The CPS Workers’ Child Removal Decision in Cases of Domestic and Witnessed Violence

Abstract

This study aimed to explore maternal and paternal risks and protective factors that may influence the Child Protection Service (CPS) workers’ child removal decision in case of domestic and witnessed violence. In all, 218 case reports of Italian CPSs were retrospectively analyzed through the Protocol of Risk and Protective Factors. The sample was then split up into two groups on the basis of the CPS professionals’ placement decision after the investigation (child removal decision versus parents support and monitoring intervention). Two statistical approaches were used to identify the patterns of risk and protective factors associated with maternal and paternal assessments: logistic regression models and decision tree analysis. Results showed that mothers who are victims of Intimate Partner Violence experienced the child removal in about half of the cases, while fathers showed a higher removal rate. Differences emerged between mothers’ and fathers’ risk profiles, suggesting that workers attributed a different weight to some factors depending on whether they concerned the mother or the father. Only the proximal risk factor poor empathy skills was significant for both mothers and fathers. For the mothers’ group, one of the most important factors was the presence of direct forms of child maltreatment in addition to witnessed violence, while for fathers’ group the drug abuse emerged a crucial relevant proximal risk factor.

Keywords

parenting assessment domestic violence witnessed violence child maltreatment removal decision

Introduction

Witnessed violence has been only recently recognized as a primary form of child maltreatment on the basis of the severe short-term and long-term consequences on the child development (Save the Children, 2011; Trout, 1999). CISMAI 2017 (Italian coordination of services against child maltreatment and abuse) defines the intra-family witnessed violence as “child’s exposure to any form of abuse performed through acts of physical, verbal, psychological, sexual, economic violence and/or persecutory acts (stalking) against significant adults or minors. [. . .] The child or the adolescent can experience violence directly (when it occurs in his/her perceptual space) indirectly (when the minor is aware of the violence) and/or perceiving its physical, psychological, acute and chronic effects [. . .]” (p. 17). A growing body of research confirms that children exposed to domestic violence are more likely to show maladjustment correlates as externalizing and internalizing problems (Carpenter & Stacks, 2009; Kitzmann et al., 2003). Moreover, a substantial body of evidence shows a significant degree of co-occurrence between witnessed violence and other form of child maltreatment and family violence (Appel & Holden, 1998; Edleson, 1999; Rumm et al., 2000; Smith Slep & O’Leary, 2005), consistent with the poly-victimization model (Finkelhor et al., 2007). As revealed by Hamby et al. (2010), about one in three of youth who were exposed to intimate partner violence (IPV) were also victims of physical and psychological abuse; in addition, maltreatment that co-occurs with witnessed violence appears to be more severe than maltreatment in the absence of IPV.

Given this high rate, it is crucial for Child Protection Service (CPS) workers to perform a safety assessment to identify the children at risk among referred cases. As stressed by Hartley (2004, p. 388), managing effectively cases where domestic violence and child maltreatment co-occur is challenging, because of the “tension between the best’s interests of the children and the best’s interests of the mother.” A growing number of studies analyzed through different approaches the workers’ evaluation of parenting skills and the subsequent child-placement decision (Font & Maguire-Jack, 2015). For example, the decision-making ecology model (Baumann et al., 2011) proposes that case factors, individual factors, organizational factors, and external factors jointly influence workers’ decision-making processes and decisions. This study is specifically focused on the case factors that may exert a risk or protective role in predicting the child removal decision.

Hartley (2004) found that workers pointed out a higher concern for safety when domestic violence is present in addition to child maltreatment and tended to opt for child removal in a higher percentage of cases. Hughes and Chau (2012) highlighted also the relevance for the CPS that the mothers leave their abusive partner to reduce the risk of removal of their children. However, this CPS’s expectations may be in conflict—or at least partially divergent—to that of other systems, as the Family Law ones, that requires women to accept custody arrangements that provide close and continued contact between themselves and their former abusive partners (Hughes & Chau, 2012).

Literature suggested also that the evaluation of parents’ dangerousness and responsibilities may be influenced by stereotypes and gendered attributions about parental roles and practices (Strega et al., 2008). In fact, the strong cultural conceptions of mothers as the primary caretaker led to harsher judgment by the CPS system than fathers, who often remain “invisible” or uninvolved (Edleson, 1998; Mills, 2000). Often battered mothers were considered responsible for a failure to protect their children from the domestic violence exposure (Bancroft & Silverman, 2002; Strega, 2006), while fathers did not receive the same type of prosecution as revealed by some studies on Canadian and United States’ cases (Fugate, 2001; Lothian, 2002). Therefore, a considerable body of literature supported the “father absence and the mother blaming” in child welfare policy, practice, and discourse. Moreover, some authors (Brewsaugh & Strozier, 2016; Brewsaugh et al., 2018; Brown et al., 2009) underlie some risks in assessing and involving fathers in child welfare policies, given that perceptions of fathers tend to be more negative (Coakley et al., 2014). This seems to be due first to the effective prevalence of male offenders in cases of IPV and witnessed violence (World Health Organization, 2012); but also, biases, such as benevolent sexism, seem to play a significant role (Brewsaugh et al., 2018). Men tend to be identified with the violent perpetrators due to perceived gender differences, based on gender identity and aggressiveness stereotypes (by which the man is seen as more capable of arousing fear and injuring another person) (Russell et al., 2019). Crawford and Bradley (2016) stated that social workers may attribute the fathers the guilt of an unintentional physical abuse as resulting from a form of discipline that went too far, hence adhering to stereotypes about the division of gender roles in parenting.

In contrast to the substantial evidence, the best practices for child maltreatment cases involving domestic violence require CPS workers to “see double” focusing on both children’s and women’s needs (Fleck-Henderson, 2000). According to Hartley (2004), CPS workers especially need to “see double” in cases where mothers have maltreated their children. Therefore, on one hand, the presence of domestic violence should not negate a mother’s responsibility for the maltreatment when it is substantiated; on the other hand, this should not exclude her from receiving support and services to address the domestic violence.

Considering the child-placement decision, studies about cases referred for child maltreatment indicated that factors most associated with the separation from the mothers are as follows: substance misuse (Andrews et al., 2018), mental health difficulties, developmental limitations (Zuravin & DePanfilis, 1997), low level of education, isolation and experience of maltreatment during the childhood (Miragoli & Verrocchio, 2008), out-of-home placement of mothers when they were children (Wall-Wieler et al., 2018), being victims of IPV (Ahmadabadi et al., 2018), the psychological and economic autonomy and independence of the mother (Milani et al., 2020), and lack of knowledge or disregard for the child development (Milani et al., 2020).

On the other hand, few studies specifically deepened factors associated with maltreating fathers, but the presence of distrust of social rules and institutions, disregard of child development, social deviance, poor empathy skills, the lack of autonomy and supportive network, the approval of violence, and punishment as educational practices significantly predicted the child removal (Grumi & Mascheroni, 2017; Milani, Grumi, Camisasca, Miragoli, Traficante, & Di Blasio, 2020).

In the light of the state of the art, this study aimed to explore factors, both risk and protective, that influence the CPS workers’ child removal intervention for mothers and fathers among cases characterized by domestic and witnessed violence. In more specific terms, it adopted a retrospective design to test the relationships between the workers’ decision and a set of risk and protective factors relevant for parents’ assessment in cases of child maltreatment (Grumi, Milani, & Di Blasio, 2017).

Method

Participants

This study constitutes a second-level analysis of a larger project about the assessment of parents referred for child maltreatment (Milani, Miragoli, Grumi, & Di Blasio, 2020). This is especially focused on families where witnessed violence was detected; therefore, analyses were run on this specific subsample.

Participants of the original project were 652 families referred to Italian CPSs due to a situation of maltreatment. Among this families, domestic and witnessed violence were present in 218 cases. Considering this subsample, the mean age of the mothers was 35.33 (SD = 7.67; range = 19–52), while the mean age of the fathers was 40.37 (SD = 8.37; range = 21–66). Overall, the 49.5% of the mothers and the 48.9% of the fathers were immigrants. Furthermore, 13% of foreign parents were from European Union states, 17% were from non-EU countries, and 70% were from other continents.

The mean age of the children was 7.52 (SD = 4.77; range = 0–17), the 52.3% of them were males and the 47.7% were females.

Measures

Protocol of risk and protective factors

The data collection was performed through an ad hoc checklist based on factors included in the Protocol of risk and protective factors (Di Blasio, 2005). This is a part of the CRidee Multi-method Assessment System (Milani, Miragoli, Grumi, & Di Blasio, 2020) and it focused on familial frailties and resources linked to maltreatment and recidivism against children, taking into consideration both distal and proximal factors, in line with the distinction introduced by Baldwin, Baldwin, and Cole (1990). Distal risk factors imply a condition of vulnerability but exert an indirect influence on parental practices and child’s developmental trajectory, while proximal factors have a direct influence on the parent–child relationship. Proximal factors include both risk and protective factors: risk factors exacerbate vulnerability induced by distal factors increasing the probability that the situation evolves into a harm for the child; on the contrary, protective factors are those proximal resources that may buffer the negative impact of distal and proximal risk factors.

The Process Oriented Model (Cummings, Davies, & Campbell, 2000), which focuses specifically on the interrelation of risk and protective factors and their mutual influence in development, formed the theoretical framework for the selection of this instrument. It posits that human development is characterized by mutual influences between different factors and environments, which lead to different developmental trajectories with adaptive or maladaptive patterns (Di Blasio, 2005). This model has been modified to fit the clinical literature on risk assessment in child protection work (Di Blasio, 2005). In particular, it takes into consideration genetic, biological, and psychological factors of individuals, which interact with familial functioning and environmental affordances into determining parental attitudes toward offspring. These contextual factors may be mediated by the daily parent’s psychological functioning, leading to adaptive or maladaptive trajectories. The adaptive trajectories characterize parents who are able to express positive parenting and to buffer the impact of eventual negative events, while maladaptive ones characterize parents who are not able to buffer the effects of personal difficulties/inabilities or negative events, exposing their children to situations of vulnerability or clear harm.

Preliminary national studies confirm the good discriminating power of factors assessed by the Protocol between maltreated minors at high risk and low risk for their safety (Milani & Gagliardi, 2013; Milani, Grumi, Gagliardi, & Di Blasio, 2016; Miragoli & Verrocchio, 2008).

The instrument comprises 36 factors classified as follows:

- 11 Distal Risk Factors (DRF): chronic poverty, low educational level, young maternal age, lack of interpersonal relationships, lack of network and social integration, one-parent family, experience of neglect, violence and/or abuse during the childhood, distrust of social rules and institutions, approval of violence and punishments as educational practices, approval of child pornography, lack of knowledge or interest relating to child development;

- 14 Proximal Risk Factors (PRF): parents’ psychopathology, parents’ social deviance, drugs abuse, no assumption of child maltreatment responsibility, compensation syndrome, poor empathy skills, impulsivity, lack of frustration tolerance, separation anxiety, unwanted pregnancy and maternity, problematic relationship with family of origin or partner’s family, dating conflict and domestic violence, child physical disease or disorder at the birth, difficult child temperament.

- 11 Proximal Protective Factors: discomfort related to the dependency on Services, psychological elaboration of experience of neglect, violence, and/or abuse during the childhood (parents), empathy, assumption of responsibility about child maltreatment, desire to improve oneself, autonomy/independence, good level of self-esteem, good relationship with at least one relative, supportive network of relatives and/or friends, conflict management skills, and child easy temperament.

A more deepened description of each factor is available in Supplemental Materials.

Procedure

Procedure and aims of the research were presented to the heads of local CPSs. All contacted services gave the authorization for the psychosocial documentation consultation in conformity with Italian privacy legislation. Research team was thus granted unlimited access to anonymized social records, which included narrative notes regarding the psychosocial evaluations of the families, the assessments performed by different professionals of CPSs, and (if available) the transcripts of meetings with family members and/or medical records.

The evaluation of the presence/absence of each factor was performed using the checklist based on the information retrospectively inferable from the social records of each referred case. Factors were measured as dichotomous variables, which take the value of 0 or 1 to indicate the absence or presence of each. Factors were measured separately for each parent. After having coded for the presence or the absence of the risk and protective factors, in the event of missing data, the research team had the opportunity to interview CPS professionals to gather the missing information.

Then, the subsample of cases characterized by IPV and witnessed violence was isolated and split up into two groups on the basis of the CPS professionals’ placement decision after the investigation (child removal decision versus parents support and monitoring intervention).

The research complied with the ethical standards of the Italian National Psychological Association and the Italian Laws regarding privacy and sensitive information treatment.

Data Analysis

This study adopted two different statistical approaches to identify the most relevant factors for the referred parents’ assessment. The first one is based on logistic regression models, while the second one used the decision tree analysis.

About the first approach, two stepwise logistic regressions for mothers and fathers were performed. Taking into consideration the small sample size, preliminary bivariate analyses (Pearson’s chi square test) were performed to reduce the overall number of predictors in the models. Therefore, factors that were not related to the CPS workers’ decision (p > .05) were excluded.Because of sample’s variability about reasons for reporting and nationality, significant factors of the Protocol were tested together with the maltreatment form (multiple forms of maltreatment versus only witnessed violence perpetrated by parents) and the status of immigrant or native in two stepwise logistic regressions. Odds ratio was calculated for each predictor.

About the second approach, the C&RT decision tree analysis was used. The C&RT algorithm is a form of binary recursive partitioning (Merkle & Shaffer, 2011) that begins from the “parent node” that is subsequently split into two subgroups, “child nodes” by a predictor variable. The predictor variable was selected through the Gini criterion, which constitute a measure of subgroup variability, and splitting continued until pre-determined stopping criteria were met. Decision trees are considered to be one of the most popular approaches for representing classifiers and are largely used in different disciplines (Maimon & Rokach, 2014; Rokach & Maimon, 2005). This method has been used by Sledjeski et al. (2008) to model the recurrence of child maltreatment using data from child welfare registers and it proved efficient and solid (for recent developments, see Thurston & Miyamoto, 2018). This data analysis constitutes an exploratory technique for investigating large quantities of categorical data and makes it possible to reduce problems due to the high number of factors to test and the small sample size (Kass, 1980; Maimon & Rokach, 2014).

If compared to more traditional methods of risk assessment such as regression models, it allows to identify more parents’ profiles of high risk and low risk. In fact, from the decision tree emerged different patterns of risk and protective factors that share the same outcome (child-out-of-home placement decision versus monitoring and support). Tested decision tree models were constructed with the same rationale of the first statistical approach: all factors of the protocol that were significantly related to the CPS workers’ decision were tested as predictors together with the maltreatment form (multiple forms of maltreatment versus only witnessed violence) and the status of immigrant or native. The CPS workers’ child placement decision was the dependent variable. The C&RT decision tree analysis was performed separately for mothers’ and fathers’ groups.

The data were analyzed with IBM SPSS 24.0.

Results

Descriptive Analyses

In particular, considering the general sample, the 54.9% of cases presented one form of maltreatment, 26.4% presented multiple/co-occurring forms of maltreatment, and 18.7% were referred for a general risk situation for minors. Cases classified as “general risk” are likely to have parents who are vulnerable to psychopathology, drugs abuse, incarceration, high level of conflict between partners, and previous referrals to CPS. Table 1 summarizes the prevalence of child maltreatment forms detected (the sum of the percentages of maltreatment typologies exceeds 100% because co-occurring forms of maltreatment characterized 26.4% of the cases).

Table 1.

Prevalence of Revealed Forms of Child Maltreatment.

Forms of child maltreatment	f_i (%)
Neglect	46.5
Witnessed violence	33.4
Emotional abuse	26.7
Physical maltreatment	25.9
Sexual abuse	7.7

Also witnessed violence often occurred in association with other direct forms of child maltreatment. In particular, among the subsample of this study, it emerged isolated in the 42.7% of the cases analyzed, while it was a part of multiple maltreatment experiences in the 57.3% of referred situations. Table 2 summarizes the prevalence of maltreatment typologies assessed in the subsample.

Table 2.

Prevalence of Child Maltreatment Typologies Associated With Witnessed Violence.

Forms of child maltreatment	f_i (%)
Witnessed violence as single form of maltreatment	42.7
Witnessed violence and emotional abuse	42.9
Witnessed violence and neglect	36.7
Witnessed violence and physical maltreatment	34.4
Witnessed violence and sexual abuse	7.8

As regards the CPS worker’s placement decisions, mothers and fathers experienced the child removal in the 53.8% and 67% of the cases, respectively. For the remaining cases, the CPSs’ intervention did not include the child removal and the out-of-home care.

Logistic Regression Models

Table 3 summarizes the bivariate relationships between the Protocol’s factors and the child safety assessment tested through the Pearson’s chi square test. As emerged, some factors appear to be predictive for the assessment of both parents, while others are specific for mothers or fathers. These preliminary analyses allowed to reduce the number of factors to test in the logistic regression models, selecting only the significant ones.

Table 3.

Bivariate Relationships Between the Protocol’s Risk and Protective Factors and the Child Removal Decision for Mothers and Fathers.

	Mothers			Fathers
Factors	χ²	p	phi	χ²	p	phi
DRF Chronic poverty	1.26	n.s.	–0.08	4.34	.037	–0.15
DRF Low educational level	0.09	n.s.	0.02	0.85	n.s.	–0.06
DRF Young maternal age	1.91	n.s.	0.10	–	–	–
DRF Lack of interpersonal relationships	0.29	n.s.	–0.04	0.003	n.s.	0.004
DRF Lack of network and social integration	0.01	n.s.	–0.01	0.90	n.s.	0.07
DRF Experience of neglect, violence, and/or abuse during the childhood	3.07	n.s.	–0.13	11.31	.001	–0.26
DRF Distrust of social rules and institutions	2.02	n.s.	0.11	1.93	n.s.	0.16
DRF Approval of violence and punishments as educational practices	5.78	.016	0.17	1.72	n.s.	0.09
DRF Approval of child pornography	0.34	n.s.	–0.04	4.93	.026	–0.16
DRF Lack of knowledge or interest relating to child development	18.6	.000	0.32	4.28	.038	0.15
PRF Parents’ psychopathology	0.55	n.s.	0.05	26.22	.000	–0.38
PRF Parents’ social deviance	3.02	n.s.	0.08	2.41	n.s.	0.12
PRF Drugs abuse	3.72	n.s.	0.14	3.83	.050	0.15
PRF No assumption of child maltreatment responsibility	13.81	.000	0.27	5.62	.018	0.18
PRF Compensation syndrome	0.31	n.s.	0.04	6.43	.011	–0.18
PRF Poor empathy skills	16.36	.000	0.29	6.49	.011	0.19
PRF Impulsivity	4.79	.028	0.16	4.84	.028	0.16
PRF Lack of frustration tolerance	1.60	n.s.	–0.09	0.03	n.s.	–0.01
PRF Separation anxiety	14.28	.000	–0.28	23.02	.000	–0.36
PRF Unwanted pregnancy and maternity	1.01	n.s.	0.07	–	–	–
PRF Problematic relationship with family of origin or partner’s family	0.80	n.s.	0.06	4.45	.035	–0.16
PRF Child physical disease or disorder at the birth	0.09	n.s.	–0.02	2.73	n.s.	–0.12
PRF Difficult child temperament	0.29	n.s.	–0.04	0.16	n.s.	–0.03
PPF Discomfort related to the dependency on Services	11.67	.001	–0.25	9.91	.002	–0.23
PPF Psychological elaboration of experience of neglect, violence, and/or abuse during the childhood	5.02	.025	–0.18	0.34	n.s.	–0.04
PPF Empathy	5.29	.021	–0.17	4.51	.034	–0.16
PPF Assumption of responsibility about child maltreatment	4.00	.045	–0.15	3.36	n.s.	–0.14
PPF Desire to improve oneself	2.98	n.s.	–0.12	3.92	.048	–0.15
PPF Autonomy/independence	7.02	.008	–0.19	1.73	n.s.	–0.09
PPF Good level of self-esteem	1.03	n.s.	–0.07	37.28	.000	–0.47
PPF Good relationship with at least one relative	5.29	.021	–0.17	10.20	.001	–0.24
PFF Supportive network of relatives and/or friends	0.04	n.s.	–0.01	5.89	.015	–0.18
PPF Conflict management skills	0.07	n.s.	0.02	3.25	n.s.	–0.13
PPF Child easy temperament	0.04	n.s.	0.01	0.01	n.s.	–0.01

Tables 4 and 5 show the logistic regression models that predict the CPS workers’ intervention for mothers (χ² = 66.309; p < .001; R² Nagelkerke = .625) and fathers (χ² = 82.808; p < .001; R² Nagelkerke = .549). Models confirmed two different pattern of factors that lead the assessment performed by CPS workers, with the exception of the proximal risk factor poor empathy skills, which enormously increases the risk of child removal for both mothers and fathers.

Table 4.

Predictive Factors of CPS Workers’ Removal Decision: Logistic Regression Model for Mothers.

Predictors	Β	P	ExpB (95% CI)
Multiple forms of child maltreatment	1.507	<.001	4.51 (2.26–9.01)
PRF Poor empathy skills	4.14	<.001	62.64 (7.2–546)
PPF Discomfort related to the dependency on Services	2.14	<.05	8.54 (1.18–61.85)

χ² = 66.309; p < .001; R² Nagelkerke = .625.

Table 5.

Predictive Factors of CPS Workers’ Removal Decision: Logistic Regression Model for Fathers.

Predictors	Β	p	ExpB (95% CI)
Immigration	–0.94	<.05	0.39 (0.16–0.96)
PRF Parents’ psychopathology	–2.04	<.001	0.13 (0.04–0.35)
PRF Drugs abuse	1.53	<.01	4.61 (1.74–12.22)
PRF Poor empathy skills	2.37	<.01	10.75 (2.52–45.9)
PRF Separation anxiety	–1.31	<.05	0.27 (0.09–0.76)
PPF Good level of self-esteem	–1.72	<.001	0.18 (0.07–0.45)

χ² = 82.808; p < .001; R² Nagelkerke = .549.

Other significant factors for the mothers were the presence of multiple forms of child maltreatment, which results in a 351% increase in the odds of child removal, and the discomfort related to the dependency on Services, which results in a 754% increase in the odds of mother–child separation. The discomfort related to the dependency on Services is presented by the protocol as a protective factor, because it promotes the parents’ engagement in the rehabilitation program to become again independent, opposing to a state aid attitude, while in this study it acts as a risk factor.

Others significant factors for the fathers were the status of immigrant and the substance abuse, which result in a 61% and 361% increase in the odds of father–child separation, respectively, while, interestingly, the presence of the risk factors parents’ psychopathology and separation anxiety predicted interventions of monitoring and support.

Decision Tree Models

The first decision tree about the mothers’ assessment (Figure 1) identified seven predictors about the child removals. In particular, it estimated that the presence or not of multiple forms of child maltreatment, the distal risk factors experience of neglect, violence and/or abuse during the childhood, low educational level, approval of violence and punishments as educational practices, the proximal risk factor poor empathy skills, and the proximal protective factors discomfort related to the dependency on Services and empathy allowed to correctly classify the 81% of the cases.

Figure 1.

Decision tree about the child placement based on the mothers’ assessment.

In particular, five profiles of high risk emerged and are characterized by:

- multiple child maltreatment together with poor empathy skills, in spite of the absence (see node 15) or presence (se node 16) of the discomfort related to the dependency on services;

- multiple child maltreatment together with poor empathy skills and approval of violence and punishment as educational practices (see node 12);

- multiple child maltreatment together with a low educational level (see node 14);

- witnessed violence and absence of mother’s experience of violence during the childhood and discomfort related to the dependency on services (see node 9).

About the low risk, four different profiles emerged:

- multiple form of child maltreatment without empathic problems and low educational level (see node 13);

- presence of only witnessed violence for the child together with mother’s experience of violence during the childhood, in spite of the absence (see node 7) or presence (see node 8) of good empathy skills;

- presence of only witnessed violence for the child together with the discomfort related to the dependency on services (see node 10).

The second decision tree about the fathers’ assessment (Figure 2) identified six predictors about the child removals. In particular, it estimated that immigration and the presence or not of multiple forms of child maltreatment, distal risk factor chronic poverty, proximal protective factors good level of self-esteem, discomfort related to the dependency on Services, and supportive network of relatives and/or friends allowed to correctly classify the 80.8% of the cases.

Figure 2.

Decision tree about the child placement based on the fathers’ assessment.

In particular, three profiles of high risk emerged and are characterized by:

- the absence of the protective factor good level of self-esteem, in spite of the foreign (see node 5) or Italian origin (see node 6). The immigrant status increased the probability of child-removal but both native and immigrants show a high-risk profile;

- good level of self-esteem together with the absence of factors chronic poverty and discomfort related to the dependency on Services (see node 7).

About the low risk, the following three profiles emerged:

- good level of self-esteem together with absence of factors chronic poverty and presence of discomfort related to the dependency on Services (see node 8);

- good level of self-esteem, chronic poverty, and witnessed violence as single form of child maltreatment, in spite of the absence (see node 11) or presence (se node 12) of a supportive network of relatives and/or friends.

The child node 10 did not appear to be sufficiently discriminative at this stage of split.

Discussion

Results of this study aimed to deepen the knowledge about factors that influence the safety assessment and child placement decision by CPS workers in cases characterized by IPV perpetrated by fathers. In line with the literature, witnessed violence emerged as a widespread form of child maltreatment in our sample. Despite the unrepresentative composition of the sample in terms of geographic sites of CPSs involved, this study found that witnessed violence constitutes the second more prevalent form of maltreatment against children, in line with the first nationwide survey of the cases in the care of CPS (Autorità garante per l’infanzia e l’adolescenza et al., 2015). Moreover, consistent with the poly-victimization model (Finkelhor et al., 2007), it emerged to be associated with other forms of child maltreatment in more of than half of referred cases. It co-occurred especially with emotional abuse, neglect, and physical maltreatment.

About the parenting skills’ assessment, mothers who are victims of IPV experienced the child removal in about half of the cases, while fathers showed a higher rate. Differences emerged about the mothers’ and fathers’ risk profiles, suggesting that workers attribute different weight to specific information cues when they characterize the maternal or the paternal role, with the exception of the proximal risk factor poor empathy skills that seems to be crucial for both mothers and fathers. In fact, good empathy skills sustain the perspective taking and the emotional tuning with the child and his or her distress, reducing the risk of the abusive behavior’s recurrence (Di Blasio, 2005; Cirillo, 2005). On the contrary, bias on this level is perceived as an element of concern in these cases since the inability to perceive the minor’s point of view reduces the possibilities of protecting him/her from the consequences of the violence exposure. This is consistent with literature that stresses the prosecution against battered mothers in terms of “failure to protect” and results about the reduced paternal perception of the violence impact on the child. As shown by Salisbury et al. (2009), few fathers perceived that their children had been negatively affected by their violent behaviors, even when they acknowledged that their children had been exposed to them. On the other hand, other scholars have noted that fathers’ concern about the impact of IPV on their children may be a powerful motivating factor in seeking and remaining in treatment (Fox et al., 2001).

Other significant factors for the mothers were the fact that the child was victims of only witnessed violence or also other forms of maltreatment and the discomfort related to the dependency on services. As noted, interestingly, the last one did not exert a protective role in the cases of battered mothers, but it increased the risk of experiencing the child removal. The Protocol of risk and protective factors showed that this factor promotes the parents’ engagement in the rehabilitation program to become again independent from Services, while in these specific cases it is interpreted by CPS workers as an element of concern. It may be associated with the strong presence of services in the care of battered mothers, including being hosted in emergency shelters, especially when they are exposed to a high risk for their health and life. These forms of intervention imply a limitation of the freedom of victims and the mothers’ discomfort may be interpreted as a risk for the reunification with the abusive partner.

The decision trees about mothers confirmed the relevance of child maltreatment typology and empathy skills. In more specific terms, the presence of only witnessed violence tended to lead to interventions that do not imply the mother–child separation, even in the presence of other risk factors as poor empathy skills and maternal unelaborated experiences of violence or abuse during the childhood; while the presence of multiple maltreatment against the child appears to be buffered only by good empathy skills.

Other significant factors for the fathers were the risk factor drug abuse and the protective factor good level of self-esteem. The substance abuse, as proximal factor, exerts a direct effect on the quality of the parent–child relationship, increasing the rates of physical and sexual abuse (Walsh et al., 2003) and, as shown by Murphy et al. (1991), parents with documented substance abuse are more likely to have their children permanently removed. Instead, a good level of self-esteem acts as a protective factor sustaining a stable identity and the willingness to ask for help in case of need, without denying obstacles and limits (Di Blasio, 2005). A good level of self-esteem emerged as a crucial factor also in the decision tree model, in fact its presence appears to buffer the impact of distal risk factors as chronic poverty leading to low-risk profiles, especially among cases without multiple child maltreatment. On the contrary, its absence is especially associated with high-risk profiles for the fathers. One controversial result emerged for father’s psychopathology: unlike expectations, father’s psychopathology acted as a protective factor, rather than as risk factor, in this subsample. Finding may be interpreted as an effect of taking charge of cases of psychopathology. In more specific terms, situations of parents with a diagnosed psychopathology are known by services and the taking in charge may play a role of both treatment and monitoring. Such increased attention by the social services may therefore contribute to contain risk situations. A second controversial results lie in the risk factor Separation Anxiety, which usually is correlated with poorer parenting skills (i.e., the parent negates the child need for autonomy), but that in this sample is unrelated with the removal of the child. As no evaluation of the attachment between parent and child has been collected, this result should be addressed in future research. However, given the relation between the parents’ own attachment experiences and the heightened risk of separation anxiety toward their infants, with subsequent impairment in parenting efficacy (cf. Lutz & Hock, 1995), the quality of the Internal Working Models should be assessed by CPS professionals whenever possible.

The immigrant status emerged as a characteristic of almost half of the sample (49.5% of the mothers, and 48.9% of the fathers), and was significant for fathers at high risk, increasing, in such cases, the probability of child removal. In this regard, it is important not to ignore some cultural considerations and possible biases related to the immigrant sample. In the CPS work, cultural characteristics could play an interfering role in the operators’ evaluation and decision-making processes: unconscious stereotypes (ethnic/cultural biases and confirmation biases) can impact the best comprehension of behaviors, history, and migration background of families referred to CPS (Middel et al., 2020; Fluke et al., 2010; Loya, 2011; Keddell & Hyslop, 2019). Ethnicity-based and immigrant stereotypes can lead to a different comprehension of family behaviors, increasing or decreasing the evaluation of risk factors (Keddell & Hyslop, 2019). Moreover, an educational practice could be considered as normative in some cultures, while creating the basis for maltreatment in another, like in western cultures (Hughes, 2006).

On the whole, results seem to indicate the following key points are to be considered when deciding upon the means of protecting children in the event of IPV:

(1) The single most relevant factor predicting the removal of the child from the family seems to be the lack of empathy skills by the parents, usually stemming from experiences of abuse and maltreatment when they were children (cf. De Paul & Guibert, 2008; Locher et al., 2014);

(2) As regards in particular the mothers, the presence of multiple maltreatment forms and the discomfort for being helped by the Services seem to be particularly relevant to decide to remove the child from the family. Literature converges on considering multiple maltreatments as potent risk factors (cf. White et al., 2015), so this result comes to no surprise. On the contrary, discomfort for being helped by the Services can be usually thought as a protective factor (i.e., the parent is not happy to be somewhat “in the spotlight” by the CPS, thus trying to ameliorate his/her condition by finding a work, getting back to study, etc.). However, in our sample, this acted as a risk factor, probably due to worse attitude of these families regarding the intervention of CPS professionals. This surprising result might be addressed in future works.

Limitations

This study showed some limitations, mainly due to methodology adopted and the partial possibility to access social records in CPS. First of all, factors related to workers’ attitudes and their working context were not considered. However, as pointed out by the Decision-Making Ecology model (Baumann et al., 2011), these factors could impact the workers’ assessment. In addition, more specific data about IPV characteristics, as duration and severity, were not available. Moreover, the checklists used to gather the data were not completed by independent judges, due to CPS restraints and management of research personnel. However, each judge was specifically trained to reduce subjective bias and to support the factors’ identification.

The second limitation regards the sample size that is limited considering the number of tested factors in the models, with a consequent negative impact on the statistical power of the analysis. Moreover, the sample size did not permit to test all the potential predictors in logistic regression models, requiring a reduction of factors based on the bivariate statistics.

The third limitation concerns the unrepresentative composition of the sample in terms of geographic provenience, as all of the CPSs involved were from only one region of Italy. However, considering that in Italy there are not a unified service system nor a national observatory or register about child maltreatment and neglect, these results provide preliminary data to address this issue.

Notwithstanding these limitations, this research tried to deepen the knowledge of CPS professionals’ decision-making processes when IPV and witnessed violence co-occur, providing preliminary data about an object of research still little investigated.

Supplemental Material

sj-docx-1-jiv-10.1177_08862605221137710 – Supplemental material for The CPS Workers’ Child Removal Decision in Cases of Domestic and Witnessed Violence

Supplemental material, sj-docx-1-jiv-10.1177_08862605221137710 for The CPS Workers’ Child Removal Decision in Cases of Domestic and Witnessed Violence by Luca Milani, Serena Grumi, Elena Camisasca, Sarah Miragoli, Martina Cattani and Paola Di Blasio in Journal of Interpersonal Violence

Footnotes

Author’s Note

Serena Grumi is also affiliated to IRCCS Fondazione Mondino, Pavia, Italy.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding

The authors received no financial support for the research and/or authorship of this article.

ORCID iD

Luca Milani

Supplemental Material

Supplemental material for this article is available online.

Author Biographies

Luca Milani, PhD, Full Professor in Developmental Psychology at the Faculty of Psychology, Catholic University of Milan, Italy. Director of the Research Center for Developmental and Educational Dynamics (C.R.I.d.e.e.). Research interests: developmental outcomes of the interaction with technology, child abuse and maltreatment, trauma, and PTSD in high-risk population.

Serena Grumi, PhD, IRCCS Fondazione Mondino, Pavia, Italy. Research interests: violence against women, intimate partner violence, child abuse, and maltreatment.

Elena Camisasca, PhD, Associate Professor in Developmental Psychology at the Faculty of Psychology, eCampus University, Novedrate (CO), Italy. Research interests: marital adjustment and marital conflict, parenting alliance, parenting stress, attachment, child abuse, and maltreatment.

Sarah Miragoli, PhD, Research Assistant in Developmental Psychology at the Faculty of Psychology, Catholic University of Milan, Italy. Director of the Research Unit for the Psychology of Trauma, Catholic University of Milan. Research interests: child abuse and maltreatment, traumatic narratives, parenting stress, sexualized behaviors, and child legal involvement.

Martina Cattani, M.Sc., Collaborator of the Research Center for Developmental and Educational Dynamics (C.R.I.d.e.e.). Research interests: child abuse and maltreatment.

Paola Di Blasio, PhD, Retired Full Professor in Developmental Psychology at the Faculty of Psychology, Catholic University of Milan, Italy. Research interests: child abuse and maltreatment, trauma related symptoms in childbirth, marital adjustment, and parenting stress.

References

Ahmadabadi

Najman

J. M.

Williams

G. M.

Clavarino

A. M.

d’Abbs

Abajobir

A. A.

(2018). Maternal intimate partner violence victimization and child maltreatment. Child Abuse & Neglect, 82, 23–33.

Andrews

N. C.

Motz

Pepler

D. J.

Jeong

J. J.

Khoury

(2018). Engaging mothers with substance use issues and their children in early intervention: Understanding use of service and outcomes. Child Abuse & Neglect, 83, 10–20.

Appel

A. E.

Holden

G. W.

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

Autorità garante per l’infanzia e l’adolescenza, Terre des homes, & CISMAI (2015). Indagine nazionale sul maltrattamento dei bambini e degli adolescenti in Italia. http://www.garanteinfanzia.org/sites/default/files/documenti/Indagine_maltrattamento_TDH_Cismai_Garante_mag15.pdf

Baldwin

A. L.

Baldwin

C. P.

Cole

(1990). Stress-resistant families and stress-resistant children. In Rolf

Masten

Cicchetti

Nuechterlein

Weintraub

(Eds.), Risk and protective factors in the development of psychopathology (pp. 257–280). Cambridge University Press.

Bancroft

Silverman

(2002). The batterer as parent. Sage.

Baumann

Dalgleish

Fluke

Kern

(2011). The decision-making ecology. American Humane Association.

Brewsaugh

Masyn

K. E.

Salloum

(2018). Child welfare workers’ sexism and beliefs about father involvement. Children and Youth Services Review, 89, 132–144. https://doi.org/10.1016/j.childyouth.2018.04.029

Brewsaugh

Strozier

(2016). Fathers in child welfare: What do social work textbooks teach our students? Children and Youth Services Review, 60, 34–41. https://doi.org/10.1016/j.childyouth.2015.11.015

10.

Brown

Callahan

Strega

Walmsley

Dominelli

(2009). Manufacturing ghost fathers: The paradox of father presence and absence in child welfare. Child & Family Social Work, 14, 25–34. https://doi.org/10.1111/j.1365-2206.2008.00578.x

11.

Carpenter

G. L.

Stacks

A. M.

(2009). Developmental effects of exposure to intimate partner violence in early childhood: A review of the literature. Children and Youth Services Review, 31, 831–839.

12.

Cirillo

(2005). Cattivi genitori. R. Cortina.

13.

CISMAI (2017) Requisiti minimi degli interventi nei casi di violenza assistita da maltrattamento sulle madri. http://cismai.it/requisiti-minimi-degli-interventi-nei-casi-di-violenza-assistita/

14.

Coakley

T. M.

Kelley

Bartlett

(2014). Exploring child welfare workers’ attitudes and practice with fathers. Journal of Family Strengths, 14, 1–17.

15.

Crawford

Bradley

M. S.

(2016). Parent gender and child removal in physical abuse and neglect cases. Children and Youth Services Review, 65, 224–230. https://doi.org/10.1016/j.childyouth.2016.04.013

16.

Cummings

Davies

Campbell

(2000). Developmental psychopathology and family process. Guilford Press.

17.

Di Blasio

. (2005). Tra rischio e protezione: la valutazione delle competenze parentali. Unicopli.

18.

De Paul

Guibert

. (2008). Empathy and child neglect: A theoretical model. Child Abuse & Neglect, 32(11), 1063-1071.

19.

Edleson

J. L.

(1998). Responsible mothers and invisible men: Child protection in the case of adult domestic violence. Journal of Interpersonal Violence, 13, 294–298.

20.

Edleson

J. L.

(1999). The overlap between child maltreatment and woman battering. Violence Against Women, 5, 134–154.

21.

Finkelhor

Ormrod

Turner

(2007). Poly-victimization: A neglected component in child victimization trauma. Child Abuse & Neglect, 31, 7–26.

22.

Fleck-Henderson

(2000). Domestic violence in the child protection system: Seeing double. Children and Youth Services Review, 22, 333–354.

23.

Fluke

Jones Harden

Jenkins

Ruehrdanz

(2010). Research synthesis on child welfare disproportionality and disparities. American Humane Association. https://casala.org/wp-content/uploads/2015/12/Disparities-and-Disproportionality-in-Child-Welfare_An-Analysis-of-the-Research-December-2011-1.pdf#page=11

24.

Font

S. A.

Maguire-Jack

(2015). Decision-making in child protective services: Influences at multiple levels of the social ecology. Child Abuse & Neglect, 47, 70–82.

25.

Fox

G. L.

Sayers

Bruce

(2001). Beyond bravado: Redemption and rehabilitation in the fathering accounts of men who batter. Marriage & Family Review, 32, 137–163.

26.

Fugate

(2001). Who’s failing whom? A critical look at failure-to-protect laws (Vol. 76, pp. 272−308). New York University Law Review.

27.

Grumi

Mascheroni

(2017). Padri maltrattanti coinvolti nella Tutela Minorile: un focus sui fattori di rischio e di protezione. Maltrattamento e abuso all’infanzia, 3, 111–123.

28.

Grumi

Milani

Di Blasio

(2017). Risk assessment in a multicultural context: Risk and protective factors in the decision to place children in foster care. Children and Youth Services Review, 77, 69–75.

29.

Hamby

Finkelhor

Turner

Ormrod

(2010). The overlap of witnessing partner violence with child maltreatment and other victimizations in a nationally representative survey of youth. Child Abuse & Neglect, 34(10), 734–741.

30.

Hartley

C. C.

(2004). Severe domestic violence and child maltreatment: Considering child physical abuse, neglect, and failure to protect. Children and Youth Services Review, 26(4), 373–392.

31.

Hughes

(2006). The neglect of children and culture: Responding to child maltreatment with cultural competence and a review of child abuse and culture: Working with diverse families. Family Court Review, 44, 501–510. https://doi.org/ 10.1111/j.1744-1617.2006.00103.x

32.

Hughes

Chau

(2012). Children’s best interests and intimate partner violence in the Canadian family law and child protection systems. Critical Social Policy, 32(4), 677–695.

33.

Kass

G. V.

(1980). An exploratory technique for investigating large quantities of categorical data. Applied Statistics, 29, 119–127.

34.

Keddell

Hyslop

(2019). Ethnic inequalities in child welfare: The role of practitioner risk perceptions. Child & Family Social Work, 24, 409–420. https://doi.org/10.1111/cfs.12620

35.

Kitzmann

K. M.

Gaylord

N. K.

Holt

A. R.

Kenny

E. D.

(2003). Child witnesses to domestic violence: A meta-analytic review. Journal of Consulting and Clinical Psychology, 71(2), 339.

36.

Locher

S. C.

Barenblatt

Fourie

M. M.

Stein

D. J.

Gobodo-Madikizela

(2014). Empathy and childhood maltreatment: A mixed-methods investigation. Annals of Clinical Psychiatry, 26(2), 97–110.

37.

Lothian

(2002). Mapping contested terrain: The doctrine of failure to protect in Canadian criminal law. [Unpublished LLM thesis]. Faculty of Law, University of British Columbia.

38.

Loya

(2011). Color-blind racial attitudes in white social workers: A cross-sectional study. Smith College Studies in Social Work, 81, 201–217. https://doi.org/10.1080/00377317.2011.589341

39.

Lutz

W. J.

Hock

(1995). Maternal separation anxiety: Relations to adult attachment representations in mothers of infants. The Journal of Genetic Psychology, 156(1), 57–72.

40.

Maimon

O. Z.

Rokach

(2014). Data mining with decision trees: Theory and applications (Vol. 81). World Scientific.

41.

Merkle

E. C.

Shaffer

V. A.

(2011). Binary recursive partitioning: Background, methods, and application to psychology. British Journal of Mathematical and Statistical Psychology, 64(1), 161–181.

42.

Middel

López

M. L.

Fluke

Grietens

(2020). The effects of migrant background and parent gender on child protection decision-making: An intersectional analysis. Child Abuse & Neglect, 104, 104479, https://doi.org/10.1016/j.chiabu.2020.104479

43.

Milani

Gagliardi

(2013). Fattori di rischio e di protezione nella valutazione delle competenze parentali di famiglie italiane e famiglie immigrate. Maltrattamento e abuso all'infanzia, 15(1), 59–80.

44.

Milani

Grumi

Camisasca

Miragoli

Traficante

Di Blasio

(2020). Familial risk and protective factors affecting CPS professionals’ child removal decision: A decision tree analysis study. Children and Youth Services Review, 109, 104687.

45.

Milani

Grumi

Gagliardi

Di Blasio

(2016). Famiglie italiane e immigrate: identificare fattori di rischio e di protezione nella valutazione delle competenze genitoriali. Psicologia della Salute, 1, 67–89.

46.

Milani

Miragoli

Grumi

Di Blasio

(2020). A multi-method assessment of risk and protective factors in family violence: Comparing Italian and immigrant families. In Balvin

Christie

D. J.

(Eds.), Children and peace: From research to action. Springer.

47.

Mills

L. G.

(2000). Woman abuse and child protection: A tumultuous marriage. Children and Youth Services Review, 22, 199–205.

48.

Miragoli

Verrocchio

M. C.

(2008). La valutazione del rischio in situazioni di disagio familiare: fattori di rischio e fattori di protezione. Maltrattamento e abuso all'infanzia, 10(3), 11–28.

49.

Murphy

J. M.

Jellinek

Quinn

Smith

Poitrast

F. G.

Goshko

(1991). Substance abuse and serious child mistreatment: Prevalence, risk, and outcome in a court sample. Child Abuse & Neglect, 15(3), 197–211.

50.

Rokach

Maimon

(2005). Decision trees. In Maimon

Rokach

(Eds.), Data mining and knowledge discovery handbook (pp. 165–192). Springer.

51.

Rumm

P. D.

Cummings

Krauss

M. R.

Bell

M. A.

Rivara

F. P.

(2000). Identified spouse abuse as a risk factor for child abuse. Child Abuse & Neglect, 24(11), 1375–1381.

52.

Russell

Kraus

S. W.

Chapleau

K. M.

Oswald

(2019). Perceptions of blame in intimate partner violence: The role of the perpetrator’s ability to arouse fear of injury in the victim. Journal of Interpersonal Violence, 34(5), 1089–1097. https://doi.org/10.1177/0886260516646999

53.

Salisbury

E. J.

Henning

Holdford

(2009). Fathering by partner-abusive men. Child Maltreatment, 14(3), 232–242.

54.

Save the Children (2011). Spettatori e vittime: i minori e la violenza assistita in ambito domestico Arti Grafiche Agostini. https://s3.savethechildren.it/public/files/uploads/pubblicazioni/spettatori-e-vittime-i-minori-e-la-violenza-assistita-ambito-domestico.pdf

55.

Sledjeski

E. M.

Dierker

L. C.

Brigham

Breslin

(2008). The use of risk assessment to predict recurrent maltreatment: A classification and regression tree analysis (CART). Prevention Science, 9(1), 28–37.

56.

Smith Slep

A. M.

O’Leary

S. G

. (2005). Parent and partner violence in families with young children: Rates, patterns, and con- nections. Journal of Consulting and Clinical Psychology, 73, 435–444.

57.

Strega

(2006). Failure to protect? Child welfare interventions when men beat mothers. In Alaggia

Vine

(Eds.), Cruel but not unusual: Violence in Canadian families (pp. 650−690). Wilfrid Laurier University Press.

58.

Strega

Fleet

Brown

Dominelli

Callahan

Walmsley

(2008). Connecting father absence and mother blame in child welfare policies and practice. Children and Youth Services Review, 30(7), 705–716.

59.

Thurston

Miyamoto

(2018). The use of model based recursive partitioning as an analytic tool in child welfare. Child Abuse & Neglect, 79, 293–301.

60.

Trout

(1999). Domestic violence and the children witnesses. Stop Domestic Violence Congress.

61.

Wall-Wieler

Almquist

Liu

Vinnerljung

Hjern

(2018). Intergenerational transmission of out-of-home care in Sweden: A population-based cohort study. Child Abuse & Neglect, 83, 42–51.

62.

Walsh

MacMillan

H. L.

Jamieson

(2003). The relationship between parental substance abuse and child maltreatment: Findings from the Ontario Health Supplement. Child Abuse & Neglect, 27(12), 1409–1425.

63.

White

O. G.

Hindley

Jones

D. P.

(2015). Risk factors for child maltreatment recurrence: An updated systematic review. Medicine, Science and the Law, 55(4), 259–277.

64.

World Health Organization. (2012). Understanding and addressing violence against women. WHO. http://apps.who.int/iris/bitstream/10665/77432/1/WHO_RHR_12.36_eng.pdf

65.

Zhang

Singer

(2010). Recursive partitioning and applications. Springer.

66.

Zuravin

S. J.

DePanfilis

(1997). Factors affecting foster care placement of children receiving child protective services. Social Work Research, 21(1), 34–42.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB