554422 research-article2014

JIVXXX10.1177/0886260514554422Journal of Interpersonal ViolenceEmery et al.

Article

Protective Informal Social Control of Child Maltreatment and Child Abuse Injury in Seoul

Journal of Interpersonal Violence 2015, Vol. 30(18) 3324­–3339 © The Author(s) 2014 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260514554422 jiv.sagepub.com

Clifton R. Emery,1 Tatiana Eremina,1 Hye Lin Yang,1 Changgeun Yoo,1 Jieun Yoo,1 and Ja Kyung Jang1

Abstract Previous findings on the relationship between neighborhood informal social control and child abuse have been mixed. We implemented a scale created by Emery, Trung, and Wu to study protective informal social control of child maltreatment (ISC_CM) by neighbors in a three-stage random cluster sample of 541 families in Seoul, South Korea. Random-effects regression models found that protective ISC_CM significantly moderated the relationship between very severe abuse and child injuries. Very severe abuse was associated with fewer injuries when levels of protective ISC_CM were higher. Implications are discussed. Keywords child abuse, injury, informal social control, ISC_CM

Introduction Proverbially, good neighbors are there when you need help. Intuitively and anecdotally, whether and how neighbors respond to child maltreatment

1Yonsei

School of Social Welfare, Seoul, Korea

Corresponding Author: Clifton R. Emery, Associate Professor, Yonsei School of Social Welfare, 134 Shinchon-dong, Seodaemun-gu, Appenzeller Hall 214, Seoul 120-749, Korea. Email: [email protected]

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3325

Emery et al.

and domestic violence can make a difference. If the popular media are indicative of mainstream opinion (cf. Davidson, 2013), we want our neighbors to intervene in cases of domestic violence and child maltreatment. Yet, with respect to effects of neighborhood response on child maltreatment, the jury on the empirical findings appears to remain in recess. In a recent article, Emery, Trung, and Wu (2013) suggest that the mixed findings on the neighborhood response to maltreatment result from a problem of measurement, and present results showing their informal social control of child maltreatment measure (ISC_CM) is associated with less very severe abuse and fewer abuse-related behavior problems. In this article, we refined Emery et al.’s (2013) ISC_CM measure and examined its relationship with very severe abuse and abuse injury in a representative sample of Seoul, South Korea.

ISC_CM Informal social control as a theoretical construct derives from the social disorganization literature in sociology and criminology (Browning, 2002; Sampson, Raudenbush, & Earls, 1997). As opposed to formal social control, informal social control consists of actions undertaken by ordinary people to prevent crime and achieve public order (Sampson et al., 1997). Although consistently related to lower rates of homicide and other crimes (cf. Browning, Feinberg, & Dietz, 2004; Sampson et al., 1997), findings on the role of informal social control with respect to child maltreatment remain mixed. Guterman, Lee, Taylor, and Rathouz (2009) found that perceived informal social control by neighbors is associated with less physical child abuse. Yonas et al. (2010) found that perceived informal social control by neighbors moderates the relationship between neglect and externalizing, but did not find this moderating effect for physical or sexual abuse and externalizing. Using the Project on Human Development in Chicago Neighborhoods data, Molnar, Buka, Brennan, Holton, and Earls (2003) found no relationship between neighborhood informal social control and physical child abuse. Emery et al. (2013) argued that these mixed findings may have occurred from a combination of inappropriate measurement and random sampling of neighborhoods. The instrument typically used, collective efficacy, measures its informal social control component as follows: You could count on your neighbors to do something if (a) “children were skipping school and hanging out on a street corner,” (b) “children were spray-painting graffiti on a local building,” (c) “a child was showing disrespect to an adult,” (d) “there was a fight in front of your house and someone was being beaten or threatened,”

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3326

Journal of Interpersonal Violence 30(18)

and (e) “because of budget cuts, the fire station closest to your house was going to be closed” (Sampson et al., 1997). Emery et al. (2013) claimed that the informal social control measure being used was appropriate for social control of disorder on the street, but that violence behind closed doors required a different measurement approach. To paraphrase their argument, imagine that there are three types of neighborhoods. In Type A neighborhoods, no one practices informal social control. In Type B neighborhoods, collective efficacy is high but traditional norms mean that neighbors think of family violence as a kind of informal social control, that is, as something necessary to keep family members in line. In Type C neighborhoods, collective efficacy is high but non-traditional norms mean that neighbors think of family violence as crime. In B neighborhoods, neighbors will act to prevent disorder on the street, but not behind closed doors. In C neighborhoods, neighbors will act to prevent disorder on the street and violence behind closed doors. Whether or not one finds the collective efficacy informal social control measure to be significantly related to child maltreatment will depend on whether more neighborhoods of the C type than the B type are randomly selected into the sample. To more accurately measure ISC_CM, Emery et al. (2013) developed the ISC_CM scale and implemented the new measure in a random sample of households, representative of Hanoi, Vietnam. The ISC_CM scale captured two types of informal social control, protective and punitive. Protective ISC_CM constituted actions recognizably aimed at protecting the child (e.g., getting in between the parent and child and trying to calm the parent down by talking). Punitive ISC_CM contained actions that were less obviously aimed at protecting the child and had some component of threat of punishment for the parent (e.g., threatening the parent or calling the police). Emery et al. (2013) found that protective ISC_CM was associated with less very severe physical abuse (i.e., beating up, choking, burning on purpose, and threatening with a knife or a gun in the past year) for a sample of children aged 1 to 18 years. Although child’s age predicted externalizing and internalizing behavior problems, it did not predict very severe abuse. Emery et al. (2013) also found that protective ISC_CM moderated the relationship between very severe abuse and the child’s externalizing behavior problems. The relationship between very severe abuse and externalizing behavior problems was weaker when protective ISC_CM was higher. That is, the main effect of very severe abuse was positive, but the interaction term with ISC_CM was negative. Emery et al. (2013) found no relationship between punitive ISC_CM and very severe abuse or child behavior problems.

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3327

Emery et al.

Appropriateness of South Korea as a Venue to Extend Study of ISC_CM South Korea and Vietnam share many similarities. Both were protectorates of China for many centuries and later freed themselves from that influence. Chinese influence in Vietnam and Korea occurred at similarly ancient times, such that many words in Vietnamese and Korean, such as the word for university (dae hak in Korean, dai hoc in Vietnamese, da xue in Mandarin), bear more similarity to one another than either does to Mandarin Chinese. This is somewhat remarkable, given that they exist at opposite ends of the Chinese land mass. Both South Korea and Vietnam are characterized by Confucian cultures. Given strong cultural similarities, it seems reasonable to examine whether protective ISC_CM may play similarly protective roles in South Korea. Child maltreatment remains a serious problem in South Korea. Emery, Kim, Song, and Song (2013) report a last-year prevalence rate of severe father-to-child physical violence (kicked or punched, hit with an object, beat up, and used or threatened to use a knife or a gun) at 19.4%, and any fatherto-child violence at 30.7%. In their retrospective study of child abuse among South Korean college students, Lee and Kim (2011) found that 42.2% of students recalled physical abuse (hit or punched, kicked, beaten with object, shaken, cut, or stabbed). Rates of physical abuse are higher for elementaryschool-aged children in South Korea than elsewhere, possibly because of strong parent–child identification combined with strong cultural emphasis on success in formal education (Hahm & Guterman, 2001; Kim, 1998). Given the cultural similarities and high prevalence rates of child abuse, we chose South Korea to replicate and extend Emery et al.’s (2013) findings. Based on the theoretical logic of informal social control, we formed the following hypotheses: Hypothesis 1: Protective ISC_CM by neighbors will be negatively associated with very severe abuse (replicates Emery et al., 2013). Hypothesis 2: Protective ISC_CM by neighbors will moderate (negative sign for the interaction term) the relationship between very severe abuse and child injury from physical abuse (extends Emery et al., 2013).

Model Figure 1 shows the hypotheses to be tested. A random-effects regression model will test Hypotheses 1 and 2 (blue arrow: main effects model, red arrow: interaction effects model).

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3328

Journal of Interpersonal Violence 30(18)

H2 Protective ISC_CM H1 1 Very Severe Abuse

Injury from Abuse

Figure 1.  The role of protective ISC_CM.

Note. ISC_CM = informal social control of child maltreatment.

Method Data The Seoul Families and Neighborhoods Study (SFNS) is a representative random probability three-stage cluster sample of 541 families in Seoul, South Korea, collected in 2012. It is part of the Families and Neighborhoods Study (FNS), an ongoing seven-city global study of informal social control of domestic violence and child maltreatment. Interviewers for the Seoul sample were recruited from the university graduate program, and were required to complete a 2-day training, written exam, and certification interview designed by the first author. The Seoul sample was drawn by probability proportional to size (PPS) sampling of 34 districts (dong) in Seoul, followed by sampling within each district (about 15 families per dong). Within-district sampling was carried out using district maps. Two random draws from the uniform distribution were used to locate a random start-point on the map. Google Earth was used to make sure that the start-point was actually near buildings rather than sitting in the middle of a forest or lake. An interviewer was dispatched to the start-point, instructed to find the nearest residential structure(s) to the start-point, and conduct interviews with about 15 households (in each of 34 districts). A refusal conversion protocol was used to increase the overall response rate from 54% to 63%. One adult was interviewed in each household. Because the FNS is also a study of intimate partner violence (IPV), to be eligible for the study, the respondent had to have been in a marriage or co-habiting relationship within the past year. Oral informed consent was obtained from participants, and contact information (household location) was stored in a different file entered data. Completed surveys are stored in a locked cabinet in the Primary Investigator’s

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3329

Emery et al. Table 1.  Sample Descriptive Statistics. Variable Protective ISC_CM Very severe physical abusea Any abuse injuryb Injuries Neighborhood solidarity Neighborhood social control Any father IPVc R female Husband age Husband education Jutaek Years resident Family size Household income (KRW1,000) Child female Child’s age

N

M

SD

509 509 502 502 509 509 479 506 483 482 458 466 500 509 456 452

9.11 7.16% 8.49% 0.1 8.46 7.89 12.17% 85.77% 46.94 14.45 58.08% 7.66 2.78 4,452 48.90% 12.04

5.20 0.29 0.24 0.46 2.64 2.37 0.32 0.35 8.97 2.66 0.49 6.60 0.87 2340 0.50 7.75

Note. ISC_CM = informal social control of child maltreatment; IPV = intimate partner violence; R female = Respondent female. Jutaek refers to a high rise Korean house.aAnswer shown is weighted for survey nonresponse. Unweighted very severe abuse rate was 6.00%. bAnswer shown is weighted for survey non-response. Unweighted injury rate was 8.19%. cAnswer shown is weighted for survey non-response. Unweighted any abuse by husband rate was 11.46%.

office. The data file has been stripped of identifying information, and is stored in a password-protected computer. Although 541 families were sampled, measures for protective ISC_CM were introduced after data collection began. For this reason, inferential analyses can only be conducted on 509 families. Descriptive statistics of the sample are reported in Table 1. The data show that the average monthly household income for the sample was KRW4,452,000 (US$4,210). As the annual household income for Seoul was KRW48,500,000 (T. K. Lee, Statistics Korea, personal communication, May 23, 2013; US$3,683 per month), this means that the sample reported higher monthly income than the Seoul average (t = 3.95, p < .001). This may be because the sample was on average better educated than the Seoul average. According to the Organisation for Economic Co-operation and Development (OECD; 2013), 80% of South Koreans have high school diploma equivalents or more. However, in our sample, 96% of husbands and 92% of wives had high school level or higher education (Z = 8.9, p < .001; Z = 6.7, p < .001). The average

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3330

Journal of Interpersonal Violence 30(18)

household size was 2.8 and the family had lived in the neighborhood for 7.7 years on average; 42% lived in high-rise apartments. The sample was 86% female. The gender imbalance probably occurred both via selective refusals (women were more likely to participate) and because not every interviewer adhered to the dictum to collect data on evenings and weekends. It is thus likely that the sample has a disproportionately high proportion of stay-athome mothers. Indeed, 31.7% of the mothers in the sample were not employed. Among mothers with college degrees or more, the unemployment rate was 33.8%. This was not significantly different for the overall unemployment rate of women with tertiary education in Korea (“South Korea Posts Lowest Employment Rate,” 2010; Z = 1.55, p = .12). This slightly higher unemployment rate for college-educated women underscores the fact that education and employment are not positively correlated among women in South Korea (Kim & Hoppe-Graff, 2001).

Measures Very severe abuse.  Three items were used from the parent–child Conflict Tactics Scale (CTS; Straus & Hamby, 1997) to capture very severe abuse. These were as follows: (a) beat him or her up, (b) grabbed him or her around the neck and choked him or her, and (c) burned or scalded him or her on purpose. Using or threatening to use a knife or a gun was removed and replaced with the injury items provided below. The respondent (parent) was asked about his or her own abuse of the child. Internal reliability for these three items was acceptable (Cronbach’s α = .76). Child abuse injury.  Injury was captured with two items: (a) child had to meet a doctor (MD) because of the punishment the parent gave and (b) child had a bruise because of the punishment the parent gave. Only two parents in the sample reported taking their child to a MD in the last year because of punishment. However, these two cases were added to the bruise report to form a scale for number of injuries. Protective ISC_CM. The original measure of protective ISC_CM used by Emery et al. (2013) had two items: “If they witnessed me using severe violence against my child, my neighbors might”: (a) get in-between my child and me and (b) try calm me down by talking. To these two items, we added another two items. These were as follows: My neighbors would (c) take my child somewhere else until I calm down and (d) ask another member of my family to help calm me down. Possible answers were my neighbors (a) would never do this, (b) might do this, (c) would probably do this, (d) would

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3331

Emery et al.

definitely do this, and (e) actually did this. A scree plot of the eigenvalues from principal components analysis of these items showed a sharp elbow after the first eigenvalue. This is consistent with the idea that they capture a one-dimensional construct. The modified protective ISC_CM also had good internal reliability (Cronbach’s α = .83). The four items were combined into a scale ranging from 1 to 5, such that the score for each individual was the average across all four items. This scoring system allows us to assess the effect of an average increase of one category across all four items. Father’s IPV.  The correlation between IPV by fathers and child abuse is well established (cf. Appel & Holden, 1998; Emery, Kim, et al., 2013), and fathers’ IPV may also be related to informal social control. The measure of violence by the father’s IPV is a modified version of the Straus and Douglas (2004) CTS Short Form (CTS2S), which has been found to have good concurrent validity with the longer form, the CTS2 (Straus & Douglas, 2004). It includes six physical violence items: (a) slapped; (b) pushed, grabbed, or shoved; (c) hit with object; (d) punched, kicked, or bit; (e) beat up; and (f) used or threatened with a knife or a gun. Possible responses were once, twice, 3 to 5 times, 6 to 10 times, 11 to 20 times, or more than 20 times in the past year. These items were condensed into a single, father IPV severity in the last-year control variable. The CTS2S had good internal reliability (Cronbach’s α = .86). To create the severity scale, each item was weighted by its log-odds of producing wife injury as estimated by bivariate logistic regressions (cf. Emery et al., 2013). Perceived collective efficacy.  Based on Sampson et al. (1997) and Zhang, Messner, and Liu (2007), collective efficacy was measured using two scales: Neighborhood Solidarity and Neighborhood Informal Social Control. Solidarity was measured by agreement with four items: (a) This is a close-knit neighborhood; (b) if your family has an important problem, people around here care; (c) people in this neighborhood can be trusted; and (d) people around here are willing to help their neighbors (Cronbach’s α = .88). Informal Social Control was measured with four items: You could count on your neighbors to do something about it if (a) children were skipping school and hanging around outside, (b) children were showing disrespect to an adult, (c) there was a fight in front of your house/apartment, or (d) you were away and someone was trying to steal your bike (Cronbach’s α = .84). These items were combined into scales ranging from 1 to 4 such that the score for each individual was the average across all four items. Control variables.  We controlled for the number of years resident in the neighborhood, housing type (high-rise apartment complex or not), household size,

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3332

Journal of Interpersonal Violence 30(18)

household income, sex of respondent, father’s age and education, and the child’s age and sex.

Analytic Issues The cluster design of the sample requires corrections for clustering in the standard errors of linear models. Random-effects regression models were hence used. These linear models estimate a random effect for each neighborhood, thus accounting for clustering in the data without using 33 degrees of freedom that would be necessary to control for each neighborhood (Hsiao, 2003; Johnston & DiNardo, 1997). Models were tested using Stata11. Model diagnostics were run using ordinary regression models to allow the models to be thoroughly vetted. Two variance inflation factors (VIFs) for the injury regression model were very large. The VIF for very severe abuse was 79.8 and the VIF for the Protective informal social control × Very severe abuse interaction was 69.2. This caused some concern about the stability of the estimates, particularly as the father IPV severity term was not significant and its sign was opposite to the predicted direction. However, as the interaction term is the chief source of interest, it is not useful to run the model without it. Removing the father IPV severity term from the model reduced the two largest VIFs to 67.3 (very severe abuse) and 67.0 (Very severe abuse × Protective informal social control).

Results Among the children in the sample, 5.8% had been very severely abused in the last year; 8.2% had been injured. When the non-response sample is used to correct for non-response bias, this number rises to 7.2% very severely abused and 8.5% injured. Any physical violence by husbands against wives was 11.5%. Correcting for non-response bias yields a 12.2% rate for any husband violence in the last year. Considering that the sample is wealthier and better educated than average, these estimates are probably minima. The average response on items for protective ISC_CM fell between neighbors might do this and neighbors would probably do this. Table 2 shows the results from the random-effects regression of frequency of very severe abuse on the model. Protective ISC_CM was not associated with very severe abuse frequency. In addition, the coefficient was positive, rather than negative as predicted by Hypothesis 1. Father IPV severity was significantly associated with very severe abuse (p < .001). A one-unit increase in father IPV severity was associated with a 0.17 increase in acts of very severe abuse.

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3333

Emery et al.

Table 2.  Very Severe Physical Abuse Random-Effects Regression Model Results (N = 343). Variable Protective ISC_CM Father IPV severity Neighborhood solidarity Neighborhood social control Female Husband age Husband education Child age Child male Family size Household income (KRW100,000) Years resident High-rise apartment

b

SE(b)

0.004 0.17*** −0.01 0.01 −0.18 −0.01 0.05* 0.004 0.04 0.01 −0.004 −0.01 −0.11

0.01 0.01 0.02 0.02 0.13 0.01 0.02 0.01 0.09 0.06 0.002 0.01 0.10

Note. ISC_CM = informal social control of child maltreatment; IPV = intimate partner violence. *p < .05. ***p < .001.

Table 3 shows the random-effects regression model for abuse-caused injury. As is to be expected, very severe abuse is strongly and positively associated with abuse-related injuries. Parents reporting very severe abuse reported, on average, 1.8 more abuse-related injuries in the past year than parents who did not (p < .001). The significant negative interaction effect for Protective ISC_ CM × Very severe abuse indicates that when levels of protective ISC_CM are higher, the impact of very severe abuse on injuries is smaller (p < .001). For example, the results imply that when very severe abuse occurs in the context of a protective ISC_CM score of 4 (would never do this for all four items), an additional act of very severe abuse is associated with about (0.62 − 0.14 =) 0.48 more injuries per year. However, when protective ISC_CM is 16 (would definitely do this for all four items), additional acts of very severe abuse are only associated with (0.62 − 0.57 =) 0.05 more injuries per year. Somewhat surprisingly, the coefficient for father IPV severity was not significant (p = .44). This anomaly will be further discussed in the discussion section. Higher household income was associated with fewer child abuse injuries (p < .05).

Discussion There are more than 2.1 million children living in this city of 10.4 million (Seoul Metropolitan Government, 2013). A 7.2% very severe abuse rate Downloaded from jiv.sagepub.com by guest on November 15, 2015

3334

Journal of Interpersonal Violence 30(18)

Table 3.  Injuries Random-Effects Regression Model Results (N = 341). Variable Very severe abuse Protective ISC_CM Protective ISC_CM × Very severe abuse Father IPV severity Neighborhood solidarity Neighborhood social control Female Husband age Husband education Child age Child female Family size Household income (KRW100,000) Years resident High-rise apartment

b

SE(b)

0.62*** 0.005 −0.04*** −0.04 −0.01 0.01 − 0.01 0.001 −0.0002 −0.004 −0.02 −0.03 −0.002† −0.002 −0.02

0.12 0.003 0.01 0.08 0.01 0.01 0.05 0.004 0.01 0.004 0.03 0.02 0.001 0.003 0.04

Note. ISC_CM = informal social control of child maltreatment; IPV = intimate partner violence. *p < .05. **p < .01. ***p < .001. †p < .10.

means that at least 150,000 Seoul children are severely beaten, choked, or burned by their parents each year. An 8.5% injury rate means that at least 178,000 Seoul children are bruised or injured, so as to need a doctor each year. These numbers are conservative estimates because the sample composition is wealthier and more educated than average. Populations with lower socioeconomic status (SES) tend to have higher abuse rates. The true rates are probably higher.

Protective ISC_CM as a Protective Factor Hypothesis 1 postulated that protective ISC_CM would be associated with less very severe abuse, and Hypothesis 2 postulated that protective ISC_CM would be associated with a protective effect against injury when very severe abuse did occur. The tables present no support for Hypothesis 1 but reasonable support for Hypothesis 2. One difficulty that occurs in estimating the relationship between protective ISC_CM and very severe abuse is reverse-causality bias. That is, protective informal social control by neighbors may in fact deter very severe abuse, but at the same time very severe abuse may be more likely to elicit a neighbor response. Hence, results can be biased in the direction of finding a positive, rather than a negative relationship (as occurred in our Downloaded from jiv.sagepub.com by guest on November 15, 2015

3335

Emery et al.

results). Better statistical controls, more statistical power, or a randomized experiment could assist in better assessing this relationship. The findings are further supported by analyses of simpler models. Simpler models (with fewer controls) consistently show protective ISC_CM to be associated with fewer abuse-related injuries when very severe abuse does occur. In light of the replication of the protective ISC_CM findings from Vietnam (Emery et al., 2013), we conclude that the evidence in support of the hypothesis on protective ISC_ CM when very severe abuse occurs (Hypothesis 2) is reasonable. One apparent anomaly in the findings is a non-significant, negative relationship between father IPV severity and child abuse injuries. We were concerned that this may have occurred due to problems of stability in the model, because the relationship is positive and significant in simpler models. However, when we examined the VIFs for the model, father IPV severity did not seem to contribute substantially to multicollinearity in the model (VIF = 3.44). Closer investigation reveals that the sign flips from positive and significant to negative and insignificant when husband’s education is controlled. The explanation may be that in more educated households with high levels of domestic violence, mothers intervene or strategize to protect their children from injurious violence by the father. Although tentative, the findings for the Protective ISC_CM × Very severe abuse interaction effect suggest that soft ISC_CM may be a protective factor against injury from very severe abuse in South Korea. The significant interaction term consistent with Hypothesis 2 suggests that protective ISC_CM by neighbors may help protect abused children’s physical health as well as the emotional health indicators seen in the initial study (Emery et al., 2013). Additional studies of protective ISC_CM are needed for larger samples of different populations and for different sequelae of abuse and different kinds of maltreatment. Likewise, studies of the antecedents of protective ISC_CM are needed to determine what factors might help augment protective neighbor intervention against child maltreatment. Results from these studies then need to be implemented in intervention design. For example, neighborhood interventions could be designed to boost ISC_CM. Taken together with Emery et al.’s (2013) findings, we believe that our findings warrant continued investigation of protective ISC_CM as a potential means of protecting more children from severe physical child abuse and its consequences. The findings also bear on the broad theoretical literature on informal social control. Our findings suggest that, with respect to child abuse, it is not the construct of informal social control that is flawed, but its measurement. We suspect our own null findings for collective efficacy and those of others (cf. Molnar et al., 2003) can be explained in terms of measurement error. Sampson et al. (1997) argued that there were two critical characteristics

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3336

Journal of Interpersonal Violence 30(18)

necessary for collective efficacy: neighborhood solidarity and informal social control. It may be worth examining the relationship between neighborhood solidarity and protective ISC_CM. Likewise, more research on ISC_CM as practiced by extended family members and friends, as well as neighbors, is needed. Increasing the responsiveness of multiple networks that surround abusive families may have synergistic effects.

Limitations There are a number of limitations that affect the strength of the conclusions that can be drawn from these analyses. First, the study is non-experimental, cross-sectional, and subject to parent reporting biases. Hence, findings are associations and not necessarily causal. However, injury, as something physically visible on a child’s body, may be somewhat less susceptible to parental response bias than the child behavior problems studied by Emery et al. (2013). Hostile attribution bias is, however, still a competing explanation that future research must rule out. Parents with hostile attribution bias may be less likely to attribute protective ISC_CM to neighbors and more likely to severely abuse and injure their children. Our results can only be generalized to Seoul. Another problem with the protective ISC_IPV model is reverse-causality bias. That is, protective ISC_ CM may have a causal impact on very severe abuse, but very severe abuse may also elicit ISC_CM. However, this bias can be said to make the findings conservative. Although the causal relationship between protective ISC_CM and very severe abuse should, in theory, be negative, the causal relationship between very severe abuse and protective ISC_CM could only logically be said to be positive. The presence of a positive reverse-causality loop would thus dampen regression estimates of the impact of protective ISC_CM on very severe abuse. This would positively bias a (true) negative regression coefficient toward values greater than 0. Hence, the fact that the coefficient is positive indicates the need for approaches that appropriately handle this issue. One effective approach would be a randomized experiment involving ISC_CM by neighbors.

Conclusion Our findings suggest that protective ISC_CM by neighbors may be a protective factor against very severe abuse and abuse-related injury. The fact that the findings from Emery et al. (2013) were extended in Seoul suggests both that their findings were probably not a random fluke, and that protective ISC_ CM may be a relevant protective factor in Korea as well as Vietnam. Our

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3337

Emery et al.

findings, together with those of Emery et al. (2013), arguably provide converging evidence for the intuitive position that when it comes to child abuse, what our neighbors do matters. Some findings bear repeating. By our estimate, more than 150,000 children in Seoul are beaten, choked, or burned by their parents each year. Nearly 180,000 children in Seoul are injured from physical child abuse. Around the world, we estimate that roughly 38,0001 children die every year from child abuse and neglect. If we are to be able to look ourselves in the mirror, we must find ways to reduce the number of children injured and killed each year by parents and caregivers. Finding ways to increase informal social control by neighbors may be part of the solution. Acknowledgment We gratefully acknowledge the H. F. Guggenheim foundation for major support.

Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding The author(s) disclosed receipt of the following financial support for the research and/ or authorship of this article: This research was made possible by a major grant from the H.F. Guggenheim Foundation. Some research assistance was funded by the National Research Foundation grant funded by the Korean government (MEST No. 616-2011-2-B000009) and the Yonsei 2010 research grant. We also thank Dr. Gary Melton for suggestions that have improved the protective ISC_CM scale.

Note 1.

This is based on a direct extrapolation of the U.S. rate, 2.10 per 100,000 children (Child Welfare Information Gateway, 2013) to the global population of 1.8 billion children aged 0 to 14 years (U.S. Central Intelligence Agency [CIA], 2013). We believe that this is reasonable, because the U.S. rate is likely to be close to the middle. The death rate from abuse and neglect is high among developed nations (Dowdy, 2003), but is likely to be low among developing countries. It would also be possible to extrapolate percentage of child homicides that are committed by parents (57% for under 12-year-olds, Dawson & Langan, 1994), but this percentage seems likely to vary wildly by developmental status and the extent to which child protective services are available.

References Appel, A., & Holden, G. (1998). The co-occurrence of spouse and physical child abuse: A review and appraisal. Journal of Family Psychology, 12, 578-599.

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3338

Journal of Interpersonal Violence 30(18)

Browning, C. (2002). The span of collective efficacy: Extending social disorganization theory to partner violence. Journal of Marriage and Family, 64, 833-850. Browning, C., Feinberg, S., & Dietz, R. (2004). The paradox of social organization: Networks, collective efficacy, and violent crime in urban neighborhoods. Social Forces, 83, 503-534. Child Welfare Information Gateway. (2013). Child abuse and neglect fatalities 2011: Statistics. Washington, DC: Children’s Bureau. Retrieved from http://www. childwelfare.gov/pubs/factsheets/fatality.pdf U.S. Central Intelligence Agency. (2013). The world factbook. Washington, DC: Author. Retrieved from https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html Davidson, A. (2013, May 7). What Charles Ramsey and Amanda Berry knew. The New Yorker. Retrieved from http://www.newyorker.com/news/amy-davidson/ what-charles-ramsey-and-amanda-berry-knew Dawson, J., & Langan, P. (1994). Murder in families (Special Report). Washington, DC: Bureau of Justice Statistics, Office of Justice Programs, U.S. Department of Justice. Dowdy, Z. (2003, September 19). U.S. among nations high in child abuse deaths. Indianapolis Star, p. A9. Emery, C., Kim, J., Song, H., & Song, A. (2013). Child abuse as a catalyst for wife abuse? Journal of Family Violence, 28, 141-152. Emery, C., Trung, H., & Wu, S. (2013). Neighborhood informal social control and child maltreatment: A comparison of protective and punitive approaches. Child Abuse & Neglect. doi:10.1016/j.chiabu.2013.05.002. Guterman, N., Lee, S., Taylor, C., & Rathouz, P. (2009). Parental perceptions of neighborhood processes, stress, personal control, and risk for physical child abuse and neglect. Child Abuse & Neglect, 33, 897-906. Hahm, & Guterman, N. (2001). The Emerging problem of physical abuse in South Korea. Child Maltreatment, 6, 169-179. Hsiao, C. (2003). Analysis of panel data (2nd ed.). New York, NY: Cambridge University Press. Johnston, J., & DiNardo, J. (1997). Econometric methods. New York, NY: McGrawHill. Kim, H., & Hoppe-Graff, S. (2001). Mothers roles in traditional and modern Korean families: The consequences for parental practices and adolescent socialization. Asia Pacific Education Review, 2, 85-93. Kim, J. (1998). Hankuk gajung eei misungnym janyu pokryuk siltae ei kwan han yungu [A study of Korean family violence against children and adolescents]. Hankuk Adongbokjihak, 6, 41-64. Lee, Y., & Kim, S. (2011). Child maltreatment in South Korea: Retrospective study. Child Abuse & Neglect, 35, 1037-1044. Molnar, B., Buka, S., Brennan, J., Holton, J., & Earls, F. (2003). A multilevel study of neighborhoods and parent-to-child physical aggression: Results from the project on human development in Chicago neighborhoods. Child Maltreatment, 8(2), 84-97.

Downloaded from jiv.sagepub.com by guest on November 15, 2015

3339

Emery et al.

Organisation for Economic Co-operation and Development. (2013). OECD better life index: Korea. Retrieved from http://www.oecdbetterlifeindex.org/countries/ korea/ South Korea posts lowest employment rate for educated women among OECD countries. (2010, June 30). People’s Daily. Retrieved from http://english.peopledaily. com.cn/90001/90778/90858/90863/7047494.html Sampson, R., Raudenbush, S., & Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918-924. Seoul Metropolitan Government. (2013). Seoul statistics (population). Retrieved from http://stat.seoul.go.kr/jsp2/Octagon/jsp/WWS00/outer_Seoul.jsp?stc_cd=1 Straus, M., & Douglas, E. (2004). A short form of the revised conflict tactics scales, and typologies for severity and mutuality. Violence and Victims, 19(5), 507-520. Straus, M., & Hamby, S. (1997). Measuring physical and psychological maltreatment of children with the conflict tactics scales. In G. Kaufman Kantor & J. Jasinski (Eds.) Out of the darkness: Contemporary perspectives on family violence. Thousand Oaks, CA: Sage. Yonas, M., Lewis, T., Hussey, J., Thompson, R., Newton, R., English, D., & Dubowitz, H.. (2010). Perceptions of neighborhood collective efficacy moderate the impact of maltreatment on aggression. Child Maltreatment, 15, 37-47. Zhang, L., Messner, S., & Liu, J. (2007). A multilevel analysis of the risk of household burglary in the city of Tianjin, China. British Journal of Criminology, 47(6), 918-937.

Author Biographies Clifton R. Emery is an associate professor in the School of Social Welfare at Yonsei University in Seoul, South Korea. He obtained his master’s degree in statistics, MSW, and his PhD in social work from the University of Chicago. His research interests include informal social control of domestic violence and child abuse, understanding the effects of domestic violence on children, and theory development. Tatiana Eremina is a graduate of the Yonsei University School of Social Welfare master’s program. Her research interests include child maltreatment and the effects of immigration policy on diversity in Korean families. Hye Lin Yang is a doctoral student at the Yonsei University School of Social Welfare’s master’s program. Her research interests include the relationship between childhood sexual abuse and PTSD and PTSD in the North Korean defector population. Changgeun Yoo is a graduate student at the Yonsei University School of Social Welfare. Jieun Yoo is a graduate student at the Yonsei University School of Social Welfare. Her research interests center on the North Korean refugee crisis. Ja Kyung Jang is a graduate student at the Yonsei University School of Social Welfare.

Downloaded from jiv.sagepub.com by guest on November 15, 2015

Protective Informal Social Control of Child Maltreatment and Child Abuse Injury in Seoul.

Previous findings on the relationship between neighborhood informal social control and child abuse have been mixed. We implemented a scale created by ...
422KB Sizes 0 Downloads 6 Views