Maternal Incarceration, Children’s Psychological Adjustment, and the Mediating Role of Emotion Regulation

Abstract

Children who live in the context of maternal incarceration (MI) are exposed to both general environmental risk and incarceration-specific risk increasing the probability of their developing externalizing and internalizing behaviors problems. Little research has examined the socio-emotional mechanisms that account for the psychological effects of MI. This research examined children’s anger and sadness regulation as mediators between environmental and incarceration-specific risk and psychological functioning. Participants were 117 children (60% Black; 52% boys; M age = 9.85 years, SD = 1.65 years), their incarcerated mother, and current caregiver. All informants completed questionnaires assessing children’s anger and sadness regulation as well as externalizing and internalizing behaviors. Mothers and caregivers provided information concerning children’s exposure to environmental risk and all three reporters provided information on incarceration-specific risk experiences (ISRE). Structural equation modeling was used to test indirect effects of risk variables (ISRE, environmental) on psychological functioning (externalizing, internalizing behaviors) via emotion regulation (anger, sadness). Gender, age, and race were covariates. The analyses revealed significant indirect effects of incarceration-specific risk on both externalizing and internalizing behavior problems via anger regulation but not via sadness regulation. The findings highlight the centrality of emotion regulation as a mechanism that helps explain the negative psychological outcomes experienced by children exposed to ISRE with implications for preventive interventions.

The number of individuals incarcerated in the United States over the last 30 years has grown from just under 2 million in 1980 to nearly 7 million in 2013 (Bureau of Justice Statistics 2016). Approximately half the inmates are parents; 60% of the incarcerated parents are mothers (Glaze and Maruschak 2008). Children and adolescents with incarcerated parents are vulnerable to a diverse array of maladaptive outcomes including aggressive and antisocial behavior (Murray et al. 2012), attachment insecurity (Poehlmann 2005), depression (Wilbur et al. 2007), diminished educational attainment (Haskins 2014), and physical health issues (Gjelsvik et al. 2013). Consequently, parental incarceration has been included as one of the key risk factors in a model that identifies adverse child experiences (ACEs) that disrupt physical and emotional development (Anda et al. 2001; Felitti et al. 1998). Most research on parental incarceration has investigated the impact of paternal incarceration, yet children whose mothers are incarcerated may experience even greater disruption in social, emotional and academic competencies (Dallaire 2007; Dallaire and Wilson 2010).

Despite accumulating evidence of the psychological and physical effects of parental incarceration on youth, little research has examined the putative socio-emotional mechanisms that account for the effects of parental incarceration, particularly maternal incarceration (MI), on children’s functioning in multiple domains (e.g., social, academic, physical). One such factor may be emotion-related processes including emotion regulation (ER) because ER skills are integrally involved in social, academic, cognitive, and health functioning (Sroufe 2005). Given ER’s central role that infuses many aspects of development, examination of these skills can help illuminate how ER functions within a high risk population of children; most research has examined ER using low risk samples (Lemerise 2016). Further, the findings of this research have the potential to provide guidance for prevention and intervention efforts for the approximately 1 million children who have an incarcerated parent (Glaze and Maruschak 2008). Thus, the overarching goal of this research was to examine the potential mediational role of anger and sadness regulation in children ages 7 to 12 years who were exposed to both incarceration-specific risk experiences related to MI and general environmental risk. We first provide information about the context to which children of MI are exposed, illuminating differences between general environmental risk versus risk experiences specific to MI. Building on this foundation, we then discuss the importance of ER in children’s functioning highlighting the gaps in the literature that this study addresses.

The Context of Maternal Incarceration

Although females comprise less than 10% of U.S. inmates, the rate of growth in the female jail and prison populations is 1.5 times higher compared to males. Furthermore, according to the Bureau of Justice Statistics, from 1980 to 2010 the rate of female incarceration grew by 646% (Guerino et al. 2011). Children with incarcerated mothers often experience greater disruption than children with incarcerated fathers because mothers are more likely to have been the primary caregiver prior to incarceration with most living with their child during the month prior to arrest and/or incarceration (Glaze and Maruschak 2008). Consequently, youth are more likely to have been exposed to stressful events related to their mother’s imprisonment, including being exposed to her criminal activity and witnessing her arrest and sentencing (Dallaire and Wilson 2010). Further, children of incarcerated mothers are more likely to be separated from both biological parents due to incarceration (Dallaire 2007; Dallaire and Wilson 2010). Adult children of incarcerated mothers are 2.5 times more likely to be incarcerated than adult children with incarcerated fathers (Dallaire 2007), and three times as likely to be incarcerated as adults compared to children whose mothers have never been incarcerated (Huebner and Gustafson 2007). A meta-analytic review of research examining the impact of parental incarceration on youth functioning found in comparison with peers, youth with incarcerated parents faced a 10% elevated risk for aggressive and antisocial outcomes (Murray et al. 2012). The review, however, did not account for differences in children’s experience of maternal and paternal incarceration.

The mechanisms of risk remain unclear because it is difficult to disentangle the impact of incarceration from the impact of other stressful negative events (Murray and Farrington 2008). The negative effects of parental incarceration on children may be due, in part, to the fact that children are exposed to general environmental risks (e.g., parental substance abuse, mental illness). Other contextual risk factors, such as poverty, substance abuse, domestic violence, family instability, and parenting stress in conjunction with the incremental stress of parental incarceration, may help explain why children with incarcerated parents are at heightened risk for psychopathology and antisocial outcomes (e.g., Mackintosh et al. 2006; Murray et al. 2012; Western and Wildeman 2009).

However, MI may confer specific risks for children and families that predict unique negative outcomes while simultaneously examining the effects of general environmental risk. Dallaire et al. (2015a) developed and validated an Index of Incarceration-specific Risk Experiences (ISRE) that delineates specific deleterious events facing children when experiencing MI that place them at risk for maladaptive psychological functioning (e.g., child separation from siblings because of MI, child living with a new caregiver because of MI). These ISREs differentiated children who responded more adaptively (i.e., less psychopathology) to their mother’s incarceration from those who showed high levels of externalizing and internalizing behaviors. Importantly, ISREs predicted children’s level of these behavior problems more strongly than their exposure to general environmental risks.

Incarceration-specific risk experiences may have a more profound impact on children’s well-being because many of them are traumatic experiences that undermine children’s ability to cope effectively with more general challenges and stresses. A mother’s incarceration can fundamentally disrupt the family system and weaken a child’s social support network, particularly if the child’s father is also incarcerated or not involved in the child’s life (Dallaire 2007). Many children with incarcerated mothers may also experience separation from siblings, friends, and other caring adults (e.g., teachers) as a result of changing residences due to their mother’s incarceration. There are often few social supports available to children and families suffering grief because of separation from and stigma due to MI, thus children may have little help processing events they may have witnessed (e.g., mother’s arrest; Arditti 2012). Feelings of sadness, loss, and anger may result in these children experiencing an elevated risk for externalizing and internalizing behaviors (Murray et al. 2012).

Anger and Sadness Regulation

In recent years there has been considerable attention to psychological outcomes, particularly externalizing behavior problems (e.g., aggression) for children of incarcerated parents (Murray et al. 2012), but very few researchers have studied emotion-related processes in this population as a potential intervening variable. Emotion regulation “consists of the extrinsic and intrinsic processes responsible for monitoring, evaluating, and modifying emotional reactions, especially their intensive and temporal features, to accomplish one’s goals” (Thompson 1994; pp. 27–28). Importantly, regulation is not synonymous with control but rather encompasses multiple ways of responding to emotional arousal that are sensitive and responsive to one’s goals and the demands of the social context. Emotion regulation is considered to be a foundational skill in emotion development (Saarni 1999) and it is integrally involved in social (e.g., Denham et al. 2007), academic (e.g., Trentacosta and Izard 2007), cognitive (e.g., Blair 2002), psychological (e.g., Zeman et al. 2002), and physical health functioning (e.g., Whitson and El Sheikh 2003).

Although the development of ER skills begins in early in development, infants’ and preschool age children’s emotion management efforts are greatly facilitated by their caregivers (Evans and Porter 2009). With increasing cognitive, social, and biological development, children gain emotional competence skills (e.g., emotion awareness and labeling; Saarni 1999) that prepare them to more effectively and independently regulate their emotions. By roughly fifth grade, children typically have become adept at managing emotional expressions to meet the demands of their social environment (Cole 2014; Saarni 1999; Zeman and Garber 1996). Studying ER in middle childhood is particularly important because it is during this stage that children hone and assimilate these skills into their various social contexts (Bariola et al. 2011). Research indicates that children who do not learn adaptive ER skills are at risk for experiencing internalizing and/or externalizing behavior problems (Kring and Sloan 2010; Zeman et al. 2002). In fact, emotion dysregulation is considered a transdiagnostic process that underlies most aspects of psychopathology (e.g., Bradley 2003; Keenan 2000; Kring and Sloan 2010). However, the majority of research on children’s ER has been conducted using samples of children in low risk environments (Lemerise 2016). It is unclear whether the role of ER in children’s functioning also applies to children developing in atypical environmental contexts, such as MI. Emotion regulation is a critical competency to examine in children of MI because they are exposed to general risk and incarceration-specific experiences that elicit strong and frequent arousal of negative affect (Poehlmann-Tynan et al. 2015). These experiences may tax children’s ability to effectively cope with the emotions elicited by these stressors that then have implications for psychological maladjustment.

The study of emotional processes benefits from examining discrete emotions (e.g., anger, sadness) rather than global negative affect as each emotion is associated with unique appraisals, goals, and action tendencies that then elicit different responses from the social context (Campos et al. 1989). For example, anger may be evoked when a child perceives that a decision was unfair, with the resultant action tendency of assigning blame that leads to conflict with the perceived perpetrator (Kuppens et al. 2003). Within the context of MI, a caregiver may limit the child’s contact and visitation with the incarcerated mother, the child may view this as unfair, and may react with anger to the caregiver. Illustrating the different relations found for anger versus sadness regulation, Sullivan et al. (2010)found that anger regulation in low-income Black children was inversely correlated with physical aggression and sadness regulation was inversely associated with relational aggression. Relations have also been documented between sadness dysregulation and depression (Chaplin 2006) as well as with externalizing behaviors (Folk et al. 2014; Walcott and Landau 2004). These findings suggest there are different pathways to externalizing and internalizing outcomes related to specific ER competencies. Thus, it is key to examine the role of anger and sadness regulation in children of MI and the links to externalizing and internalizing outcomes given the high incidence of these problem behaviors in this population (Murray et al. 2012).

Despite mounting evidence of the psychological and physical toll of MI on youth, virtually no research has examined putative emotional and psychological mechanisms that account for the effects of MI on maladaptation. Only two published studies have examined ER in children exposed to parental incarceration (Dallaire and Wilson 2010; Lotze et al. 2010). Using a sample of 32 children with one (57% mothers) or both parents incarcerated, Dallaire and Wilson (2010) found that children’s exposure to incarceration-related experiences was associated with poorer ER but not with psychological maladjustment. However, the correlations were in the expected direction for externalizing problems suggesting that there may have been inadequate power to detect significant differences. Lotze et al. (2010) examine whether ER predicted externalizing and internalizing problems and callous-unemotional traits in 50 children with MI enrolled at a summer camp. Child report of anger dysregulation and frustration predicted both externalizing and internalizing behaviors but not callous-unemotional traits. Low levels of ER predicted internalizing behaviors and callous-unemotional traits, whereas emotional dysregulation predicted externalizing behaviors. These studies are an important first step that document the linkages between ER and psychological outcomes in children exposed to parental incarceration. More in depth investigations using larger sample sizes that allow for more sophisticated data analyses are needed to investigate the role of children’s anger and sadness regulation skills in the context of MI.

The Current Study

The goal of this research was to examine the role of anger and sadness regulation in 7–12 year-old children who are exposed to varying degrees of ISREs associated with a current MI. This research builds on prior research (Dallaire et al. 2015a, b) by examining ER in the relation between ISREs and psychological functioning. Using a multi-informant approach, children and their current caregivers and incarcerated mothers provided their perceptions of the child’s anger and sadness regulation skills, the presence of externalizing and internalizing behavior problems, and the number of ISREs and general risk experiences. The adequate sample size and multiple reporters of the same information permitted the use of structural equation modeling (SEM) to examine the relations among ISREs and general environmental risk to psychological outcomes and the role of anger and sadness regulation as mediators. We hypothesized that anger and sadness regulation would mediate the relation between ISREs, general environmental risk, and children’s externalizing and internalizing behavior problems.

Method

Participants

Participants included 117 children and their incarcerated mother and current caregiver. Children (52% boys; M age = 9.85 years, SD = 1.65 years, range = 7.40–12.98 years) were ethnically diverse (60% African American, 31% Caucasian, 9% Other). Children’s mothers (M = 32.85 years, SD = 5.91 years, range = 24–50 years; 64% Black) had three biological children on average (range = 1–7). If mothers had multiple children in the eligible age range, one child was selected at random for inclusion in these analyses. Additional detailed information about participants can be found in Dallaire et al. (2015a, b). Many mothers reported low educational attainment, with 35% reporting not completing the 12th grade or the high school equivalency exam. Mothers were incarcerated for various reasons, which, according to their self-reports, included contempt of court (e.g., parole violations, 32%), property crimes (e.g., larceny, 27%), drug crimes (e.g., prescription fraud, distribution of illegal substances, 16.7%), and other less frequently reported reasons like fraud (e.g., identity theft, 12%) and violent crimes (e.g., armed robbery, 6%). The number of times a mother had been incarcerated ranged from 1 to 11 (M = 2.65, SD = 1.64), with 25% of mothers reporting that this was their first incarceration.

Caregivers included children’s grandparent (55%), father (18%), relative (e.g., aunt, sibling, 24%), and stepparent (3%). The majority (75%) of caregivers were female (M = 47.64 years, SD = 11.54 years, range: 19–70 years). Although 25% reported having not completed 12th grade, 22% had taken some college courses and 11% had graduated from college. Caregivers identified as primarily Black (60%), White (30%), and Other (10%).

Procedure

Ethics approval was obtained by the authors’ university’s protection of human subjects committee and cooperating jail facilities reviewed and approved the research protocol. Informed consent was obtained from all participants with children providing their assent. Eligible women incarcerated at one of six jail facilities were recruited to participate by project staff. Eligibility recruitments included having one child within the specified age range (7–12 years), maintenance of parental rights, and no documented history of abuse or neglect to the target child. Mothers participated by completing a privately conducted 1-h individual interview with a research assistant at the jail facility in which they answered questions about themselves and their children. Mothers provided consent to contact the child’s caregiver along with contact information.

In total, 236 mothers participated by completing an interview and providing information to contact and recruit the child’s caregiver and child to participate. Lower participation rates of families of the incarcerated mother were expected (Farrington et al. 2001), thus, we over-sampled incarcerated mothers to achieve our target sample of 150 children. Caregiver participation rates were similar as to what we have seen in previous research with this population (see Dallaire and Wilson 2010), with approximately 50% of caregivers participating. The primary reason for caregiver non-participation included being unable to contact the caregiver (e.g., phone number had been disconnected), some caregivers refused because they reported that the mother was no longer considered part of the family, and some expressed a general lack of interest in the project. We recruited and interviewed 118 (50%) caregivers of 151 children with incarcerated mothers. For the current analyses, we used the data from only one sibling per family to participate to avoid violation of the statistical assumption of independent sampling. Thus, we present data from 117 children and their current caregivers. The caregiver sample is reduced by one to 117 because one sibling pair was separated from each other during the incarceration period and had different caregivers.

After contact was made with a caregiver, interviews were scheduled at the caregiver’s homes (64%), local libraries (30%), other public locations (e.g., restaurant, 3%), and university lab facility (3%). After obtaining caregiver consent and child assent, caregivers and children participated in separate locations to ensure privacy. Interviews took approximately one hour. For their participations, caregivers received $50 and children chose a small toy and received $10.

Measures

Incarceration-Specific Risk

An Incarceration-specific Risk Index (ISRI) was created for children, mothers, and caregivers based on their report of events related specifically to incarceration (e.g., child witnessed mother’s arrest) and/or events that occurred as a result of the mother’s incarceration (e.g., child is with a new caregiver). The following events were included: lack of contact with the child, three or more maternal incarcerations during the child’s lifetime, separation from siblings because of mother’s incarceration, child changed schools because of mother’s incarceration, child witnessed mother’s arrest, child witnessed mother’s sentencing, and child’s biological father was also incarcerated. Mothers reported whether or not their biological mother had been incarcerated and caregivers reported whether they were a new caregiver to this child. Additional details about the ISRI are reported elsewhere (see Dallaire et al. 2015a, b for supplementary material). Summed across the nine variables, scores ranged from 0 to 6 for mothers and caregivers.

Children’s ISRI was calculated based on their responses to 10 items from a life events checklist. Children indicated whether any of the following events occurred in them in the previous six months by answering either “Yes” (1) or “No” (0): A close family member was arrested or in jail, changed schools because of a change in residence, no longer living with your mother, no longer living with one of his/her siblings, no longer living with your father, witnessed criminal activity in the home, witnessed mother’s arrest, witnessed father’s arrest, witnessed mother’s criminal sentencing, witnessed father’s criminal sentencing. The sum of children’s responses on this measure constituted their ISRI and was used in all subsequent analyses.

Environmental Risk

An Environmental Risk Index variable was created based on information provided by mothers and caregivers. We adopted a cumulative risk approach and based the construction of the Environmental Risk Index on past available research (see Aaron and Dallaire 2010; MacKenzie et al. 2011; Sameroff et al. 1993). The index included: (a) traditional child rearing beliefs, (b) parental hostility, (c) stressful life events, (d) maternal/caregiver anxiety, (e) maternal/caregiver psychopathology, (f) maternal/caregiver non-completion of high school (g) maternal/caregiver un/underemployment, (h) large family size, (i) no father figure in child’s household, and (j) ethnic minority status. Each of these variables were dichotomized to indicate if the risk was present or absent. Additional details about these items are presented in Dallaire et al. (2015a, b). Each of 10 different items were dichotomized to indicate if the risk was present or absent. For ease of presentation, psychometric and scoring information for five of the items are summarized in Table 1 and the other five items are described below.

Table 1.

Instrument documentation and scoring information for items used to create the environmental risk index

Instrument Name	Citation	Sample Item	Maternal Reliability	Caregiver Reliability	Scoring Information for Environmental Risk Index
Parental Modernity Scale of Child-rearing and Educational Beliefs: Traditional Beliefs subscale	Schaefer and Edgerton 1985	“The most important thing to teach children is absolute obedience to parents”	0.83	0.84	Mothers and caregivers scoring in top 25% received a score of 1.
Parenting Behavior Inventory: Hostile/Coercive subscale	Lovejoy et al. 1999	“I grab or handle my child roughly”	0.70	0.74	Mothers and caregivers scoring in top 25% received a score of 1.
Life Events Checklist	Work et al. 1990	“Sometimes your family had little food to eat.”	0.73	0.76	Mothers and caregivers scoring in top 25% received a score of 1.
Psychiatric Diagnostic Screening Questionnaire: Anxiety subscale	Zimmerman and Chelminski 2006	“Was it hard for you to control or stop worrying on most days?”	0.88	0.89	The mothers (44%) and caregivers (18%) scoring at or above the clinical cutoff for generalized anxiety received a score of 1.
Psychiatric Diagnostic Screening Questionnaire: Depression, post-traumatic stress, obsessive compulsions, panic attacks, psychosis, agoraphobia, social phobia, and somatization subscales	Zimmerman and Chelminski 2006	“Did reminders of an event make you to feel intense distress?”	Ranged from 0.25 (psychosis) - 0.93 (social phobia)	Ranged from 0.37 (psychosis) - 0.89 (post-traumatic stress; panic attacks)	The mothers (71%) and caregivers (31%) scoring at or above the clinical cutoff on 2 or more of the sub scales received a score of 1.

In addition to the items summarized in Table 1, five additional environmental risk items were assessed with a demographic and background interview in which participants provided information about their education, occupation, household composition, and ethnicity. Mothers (34%) and caregivers (25%) who did not complete 12th grade (or GED) received a score of 1 (education risk present). Mothers’ and caregivers’ occupations over the previous 12 months were coded based on the Hollingshead (1975) system. Mothers (79%) and caregivers (75%) who reported being unemployed/retired or working in menial (e.g., dishwasher) or unskilled (e.g., garbage worker) positions received a score of 1 (occupation risk present). Mothers (29%) and caregivers (30%) who reported that four or more individuals under the age of 18 resided in the household received a score of 1 (large family size risk present). Mothers (67%) and caregivers (82%) who indicated that no father figure to the child (grandfather, step-father, biological father) lived in the household received a score of 1 (lack of father figure risk present). Lastly, mothers (64%) and caregivers (70%) who identified their race as nonWhite received a score of 1 (i.e., minority status risk present).

Children’s Externalizing and Internalizing Behavior Problems

Mothers and caregivers completed the 113-item Child Behavior Checklist (CBCL; Achenbach and Rescorla 2001). The CBCL provides three broad-band scales and eight syndrome scales. Only the total raw scores from the externalizing and internalizing broad-band scales were used in the current study. The CBCL has demonstrated strong internal consistency with adequate content, criterion-related, and construct validity (Achenbach and Rescorla 2001). Reliabilities in the current study for Externalizing and Internalizing problems were strong for mothers/caregivers (Externalizing: α = 0.75/0.79, Internalizing: α = 0.87/0.87). According to mothers’/caregivers’ reports, respectively, 34.3/38.6% of children were in the borderline or clinical range for Externalizing problems whereas 25.9/28.9% of children were in the borderline or clinical range for Internalizing problems. For mother-report of internalizing problems, one outlier was far greater than all other values in the data set. This outlier was decreased so that it was equal to the next highest value plus one.

Children completed the Children’s Depression Inventory (CDI; Kovacs 1992). Due to IRB concerns, the suicide item was omitted. The psychometric properties of the CDI have been well established and are acceptable (Carey et al. 1987; Kovacs 1985). Within this sample, 6 children (4%) met the clinical cutoff and 24 children (21%) receiving total raw scores of 13 or higher, indicating mild to moderate levels of depression (Smucker et al. 1986). Internal consistency was strong (α = 0.86).

Children’s self-report of externalizing behavior was assessed with the 19-item Things You Do subscale from the Risky Behavior Protocol (RBP, Conger and Elder 1994). The questions assess major risk-taking and delinquency (e.g., purposely set fire in a building or in any other space), minor risk-taking (e.g., ridden in a car without a seat belt), and other risk-taking behavior falling between minor and major risk (e.g., smoked cigarettes or used tobacco). Children indicate on a 3-point scale the frequency of the behavior. One outlier was far greater than all other values in the data set. This outlier was decreased so that it was equal to the next highest value plus one. The RPB has acceptable reliability and validity (NICHD SECCYD 2007). In our sample, internal consistency was strong (α = 0.80).

Children’s Emotion Regulation

Mothers, caregivers, and children completed the parent and child versions of Children’s Emotion Management Scales (CEMS – Parent; Cassano et al. 2007; CEMS; Zeman, Shipman, & Penza-Clyve, 2001) for Anger (CAMS) and Sadness (CSMS). The 11-item CAMS, and the 12-item CSMS are designed to assess strategies for regulating anger and sadness using a 3-point Likert scale (1 = a little, 3 = often). The 4-item Anger and 5-item Sadness Regulation Coping scales were used that assess strategies for the management of emotional arousal (e.g., mother/caregiver report: “When my child is feeling sad, he/she does something totally different until he/she calms down”; child report: “When I am feeling sad, I do something totally different until I calm down”). Previous studies have reported construct and criterion validity and adequate internal consistency for the CEMS-P/T for sadness management in a sample of 6- to 11-year-old children (Cassano et al. 2007) and for the child report version (e.g., Wills et al. 2006). The internal consistencies were acceptable for mother- and caregiver- report but weaker for child-report (anger: α = 0.83/83/0.51; sadness: α = 0.65/0.67/0.50).

Results

Preliminary Analyses

Descriptive data and correlations among study variables are presented in Table 2 (raw, untransformed data). There were few differences between mothers whose children and caregivers did and did not participate. Mothers whose children participated reported significantly higher levels of environmental risk (M = 4.50 experiences, SD = 2.05) than mothers of nonparticipating children (M = 3.97 experiences, SD = 1.66), t(222.79) = −2.18, p = 0.03; there were no significant differences in maternal-reported ISREs or maternal report of children’s externalizing and internalizing behaviors. Regarding children who were and were not included in the sample (based on random selection of one child from each household), there were also minimal differences. Children who were included in the sample had higher sadness regulation scores based on caregiver report, t(142) = −2.05, p = 0.04, had experienced fewer ISREs based on caregiver report, t(64.27) = −3.38, p = 0.001, and mother report, t(58.48) = −3.57, p = 0.001, and experienced fewer environmental risk experiences based on caregiver report, t(152) = 2.41, p = 0.02.

Table 2.

Descriptive and correlational data

Variable (reporter)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	M (SD)
1. Child age (in months)	-	−0.05	0.05	0.07	0.13	0.03	0.24**	0.28**	0.03	0.11	−0.21*	0.19*	0.07	0.08	0.02	0.10	0.22*	0.11	0.20*	−0.09	118.17 (19.78)
2. Child gendera		-	−0.01	−0.14	0.23*	−0.03	−0.03	−0.10	0.17t	−0.01	0.15	0.16t	−0.08	−0.04	−0.12	−0.08	−0.11	0.06	0.06	0.08	0.48 (0.50)
3. Child Raceb			-	0.13	−0.25*	−0.01	0.50**	0.26**	0.25**	0.28**	0.09	0.03	−0.08	0.01	−0.21*	−0.19*	−0.03	−0.09	−0.12	−0.06	0.57 (0.50)
4. ISRE (mother)				-	0.50**	0.06	0.00	−0.00	−0.20t	0.03	−0.06	−0.16	0.16	0.07	0.17	0.11	0.15	0.03	0.15	−0.02	2.42 (1.36)
5. ISRE (caregiver)					-	0.24*	−0.21t	0.00	−0.15	−0.17	−0.43**	−0.21t	0.07	−0.01	0.19t	0.31**	0.15	0.15	0.24*	0.00	2.92 (1.44)
6. ISRE (child)						-	.06	0.03	−0.04	−0.11	−0.18t	0.05	0.03	−0.03	0.09	0.21*	0.30**	0.09	0.15	0.36**	2.60 (1.71)
7. Environmental Risk (mother)							-	0.40**	0.02	0.13	−0.05	0.07	−0.07	−0.08	0.08	−0.02	0.18	0.24*	0.03	0.03	4.50 (2.05)
8. Environmental Risk (caregiver)								-	−0.02	−0.13	−0.22*	0.08	−0.08	−0.14	0.08	0.43**	0.21*	0.07	0.48**	−0.01	3.90 (1.83)
9. Anger regulation (mother)									-	0.29**	0.24*	0.59**	0.04	0.13	−0.65**	−0.36**	−0.15	−0.44**	−0.12	−0.29**	8.02 (2.10)
10. Anger regulation (caregiver)										-	0.35**	0.14	0.49**	0.15	−0.34**	−0.50**	−0.04	−0.20*	−0.33**	−0.34**	7.84 (2.18)
11. Anger regulation (child)											-	0.17*	0.10	0.24*	−0.31**	−0.34**	−0.32**	−0.10	−0.23*	−0.21*	8.50 (1.76)
12. Sadness regulation (mother)												-	0.13	0.22*	−0.29**	−0.15	0.09	−0.32**	−0.02	−0.25*	9.71 (2.11)
13. Sadness regulation (caregiver)													-	0.19t	−0.09	−0.17t	−0.02	−0.10t	−0.20*	−0.24*	10.19 (2.29)
14. Sadness regulation (child)														-	−0.17t	−0.13	−0.19*	−0.12	−0.15	−0.36**	10.80 (2.15)
15. Externalizing Behavior (mother)															-	0.56**	0.33**	0.64**	0.24*	0.30**	10.60 (10.02)
16. Externalizing Behavior (caregiver)																-	0.34**	0.37**	0.59**	0.28**	11.75 (10.11)
17. Risky Behavior Protocol (child)																	-	0.19	0.28**	0.23*	4.28 (3.89)
18. Internalizing Behavior (mother)																		-	0.40**	0.26**	7.37 (6.43)
19. Internalizing Behavior (caregiver)																			-	0.18t	7.70 (7.38)
20. CDI (child)																				-	8.06 (7.33)

ISRE Incarceration-specific risk experiences, CDI Children’s Depression Inventory

^aChild gender was coded as 1 = female, 0 = male

^bChild Race was coded as 1 = Black, 0 = Not Black

^tp < 0.10

^*p < 0.05

^**p < 0.01

Data Preparation and Plan of Analysis

In order to remove skew, all three indicators of Externalizing problems and all three indicators of Internalizing problems were transformed using a square root transformation. In addition, child age and child report of externalizing problems were rescaled in order to bring all variances in the 1–10 range (Muthén and Muthén 2012).

Analyses were conducted using Mplus 7.3 (Muthén and Muthén 2012). Latent constructs were identified using the marker method (Kline 2005). We used Bayesian estimation with uninformative priors (Asparouhov and Muthén 2010). Bayesian estimation has been recommended when analyzing relatively small samples (Asparouhov and Muthén 2010) such as the sample used in this research. In addition, Bayesian estimation allows unbiased testing of indirect effects with asymmetric credible intervals, similar to boostrapping methods (Yuan and MacKinnon 2009). The Markov Chain Monte Carlo (MCMC) algorithm was used to generate posterior distributions. Two MCMC chains were used, and the second half of each was retained. Potential Scale Reduction was used as the MCMC convergence criterion. All analyses were repeated with the minimum number of iterations set at four times the number of iterations from the initial analysis (Muthén and Asparouhov 2012). We report the unstandardized estimate and the 95% credible interval (CI) in brackets following the estimate (standardized estimates are available upon request). An estimate is considered significant if zero falls outside of the CI. Significant estimates are indicated in bold in the Tables. Model fit is estimated by the Posterior Predictive P-Value (PPP). Similar to the chi-square p value in maximum likelihood estimation, PPP values greater than 0.05 indicate good model fit (Asparouhov and Muthén 2010).

In the structural models indirect paths through ER were tested. See Fig. 1 for a general overview of the models tested. Residual error terms for observed variables reported by the same individual (mother, caregiver, or child) were inter-correlated in order to remove the effect of shared method variance. Child age, gender, and race were controlled by specifying these variables as exogenous predictors of mediators and outcomes.

Fig. 1.

Structural model for children’s emotion regulation as a mediator between risk and psychopathology

Note. Children’s Emotion Regulation = Anger or Sadness Regulation; Children’s Psychopathology = Externalizing or Internalizing Behavior.

Measurement Models

Six latent constructs (indicated with capital letters throughout the remainder of the manuscript) were specified to capture the independent variables (Incarceration-specific Risk Experiences, Environmental Risk), mediator (children’s Anger Regulation, children’s Sadness Regulation), and dependent variables (children’s Externalizing Behavior, and children’s Internalizing Behavior). A Confirmatory Factor Analyses (CFA) indicated acceptable model fit with a PPP of 0.22. Unstandardized factor loadings are presented in Table 3. Standardized loadings ranged from 0.18 (child report of ISRI on the latent variable) to 0.79 (mother report of externalizing behavior on the externalizing latent variable). Correlations among latent variables are presented in Table 4.

Table 3.

Model parameter estimates

	B	Lower 2.5% CI	Upper 2.5% CI
Incarceration-specific Risk Experiences
Mother Report	1.00	1.00	1.00
Caregiver Report	1.29	0.62	2.44
Child Report	0.36	−0.11	0.98
Environmental Risk
Mother Report	1.00	1.00	1.00
Caregiver Report	0.59	0.34	1.25
Anger Regulation
Mother Report	1.00	1.00	1.00
Caregiver Report	0.85	0.44	1.54
Child Report	0.57	0.23	1.06
Sadness Regulation
Mother Report	1.00	1.00	1.00
Caregiver Report	0.33	−0.31	1.36
Child Report	0.31	−0.18	1.12
Externalizing Behavior
Mother Report	1.00	1.00	1.00
Caregiver Report	0.79	0.55	1.13
Child Report	0.44	0.13	0.78
Internalizing Behavior
Mother Report	1.00	1.00	1.00
Caregiver Report	0.83	0.46	1.40
Child Report	0.46	0.09	1.08

N = 117; for effects in bold, 95% credible intervals (CI) exclude zero, Posterior predictive p-value - 0.220

Table 4.

Correlations among latent variables

	1	2	3	4	5	6
1. ISRE		−0.23	−0.43t	−.33 t	0.34*	0.23
2. Environmental Risk			0.08	0.11	0.06	0.05
3. Anger Regulation				0.55**	−0.81**	−0.50**
4. Sadness Regulation					−0.34t	−0.32
5. Externalizing Behavior						0.63**
6. Internalizing Behavior

ISRE Incarceration-specific risk experiences

^tp < .10

^*p < .05

^**p < .01

Does Anger Regulation Mediate the Relation between ISRE and Externalizing and Internalizing Behaviors?

A structural model examining Anger Regulation as a mediator between risk variables and Externalizing Behavior fit the data acceptably (PPP = 0.22). As hypothesized, ISRE was negatively related to children’s Anger Regulation, B = −0.68, 95% CI [−2.09, −0.02]. The indirect effect of ISRE on children’s Externalizing Behavior through children’s Anger Regulation was also significant, B = 0.50, 95% CI [0.02, 1.72]. No other structural equation pathways were significant (see Table 5).

Table 5.

Structural model parameter estimates

	Anger and Externalizing B [95% CI]	Anger and Internalizing B [95% CI]	Sadness and Externalizing B [95% CI]	Sadness and Internalizing B [95% CI]
Structural Model Pathways
ISRE → Regulation	−0.68 [−2.09, −.02]	−0.81 [−2.26, −0.08]	−0.65 [−2.07, 0.51]	−0.59 [−1.91, 0.70]
Environmental Risk → Regulation	−0.38 [−1.75, 0.20]	−0.58 [−2.25, 0.02]	−0.36 [−1.47, 1.00]	−0.24 [−1.24, 1.11]
Regulation →Psychopathology	−0.74 [−1.16, −0.46]	−0.30 [−0.55, −0.10]	−0.42 [−1.14, −0.04]	−0.24 [−0.71, 0.02]
*ISRE → Regulation →Psychopathology	0.50 [0.02, 1.72]	0.24 [0.02, 0.72]	0.28 [−0.13, 0.99]	0.14 [−0.12, 0.54]
*Environmental Risk → Regulation → Psychopathology	0.28 [−0.14, 1.52]	0.17 [−0.01, 0.65]	0.16 [−0.25, 0.67]	0.05 [−0.25, 0.33]
Control Pathways
Age → ISRE	0.12 [−0.20, 0.51]	0.15 [−0.15, 0.56]	0.12 [−0.19, 0.50]	0.13 [−0.18, 0.52]
Age → Environmental Risk	0.83 [0.30, 1.47]	0.79 [0.27, 1.42]	0.82 [0.29, 1.46]	0.76 [0.25, 1.38]
Age → Regulation	0.15 [−0.21, 0.80]	0.28 [−0.11, 1.17]	0.45 [−0.20, 1.01]	0.38 [−0.22, 0.87]
Age → Psychopathology	0.03 [−0.17, 0.19]	0.09 [−0.03, 0.22]	0.16 [−0.06, 0.44]	0.16 [0.03, 0.36]
Sex → ISRE	−0.12 [−0.29, −0.01]	−0.12 [−0.27, −0.01]	−0.12 [−0.28, −0.01]	−0.12 [−0.28, −0.01]
Sex → Environmental Risk	−0.06 [−0.28, 0.15]	−0.06 [−0.27, 0.14]	−0.07 [−0.29, 0.14]	−0.06 [−0.28, 0.15]
Sex → Regulation	0.14 [−0.99, 0.96]	0.09 [−1.27, 0.95]	0.02 [−1.09, 1.21]	0.24 [−0.80, 1.37]
Sex → Psychopathology	0.04 [−0.39, 0.55]	0.32 [−0.01, 0.67]	−0.21 [−0.75, 0.40]	0.27 [−0.09, 0.75]
Race → ISRE	−0.08 [−0.21, 0.07]	−0.07 [−0.21, 0.06]	−0.08 [−0.21, 0.06]	−0.07 [−0.21, 0.06]
Race → Environmental Risk	0.51 [0.30, 0.79]	0.49 [0.29, 0.77]	0.51 [0.30, 0.79]	0.50 [0.29, 0.77]
Race → Regulation	1.63 [0.38, 3.80]	2.00 [0.71, 5.08]	0.53 [−1.67, 2.47]	0.36 [−1.82, 2.07]
Race → Psychopathology	0.26 [−0.25, 0.90]	0.20 [−0.17, 0.64]	−0.55 [−1.11, 0.01]	−0.13 [−0.49, 0.24]
Correlated Error Terms
Latent ISRE and Environmental Risk	−0.33 [−0.90, 0.16]	−0.35 [−0.89, 0.10]	−0.36 [−0.95, 0.11]	−0.35 [−0.91, 0.11]
Mom: ISRE and Environmental Risk	0.16 [−0.38, 0.74]	0.21 [−0.32, 0.78]	0.16 [−0.39, 0.76]	0.19 [−0.35, 0.76]
Mom: ISRE and Regulation	−0.36 [−1.01, 0.24]	−0.49 [−1.16, 0.12]	−0.06 [−0.79, 0.67]	−0.07 [−0.81, 0.65]
Mom: Environmental Risk and Regulation	−0.16 [−0.95, 0.59]	0.01 [−0.79, 0.81]	0.25 [−0.71, 1.16]	0.23 [−0.74, 1.19]
Mom: ISRE and Psychopathology	0.12 [−0.26, 0.52]	−0.08 [−0.39, 0.21]	0.13 [−0.26, 0.55]	−0.09 [−0.41, 0.21]
Mom: Environmental Risk and Psychopathology	0.29 [−0.17, 0.81]	0.48 [0.13, 0.90]	0.33 [−0.14, 0.85]	0.52 [0.15, 0.96]
Mom: Regulation and Psychopathology	−0.87 [−1.60, −0.20]	−0.64 [−1.18, −0.15]	−0.36 [−1.04, 0.25]	−0.47 [−1.06, 0.08]
Caregiver: ISRE and Environmental Risk	0.31 [−0.22, 0.89]	0.33 [−0.18, 0.90]	0.34 [−0.19, 0.92]	0.31 [−0.20, 0.88]
Caregiver: ISRE and Regulation	−0.09 [−0.79, 0.57]	−0.11 [−0.83, 0.56]	0.04 [−0.70, 0.80]	−0.07 [−0.82, 0.70]
Caregiver: Environmental Risk and Regulation	−0.67 [−1.46, 0.04]	−0.63 [−1.44, 0.08]	−0.20 [−1.06, 0.64]	−0.24 [−1.11, 0.59]
Caregiver: ISRE and Psychopathology	0.27 [−0.11, 0.70]	0.12 [−0.20, 0.47]	0.35[−0.01, 0.78]	0.19 [−0.14, 0.54]
Caregiver: Environmental Risk and Psychopathology	1.03 [0.63, 1.55]	1.05 [0.68, 1.56]	1.07 [0.67, 1.60]	1.09 [0.71, 1.61]
Caregiver: Regulation and Psychopathology	−0.68 [−1.32, −0.13]	−0.44 [−1.02, 0.06]	−0.29 [−0.90, 0.28]	−0.42 [−1.03, 0.14]
Child: ISRE and Regulation	−0.30 [−0.97, 0.31]	−0.28 [−0.94, 0.33]	−0.09 [−0.87, 0.67]	−0.12 [−0.91, 0.64]
Child: ISRE and Psychopathology	0.64 [0.11, 1.27]	0.72 [0.34, 1.20]	0.66 [0.12, 1.29]	0.75 [0.37, 1.23]
Child: Regulation and Psychopathology	−0.62 [−1.26, −0.10]	−0.28 [−0.71, 0.10]	−0.66 [−1.44, 0.03]	−0.77 [−1.34, −0.29]

For effects in bold, 95% credible intervals (CI) exclude zero

ISRE Incarceration-specific risk experiences

^*Indicates test of hypothesized mediated pathways

The structural model examining Anger Regulation as a mediator between risk variables and Internalizing Behavior fit the data acceptably (PPP = 0.20). As hypothesized, ISRE was negatively related to children’s Anger Regulation, B = −0.81, 95% CI [−2.26, −0.08]. The indirect effect of ISRE on children’s Internalizing Behavior through children’s Anger Regulation was significant, B = 0.24, 95% CI [0.02, 0.72]. No other structural equation pathways were significant (see Table 5).

Does Sadness Regulation Mediate the Relation between ISRE and Externalizing and Internalizing Behaviors?

The structural model examining Sadness Regulation as a mediator between risk and Externalizing Behavior fit the data acceptably (PPP = 0.28). Although the mediated pathway was not significant, the relation between sadness regulation and externalizing behavior was significant, B = −0.042, 95% CI [−1.14, −0.04]. No other structural model pathways were significant (see Table 5).

The structural model examining Sadness Regulation as a mediator between risk and Internalizing Behavior fit the data acceptably (PPP = 0.43). None of the structural model pathways, however, were significant (see Table 5).

Discussion

This research provides preliminary evidence that anger regulation is a mechanism that helps explain the relation between exposure to risks associated with MI and children’s psychological problems. Specifically, the total indirect effect of ISRE on children’s externalizing and internalizing behaviors through anger regulation was significant. These patterns of associations were not present for sadness regulation nor for general environmental risk. These findings replicate and extend our previous work with ISREs and general environmental risk within this population and contribute to the study of ER in the context of risk.

The bio-ecological model (Bronfenbrenner and Ceci 1994) provides a useful framework for understanding and interpreting these findings. As children cope with the multifaceted risk of MI, specific risks or threats to family functioning that occur as a results of MI (e.g., separation from siblings, changes in residence) may have a more pronounced effect on psychological outcomes and emotional processes than more general chronic stressors in a child’s environment. In the current paper, we incorporate the central component of the bioecological model by accounting for children’s emotion regulatory capacities. Previous research has demonstrated a strong link between parental incarceration and youth’s increased risk for antisocial and aggressive outcomes (see Murray et al. 2012). In the current study, children’s ISRE related to anger regulation which in turn predicted increased externalizing and internalizing behavior. Importantly, the results of both models indicated the indirect path through anger regulation was significant but the direct path from ISRE to externalizing and internalizing behavior was non-significant.

Regarding the relation between anger regulation and externalizing behaviors, our findings replicate two prior studies that recruited youth exposed to general environmental risk (i.e., high rates of poverty, exposure to violence), used child report on the CEMS anger regulation scale, and tested the relation concurrently (Sullivan et al. 2010) and longitudinally (Folk et al. 2014). Other research also using the CEMS anger regulation scale has indicated this same pattern of relations using middle-class, low-risk samples (Cui et al. 2015; Zeman et al. 2002). The findings from the current study strengthen this small corpus of research by its incorporation of multiple reporters for all constructs and its use of SEM to analyze relations in a statistically stringent manner. Taken together, children’s exposure to ISREs was associated with poor anger regulation that, in turn, was associated with more engagement in destructive, delinquent, and other types of externalizing behavior. As such, difficulties with anger regulation may be a contributing factor to the high rate of intergenerational transmission of incarceration among children of MI (Dallaire 2007; Huebner and Gustafson 2007).

Anger regulation also mediated the link between ISREs and children’s internalizing behavior. That is, children who experienced more ISRES also had poorer anger regulation and more internalizing symptoms. Zahn-Waxler et al. (2000)challenged researchers to consider emotions other than sadness when examining the role of negative emotions in internalizing behaviors. For example, irritability is a diagnostic marker of both internalizing and externalizing emotions. Our findings point to the important role of anger regulation in internalizing behaviors in a high-risk sample. Similar relations between anger regulation and internalizing symptoms have been found in other research that have used a low income, ethnically diverse sample of adolescents (Criss et al. 2015), an inner city, primarily Black sample of children with a 2-year longitudinal design (Folk et al. 2014), and a sample of middle-class, White children (Zeman et al. 2001; Zeman et al. 2002). However, the relations examined in these other studies primarily relied on sole reporters and used correlational or regression models to document these linkages.

In contrast to the cohesive set of findings regarding the mediational role of children’s anger regulation, the results of models examining sadness regulation did not provide evidence of mediation between ISREs and either externalizing or internalizing behaviors. Although children’s sadness regulation was associated with fewer externalizing behaviors, there were no significant links between any of the other constructs in the externalizing model. This is perhaps not altogether surprising. Little prior research has provided evidence for the linkage between sadness regulation and externalizing behaviors in high risk samples with the exception that sadness regulation was found to be inversely related to relational aggression (Sullivan et al. 2010). The measure of externalizing behavior in the current study did not directly examine this form of aggression. Zeman et al. (2002) found that sadness dysregulation (i.e., crying, pouting) was related to peer ratings of aggressive behavior (e.g., starts fights) which provides another lens from which to view aggressive behavior that differs from the mother, caregiver, and child report information solicited in this study.

The experience and expression of sadness in a higher risk environment may differ in fundamental ways than in lower risk contexts. Within a high-risk population, expressing sadness may exude vulnerability and thus, it may not be adaptive to express sadness openly. For example, Miller and Sperry (1987) indicated that in low-income environments, children are socialized by their mothers at an early age to express anger and aggression to project an image of “toughness” that is adaptive within this particular context. Alternatively, the lack of mediation in the current study could be due to measurement issues. Perhaps there is a particular vocabulary that is used to express sadness that was not captured in the sadness regulation scale that was developed using White, middle-class children (Zeman et al. 2001). Interestingly, the internal consistencies for this scale in the current study were higher for the adult than child reporters and similar to the validation study (Zeman et al. 2001), providing initial support for this supposition. It could also be that caregivers and incarcerated mothers, in particular, may not be the most reliable reporters of the child’s less observable internalizing behaviors (De Los Reyes et al. 2015), particularly if they are not in close contact with the child or are overwhelmed by the new caregiving responsibilities.

In a meta-analytic review, Murray et al. (2012) demonstrated a strong link between the experience of parental incarceration and the development of externalizing behaviors; the associations between parental incarceration and internalizing behavior were less strong. A similar finding was obtained in the current study. In the sadness regulation models, it is interesting that the ISRI no longer predicted to internalizing or externalizing behaviors, as it has in previous work (Dallaire et al. 2015a). Although mediation was not demonstrated here, the relations between ISREs, environmental risk, sadness regulation and psychological outcomes warrant further investigation into the potentially mediating role of sadness ER in this context.

This study contributes to the children’s ER literature by examining the mediating role of anger and sadness regulation in a sample of high-risk children exposed to the deleterious effects of MI. The use of multiple reporters and SEM provides a more stringent test of the relations among the predictors, outcomes, and mediators that has not been present in prior research with children experiencing MI. Despite these strengths, the results must be interpreted with the limitations of this study in mind. First, a cross-sectional design was used that limits inferences of causality. Second, the sample size did not permit the use of multi-group models to test variables such as race/ethnicity or gender that may play an important role in elucidating how ER may be a protective factor against the impact of MI for specific subgroups of children. Third, as discussed previously, the assessment of children’s sadness regulation may have not been sensitive to the ways in which sadness is managed within this unique high-risk context. Future research should consider using behavioral tasks and observational paradigms in addition to self- and other-reports to more completely discern how ER may be a protective mechanism.

Clinical Implications

This research draws attention to a promising avenue for future research that may guide preventive intervention efforts; namely, the potential value of teaching ER skills. The literature is replete with evidence that poor ER skills portend the development of most forms of psychopathology and are also implicated in negative social, scholastic, and physical health outcomes (e.g., Chaplin and Cole 2005). Thus, developing or adapting existing interventions (e.g., Ehrenreich et al. 2007; Suveg et al. 2007) to teach children emotional competency skills, including ER, may be a “best practices” approach that will also benefit children in high-risk environments. The challenge to implement such interventions, however, lies in developing interventions that are sensitive to the unique challenges of high risk contexts including identifying emotion socialization models in the child’s environment. Only one study has examined emotion socialization in the context of MI and found that emotion socialization strategies typically thought to be adaptive (i.e., problem-focused, emotion-focused) do not provide the same psychosocial benefits to children exposed to high compared to low levels of MI risk (Zeman et al. 2016). Thus, intervention efforts will need to carefully consider the role of contextual factors in order to develop culturally-sensitive practices that teach children ER strategies that will be constructive in their particular context.