|Home | About | Journals | Submit | Contact Us | Français|
This paper examines the impact of emotional maltreatment on the psychosocial functioning of youth placed in out-of-home care as a result of maltreatment. Participants included 243 children participating in a randomized controlled trial of a preventive intervention for preadolescent youth placed in out-of-home care. This study analyzed baseline data collected pre-randomization from interviews with children and their out-of-home caregivers and data from child welfare records. Bivariate and regression analyses were used to explore the relationships between emotional maltreatment and interpersonal functioning, self-perception, mental health, and behavioral problems for the total sample and by gender. Findings suggest that subtypes of emotional maltreatment are associated with different outcomes and that males are more negatively impacted by emotional maltreatment than are females.
Although emotional maltreatment has been studied less frequently than physical and sexual abuse, evidence suggests that emotional maltreatment is associated with long-term deleterious consequences in adulthood. In studies with adult samples that rely on retrospective self-reports of maltreatment, a childhood history of emotional maltreatment is associated with low self-esteem, depression, anxiety, trauma symptoms, and the perpetration of or victimization from interpersonal violence (Berzenski & Yates, in press; Briere & Runtz, 1990; Crawford & Wright, 2007; Gibb, Chelminski, & Zimmerman, 2007; Dodge Reyome, in press; Dodge Reyome, Ward, & Witkiewitz, in press; Higgins & McCabe, 2000; Lewis et al., 2006; Mullen, Martin, Anderson, Romans, & Herbison, 1996; Riggs & Kaminski, this issue; Sachs-Ericsson, Verona, Joiner, & Preacher, 2006; Spertus, Yehuda, Wong, Halligan, & Seremetis, 2003; Zurbriggen, Gobin, & Freyd, in press). In many of these retrospective studies, emotional maltreatment serves as a reliable predictor of outcomes over and above the effects of physical and/or sexual abuse (Crawford & Wright, 2007; Higgins & McCabe, 2000; Sachs-Ericsson et al., 2006; Spertus et al., 2003). As many have noted, however, the reliance on retrospective self-reports is problematic as it raises the possibility that the association between maltreatment and outcomes is attributable to shared method variance and/or response bias. Two studies that estimated the association between maltreatment and outcomes found that these associations differed depending upon whether maltreatment was measured retrospectively or prospectively (Raphael, Widom, & Lange, 2001; Widom, Weiler, & Cottler, 1999).
A small number of prospective and cross-sectional studies have examined the relationships between emotional maltreatment and psychosocial development in childhood and adolescence using teacher reports (Gracia, 1995), parent reports (Vissing, Straus, Gelles, & Harrop, 1991), caseworker reports (Chamberland et al., 2005), child welfare records (Crittenden, Claussen, & Sugarman, 1994; Kim & Cicchetti, 2006; Manly, Kim, Rogosch, & Cicchetti, 2001), and observational data (Linder & Collins, 2005) to measure maltreatment experiences. These studies, which reduce the risk of shared method variance and response bias, suggest that emotional maltreatment is associated with significant deviations from normative socio-emotional development among children and adolescents. In these studies, emotional maltreatment was associated with internalizing symptoms, externalizing behavior problems, low self-esteem, and interpersonal problems (Chamberland et al., 2005; Crittenden et al., 1994; Gracia, 1995; Kim & Cicchetti, 2006; Linder & Collins, 2005; Manly et al., 2001; Vissing et al., 1991). Cross-sectional (Gibb & Abela, 2008; Solomon & Serres, 1999) and longitudinal (Stuewig & McCloskey, 2005) studies that use youth self-reports to assess maltreatment also suggest that emotional maltreatment is associated with maladaptive developmental trajectories such as delinquency and depression.
Some studies suggest that emotional maltreatment may have more negative effects on development than do other types of abuse (Crittenden et al., 1994; McGee, Wolfe, & Wilson, 1997). For example, in their study of 160 adolescents randomly selected from the open caseload of a child protection agency, McGee et al. (1997) found that self-report severity ratings of emotional maltreatment were associated with self-reported internalizing and externalizing symptoms, even after controlling for self-reported severity ratings on four other types of maltreatment (physical abuse, sexual abuse, neglect, and family violence). Among the other types of maltreatment, only physical abuse was associated with self-reported symptoms, and it was correlated only with externalizing and not internalizing symptoms.
Although the aforementioned research suggests that the association between emotional maltreatment and adverse outcomes is robust, the interpretation of these findings is complicated by the fact that there is no universally-used, operational definition of emotional maltreatment. The terms emotional abuse, psychological abuse, verbal abuse, emotional maltreatment, and psychological maltreatment have been used interchangeably to describe a wide variety of parenting behaviors believed to be psychologically damaging. In one study, emotional maltreatment may be defined as marital violence, harsh parenting, and parental rejection. In another study, emotional maltreatment may be measured solely by verbal aggression, and in yet another study, emotional maltreatment may include abandonment and exploitation. Furthermore, such studies rarely assess the differential effects of these subtypes of emotional maltreatment. In order to gain a clear understanding of the impact of emotional maltreatment on psychosocial development and adult adjustment, it is important to understand which subtypes of maltreatment are associated with which outcomes.
As a first step, a clear delineation of those behaviors that are universally accepted as forms of emotional maltreatment is needed. Several authors have proposed different typologies for categorizing emotional maltreatment (Hart, Brassard, & Karlson, 1996; Garbarino, Guttman, & Seeley, 1986; McGee & Wolfe, 1991). In their system for coding maltreatment types, Barnett, Manly, and Cicchetti (1993) provide a definition of emotional maltreatment that includes the following: (a) verbal abuse that thwarts self-esteem and belittles or blames a child, (b) intimidation or threats to a child, (c) acts that prevent age-appropriate autonomy, (d) acts that put children in inappropriate roles or provide them with inappropriate levels of responsibility, (e) abandonment, or (f) exposure to violence. These typologies provide a starting place for studies designed to identify the differential effects of emotional maltreatment subtypes on outcomes.
A few recent studies have begun to explore the possibility that subtypes of maltreatment may differentially predict outcomes (Allen, 2008; Linder & Collins, 2005; Schneider, Ross, Graham, & Zielinski, 2005; Stuewig & McCloskey, 2005). For example, in a retrospective study using a college sample, Allen (2008) used simultaneous regression to explore the differential effects of several parent behaviors: acts designed to terrorize children, acts that ignored children, acts that degraded children, and acts that exposed children to domestic violence. After controlling for physical abuse and all other emotional maltreatment subtypes, terrorizing behaviors predicted anxiety and somatic complaints, ignoring predicted depression and features of personality disorders, degrading behaviors predicted features of personality disorders, and exposure to domestic violence did not predict any of the dependent variables.
The current study is designed to expand on this early research in several ways. First, it provides descriptive information on and estimates the differential effects of an overall measure of emotional maltreatment and four subtypes of emotional maltreatment, using data collected from a sample of 243 preadolescent youth placed in foster care as a result of abuse and/or neglect. Second, this study estimates the relationships between each of these emotional maltreatment classifications and several indices of psychosocial functioning (social functioning, self-perception, mental health functioning, life satisfaction), as reported by both youth and their current caregivers.
Participants in this study included youth and their current caregivers who were recruited for a randomized controlled trial of a preventive intervention for preadolescent youth placed in out-of-home care. This preventive intervention is known as the Fostering Healthy Futures prevention program and is described elsewhere (Taussig, Culhane, & Hettleman, 2007). Children, ages 9-11, were eligible for the randomized controlled trial if the following criteria were met: (a) they had been placed in out-of-home care by court order in one of four participating counties within the preceding year, (b) placement was ordered due to maltreatment, and (c) they still resided in out-of-home care at the time of the baseline interview. Ninety-three percent of those meeting eligibility requirements were enrolled. The current study analyzed baseline data that were collected pre-randomization from interviews with children and their out-of-home caregivers. Although 286 children were interviewed at baseline, the sample for this study includes only 243 children. Data from four children were excluded because they were not sufficiently proficient in English to understand the study questions, data from 11 children were excluded because their scores on achievement and intelligence tests indicated significant cognitive impairment, and data from 28 children were excluded because they were siblings of others included in the sample (when siblings were interviewed, one of the siblings was chosen at random to be included in the current study's analyses).
The final sample of 243 children was 48.1% female, with a mean age of 10.4 (SD = .91). The racial/ethnic distribution of children (non-exclusive categories) was 46.9% Hispanic, 48.6% Caucasian, and 28.0% African-American. The sample of caregivers was primarily female (89%). The majority of youth participants were placed in foster (47.3%) or kinship (46.9%) care, with the remaining 5.8% residing in shelters, group homes, and residential treatment centers. About half (49.0%) of the youth were removed from their mothers only, 7.4% from their fathers only, 33.7% from a two-parent household, and the remaining 9.9% from another caregiver(s). The average length of time youth had been in out-of-home care at the time of the interview was 6.4 months (SD = 3.7). Children had to be living with their current caregivers for at least 3 weeks prior to being interviewed. The children in our study had been living in their current placement for an average of 5.3 months (SD = 3.8) at the time of the interview.
This study was approved by the local institutional review board, and informed consent and assent were obtained from all participants. Youth and their current caregivers were interviewed by separate interviewers, typically at the child's residence. Structured youth interviews took between 3 and 4 hours and were typically completed over two sessions. Structured caregiver interviews took approximately 2 hours and were typically completed in one session. Children and caregivers were each paid $40 for their participation.
Trained research assistants coded each child's legal petition and social history (child welfare records' narrative of the history and events preceding the legal filing that led to the child's removal from the home) using the Maltreatment Classification System (MCS; Barnett et al., 1993). The developers of the rating system report adequate estimates of interrater agreement, with kappas ranging from .60-1.0 (Manly, Cicchetti, & Barnett, 1994). In the current study, emotional maltreatment, physical abuse, sexual abuse, failure to provide, and lack of supervision were coded as present or absent for each participating child. Only maltreatment that occurred within the previous two years was coded because of concerns that not all caseworkers would reliably include information about past history of abuse. The relationship of the perpetrator to the victim and the severity of each maltreatment type were coded using the MCS. Also coded was the presence or absence of maternal history for criminal behavior, alcohol abuse/dependence, other substance use, mental illness, and history of being abused or neglected as a child. Maternal characteristics were coded because most youth (88.8%) had been removed from their biological mother. All records were consensus coded by at least two trained staff and discrepancies were resolved through consultation with one of the senior investigators.
In our sample of 243 children, 160 (65.8%) were coded as having been emotionally maltreated based on the MCS. Perpetrators of emotional maltreatment were predominantly mothers (54.3%), followed by both parents/caregivers (27.6%), other caregivers (10.6%), and fathers (7.5%).
The legal petitions and social histories of the 160 youth who were coded for emotional maltreatment were recoded for the presence or absence of each of the following, more specific, subtypes of emotional maltreatment (all of which were included in the broader Emotional Maltreatment coding of the MCS): (a) verbal abuse, thwarts self-esteem, and/or blames child (n = 37); (b) intimidation and/or threats to child (n = 21); (c) age-inappropriate autonomy (n = 5); (d) inappropriate level of responsibility and/or parentification (n = 49); (e) abandonment (n = 48); (f) exposure to parental suicide gestures/attempts (n = 8); (g) exposure to violence (n = 69); (h) physical abuse with an emotional maltreatment component (e.g., binding, smothering; n = 10); (i) failure to protect child from a known sexual perpetrator and/or sexual abuse with emotional component (n = 29); and (j) extreme, unpredictable, and/or inappropriate behaviors (n = 28). These codes were non-exclusive, and of the 160 children who were re-coded the majority (n = 80) had one subtype code, 36 had two subtype codes, 30 had three subtype codes, and in one extreme case, the child had eight subtype codes.
Some of the newly created emotional maltreatment subtype categories did not have a large enough n (i.e., age-inappropriate autonomy, exposure to parental suicide gestures/attempts, physical abuse with an emotional maltreatment component) or were conceptually unclear (i.e., failure to protect child from a known sexual perpetrator/sexual abuse with emotional component and extreme, unpredictable, and/or inappropriate behaviors), and therefore were excluded from future analyses. Also, replicating what others have found (Allen, 2008; Linder & Collins, 2005), the data of the current study suggested that those children who were coded as being emotionally maltreated solely due to their exposure to violence (n = 25), were faring better on self-reports of functioning. These children, therefore, were removed from the overall emotional maltreatment group before analyses were conducted (although they were retained in the violence exposure category). Finally, because the first two subtypes (i.e. verbal abuse and intimidation/threats) seemed conceptually similar and are often reported together as verbal aggression, we combined these subtypes into a verbal aggression category for subsequent analyses.
Consequently, five different categories of emotional maltreatment were used in all subsequent analyses: (a) overall emotional maltreatment (n = 135, excluding those who were coded as emotionally maltreated due solely to violence-exposure), (b) verbal aggression (n = 48), (c) inappropriate level of responsibility/parentification (n = 49), (d) abandonment (n = 48), and (e) violence exposure (n = 69).
With the exception of the measure of life satisfaction, which was modified for use in the current study, all of the dependent measures have demonstrated validity and reliability with racially and ethnically diverse samples.
Attachment to caregivers, parents, and peers was assessed using a short form of the People in My Life measure (PML; Gifford-Smith, 2000). The PML is a 30-item, self-report measure, which was derived from a longer version of the measure (Cook, Greenberg, & Kusche, 1995). It includes two subscales, one measuring attachment to parents (which was used to measure attachment to substitute caregivers as well as parents in the current study) and one measuring attachment to peers. The short form has demonstrated high internal consistency (α = .80 to .83) and its subscales reliably discriminate between samples of high risk and normative youth. Sample PML items include questions asking youth to rate whether their caregivers, parents, and peers “help me with my problems,” “care about me,” “listen to me,” “accept me,” and “are proud of me.” Sample items also include questions asking whether youth “share thoughts and feelings” with caregivers/parents/friends and “feel angry” with caregivers/parents/friends. Mean scores range from 1 (not true) to 3 (often true), with higher scores indicating greater perceived attachment.
Social acceptance and self-esteem were assessed using the Social Acceptance and Global Self Worth subscales of the Self-Perception Profile for Children (SPPC; Harter, 1982, 1985). The SPPC is a widely used self-report measure of perceived self-competence for 3rd through 8th graders that assesses self-esteem in six domains. All subscales have demonstrated adequate reliability and validity (Harter, 1985). Estimates of internal consistency for both the Social Acceptance (α = .75 to .80) and Global Self Worth (α = .78 to .84) scales are acceptable, and the two scales are moderately correlated with each other and with the other subscales measuring athletic competence, physical appearance, behavioral conduct, and scholastic competence (r = .20 to .73). Mean scores on the Social Acceptance and Global Self Worth scales range from 1 to 4, with higher scores indicating greater perceived social acceptance and self worth.
Anxiety was assessed using the Total Anxiety score from the Revised Children's Manifest Anxiety Scale (RCMAS; Reynolds & Paget, 1981; Reynolds & Richmond, 1978, 1979, 2000, 2008). The RCMAS is a self-report measure of anxiety that has demonstrated good reliability and validity. It is normed for male and female youth aged 6-19. Standardized scores are T scores, having a mean of 50 and a standard deviation of 10.
Posttraumatic stress symptoms were assessed using the Posttraumatic Stress subscale of the Trauma Symptom Checklist for Children (TSCC; Briere, 1996). The TSCC is a 54-item, 6-scale, self-report measure of posttraumatic stress and related symptomatology, which includes normative data on 3,008 children ages 8 to 16 sampled from the general population. All of the TSCC subscales, including the Posttraumatic Stress subscale, have demonstrated adequate reliability (α = .77 to .89) and subscale scores covary in the predicted directions with a number of other measures sharing similar content. As with RCMAS scores, standardized scores on the TSCC are T scores.
Life satisfaction was assessed using a modified version of the Delighted-Terrible Scale (Andrews & Withey, 1976). The 11-item self-report scale asks respondents to rate their satisfaction in several different domains (e.g., school, home, health, friendships, leisure activities). The items have good internal consistency and construct validity (Andrews & Withey, 1976). Mean scores range from 1 to 3, with higher scores indicating greater life satisfaction.
Social, emotional, and behavioral problems were assessed with the Social Problems and Total Problems subscales of the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001). The CBCL is a widely used caregiver-report measure of child behavior problems. Caregivers rate 113 items on a 3-point scale to yield data on a variety of childhood behavior problems. The CBCL has been standardized separately for males and females in different age groups using a national probability sample that is racially and ethnically diverse. Standardized scores are t scores, which have a mean of 50 and a standard deviation of 10.
Since little is known about the subtypes of emotional maltreatment, a series of analyses was first run to determine whether the prevalence of distinct subtypes differed by demographic characteristics, other subtypes of emotional abuse, maternal characteristics, and/or by other types of maltreatment. Differences were identified using t-tests and chi-square statistics. Next, t tests were used to estimate the bivariate relationships between each of the emotional maltreatment types and the 10 dependent variables. These bivariate analyses were run for the total sample and for each gender separately. Finally, when the bivariate analysis suggested that an emotional maltreatment type was associated with a dependent variable, multiple regression was used to estimate the effect of the emotional maltreatment type on the dependent variable after controlling for the other four types of maltreatment (sexual abuse, physical abuse, failure to provide, and lack of supervision). All analyses were conducted using the SPSS statistical program, version 16.
There were no age or racial/ethnic differences as a function of any type of emotional maltreatment. Males, however, were more likely to be abandoned, χ2(1, N = 243) = 3.88, p = .049, and there was a trend suggesting that males were more often exposed to violence, χ2(1, N = 243) = 3.14, p = .076.
As shown in Table 1, children who experienced verbal abuse were more likely to have experienced abandonment, χ2(1, N = 243) = 9.3, p = .002. Children who experienced inappropriate levels of responsibility were also more likely to have been exposed to violence, χ2(1, N = 243) = 6.31, p = .012, and there was a trend suggesting that these children to were more likely to have experienced abandonment, χ2(1, N = 243) = 3.01 p = .083.
As summarized in Table 2, our findings suggest that emotional maltreatment types are, for the most part, not related to maternal characteristics including criminal history, alcohol or substance use problems, mental illness, or maltreatment history. There were only two exceptions. Youth who were identified as emotionally maltreated were less likely to have mothers with a maltreatment history, χ2(1, N = 243) = 4.69, p = .030, and youth who were forced to assume inappropriate levels of responsibility were more likely to have mothers with substance abuse problems, χ2(1, N = 243) = 3.99, p = .046.
The findings in Table 2 suggest that there are significant associations between emotional maltreatment types and other types of maltreatment. Youth who were identified as emotionally-maltreated were more likely to have been physically abused, χ2(1, N = 243) = 7.36, p = .007, and sexually abused, χ2(1, N = 243) = 7.50, p = .006. Youth who were identified as victims of verbal aggression were more likely to have been physically abused, χ2(1, N = 243) = 40.56, p = .000, and sexually abused, χ2(1, N = 243) = 4.06, p = .044, but less likely to have been coded for failure to provide, χ2(1, N = 243) = 4.94, p = .026, or lack of supervision, χ2(1, N = 243) = 5.16, p = .023. Youth who were forced to assume inappropriate levels of responsibility were more likely to have been coded for lack of supervision, χ2(1, N = 243) = 5.71, p = .017, and there were trends suggesting that they were less likely to be sexually abused χ2(1, N = 243) = 2.81, p = .093, and more likely to have experienced failure to provide, χ2(1, N = 243) = 3.20, p = .074. There was a third trend suggesting that youth who had been abandoned were less likely to have been sexually abused, χ2(1, N = 243) = 2.67, p = .102. Finally, exposure to violence was unrelated to the other types of maltreatment.
Table 3 summarizes the bivariate associations between the five emotional maltreatment types and each of the 10 dependent variables for the total sample and for each gender separately.
In the total sample of maltreated children, emotional maltreatment was associated with higher levels of self-reported posttraumatic stress symptoms, t(241) = 2.68, p = .008, but was not associated with any of the other dependent variables. Verbal aggression was associated with lower levels of self-reported social acceptance t(239) = -2.22, p = .027, and self-esteem, t(239) = -2.91, p = .004, and with higher levels of self-reported posttraumatic stress symptoms, t(241) = 2.27, p = .024, but was not associated with the caregiver-report measures of internalizing and externalizing symptoms. There was also a trend suggesting that verbal aggression may be associated with higher levels of self-reported anxiety, t(241) = 1.71, p = .089. Inappropriate responsibility was associated with fewer caregiver-reported social problems, t(240) = -2.18, p = .030, but was not associated with any of the other dependent variables. Abandonment was associated with lower levels of attachment to peers, t(240) = -1.66, p = .098, lower life satisfaction, t(240) = -1.67, p = .096, and higher levels of self-reported anxiety, t(241) = 2.11, p = .036. Violence exposure was associated with lower levels of self-reported posttraumatic stress symptoms, t(241) = -1.99, p = .048. In addition, there was a trend suggesting that violence exposure may also be associated with fewer self-reported symptoms of anxiety, t(241) = -1.79, p = .074.
In the female sample, there were few associations between emotional maltreatment types and dependent variables. Among females, there was a trend suggesting that emotional maltreatment may be associated with higher levels of self-reported posttraumatic stress symptoms, t(115) = 1.75, p = .083, while inappropriate responsibility was significantly associated with lower levels of social problems, t(114) = -2.54, p = .012. There was also a trend suggesting that females' exposure to violence was associated with fewer self-reported symptoms of anxiety, t(115) = -1.69, p = .094. These findings were consistent with findings from analyses using the total sample. A finding not replicated with the total sample suggests that when females were forced to assume inappropriate levels of responsibility, they had fewer emotional and behavioral problems as indexed by caregiver-report, t(114) = -2.34, p = .021. Finally, a trend suggested that females who were abandoned may also have fewer emotional and behavioral problems, t(114) = -1.76, p = .081.
There were many more associations between the emotional maltreatment types and dependent variables for males, suggesting that emotional maltreatment may have more of an impact on the psychosocial development of males. In the male sample, as with the total sample, emotional maltreatment was associated with more self-reported symptoms of posttraumatic stress, t(124) = 2.01, p = .046. Abandonment was associated with higher levels of self-reported anxiety t(124) = 2.22, p = .028, and there was a trend suggesting that abandonment was also associated with lower levels of life satisfaction t(124) = -1.91, p = .058. Verbal aggression was associated with lower levels of self-esteem, t(123) = -3.69, p = .000, and there were trends suggesting that it was associated with higher levels of self-reported anxiety, t(124) = 1.80, p = .074, and lower levels of social acceptance, t(123) = -1.66, p = .100. Findings not replicated with the total sample suggest that emotionally maltreated males reported lower levels of attachment to peers, t(124) = -2.22, p = .028, and lower life satisfaction, t(124) = -2.11, p = .037, and that males who were subjected to verbal aggression reported lower levels of attachment to parents, t(121) = -1.96, p = .052.
When bivariate analyses suggested that there was an association between an emotional maltreatment type and a dependent variable, multiple regression was used to estimate that relationship controlling for physical abuse, sexual abuse, failure to provide, and lack of supervision.
After controlling for the other four abuse types, emotional maltreatment was associated with higher self-reported posttraumatic stress symptoms, β = .149, p = .024. There was a trend suggesting that verbal aggression was associated with lower levels of social acceptance, β = -.122, p = .083, and with lower self-esteem, β = -.123, p = .076, after controlling for the other maltreatment types. However, verbal aggression was no longer associated with self-reported anxiety or with self-reported posttraumatic stress symptoms. A trend suggested that inappropriate responsibility was associated with fewer social problems, β = -.125, p = .055, after controlling for other types of maltreatment. There were also trends suggesting that abandonment was associated with lower levels of life satisfaction, β = -.108, p = .094, and with higher self-reported anxiety, β = .121, p = .062. However, after controlling for other maltreatment types, abandonment was no longer associated with attachment to friends. Violence exposure was associated with lower levels of posttraumatic stress symptoms, β = -.129, p = .045, and a trend suggested that it was also associated with lower levels of self-reported anxiety symptoms, β = -.122, p = .057.
In the female sample, inappropriate responsibility was associated with fewer caregiver-reported social problems after controlling for other types of maltreatment, β = -.236, p = .014. There were trends suggesting that inappropriate responsibility, β = -.184, p = .057, and abandonment, β = -.186, p = .056, were associated with fewer emotional and behavioral problems after controlling for other maltreatment types. Emotional maltreatment was no longer associated with self-reported posttraumatic stress symptoms, and violence exposure was no longer associated with self-reported anxiety when controlling for other types of maltreatment.
In the male sample, after controlling for other maltreatment, emotional maltreatment was associated with higher self-reported posttraumatic symptoms, β = .171, p = .054, lower life satisfaction, β = -.186, p = .040, and lower levels of attachment to friends, β = -.198, p = .029. Verbal aggression was associated with lower self-esteem, β = -.257, p = .008, and lower levels of attachment to parents, β = -.195, p = .052. Verbal aggression was no longer associated with social acceptance or with self-reported anxiety after controlling for the other types of maltreatment. Abandonment was associated with lower life satisfaction, β = -.192, p = .032, and higher levels of self-reported anxiety, β = .195, p = .031.
The current study extends our knowledge about the effects of childhood emotional maltreatment in several ways. First, using a prospective design, the impact of emotional maltreatment was analyzed in a representative sample of maltreated children with a recent entry into out-of-home care. Second, the focus of this study was not only on the relationships between emotional maltreatment and mental health symptoms, but also on the relationships between emotional maltreatment and interpersonal functioning, employing both youth self-report and caregiver reports of functioning. The use of subtypes to describe emotional maltreatment and its associations with functioning also broadens the knowledge-base of the field, as most previous research has examined emotional maltreatment as a homogeneous construct.
The descriptive analyses that were conducted suggest that males and females were equally likely to experience emotional maltreatment of any type (including verbal abuse, inappropriate responsibility, and violence exposure), except that males were more likely to be abandoned. There was some overlap among the emotional abuse subtypes, most notably between verbal abuse and abandonment and between inappropriate responsibility and violence exposure. There was no evidence of an association between age and emotional maltreatment or emotional maltreatment subtypes, but our failure to find an effect may be attributable to the fact that our sample had a limited age range (9- to 11-year-old children). Future studies should examine the rate of emotional maltreatment across the age spectrum, as differences in prevalence by age would have implications for prevention efforts.
In informing prevention efforts, it is important to identify those risk factors associated with the perpetration of emotional maltreatment. In the current study, the associations between maternal characteristics (as coded from child welfare records) and perpetration of emotional maltreatment were examined. Maternal characteristics were targeted because the majority of the emotional maltreatment perpetrators in the current study were mothers and most children were removed from their mothers. In general, maternal characteristics, which included criminal history, alcohol and substance use problems, mental illness, and maltreatment history, were not related to emotional maltreatment perpetration. Future studies should examine the associations between maternal characteristics and emotional maltreatment with a sample that is restricted to those children who have been emotionally maltreated by their mothers. Preliminarily, however, these analyses suggest that one cannot identify which children are at risk for emotional maltreatment based on their mothers' risk factors. This differs from other studies, which find that maternal characteristics predict perpetration of emotional maltreatment (Chamberland et al., 2005; Mullen et al., 1996). Finally, the overlap between emotional maltreatment and other types of maltreatment was examined. As other studies have found (Chamberland et al., 2005; Crittenden et al., 1994; McGee et al., 1997; Rodgers et al., 2004; Wolfe & McGee, 1994), there was a great deal of overlap between emotional maltreatment types and other types of maltreatment, although there was less overlap between abandonment and violence exposure and other types of maltreatment.
After examining the descriptive characteristics, bivariate and multivariate analyses were conducted to examine the impact of emotional maltreatment and subtypes of emotional maltreatment on social and related functioning. Regression results, after controlling for other types of maltreatment, suggest that it is important to evaluate the effects of emotional maltreatment subtypes. An exclusive focus on the effects of an overall measure of emotional maltreatment would have masked some key subtype differences. For example, it was found that verbal aggression was related to lower perceived social acceptance and self-esteem, but overall emotional maltreatment was not. This theoretically makes sense as verbal aggression, which often belittles children, may make children feel bad about themselves and may make them feel that others do not like them. Abandonment was related to greater anxiety and less life satisfaction, whereas overall emotional maltreatment was not. Similarly, inappropriate responsibility was related to fewer social problems as reported by current caregivers, but overall emotional maltreatment was not. Children who have been forced to care for other children and to maintain household functioning may display parentified behaviors that are viewed by other caregivers as helpful or mature. Finally, exposure to violence was associated with fewer self-reported posttraumatic stress symptoms. Although this may seem counterintuitive, it is important to remember that in this sample, all children had experienced maltreatment. Those children who were coded as having been emotionally maltreated solely because they had been exposed to violence displayed the fewest problems. Thus, witnessing violence may be less traumatic than being the subject of violence, whether verbal, physical, or sexual. These findings are similar to those observed in other studies that compare children who have been exposed to domestic violence with children who have been abused or neglected (Allen, 2008; Chamberland et al., 2005; Lewis et al., 2006; Linder & Collins, 2005).
As the findings of the current study suggest, some emotional maltreatment subtypes (e.g., verbal aggression, abandonment) are positively associated with psychosocial problems, while others (e.g., inappropriate responsibility, violence exposure) are negatively associated with such problems, relative to other types of maltreatment. The examination of a global category of emotional maltreatment, therefore, may hide critical findings. It is also important to examine a range of outcomes from multiple reporters (Morimoto & Sharma, 2004; McGee et al., 1997), as some important, yet subtle differences were found in the current study. Finally, given the overlap between emotional maltreatment and other types of maltreatment, it is critical to control for the impact of other types of maltreatment in order to isolate the unique effects of emotional maltreatment (Berzenski & Yates, in press; Sachs-Ericsson et al., 2006; Vissing et al., 1991; Zurbriggen, Gobin, & Freyd, in press).
Studies should also examine outcomes by gender, as previous research has found inconsistent results with regard to gender differences concerning the impact of emotional maltreatment (Kim & Cicchetti, 2006; McGee et al., 1997; Morimoto & Sharma, 2004; Vissing et al., 1991; Wolfe & McGee, 1994). In the current study, females who had been emotionally maltreated were faring better on some indices of mental health and social functioning than maltreated females who experienced other types of maltreatment. On the other hand, males who were emotionally maltreated endorsed more posttraumatic stress and anxiety symptoms and reported lower levels of life satisfaction, less attachment to friends, lower self-esteem, and less attachment to their biological parents than did maltreated males who experienced other types of maltreatment. In an effort to understand why males might be more negatively impacted by emotional maltreatment than females, a closer examination of the data revealed that the males in our sample experienced a greater severity of emotional maltreatment than did females, t(132) = 2.58, p = .011. A recent study found that the association between emotional abuse and adolescent competence was mediated by social withdrawal, but only for males (Shaffer, Yates, & Egeland, 2009). Future studies should examine gender by emotional maltreatment interactions while controlling for severity of emotional maltreatment. In addition, an examination of the relationship between the gender of the perpetrator (mother, father) and the gender of the victim may illuminate any differential impact of emotional abuse.
One of the limitations of the current study is that each of the subtypes of emotional maltreatment was not coded for severity. These data would have been helpful in understanding the topography of emotional maltreatment and for analyzing the data to better understand the impact of different subtypes. It is recommended that a standardized method for coding emotional maltreatment subtypes and associated severity be developed because emotional maltreatment, as currently defined in the literature, is a very heterogeneous construct and is often operationalized differently in each study. This would enhance researchers' ability to compare findings across studies and to examine the impact of subtypes within a developmental framework.
Another limitation of the current study was that multiple analyses were conducted and the level of significance was not adjusted, thereby increasing the possibility that our findings may be attributable to chance. The significance levels were not adjusted because of the exploratory nature of this study and because the associations between emotional maltreatment and psychosocial functioning have not been previously studied with a sample of preadolescent youth placed in out-of-home care. For this reason, trends were also noted in the tables and accompanying text. Another limitation was the use of caregivers as informants, as the length of time that they knew the child was variable. In addition, the generalizability of these findings are limited by the fact that the sample only included youth within a limited age range and those who were living in out-of-home care as a result of court order.
Finally, the current study did not include a control group of non-maltreated children who were matched on other, important sociodemographic variables. It would have been helpful to compare three groups of children: those who experienced emotional maltreatment, those who were maltreated but did not experience emotional maltreatment, and a non-maltreated group. This may have better clarified the unique impact of emotional maltreatment on psychosocial outcomes for youth. A strength of the current design, however, was its ability to control for other types of maltreatment, as only eight children (3% of the sample) experienced solely emotional maltreatment.
The results of the current study suggest that emotional maltreatment may impair psychosocial functioning and related outcomes. Adult reports of the impact of childhood emotional maltreatment suggest that problems experienced by youth who are emotionally maltreated will not resolve on their own (see Carbone, this issue; Dodge Reyome, in press; Riggs, this issue; Riggs & Kaminski, this issue). Longitudinal studies are needed to examine not only the impact of emotional maltreatment over time, but salient risk and protective factors developmentally. Analyses of mediation and moderation (e.g., Morimoto & Sharma, 2004; Sachs-Ericsson et al., 2006; Wright, 2007), as well as analyses evaluating the impact of preventive interventions, should help to identify key mechanisms in the development of healthy social functioning for children whose lives have been impacted by grossly inappropriate parenting.
This project was supported by grants from the National Institute of Mental Health (1 R21 MH067618, and 1 R01 MH076919-03, H. Taussig, PI) and the Kempe Foundation, Pioneer Fund, and Daniels Fund. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Mental Health or the National Institutes of Health. We thank Edward Garrido and Tali Raviv for their review of this manuscript, and Laura Mann, Jennifer Hughes, and Tabitha Webster for their invaluable support in the preparation of the manuscript. We also thank the Denver, Adams, Arapahoe, and Jefferson County Departments of Social Services for their partnership in our research efforts. Finally, we wish to express our appreciation to the children and families who made this research possible.