You are on page 1of 22

Child Development, xxxx 2019, Volume 00, Number 0, Pages 1–22

Emotion Reactivity and Regulation in Maltreated Children: A Meta-Analysis


Iris Lavi Lynn Fainsilber Katz
University of Haifa University of Washington

Emily J. Ozer James J. Gross


University of California, Berkeley Stanford University

The many adverse effects of child maltreatment make the scientific investigation of this phenomenon a matter
of vital importance. Although the relationship between maltreatment and problematic emotion reactivity and
regulation has been studied, the strength and specificity of these associations are not yet clear. We examine
the magnitude of the maltreatment—child-emotion reactivity/regulation link. Studies with substantiated mal-
treatment involving children aged up to 18 were included, along with a smaller number of longitudinal stud-
ies (58 papers reviewed, encompassing more than 11,900 children). In comparison to nonmaltreated children,
maltreated children experience more negative emotions, behave in a manner indicative of more negative emo-
tion, and display emotion dysregulation. We outline several theoretical implications of our results.

The topic of child maltreatment has occupied Violato, 2001). In this review, we will follow the
researchers in the past 6 decades, since the term classic distinction between sexual abuse and other
“battered child syndrome” gained currency forms of maltreatment (Belsky, 1993) and address
(Kempe, Silverman, Steele, Droegemueller, & Silver, outcomes related to physical abuse, neglect, and
1962). Interest in this complex, significant issue emotional maltreatment.
remains high, with more than 300 publications with One central focus in research on child maltreat-
the term “child maltreatment” in the title in 2017 ment has been emotional processes of maltreated
alone (Google Scholar database, retrieved on children, including emotion reactivity and regula-
August 9, 2018). Child maltreatment includes acts tion (e.g., Kim & Cicchetti, 2010; Shields & Cicchetti,
or omission of acts by a parent or other caregiver 2001; Skowron, Cipriano-Essel, Gatzke-Kopp, Teti,
that result in harm, potential for harm, or threat of & Ammerman, 2014). Emotion reactivity is the reac-
harm to a child (Leeb, Paulozzzi, Melanson, Simon, tion when an emotion arises—the subjective feeling,
& Arias, 2008). Acts of child maltreatment are fre- behavioral manifestation, and bodily response. Emo-
quently divided into four subtypes: physical abuse, tion reactivity varies in its valence (negative or posi-
neglect, emotional maltreatment, and sexual abuse tive) and intensity (strong or weak; Gross & Jazaieri,
(Barnett, Manly, & Cicchetti, 1993; Cicchetti & Toth, 2014; Gross & Thompson, 2007). Emotion regulation
2005). Most studies addressing maltreatment is the way individuals influence which emotions
addressed several subtypes of maltreatment while they have, when they have them, and how they
excluding sexual abuse (e.g., Schofield, Conger, & experience and express them (Gross, 1998).
Conger, 2017; Teisl & Cicchetti, 2008; Young & In this study, we examine the reactivity–maltreatment
Widom, 2014), recognizing that, in the literature, and regulation–maltreatment relationships sepa-
sexual abuse frequently incorporates acts commit- rately, and compare the magnitudes of both connec-
ted by other persons close to the child and not tions. We also investigate the relationships between
exclusively by the parents (e.g., Paolucci, Genuis, & maltreatment and individual subprocesses within
reactivity and regulation, such as intensity of anger,
This research was supported in part by the Haruv Institute and compare magnitudes of the connections. These
Postdoctoral Fellowship and by the Marie Curie International comparisons shed light on which specific emotional
Outgoing Fellowship awarded to Iris Lavi. The funding source processes are most closely linked to maltreatment,
did not influence the findings or the report and the authors
report no conflict of interest.
Correspondence concerning this article should be addressed to
Iris Lavi, University of Haifa, 199 Aba Khoushy Ave. Mount Car- © 2019 Society for Research in Child Development
mel, Haifa 3498838, Israel. Electronic mail may be sent to All rights reserved. 0009-3920/2019/xxxx-xxxx
iris.lavi.01@gmail.com. DOI: 10.1111/cdev.13272
2 Lavi, Katz, Ozer, and Gross

which has several potential benefits, namely: (a) during the early years influences many adjustment
better understanding of long-term effects of child outcomes, including the future ability to self-regu-
maltreatment, (b) clearer theoretical understanding late emotions (Mikulincer et al., 2003). In the first
of emotional processes in complex social conditions, year of life, the child uses proximityseeking to care-
and (c) more effective programs to help victims of givers, the primary and innate attachment strategy
maltreatment that target specific areas of difficulties (Bowlby, 1988). When caregivers respond posi-
to which they are vulnerable. tively, successful proximity-seeking leads to a
secure attachment: the feeling that the world is safe,
and that one can explore it and other humans
Child Maltreatment and Children's Emotion Reactivity
safely (Mikulincer et al., 2003). A securely attached
and Regulation
child has a secure base script (Waters, Rodrigues, &
Multiple theoretical models have sought to Ridgeway, 1998)—a regulation strategy that
define and explain the development of child emo- includes acknowledgment and display of distress,
tion reactivity and regulation following experiences support seeking, and engagement in instrumental
of maltreatment. The first model, focusing on problem solving.
attachment, addresses the impact of shortfalls of the It has been proposed that maltreatment increases
parent–child emotional bond on emotion reactivity the probability of insecure attachment via two
and regulation of children (Bowlby, 1988; Cicchetti paths (Mikulincer et al., 2003, figure 3): parental
& Lynch, 1995; Cyr, Euser, Bakermans-Kranenburg, rejection and parental intrusiveness. These parental
& Van Ijzendoorn, 2010; Main & Hesse, 1990; behaviors are characteristic of many, if not most,
Mikulincer & Shaver, 2007; Mikulincer, Shaver, & maltreatment subtypes (e.g., Belsky & Jaffee, 2006).
Pereg, 2003). Next, the social information process- Maltreating-rejecting parents punish the child for
ing theory suggests that maltreatment leads to the proximity seeking and emphasize self-reliance,
development of dysfunctional encoding of external encouraging the child to refrain from exhibiting
cues, resulting in high levels of child hostility, needs and emotions. The child thus distances her-
anger, and aggression (Crick & Dodge, 1994; Cum- self from her own emotions in an attempt to avoid
mings, Pellegrini, Notarius, & Cummings, 1989; painful experience of negative affect, that is, she
Grych & Fincham, 1990). A third model suggests develops hyporeactivity to both negative and posi-
that maltreated children learn dysfunctional reactiv- tive emotions, coupled with an inability to regulate
ity and regulation from their parents (Eisenberg emotions when these are activated. Maltreating-in-
et al., 2001); this learning process is enhanced by truding parents compulsively interfere with child’s
the fact that maltreating parents show a rigid and exploration and emphasize a sense of helplessness
restricted repertoire of emotional processes, and by and deficits in self-regulation. Children react with
the social isolation that is characteristic of maltreat- hyperactivating strategies of anxiety and fear, and
ing families and which limits other available role experience fewer opportunities to develop adequate
models (summary of models and predictions is pre- abilities to self-soothe, such that they are character-
sented in Supporting Information). ized by hyper-reactivity to negative emotions and
hyporeactivity to positive emotions (Mikulincer
et al., 2003).
Attachment-Focused Theories
Further research has indicated the relationship
Attachment is the inborn tendency of infants to between experiences of maltreatment and the devel-
seek proximity to a caring individual in times of opment of insecure attachment is specifically rele-
stress and/or discomfort (Bakermans-Kranenburg vant to the emergence of disorganized attachment
& van IJzendoorn, 2016). It is the emotional bond (Baer & Martinez, 2006; Cyr et al., 2010). Disorga-
that connects a child to his/her caregivers (Ains- nized attachment is characterized by a condition of
worth, 1973; Bowlby, 1951, 1969). These early bonds "fright without solution" for the child, indicating an
are immensely important for the development of absence of a strategy to deal with stress (Main &
the child and have an impact on physical, cognitive, Hesse, 1990). A child displaying disorganized
emotional, and social well-being (Ainsworth, 1973; attachment is nervous and anxious, and is unre-
Bowlby, 1951, 1969). A sensitive and caring attach- sponsive when rejoining the parent after separation
ment figure is regarded a safe haven and as a (Main, Kaplan, & Cassidy, 1985). Maltreatment has
secure base for the child, and this function is a key been found to be strongly related to the develop-
component of a well-functioning attachment bond ment of disorganized attachment in children (Cyr
(Bowlby, 1988). The development of attachment et al., 2010). In a meta-analysis, Cyr et al. (2010)
Emotion in Maltreated Children 3

reported that maltreated children were less secure develop sufficient social interaction skills through
and more disorganized than other high-risk chil- trial and practice (Cicchetti, 1996). This constellation
dren. Maltreatment leads the child to experience a of factors may contribute further to the develop-
paradox: The parent is the only source of comfort ment of insufficient skills in the regulation of emo-
and at the same time is abusive and frightening tions.
(Van Ijzendoorn, Schuengel, & Bakermans-Kranen- The three theories presented show the different
burg, 1999). Patterns of disturbances in attachment facets of a process by which maltreated -children
lead to elevated sensitivity to stress, dysregulation develop a distinct pattern of emotion reactivity and
of emotions, and difficulties in self-regulation of regulation. Attachment theory -stresses the prob-
emotions (Cook et al., 2017; Lyons-Ruth & Jacob- lematic emotional bond between child and parent,
vitz, 2016). a bond that would result in difficulties developing
self-regulation, learning theory addresses -the lack
of models for such behaviors, and the social-infor-
Social Information Processing
mation theory addresses biases in -encoding and
The second major theoretical framework linking interpretation of external cues. The theories con-
maltreatment to the development of emotion reac- verge in the prediction that -maltreated children
tivity and regulation focuses on problems in cogni- would be highly reactive to negative emotions. For
tive processing. In the social information processing example, we would -expect these children to show
model (Crick & Dodge, 1994), processing of social high levels of sadness and anger. We would also
information begins with encoding and interpreta- predict that these -children would show high levels
tion of cues; accurate execution of these two initial of behaviors indicative of such emotions. Indeed,
steps is mandatory for later stages of the informa- according to -attachment theory, when children are
tion processing. Maltreatment in the early years can faced with consistent rejection from their parents,
lead to alterations in this initial stage (Crick & hypo-sensitivity to positive and negative emotions
Dodge, 1994), for instance, greater accuracy in per- might develop. In addition, the models also predict
ception of anger in adults’ faces (Pollak, Messner, dysregulation: -difficulty in self-management of
Kistler, & Cohn, 2009). That is, we would expect to emotions, coping, and problem solving.
see hyper-reactivity to negative emotions in mal- Importantly, the theoretical models suggest a
treated children, specifically, anger, fear, and causal link between parental behaviors -and chil-
aggression. dren’s emotional processing. It is also likely that
there are reverse causal or bi-directional links—
from child to parent and from parent to child (Avi-
Social Learning
nun & Knafo, 2014).
According to social learning perspectives, children
observe their parents as models, including patterns
Research Questions and Hypotheses
of reactivity and regulation (Bandura, 1997). In non-
maltreating families, parents aid the child in devel- Multiple models indicate that children exposed
oping skills such as inhibiting negative emotions, to maltreatment are likely to develop dysfunctions
self-soothing, and directing attention in social con- in emotion reactivity and regulation. The models
texts (Eisenberg et al., 1999; Gottman, Katz, & Hoo- described suggest three distinct paths linking mal-
ven, 1996). In maltreating families, parents who are treatment and child reactivity and regulation: (a)
hostile toward their children and exhibit negative dysregulation via insecure attachment with the mal-
emotions model dysregulated behavior for their treating parent; (b) dysfunctions in interpretation of
children (Eisenberg et al., 2001). These learned neg- cues, which result in hyper-reactivity to negative
ative affect and poor regulation strategies are trans- emotions; and (c) social learning of dysfunctional
ferred to the child’s interactions with peers (Parke, reactivity and regulation by observing these pat-
Cassidy, Burks, Carson, & Boyum, 1992). The ten- terns in the maltreating parent. The fact that many
dency to react aggressively may cause further diffi- maltreating families are socially isolated potentially
culties in learning emotion regulation techniques enhances all of these patterns, as the children have
(Repetti, Taylor, & Seeman, 2002). In addition, mal- few other role models and are highly dependent on
treated children often live in chaotic and disorga- their (maltreating) parents. Although existing mod-
nized homes with little social support in or outside els distinguish among various parental behaviors,
the family (Cicchetti, 1996; Stith et al., 2009). This these theories do not articulate the relationship of
social isolation further reduces their opportunity to these parenting behaviors to the major subtypes of
4 Lavi, Katz, Ozer, and Gross

maltreatment such as physical abuse and neglect. author and two research assistants examined titles
Rather, extant theories focus on the consequences of and abstracts independently and 10% in duplicate.
specific parental behaviors that cut across the vari- Papers were excluded for three reasons: (a) did not
ous maltreatment subtypes (e.g., rejection). To our include relevant predictor variable of child mal-
knowledge, hypothesized links between subtypes of treatment and relevant outcome variables of either
maltreatment and various emotion reactivity/regu- emotion reactivity or emotion (dys)regulation; (b)
lation processes have not yet been tested. included only sexual abuse, not other subtypes of
We present the following two primary hypothe- child maltreatment; or (c) did not examine the rela-
ses, based on the theoretical models described. Our tionship between child maltreatment and emotion
first primary hypothesis is that in comparison to reactivity and regulation. This stage resulted in
nonmaltreated children, maltreated children will exclusion of 4,920 papers, leaving 646. In the sec-
show hyperreactivity to negative emotions and ond stage, the research team examined remaining
hyporeactivity to positive emotions. In secondary papers in their entirety independently and 10% in
analyses, we explore the link between child mal- duplicate. To ensure articles were relevant to our
treatment and specific constructs within child emo- study, we used the following inclusion/exclusion
tion reactivity and examine differences in emotion criteria: (a) included if examined maltreatment
reactivity between neglected and abused children. directed toward children aged 0–18; (b) included if
Our second primary hypothesis is that in compar- examined the relationship between emotion regula-
ison to nonmaltreated children, maltreated children tion (of the maltreating parent) and child maltreat-
will show dysregulation of emotion. We also ment—316 papers excluded. Last, because cross-
explore the strength of the link between child mal- sectional and longitudinal studies constitute the lit-
treatment and specific constructs within child emo- erature (as there are no randomized controlled trials
tion regulation and examine differences in emotion given ethical considerations), we chose to include
regulation between neglected and abused children. studies of any design that either (c1) used a
prospective design and/or (c2) used a cross-sec-
tional design and relied on official records of his-
Method tory of child maltreatment—240 papers were
excluded (of which 170 addressed reactivity/regula-
Identification of Articles in Sample
tion of children and 70 addressed reactivity/regula-
Articles were located using the search host Pro- tion of parents); further details regarding search
Quest to conduct computerized searches of four strategies and criteria appears in Supporting Infor-
databases (conducted in October 2017): PsycINFO, mation; (d) excluded if was qualitative with no sta-
ERIC, Social Work, and PubMed. The search focused tistical data—16 papers excluded; (e) excluded if
on three major constructs: reactivity, emotion regu- the same sample, variables and results were pre-
lation/emotion dysregulation, and child maltreat- sented in more than one article—6 papers excluded;
ment; list of search terms appears in Supporting in most cases, effect sizes were identical in the two
Information. We chose highly inclusive search articles, when this was not the case, the more recent
terms in order not to miss relevant papers, did not article was included; (f) excluded if emotion regula-
have a publication date limit, and included pub- tion was of the maltreating parent and not the child
lished and unpublished manuscripts. Given —10 papers excluded. An outline of the exclusion
resource limitations, only studies published in Eng- criteria used at this stage is provided in Supporting
lish were included. The initial search resulted in a Information. At this stage, 588 papers were
pool of 6,003 papers. We also used the reference excluded. The remaining 58 papers were included
lists of papers in the meta-analysis to create a list of in the meta-analysis.
prominent authors and then conducted a focused To test reliability of the screening process, a sub-
search for relevant articles by these authors. We set of independent second screening and coding
found 59 papers during this search, including dis- was conducted. Two master’s students screened
sertations and reports. We then excluded dupli- and coded a sample that included 10% of the
cates, resulting in a pool of 5,566 papers. A flow papers. The master’s students included 100% of the
diagram for location of papers appears in Figure S1 papers that the research team included in the first
in Supporting Information. stage and included additional papers as well. At
The research team screened the pool of papers to the second stage—the full-text eligibility screening—
determine which were relevant to this meta-analy- there was 65.34% agreement, with higher rates of
sis. This was done in two stages. First, the first overinclusion by the students. Reliability of coding
Emotion in Maltreated Children 5

was between 0.61 and 1.00, with 54% of reliability established distinction -among experience, behavior,
equaling 1.00. Coding reliability of statistical data and physiology, while addressing the valence of the
was between 0.97 and 1.00 (Cohen’s kappa was cal- emotion and, -where indicated, the specific emotion
culated for nominal measures and interclass correla- addressed (Gross & Jazaieri, 2014; Gross & Thomp-
tion for continuous measures). All disagreements son, 2007). Thus, for example, we coded indications
were resolved in subsequent coding meetings. of behaviors of anger and experiences of anger as
two distinguishable manifestations of emotions
(both in the category of reactivity, the first under
Sample of Studies
the major construct of behavior and the second
Majority of studies included reported multiple under major construct of experience); within these
analyses, for a total of 131 separate effect sizes. major constructs, these emotions are under the
However, as described below, each meta-analysis minor construct of anger. Another set of examples
included just one effect size per independent sam- pertain to the valence of emotion. Such coding
ple, to ensure the statistical independence of all entails, for instance, -positive emotions (e.g., experi-
effect sizes. Studies are listed in Table 1, with char- ence low level of happiness, under the major con-
acteristics of each study. Variables coded from each struct of -experience), or negative emotions
study appears in Supporting Information. (behaving in a way indicative of high levels of
anger, under the -major construct of behavior). -
These categories and constructs appear in Table 2,
Measures of Maltreatment
along with justifications regarding allocation of
Most of the studies included in the meta-analysis minor constructs to the various major constructs;
used official records of maltreatment or observa- the scales used appear in Supporting Information.
tional data taken by child welfare professionals as
an indication of past maltreatment, either as an
Emotion Reactivity Category
indication that the child had been maltreated or as
an indication that the parent was maltreating. Of This category includes three major constructs:
the 58 studies, four utilized scales to measure mal- experience, physiology, and behavior. Experience
treatment, and these were the only studies that did was determined using scales measuring for nega-
not utilize official records to establish child mal- tive emotions, such as anger, sadness, and guilt,
treatment. These studies used prospective designs, and positive emotions, such as happiness. The
and the maltreatment was measured in the first physiology construct was examined by looking at
wave of the longitudinal design. The maltreatment physiological reactions such as skin conductance
scales used in these four studies were the Abusive and respiratory sinus arrhythmia (RSA). The behav-
Behavior Cluster that uses home observation and ior indicators construct was examined using scales
the family problem-solving task, the Conflict Tactics measuring the behavioral manifestations of emo-
Scales (CTS-child report on parent; Straus, 1979), tions; specifically, the intensity and frequency of
the CTS (CTSPC-parent self-report; Straus, Hamby, aggression, anger, antisocial behavior, low positive
& Warren, 2003), and the Child Abuse and Neglect effect, and high negative effect.
Risk Evaluation—Short version (Schatz & Burke-
Lefever, 2003).
Emotion Dysregulation Category
Dysregulation was measured directly, with scales
Measures of Emotion Reactivity and Emotion
designed for this purpose, and indirectly, with
Regulation
scales designed to measure use of dysfunctional
In our examination of the child maltreatment lit- problem solving.
erature, we considered a range of variables relevant
to the categories of emotion reactivity and regula-
Calculation of Effect Sizes
tion. Although there is considerable -debate regard-
ing the distinctness of emotions, and the Given that most outcomes reported were based
differentiation between reactivity and -regulation on group differences in means (Johnson & Eagly,
(Gross & Feldman Barrett, 2011), we make use of 2000), we calculated standardized mean difference
the general distinction between -the process of reac- (SMD) effect sizes (Cohen’s d) to index differences
tivity and the process of regulation. -Within the in mean levels of emotion reactivity and regulation
general category of reactivity, we follow the well- for children who were or were not maltreated. All
6
Table 1
Studies in Meta-Analysis

Cs versus Parents’ Child's Child age in Total Maltreatment Reactivity/reg-


Study Location long gender gender yearsa n sub-type ulation ES (95% CI)

Bank and Burraston (2001) USA L Both Boys 10 182 PA Beh. Agg. 0.56 (0.33, 0.79)
N Beh. Agg. 0.32 ( 0.02, 0.66)
Bennett, Sullivan, and Lewis (2005) USA CS Mots Both 5 (0.8) 177 PA Beh. Neg. 0.18 ( 0.22, 0.57)
PA Beh. Ang. 0.09 ( 0.65, 0.48)
N Beh. Neg. 0.20 ( 0.14, 0.53)
N Beh. Ang. 0.04 ( 0.51, 0.43)
Lavi, Katz, Ozer, and Gross

PA, N Beh. Neg. 0.20 ( 0.23, 0.62)


PA, N Beh. Ang. 0.34 ( 0.2, 0.88)
Bennett, Sullivan, and Lewis (2010) USA L Mots Both 5 111 N Exp. Neg. 0.33 (0.06, 0.59)
Bousha and Twentyman (1984) USA CS Mots N/A 4.4 36 PA Beh. Agg. 1.4 (0.47, 2.33)
N Beh. Agg. 1.07 (0.33, 1.8)
Chavez (2006) USA L Both Both 11 or less 1,196 PA, N, SA Dys. 0.06 ( 0.01, 0.14)
Cipriano (2010) USA CS Mots Both 3.75 (0.74) 85 PA, N Beh. Ang. 0.29 ( 0.74, 0.16)
PA, N Beh. Neg. < 0.01 ( 0.45, 0.45)
PA, N RSA 0.24 ( 0.24, 0.72)
PA, N Dys. 0.07 ( 0.37, 0.52)
Connor, Doerfler, Volungis, Steingard, and USA CS Both Both 12.76 (3.24) 58 PA, SA Beh. Agg. 1.18 (0.92, 1.45)
Melloni (2003) PA, SA Beh. Neg. 0.54 (0.02, 1.07)
Dadds, Mullins, McAllister, and Atkinson Australia CS Mots Both 3.3 (1.33) 60 PA Beh. Pos. 1.92 (1.29, 2.56)
(2003) PA Beh. Neg. 0.59 (0.04, 1.14)
PA Exp. Neg. 0.44 ( 0.11, 0.98)
Darwish (1999) USA CS Both Both 4 (0.44) 30 PA, PN, EM, SA Beh. Agg. 0.54 ( 0.19, 1.27)
PA, PN, EM, SA Dys. 0.84 (0.1, 1.59)
de Pa
ul and Arruabarrena (1995) Spain CS Both Both 5–11 66 PA Beh. Agg. 0.59 ( 0.16, 1.34)
N Beh. Agg. 0.67 ( 0.03, 1.37)
Delker, Noll, Kim, and Fisher (2014) USA L Mots Both 3 240 PA, EA Dys. 0.26 (0.01, 0.52)
Dileo, Brewer, Northam, Yucel, and Australia CS N/A Both 8.48 (1.75) 50 PA, N, AE, SA Beh. Agg. 0.51 ( 0.07, 1.08)
Anderson (2017) PA, N, AE, SA Beh. Neg. 0.91 (0.32, 1.5)
PA, N, AE, SA Dys. 0.66 (0.08, 1.24)
Ewing-Cobbs, Prasad, Kramer, and Landry USA CS Both Both 3.5 or less 56 PA Dys. 1.09 ( 0.47, 2.65)
(2000)
Finzi, Har-Even, and Weizman (2003) Israel CS Both Both 9.77 (1.8) 114 PA Exp. Neg. 2.94 (2.17, 3.72)
PA Dys. 1.26 (0.65, 1.86)
N Exp. Neg. 1.71 (1.06, 2.36)
N Dys. 0.09 ( 0.48, 0.66)
Finzi et al. (2002) Israel CS Both Both 9.77 (1.8) 114 PA Beh. Agg. 1.93 (1.26, 2.6)
N Beh. Agg. 0.46 (0.17, 0.75)
Ghirmay (2015) USA L N/A Both 0–11 807 PA, N, SA Dys. 0.20 (0.06, 0.34)
Continued

Cs versus Parents’ Child's Child age in Total Maltreatment Reactivity/reg-


Study Location long gender gender yearsa n sub-type ulation ES (95% CI)

Gordis, Feres, Olezeski, Rabkin, and USA CS Both Boys 12.1 (1.21) 187 PA, PN, EM, SA Beh. Agg. (B) 0.29 (0, 0.59)
Trickett (2009) PA, PN, EM, SA Beh. Agg. (G) 0.66 (0.34, 0.99)
PA, PN, EM, SA Phy. (B) 0.02 ( 0.28, 0.31)
PA, PN, EM, SA Phy. (G) 0.16 ( 0.48, 0.17)
PA, PN, EM, SA RSA (B) 0.15 ( 0.14, 0.45)
PA, PN, EM, SA RSA (G) 0.11 ( 0.21, 0.44)
Harris (2010) USA CS Both Both 12.63 (2.2) 80 PA Exp. Ang. 0.66 (0.21, 1.11)
Haskett, Allaire, Kreig, and Hart (2008) USA CS Both Both 7.2 (1.4) 153 PA Beh. Agg. 0.34 (0.02, 0.66)
Haskett (1990) USA CS Both Both 4.86 (0.51) 18 PA Dys. 1.32 (0.36, 2.29)
Hennessy et al. (1994) USA CS Both Both 8.73 (1.47) 88 PA Exp. Neg. 0.47 ( 0.04, 0.97)
Herrenkohl, Klika, Herrenkohl, Russo, and USA L Both Both 1.5–6 357 PA Beh. Ang. 0.43 (0.22, 0.64)
Dee (2012) N Beh. Ang. 0.18 ( 0.03, 0.39)
Hjorth and Ostrov (1982) USA CS Both Both 12–16 49 PA Dys. 0.70 (0.12, 1.28)
Houck and King (1993) USA CS Both Both 6–11 225 PA, N, SA Beh. Agg. 0.41 ( 0.01, 0.83)
PA, N, SA Dys. 0.58 ( 0.29, 1.44)
Jaffee, Caspi, Moffitt, and Taylor (2004) USA L Both Both 5 1,116 PA Beh. Agg. 0.32 (0.21, 0.44)
Keil and Price (2009) USA CS Both Both 6.5 188 N Beh. Agg. 0.34 ( 0.11, 0.79)
PA, N Beh. Agg. 0.57 (0.11, 1.03)
Kim and Cicchetti (2010) USA CS Both Both 8.11 (1.77) 421 PA, PN, EM, SA Dys. 0.44 (0.25, 0.63)
Kinard (1980) USA CS Both Both 5–12 60 PA Beh. Agg. 0.34 ( 0.02, 0.7)
Klorman, Cicchetti, Thatcher, and Ison USA CS Both Both 6.65 (2.37) 212 PA, N, EM, SA Phy. 0.10 ( 0.29, 0.09)
(2003)
Koenig, Cicchetti, and Rogosch (2000) USA CS Mots Both 3.61 (0.11) 89 PA Beh. Neg. 0.20 ( 0.23, 0.63)
PA Beh. Pos. 0.59 ( 0.03, 1.22)
N Beh. Neg. 0.39 ( 0.09, 0.87)
N Beh. Neg. 0.28 ( 0.3, 0.85)
Lansford et al. (2002) USA L Mots Both 5 444 PA Beh. Agg. 0.77 (0.48, 1.06)
Ma & Li, 2014) Hong CS Both Both 9–15 283 PA, SA Dys. 0.46 (0.26, 0.65)
Kong
Main and George (1985) USA CS Both N/A 1–3 20 PA Beh. Agg. 1.79 (0.76, 2.83)
Manly, Kim, Rogosch, and Cicchetti (2001) USA CS Both Both 7.51 (1.45) 814 PA Beh. Agg. 0.56 (0.29, 0.84)
N Beh. Agg. 0.47 (0.17, 0.76)
EM Beh. Agg. 0.65 (0.27, 1.03)
N EM Beh. Agg. 0.43 (0.13, 0.73)
PA Dys. 0.62 (0.42, 0.81)
N Dys. 0.41 (0.08, 0.73)
EM Dys. 0.48 (0.22, 0.75)
Emotion in Maltreated Children

N EM Dys. 0.39 ( 0.37, 1.15)


7
8
Continued

Cs versus Parents’ Child's Child age in Total Maltreatment Reactivity/reg-


Study Location long gender gender yearsa n sub-type ulation ES (95% CI)

Maughan and Cicchetti (2002) USA CS Mots Both 5.31 (1.04) 139 PA Beh. Agg. 0.60 (0.25, 0.95)
N Beh. Agg. 0.45 (0.11, 0.80)
McGinn (2014) Canada CS Fats Both 9.26 (2.45) 31 PA Dys. 1.28 (0.19, 2.36)
McLaughlin, Peverill, Gold, Alves, and USA CS N/A Both 16.58 (1.39) 42 PA, SA Dys. 0.51 ( 1.04, 2.05)
Sheridan (2015)
Milojevich (2016) USA CS N/A Both 12.51 (3.27) 395 PA, N. SA Beh. Agg. 0.35 (0.14, 0.56)
Lavi, Katz, Ozer, and Gross

PA, N. SA Dys. 0.22 ( 0.02, 0.47)


Moretti and Craig (2013) Canada L Both Both 15.34 (1.53) 179 PA, EA Dys. (Mthr) 0.35 (0.05, 0.64)
PA, EA Dys. (Fthr) 0.12 ( 0.42, 0.18)
Reidy (1977) USA CS Both Both 6.61 (1.05) 58 PA Beh. Agg. 0.89 (0.44, 1.33)
N Beh. Agg. 0.27 ( 0.39, 0.93)
Robinson et al. (2009) USA CS Mots Both 2.73 (0.89) 123 PA, PN, EM, SA Beh. Ang. 1.95 (1.52, 2.38)
PA, PN, EM, SA Beh. Pos. 1.06 (0.68, 1.43)
PA, PN, EM, SA Dys. 0.29 ( 0.07, 0.65)
Schatz, Smith, Borkowski, Whitman, and USA L Mots Both 3 126 CM Dys. 0.22 ( 0.03, 0.47)
Keogh (2008)
Schofield et al. (2017) USA L Both Both About 18 290 PA, EM Dys. 0.37 (0.08, 0.66)
Shenk, Noll, and Cassarly (2010) USA CS Both Girls 15.75 (1.11) 211 PA, PN, SA Dys. 0.54 (0.26, 0.82)
Shenk, Putnam, Rausch, Peugh, and Noll USA CS Both Girls 17 (1.16) 110 PA, PN, SA RSA 0.3 ( 0.08, 0.69)
(2014)
Shenk, Griffin, and O'Donnell (2015) USA L N/A Girls 17 (1.17) 110 PA, N, SA Exp. Neg. 0.43 (0.05, 0.81)
PA, N, SA Dys. 0.61 (0.22, 0.99)
Shields and Cicchetti (1998) USA CS Both Both 8.97 (1.87) 228 PA Beh. Agg. 0.45 (0.06, 0.84)
PA Beh. Neg. 0.52 (0.13, 0.91)
PA Dys. 0.53 (0.14, 0.92)
N Beh. Agg. 0.06 ( 0.33, 0.46)
N Beh. Neg. 0.26 ( 0.14, 0.65)
N Dys. 0.24 ( 0.16, 0.63)
Shields and Cicchetti (2001) USA CS Both Both 8.98 267 PA, N, EM, SA Dys. 0.42 (0.17, 0.67)
Shipman and Zeman (2001) USA CS Mots Both 9.07 (2.03) 50 PA Beh. Neg. 0.83 (0.25, 1.41)
PA Dys. 0.74 (0.40, 1.07)
Shipman, Edwards, Brown, Swisher, and USA CS Mots Both 9.17 (2.09) 48 N Beh. Neg. 0.52 ( 0.18, 1.23)
Jennings (2005) N Dys. 0.63 (0.22, 1.04)
Shipman (1990) USA CS Mots Both 8.79 (2.04) 40 PA Exp. Ang. 0.07 ( 0.55, 0.69)
PA Exp. Neg. 0.57 (0.33, 0.81)
PA Beh. Agg. 0.91 (0.25, 1.56)
PA Beh. Neg. 0.96 (0.31, 1.62)
PA Dys. 0.86 (0.40, 1.32)
Continued

Cs versus Parents’ Child's Child age in Total Maltreatment Reactivity/reg-


Study Location long gender gender yearsa n sub-type ulation ES (95% CI)

Skowron et al. (2014) USA CS Mots Both 3.71 (0.72) 161 PA, PN, EM RSA 0.08 ( 0.26, 0.1)
Smith and Walden (1999) USA CS Mots Both 4.55 (0.67) 30 PA, N Exp. Neg. 0.22 ( 1.13, 1.56)
PA, N Beh. Agg. 0.10 ( 0.41, 0.61)
PA, N Phy. 0.14 ( 0.38, 0.66)
PA, N Dys. 0.25 ( 0.02, 0.51)
Stoddart (1993) USA CS Both Both 4.9 34 PA Exp. Neg. 0.93 (0.19, 1.66)
PA Beh. Agg. 0.73 (0.01, 1.46)
Teisl and Cicchetti (2008) USA CS Both Both 8.04 (1.63) 267 PA Beh. Agg. 0.40 (0.15, 0.66)
PA Dys. 0.49 (0.13, 0.85)
CM Beh. Agg. 0.16 ( 0.11, 0.43)
CM Dys. 0.31 ( 0.03, 0.66)
Veloz (2005) USA CS Mots Both 8.56 (1.97) 44 PA, N Dys. 0.61 (< 0.01, 1.22)
Vorderstrasse (2013) USA CS Both Both 10.54 (0.96) 200 PA Beh. Agg. 0.34 ( 0.08, 0.76)
Young and Widom (2014) USA L Both Both 0–11 547 PA Exp. Neg. 0.27 ( 0.08, 0.62)
N Exp. Neg. 0.29 (0.07, 0.51)

Note. CI = confidence interval; L = longitudinal; CS = cross-sectional; B = boys; G = girls; Mots = mothers; Fats = fathers; PA = physical abuse; N = neglect, EM = emotional mal-
treatment; SA = sexual abuse; CM = child maltreatment (not further detailed); Beh. = behavioral; Exp. = experience; Agg. = aggression; Ang. = Anger; Neg. = high negative affect;
Pos. = low positive affect; Dys. = dysregulation; RSA = respiratory sinus arrhythmia; Phy. = physiological reactivity; ES = effect size.
a
Mage (SD) or range.
Emotion in Maltreated Children
9
10 Lavi, Katz, Ozer, and Gross

Table 2
Categories, Major and Minor Constructs of Emotion Reactivity, and Emotion Regulation

Major construct Minor construct Description of minor construct

Category: emotion reactivity


Experience Anger Subjective feeling of annoyance and irritation
High negative affect Subjective feelings such as hostility, sadness, and disappointment
Low positive affect Subjective feelings such as joy, happiness, satisfaction
Behavior Aggression and antisocial behavior Behaviors indicative of verbal and/or physical hostility and violence toward
others (not the child; Aggression has been described as an indicator to
emotion reactivity, Averill, 1983)
Anger Behaviors indicative of annoyance and irritation
High negative affect Behaviors indicative of emotions such as hostility, sadness, and
disappointment
Low positive affect Behaviors indicative of emotions such as joy, happiness, satisfaction
Physiology Physiological reactivity Bodily responses indicative of changes in arousal
Respiratory sinus arrhythmia (RSA) An index of parasympathetic activity that addresses the vagal nerve influences
on the heart; achieved by measuring periodic fluctuations in heart rate that
are linked to breathing (Literature indicates RSA is an indicator of emotion
reactivity, Butler, Wilhelm, & Gross, 2006)

Description

Category: emotion dysregulation


Emotion dysregulation, dysfunctional coping, and problem solving
Attempts to change emotions and/or regulate emotions in a constructive way
Difficulty controlling emotions

effect sizes were coded in a unified direction such the statistical independence of effect sizes. When
that positive values (> 0) indicated that maltreat- studies reported outcomes for multiple, highly simi-
ment was associated with worse child outcomes, lar contexts, we calculated a weighted average
for example, higher negative emotion reactivity, effect size (weighting for dfs) and used this average
lower positive emotion reactivity, and higher emo- as a single, overall effect size for the construct for
tion dysregulation (i.e., lower emotion regulation). the study (using formulas in Borenstein, Hedges,
Where possible, we included a substantive transfor- Higgins, & Rothstein, 2009, pp. 227–228). This was
mation of the effect sizes, based on the measures done for all minor constructs, except reactivity-be-
utilized. Ten studies reported correlational data; havior-anger (in which there were no studies
sensitivity analysis, presented in Supporting Infor- reporting similar variables). Thus, we ensured that
mation, was conducted to examine aggregated for each meta-analysis, each unique sample con-
effect sizes while excluding these studies. In cases tributed only one effect size to the synthesis, mean-
where a study was relevant to the meta-analysis ing that all effect sizes in the synthesis were
but the data given were insufficient for our statisti- statistically independent. As there were too few
cal analysis, the authors were contacted and asked studies to enable a multilevel meta-analysis, we
to provide the missing data. The authors of five elected to structure major constructs conceptually
papers were contacted; only in one case was the and to ensure that each unique sample only con-
data provided by the author (Finzi, Har-Even, tributed one effect size to an outcome.
Shnit, & Weizman, 2002). In addition, although we considered using a
Effect size calculations were derived using multilevel approach to average across all the reac-
STATA (StataCorp., 2017). The studies included in tivity outcomes and create one larger reactivity
the analysis used a wide range of constructs and pooled effect (to directly compare to dysregulation
research designs. In order to use conventional outcomes), the results from the individual models
meta-analytic techniques, which assume the statisti- indicated substantial heterogeneity in some models
cal independence of effect sizes (Rosenthal, 1994), and a mix of both significant and nonsignificant
we composed a set of decision principles to ensure findings. Thus, we anticipated that any multilevel
Emotion in Maltreated Children 11

model across all these reactivity outcomes would participants was 8.12 years (MSD = 1.48). Most
also show significant variability and thus be some- studies were conducted in the United States, with
what unstable for a discussion comparing dysregu- 77.6% of the studies being cross-sectional (k = 45).
lation to reactivity (without explaining the Of the 58 papers located, one was published during
particular subtypes). Following this, we address the the 1970s (Reidy, 1977), four were published during
differences between reactivity and dysregulation the 1980s (Bousha & Twentyman, 1984; Hjorth &
descriptively and as ranges in the interpretation of Ostrov, 1982; Kinard, 1980; Main & George, 1985),
the results. nine during the 1990s (Darwish, 1999; de Pa ul &
We estimated pooled effect sizes using a ran- Arruabarrena, 1995; Haskett, 1990; Hennessy, Rabi-
dom-effects model, based on the assumption that deau, Cicchetti, & Cummings, 1994; Houck & King,
variability in effect sizes was due to sampling 1993; Shields & Cicchetti, 1998; Shipman, 1990;
error in addition to true variability (Lipsey & Wil- Smith & Walden, 1999; Stoddart, 1993), and the rest
son, 2001) because of the likely heterogeneity of of the reviewed papers (k = 44) during the 2000s.
the populations, samples, and constructs appearing This marked increase in the number of studies dur-
in the included studies (Hedges & Vevea, 1998). ing the 2000s suggests a growing interest in emo-
We examined heterogeneity in effect sizes by esti- tion reactivity and regulation of maltreated
mating Q (used to test the null hypothesis of effect children, as well as growing specificity in the theo-
sizes homogeneity), s2 (an estimate of the variance rizing of emotion reactivity, emotion regulation,
of the distribution of effect sizes), and I2 (the pro- and child maltreatment. Descriptive data of
portion of observed variability that is due to true reviewed studies appear in Table 1. Forest plots
heterogeneity rather than chance). Although we appear in Supporting Information.
had not prespecified moderator analyses, some of
our models had enough studies, and enough vari-
Reactivity-Experience-Anger
ability, to allow us to conduct -bivariate meta-re-
gressions for up to five variables, depending on Only two studies reported anger experience and
model (percent female, geography of study the weighted mean effect size for the relationship
[USA = 1 vs. Non-US = 0], child mean age, study between child maltreatment and child’s experience
design [Correlational = 0 vs. Longitudinal = 1], of anger was nonsignificant: SMD = .41 (95% CI
and maltreatment type [Neglect = 1, Abuse = 0]). [ 0.16, 0.98)]. This model had no remaining hetero-
Given the way some of these moderators were geneity, and thus, possible moderators were not
reported in primary studies, the moderator analy- examined.
ses is somewhat crude. For instance, although we
have conducted moderator analysis for percent
Reactivity-Experience-Negative Affect
gender, these outcomes represent the proportion of
a group in a study—that is, any analysis that indi- The weighted mean effect size for the relationship
cates this is a significant moderator suggests that between child maltreatment and child’s experiences
studies with a certain proportion of that gender of negative affect, averaged over 11 effect sizes, was
are more (or less) likely to demonstrate the out- significant: SMD = .71 (95% CI [0.41, 1.01]). This
comes rather than the ability to say something model had a significant amount of heterogeneity
about maltreatment outcomes by gender more (Q = 60.43, p < .001, s2 = .193, I2 = 83.5%); however,
specifically. this heterogeneity appeared largely driven by a sin-
gle study, which when removed resulted in a signif-
icant effect size (SMD = .40, 95% CI [0.29, 0.51]) and
no remaining heterogeneity.
Results
Prior to presenting the main sections of the results,
Reactivity-Behavior-Aggression
descriptive data of the studies included in the
meta-analysis are described in Supporting Informa- As shown in Table 3, the weighted mean effect
tion and will be addressed. More than 11,900 chil- size for the relationship between child maltreatment
dren were assessed in studies included in our meta- and child’s aggressive behavior, averaged over 38
analysis, with a mean number of children in each effect sizes, was 0.53 (95% CI [0.43, 0.62]). This indi-
study being 205.45 (SD = 247.65, range = 18–1,196). cates that maltreated children were significantly
Across studies, the mean age of parent-participants more likely to display aggression in comparison to
was 33.36 years (MSD = 6.00) and of child- nonmaltreated children. This model also had a
12 Lavi, Katz, Ozer, and Gross

Table 3
Meta-Analysis Results: ESs by Category and by Major and Minor Constructs

K (n) ES LL UL v2(df) p I2 s2

Reactivity
Experience
Experience—anger 2 (120) 0.41 0.16 0.98 2.25 (1) .13 55.70 .10
Experience—high neg. affect 11 (1,795) 0.71 0.41 1.01 60.43 (10) < .01 83.50 .19
Behavior
Behavior—aggression 38 (9.029) 0.53 0.43 0.62 95.48 (37) < .01 61.20 .05
Behavior—anger 7 (1,453) 0.36 0.09 0.82 69.15 (6) < .01 91.30 .33
Behavior—high neg. affect 14 (1,556) 0.38 0.24 0.53 16.70 (13) .21 22.20 .02
Behavior—low pos. affect 4 (361) 0.96 0.34 1.58 15.78 (3) < .01 81.00 .32
Physiology
Physiological reactivity 4 (604) 0.07 0.21 0.07 1.32 (3) .72 0.00 .00
Physiological—RSA 5 (718) 0.08 0.07 0.22 4.90 (4) .30 18.30 .01
Dysregulation 38 (10,099) 0.42 0.32 0.51 118.92 (37) < .01 68.90 .04

Note. ES is standard difference in means. Effects reported are under random model. Negative affect = Reactivity to other negative affect
(e.g., shame, guilt, disgust, sadness, hostility); Q (Cochran’s Q) = squared deviations of each study’s estimate from the overall meta-ana-
lytic estimate, weighting each study’s contribution (Higgins, Thompson, Deeks, & Altman, 2003). RSA = respiratory sinus arrhythmia;
ES = effect size.

significant amount of heterogeneity (Q = 95.48,


Reactivity-Behavior-Anger
p < .001, s2 = .05, I2 = 61.2%) and so potential mod-
erators were examined. Across the five bivariate As shown in Table 3, the weighted mean effect
meta-regression models, none of the moderators size for the relationship between child maltreatment
explained any of the variability in outcomes (see and child’s angry behavior, averaged over seven
Table 4), indicating that this effect may not be stable effect sizes, was nonsignificant: SMD = .36 (95% CI
for certain children with certain characteristics. [ 0.09, 0.82]). This model also had a significant

Table 4
Moderator Analysis: Effect Sizes by Category and by Major and Minor Constructs

Effect modifier k b [95% CI] Constant s2 Residual I2 (%) Adjusted R2

Reactivity-behavior-aggression
Gender (% female) 17 0.05 [ 2.12, 2.02] 0.48 [ 0.41, 1.37] .01 50.1 64.57
Geography (USA = 1) 38 0.24 [ 0.61, 0.12] 0.74 [0.40, 1.09] .04 61.18 0.54
Child age (mean of sample) 38 0.00 [ 0.05, 0.04] 0.55 [0.17, 0.93] .05 61.22 8.21
Design: cross-sectional versus longitudinal 38 0.05 [ 0.35, 0.26] 0.53 [0.42, 0.65] .05 61.32 8.13
(longitudinal = 1)
Maltreatment type (abuse = 0, neglect = 1) 28 0.24 [ 0.54, 0.05] 0.66 [0.50, 0.82] .06 67.4 9.67
Reactivity-behavior-anger
Child age (mean of sample) 7 0.58 [ 1.30, 0.14] 2.75 [ 0.27, 5.77] .31 87.61 39.36
Design: cross-sectional versus longitudinal 7 0.08 [ 1.80, 1.65] 0.38 [ 0.57, 1.33] .62 92.63 22.62
(longitudinal = 1)
Dysregulation
Gender (% female) 14 0.12 [ 0.71, 0.73] 0.49 [0.16, 0.83] 0 21.59 0
Geography (USA = 1) 38 0.01 [ 0.23, 0.25] 0.40 [0.18, 0.61] .03 69.25 5.31
Child age (mean of sample) 38 0.01 [ 0.01, 0.03] 0.34 [0.13, 0.54] .03 64.66 1.13
Design: cross-sectional versus longitudinal 38 0.29 [ 0.44, 0.15] 0.44 [0.39, 0.55] .01 34.27 71.11
(longitudinal = 1)
Maltreatment type (abuse = 0, neglect = 1) 20 0.18 [ 0.55, 0.20] 0.54 [0.37, 0.71] .05 56.94 13.05

Note. Effect size is standard difference in means. Effects reported are under random model. Negative affect = reactivity to other nega-
tive affect (e.g., shame, guilt, disgust, sadness, hostility); Q (Cochran’s Q) = squared deviations of each study’s estimate from the overall
meta-analytic estimate, weighting each study’s contribution (Higgins et al., 2003).
Emotion in Maltreated Children 13

amount of heterogeneity (Q = 69.15, p < .001, This moderator explained a large portion of the vari-
s2 = .33, I2 = 91.3%), and so potential moderators ability (residual I2 = 33.27%, adj R2 = 71.11%; see
were examined. One study appeared to be a poten- Table 4).
tial outlier so the model was rerun with this study Substantive transformation of the effect sizes was
removed. The overall mean effect did not substan- possible for only a small portion of the effect sizes
tially change, and there was still a large amount of due to the high variability in measures used (see
remaining heterogeneity so we used the original Supporting Information). Analyzing the measure
model to conduct moderator analysis. Only child used most prominently as an indication of negative
age and study design (cross-sectional vs. longitudi- emotion reactivity and of dysregulation, the Emo-
nal) had sufficient number of studies reporting data tion Regulation Checklist (Shields & Cicchetti, 1995,
on moderators, and both variables were not a sig- 1997), scored on a scale of 1–4, reveals that mal-
nificant moderators of the relationship between treated children scored 1.42 points higher on the
maltreatment and child angry behavior (see negative affect behavior scale and 0.57 points
Table 4). higher on the dysregulation scale, in comparison to
nonmaltreated children. This result is in accordance
with results indicating that reactivity effect sizes are
Reactivity-Behavior-Negative Affect and Positive Affect
higher than dysregulation effect sizes.
As shown in Table 3, the relationship for child
maltreatment and affect was significant for both
Summary Across Constructs
high negative affect (.38, 95% CI [0.241, 0.53],
k = 14) and low positive affect (.96, 95% CI [0.34, Results indicate that, in comparison to nonmal-
1.58]). This indicated that, in comparison to non- treated children, maltreated children are signifi-
maltreated children, children experiencing maltreat- cantly more likely to have negative effects on
ment were significantly likely to display both high regulation, aggression, high negative affect behav-
negative and low positive affect. Tests for homo- iors and experience, and low positive affect behav-
geneity indicated that there was no significant iors with effect sizes ranging from 0.41 to 0.96;
heterogeneity among high negative affect outcomes, most of the reactivity effect sizes are slightly higher
and although there was heterogeneity for low posi- in comparison to the dysregulation outcomes. Of
tive affect, there were too few studies to explore it. note, the physiological outcomes did not appear to
be significantly affected for maltreated children and
these effect sizes were less than half those for
Reactivity-Physiological Reactivity
behavior and experience outcomes. Overall, there is
Only a few studies reported physiological out- a great degree of variability in child outcomes
comes including reactivity and RSA; neither of the across the various constructs of interest.
models for these outcomes were significant, and
there was no heterogeneity to explore (see Table 3).
Publication Bias Analysis
Three methods were used to examine the possi-
Dysregulation
bility of small study bias, or bias due to the omis-
As shown in Table 3, the weighted mean effect sion of studies with negative results or small
size for the relationship between child maltreatment sample sizes: a funnel plot, the Egger test for funnel
and child’s dysregulation, averaged over 38 effect plot asymmetry, and a trim and fill analysis. These
sizes, was 0.42 (95% CI [0.32, 0.51]). This indicates were only conducted for the analyses with more
that, in comparison to children who were not mal- than 20 studies (dysregulation and aggression). The
treated, children who were maltreated were signifi- funnel plot was asymmetrical (i.e., there were few
cantly more likely to display emotion dysregulation. included studies with small sample sizes and nega-
This model also had a significant amount of hetero- tive findings) and Egger’s test was significant
geneity (Q = 118.92, p < .001, s2 = .04, I2 = 68.9%), (Egger b = 2.02, 95% CI [1.27, 2.77], p < .01 for dys-
and so potential moderators were examined. Across regulation and Egger b = 1.45, 95% CI [0.33, 2.56],
the five bivariate meta-regression models, only the p = .012), indicating possible small study bias. In
study design was a significant moderator the trim and fill analysis, the dysregulation out-
(b = 0.29, 95% CI [ 0.44, 0.15]), indicating that comes remained significant but were smaller (.24,
longitudinal studies were more likely to demon- 95% CI [0.15, 0.34]), and the aggression outcomes
strate smaller effects than cross-sectional studies. remained similar to the original analysis (.41, 95%
14 Lavi, Katz, Ozer, and Gross

CI [0.29, 0.52]). Thus, these results suggest the of emotion or of physiology. Below, we review
potential for small study bias in the effect sizes results of behavioral manifestations of emotions,
included in the synthesis, such that the magnitude followed by subjective reports on emotions, and
of this mean effect size may be upwardly biased; finally review bodily responses.
although the gaps in the funnel plot may simply be Results of the behavioral indicators of affect
due to a lack of studies indicating a reverse rela- show that, in comparison to nonmaltreated chil-
tionship than would be expected (i.e., maltreated dren, maltreated children exhibit much higher
children have more positive outcomes compared to levels of aggression. Maltreated children might
their nonmaltreated peers). Additionally, these tests react in a violent and destructive manner to their
could be driven to some extent by existing hetero- environment. Interestingly, results regarding anger
geneity in the included studies or other factors not were nonsignificant. Our findings are aligned with
related to publication bias (e.g., Lau, Ioannidis, Ter- prior research demonstrating the link between mal-
rin, Schmid, & Olkin, 2006). Given that we could treatment and conduct disorder and violent tenden-
not test for potential publication bias across all the cies (Andreou, Comasco, Aslund,  Nilsson, &
outcomes, it seems possible that there will be Hodgins, 2018; Docherty, Kubik, Herrera, & Boxer,
reporting bias or small study bias in this field. The 2018). In comparison to nonmaltreated children,
relevant funnel plots appears in Supporting Infor- maltreated children also show higher levels of
mation. behavioral manifestations of negative affect. None
of the reviewed studies addressed experiences of
positive emotions. This neglect of positive emotions
might represent a focus on the negative conse-
Discussion
quences of exposure to maltreatment.
This study addressed the question of which emo- Turning to experiences of emotions, in compar-
tional processes are the most vulnerable to the ison to nonmaltreated children, children who have
experience of growing up in a family challenged by been maltreated experience higher levels of sadness,
maltreatment. Our goal was to address this ques- hostility, and fear. Most of the effect sizes for this
tion via a theoretically informed synthesis of the analysis are based on informant-report measures
extant empirical literature on emotion reactivity (e.g., parent, teacher; only 3 of the 11 effects are
and regulation of maltreated children. Our meta- self-report). We further examined here whether, in
analysis is the first to our knowledge to examine comparison to nonmaltreated children, children
this relationship between children’s experience of who have been maltreated exhibit higher levels of
maltreatment and the development of emotion reac- physiological representations of emotion. Physiolog-
tivity and emotion regulation. Our emotion reactiv- ical data were presented in nine effect sizes from
ity findings indicate that, in comparison to six studies included in this meta-analysis. The
nonmaltreated children, maltreated children show aggregated average effect size for the physiological
aggressive behaviors, high levels of negative emo- data was not statistically significant.
tions, and low levels of positive emotion and expe- To summarize, we found that maltreated chil-
rience high levels of negative emotions. Examining dren tend to be aggressive, show high levels of
emotion regulation reveals that maltreated children negative affect, and experience high levels of sad-
have dysfunctional coping and problem-solving ness, fear and hostility. Our results are aligned
abilities and are emotionally dysregulated. Our with key theoretical models discussed previously.
results indicate that, for maltreated children, dys- Specifically, attachment theories (Bowlby, 1988;
functions in the various emotion reactivity con- Cicchetti & Lynch, 1995; Main & Hesse, 1990;
structs examined are stronger than dysfunctions in Mikulincer & Shaver, 2007), social information the-
emotion regulation. ories (Crick & Dodge, 1994; Cummings et al., 1989;
Grych & Fincham, 1990) and learning theories
(Eisenberg et al., 2001) are consistent in predicting
Emotion Reactivity in Maltreated Children
that maltreated children will have higher levels of
Addressing the components of emotions (experi- negative affect. Our results further demonstrate
ence, behavior, and physiology, Gross & Jazaieri, that low levels of positive emotion is the construct
2014; Gross & Thompson, 2007), we note that more that shows the greatest divergence between non-
studies examined the behavioral indicators of emo- maltreated and maltreated children. That is,
tion (e.g., crying as indicative of sadness, clenching although aggressive behaviors of children have
your jaws as indicative of anger) than experiences been studied extensively, the greatest difference
Emotion in Maltreated Children 15

between maltreated children and nonmaltreated exhibit more flexible and varied social behavior
children is in the levels of behavioral manifesta- (Cicchetti, 1996).
tions of positive emotion. Interestingly, for this outcome study design (con-
These results support the outlined hypotheses. trasting cross-sectional to longitudinal studies) was
All three reviewed theories converge in -predicting a significant moderator. That is, effect sizes for
hyper-reactivity to negative affect, describing how cross-sectional studies were significantly higher
experiences of maltreatment result -in the child than effect sizes for longitudinal data, with this
experiencing and expressing negative emotions. moderator explaining a significant portion of the
Attachment -theory (Mikulincer et al., 2003) also variance in dysregulation. We can compare this
describes specific patterns that indicate high levels result to the nonsignificant moderation of study
of hyporeactivity to positive emotions: this pattern design for behavioral indicators of aggression. This
is described as resulting from continuous -intrusive- might lead us to postulate that the relationship
ness on the part of the parents. The results of this between maltreatment and behavioral aggression is
study indicate that indeed, low -levels of behaviors long lasting, whereas the relationship between mal-
indicating positive emotions are characteristic of treatment and child dysregulation decreases with
maltreated children. Thus, -not only do maltreated time. It is possible that maltreated children adapt
children show high levels of sadness, anger, and more regulatory techniques with time/age, and this
hostility; these children -also show low levels of joy is represents the impact of maturation with age.
and happiness.
Reactivity and Regulation of Abused Versus Neglected
Emotion Regulation in Maltreated Children Children
Our second set of meta-analyses focused on emo- Our last set of findings concerning maltreated
tion regulation of maltreated and nonmaltreated children focused on the question of whether abused
children. Examining the different manifestations of and neglected children differ in levels of reactivity
emotion regulation, our results indicate that mal- and regulation. Only partial comparisons between
treated children exhibit dysregulation more than the groups were possible, due to the small number
nonmaltreated children. These children find it hard of studies comparing these groups. The low num-
to control their emotions—as reported by the child ber of studies reflects the high comorbidity between
and others in his/her environment. In addition, abuse and neglect. It also represents the under-
these children have difficulties with coping and standing that there might be differences between
with finding solutions to problems. abuse that is accompanied by neglect and abuse
Our results for emotion regulation of maltreated that is not accompanied by neglect (and similarly
and nonmaltreated children are consistent with the regarding neglect)—and thus many studies do not
theoretical frameworks discussed earlier. These compare between the two subtypes. With this con-
theories converge in predicting dysregulation for sideration, below we review studies that have
maltreated children. Attachment theories predict examined groups of families that were reported to
that dysregulation would result from child mal- demonstrate only abuse or only neglect.
treatment (Bowlby, 1988; Cicchetti & Lynch, 1995; In all constructs examined, fewer studies exam-
Main & Hesse, 1990; Mikulincer & Shaver, 2007; ined neglected children, in comparison to abused
Mikulincer et al., 2003) because maltreating parents children. This reflects the “neglect of neglect” that
fail to form a secure base for the child. This ren- has been reported when addressing child maltreat-
ders the child unable to develop beneficial emotion ment (Hobbs & Wynne, 2002). In two cases, com-
regulatory capabilities. In addition, the prediction parisons between abuse and neglect were possible,
made by learning theories, specifically addressing that is, in heterogeneous constructs and when the -
difficulties in problem solving (Eisenberg et al., number of studies was sufficient: reactivity-behav-
2001), was also supported. According to Eisenberg ior-aggression and dysregulation. In both cases,
et al. (2001), the child learns poor problem-solving maltreatment subtype (abuse vs. neglect) was non-
techniques from his/her maltreating parents, who significant. However, the results show a trend
resort to a very narrow set of disciplinary strate- toward effect sizes being lower in cases of neglect,
gies. Furthermore, many if not most maltreating in comparison to cases of abuse.
families are socially isolated (Cicchetti, 1996; Stith Our results demonstrate the relationship between
et al., 2009). This isolation can prevent the child exposure to maltreatment and alterations in the
from establishing other role models that would emotional processes of reactivity and regulation.
16 Lavi, Katz, Ozer, and Gross

The theories of attachment, social information -pro- together with understanding that families undergo
cessing, and learning all converge in their developmental stages that are more conflictprone
description of the outcomes of maltreatment— and in which escalating influences may become
hypernegativity and dysregulation. Although all even more salient.
theories predict similar outcomes, they indicate dif- Our results indicate a possible path for intergen-
ferent pathways for these outcomes. The theories erational transmission of maltreatment, a process
reviewed stress the main role of the parent as a that has been under study in recent decades, with
model for the child: thus, these theoretical frame- studies indicating such transmission in various cir-
works -indicate that the child is not only exposed cumstances (e.g., Berlin, Appleyard, & Dodge,
to a violent, abusive, and neglecting relationship, 2011). Our results suggest a possible mediator for
but is -also exposed to a distorted role model. this link: It is possible that when maltreatment
Neglect of the basic relational needs of the child results in difficulties in emotion reactivity and reg-
results in a significant vulnerability: results of our ulation, these difficulties characterize the child as
analyses indicate that this vulnerability is expressed s/he grows into an adult. These difficulties and
in greater negative emotions, lower positive emo- biases could be a risk factor when this child
tions, and difficulty controlling and managing these becomes a parent (Lavi, Ozer, Katz, & Gross, in
emotions. Our results further indicate that these submission).
alterations in reactivity are, in most cases, stronger
than alterations in regulation. Last, our results sug-
Limitations and Directions for Further Research
gest that, with time, regulation abilities might
improve while reactivity patterns remain stable. We note that the literature itself is characterized
These results have consequences for our under- by design limitations such as a relatively low pro-
standing of the development and maturation of portion of longitudinal designs and lack of speci-
emotional processes in children. ficity in the assessment of subcategories of
maltreatment and subcategories of emotional pro-
cesses. For example, only 13 of the 58 papers in the
Implications for Future Practice and Research
meta-analysis employed a longitudinal design,
Our results underscore the strong link between likely owing to challenges in following populations
maltreatment and children’s difficulties in the emo- of highly stressed families and children over time.
tional processes of reactivity and regulation. These It is also possible that although these families are
results could inform treatment of maltreated chil- already registered within the child welfare system,
dren. Therapists should be informed regarding the they are difficult to approach or it is challenging to
consequences of maltreatment of children and set gain their trust or cooperation due to the multiple
treatment goals accordingly. Understanding emo- problems they are facing. The significant differences
tion reactivity/regulation developmental phases in emotion reactivity/regulation found in our
and the relationship between emotion processes results underscore the need to examine these rela-
and attachment (Shaver & Mikulincer, 2014) could tionships while utilizing prospective designs. A
guide therapists in setting therapeutic goals for prospective design could help establish which cau-
maltreated children. For instance, the fact that low sal direction has more empirical support: Is it that
levels of positive affect is a characteristic of mal- maltreated children develop difficulties in emotion
treated children should be taken into account reactivity and regulation, or is it that children who
when setting therapeutic goals. Indeed, savoring are highly emotionally reactive and are emotionally
positive affect has been reported to be a therapeu- dysregulated are difficult to raise, and hence are at
tic strategy in family therapy (Lunkenheimer, greater risk of maltreatment? It is likely that both
Shields, & Cortina, 2007) and could be used to scenarios co-occur, as it has been noted that chil-
increase positive affect in maltreated children. Last, dren’s difficult behavior is a strong risk factor for
it is known that certain characteristics of the child maltreatment (Black, Slep, & Heyman, 2001; Stith
(e.g., difficult temperament) are risk factors for et al., 2009). Carefully designed prospective studies
maltreatment (Stith et al., 2009). Thus, maltreat- could help advance our knowledge regarding this
ment at a young age could lead to the develop- issue.
ment of aggressive and negative behavior of the Further, more attention should be given to the
child, leading to further maltreatment. Such poten- subcategories within child maltreatment—as our
tially escalating interaction patterns between the results indicate that the different subtypes could be
parent and child could be taken into account related to differential child outcomes. The
Emotion in Maltreated Children 17

understudied nature of emotional maltreatment (ne-


References
glect and abuse) is especially apparent (see also
Lavi et al., 2019). More nuanced examination of the
References marked with an asterisk indicate studies
emotional processes is also necessary: Many studies
included in the meta-analysis.
examined factors that pertain to emotional reactiv-
ity and emotion regulation as a single dimension, Ainsworth, M. D. S. (1973). The development of infant–
whereas others did not differentiate between the mother attachment. In B. Cardwell & H. Ricciuti (Eds.),
various processes within regulation. Last, there was Review of child development research (Vol. 3, pp. 1–94).
little use of up-to-date measurement approaches of Chicago, IL: University of Chicago.

Andreou, D., Comasco, E., Aslund, C., Nilsson, K. W., &
emotion reactivity and regulation (i.e., RSA).
Hodgins, S. (2018). Maltreatment, the oxytocin receptor
A second set of limitations concerns under-at-
gene, and conduct problems among male and female
tention to questions of psychological resilience in teenagers. Frontiers in Human Neuroscience, 12, 112.
the context of maltreatment. The reviewed studies https://doi.org/10.3389/fnhum.2018.00112
explored a variety of demographic and resilience Averill, J. R. (1983). Studies on anger and aggression:
factors (e.g., gender, peer relationship, e.g., Kim & Implications for theories of emotion. American Psycholo-
Cicchetti, 2010; Moretti & Craig, 2013). However, gist, 38, 1145–1160. https://doi.org/10.1037/0003-066X.
the great variability in these resilience factors and 38.11.1145
their measurement precludes the possibility of Avinun, R., & Knafo, A. (2014). Parenting as a reaction
generating a unified effect size for these moder- evoked by children's genotype: A meta-analysis of chil-
ated relations. We were only able to address a dren-as-twins studies. Personality and Social Psychology
Review, 18, 87–102. https://doi.org/10.1177/10888683
narrow range of moderators. Further studies are
13498308
needed in order to be able to address the impact
Baer, J. C., & Martinez, C. D. (2006). Child maltreatment
of such factors on the maltreatment-reactivity/reg- and insecure attachment: A meta-analysis. Journal of
ulation link. Reproductive and Infant Psychology, 24, 187–197.
Furthermore, it would be illuminating- to exam- https://doi.org/10.1080/02646830600821231
ine the moderating role of protective factors in the - Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H.
maltreatment-reactivity/regulation link while using (2016). Attachment, parenting, and genetics. In J. Cas-
the differential susceptibility (Belsky, 2005) and bio- sidy & P. R. Shaver (Eds.), Handbook of attachment: The-
logical sensitivity to context (Boyce & Ellis, 2005) ory, research, and clinical applications (pp. 155–179). New
frameworks. According to these -frameworks, out- York, NY: Guilford.
comes for children such as well-being and psycho- Bandura, A. (1997). Self-efficacy: The exercise of control.
New York, NY: W. H. Freeman.
logical distress are determined by the -interplay
*Bank, L., & Burraston, B. (2001). Abusive home environ-
between environmental stress (here, maltreatment
ments as predictors of poor adjustment during adoles-
vs. -benevolent parenting) and children’s sensitivity cence and early adulthood. Journal of Community
(here, reactivity/regulation). For a relationship -to Psychology, 29, 195–217. https://doi.org/10.1002/jcop.
match patterns of differential susceptibility, the 1014
environmental stressor and the sensitivity -should Barnett, D., Manly, J. T., & Cicchetti, D. (1993). Defining
not be related. To pursue this line of investigation, child maltreatment: The interface between policy
in future studies it will be important to explore and research. In D. Cicchetti & J. T. Manly (Eds.),
whether children -who are sensitive (here, highly Child abuse, child development, and social policy (Vol. 8,
reactive and dysregulated) would gain greater ben- pp. 7–73). Norwood, NJ: Ablex.
efit from -parental interventions, in comparison to Belsky, J. (1993). Etiology of child maltreatment: A
developmental-ecological analysis. Psychological Bulletin,
less sensitive children. Initial findings point in that -
114, 413–434. https://doi.org/10.1037/0033-2909.114.3.
direction (Scott & O’Connor, 2012).
413
Our results indicate that there is a significant Belsky, J. (2005). Differential susceptibility to rearing
relationship between exposure to child maltreat- influence. In B. Ellis & D. Bjoklund (Eds.), Origins of the
ment and development of hyper-reactivity to nega- social mind: Evolutionary psychology and child development
tive emotions, hyporeactivity to positive emotions (pp. 139–163). Now York, NY: Guilford.
and dysregulation. Emotion reactivity and regula- Belsky, J., & Jaffee, S. R. (2006). The multiple determi-
tion are crucial for the social, psychological, and nants of parenting. In D. Cicchetti & D. J. Cohen (Eds.),
behavioral development of children. Our study Developmental psychopathology. Vol. 3: Risk, disorder and
underscores the need to consider the consequences adaptation (pp. 38–85). New York, NY: Wiley.
of maltreatment for the basic, core emotional pro- *Bennett, D. S., Sullivan, M. W., & Lewis, M. (2005).
Young children's adjustment as a function of
cesses of children.
18 Lavi, Katz, Ozer, and Gross

maltreatment, shame, and anger. Child Maltreatment, 10, * Cipriano, E. A.( 2010). Emotion regulation in children
311–323. https://doi.org/10.1177/1077559505278619 exposed to maltreatment: The interplay between mal-
*Bennett, D. S., Sullivan, M. W., & Lewis, M. (2010). treatment status, parenting quality, and temperament.
Neglected children, shame-proneness, and depressive Doctoral dissertation. The Pennsylvania State Univer-
symptoms. Child Maltreatment, 15, 305–314. https://doi. sity.
org/10.1177/1077559510379634 *Connor, D. F., Doerfler, L. A., Volungis, A. M., Stein-
Berlin, L. J., Appleyard, K., & Dodge, K. A. (2011). Inter- gard, R. J., & Melloni, R. H. (2003). Aggressive behav-
generational continuity in child maltreatment: Mediat- ior in abused children. Annals of the New York Academy
ing mechanisms and implications for prevention. Child of Sciences, 1008, 79–90. https://doi.org/10.1196/anna
Development, 82, 162–176. https://doi.org/10.1111/j. ls.1301.009
1467-8624.2010.01547.x Cook, A., Spinazzola, J., Ford, J., Lanktree, C., Blaustein,
Black, D. A., Slep, A. M. S., & Heyman, R. E. (2001). Risk M., Cloitre, M., & Liautaud, J. (2017). Complex trauma
factors for child psychological abuse. Aggression and in children and adolescents. Psychiatric Annals, 35, 390–
Violent Behavior, 6, 189–201. https://doi.org/10.1016/ 398.
S1359-1789(00)00022-7 Crick, N. R., & Dodge, K. A. (1994). A review and refor-
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Roth- mulation of social information-processing mechanisms
stein, H. R. (2009). Introduction to meta-analysis. West in children's social adjustment. Psychological Bulletin,
Sussex, UK: Wiley. https://doi.org/10.1002/ 115, 74–101. https://doi.org/10.1037/0033-2909.115.1.
9780470743386 74
*Bousha, D. M., & Twentyman, C. T. (1984). Mother-child Cummings, J. S., Pellegrini, D. S., Notarius, C. I., & Cum-
interactional style in abuse, neglect, and control groups: mings, E. M. (1989). Children's responses to angry
Naturalistic observations in the home. Journal of Abnor- adult behavior as a function of marital distress and his-
mal Psychology, 93, 106–114. https://doi.org/10.1037/ tory of interparent hostility. Child Development, 60,
0021-843X.93.1.106 1035–1043. https://doi.org/10.2307/1130777
Bowlby, J. (1951). Maternal care and mental health (Vol. 2). Cyr, C., Euser, E. M., Bakermans-Kranenburg, M. J., & Van
Geneva, Switzerland-: World Health Organization. Ijzendoorn, M. H. (2010). Attachment security and disor-
Bowlby, J. (1969). Attachment and loss: Attachment. Vol. 1: ganization in maltreating and high-risk families: A series
Loss. New York, NY: Basic. of meta-analyses. Development and Psychopathology, 22,
Bowlby, J. (1988). A secure base: Clinical applications of 87–108. https://doi.org/10.1017/S0954579409990289
attachment theory. London, UK: Routledge. *Dadds, M. R., Mullins, M. J., McAllister, R. A., & Atkin-
Boyce, W. T., & Ellis, B. J. (2005). Biological sensitivity to son, E. (2003). Attributions, affect, and behavior in
context: I. An evolutionary–developmental theory of abuse-risk mothers: A laboratory study. Child Abuse and
the origins and functions of stress reactivity. Develop- Neglect, 27, 21–45. https://doi.org/10.1016/S0145-2134
ment and Psychopathology, 17, 271–301. https://doi.org/ (02)00510-0
10.1017/S0954579405050145 * Darwish, D.( 1999). A comparison of maltreated and
Butler, E. A., Wilhelm, F. H., & Gross, J. J. (2006). Respi- non-maltreated preschoolers’ social skills and play in
ratory sinus arrhythmia, emotion, and emotion regula- peer interactions. Doctoral dissertation. Fordham
tion during social interaction. Psychophysiology, 43, University, New York, NY.
612–622. https://doi.org/10.1111/j.1469-8986.2006. *de Pa ul, J., & Arruabarrena, M. I. (1995). Behavior prob-
00467.x lems in school-aged physically abused and neglected
* Chavez, J. M.( 2006). An examination of race/ethnicity children in Spain. Child Abuse and Neglect, 19, 409–418.
differences in depression and violence as conse- https://doi.org/10.1016/0145-2134(95)00009-W
quences of childhood maltreatment. Doctoral disserta- *Delker, B. C., Noll, L. K., Kim, H. K., & Fisher, P. A.
tion. University at Albany, State University of New (2014). Maternal abuse history and self-regulation diffi-
York. culties in preadolescence. Child Abuse and Neglect, 38,
Cicchetti, D. (1996). Child maltreatment: Implications for 2033–2043. https://doi.org/10.1016/j.chiabu.2014.10.
developmental theory and research. Human Develop- 014
ment, 39, 18–39. https://doi.org/10.1159/000278377 *Dileo, J. F., Brewer, W., Northam, E., Yucel, M., &
Cicchetti, D., & Lynch, M. (1995). Failures in the expect- Anderson, V. (2017). Investigating the neurodevelop-
able environment and their impact on individual devel- mental mediators of aggression in children with a his-
opment: The case of child maltreatment. In D. C. D. J. tory of child maltreatment: An exploratory field study.
Cohen (Ed.), Developmental psychopathology: Risk, disor- Child Neuropsychology, 23, 655–677. https://doi.org/10.
der, and adaptation (Vol. 2, 2nd ed., pp. 32–71). New 1080/09297049.2016.1186159
York, NY: Wiley. Docherty, M., Kubik, J., Herrera, C. M., & Boxer, P.
Cicchetti, D., & Toth, S. L. (2005). Child maltreatment. (2018). Early maltreatment is associated with greater
Annual Review of Clinical Psychology, 1, 409–438. risk of conduct problems and lack of guilt in adoles-
https://doi.org/10.1146/annurev.clinpsy.1.102803. cence. Child Abuse and Neglect, 79, 173–182. https://doi.
144029 org/10.1016/j.chiabu.2018.01.032
Emotion in Maltreated Children 19

Eisenberg, N., Fabes, R. A., Shepard, S. A., Guthrie, I. K., Grych, J. H., & Fincham, F. D. (1990). Marital conflict and
Murphy, B. C., & Reiser, M. (1999). Parental reactions children's adjustment: A cognitive-contextual frame-
to children's negative emotions: Longitudinal relations work. Psychological Bulletin, 108, 267–290. https://doi.
to quality of children's social functioning. Child Develop- org/10.1037/0033-2909.108.2.267
ment, 70, 513–534. https://doi.org/10.1111/1467-8624. * Harris, L. S.( 2010). Influence of maltreatment on execu-
00037 tive functions that underlie problem solving. Doctoral
Eisenberg, N., Gershoff, E. T., Fabes, R. A., Shepard, S. A., dissertation. University of California, Davis.
Cumberland, A. J., Losoya, S. H., & Murphy, B. C. *Haskett, M. E. (1990). Social problem-solving skills of
(2001). Mother's emotional expressivity and children's young physically abused children. Child Psychiatry and
behavior problems and social competence: Mediation Human Development, 21, 109–118. https://doi.org/10.
through children's regulation. Developmental Psychology, 1007/BF00706119
37, 475–490. https://doi.org/10.1037/0012-1649.37.4.475 *Haskett, M. E., Allaire, J. C., Kreig, S., & Hart, K. C.
*Ewing-Cobbs, L., Prasad, M., Kramer, L., & Landry, S. (2008). Protective and vulnerability factors for physi-
(2000). Inflicted traumatic brain injury: Relationship of cally abused children: Effects of ethnicity and parenting
developmental outcome to severity of injury. Pediatric context. Child Abuse and Neglect, 32, 567–576. https://d
Neurosurgery, 31, 251–258. https://doi.org/10.1159/ oi.org/10.1016/j.chiabu.2007.06.009
000028872 Hedges, L. V., & Vevea, J. L. (1998). Fixed-and random-
*Finzi, R., Har-Even, D., Shnit, D., & Weizman, A. (2002). effects models in meta-analysis. Psychological Methods,
Psychosocial characterization of physically abused chil- 3, 486–504. https://doi.org/10.1037/1082-989X.3.4.486
dren from low socioeconomic households in compar- *Hennessy, K. D., Rabideau, G. J., Cicchetti, D., & Cum-
ison to neglected and nonmaltreated children. Journal of mings, E. M. (1994). Responses of physically abused
Child and Family Studies, 11, 441–453. https://doi.org/ and nonabused children to different forms of interadult
10.1023/A:1020983308496 anger. Child Development, 65, 815–828. https://doi.org/
*Finzi, R., Har-Even, D., & Weizman, A. (2003). Compar- 10.2307/1131420
ison of ego defenses among physically abused children, *Herrenkohl, T. I., Klika, J. B., Herrenkohl, R. C., Russo,
neglected, and non-maltreated children. Comprehensive M. J., & Dee, T. (2012). A prospective investigation of
Psychiatry, 44, 388–395. https://doi.org/10.1016/S0010- the relationship between child maltreatment and indi-
440X(03)00106-8 cators of adult psychological well-being. Violence and
*Ghirmay, M.( 2015). Three potential mediators of the Victims, 27, 764776. https://doi.org/10.1891/0886-6708.
effects of child abuse and neglect on smoking in middle 27.5.764
adulthood. Doctoral dissertation, The City University of Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman,
New York. D. G. (2003). Measuring inconsistency in meta-analyses.
*Gordis, E. B., Feres, N., Olezeski, C. L., Rabkin, A. N., & British Medical Journal, 327, 557–560. https://doi.org/
Trickett, P. K. (2009). Skin conductance reactivity and 10.1136/bmj.327.7414.557
respiratory sinus arrhythmia among maltreated and *Hjorth, C. W., & Ostrov, E. (1982). The self-image of
comparison youth: Relations with aggressive behavior. physically abused adolescents. Journal of Youth and Ado-
Journal of Pediatric Psychology, 35, 547–558. lescence, 11, 71–76. https://doi.org/10.1007/BF01834704
Gottman, J. M., Katz, L. F., & Hooven, C. (1996). Parental Hobbs, C. J., & Wynne, J. M. (2002). Neglect of neglect.
meta-emotion philosophy and the emotional life of Current Paediatrics, 12, 144–150. https://doi.org/10.
families: Theoretical models and preliminary data. Jour- 1054/cupe.2001.0266
nal of Family Psychology, 10, 243–268. https://doi.org/ *Houck, G. M., & King, M. C. (1993). Cognitive function-
10.1037/0893-3200.10.3.243 ing, and behavioral and emotional adjustment in mal-
Gross, J. J. (1998). The emerging field of emotion regula- treated children post-intervention. Journal of Child and
tion: An integrative review. Review of General Psychol- Adolescent Psychiatric Nursing, 6, 5–17. https://doi.org/
ogy, 2, 271–299. https://doi.org/10.1037/1089-2680.2.3. 10.1111/j.1744-6171.1993.tb00154.x
271 *Jaffee, S. R., Caspi, A., Moffitt, T. E., & Taylor, A. (2004).
Gross, J. J., & Feldman Barrett, L. (2011). Emotion genera- Physical maltreatment victim to antisocial child: Evi-
tion and emotion regulation: One or two depends on dence of an environmentally mediated process. Journal
your point of view. Emotion Review, 3, 8–16. https://d of Abnormal Psychology, 113, 44–55. https://doi.org/10.
oi.org/10.1177/1754073910380974 1037/0021-843X.113.1.44
Gross, J. J., & Jazaieri, H. (2014). Emotion, emotion regu- Johnson, B. T., & Eagly, A. H. (2000). Quantitative synthe-
lation, and psychopathology: An affective perspective. sis of social psychological research. CHIP Documents.
Psychological Science, 2, 387–410. https://doi.org/10. Retrieved from http://digitalcommons.uconn.edu/
1177/2167702614536164 chip_docs/13
Gross, J. J., & Thompson, R. A. (2007). Emotion regula- *Keil, V., & Price, J. M. (2009). Social information-process-
tion: Conceptual foundations. In J. J. Gross (Ed.), Hand- ing patterns of maltreated children in two social
book of emotion regulation (pp. 3–24). New York, NY: domains. Journal of Applied Developmental Psychology, 30,
Guilford. 43–52. https://doi.org/10.1016/j.appdev.2008.10.003
20 Lavi, Katz, Ozer, and Gross

Kempe, C. H., Silverman, F. N., Steele, B. F., Droege- correlates, parenting contexts, and pathways to disor-
mueller, W., & Silver, H. K. (1962). The battered der. In J. Cassidy & P. R. Shaver (Eds.), Handbook of
child syndrome. Journal of the American Medical Associa- attachment: Theory, research, and clinical applications (Vol.
tion, 181, 17–24. https://doi.org/10.1001/jama.1962. 3, pp. 667–695). New York, NY: Guilford.
03050270019004 *Ma, E. Y. M., & Li, F. W. S. (2014). Developmental
*Kim, J., & Cicchetti, D. (2010). Longitudinal pathways trauma and its correlates: A study of Chinese children
linking child maltreatment, emotion regulation, peer with repeated familial physical and sexual abuse in
relations, and psychopathology. Journal of Child Psychol- Hong Kong. Journal of Traumatic Stress, 27, 454–460.
ogy and Psychiatry, 51, 706–716. https://doi.org/10. https://doi.org/10.1002/jts.21944
1111/j.1469-7610.2009.02202.x *Main, M., & George, C. (1985). Responses of abused and
*Kinard, E. (1980). Emotional development in physically disadvantaged toddlers to distress in agemates: A
abused children. American Journal of Orthopsychiatry, 50, study in the day care setting. Developmental Psychology,
686–696. https://doi.org/10.1111/j.1939-0025.1980.tb 21, 407–412. https://doi.org/10.1037/0012-1649.21.3.
03332.x 407
*Klorman, R., Cicchetti, D., Thatcher, J. E., & Ison, J. R. Main, M., & Hesse, E. (1990). Parents’ unresolved trau-
(2003). Acoustic startle in maltreated children. Journal of matic experiences are related to infant disorganized
Abnormal Child Psychology, 31, 359–370. https://doi. attachment status: Is frightened and/or frightening par-
org/10.1023/A:1023835417070 ental behavior the linking mechanism? In M. T. Green-
*Koenig, A. L., Cicchetti, D., & Rogosch, F. A. (2000). berg, D. Cicchetti, & E. M. Cummings (Eds.),
Child compliance/noncompliance and maternal con- Attachment in the preschool years: Theory, research, and
tributors to internalization in maltreating and nonmal- intervention (pp. 161–182). Chicago, IL: University of
treating dyads. Child Development, 71, 1018–1032. Chicago.
https://doi.org/10.1111/1467-8624.00206 Main, M., Kaplan, N., & Cassidy, J.( 1985). Security in
*Lansford, J. E., Dodge, K. A., Pettit, G. S., Bates, J. E., infancy, childhood, and adulthood: A move to the level
Crozier, J., & Kaplow, J. (2002). Long-term effects of of representation. Monographs of the Society for Research
early child physical maltreatment on psychological, in Child Development, 50(1-2), 66 – 104. https://doi.org/
behavioral, and academic problems in adolescence: A 10.2307/3333827
12-year prospective study. Archives of Pediatrics and *Manly, J. T., Kim, J. E., Rogosch, F. A., & Cicchetti, D.
Adolescent Medicine, 156, 824–830. https://doi.org/10. (2001). Dimensions of child maltreatment and chil-
1001/archpedi.156.8.824 dren's adjustment: Contributions of developmental
Lau, J., Ioannidis, J. P. A., Terrin, N., Schmid, C. H., & timing and subtype. Development and Psychopathology,
Olkin, I. (2006). Evidence based medicine: The case of 13, 759–782.
the misleading funnel plot. British Medical Journal, *Maughan, A., & Cicchetti, D. (2002). Impact of child mal-
333, 597–600. https://doi.org/10.1136/bmj.333.7568. treatment and interadult violence on children's emotion
597 regulation abilities and socioemotional adjustment.
Lavi, I., Manor-Binyamini, I., Seibert, E., Katz, L. F., Ozer, Child Development, 73, 1525–1542. https://doi.org/10.
E. J., & Gross, J. J. (2019). Broken bonds: A meta-analy- 1111/1467-8624.00488
sis of emotion reactivity and regulation in emotionally * McGinn, H. C.( 2014). The role of paternal emotion
maltreating parents. Child Abuse and Neglect, 88, 376– socialization in the development of children's emotion
388. https://doi.org/10.1016/j.chiabu.2018.11.016 regulation in the context of physical maltreatment. Doc-
Lavi, I., Ozer, E. J., Katz, L. F., & Gross, J. J. (in submis- toral disertation. Ontario Institute for Studies in Educa-
sion). The role of parental emotion reactivity and regu- tion of the University of Toronto.
lation in child maltreatment: A meta-analytic review. *McLaughlin, K. A., Peverill, M., Gold, A. L., Alves, S., &
Leeb, R. T., Paulozzzi, L., Melanson, C., Simon, T., & Sheridan, M. A. (2015). Child maltreatment and neural
Arias, I. (2008). Child maltreatment surveillance: Uniform systems underlying emotion regulation. Journal of the
definitions for public health and recommended data elements. American Academy of Child & Adolescent Psychiatry, 54,
Atlanta, GA: Centers for Disease Control and Preven- 753–762. https://doi.org/10.1016/j.jaac.2015.06.010
tion, National Center for Injury Prevention and Con- Mikulincer, M., & Shaver, P. R. (2007). Attachment in
trol. adulthood: Structure, dynamics, and change. New York,
Lipsey, M. W., & Wilson, D. B. (2001). Applied social NY: Guilford.
research methods series: Practical meta-analysis (Vol. 49). Mikulincer, M., Shaver, P. R., & Pereg, D. (2003). Attach-
Thousand Oaks, CA: Sage. ment theory and affect regulation: The dynamics,
Lunkenheimer, E. S., Shields, A. M., & Cortina, K. S. development, and cognitive consequences of attach-
(2007). Parental emotion coaching and dismissing in ment-related strategies. Motivation and Emotion, 27, 77–
family interaction. Social Development, 16, 232–248. 102. https://doi.org/10.1023/A:1024515519160
https://doi.org/10.1111/j.1467-9507.2007.00382.x * Milojevich, H. M.( 2016). The role of maltreatment in
Lyons-Ruth, K., & Jacobvitz, D. (2016). Attachment disor- the development of emotion regulation. Doctoral dis-
ganization from infancy to adulthood: Neurobiological sertation. University of California, Irvine, Irvine, CA.
Emotion in Maltreated Children 21

*Moretti, M. M., & Craig, S. G. (2013). Maternal versus Psychology and Psychiatry, 53, 1184–1193. https://doi.
paternal physical and emotional abuse, affect regulation org/10.1111/j.1469-7610.2012.02586.x
and risk for depression from adolescence to early adult- Shaver, P. R., & Mikulincer, M. (2014). Adult attachment
hood. Child Abuse and Neglect, 37, 4–13. https://doi. and emotion regulation. In J. J. Gross (Ed.), Handbook of
org/10.1016/j.chiabu.2012.09.015 emotion regulation (pp. 237–250). New York, NY: Gil-
Paolucci, E. O., Genuis, M. L., & Violato, C.( 2001). A ford.
meta-analysis of the published research on the effects *Shenk, C. E., Griffin, A. M., & O'Donnell, K. J. (2015).
of child sexual abuse. Journal of Psychology , 135 , 17–36. Symptoms of major depressive disorder subsequent to
https://doi.org/10.1080/00223980109603677 child maltreatment: Examining change across multiple
Parke, R. D., Cassidy, J., Burks, V. M., Carson, J. L., & levels of analysis to identify transdiagnostic risk path-
Boyum, L. (1992). Familial contribution to peer compe- ways. Development and Psychopathology, 27, 1503–1514.
tence among young children: The role of interactive https://doi.org/10.1017/S0954579415000905
and affective processes. In R. D. Parke & G. W. Ladd *Shenk, C. E., Noll, J. G., & Cassarly, J. A. (2010). A mul-
(Eds.), Family-peer relationships: Modes of linkage (pp. tiple mediational test of the relationship between child-
107–134). Hillsdale, NJ: Erlbaum. hood maltreatment and non-suicidal self-injury. Journal
Pollak, S. D., Messner, M., Kistler, D. J., & Cohn, J. F. of Youth and Adolescence, 39, 335–342. https://doi.org/
(2009). Development of perceptual expertise in emotion 10.1007/s10964-009-9456-2
recognition. Cognition, 110, 242–247. https://doi.org/ *Shenk, C. E., Putnam, F. W., Rausch, J. R., Peugh, J. L.,
10.1016/j.cognition.2008.10.010 & Noll, J. G. (2014). A longitudinal study of several
*Reidy, T. J. (1977). The aggressive characteristics of potential mediators of the relationship between child
abused and neglected children. Journal of Clinical Psychol- maltreatment and posttraumatic stress disorder symp-
ogy, 33, 1140–1145. https://doi.org/10.1002/1097-4679 toms. Development and Psychopathology, 26, 81–91.
(197710)33:4<1140:AID-JCLP2270330449>3.0.CO;2-G https://doi.org/10.1017/S0954579413000916
Repetti, R. L., Taylor, S. E., & Seeman, T. E. (2002). Risky Shields, A. M., & Cicchetti, D. (1995). The development of
families: Family social environments and the mental an emotion regulation assessment battery: Reliability
and physical health of offspring. Psychological Bulletin, and validity among at-risk grade-school children. Paper
128, 330–366. https://doi.org/10.1037/0033-2909.128.2. presented at the Society for Research in Child Develop-
330 ment, Indianapolis, IN.
*Robinson, L. R., Morris, A. S., Heller, S. S., Scheeringa, Shields, A. M., & Cicchetti, D. (1997). Emotion regulation
M. S., Boris, N. W., & Smyke, A. T. (2009). Relations among school-age children: The development and vali-
between emotion regulation, parenting, and psy- dation of a new criterion Q-sort scale. Developmental
chopathology in young maltreated children in out of Psychology, 33, 906–916. https://doi.org/10.1037/0012-
home care. Journal of Child and Family Studies, 18, 421– 1649.33.6.906
434. https://doi.org/10.1007/s10826-008-9246-6 *Shields, A. M., & Cicchetti, D. (1998). Reactive aggres-
Rosenthal, R. (1994). Parametric measures of effect size. sion among maltreated children: The contributions of
In H. Cooper & L. V. Hedges (Eds.), The handbook of attention and emotion dysregulation. Journal of Clinical
research synthesis (pp. 231–244). New York, NY: Russell Child Psychology, 27, 381–395. https://doi.org/10.1207/
Sage Foundation. s15374424jccp2704_2
Schatz, J. N., & Burke-Lefever, J. (2003). Understanding *Shields, A. M., & Cicchetti, D. (2001). Parental maltreat-
maltreatment: The interplay between mothers’ parent- ment and emotion dysregulation as risk factors for bul-
ing orientations and children's vulnerability. Paper pre- lying and victimization in middle childhood. Journal of
sented at the Gatlinburg Conference on Research and Clinical Child Psychology, 30, 349–363. https://doi.org/
Theory in Intellectual and Developmental Disabilities, 10.1207/s15374424jccp1801_9
Annapolis, MD. * Shipman, K. L.( 1990). The socialization of affect man-
*Schatz, J. N., Smith, L. E., Borkowski, J. G., Whitman, T. agement skills: A developmental psychopathology per-
L., & Keogh, D. A. (2008). Maltreatment risk, self-regu- spective. Doctoral dissertation. University of Maine.
lation, and maladjustment in at-risk children. Child *Shipman, K. L., Edwards, A., Brown, A., Swisher, L., &
Abuse and Neglect, 32, 972–982. https://doi.org/10. Jennings, E. (2005). Managing emotion in a maltreating
1016/j.chiabu.2008.09.001 context: A pilot study examining child neglect. Child
*Schofield, T. J., Conger, R. D., & Conger, K. J. (2017). Abuse and Neglect, 29, 1015–1029. https://doi.org/10.
Disrupting intergenerational continuity in harsh parent- 1016/j.chiabu.2005.01.006
ing: Self-control and a supportive partner. Development *Shipman, K. L., & Zeman, J. (2001). Socialization of chil-
and Psychopathology, 29, 1279–1287. https://doi.org/10. dren's emotion regulation in mother–child dyads: A
1017/S0954579416001309 developmental psychopathology perspective. Develop-
Scott, S., & O'Connor, T. G. (2012). An experimental test ment and Psychopathology, 13, 317–336. https://doi.org/
of differential susceptibility to parenting among emo- 10.1017/S0954579401002073
tionally-dysregulated children in a randomized con- *Skowron, E. A., Cipriano-Essel, E., Gatzke-Kopp, L. M.,
trolled trial for oppositional behavior. Journal of Child Teti, D. M., & Ammerman, R. T. (2014). Early
22 Lavi, Katz, Ozer, and Gross

adversity, RSA, and inhibitory control: Evidence of chil- * Veloz, C. Y.( 2005). Play behaviors and social skills
dren's neurobiological sensitivity to social context. among maltreated foster children. Doctoral dissertation.
Developmental Psychobiology, 56, 964–978. https://doi. Fordham University, New York. NY.
org/10.1002/dev.21175 * Vorderstrasse, V. M.( 2013). Hostile attribution biases
*Smith, M., & Walden, T. (1999). Understanding feelings and problematic peer relationships in physically abused
and coping with emotional situations: A comparison of school-age children. Doctoral dissertation. Department
maltreated and nonmaltreated preschoolers. Social of Psychology, Roosevelt University.
Development, 8, 93–116. https://doi.org/10.1111/1467- Waters, H. S., Rodrigues, L. M., & Ridgeway, D. (1998).
9507.00082 Cognitive underpinnings of narrative attachment
StataCorp. (2017). Stata statistical software: Release 15. Col- assessment. Journal of Experimental Child Psychology, 71,
lege Station, TX: StataCorp LLC. 211–234. https://doi.org/10.1006/jecp.1998.2473
Stith, S. M., Liu, T., Davies, L. C., Boykin, E. L., Alder, M. *Young, J. C., & Widom, C. S. (2014). Long-term effects of
C., Harris, J. M., & Dees, J. (2009). Risk factors in child child abuse and neglect on emotion processing in adult-
maltreatment: A meta-analytic review of the literature. hood. Child Abuse and Neglect, 38, 1369–1381. https://d
Aggression and Violent Behavior, 14, 13–29. https://doi. oi.org/10.1016/j.chiabu.2014.03.008
org/10.1016/j.avb.2006.03.006
Stoddart, R. M. (1993). Social cognitive skills, social
adjustment and perceived competence in maltreated Supporting Information
preschoolers. Paper presented at the Biennial Meeting Additional supporting information may be found in
of the Society for Research in Child Development, New
the online version of this article at the publisher’s
Orleans, LA.
website:
Straus, M. A. (1979). Measuring intrafamily conflict and
violence: The conflict tactics (CT) scales. Journal of Mar- Appendix S1. Summary of Theories of Emotion
riage and the Family, 41, 75–88. https://doi.org/10. Regulation and Maltreatment
2307/351733 Appendix S2. Search Phrases, Details Regarding
Straus, M. A., Hamby, S. L., & Warren, W. L. (2003). The Search and Criteria, and Variables Coded From
conflict tactics scales handbook. Los Angeles, CA: Western Each Study
Psychological Services. Appendix S3. Emotion Reactivity and Emotion
*Teisl, M., & Cicchetti, D. (2008). Physical abuse, cognitive Regulation Scales
and emotional processes, and aggressive/disruptive Appendix S4. Sensitivity Analysis
behavior problems. Social Development, 17, 1–23. Appendix S5. Flow Diagram
https://doi.org/10.1111/j.1467-9507.2007.00412.x
Appendix S6. Descriptive Data of Studies
Van Ijzendoorn, M. H., Schuengel, C., & Bakermans-
Included
Kranenburg, M. J. (1999). Disorganized attachment in
early childhood: Meta-analysis of precursors, Appendix S7. Forest Plots of Meta-Analysis
concomitants, and sequelae. Development and Psy- Results
chopathology, 11, 225–250. https://doi.org/10.1017/ Appendix S8. Funnel Plot of Effect Sizes of
S0954579499002035 Meta-Analysis

You might also like