e.g., percent age of foreign -born residents ). Second, factors for neighborhood social processes , based on individual -leve l survey data, will include perceptions on collective efficacy (social cohesions and trust, informal social control), support network (e.g., having someone who can help with emergency child care), and community participation (e.g., participation in civic a ctivities/organization s). Last , this study will test if neighborhood social processes mediate the effects of neighborhood structural characteristics on child maltreatment among immigrant families in the manner proposed by social disorganization theory. Prevalence of Child Maltreatment Among I mmigrant Families According to the most recent data from the National Child Abuse and Neglect Data System (NCANDS), during federal fiscal year 2016 CPS officials investigated approximately 3.5 million child maltreatment referrals. They confirmed 676,000 victims of child abuse and neglect , with a rate of 9.1 victims per 1,000 children in the population, nearly three -quarters (74.8%) having suffered neglect, 18.2% physical abuse, 8.5% sexual abuse, and 6% psych ological abuse (USDHHS, 2018). These victim rates of child maltreatment differ by race and Hispanic origin. Non -Hispanic Black, American Indian/Alaska Native, and children of two or more races had higher rates of child maltreatment than other children. In 2016, the victim rates of child maltreatment for non -Hispanic Black and American Indian/Alaska Native children was 14 per 1,000 children, and 11 per 1,000 for non -Hispanic children of two or more race s. This compares with rates of 9 per 1,000 for Hispanic children, 8 per 1,000 for non -Hispanic White children, and 2 per 1,000 for non -Hispanic Asian children (USDHHS, 2018). !!!!16 While these estimates are considerable, substantiated reports of child abuse and neglect are prone to un derestimating the actual inciden ce of maltreatment , because not all child maltreatment cases come to the attention of CPS. Further, the number of these children who are living in immigrant families remains unknown, as these data are not collected uniformly by any national - or state -level child welfare reporting system (Dettlaff & Earner, 2012 ; Dettlaff et al., 2009), despite the se familiesÕ substantial portion of the overall U.S. population (13.7%, Census Bureau, 201 7b). Among the approximately 69.9 m illion children in the United States under age 18, 18.2 million Ñone out of every four Ñhave at least one parent who is foreign -born (Migration Policy Institute, 2019; U.S. Census Bureau, 201 7a) . Literature suggest s that immigration may be a risk factor for child maltreatment , given the stress it places on parents and on parent -child relationships (Dettlaff & Earner, 2012; Dettlaff & Finno -Velasquez, 2013; LeBrun et al., 2015; Pine & Drachman, 2005). Immigration experiences vary depending on country of origin, type of migration , and individual motivations; however, the decision to migrate is often driven by financial necessity or dangerous political climates that pose a risk of exposure to robbery, physical violence , and sexual assault (Hipsman, & Meissner, 2013) . Thus , the process of migration to the U nited States often is characterized by loss, trauma, and fear (Segal & Mayadas, 2005). Many difficulties that immigrants face Ñfinancial hardship , personal dissatisfactio n, depression, and social isolation Ñare known as factors a ssociated with child maltreatment risk (Dettlaff & Earner, 2012; Dettlaff et al., 2009; Dettlaff & Finno -Velasquez, 2013; Johnson, 2007; LeBrun et al., 2015 ). Paired with these challenges, d iscrepancies in the cultural values, language barriers, and behavior norms of less acculturated immigrant parents and their more acculturated children generate intergenerational conflict, which may also place immigrant families at higher risk of child maltreatment (Johnson, 2007) . However, the few studies that have !!!!17 examin ed child maltreatment among families that immigrate to the U nited States have provide d little support for this idea (Altschul & Lee, 2014; Dettlaff & Earner, 2012; Johnson -Motoyama, Dettlaff, & Finno, 2012 ), perhaps because they measure d only maltreatment reported by mandatory reporters , rather than actual behavior s regardless of the form of detection (Dettlaff & Finno -Velasquez, 2013). Estimates from the National Survey of Child and Adolescent Well -Being ( NSCAW) indicated no significant difference s in the overall rates of substantiated maltreatment between children of immigrants and children of U.S. -born parents (22.7% vs. 22.0%) (Dettlaff & Earner, 2010; Dettlaff & Earner, 2012 ). This suggests that children of immigrants to the U nited States are at similar risk of being maltreated as children of U.S. -born parents , or at least that they are underrepresented in the child welfare system . However, there were significant differences in the types of substantiated maltreatment. Specifically, children of immigrants were eight times less frequently physically abused than children of natives !(2.1% vs. 16.4%), while children of immigrants were more than twice as frequently emotionally abused than children of natives (25.1% vs. 11.1 %). Studies looking at particular race/ethnicit y reported differences in the types of child maltreatment between children of immigrants and children of natives. For example, Dettlaff et al. (2009) found from NSCAW that children of Latino immigrants reported to the chi ld welfare system were less likely than Latino children of native parents to be the subject of a confirmed physical neglect allegation (1.4% vs. 1 8.4%), although sexual abuse was the inverse ; Latino children of immigrants were nearly four times as likely to be the subject of a sexual abuse report than Latino children of native parents in the child welfare system (22.1% vs. 5.8%). Johnson -Motoyama et al. (2012) found from NSCAW II that Latino families with one foreign -born parent had significantly lower odds of substantiations of child maltreatment, compared to !!!!18 U.S. -born families. Altschul and Lee (2014) found from the Fragile Families and Child Wellbeing Study that foreign -born Hi spanic mothers , compared with U.S. -born Hispanic mothers , spank ed their young children less frequently . Despite the assumptions of higher risk of maltreatment of children in immigrant families, overall rates of substantiated maltreatment of children of im migrants were not significantly different from those of children with U.S. -born parent s, with some variations in the certain types of maltreatment. While the exact scope of child maltreatment among immigrant families in the United States remains un clear, the best available data indicate that it is hardly a rare phenomenon; rather, child abuse and neglect affect a significant number of children of immigrants . Neighborhood Effects on Child Maltreatment The majority of studies investigating child maltreatment ha ve focused on characteristics of children and parents that lead them to abuse or neglect children. In comparison, relatively little research has been conducted that examines the role of neighborhood in child maltreatment. The following discu ssion will review neighborhood factors that have been found to be associated with child maltreatment, such as neighborhood structural characteristics and social processes, and how they affect maltreatment among immigrant families. Neighborhood structural characteristics. To date, the majority of studies on neighborhood and child maltreatment have examined structural characteristics by analyzing demographic and socioeconomic features of communitiesÕ residents, such as neighborhood rates of poverty, residen tial i nstability , racial/ethnic heterogeneity , unemployment, child care burden, alcohol and drug availability, and social service availability (Coulton et al., 2007; Freisthler et al., 2006). Poverty. Literature on child maltreatment strongly suggests that concentrated poverty !!!!19 increases child maltreatment , despite differences in measures and study populations (Coulton et al, 1995; Coulton et al., 1999; Deccio et al., 1994; Drake & Pandey, 1996; Ernst, 2000, 2001; Freisthler, 2003; 2013; Freisthler et al ., 2005; Freisthler et al, 2007 ; Klein, 2011; Klein & Merritt, 2014; Korbin et al., 1998; Lery, 2 009; Molnar et al., 2016; Morton et al., 2014). Several different measurements for poverty were examined to assess its influence on child abuse and neglect. When loaded as a neighborhood -concentrated disadvantage factor in the analysi s models, poverty often included not only lack of money but also high rates of unemployment and high rates of families headed by females (Ben -Arieh, 2010; Coulton et al., 1999; Deccio et al., 1994; Gillham et al., 1998; Korbin et al., 1998; Weissman, Jogerst, & Dawson, 2003) , which may also indicate structural inequalities in neighborhoods affecting child maltreatment. For example, Coulton et al. (1995) used factor scores with lo adings of the percentage of families with incomes below the federal poverty level; the percentage of unemployed population; the percentage of vacant housing; the percentage of population loss; the percentage of African American residents; and the percentag e of female -headed households. Using the factor scores for concentrated poverty , they found that neighborhoods with higher poverty had higher child maltreatment rate s (Coulton et al., 1995) . Other studies employing similar approaches for impoverishment fac tor scores produced positive relationships between poverty and child maltreatment (Korbin et al., 1998; Lery, 2009; Morton et al., 2014) . A few studies investigated the effects of neighborhood poverty on a specific type of child maltreatment (Drake & Pandey, 1996; Ernst, 2000; Freisthler et al., 2 004; Morton et al., 2014) . While Drake and Pandey (1996) found a positive relationship between poverty and sexual abuse, Ernst (2000) found no relations hip. Additionally, while Morton et al. (2014) found pover ty as a significant predictor for both physical abuse and neglect, Freisthler et al. (2004) reported that !!!!20 poverty predicted physical abuse, but not neglect. One multilevel study by Coulton et al. (1999) reported the effect of impoverishment on caregiversÕ child abuse potential scores, indicating that even after controlling for family -level poverty, neighborhood poverty still remained positively associ ated with child abuse (Coulton et al., 1999). Residential instability. Residential instability refers to the degree that residents move in and out of the neighborhood. Stable neighborhoods are those where residents have lived for long periods of time. Findings have been mixed regarding the effects of residential instability on child maltreatment Ñthis may be due to the different types of measurements and race/ethnicit y of the study population s (Coulton et al., 1995; Deccio et al., 1994; Ernst, 2001; Freisthler et al., 2007; Klein & Merrit t, 2014; Korbin et al., 1998; Lery, 2009) . Measures of resident ial instability ranged from a single variable gauging the percentage of residents living in their current residence for less than five years (Klein & Merrit t, 2014) or less than 10 years (Lery, 2009) to a factor score, which incorporates measures for the percentage of residents living in the neighborhood for less than 10 years; the percentage that had moved in the last five years; and the percentage that had moved in the las t year (Coulton et al., 199 5; E rnst, 2001). While some studies found that r esidential instability increased child maltreatment (Coulton et al ., 1995, Deccio et al., 1994; E rnst, 2001; Lery, 2009) , others found no relationship (Freisthler, 2004; Freisthler et al., 2005; Klein, 2011; Merri tt, 2009; Molnar et al., 2003; Morton, 2013). Several studies reported that the effects of residential instability on child maltreatment vary , depending on rac ial/ethnic differences. For example, Korbin et al. (1999) found that residential instability was positively related to higher rates of child maltreatme nt only for European American (W hite) neighborhoods. Similarly, Freisthler et al. (2007) reported that residential instability inc reased child maltreatment for neighborhood s where the W hite !!!!21 population was concentrated, but it decreased maltreatment for neighborhood s with a concentrated Black population . Contrary to findings in the study by Freisthler et al. (2007), Klein and Merrit t (2014) found th at residen tial instability in neighborhood s with concentrated Whites decreased child maltreatment; they found no relationship in neighborhoods with concentrated Blacks . Racial /ethnic heterogeneity. Immigration concentration , as a proxy for race/ethnicity, is often considered as an indicator of undeveloped social support in neighborhoods, since families settling into a new count ry may struggle to obtain and build social support upon their arrival. This struggle could negatively impact their child rearing (Freisthler et al., 2006) . Also, it may influence child maltreatment , because different languages and cultures among neighborhood residents are likely to disrupt the deve lopment of social norms regarding appropriate behaviors (Klein & Merrit t, 2014) . Thus, it may be that when a neighborhood is predominantly composed of immigrants, child maltreatment is likely to be higher. Findings from several neighborhood studies of child maltreatment are mixed, due perhaps to different measures of immigrant concentr ation, type of child maltreatment, and rac ial/ethnic composition of the children (Ernst, 2000; Freisthler, 2004; Freisthler et al., 2004; Gracia, LŠpez -Qu™lez, Marco, & Lila, 2017; Klein, 2011; Klein & Merrit t, 2014; Lery, 2009; Molnar et al., 2003; Morton , 2013). For example, Ernst (2000) found that immigrant population decreases the rates of child sexual abuse by measuring immigrants as residents who had arrived in the United States in the past five years, and Molnar et al. (2003) found that immigrant con centration predicted lower rates of physical abuse using factor scores with l oading s of percentage of foreign -born residents and percentage of Hispanic residents . Klein (2011) examined immigrant population by measuring the U.S. -born population and found that ne ighborhoods where immigrants were concentrated had lower rates of both CPS !!!!22 referrals and substantiations. The negative relationship between these two variables looks different when categorizing by childrenÕs race/ethnicity. For example, Klein and Me rrit t (2014) found that immigrant concentration reduced CPS referrals only for White children, not for Hispanic or Black children. Conversely, Freisthler et al. (2004) found a positive relationship between immigrant concentration and child maltreatment, re porting that neighborhoods with higher concentration s of Hispanic residents (measured as a proxy for immigrant population s) had higher rates of physical abuse. Additionally, some studies found no relationship for first foster care entries (Lery, 2009) and CPS referrals (Morton, 2013) . Detalaff et al. (2009) and Klein and Merrit t (2014) speculated that lower rates of child maltreatment in neighborhoods with higher immigrant concentration s is possibly due to the development of positive social processes within the neighborhoods , such as social cohesion and social support , which may alleviate the risk for child maltreatment. However, how the actual neighborhood social processes influence chil d maltreatment among immigrant families has rarely been examined. Unemployme nt. In an analysis of ecological reviews of neighborhood studies on child maltreatment, Zuravin (1989) concluded that a higher percent age of unemployed residents in neighborhood s increased the risk of child maltreatment. Since then, it has been hypothesized that unemployment not only affects the lives of individual parents, it also changes the economic structure of neighborhoods, which , in turn , influences the larger environment f or families (Freisthler et al., 2006) . The l iterature suggests that higher rates of unemployed residents increase child maltreatment (Deccio et al., 1994; Freisthler, 2004; Freisthler et al., 2004; Molnar et al., 2003) . The majority of studies that have examined unemployment predictors have used a single yearÕs unemployment rates, while Molna r et al. (2003) measured unemployment rates over !!!!23 five years. Freisthler (2004) found a positive relationship between unemployment and substantiated child maltreatment rates in California, but Freisthler et al. (2007) found th at this positive relationship was observe d only for children living in neighborhoods with concentrated Hispanic population s, indicating that the effect of unemployment is differentially associated with child maltreatment by race/ethnicity. One study has suggest ed that neighborhood unemployment affects different types of child maltreatment. For example, Freisthler et al. (2004) reported that higher rates of unemployment significantly predicted child neglect, but not physical abuse. Additionally, a few studies have reported no significant relationship between neighborhood unemployment rates and substantiated child maltreat ment (Freisthler et al., 2005) and CPS referrals (Morton, 2013) . Child care burden . Child care burden refers to the resources and supervision available to care for children in neighborhoods (Coulton et al., 1995, Freisthler et al., 2006; Klein, 2011) . The literature indicates that child care burden in creases child maltreatment (Coulton, et al., 1999; Freisthler, 2004; Freisthler, 2013; Freisthler et al., 2005, 2007; Klein, 2011; Klein & Merri tt, 2014; Korbin et al., 1998; Lerry, 2009; Merri tt, 2009) . It ha s been measured as a score factor with a loading of ratio of children to adults; ratio of male to female; and percentage of elderly population older than 65 ; or examined these factors separately . Specifically, studies using factor scores to investigate the effects of child care burden on child maltreatment found that child care burden significantly predicted elevated child maltreatment (Coulton et al., 1995; Lery, 2009) . However, Korbin et al. (1998) report ed that child care burden was positively related to child maltreatmen t only for children living in neighborhood s with concentration s of European American ( White) residents, indicating that there are rac ial/ethnic difference s on this effect. !!!!24 Subsequent studies have suggest ed that this relationship also differ s by race/ethnicity , despite using a single measure of child care burden (i.e., the ratio to children to adults) . For example, Freisthle r et al. (2007) found a positive relationship between child care burden and child maltreatment, particularly for White children. Also, Klein and Merrit t (2014) reported that child care burden was positively related to CPS referrals in predominantly White and Hispanic neighborhoods. Ernst (2000) approached the effects of child care burden differently than other researchers , by measuring female labor force participation and focusing on femalesÕ role as primary caregiver s. While Ernst (2000) antici pated that this would produce a positive relationship to child maltreatment, because working women as primary caregiver s would feel a heightened child care burden, which would, in turn , increase child maltreatment . However, female labor force participation significantly predicted lower rates of physical and sexual abuse. This may be because working women have more social s upport and child care services available, which lessens the child care burden and therefore decreases child maltreatment (Young & Gately, 1988; Freisthler et al., 2006) . However, one multilevel study by Coulton et al. (1999) found that child care bur den at the neighborhood -level was still positively related to higher scores of child abuse potential, controlling for individual sÕ characteristics , such as level of social support. Alcohol and drug availability . The literature consistently has suggest ed that higher levels of alcohol availability increase child maltreatment , offering an understanding of the impact of parentsÕ substance abuse on child abuse and neglect (Freisthler, 2004; Feisthler et al., 2004; Feisthler et al., 2005; Freisthler & Maguire -Jack, 2015; Freisthler et al., 2007; Morton, 2013; Morton et al., 2014) . However, this positive relationship differs by measures of alcohol availability, types of child maltreatment, and race/ethnicity of the study sample. Studies !!!!25 measuring alcohol availa bility typically have use d two categories of alcohol outlets: First, on-premises outlets, or drinking venues where alcohol consumption occurs on site (e.g., bars or restaurants that serve alcohol), and second, off -premises outlets, drinking venues where the consumption of alcohol occurs away from the site (e.g., liquor stores ; Freisthler, 2004). Literature on this topic suggests that on -premises alcohol outlets are particularly associated with higher level s of child neglect (Freisthler et al., 2004; Morto n et al., 2013) . For example, Freisthle r et al. (2004) found a positive relationship between higher density of on -premises alcohol outlets and higher rates o f child neglect in California. Morton et al. (2014) found the same direction of the relationship in New Jersey. These findings indicate that when there are greater numbers of bars or restaurants in neighborhoods, parents may spend more time and money in these venues than at home, thus neglecting the basic necessities of their children. In the case of ne ighborhoods with higher density of off -premises alcohol outlets, higher rates of physical abuse were observed (Freisthler, 2004; Freisthler et al., 2005) , indicating that parents living in neighborhoods with great er numbers of liquor stores may purchase alcohol at the stores and drink at home, leading to a greater possibility of child physical abuse. However, Freisthler and Maguire -Jack (2015) found that neither density of off -premise nor density of on -premise alcoho l outlets w as associated with physical abuse. Moreover, the effects of neighborhood alcohol availability on child maltreatment var ied, depending on the race/ethnicity of the children (Freisthler et al., 2007) . For example, Freisthler et al. (2007) found th at higher density of off -premises alcohol outlets significantly predicted maltreatment for African American children, but not for Hispanic or Caucasian children . However, on -premises alcohol outlets did not predict maltreatment of children in any of these racial/ethnic groups. In terms of a relationship between neighborhood drug availability and child maltreatment, !!!!26 only a few studies have e xamined this relationship. They hypothesized that greater number s of drug possessions and drug sales in neighborhoods would predict an increase in child maltreatment (Frei sthler et al., 2005; Freisthler, Kepple, & Holmes, 2012). However, unlike the hypothesis, the findings of two studies differ by analytical approach an d type of maltreatment. For example, Freisthler et al. (2005) found that, in one northern California city, higher incidence of drug possession was positively related to higher rates of substantiated child maltreatment after controlling for neighborhood dem ographic variables. However, Freisthler et al. ( 2012) investigated the relationship between geography of drug market activities and child maltreatment and found that while current n eighborhood drug possessions were negatively related to rates of child malt reatment referrals, past -year local and spatially lagged drug sales were positively related to the rates of child maltreatment referrals. Regarding child maltreatment substantiatio ns, only current local sales were positively related to substantiated rates of child maltreatment, and only past -year sp atially adjacent possessions were negatively related to substantiation cases. For foster care entries, current local drug possess ions were positively related to rates of foster care entries, while past -year local possessions were negatively related to foster care entries. Social service availability . More recent neighborhood studies of child maltreatment have explor ed how social service availability influences child maltreatment. The researchers have hypothesize d that geographic proximity of social services facilitates family participation in child welfare services, and thus decreases the potential risk of child maltreatment (Freisthler, 2013; Klein, 2011) . Only a few studies have examined the relationship between social service availability and child maltreatment (Freisthler, 2013; Klein, 2011, Maguire -Jack & Klein, 2015; Maguire -Jack & Negash, 2016; Morton, 2013; Morton et al., 2014; Negash & Maguire -Jack, 2015). Findings for this topic differ by types of social services, types of child maltreatment, !!!!27 childrenÕs age s, and measures of geographic availability of social services. For example, Freisthler (2013) investigated the impacts of eight child welfare -related service locations for substance abuse, adoption, ment al health, housing, pregnant and parenting teens, special needs, youth independent living, and domestic violence on CPS referrals and foster care entries. Results indicated that only housing services decrease d both CPS referrals and foster care entries. Similarly, Negash and Maguire -Jack (2015) measured the aggregated eight types of social services, including domestic violence, mental health, alcohol and other drug abuse, housing, basic needs, childcare, child and parenting, and medical services, and found that greater availability of social services wa s related to lowe r levels of child abuse and ne glect . Klein (2011) researched location, capacity, and total number of early care and education resources , particularly for maltreatment of young children from bi rth to age five , and found that higher percentages of children attending preschool in local and adjacent neighborhoods were associated with lower rates of CPS referrals and maltreatment substantiations. Maguire -Jack and Klein (2015) focused on access to me ntal and substance abuse services and found that they decreased child neglect. Both Morton (2013) and Morton et al. (2014) examined the relationship between substance abuse service access and child maltreatment by measuring the distance from the center of each census tract to the nearest service facility, and found that neighborhood proximal accessibili ty of substance abuse services was negatively related to CPS referrals (Morton, 2013) and physical abuse (Morton et al., 2014). Neighborhood social processes. Compared to neighborhood structural characteristics, much less is known about whether neighborhood social processes play a positive or negative role in child maltreatment (Coulton et al., 2007 ; Freisthler et al., 2006 ). While it has been hypothe sized that higher social support, social network, collective efficacy , and community !!!!28 participation , as indicators of social processes , would decrease neighborhood child maltreatment, findings on this topic are mixed. Social support. Earlier neighborhood studies that compared two contrasting levels of child maltreatment ( high vs. low) indicate d that low -risk neighborhoods had more neighborhood support and resources than high -risk neighborhoods (Garbarino & Kostelny, 1992; Garbarino, & Sherman, 1980) . Subse quent studies looking at the role of social support for child maltreatment also found that greater social support significantly predicted lower child abuse potential scores (Korbin et al., 1999 , Merritt, 2009) and child neglect (Negash & Maguire -Jack, 2015 ). Although most of the studies on social support strongly highlighted its positive impacts on child maltreatment, Freisthler et al. (2014) focused on possible negative influences of social support on child maltreatment. They measured how three types of so cial support Ñtangible (i.e., money or babysitting), emotional, and social companionship (i.e., belongingness) Ñaffect child physical abuse. They found that social companionship was related to more frequent child physical abuse, suggesting that not all socia l support is beneficial in preventing child maltreatment (Freisthler, Holmes, & Wolf, 2014). Deccio et al. (1994) did not find evidence of such a relationship. Social network. Another indicator of social processes contributing to child maltreatment is the availability of neighborhood social network s, which are known to decrease child maltreatment. Only a handful of studies have examined the relationship between social network and child maltreatment (Vinson, Baldry, & Hargreaves, 1996; Molnar et al., 2003; Molnar et al., 2016). Vinson et al. (1996) compared two different neighborhoods, which were at low or high risk of child maltr eatment in Sydney, Australia, and found that neighborhoods with weaker social network s had higher rates of child abuse. Molna r et al. (2003) found a negative relationship between size of social network (i.e. , the average number of friends and relatives !!!!29 living in the neighborhood ) and self -reported parent -to-child physical aggression for Hispanic families in Chicago . Molnar et al. (2016) also investigated social network s in Chicago , measuring the total number of friends and relatives the subjects had living nearby and four types of child maltreatment : neglect, sexual abuse, physical abuse, and substance -exposed infants. They found that neighborhoods with more social networks had lower rates of all four types of child maltreatment after controlling for neighborhood structural factors. Collective efficacy . Collective efficacy is the glue that binds neighborhoods together (Sampson, Morenoff, & Gannon -Rowley, 2002 ), which has been conceptualized by two constructs: social cohesion , which refers to mutual trust among neighborhood residents, and social control, which refers to social norms regarding appropriate behaviors (Coulton et al., 1999; Guterman et al., 2009; Molnar et al., 2003; Sampson et al., 2002) . Researchers expected to find a negative relationship between collective efficacy and child maltreatment, suggesting that neighborhoods with higher levels of collective efficacy have less child maltreatment (Emery, Trung, & Wu, 2015; Emery et al., 2014; Freisthler & Maguire -Jack, 2015; Guterman et al., 2009; Kim & Maguire -Jack, 2015; Maguire -Jack & Showalter, 2016; Molnar et al., 2016) . However, results from each study differ by maltreatment type and focus of measures. For example, Guterman et al. (2015) found that psychological aggressi on and ph ysical assault toward children were higher among neighborhood residents who had lower perceptions of collective efficacy. Molnar et al. (2003) found that neighborhood collective efficacy significantly predicted lower rates of child neglect, physic al abuse, sexual abuse, and substance -exposed infants. There are studies focusing on sub -construct s of collective efficacy : social cohesion (Ma et al., 2018; McLeigh et al., 2018; Maguire -Jack & Showalter, 2016; Maguire -Jack & Wang, 2016) and informal so cial control (Freisthler & Maguire -Jack, 2015; Emery et al., 2014; Emery !!!!30 et al., 2015). First, four studies examined the relationship between social cohesion and child maltreatment. For example, Maguire -Jack and Showalter (2016) found that social cohesion was associated with lower child neglect, but not with lowe r corporal punishment and severe assault toward children. Maguire -Jack and Wang (2016) extended Maguire -Jack and Showalter (2016)Õs study by including social support and parental stress as mediators . They found that while social cohesion was not directly associated with neglect, social support was associated with higher level s of social cohesion, which, in turn, was associated with lower levels of parenting stress and , consequently, lower levels of n eglect. Ma et al. (2018) found that social cohesion was associated with lower odds of CPS involvement. McLeigh et al. (2018) reported that social cohesion mediated the relationship between neighborhood poverty and child abuse, but not the relationship between neighborhood poverty and neglect. Second, t hree studies examine d the particular role of informal social control for child maltreatment . Freisthler and Maguire -Jack (2015) measured informal social control, such as neighborhood re sidentsÕ willingness to intervene when undesirable behaviors happen in their neighborhoods, and found that higher level s of informal social control were related to less frequent child physical abuse. Emery et al. (2014) investigated informal social control for child maltreatment by asking how neighborhood residents react ed to child -related violent situations. They found a negative relationship between informal social control and very severe child physical abuse. Emery et al. (2015) used measures similar to those employed by Emery et al. (2014) for informal social control and found that informal social control was related to lower severe physical abuse and lower externalizing behavior problems. Kim and Maguire -Jack (2015) incl uded a motherÕs involvement in her community and informal social control to measure collective efficacy. They found that a moth erÕs involvement in her community was associated !!!!31 with lower levels of psychological aggression and that a more positive percepti on of informal social control was associated with lower levels of physical assault. Community participation. Community participation was measured by caregiversÕ attendance at social activities and parenting educational events within their communities . Gra cia and Musitu (2003) surveyed 670 non -abusive families and 166 abusive families in Spain and Colombia to examine the degree of family participation in their communities. They found that abusive parents who were suspected of physical neglect and psychologi cal abuse showed lower levels of participation in community social activities than non -abusive parents in both cultures (Gracia & Musitu, 2003) . Kim and Maguire -Jack (2015) analyzed 2,291 in -home interviews with mothers from the Fragile Families and Child Wellbeing Study. They sought to understand the interaction of parents and their environments by examining both perceptions of community a nd participation in community in relation to child maltreatment. They found that higher levels of community involvement , such as motherÕs attendance at educational events, were associated with lower levels of self -reported psychological abuse. They concluded that encouraging parentsÕ community involvement can improve the environment in which families and children develop and decrease child maltreatment (Kim & Maguire -Jack, 2015) . Cao and Maguire -Jack (2016) examined whether neighborhood processes and community participation relate to internal control, and whether these three variables are associated with child maltreatment. They found that greater community participation was associated with lower levels of physical assault and lower levels of neglect , via the pathway of mothersÕ internal control. Neighborhood and Child Maltreatment Among Immigrant Families As a major gap in the literature on neighborhood and child maltreatment, an examination of immigrant families in the U nited States is mostly missing. Although children in immigrant !!!!32 families ar e often at higher risk of maltreatment , stemming from the challenges experienced following immigration to the United States (Dettlaff & Earner, 2012; Dettlaff et al. , 2009; Dettlaff & Finno -Velasquez, 2013; Dettlaff, & Johnson, 2011; LeBrun et al., 2015), little is known about their presence in the child welfare system due to unavailable information in the CPS data . As a result, research looking at contributing factors , including neighborhood contexts, for child maltreat ment among immigrants is limited . The few studies investigating contributing factors for chil d maltreatment among immigrant families have primarily focused on risk factors at the family level that belong to the microsystem of the ecological framework. These risk factors include caregiverÕs older age (Dettlaff & Earner, 2012 ; Dettlaff et al., 2009 ; Osterling & Han, 2011 ), high er level of family stress (Johnson -Motoyama et al., 2012), living in a step -parenthood family or a single -parent family (Alink, Euser, IJzendoorn, & Bakermans !Kranenburg, 2013 ), living in a family with lower education and incom e level s (Alink et al. , 2013), unemployment status (Maiter, Stalker, & Alaggia, 2009; Rhee, Chang, Berthold, & Mar, 2012 ), presence of domestic violence (Dettlaff & Earner, 2012) , and parentsÕ history of child maltreatment (Tajima & Harachi, 2010 ). As an indicator of acculturation, shorter length of residence in the host country was associated with severe physical abuse (Rhee, Chang, Weaver, & Wong, 2008 , Tajima & Harachi ; 2010 ). It was speculated that immigrant families might be at higher risk of child maltreatment during the ir first years of settlement in the U nited States, due to their higher level of stress of acculturation because they could be less familiar with the laws, norms , and values regarding parenting in the host country (Rhee et al., 2012) . In addition, immigration status such as being undocumented or refugee was also related to increased child maltreatment (Alink et al., 2013; Euser et al., 2011 ). Compared with the risk factors, protective factors associated with child maltreatment !!!!33 among immigrant families are relatively lesser known and also limited to the family level , such as two -parent household (Johnson -Motoyama et al., 2012 ), motherÕs lower use of alcohol (Altschul & Lee, 2011) , and caregiversÕ higher education level s (Johnson -Motoyama et al., 2012). A protective factor for child maltreatment, which is particularly unique to immigrant families, is having a foreign -born mother , which was related to a lower likelihood of child maltreatment (Altschul & Lee, 2011; Putnam -Hor nstein & Needell, 2011 ). Research describing the contexts of neighborhoods in the exosystem of the ecological framework for child maltreatment among immigrants is scant. Within the small number of studies, one study found a lower level of social support as a risk factor for maltreatment of children in immigrant families (Dettlaff & Earner, 2012) . In terms of protective factors, two studies found that perceptions on neighborhood s were associated with child maltreatment among immigrant families (Dettlaff et al., 2009; Tajima, & Harachi, 2010 ). For example, Dettlaff et al. (2009) found from the NSCAW that Latino immigrants, compared to native -born Latino families, were more likely to describe their neighborhoods as safe with helpful neighbors, whi ch may be protective against maltreatment of La tino children of immigrants . However, this positive perception on neighborhoods (e.g., neighborhood support) was found to be protective for Cambodian immigrant families but not for Vietnamese families in a study using a longitudinal study by the Cross -Cultural Families Project in Washington State (Tajima, & Harachi, 2010 ). These two studies shed light on the relationship between perceptions on neighborhood social processes and child maltreatment , particularly in immigrant families, despite their different focuses , such as descriptive analysis of Latino children of immigrants in the child welfare system (Dettlaff et al., 2009) and the intergenerational transmission of violence among South Asian immigrants (Tajima, & Harachi, 2010 ). !!!!34 Although these studies provided significant findings for what influenc es child maltreatment in immigrant families, the majority of studies have focused on Hispanic families and have been limited to family -level factors (Altschul & Lee, 2011 ; Dettlaff & Earner, 2012 ; Johnson -Motoyama et al., 2012; Osterling & Han, 2011 ). This supports the need for further studies to examine more diverse racial/ethnic groups of immigrants and larger neighborhood contexts affecting child maltreatment among them. Research Questions and Hypotheses Based on the theoretical background and the gaps in the existing research , this study proposed three research questions and corresponding hypotheses. Figure 1 illustrates a proposed study model. Research Question 1 : What are the relationship s between neighborhood structural characteristics and child maltreatment among immigrant families? Hypothesis 1a : Negative neighborhood structural characteristics will increase child physical assault among immigrant families. Hypothesis 1b : Negative neighborhood structural characteristics will increase child psychological aggression among immigrant families. Hypothesi s 1c: Negative neighborhood structural characteristics will increase child neglect among immigrant families. Research Question 2 : What are the relationships between neighborhood social processes and child maltreatment among immigrant families? Hypothesis 2 a: Positive n eighborhood social processes will decrease child physical assault among immigrant families. Hypothesis 2 b: Positive n eighborhood social processes will decrease child psychological !!!!35 aggression among immigrant families. Hypothesi s 2 c: Positive n eighborhood social processes will decrease child neglect among immigrant families. Research Question 3 : Do neighborhood social processes mediate the relationship between neighborhood structural characteristics and maltreatment of children in immigrant fami lies? Hypothesis 3a : Neighborhood social processes will mediate the relationship between neighborhood structural characteristics and child physical assault among immigrant families . Hypothesis 3b : Neighborhood social processes will mediate the relationship between neighborhood structural characteristics and child psychological aggression among immigrant families . Hypothesis 3 c: Neighborhood processes will mediate the relationship between neighborhood structural characteristics and child neg lect among immigrant families . !!!!36 Figure 3 . Proposed Study M odel !!!!37 CHAPTER 3: METHOD Data and Study Sample This study used data from the Fragile Families and Child Wellbeing Study (FFCWS). Prior to developing this study, the Institutional Review Board at Michigan State University approved it as Òexempt research ,Ó meaning that as a study that uses secondary data in which respondents cannot be identified, this study was determined exempt from federal regulations governing the ethical conduct of rese arch on human subjects. Following the IRB approval, application for the contract data license that granted access to restricted neighborhood contextual variables (i.e., census trac t information) for this study was approved by the Bendheim -Thoman Center for Research on Child Wellbeing at Princeton University. The FFCSW is a longitudinal birth cohort study of 4,898 children born in 20 large U.S. cities between 1998 and 2000. FFCWS collects information by conducting interviews with both mothers and fathers an d taking in -home assessments at birth (baseline, Wave 1), and when children reach one (Wave 2), three (Wave 3), five (Wave 4), nine (Wave 5), and 15 (Wave 6) years of age. Rich information on respondentsÕ characteristics and experiences at the child, paren t, family, and neighborhood level s were obtained, which are pertinent to this study. To be specific, the FFCWS core interviews with parents consist of questions about parentsÕ nativity, relationships between biological mothers and fathers, parenting behavi ors, demographic characteristics, mental and physical health, economic and employment status, perceptions on neighborhood , and community participation. The FFCWS in -home assessments include information on childrenÕs cognitive and emotional development, hea lth, parentÕs maltreating behaviors, CPS contacts, and home environments. Additionally, census trac t information for mothersÕ and fathersÕ addresses at each wave is available, via a restricted -use data contract, such !!!!38 as demographics (e.g., percent age of di fferent racial/ethnic groups, foreign -born), employment (e.g., percent age of labor force [16+] who are unemployed), housing (e.g., percent age of housing vacant), and poverty (e.g., percent age of families below poverty level). This study used information from the baseline and Wave 4 (child age five ) FFCWS core interviews, in -home assessment s, and census trac t data. These subsets of data contain the most complete information on the variables for the research questions of interest , including neighborhood structural characteristics (census trac t information), perceptions on neighborhood social processes (core interview) and child maltreatment (in -home assessment). This study further limited the analysis to the sample of mothers , due to the high amount of missing data among fathers. The target sample included 2,957 mothers who responded to both core interview and in -home assessment at Wave 4 and for whom census trac t information was available. Therefore, the final sample for this stud y consisted of 372 immigrant mothers not born in the United States screened by a piece of information obtained at the baseline core interview: Òwhether the mother was born in the U.S. (0 = no, 1 = yes) (12.65 % out of 2,957 mothers at Wave 4 in FFCSW).Ó Am ong the 372 immigrant mothers, 70.3% w ere Hispanic, followed by non-Hispanic Black (12.2%), other (11.9 %), and non -Hispanic White (5.7%). Measures Child maltreatment. Child maltreatment was measured by three types of mothersÕ self -reported maltreating behaviors toward their children using 15 questions based on the Conflict Tactics Scale: Parent -Child Version (CTS -PC) (Straus, Hamby , Finkelh or, Moore, & Runyan, 1998): physical assault, psychological aggression, and neglect from the home assessment at Wave 4. Eight questions regarding severe physical maltreatment (e.g., grabbing a child around the neck and choking the child) from the original CTS -PC were excluded in the FFCSW. The !!!!39 FFCWS included five items for physical assault (i.e., Òspanked the child on the bottom with bare hand ,Ó Ò hit the child on the bottom with some hard object (belt, hairbrush, stick ),Ó Òslapped the child on the hand, arm, or leg ,Ó Òpinched the child ,Ó and Òshook the childÓ) ; five items for psychological aggression (i.e., Ò shout ed, yelled, screamed at the child, Ó Òthreatened to spank or hit but didnÕt actually do it ,Ó Òswore or cursed at the child ,Ó Òcalled the child dumb or lazy or some other name like that ,Ó and Òsaid you would send the child away or would kick out of the houseÓ) ; and five items for neglect (i.e., Òhad to leave the child home alone , even when you thought some adults should be with the child ,Ó Òw ere so caught up with your own problems that you were not able to show or tell the child that you loved the child ,Ó Òw ere not able to make sure the child got food when the child needed it,Ó Òw ere not able to make sure the child got to a doctor or hospital when the child needed it ,Ó and Òw ere so drunk or high that you had a problem taking care of the childÓ). Mothers were asked , ÒHow many times have you done the behavior in the past year?Ó A 7-point ordinal scale was used (0 = never happened , 1 = once , 2 = twice , 3 = 3Ð5 times , 4 = 6Ð10 times , 5 = 11Ð20 times , 6 = more than 20 times ), with additional one response category (7 [yes, but not in the past year ]; ! = .91). Based on Straus and colleagues Õ (1998) recommendations on scoring for CTS -PC, each item of the three types of maltreating behaviors Ñphysical assault, psychological aggression, and neglect Ñwas weighted with a mid -point value of each scale (e.g., 0 = never happened to 0, 1 = once and yes, but not in the past year to 1, twice = 2 to 2, 3 = 3Ð5 times to 4, 4 = 6 Ð10 times to 8, 5 = 11 Ð20 times to 15, 6 = more than 20 times to 25) to indicate frequencies of maltreating behaviors. The additional response category, Ò7 ( yes, but not in the past year ),Ó was recoded as 1 (once ) to capture the incident regardless of the time reference (Straus et al ., 1998). All 15 items for child maltreatment were continuous variables, ranging from 0 to 25. Higher scores indicated !!!!40 more frequent child maltreatment. Neighborhood structural characteristics. This study used census trac t-level information to measure stru ctural characteristics of neighborhoods where the mothers were residing at the time of Wave 4. Census trac ts, areas with a population ranging from 1,200 to 8,000 with an average of 4,000 residents and 1,500 households (U.S. Census Bureau, 200 0), are common ly used by social scientists as proxies for neighborhood boundaries (Hart & Waller, 2013). They reflect homogeneity of population characteristics, economic status, and physical living conditions (U.S. Census Bureau, 200 0), and closely overlap with resident sÕ perceived neighborhood boundaries (Kohen, Dahinten, Leventhal , & McInto sh, 2008) . The FFCSW geo -coded the respondentsÕ residential addresses in Wave 4 into 2000 U.S. Census trac ts and created pseudo census trac t identifiers (Bendheim -Thoman Center for Research on Child Wellbeing , 2018). Therefore, the unit of neighborhood for structural characteristics in this study is census trac t. Based on the census -tract information at Wave 4, this study used six neighborhood rates of structural characteristics. T hey included percent age of families below poverty level, percentage of housing units vacant, and percentage of foreign -born residents . These are major constructs of social disorganization (Shaw & McKay, 1942) , which, in turn, may increase child maltreatmen t. In addition, this study also included percentage of households on public assistance (e.g., Temporary Assistance to Needy Families and general assistance such as food stamps), percentage of labor force (16 +) who are unemployed, and percentage of families headed by females, which are known from previous studies to be indicators of neighborhood inequalities affecting child maltreatment (Coulton et al., 2007; Freisthler et al., 2006; Molnar et al., 2016) . These si x neighborhood structural characteristics were continuous variables, as provided !!!!41 in the FFCWS restricted data. Higher scores of structural characteristics indicate d more disorganized neighborhoods. Neighborhood social processes. This study used three com ponents of neighborhood social processes known to mediate the relationship between neighborhood structural characteristics and child maltreatment (Sampson & Grove, 1989 ) from the core interview s at Wave 4 : collective efficacy, social network, and community participation. Collective efficacy. Collective efficacy, adopted from two subscales in the Project on Human Development in Chicago Neighborhoods, measured two types of mothersÕ perceptions on the degree of collective efficacy in their neighborhoods (Sampson et al., 1997): informal social control and social cohesion and trust. The i nformal social c ontrol subscale asked the mothers about their perception s on neighborsÕ willingness to intervene in certain scenarios using five items (i.e., Òif children were skipping school and hanging out on the street ,Ó Òif children were spray painting buildings with graffiti ,Ó Òif children were showing disrespect to an adult ,Ó Òif a fight broke out in front of the house ,Ó and Òif the fire station closet to t he neighborhood was threatened and its budge t was cutÓ). A 4-point ordinal scale was used (1 = very likely , 2 = somewhat likely , 3 = not likely , 4 = very unlikely ; ! = .84). The five items of informal social control were reverse -coded, summed, and divided by the number of items to create a continuous variable, ranging from 1 to 4. Higher scores indicated higher levels of perceived informal social control. Social cohesion and trust was measured using five items indicating mothersÕ agreement with the five statements, such as Òpeople around here are willing to help their neighbors ,Ó Òthis is a close -knit neighborhood ,Ó Òpeople in this neighborhood generally do not get along with each othe r,Ó Òpeople in this neighborhood do not share the same values,Ó and Ògangs are a problem in !!!!42 this neighborhood ,Ó using a 4 -point ordinal scale (1 = strongly agree , 2 = agree , 3 = disagree , 4 = strongly disagree ; ! = .78). The first three items were reverse -coded. The five items of social cohesion and trust were summed, and divided by the number of items to create a continuous variable, ranging from 1 to 4. Higher scores indicated higher levels of perceived social control and cohesion. Social network. Three dichotomous questions measured mothersÕ supportive social network of family and friends. Mothers were asked, ÒCould you count on someone to help with emergency child care?Ó Ò Could you count on someone to loan you $ 200?Ó and Ò Is there someone you coul d count on to provide you with a place to live? Ó (0 = no, 1 = yes). The t hree items of social network were summed to create a continuous variable, ranging from 0 to 3. Higher scores indicated a greater social network. Community participation. Three dichotomous questions measured mothersÕ participation in community activities and organizations, and their attendance at educational events regarding child care in their neighborhoods. First, mothers were asked, ÒD o you participate in any groups (e.g., sen ior center, social or work group, church, charity, public service, or community group ; 0 = no, 1 = yes)?Ó Second, mothers were asked if they attended any workshop s, talks, or information sessions provided by child care settings in their neighborhoods (0 = no, 1 = yes). Responses from mothers living in neighborhoods having none of those events available (8%) were coded as 0. Last, mothers were asked if they participated in parenting group meetings using a 4-point ordinal scale (0 = never , 1 = rarely , 2 = som etimes, 3 = always ). Zero (never ) indicated no participation, and 1 ( sometimes ), 2 (rarely ), and 3 ( always ) were recoded as 1 to indicate any participation in parenting group meetings. T he t hree items of community participation were summed to create a continuous variable, ranging from 0 to 3. Higher scores indicated greater !!!!43 community participation. Control variables. This study included several control variables that have been found by previous studies to be associated with maltreatment of children in immigrant families (Dettlaff & Earner, 2012; Dettlaff et al., 2009; Dettlaff & Johnson, 2011; Lebrun et al., 2015; Lee & Altschul, 2015; Maiter et al., 2009; Rhee et al., 2012; Segal, 2000; Tajima & Harachi, 2010 ). MothersÕ demographic and socioeconomic characteristics. This study controlled for mothersÕ demographic and socioeconomic characteristics from the core interview s at baseline and Wave 4. The baseline core interview information included years when mothers first came to the United States as a categorical variable (1 = before 1970 , 2 = between 1970 and 1979 , 3 = between 1980 and 2000 ), and race/ethnicity as a categorical variable ( 1 = non-Hispanic White , 2 = non-Hispanic Black , 3 = Hispanic , 4 = other race/ethnicity ). In addition, information from the Wave 4 core interview included !mother Õs age as a continuous variable ; motherÕs marital status as a categorical variable ( 1 = married , 2 = unmarried ); motherÕs level of education status as a categorical variable (1 = less than high school , 2 = high school degree or GED , 3 = some college/technical school , 4 = college degree or higher ); and mother Õs employment status as a binary variable (0 = no, 1 = yes). Last , household income with 10% of missing data imputation by the FFCSW was included as a continuous variable. Annual household income was measured by asking , ÒThinking about your income and the income of everyone else who lives with you, what was your total house hold income before taxes in the past 12 month s?Ó Mothers were asked to answer with an exact amount , or a range if they could not provide the exact amount of household income. MothersÕ behavioral health . This study controlled for maternal parenting stress, depression, and alcohol use from the Wave 4 core interview. Maternal parenting stress was !!!!44 measured by four questions ( i.e., Òb eing a parent is ha rder than I thought it would be ,Ó ÒI feel trapped by my r esponsibilities as a parent ,Ó ÒI find that taking care of my child is much more work than pleasure,Ó and ÒI am often tired, wo rn out, exhausted from raising a familyÓ) , using the Parental Burden of Caregiving Scale, a subscale of the Parental Stress Index, on a 4 -point ordinal scale (1 = strongly agree , 2 = agree , 3 = dis agree , 4 = strongly dis agree ; ! = .80) (Abidin 1995 ). The four items were reverse -coded, summed, and divided by the number of items to create a continuous variable, ranging from 1 to 4. Higher scores indicated greater maternal parenting stress. ParticipantsÕ d epression was measured using the Composite International Diagnostic Interview -Short Form (CIDI -SF), Section A (Kessler, Andrew s, Mroczek, Ustun, & Wittchen, 1998). Mothers were asked one dichotomous stem question : ÒDuring the past 12 months, has there ever been a time when you felt sad, blue, or depressed for two or more weeks in a row (0 = no, 1 = yes)?Ó If they responded yes, they were asked following seven specific symptom questions (0 = no, 1= yes including frequencies (e.g., all day long, most of the day, about half of the day ) about (1) losing interest, (2) feeling tired, (3) change in weight, (4) trouble sleeping, (5) trouble concentrating, (6) feeling worthless, and (7) thinking about death . This study used a constructed maternal depression variable, which was created by the FFCSW. The FFCSW calculated whether mothers met conservative major depression criteria as a binar y variable, such as 0 = no major depression, 1 = major depression , following Walter and colleagues Õ (2002) guidelines on scoring the CIDI -SF. Specifically, mothers were categorized as having major depression when they had experienced feelings of depression and los s of interest in things that lasted Ò most of the day Ó over a two-week period during the past year (Walter, Kessler, Nelson, & Mroczek , 2002). !!!!45 Alcohol use was measured by asking the largest number of drinks the mothers had consumed in any single day during the past 12 months (ÒdrinkÓ means either a bottle of beer, a wine cooler, a glass of wine, a shot of liquor, or a mixed drink) on a 5 -point scale (0 = none, 1 = oneÐthree, 2 = 4Ð10, 3 = 11Ð20, 4 = more than 20) , using the CIDI -SF (Kessler et al., 1998). Alcohol use data were recoded to generate a binary variable of 0 = none and 1 = alcohol user (from one to more than 20). ChildrenÕs gender and health. This study controlled for childrenÕs gender (0 = male , 1 = female ) as a binary variable from mothersÕ core interview s at baseline. In addition, mothersÕ self -reports of childrenÕs health problems from the home assessment data at Wave 4 were also controlled , including the following 14 dichotomous health conditions (0 = no, 1 = yes): attention deficit disorder or attention deficit hyperactivity disorder, mental retardation or developmental delay, Down syndrome, cerebral palsy, sickle cell anemia, autism, congenital health disease or other heart condition, asthma, total bli ndness, partial blindness, total deafness, partial deafness, speech or language problem, and problems with limbs. The 14 items were summed and recoded to a binary variable of 0 = none and 1 = yes (at least one health problem ). Table 1 lists study variables, types of variables, and data sources. !!!!46 Table 1 . Study Variables, Types, and Data Sources Variable Type FFCWS Data Source Nativity Binary Core -interview at Baseline Dependent variables: Child maltreatment measures Maltreating behaviors Continuous Home -assessment at Wave 4 Physical assault Psychological aggression Neglect Independent variables: Neighborhood measures Neighborhood structural characteristics Continuous Wave 4 using 2000 U.S. Census track boundary Percent of families below poverty line Percent of housing units vacant Percent of foreign born Percent of household on public assistant Percent of unemployment Percent of families headed by females Neighborhood social processes Collective efficacy Continuous Core -interview at Wave 4 Social network Participation in community Control variables MotherÕs demographic and socioeconomic characteristics Years when first came to the U.S. Categorical Core -interview at Baseline Race/ethnicity Categorical Core -interview at Baseline Age Continuous Core -interview at Wave 4 Marital status Categorical Core -interview at Wave 4 Education level Categorical Core -interview at Wave 4 Household income Continuous Core -interview at Wave 4 Employment status Binary Core -interview at Wave 4 MotherÕs behavioral health Parenting stress Continuous Core -interview at Wave 4 Depression Binary Core -interview at Wave 4 Alcohol use Binary Core -interview at Wave 4 Child Õs characteristics Gender Binary Core -interview at Baseline Health Problems Binary Home -assessment at Wave 4 Note. FFCSWS=Fragile Families and Child Wellbeing Study. !!!!47 Data Analyses Descriptive analyse s. This study tested a set of pathways linking neighborhood structural characteristics, neighborhood social processes, and three types of child maltreatment, including physical assault, psychological aggression, and neglect. Descriptive analyses of all of the variables of this study, such as percent age , mean, standard deviation, and range, are presented. This study conducted confirmatory factor analysis and multilevel structural equation modeling (MSEM) using Mplus 7.0 (Muth”n & Muth”n, 2012) . This study used a two -step process, guided by Anderson and Gerbing (1988) . The first step was to assess the fit between all theorized latent constructs and their corresponding observed indicators using confirmatory factor analyses, and the second step was to test the hypothesized direct and indirect relationships between neighborhood structural characteristics, social processes, and child maltreatment using the latent variables by employing MSEM. This two -step process helped decreas e the possibility that the structure of latent variables could be the source of misfit when testing the structural model (Anderson & Gerbing, 1988) . Confirmatory factor analyses . As the first step of the MSEM , this study conducted confirmatory factor anal ysis (CFA) and multilevel confirmatory factor analysis (MCFA) to examine the model fit to the data of this study. To be specific, this study conducted CFA to confirm whether or not six observed indicators of neighborhood structural characteristics, includi ng percent age of families below poverty line , percent age of housing units vacant , percent age of foreign born , percent age of household s on public assistan ce, percent age of unemployment , and p ercent age of families headed by females , load ed on a single latent variable. In addition, this study conducted MCFA to explore the underlying structures among a set of items for neighborhood social processes and three types of child maltreatment to determine how !!!!48 those observed items fit theorized !measurement models. MCFA , compared to CFA, incorporates the hierarchical structure of data and permits assessment of the factor structure across the two levels (Muth”n , 1994). Given the multilevel nature of the FFCWS data in this study, with children and mothers ( Level 1) nested within neighborhoods ( Level 2), MCFA is an appropriate analytic approach to prevent against yielding biased parameter estimates (Muth”n , 1990, 1994). Therefore, this study conducted MCFA to confirm whether or not the four observed indicators of neighborh ood social processes, which are nested in neighborhood structural characteristics including informal social control, social cohesion and trust, social network, and community participation, can be reduced to a single latent variable for neighborhood social processes. Also, this study conducted MCFA to confirm how the five observed indicators of each of the three types of child maltreatment can be loaded to each single latent variable to represent three latent variables for each type of child maltreatment : physical assault, psychological aggression, and neglect. By conducting MCFA for neighborhood social processes and child maltreatment, nested data nature (e.g., mothers and children within neighborhoods) were properly accounted for. Table 2 lists observed ind icators, types of confirmatory factor analyses, and latent variables of this study. !!!!49 Table 2 . Variable Summary of Confirmatory Factor Analyses Observed Indicator Analysis Type Latent Variable Dependent variables: Child maltreatment measures >Ì Maltreating behaviors MCFA Physical assault Physical assault Psychological aggression Psychological aggression Neglect Neglect >ÌIndependent variables: Neighborhood measures Neighborhood structural characteristics CFA Neighborhood structural characteristics Percent of families below poverty line Percent of housing units vacant Percent of foreign born Percent of household on public assistant Percent of unemployment Percent of families headed by females Neighborhood social processes MCFA Neighborhood social processes Collective efficacy Informal social control Social cohesion and trust Social network Participation in community Note. MCFA=Multilevel Confirmatory Factor Analysis, CFA=Confirmatory Factor A nalysis. !!!!50 Missing data (0 ! 2.4%) were handled using full information maximum likelihood (FIML) estimation. This method offered less biased estimators by preserving all available data (Acock, 2005), rather than estimating missing values by replac ement or imputation (Arbuckle, 1996) . This means that, in FIML, missing values are not replaced or imputed, but the missing data are handled within the analysis model by estimating parameters that would most likely produce the estimates from the sample dat a. This study used a maximum likelihood with robust standard error estimator to adjust non -normality and non -independence of indicators due to the ne sted neighborhood census tracts (Muth”n & Muth”n , 2012). In order to reduce subjectivity of interpretation on the latent variables in CFA and MCFA, indicators with factor loadings greater than .40 were considered as Òsignificant Ó and used to define the latent variables (Ford, MacCallum, & Tait, 1986 ). The meas urement model fit for the final five latent variables, including neighborhood structural characteristics, neighborhood social processes, and the three types of child maltreatment (physical assault, psychological aggression, and neglect ) and their factor lo adings were examined using multiple criteria of model fit indices (Kline, 2005): (1) the p-value of chi -square statistics ( "2) should be larger than .05; (2) a comparative f it index (CFI) of .90 or higher; (3) a Tucker -Lewis index (TLI) of .90 or higher; (4) a standardized r oot mean square residual (SRMR) of less than .08, and (5 ) a root mean square error of approximation (RMSEA ) of less than .06 indicate an adequate model fit (Browne & Cudeck, 1993) . Multilevel structural equation modeling. Using the measurement models specified by CFA and MCFA, this study conducted MSEM to simultaneously test the research questions. MSEM has advantages over the traditional multilevel modeling and structural equation modeling in that it allows simul taneous examination of the direct and indirect relationships among latent variables while controlling for the nested data structure (Preacher, Zyphur, & Zhang, 2010 ). In !!!!51 all MSEM models in this study, between -neighborhoods variables (Level 2) included inde pendent variables (e.g., neighborhood structural characteristics). The within -neighborhoods (Level 1) variables included a mediator (e.g., neighborhood social processes), dependent variables (e.g., child physical assault, psychological aggression, and negl ect), and covariates (e.g., motherÕs demographics and behavioral health and childÕs gender and health ; i.e., 2 -1-1 design, meaning that independent variables were assessed at Level 2 and mediator and dependent variables were assessed at Level 1). Intracl ass correlation. Prior to conducting the MSEM to test the hypothesized models, this study examined the intraclass correlation (ICC) to confirm whether the multilevel model was an appropriate analytic approach for this study (Rau denbush & Bryk, 2002). ICC quantifies how much correlatedness is in a hierarchical data set by calculating a conventional one -way ANOVA (Raudenbush & Bryk, 2001). For instance, this study examine d how much variance in individual mothersÕ report s of child maltreatment (Level 1) stemmed from neighborhoods (Level 2). If the ICC is low, all of the variance in child maltreatment would be expected to lie among individual mothers. That is, child maltreatment would not be different across neighborhoods, and thus, a multilevel model wou ld not be regarded as appropriate. !This study reported whether the ICC coefficient was closer to 1 (higher statistical dependency) or closer to 0 (no statistical dependency) due to the fact that there is no widely accepted guideline for interpreting the coefficient (Raykov , 2011). Multilevel mediation model. After examining the ICC coefficient, this study confirmed model fit for the multilevel mediation model to simultaneously examine research questions 1, 2, and 3. Next, four parameters were tested fo r direct and indirect relationships between neighborhood structural characteristics, social processes , and child maltreatment: (1) the direct !!!!52 effect of neighborhood structural characteristics on child physical assault, psychological aggression, and neglect controlling for covariates ( path 3 ) (H1 -a, b, c); (2) the effect of neighborhood structural characteristics on neighborhood social processes controlling for covariates ( path 1 ); (3) the direct effect of neighborhood social processes on child physical assault, psychological aggression, and neglect controlling for covariates ( path 2 ) (H2-a, b, c ); and (4) the indirect effect of neighborhood structural characteristics on child physical assault, psycholog ical aggression, and neglect via neighborhood social processes ( path 1 ! 2) (H3 -a, b, c). This procedure followed recent work on mediation analysis (MacKinnon, 2008 ), which simultaneously tests direct path 2 and path 3, and the product of 1 and 2 quantifie s the indirect path 1 ! 2 through neighborhood social processes (Hayes, 2009 ), rather than following the conservative causal steps of Baron and Kenny (1986 ). Unlike Baron and KennyÕs approach for the mediation process using a series of standard regression method s, mediation analysis in SEM simplifies the mediation test in a single analysis. Also, it provides model fit information about the consistency of the theorized mediation model to the data, made when constructing the model. The standard regression pro cedure of mediation recommended by Baron and Kenny has been shown to be low powered (MacKinnon, 2008) . Finally, Bollen and Pearl (2013) noted that even when the same mediation equation is used in SEM and in regression analysis, the results will be differen t because they use completely different assumptions. Standard regression analyzes a relationship based on a conditional expected value, while SEM estimates a relationship via a theoretically constructed model, path diagrams, and mathematical equations (Bollen & P earl, 2013; MacKinnon, 2008) . Thus , this studyÕs hypothesized mediation process in the simultaneous nature of the direct and indirect effects can be more appropriately calculated using SEM than using regression procedure. !!!!53 This study reported di rect effects of path 2 and path 3 , and indirect effect s of path 1 ! 2 (the amount of mediation ), using standardized coefficients and a cutoff of p < .05 to determine the statistical significance of each hypothesized path. The standardized coefficients for the dependent variables , child physical assault, psychological aggression, and neglect, were considered as effect sizes (MacKinnon, 2008 ; MacKinnon, Fairchild, & Fritz, 2007). Figure 4 illustrates a proposed multilevel structural equation model. !!!!54 Note. NSC=Neighborhood Structural Characteristics , NSP=Neighborhood Social Processes , CPH=Child Physical Assault , CPS=Child Psychological Aggression , CN=Child Neglect . Figure 4. Proposed M ultilevel Structural Equation M odel !!!!55 CHAPTER 4: RESULTS Descriptive Statistics Table 3 displays descriptive statistics of the study sample of 372 foreign -born mothers and their children. Nearly 90% of mothers first came to the U nited States between 1980 and 2000. The majority were Hispanic (70.3%), followed by non -Hispanic Black (12.2%), other (11.9 %), and non -Hispanic White (5.7%). Their average age was 32.44 with a range from 21 to 49. Almost half were married (51.1%) and had less than a high school degree (46.1%). Nearly 60% of mothers were employed. On average, the annual household income was $43,006 (range: 0 to $68,884). In ter ms of mothersÕ behavioral health, the mean of parenting stress was 2.20 (range : 1 to 4), which falls on the response option Ò agree .Ó While the majority (93.5%) did not suffer from depression, 62.4% were alcohol users. Regarding the child ren Õs characterist ics, nearly half were boy s. Just under one -fourth had at least one type of health problem. In regards to the dependent variables, the most frequently reported average level of physical assault was Ò hit the child on the bottom with some hard object (e.g., belt, hairbrush, stick)Ó at 3.35, indicating about three times in the past year. The most frequently reported average level of psychological aggression were Ò shouted, yelled, screamed at child Ó and Òthreatened to spank or hit but did not actually do it Ó at 8.23 and 6.91, respectively, meaning about eight and seven times in the past year, respectively. Neglect was the least frequently reported maltreating behavior among foreign -born mothers , with a range from 0.13 to 0.15, indicating less than one time i n the past year. In terms of the independent variables, average neighborhood structural characteristics ranged from 5% to 28 %. Specifically, the average poverty rate was 18 %, with a minimum of zero and a maximum of 53 %; the mean vacant housing rate was 5% (range: 0 to 38%) ; and the !!!!56 mean foreign -born rate was 28 % (range: 0 to 75%). In addition, on average, 7% of households received public assistance (range: 0 to 32%), 9% of residents were unemployed (range: 0 to 36%), and 15 % of families with children were headed by women (range: 0.1 % to 46%). In terms of neighborhood social processes, average informal social control and social cohesion and trust ranged from 2.44 to 2.92, which falls on the response options Ò likely to interve neÓ and Ò agree ,Ó respectively. Mothers had large social network s in their neighborhoods ( M = 2.44, SD = 0.93) with a range from 0 to 3, but very minimal community participation ( M = 0.85, SD = 0.30), with a range from 0.33 to 1.33. !!!!57 Table 3 . Descriptive Statistics of Study Sample ( N=372) Variable % Mean (SD) Range n Missing (%) MotherÕs demographic and socioeconomic characteristics First came to the U.S. 363 2.4 Before 1970 2.2 1970-1979 8.8 1980-2000 89.0 Race/ethnicity 370 0.5 White, non-Hispanic !5.7 Black, non -Hispanic 12.2 Hispanic 70.3 Other 11.9 Age 32.44 (6.22) 21-49 371 0.3 Marital status 372 0 Not married 48.9 Married 51.1 Education level 371 0.3 Less than high school 46.1 High school degree or GED 21.3 Some college/technical school 18.1 College degree or higher 14.6 Household income 43,006 (59,891) 0-68,844 372 0 Employment status 371 0.3 No 41.8 Yes 58.2 MotherÕs behavioral health Parenting stress 2.20 (0.75) 1-4 371 0.3 Depression 372 0 No 93.5 Yes 6.5 Alcohol use 372 0 Non -alcohol user 62.4 Alcohol user 37.6 Child's characteristics Gender 372 0 Boy 51.6 Girl 48.4 Health Problems 372 0 No 77.7 Yes 22.3 >Ì >Ì >Ì >Ì Note. SD=Standard Deviation . !!!!58 Table 3. (contÕd) Variable Mean (SD) Range n Missing (%) Dependent variables: Child maltreatment measures Physical assault Shook the child 0.50 (2.17) 0-25 369 0.8 Spanked the child on the bottom with bare hand 0.74 (2.62) 0-25 367 1.3 Hit the child on the bottom with some hard object 3.35 (5.53) 0-25 368 1.1 Slapped the child on the hand, arm, or leg 2.22 (4.92) 0-25 368 1.1 Pinched the childÓ, and Òshook the child 0.43 (1.96) 0-25 370 0.5 Psychological aggression Shouted, yelled, screamed at the child 8.23 (9.14) 0-25 370 0.5 Threatened to spank or hit but didnÕt actually do it 6.91 (9.10) 0-25 369 0.8 Swore or cursed at the child 0.75 (3.02) 0-25 370 0.5 Called the child dumb or lazy or some other name like that 0.90 (3.10) 0-25 371 0.3 Said you would send the child away or would kick out of the house 0.58 (2.86) 0-25 370 0.5 Neglect Had to leave the child home alone, even when you thought some adults should be with the child 0.65 (0.42) 0-4 370 0.5 Was caught up with your own problems that you were not able to show or tell the child that they loved the child 0.38 (2.15) 0-25 370 0.5 Was not able to make sure the child got the food when the child needed 0.13 (1.40) 0-25 370 0.5 Was not able to make sure the child got to a doctor or hospital when the child needed it 0.46 (0.35) 0-4 368 1.1 Was so drunk or high that you had a problem taking care of the child 0.19 (0.23) 0-4 367 1.3 Independent variables: Neighborhood measures Neighborhood structural characteristics Percent of families below poverty line 0.15 (0.12) 0-0.53 372 0 Percent of housing units vacant 0.05 (0.49) 0-0.38 372 0 Percent of foreign born 0.28 (0.17) 0-0.75 372 0 Percent of household on public assistant 0.07 (0.06) 0-0.32 372 0 Percent of unemployment 0.09 (0.06) 0-0.36 372 0 Percent of families headed by females 0.15 (0.10) 0.01 -0.46 372 0 Neighborhood social processes Collective efficacy Informal social control 2.92 (0.95) 0.80 -4 364 2.2 Social cohesion and trust 2.60 (0.59) 0.40 -4 370 0.5 Social network 2.44 (0.93) 0-3 372 0 Participation in community 0.85 (0.30) 0.33 -1.33 372 0 Note. SD=Standard Deviation . !!!!59 Confirmatory Factor Analyses Table 4 describes the results of CFA for neighborhood structural characteristics. The CFA showed that five observed indicators of neighborhood structural characteristics, including percent age of families below poverty line, percent age of housing units vacant, percent age of household s on public assistan ce, percent age of unemployment, and percent age of families headed by females, loaded on a single latent variable with a good measurement model fit ("2 = 217.21, p < .001, CFI = .90, TLI = .96, RMSEA = .02, SRMR = .06 ). One indicator, percent age of foreign born , with low factor loading ( b = .18, p < .01), was dropped from the final measurement model for neighborhood structural characteristics, because it substantially decreased the model fit ("2 = 380.01, p < .001, CFI = .67, TLI = .53, RMSEA = .09, SRMR = .09). Table 4 . Confirmatory Factor Analysis for Neighborhood Structural C haracteristics Variable Standardized factor loading P Value Percent of families below poverty line 0.89 < .001 Percent of housing units vacant 0.61 < .001 Percent of household on public assistant 0.74 < .001 Percent of unemployment 0.66 < .001 Percent of families headed by females 0.85 < .001 Model fit "2 = 217.21, p < .001, CFI = .90, TLI = .96 , RMSEA = .02, SRMR = .06 Variable Standardized factor loading P Value Percent of foreign born 0.18 < .01 Table 5 illustrates the results of MCFA for neighborhood social processes. The MCFA revealed that four observed indicators of neighborhood social processes, including informal !!!!60 social control, social cohesion and trust, social network, and community partici pation, loaded on a single latent variable with a good model fit ("2 = 108.32, p < .001, CFI = .99, TLI = .90, RMSEA = .06, SRMR within = .02, SRMR between = .05). Table 5 . Multilevel Confirmatory Factor Analysis for Neighborhood Social Processes Variable Standardized factor loading P Value Collective efficac y Informal social control 0.88 < .001 Social cohesion and trust 0.69 < .001 Social network 0.67 < .001 Community participation 0.38 < .001 Model fit "2 = 108.32 , p < .001, CFI = .99, TLI = .90, RMSEA = .06, SRMR within = .02, SRMR between = .05 Another MCFA was conducted to test the separate measurement models for each of the three types of child maltreatment, physical assault, psychological aggression, and neglect. The five observed indicators of physical assault and psychological aggression loa ded on two single latent variables with a good model fit, respectively (physical assault: "2 = 376.96, p < .001, CFI = 1.00, TLI = .95, RMSEA = .05, SRMR within = .03, SRMR between = .04, psychological aggression: "2 = 215.12, p < .001, CFI = .93, TLI = .92, RMSEA = .05, SRMR within = .01, SRMR between = .02 ). The m easurement model for neglect showed an acceptable model fit ("2 = 100.26, p < .001, CFI = .94, TLI = .91, RMSEA = .06, SRMR within = .06, SRMR between = .06), having all factor loadings above .04 with only one indicator exception, Ò Was not able to make sure the child got to a doctor or hospital when the child needed it Ó (factor loading = .22, p < .05). Thus, this study removed the indicator of neglect with low factor loading in the model and re -ran the MCFA, generating a model fit similar to the initial one ("2 = 208.36, p < .001, CFI = .90, TLI = .89 , RMSEA = .06, !!!!61 SRMR within = .07, SRMR between = .06), although this did not suggest critical improvement for the measurement model fit for neglect. This study, therefore, decided to still include that one indicator , ÒWas not able to make sure the child got to a doctor or hospital when the child needed it,Ó with low factor loading to capture all aspects of neglect. The results of MCFA for child maltreatment are presented in Table 6. !!!!62 Table 6 . Multilevel Confirmatory Factor Analysis for Child Maltreatment Variable Standardized factor loading P Value Physical assault Shook the child 0.89 < .01 Spanked the child on the bottom with bare hand 0.69 < .01 Hit the child on the bottom with some hard object 0.77 < .01 Slapped the child on the hand, arm, or leg 0.82 < .01 Pinched the childÓ, and Òshook the child 0.85 < .01 Model fit "2 = 376.96, p < .001, CFI = 1.00, TLI = .95, RMSEA = .05, SRMR within = .03, SRMR between = .04 Variable Standardized factor loading P Value Psychological aggression Shouted, yelled, screamed at the child 0.87 < .01 Threatened to spank or hit but didnÕt actually do it 0.76 < .05 Swore or cursed at the child 0.96 < .01 Called the child dumb or lazy or some other name like that 0.80 < .01 Said you would send the child away or would kick out of the house 0.64 < .01 Model fit "2 = 215.12, p < .001, CFI = .93, TLI = .92, RMSEA = .05, SRMR within = .01, SRMR between = .02 Variable Standardized factor loading P Value Neglect Had to leave the child home alone, even when you thought some adults should be with the child 0.61 < .01 Was caught up with your own problems that you were not able to show or tell the child that they loved the child 0.63 < .01 Was not able to make sure the child got the food when the child needed 0.40 < .05 Was not able to make sure the child got to a doctor or hospital when the child needed it 0.22 < .05 Was so drunk or high that you had a problem taking care of the child 0.47 < .05 Model fit "2 = 208.36, p < .001, CFI = .90, TLI = .89, RMSEA = .06, SRMR within = .07, SRMR between = .06 !!!!63 Multilevel Structural Equation Modeling Intraclass correlation . Table 7 summarizes ICC coefficients for the three types of child maltreatment, physical assault, psychological aggression, and neglect. The ICCs ranged from .07 to .44 ( p < .001), indicating that there were 7% ! 44% variances in individual foreign -born mot hersÕ reports of child maltreatment across clustered neighborhoods, with an average ICC of 29%. Given these relatively considerable ICC coefficients (Dyer, Hanges, & Hall, 2005), this study concluded that there were sufficient within - and between -level variations in dependent variables to statistically warrant the use of multilevel modeling. Table 7 . Int raclass Correlations for Three T ypes of Child Maltreatment Variable ICC P value Physical assault 0.37 < .001 Psychological aggression 0.44 < .001 Neglect 0.07 < .001 Multilevel mediation model. To simultaneously test Hypotheses 1, 2, and 3, this study structured a multilevel mediation model using the results from CFA for neighborhood structural characteristics and MCFA for neighborhood social processes and the three types of child maltreatment (physical assault, psychol ogical aggression, and neglect ). The f inal MSEM for the multilevel mediation test (Figure 5) differs from the in itially proposed model (Figure 4 ) in that the final MSEM dropped one observed indicator, percent age of foreign born, from the six indicators of neighborhood structural characteristics to improve the final model fit. Thus, the latent variable for characteristics of neighborhood structure included five observed indicators, not six. Except for this discrepancy, the variables for neighborhood social p rocesses and three !!!!64 types of child maltreatment were successfully loaded on the final model as proposed. Figure 5 illustrates the final MSEM, which was used for the multilevel mediation analysis. The mu ltilevel mediation model showed a good model fit ("2 = 361.56, p < .001, CFI = .91, TLI = .93 , RMSEA = .06, SRMR within = .05, SRMR between = .06). !!!!65 Note. NSC=Neighborhood Structural Characteristics, NSP=Neighborhood Social Processes, CPH=Child Physical Assault, CPS=Child Psychological Aggression, CN=Child Neglect. Figure 5. Multilevel Structural Equation M odel !!!!66 Table 8 and Figure 6 provide results from the mediation model in MSEM. Hypothesis 1, the relationship between neighborhood structural characteristics and child maltreatment, was supported for all three types of child maltreatment (H1 -a, H1 -b, H1 -c), including physical assault (# = . 30, p < .01), psychological aggression ( # = . 13, p < .05), and neglect ( # = . 03, p < .05). Negative neighborhood structural characteristics were positively associated with child maltreatment among immigrant families, indicating that foreign -born mothers living in neighborhoods with more negative structural characteristics reported higher leve ls of physical assault, higher psychological aggression, and higher neglect, respectively, after controlling for motherÕs demographic and socioeconomic characteristics and childÕs gender and health problems. Hypothesis 2, relationship between neighborhoo d social processes and child maltreatment, was supported for all three types of child maltreatment (H2 -a, H2 -c, H3 -c), including physical assault ( # = -.15, p < .05), psychological aggression ( # = -.08, p < .05), and neglect ( # = -.04, marginally significa nt p = .08) among immigrant families. Positive neighborhood social processes were inversely associated with physical assault, psychological aggression, and neglect, indicating that foreign -born mothers having positive social processes reported lower levels of physical assault, lower psychological aggression, and lower neglect, respectively, after controlling for motherÕs demographic and socioeconomic characteristics and childÕs gender and health problems. Hypothesis 3, indirect effects of neighborhood soci al processes in the relationship between neighborhood structural characteristics and child maltreatment, was supported for only two types of child maltreatment , physical assault (H3 -a) and psychological aggression (H3 -b). Specifically, neighborhood social processes mediated the relationship between neighborhood structural characteristics and physical assault among immigrant families ( # = -.05, p < .05). Also, !!!!67 neighborhood social processes mediated the relationship between neighborhood structural characteris tics and psychological aggression ( # = -.02, p < .05) among immigrant families. Both indirect effects controlled for motherÕs demographic and socioeconomic characteristics and childÕs gender and health problems. It indicates that foreign -born mothers livin g in neighborhoods with more negative neighborhood structur al characteristics reported lower levels of physical assault via the pathway of neighborhood social process es. Similarly, it indicates that foreign -born mothers living in neighborhoods with more ne gative neighborhood structural characteristics revealed lower levels of psychological aggression through the pathway of neighborhood social processes. However, the indirect effect of neighborhood social processes o n psychological aggression was to lesser e xtent , compared to th e indirect effect on physical assault. In terms of control variables, household income was significantly inversely associated with physical assault ( # = -.15, p < .05) and psychological aggression ( # = -.14, p < .05), respectively, in dicating that foreign -born mothers having higher annual household income reported lower levels of child physical assault and lower psychological aggression, respectively. There were significant positive relationships between mothersÕ parental stress and ph ysical assault ( # = -.08, p < .01), and psychological aggression ( # = .10, p < .01), respectively, indicating that foreign -born mothers having higher levels of parental stress reported higher levels of physical assault and higher psychological aggression, respectively. !!!!68 Table 8 . Results from Mediation Model in Multilevel Structural Equation Modeling Variab les Physical assault Psychologi cal aggression Neglect # SE # SE # SE Covariates >Ì >Ì >Ì >Ì >Ì >Ì MotherÕs demographic and socioeconomic characteristics Years in the U.S. 0.74 0.22 0.77 0.15 0.16 0.17 Race/ethnicity 0.02 0.79 0.25 0.08 0.24 0.08 Age 0.08 0.06 0.67 0.17 0.05 0.17 Marital status 0.06 0.22 0.66 0.23 0.54 0.24 Education level -0.01 0.04 -0.04 0.13 -0.02 0.09 Household income -0.15* 0.12 -0.14* 0.27 -0.18 0.15 Employment status -0.03 0.41 -0.04 0.15 -0.01 0.01 MotherÕs behavioral health Parenting stress 0.08** 0.23 0.10** 0.19 0.09 0.08 Depression 0.81 0.08 0.98 0.20 0.81 0.21 Alcohol use 0.29 0.95 0.11 0.87 0.30 0.71 Child's characteristics Gender 0.68 0.12 0.48 0.09 0.26 0.23 Health Problems 0.38 0.81 0.27 0.74 0.32 0.71 Independent variable Neighborhood structural characteristics 0.30** 0.28 0.13* 0.60 0.03* 0.02 Mediator Neighborhood social processes -0.15* 0.05 -0.08* 0.08 -0.04€ 0.05 Indirect effects Neighborhood structural characteristics # Neighborhood social processes -0.05* 0.04 -0.02* 0.03 -0.01 0.02 R2 0.35 0.29 0.14 Note. Standardized estimates are shown, and €p < .10, * p < .05, ** p < .01. !!!!69 Note. Standardized estimates are shown, and €p < .10, *p < .05 , **p < .01 . Solid lines indicate direct effects, and dotted lines suggest indirect effects. Figure 6. Results from Mediation Model in Multilevel Structural Equation M odeling !!!!70 CHAPTER 5: DISCUSSION This study examined the pathways that explain relationships between neighborhood structural characteristics, social processes, and child maltreatment among immigrant families, in the manner suggested by social disorganization theory. In doing so, it offers several contributions to social work research on the ecological context of child abuse and neglect. This study is the first of its kind to consider the impact of both neighborhood structures and social processes on child maltreatment among immigrant famil ies. Although immigrant families in the United States are considered at greater risk of child abuse and neglect, they are understudied in child maltreatment research. By analyzing the relationships between neighborhood structures and social processes and c hild maltreatment focusing specifically on immigrant families, this study provides a number of insights into factors that may reduce and prevent child maltreatment for an underrepresented group. It also contributes to the empirical and theoretical literat ure by scrutinizing the mediating role of neighborhood social processes in affecting neighborhood structural characteristics and child maltreatment. In this study, neighborhood social processes, measured by collective efficacy, social network s, and communi ty participation, are viewed as protective resources for parenting and child maltreatment that is embedded within neighborhoods. Few other studies of child maltreatment have focus ed either on descriptive information about social processes in relation to child abuse and neglect or on direct associations with child maltreatment without considering the influences of characteristics of neighborhood structure. However, this study built MSEM and conducted multilevel mediation analysis to identify specific mechani sms by and contexts in which neighborhood -level interventions might be effective for child abuse and neglect prevention for immigrant families. !!!!71 Last, findings from this study further extended our understanding o f how the key principles of social disorgani zation theory work for child abuse and neglect for immigrant families . The analysis included the verification of the mechanisms that link structural characteristics of neighborhoods, such as concentrate d poverty and vacant housing, to child maltreatment through incorporating the interactional role of social processes. Unlike the previous studies that used separate or aggregated predictors for neighborhood measures, this study identif ied whether or not major indicators of neighborhood structural characteri stics and social processes represent the theorized measurement models for examination of child maltreatment by conducting CFA and MCFA. It is par ticularly useful in testing whether the percentage of foreign -born, one of the controversial indicators of nega tive characteristics of neighborhood structures, is consistent with our understanding of how racial/ethnic heterogeneity is related or unrelated to one another in neighborhood structural measures for immigrant samples. The following will evaluate each stu dy hypothesis and discuss social work implications of the study findings. Evaluation and Interpretation of Research Hypotheses The first hypothesis was supported by the study results: Foreign -born mothers living in neighborhoods with more negative structural characteristics reported significantly higher level s of child physical assault, higher psychological aggression, and higher neglect. These findings are consistent with a set of neighborhood studies on child maltreatment that examined the relationships between child maltreatment and neighborhood disadvantages, such as concentrated poverty, residential instability, racial/ethnic h eterogeneity, unemployment, child care burden , availability of alcohol and d rugs, and lack of social service s availability (Coulton et al., 1995 ; Coulton et al., 1999; Deccio et al., 1994; Drake & Pandey, 1996; Ernst, 2000, 2001; Frei sthler, !!!!72 2003; 2013; Freisthler et al., 2005; Freisthler et al., 2007; Freisthler et al., 2012 ; Kle in, 2011; Klein & Merritt, 2014; Korbin et al., 1998 ; Lery, 2009 ; Merritt, 2009; Molnar et al., 2016; Morton et al., 2014 ; Zuravin , 1989). Even though the majority of these studies investigated child abuse and neglect among non -immigrant families or did not identify the nativity of the samples, this study confirms the same direction , suggesting that the more negative neighborhoo d structural characteristics, the more maltreatment of children in immigrant families. Also, it supports a long -standing point of social disorganization theory in child maltreatment research. Despite a scarcity of research for immigrant families, a few studies have compared the impacts of structural indicators of neighborhood disorganization on child maltreatment by neighbor hood race/ethnicity (Korbin et al., 1998; Freisthler et al., 2007; Klein & Merritt, 2014) . Each study examined different neighborhoods with predominantly Black, Hispanic, and White population s and found that higher levels of neighborhood disadvantages (e. g., impoverishment, residential instability, population density, immigrant concentration, child care burden, female -headed family, unemployment) were positively related to hig her rates of child maltreatment . One exception, however, is notable from their fi ndings : impoverishment in particular had a significantly weaker impact on maltreatment rates in African American neighborhoods than in Eu ropean American neighborhoods (Korbin et al., 1998) . Moreover, other indicators of neighborhood disorganization, such a s residential instability, were found as protective for maltreatment in minority population s, but not for Whites (Freisthler et al., 2007 ). These findings suggest that the effects of characteristics of neighborhood structures on child abuse and neglect may differ by race/ethnicity or nativity and thus suggest that future research should compare how neighborhood structural measures influence child maltreatment by nativity. This study looked only at immigrant families. Investigation of such difference is a worthy consideration to find what moderates neighborhood !!!!73 structural factors for malt reatment in minorities. In keeping with this conclusion of the positive relationship between negative neighborhood structural characteristics and higher child maltreatment among immigrant families, effect size s of the relationships varied by types of malt reatment. The majority of earlier studies on neighborhood structures and child maltreatment tended to aggregate different types of maltreatment, such as CPS referrals or substantiations and foster care entries (Coulton et al., 1995; Coulton et al., 1999; D eccio et al., 1994 ; Frei sthler, 2003; 2013; Freisthler et al., 2005; Freisthler et al., 2007; Freisthler et al., 2012 ; Klein, 2011; Klein & Merritt, 2014; Korbin et al., 1998; Lery, 2009 ; Molnar et al., 2016) . This hinders the explanation about which type of maltreatment is more or less affected by characteristics of neighborhood structures . However, this study reported the sizes of effect differences by assessing dependent variables separately for physical assault, psychological aggression, and neglect . In this study, the effect size of the relationship between neighborhood structural characteristics and physical assault (# = .30, p < .01) was greater than that of the other two types (psychological aggression: # = . 13, p < .05, neglect: # = . 03, p < .05). This may be because socioeconomically disadvantaged neighborhoods lack resources that can be beneficial to parenting and family functioning , putting foreign -mothers specifically at higher risk of child physical assa ult, which has been examined in several studies (Freisthler & Maguire -Jack, 2015; McLeigh et al., 2018; Molar et al., 2003; Morton et al., 2014 ). Alternately, it is possible that having beliefs and practices in the use of excessive physical discipline among foreign -born mothers (Chang Rhee, & Weaver, 2006; Johnson -Motoyama et al., 2012; Rhee et al., 2012) , paired with the negative effects of neighborhood structures, may be reflected as a risk factor for physica l assault. Research has found greater relat ionship s between socioeconomic factors, such as living conditions and economic hardships, and physical abuse !!!!74 than other types of child maltreatment among immigrant families (Chang et al. , 2006; Rhee et al., 2012; Rhee et al., 2008). The second hypothesis was supported by the study results: Foreign -born mothers having more positive neighborhood social processes had significantly lower level s of physical assault, lower psychological aggression, and lower neglect. These findings are in line with emerging stud ies on the positive role of social processes in reducing child maltreatment (Cao & Maguire -Jack, 2016; Coulton et al., 1999 ; Emery et al., 2014; Emery et al., 2015; Freisthler & Maguire -Jack, 2015; Garbari no & Kostelny, 1992; Garbarino & Sherman, 1980 ; Guterman et al., 2009 ; Kim & Maguire -Jack, 2015 ; Korbin et al., 1999 ; Ma et al., 2018; Maguire -Jack & Showalter, 2016; Maguire -Jack & Wang, 2016 ; McLeigh et al, 2018; Molnar et al., 2003; Molnar et al., 2016; Negash & Maguire -Jack, 2015 ; Vinson et al., 199 6). However, findings of this study add to these studies by analyzing immigrant families. Several studies have suggested that lower levels of neglect of Latino children in immigrant families (Dettlaff et al., 2009 ), or lower impacts of neighborhood poverty on child maltreatment in African American neighborhoods (Korbin at al., 1998) , or an inverse relationship between immigration concentration and neighborhood rates of maltreatment among Whites (Klein & Merritt, 201 4) may partially be attributed to the development of positive social processes among neighbors , such as social support. As posited, this study confirms neighborhood social processes as a protective factor for maltreatment of children in immigrant families . Also, it provide s support for social disorganization theory applied to child maltreatment research. !The varied finding s for neighborhood social processes across maltreatment types merit further interpretation . Although the NSCAW reported that children of immigrants, compared to children of native parents, are significantly more likely to experience psychological aggression !!!!75 (Dettlaff & Earner, 2010; Dettlaff & Earner, 2012 ), previous studies on child maltreatment among immigrant families largely focused on physical discipline or physical abuse (Altschul & Lee, 2011; Alink et al. , 2013; Chang et al., 2006; Dettlaff et al., 2011; Rhee et al., 2012; Rhee et al., 2008; Tajima & Harachi, 2010), rather than psychological aggression (Euser et al, 2011; Osterling & Han, 2011 ). However, this study showed that psychological aggression was the type of child maltreatment most frequ ently reported by foreign -born mothers and that it was inversely associated with neighborhood social processes. This finding is consistent with studies looking at the positive role of neighborhood social processes in psychological aggression using data from the FFCSW study (Cao & Maguire -Jack, 2016; Guterman et al., 2009; Kim & Maguire -Jack, 2013). Guterman et al. (2009) found that while negative neighborhood social processes were directly significantly related to more frequent psychological aggression, the path coefficient for Whites ( # = .22) was greater than for African Americans ( # = .13) and Hispanics ( # = .13). It suggests that the impact of neighborhood social processes on psychological aggression might be lower among minority families. Although this study was unable to compare the findings with native -born families b ecause it looked solely at immigrant famili es, future research should examine how the degrees of influence of social processes on psychological aggression differ by nativity. The relationship between neighborhood social processes and child neglect was statistically marginal ( # = -.04, p < .10). This might be explained with the measure used for neighborhood social processes. This study used foreign -born mothersÕ perceptions on the aspects of social processes, including collective efficacy, social network, and community participation, which are not objective and may be affected by their own background s, experience s, environmental circumstances, and individual characteristics. Polansky et al. (1985) found that !!!!76 neglectful caregivers reported fewer support networks than those who were not neglectful to ward their children within the same neighborhoods. They argued that this is because caregiversÕ perspectives on neighborhood social processes are influenced by individualsÕ attributes that may also be related to maltreatment, so there is a possibility that the relationship between neighborhood social processes and child neglect could be misrepresented (Cao & Maguire -Jack, 2016; Kim & Maguire -Jack, 2015) . It is also possible that immigrant families may interact more in their informal social networks , which may affect neglectful behaviors, rather than in formal social support resources that can be more easily replaced upon moving to a new neighborhood (Klein & Merritt, 2014; Spearly & Lauderdale, 1983; Tajima & Harachi, 2010). This hypothesis may be re flected in this study by the fact that foreign -born mother respondents reported larger social network s from their famil ies and friends ( M = 2.44) than from community participation in formal workshops and information sessions provided by child care settings in their neighborhoods ( M = 0.85). Therefore, the marginal effect of neighborhood social processes on neglect of children in this sample of immigrant families may be due to greater use of informal support (e.g., child care) among them. A s mall number of stud ies support this finding that child neglect is not likely to be affected by neighborhood social processes, including neighborhood social processes using joint indicators of informal social control, social cohesion, and social disorder (Cao & Maguire -Jack, 2016; Guterman et al., 2009) , community involvement and perception (Kim & Maguire -Jack, 2015) , and , social cohesion (McLeigh et al, 2018) . The last hypothesis was partially supported by the study results: Foreign -born mothers living in neighborhoods with negative structural characteristics had lower level s of physical assaults via the pathway of neighborhood social process es, and foreign -born mothers living in neighborhoods with negative structural characteristics had lower psychological aggression via !!!!77 the pathway of neighborhood social process es. However, the relationship between characteristics of neighborhood structure and child neglect through the pathway of social processes was not significant. Findings from this study for physical assault and psychological aggression are congruent with those from previous studies that examined the impacts of neighborhood social processes on child maltreatment in conjunction with structural characteristics (Deccio et al., 1994; Fre isthler & Maguire -Jack, 2015; Fromm, 2004 ; Korbin et al., 1998; McLeigh et al., 2018), highlighting the importance of the buffering role of neighborhood social processes in child maltreatment (e.g., social integration , social connectedness , intergenerational relationship s between adults and children , social cohesion). In addition, it supports social disorganization theory applied to child maltreatment that posits caregivers living in socially cohesive neighborhoods that foster appropriate par enting norms would be less likely to maltreat their children even in disadvantaged neighborhoods (Freisthler & Maguire -Jack, 2015; McLeigh et al., 2018; Molnar et al., 2016) . However, this study augments these studies by discerning an indirect pathway from neighborhood structural characteristics to child physical assault, psychological aggression, and neglect via perceived neighborhood social processes among immigrant families. Specifically, findings of this study are consistent w ith the proposition of McLe igh et al. (2018) that positive neighborhood social processes in highly disadvantaged neighborhoods may help decrease child abuse. Findings from this study differ from those of McLeigh et al. (2018) in that this study used combined measure s for neighborhoo d social processes, including collective efficacy, social network, and community participation , and measured three types of self -reported maltreatment , including physical assault, psychological aggression, and neglect, while McLeigh et al. (2018) examined social cohesion for two types of neighborhood rates of maltreatment from official CPS records, abuse and neglect. Despite these discrepancies, based on the evidence from !!!!78 the two studies, it can be inferred that the protective mechanism of neighborhood soci al processes in the association between neighborhood structural disadvantages and child abuse appears to operate regardless of nativity of caregivers, because this study examined foreign -born mothers in 20 large U.S. cities and McLeigh et al. (2018) conducted the ir study for randomly selected neighbors residing in South Carolina. However, it should be noted that the reliability of McLeigh and colleagues Õ (2018) study is in question because they did not account for potential spatial autocorrela tion , even though they used contiguous census block group s as a unit of analysis. In neighborhood analyses of contiguous units, spatial autocorrelation , or the correlation of a variable with itself through space , is common because neighborhood units connec ted to each other share permeable boundaries (Freisthler, Lery, Gruenewald, & Chow, 2006). Spatial autocorrelation violates the primary assumption of ordinary least squares that !"#$! values of !%&'"!observations are independent of one another. When this assumption is violated , it biases the coefficients, resulting in an increased risk of Type I or Type II error s (Cliff & Ord, 1981). Child maltreatment scholars who have examined neighborhood units (e.g., census block s, census trac ts, zip code s) have been applying the spatial error regression method to account for this potential source of bias (Freisthler, 2004, 2013; Freisthler et al., 2006; Klein, 2011; Klein & Merritt, 2014; Lery, 2009) . Contrary to this studyÕs hypothesis, neighborhood social processes did not mediate the relationship between structural characteristics of a neighborhood and child neglect among immigrant families. This could mean that promoting social processes for foreign -born mothers living in higher levels of neighborhood structural c haracteristics may not be effective for reducing neglect of children in immigrant families. One possible explanation for this result is that causes of child neglect are often considered as circumstantial (e.g., lack of social services !!!!79 within neighborhoods to provide adequate basic needs for a child), rather than intentional ( an individual Õs motive to physically abuse a child) (Golden, Samuels , & Southall, 2003) , so how caregivers individually interact and connect with neighbors might not be as critical for child neglect as whether or not the neighborhoods where the caregivers live are disadvantaged. The result a lso may be explained by measurement issues for neighborhood social processes. As described earlier for Hypothesis 2, neglectful caregivers tend to re port fewer social networks (Polansky, Ammons, & Gaudin , 1985), which may affect the insignificant path from characteristics of neighborhood structures to child neglect via neighborhood social processes. One study support ed no indirect effect of social proc esses (e.g., social cohesion) on the relationship between neighborhood structural characteristics and child neglect (McLeigh et al., 2018). Limitations !! These findings should be interpreted with caution due to several limitations . First, this study used self -reported measures for child maltreatment and neighborhood social processes, and thus is vulnerable to social desirability and underestimation for reports of maltreatment. Second, the FFCWS studyÕs sample design, with geographi cally dispersed study respondents within census trac ts (Reichman, Teitler, Garfinkel, & Mc Lanahan, 2001 ), limits the ability to detect wide variations across within - and between -levels with a low average cluster size ( 1.08, meaning that there was , on avera ge, approximately one foreign -born mother in each neighborhood defined by the census trac t). Also, census trac ts may not correspond with the perceptions of neighborhood s among foreign -born mothers. Third , this study used FFCWSÕs data collected between 1998 and 2000 (baseline) and 2003 and 2006 (Wave 4). Considering these past time frames, findings of this study may not be generalized to the present time. The c haracteristics of !!!!80 neighborhood structures and social international mechanisms might have changed since th e data were collected . Fourth, this study was not able to conduct multi -group analyses by race/ethnicity among immigrant families , because the majority of the sample was Hispanic (70.3%) and there was an insufficient number of survey respondents from other race/ethnic groups (e.g., Asian). This inhibits our understanding of how the effects of neighborhood on child maltreatment differ across race/ethnicity among immigrant families. Specifically, aspe cts of neighborhood social processes may differ significantly by race/ethnicity, which are likely to produce different path coefficients for study Hypothes es 2 and 3 across race/ethnicity groups. Last, as this study is of a cross -sectional survey nature, i t could not analyze temporal directions among study variables. It is possible, for instance, that foreign -born mothersÕ greater felt stress stemming from living in socioeconomically disadvantaged neighborhoods may lead to greater isolation and lower intera ction with neighbors, which, in turn, may affect maltreatment of children in immigrant families. However, t his study could not elucidate the ordering of these possible pathways. Despite these limitations, findings from this study have important implicatio ns for social work practice, policy, and research. Implications Social work practice and policy. Findings from this study provide evidence of how neighborhood factors can help reduce and prevent maltreatment of children in immigrant families. Given that negative characteristics of neighborhood structure predicted higher level s of child maltreatment, while positive social processes predicted lower level s of child maltreatment among immigrants, ecological multitier social work interventions may prove more effective than a single program designed to affect only foreign -born mothers, immigrant families, or neighborhoods as separate entities. An example of this multitier approach might be an !!!!81 intervention to reduce income inequality in neighborhoods while incre asing connections across immigrant families . Even when both parents of the family work, one out of four children of immigrant families are still low income (Capps, Horowitz, Fortuny, Bronte-Tinkew, & Zaslow, 2009), a factor that may influence their neighborhoods , where families are more susceptible to violence, substance use, and disruptive social circumstances (Johnson -Motoyama & Wu, 2018) . The result may be that immigrant parents raising their children in neighborhoods where poverty is concentrated might have more difficulty reducing environmental risks , such that even greater provision of interventions (e.g., parental monitoring) are less likely to be effective (Freisthler & Maguire -Jack, 2015; Johnson -Motoyama & Wu, 2018) . It bolsters a need for c hild maltreatment prevention programs in neighborhoods targeting both individual and community level s. Th is suggestion is in line with those of Kim and Maguire -Jack (2013) and Sabol, Coulton, and Korbin (2004), who highlighted building community capacity f or violence prevention (e.g., intimate partner violence, !child maltreatment ). They recommended three program strategies to prevent violence: (1) interventions as social integration to increase linkages between individuals and community institutions such as schools; (2) interventions increasing resources and ties within communities; and (3) interventions designed to connect individuals, families, communities, institutions, and related policies. This multi -layered approach may help create a healthy neighborho od in which immigrant families feel more integrated and able to access needed supports that may reduce and prevent maltreatment of their children. Social work policy must support this approach of ecological multitier social work programs that intervene for neighborhood economic development and individual opportunities for social integration in the community and /or parenting programs to be located within disadvantaged neighborhoods. This will be especially appealing for immigrant families, because it does not require social work !!!!82 practitioners to identify families who are Òat risk Ó but instead focuses on building a safe environment for all children and their caregivers. Therefore, immigrant families need not be stigmatized when participating in these social services because they are available to all families in the neighborhood. If community -policing efforts that invest in neighborhood disadvantages and formal resources of connection and support are put in neighborhoods, immigrant families may be more likely to utilize these services , thereby reducing the maltreatment of children in those families. The f indings of this study suggest that immigrant families with limited social interactions and activities that reside in socioeconomically disadvantaged neighborhoods would appear to benefit the most from programs that provide a combination of improvement for community safety, concrete social network s, and meaningful activities in which foreign -born mothers may participate. However, hostile anti -immigrant policies are typically related to downward impacts on the use of social services among immigrant families , due to their distrust of authorit y (Hagan, Rodrigue z, Capps, & Kabiri, 2003), which generates additional challenges for immigrants (Johnson -Motoyama & Wu, 2018) . The c ommunity health workers (CHW) model, for example, is a possible approach that social workers may use to overcome these barriers. As a public health outreach, CHW is defined as a frontline public health worker who is a trusted member of an d/or has a deep understanding of the community served (American Public Health Association, 2011) , although the se workers generally have shorter trainings than professional workers (World Health Organization , 1989). The history of CHW traces back to the 1970s; its introduction aimed to improve maternal and child health and management of common infectious diseases in areas with limited health workforce s and low access to basic health services (Perry, Zulliger, & Rogers, 2014). Since then, CHW has been succ essful in offering a wide range of child health !!!!83 interventions by bridging cultural context s between health/social services and communities that are low income and racially an d ethnically diverse (Gibbons & Tyus, 2007; Javanpar ast, Windle, Freeman, & Baum, 2018; Nunnery & Dharod, 2015). Social work interventions that reinforce neighborhood social processes for immigrant families might be incorporated into the CHW model, which has been built in communit ies with similar cultural backgrounds to establish residentsÕ trust in service providers. As suggested by Javanparast et al. (2018), the responsibilities of CHW in child welfare sett ings may focus on education and advocacy/engagement roles, including (1) delivering information sessions for imm igrant families about access to child welfare services; (2) strengthening connections between immigrant families and child welfare services that provide interventions to promote social support and community participation ; and (3) supporting child welfare p ractitionersÕ cultural competence and the cultural appropriate ness of child welfare services . Child maltreatment prevention that is effective and efficacious for immigrant families at the individual and community level s will be possible only through collaborative efforts with continued advocacy considering cultural diversity, community endeavor s, and political will. Social work research .!The findings of this study suggest directions for future research related to the role of n eighborhoods in child maltreatment. First, although social disorganization theory has long been employ ed in neighborhood studies of child maltreatment, it must be reevaluated in light of the empirical evidence. Despite the theoryÕs predictive power for the effect of neig hborhood structures and social processes on child abuse and neglect, this study suggests that there is room for reconsidering, particularly when it comes to specifying the characteristics of neighborhood structure, such as rac ial/ethnic heterogeneity. Shaw and McKay (1942) argued that communities with low socioeconomic status were characterized by diversity !!!!84 in norms and values of behaviors, rather than uniformity. The result was that youth in impoverished communities, who were exposed to a wide variety of c ontradictory norms, were more likely to engage in deviant behaviors. In this sense, Shaw and McKay (1942) determined that racial/ethnic heterogeneity, as an indicator of the negative structure of a neighborhood, resulted in disrupting social cohesion, whic h, in turn, elevate d community crime rates. Child maltreatment researchers have been using immigrant concentration as a proxy for rac ial/ethnic heterogeneity with an expectation of find ing that immigrant concentration predicts higher rates of child maltreatment. However, some studies examining this relationship have failed to find the relationship (Freisthler et al., 2007) or found inverse relationships (Ernst, 2000; Freisthler & Maguire -Jack, 2 015; Klein & Merritt, 2014; Molnar et al., 2003). Th is suggest s that rac ial/ethnic heterogeneity may be a poor predictor for neighborhood structural disadvantages. This study conducted CFA to confirm or reject whether the traditional measurements for negat ive characteristics of neighborhood structure fit into neighborhood structural disadvantages using six variables, including neighborhood rates of poverty, vacant housing, foreign -born, household s on public assistance, unemployment, and families headed by f emales. Among these six indicators, the percent age of foreign -born did not statistically fit well into the theorized measurement model for negative characteristics of neighborhood structure, indicating that immigrant concentration was less related to the o ther indicators. This may be because this study tested the foreign -born measure in CFA exclusively for immigrant families, who might not be so negatively affect ed by living in predominantly immigrant neighborhoods. Another possibility is that , as Klein and Merritt (2014) suggested, immigrant concentration may be a proxy for social cohesion, rather than social disorganization, which help s alleviate child maltreatment. Therefore, future research must continue to evaluate how rac ial/ethnic heterogeneity consti tutes a vital component of social !!!!85 disorganization theory and reflect empirical evidence in child maltreatment research. Also, future research needs to take into account cultural contexts in examination of the theory, as a growing number of studies on social disorganization theory have highlight ed that neighborhood subculture , as well as neighborhood conditions, has an impact on crime and delinquency (Kubrin, & Weitzer, 2003; Kubrin & Wo, 2015). This is particularly essential for immigrant fa milies or immigrant enclaves that could develop and practice unique techniques with regard to their parenting and child rearing. ! In addition , future neighborhood research should incorporate characteristics of immigrant families into the analysis model, su ch as length of residence in the U nited States and parenting stress , which were controlled in this study , to explore how their experience of migration and settlement influences child maltreatment in addition to neighborhood context. A recent review of literature on child maltreatment in immigrant and refugee families reported that more years liv ed in the U nited States was a risk factor for severe child physical abuse (LeBrun et al., 2015) . This was explained by the acculturative stress, hypothesiz ing that immigrant families may be at higher risk of child maltreatment during the first years of settlement because they face challenges for acculturation and are less familiar with the laws, values, and norms in the U nited States (LeBrun et al., 2015; Rhee et al, 2012) . Thus, an examination of length of residence in the U nited States among immigrant families may help us understand how long settled immigrants are more or less likely to take in develop ing supportive resources in their neighborhoods , which may affect the risk of maltreatment of children in immigrant families . On a related note, the role of parental stress in immigrant families should be investigated in combination with the context of neighborhood s, given that this study found that parental stre ss predicted more frequent child physical assault and more frequent psychological aggression. Acculturation processes are !!!!86 relevant to parenting stress because generations tend to experience acculturation differently. Children often learn and accept the soc ial norms and language of their new culture more rapidly than their parents do, which can cause family conflicts or parenting stress (Dinh, & Nguyen, 2006; Phinney, Ong, & Madden, 2000). Two studies found significant pathways from neighborhood social processes to child maltreatment through parental stress (Guterman et al., 2009; Maguire -Jack & Wang, 2016 ). However, th ese studies examined the general population. Therefore, further empirical evidence for immigrants is necessary to demonstrate possible pathways link ing neighborhood context, parental stress, and child maltreatment to effectively address preventive strategies for maltreatment of children in immigrant families. !!!!87 REFERENCES !!!!88 REFERENCES Abidin, R. (1995). Parent stress inventory, Odessa, FL: Psychological Assessment Resources. Acock, A. C. (2005). Working with missing v alues. Journal of Marriage and Family, 67 (4), 1012Ð1028. doi: 10.1111/j.1741 -3737.2005.00191.x Alink, L. R., Euser, S., IJzendoorn, M. H., & Bakermans !Kranenburg, M. J. (2013). Is elevated risk of child maltreatment in immigrant families associated with socioeconomic status? Evidence from three sources. Internationa l Journal of Psychology, 48 (2), 117Ð127. doi: 10.1080/00207594.2012.734622 Altschul, I., & Lee, S. J. (2011). Direct and mediated effects of nativity and other indicators of acculturation on Hispanic mothersÕ use of physical aggression. Child Maltreatment, 16(4), 262Ð274. doi: 10.1177/1077559511421523 American Public Health Association. (2001). Recognition and support for community health workers' contributions to meeting our nation's health care needs . Retrieved from https://www.apha.org/policies -and -advocacy/public -health -policy -statements/policy -database/2014/07/15/13/24/recognition -and -support -community -health -workers -contrib -to-meeting -our-nations -health -care -needs Anderson, J. C., & Gerbing, D. W. (1988). Structur al equation modeling in practice: A review and recommended two -step approach. Psychological Bulletin, 103 (3), 411Ð423. Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling (pp. 243Ð277). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Baron, R. M., & Kenny, D. A. (1986). Moderator -mediator variables distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51 (6), 1173Ð1182. doi: 10.1037//0022 -3514.51.6.1173 Belsky, J. W. (1980). Child maltreatment: An ecological integration. American Psychologist , 35(4), 320Ð335. doi: 10.1037/0003 -066x.35.4.320 Ben -Arieh, A. (2010). Socioeconomic correlates of rates of child maltreatment in small communities. American Journal of Orthopsychiatry, 80 (1), 109Ð114. doi:10.1111/j.1939 -0025.2010.01013.x Bendheim -Thoman Center for Research on Child Wellbeing . (2018). Fragile families census tract measures restricted use data appendage : Baseline, 1, 3, 5, 9, and 15 year waves . Princeton, NJ: Bendheim -Thoman Center for Research on Child Wellbeing . !!!!89 Bollen, K. A., & Pearl, J . (2013). Eight myths about causality and s tru ctural equation m odels . In S. L. Morgan. (Ed.), Handbook of causal analysis for social r esearch (pp. 301Ð28). Dordrecht, The Netherlands: Springer. Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 136Ð162). Newbury Park, CA: Sage. Cao, Y., & Maguire -Jack, K. (2016). Interactions with community members and institutions: Preventive pathways for child maltreatment. Child Abuse & Neglect, 62 , 111Ð121. doi:10.1016/j.chiabu.2016.10.012 Capps, R., Horowitz, A., Fortuny, K., Bronte -Tinkew, J., & Zaslow, M. (2009). Young children in immigrant families face higher risk of food insecurity . Retrieved from https://www.childtrends.org/wp -content/uploads/ 01/Immigrant -Food -Security.docx Chang, J., Rhee, S., & Berthold, S. M. (2008). Child abuse and neglect in Cambodian refugee families: Characteristics and implications for practice. Child Welfare , 87(1), 141Ð160. Chang, J., Rhee, S., & Weaver, D. (2006). Characteristics of child abuse in immigrant Korean families and correlates of placement decisions. Child Abuse & Neglect , 30(8), 881Ð891. doi: 10.1016/j.chiabu.2006.03.002 Cliff, A. D., & Ord, J. K. (1981). Spatial processes: M odels and applications . London, UK: Pion Limited. Coulton, C. J., Crampton, D. S., Irwin, M., Spilsbury, J. C., & Korbin, J. E. (2007). How neighborhoods influence child maltreatment: A review of the literature and alternative pathways. Child Abuse & Neglect , 31(11), 1117Ð1142. doi: 10.1016/j.chiabu.2007.03.023 Coulton, C. J., Korbin, J. E., & Su, M. (1999). Neighborhoods and child maltreatment: A multi -level study. Child Abuse & Neglect, 23 (11), 1019Ð1040. doi: 10.1016/s0145 -2134(99)00076-9 Coulton, C. J., Korbin, J. E., Su, M., & Chow, J. (1995). Com munity level factors and child maltreatment rates. Child Development , 66(5), 1262Ð1276. doi: 10.2307/1131646 Deccio, G., Horner, W. C., & Wilson, D. (1994). High -risk neighborhoods and high -risk families: Replication research related to the human ecology of child maltreatment. Journal of Social Service Research , 18(3Ð4), 123Ð137. doi: 10.1300/j079v18n03_06 Dettlaff, A. J., & Finno -Velasquez, M. (2013). Child maltreatment and immigration enforcement: Considerations for child welfare and legal systems working with immigrant families. ChildrenÕs Legal Rights Journal , 33(1). 37Ð63. doi: 10.1111/j .1755-6988.2011.01069.x !!!!90 Dettlaff, A. J., & Johnson, M. A. (2011). Child maltreatment dynamics among immigrant and U.S. born Latino children: Finding s from the National Survey of Child and Adolescent Well-being (NSCAW). Children and Youth Services Review , 33(6), 936Ð944. doi: 10.1016/j.childyouth.2010.12.017 Dettlaff, A. J., Earner, I., & Phillips, S. D. (2009). Latino children of immigrants in the chil d welfare system: Prevalence, characteristics, and risk. Children and Youth Services Review, 31(7), 775Ð783. doi: 10.1016/j.childyouth.2009.02.004 Dettlaff, A., & Earner, I. (2012). Children of immigrants in the child welfare system: Characteristics, risk, and maltreatment. Families in Society: The Journal of Contemporary Social Services , 93(4), 295Ð303. doi: 10.1606/1044 -3894.4240 Dinh, K. T., & Nguyen, H. H. (2006). The effects of acculturative variables on Asian American parent -child relationships. Journal of Social & Personal Relationships, 23 (3), 407Ð426. doi: 10.1177/0265407506064207 Dinh, K. T., & Nguyen, H. H. (2006). The effects of acculturative variables on Asian American parent -child relationships. Journal of Social & Personal Relationships, 23 (3), 407Ð426. doi:10.1177/0265407506064207 Drake, B., & Pandey, S. (1996). Understanding the relationship between neighborhood poverty and specific types of child maltreatment. Child Abuse & Neglect, 20 (11), 1003Ð1018. doi: 10.1016/0145 -2134(96)00091-9 Dyer, N. G., Hanges, P. J., & Hall, R. J. (2005). Applying multilevel confirmatory factor analysis techniques to the study of leadership. The Leadership Quarterly, 16 (1), 149Ð167. doi:10.1016/j.leaqua.2004.09.009 Emery, C. R., Eremina, T., Yang, H. L., Yoo , C., Yoo, J., & Jang, J. K. (2014). Protective informal social control of child maltreatment and child abuse injury in Seoul. Journal of Interpersonal Violence , 30(18), 3324Ð3339. doi:10.1177/0886260514554422 Emery, C. R., Trung, H. N., & Wu, S. (2015). N eighborhood informal social control and child maltreatment: A comparison of protective and punitive approaches. Child Abuse & Neglect , 41, 158Ð169. doi:10.1016/j.chiabu.2013.05.002 Ernst, J. S. (2000). Mapping child maltreatment: Looking at neighborhoods i n a suburban county. Child Welfare, 79 (5), 555Ð572. Ernst, J. S. (2001). Community -level factors and child maltreatment in a suburban county. Social Work Research, 25 (3), 133Ð142. doi: 10.1093/swr/25.3.133 Euser, E. M., van IJzendoorn, M. H., Prinzie, P., & Bakermans -Kranenburg, M. J. (2011). Elevated child maltreatment rates in immigrant families and the role of socioeconomic differences. Child Maltreatment , 16(1), 63Ð73. doi: 10.1177/1077559510385842 !!!!91 Ford, J. K., MacCallum, R. C., & Tait, M. (1986). The app lication of exploratory factor a nalysis in applied psychology: A critical review and analysis. Personnel Psychology, 39 (2), 291Ð314. doi: 10.1111/j.1744 -6570.1986.tb00583.x Freisthler, B. (2004). A spatial analysis of social disorganization, alcohol access, and rates of child maltreatment in neighborhoods. Children and Youth Services Review, 26 (9), 803Ð819. doi: 10.1016/j.childyouth.2004.02.022 Freisthler, B. (2013). Need for and access to supportive services in the child welfare system. GeoJournal, 78 (3), 429Ð441. doi: 10.1007/s10708 -011-9426-6 Freisthler, B., & Maguire -Jack, K. (2015). Understanding the interplay between neighborhood structural factors, social processes, and a lcohol outlets on child physical a buse. Child Maltreatment, 20 (4), 268Ð277. doi:10.1177/1077559515598000 Freisthler, B., Bruce, E., & Needell, B. (2007). Understanding the geospatial relationship of neighborhood characteristics and rates of maltreatment fo r Black, Hispanic, and White children. Social Work, 52 (1), 7Ð16. doi: 10.1093/sw/52.1.7 Freisthler, B., Gruenewald, P. J., Remer, L. G., Lery, B., & Nee dell, B. (2007). Exploring the spatial dynamics of alcohol outlets and child protective services referrals, substantiations, and foster care e ntries. Child Maltreatment, 12 (2), 114Ð124. doi:10.1177/1077559507300107 Freisthler, B., Holmes, M. R., & Wolf, J. P. (2014). The dark side of social support: Understanding the role of social support, drinking b ehaviors and alcohol outlets for child physical abuse. Child Abuse & Neglect , 38(6), 1106Ð1119. doi: 10.1016/j.chiabu.2014.03.011 Freisthler, B., Kepple, N. J., & Holmes, M. R. (2012). The geography of drug market activities and child maltreatment. Child Ma ltreatment, 17 (2), 144Ð152. doi: 10.1177/1077559512443124 Freisthler, B., Lery, B., Gruenewald, P.J., & Chow, J. (2006). Methods and challenges of analyzing spatial data for social work problems: The case of examining child maltreatment geographically. Social Work Research, 30 (4), 198Ð210. doi: 10.1093/swr/30.4.198 Freisthler, B., Merritt , D. H., & LaScala, E. A. (2006). Unde rstanding the ecology of child maltreatment: A review of the literature and directions for future research. Child Maltreatment , 11(3), 263Ð280. doi: 10.1177/1077559506289524 Freisthler, B., Midanik, L. T., & Gruenewald, P. J. (2004). Alcohol outlets and child physical abuse and neglect: Applying routine activities theory to the study of child maltreatment. Journal of Studies on Alcohol, 65(5), 586Ð592. doi: 10.15288/jsa.2004.65.586 !!!!92 Freisthler, B., Needell, B., & Gruenewald, P. J. (2005). Is the physical availability of alcohol and illicit drugs related to neighborhood rates of child maltreatment? Child Abuse & Neglect, 29(9), 1049Ð1060. doi: 10.1016/j.chiabu.2004.12.014 Fromm, S. (2004). The processes tha t moderate the effect of commu nity structural factors on neighbo rhood child maltreatment rates ( Unpublished doctoral dissertation). North Carolina State University, Raleigh, NC. Garbarin o, J. (1976). A preliminary study of some ecological correlates of child abuse: The impact of socioeconomic stress on mothers. Child Development, 47 (1), 178Ð185. doi: 10.1111/j.1467 -8624.1976.tb03407.x Garbarino, J., & Kostelny, K. (1992). Child maltreatment as a community problem. Child Abuse & Neglect , 16(4), 455Ð464. doi:10.1016/0145 -2134(92)90062-v Garbarino, J., & Sherman, D. (1980). High -risk neighborhoods and high -risk families: The human ecology of child m altre atment. Child Development, 51 (1), 188Ð199. doi:10.2307/1129606 Gibbons, M. C., & Tyus, N. C. (2007). Systematic review of U.S. -based r andomize d controlled trials using community health w orkers. Progress in Community Health Partnerships: Research, Education, and Action, 1 (4), 371Ð381. doi:10.1353/cpr.2007.0035 Golden, M. H. , Samuels , M. P., & Southall, D. P. (2003). How to distinguish between neglect and deprivational abuse. Archives of Disease in Childhood, 88 (2), 105Ð107. doi:10.1136/adc.88.2.105 Gracia, E., & Musitu, G. (2003). Social isolation from communities and child maltreatment: A cross -cultural comparison. Child Abuse & Neglect, 27 (2), 153Ð168. doi: 10.1016/s0145 -2134(02)00538-0 Gracia, E., LŠpez -Qu™lez, A., Marco, M., & Lila, M. (2017). Ma pping child maltreatment risk: A 12-year spatio -temporal analysis of neighborhood influences. International Journal of Health Geographics , 16(1), 1Ð12. doi: 10.1186/s12942 -017-0111-y Guterman, N. B., Lee, S. J., Taylor, C. A., & Rathouz, P. J. (2009). Paren tal perceptions of neighborhood processes, stress, personal control, and risk for physical child abuse and neglect. Child Abuse & Neglect, 33 (12), 897Ð906. doi:10.1016/j.chiabu.2009.09.008 Hagan, J., Rodriguez, N., Capps, R., & Kabiri, N. (2003). The eff ects of recent welfare and immigration reforms on immigrantsÕ access to health care 1. International Migration Review, 37 (2), 444Ð463. doi: 10.1111/j.1747 -7379.2003.tb00144.x Hart, T. C., & Waller, J. (2013). Neighborhood boundaries and structural determina nts of social disorganization: E xamining the validity of commonly used measures. Western Criminology Review 14 (3). 16Ð33. doi: 10.1080/03637750903310360 !!!!93 Hayes, A. F. (2009). Beyond Bar on and Kenny: s tatistical mediation analysis in the new millennium. Commu nication Monographs, 76 (4), 408Ð420. Hipsman, F., & Meissner, D. (2013) Immigration in the United States: New economic, social, political landscapes with legislative reform on the horizon. Retrieved from https://www.migrationpolicy.org/article/immigration -united -states-new -economic -social -political -landscapes -legislative -reform Javanparast, S., Windle, A., Freeman, T. , & Baum, F. (2018). Community health worker programs to improve healthcare access and equ ity: Are they only relevant to low - and middle-income c ountries? International Journal of Health Policy and Management, 7(10), 943Ð954. doi:10.15171/ijhpm.2018.53 Johnson, M. A. (2007). The social ecology of acculturation: Implications for child welfare services to children of immigrants. Children and Youth Services Review, 29 (11), 1426!1438. doi:10.1016/j.childyouth.2007.06.002 Johnson -Motoyama, M., & Wu, W. (2018). Unraveling disparities in child neglect risk between Hispanics who are immigrants and those born in the United States: A social -ecological approach using structural equation m odeling. Child Welfare , 96(5). 51Ð80. Johnson -Motoyama, M., Dettlaff, A. J., & Finno, M. (2012). Parental nativity and the decision to substantiate: Findings from a study of Latino children in the second National Survey of Child and Adolescent Well -being (NSCAW II). Children and Youth Services Review, 34(11), 2229Ð2239. doi: 10.1016/j.childyouth.2012. 07.017 Jıreskog, K. G. (1971). Simultaneous factor analysis in several populations. Psychometrika, 36(4), 409Ð426. doi:10.1007/bf02291366 Kempe, C. H., Silverman, F., Steele, B., Droegemueller, W., & Silver, H. (1962). The battered child syndrome. Journal of the American Medical Association , 181, 17Ð24. doi: 10.1001/jama.1962.03050270044010 Kessler, R. C., Andrews, G., Mroczek, D., Ustun, T. B., & Wittchen, H. U. (1998). The Kim, B., & Maguire -Jack, K. (2015). Community interaction and child maltreatment. Child Abuse & Neglect , 41, 146-157. doi:10.1016/j.chiabu.2013.07.020 Klein, S. (2011). The availability of neighborhood early care and education resources and the maltreatment of young children. Child Maltreatment, 16 (4), 300Ð311. doi: 10.1177/1077559511428 801 Klein, S., & Merritt , D. (2014). Neighborhood racial & ethnic diversity as a predictor of child welfare system involvement. Children and Youth Services Review, 41 , 95Ð105. doi: 10.1016/j.childyouth.2014.03.009 Kline, R.B. (2005), Principles and practice of structural equation m odeling. New York , NY : The Guilford Press. !!!!94 Kohen, D. E., Dahinten, V. S., Leventhal, T., & McInto sh, C. N. (2008). Neighborhood disadvantage: Pathways of effects for young c hildren. Child Development , 79(1), 156Ð 169. doi: 10.1111/j.1467 -8624.2007.01117.x Korbin, J. E., Coulton, C. J., Chard, S., Platt -Houston, C., & Su, M. (1998). Impoverishment and child maltreatment in African American and European American neighborhoods. Development and Psychopathology, 10 (2), 215Ð233. doi: 10.1017/s0954579498001588 Kubrin, C. E., & Weitzer, R. (2003). New directions in social d isorganization Theory. Journal of Research in Crime and Delinquency , 40(4), 374Ð402. doi:10.1177/0022427803256238 Kubrin, C. E ., & Wo, J. C. (2015). Social disorgan ization theory's g reatest challenge. In A. R. Piquero (Ed.) The handbook of criminological theory (pp. 121Ð136). Hoboken, NJ: John Wiley & Sons. doi:10.1002/9781118512449.ch7 LeBrun, A., Hassan, G., Boivin, M., Fraser, S. L., & Dufour, S. (2015). Review of child maltreatment in immigrant and refugee families. Canadian Journal of Public Health , 106(7), eS45 ÐeS56. doi: 10.17269/cjph.106.4838 Lery, B. (2009). Neighborhood structure and foster care entry risk: The role of spatial scale in defining neighborhoods. Children and Youth Services Review, 31 (3), 331Ð337. doi: 10.1016/j.childyouth.2008.08.001 Ma, J., Grogan -Kaylor, A., & Klein, S. (2018). Neighborhood collective efficacy, parental spanking, and subsequent risk of household child protective services involvement. Child Abuse & Neglect , 80, 90Ð98. doi: 10.1016/j.chiabu.2018.03.019 MacKinnon, D. P. (2008). Introduction to statistical mediation analysis . New York, NY: Erlbaum. MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review of Psychology, 58, 593Ð614. doi:10.1146/ annurev.psych.58.110405.085542 Maguire -Jack, K., & Klein, S. (2015). Parenting and proximity to social services: Lessons from Los Angeles County in the community context of child neglect. Child Abuse & Neglect , 45, 35Ð45. doi: 10.1016/j.chiabu.2015.04.020 Maguire -Jack, K., & Negash, T. (2016). Parenting stress and child maltreatment: The buffering effect of neighborhood social service availability and accessibility . Children and Youth Services Review, 6 0, 27Ð33. doi:10.1016/j.childyouth.2015.11.016 Maguire -Jack, K., & Showalter, K. (2016). The protective effect of neighborhood social cohesion in child abuse and neglect. Child Abuse & Neglect, 52 , 29Ð37. doi:10.1016/j.chiabu.2015.12.011 Maguire -Jack, K., & Wang, X. (2016). Pathways from neighborhood to neglect: The mediating effects of social support and parenting stress . Children and Youth Services Review, 66 , 28Ð34. doi:10.1016/j.childyouth.2016.04.017 !!!!95 Maiter, S., Stalker, C., & Alaggia, R. (2009). The e xperiences of minority immigrant families receiving child welfare services: Seeking to understand how to reduce risk and increase protective factors. Families in Society: The Journal of Contemporary Social Services , 90(1), 28Ð36. doi: 10.1606/1044 -3894.3842 McLeigh, J. D., McDonell, J. R., & Lavenda, O. (2018). Neighborhood poverty and child abuse and neglect: The mediating role of social cohesion. Children and Youth Services Review, 93, 154Ð160. doi: 10.1016/j.childyouth.2018.07.018 Merrit t, D. H. (2009). Child abuse potential: Correlates with child maltreatment rates and structural measures of neighborhoods. Children and Youth Services Review, 31 (8), 927Ð934. doi: 10.1016/j.childyouth.2009.04.009 Migration Policy Institute . (2019). Children in U.S. immigra nt families by age group and state, 1990 versus 2017 . Retrieved from https://www.migrationpolicy.org/programs/data -hub/charts/children -immigrant -families Molnar, B. E., Buka, S. L., Brennan, R. T., Holton, J. K., & Earls, F. (2003). A multilevel study of n eighborhoods and parent -to-child physical aggression: Results from the project on human development in Chicago neighborhoods. Child Maltreatment, 8 (2), 84Ð97. doi: 10.1177/1077559502250822 Molnar, B. E., Goerge, R. M., Gilsanz, P., Hill, A., Subramanian, S. V., Holton, J. K., . . . Beardslee, W. R. (2016). Neighborhood -level social processes and substantiated cases of child maltreatment. Child Abuse & Neglect, 51 , 41Ð53. doi: 10.1016/j.chiabu.2015.11.007 Morton, C. M. (2013). The moderating effect of substanc e abuse service accessibility on the relationship between child maltreatment and neighborhood alcohol availability. Children and Youth Services Review , 35(12), 1933Ð1940. doi: 10.1016/j.childyouth.2013.09.019 Morton, C. M., Simmel, C., & Peterson, N. A. (2014). Neighborhood alcohol outlet density and rates of child abuse and neglect: Moderating effects of access to substance abuse services. Child Abuse & Neglect, 38 (5), 952Ð961. doi: 10.1016/j.chiabu.2014.01.002 Muthen, B. O. (1990). Mean and covariance structure analysis of hierarchical data. Retrieved from http://www.statmodel.com/bmuthen/articles/Article_032.pdf Muthen, B. O. (1994). Multilevel covariance structure analysis. Sociological Methods and Research, 22, 376Ð398. doi: 10.1177/004 9124194022003006 Muth”n, L. K., & Muth”n, B. O. (2012). Mplus user's guide . Los Angeles, CA: Muth”n & Muth”n. Osterling, K. L., & Han, M. (2011). Reunification outcomes among Mexican immigrant families in the child welfare system. Children and Youth Services Review , 33(9), 1658Ð1666. doi: 10.1016/j.childyouth.2011.04.020 !!!!96 Perry, H. B., Zulliger, R., & R ogers, M. M. (2014). Community health workers in low -, middle -, and high -income countries: An overview of their history, recent evolution, and c urrent effectiveness. Annual Review of Public Health, 35 (1), 399Ð421. doi:10.1146/annurev -publhealth -032013-182354 Phinney, J. S., Ong, A., & Madden, T. (2000). Cultural values and intergenerational value discrepancies in immigrant and non -immigrant families. Child Development , 71(2), 528Ð539. doi: 10.1111/1467 -8624.00162 Pine, B. A., & Drachman, D. (2005). Effective child welfare practice with immigrant and refugee children and their families. Child Welfare, 84, 537Ð562. Polansky, N. A., Ammons, P. W., & Gaudin , J. M. (1985). Loneliness and isolation in c hild neglect. Social Casework, 66 (1), 38Ð47. doi:10.1177/104438948506600105 Preacher, K. J., Zyphur, M. J., & Zhang, Z. (2010). A general multilevel SEM framework for assessing multilevel mediation. Psychological Met hods, 15 (3), 209Ð233. doi: 10.1037/a0020141 Putnam -Hornstein, E., & Needell, B. (2011). Predictors of child protective service contact between birth and age five: An examination of California's 2002 birth cohort. Children and Youth Services Review , 33(8), 1337Ð1344. doi: 10.1016/j.childyouth.2011.07.010 Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage. Raykov, T. (2011) Intraclass correlation coefficients in hierarchical designs: Evaluation using latent variable m odeling, Structural Equation Modeling, 18 (1). 73Ð90, doi: 10.1080/10705511.2011.534319 Reichman, N. E., Teitler, J. O., Garfinkel, I., & Mc Lanahan, S. S. (2001). Fragile f amilies: Sample and design. Children and Youth Services Review, 23 (4Ð5), 303Ð326. doi: 10.1016/s0190 -7409(01)00141-4 Rhee, S., Chang, J., Berthold, S. M., & Mar, G. (2012). Child maltreatment among immigrant Vietnamese families: Characteristics and implications for practice. Child and Adolescent Social Work Journal , 29(2), 85Ð101. doi: 10.1007/s10560 -011-0253-x Rhee, S., Chang, J., Weaver, D., & Wong, D. (2008). Child Maltreatment among immigrant Chinese families: Characteristics and patterns of p lacement. Child Maltreatment, 13 (3), 269Ð279. doi:10.1177/1077559507313461 Sabol, W. J., Coulton, C. J., & Korbin, J. E. (2004). Building community capacity for violence prevention. Journal of Interpersonal Violence, 19 (3), 322Ð340. doi:10.1177/0886260503261155 !!!!97 Sampson (2001). How do communities undergird or undermine human development? Relevant contexts and social mechanisms. In A. Booth & A. C. Crouter (Eds .), Does it take a village? Community effects on children adolescents and families (pp. 3 Ð30). Mahwah, NJ: Lawrence Erlbaum Associates. Sampson, R. J. (1988). Local friendship ties and community attachment in mass society: A multilevel systemic model. American Sociological Review 53(5), 766Ð779. doi:10.2307/2095822 Sampson, R. J., & Groves, W. B. (1989). Community struc ture and crime: Testing social disorganization theory. American Journal of Sociology , 94(4). 774Ð802. Sampson, R. J., Morenoff, J. D., & Earls, F. (1999). Beyond social capital: Spatial dynamics of collective efficacy for children. American Sociological R eview, 633Ð660. Sampson, R. J., Morenoff, J. D., & Gannon -Rowley, T. (2002). Assessing neighborhood effects: Social processes and new directions in research. Annual Review of Sociology , 28, 443Ð478. doi: 10.1086/229068 Sampson, R. J., Raudenbush, S. W., & E arls , F. (1997). Neighborhoods and violent c rime: A multilevel s tudy of collective e fficacy. Science , 277(5328): 918 Ð924. doi:10.1126/science.277.5328.918 Segal, U. A. (2000). Exploring child abuse among Vietnamese r efugees. Journal of Multicultural Social Work, 8 (3Ð4), 159Ð191. doi: 10.1300/j285v08n03_01 Segal, U. A., & Mayadas, N. S. (2005). Assessment of issues facing immigrant and refugee families. Child W elfare , 84(5). 564Ð566. Shaw, C. R., & McKay, H. D. (1942). Juvenile delinquency and urban areas . Chicago, IL: University of Chicago Press. Skogan, W. (1986) . Fear of crime and neighborhood c hange. Crime and Justice, 8 , 203Ð229. doi:10.1086/449123 Spearly, J. L., & Lauderdale, M. (1983). Community characteristics and ethnicity in the prediction of chi ld maltreatment rates. Child Abuse & Neglect, 7 (1), 91Ð105. doi:10.1016/0145 -2134(83)90036-4 Straus, M.A., Hamby, S.L., Finkelhor, D., Moore, D.W., & Runyan, D. (1998). Identification of child malt reatment with the Parent -Child C onflict Tactics Scales: Dev elopment and psychometric data for a national sample of American parents. Child Abuse & Neglect, 22(4), 249Ð270. doi: 10.1016/s0145 -2134(97)00174-9 Tajima, E. A., & Harachi, T. W. (2010). Parenting beliefs and physical discipline practices among Southeast Asian immigrants: Parenting in the context of cultural adaptation to the United States. Journal of Cross -Cultural Psychology , 41(2), 212Ð235. doi: 10.1177/0022022109354469 !!!!98 U.S. Bureau of the Census. (2000). Geographic areas reference manual : Census tracts a nd block numbering areas. Washington, DC: U.S. Department of Commerce, Bureau of the Census. Retrieved from https://www2.census.gov/geo/pdfs/reference/GARM/Ch10GARM.pdf U.S. Census Bureau. (2017a ). American commun ity survey 1 -Year estimates 2017: A ge and nativity of own children under 18 years in families and subfamilies by number and nativity of parents. Retrieved from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_ 14_1YR_B05009&prodType=table U.S. Census Bureau. (2017b). American community survey 5 -year estimates: Selected characteristics of the native and foreign -born populations 2013 -2017. Retrieved from https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_ 17_5YR_S0501&prodType= table U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, ChildrenÕs Bureau. (2018). Child maltreatment 2016. Retrieved from https://www.acf.hhs.gov/cb/research -data -technolo gy/statistics -research/child -maltreatment U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, ChildrenÕs Bureau. (2010). The child abuse prev ention and treatment act 2010: Incl uding adoption opportunities & t he abandoned infants assistant act. Retrieved from https://www.acf.hhs.gov/sites /default/files/cb/capta2010.pdf Vinson, T., Baldry, E., & Hargreaves, J. (1996). Neighbourhoods, networks and child abuse. British Journal o f Social Work, 26 (4), 523Ð543. doi: 10.1093/oxfordjournals.bjsw.a011123 Walters, E.E., Kessler, R.C., Nelson, R.C., & Mroczek, D. (2002). Scoring the World Health Organization's Composite International Diagnostic Interview Short Form (CIDI -SF; Dec 2002). Weissman, A. M., Jogerst, G. J., & Dawson, J. D. (2003). Community characteristics associated with child abuse in Iowa. Child Abuse & Neglect, 27 (10), 1145Ð1159. doi:10.1016/j.chiabu.2003.09.002 World Health Organization Composite International Diagnostic Interview Short - Form (CIDI -SF). International Journal of Methods in Psychiatric Research, 7, 171Ð185. doi: 10.1002/mpr.47 World Health Organization. (1989). Strengthening the performance of community health workers in primary health care: Report of a WHO Study Group . Retrieved from https://apps.who.int/iris/bitstream/handle/10665/39568/WHO_TRS_780.pdf?sequence=1 &isAllowed=y !!!!99 Xu, Y., Wang, X., Ahn, H., & Harrington, D. (2018). Predictors of non -US born mothers' parenting stress across early childhood in fragile families: A longitudinal analysis. Children and Youth Services Review , 89, 62Ð70. doi: 10.1016/j.childyouth.2018.04.012 Young, G., & Gately, T. (1988). Neighborhood i mpoverishmen t and child m altreatment. Journal of Family Issues, 9 (2), 240Ð254. doi:10.1177/019251388009002006 Zuravin, S. J. (1989). The ecology of child abuse and neglect: Review of the literature and presentation of data. Violence and Victims , 4, 101Ð120.