Parental Emotion Regulation Strategies and Parenting Quality Predict Child Internalizing Symptoms in Families Experiencing Homelessness

Abstract

Adaptive emotion regulation (ER) in parents has been linked to better parenting quality and social-emotional adjustment in children from middle-income families. In particular, early childhood may represent a sensitive period in which parenting behaviors and functioning have large effects on child social-emotional adjustment. However, little is known about how parent ER and parenting are related to child adjustment in high-risk families. In the context of adversity, parents may struggle to maintain positive parenting behaviors and adaptive self-regulation strategies which could jeopardize their children’s adjustment. The current study investigated parents’ own cognitive ER strategies and observed parenting quality in relation to young children’s internalizing and externalizing problems among families experiencing homelessness. Participants included 108 primary caregivers and their four- to six-year-old children residing in emergency shelters. Using multiple methods, parenting and parent ER were assessed during a shelter stay and teachers subsequently provided ratings of children’s internalizing and externalizing difficulties in the classroom. Parenting quality was expected to predict fewer classroom internalizing and externalizing behaviors as well as moderate the association between parent ER strategies and child outcomes. Results suggest that parenting quality buffered the effects of parent maladaptive ER strategies on child internalizing symptoms. The mediating role of parenting quality on that association was also investigated to build on prior empirical work in low-risk samples. Parenting quality did not show expected mediating effects. Findings suggest that parents experiencing homelessness who use fewer maladaptive cognitive ER strategies and more positive parenting behaviors may protect their children against internalizing problems.

Dynamic developmental systems models of resilience in children highlight the interdependence of resilience across systems and development (Hostinar & Miller, 2019; Masten & Cicchetti, 2016). For example, the resilience of a child at any time can depend on the adaptive systems within the child, the quality of their relationships, and the resilience of their family members. In particular, parents have been highlighted as a vital relationship that promotes child resilience. Parents can nurture and protect their children from environmental harm, while also serving as primary socialization agents. Further, how well parents’ function as independent entities can cascade to benefit or harm their children through multiple mechanisms. The resilience of parents and their ability to parent effectively could be a key leveraging point for promoting adaptive outcomes in children (Doty, Davis & Arditti, 2017; Masten & Palmer, 2019).

Parenting behaviors have been underscored as playing a particularly salient role in the development of social-emotional abilities in early school-aged children (Landry, Smith & Swank, 2011). Animal and human models suggest that early childhood is a sensitive period, in which enhancing positive parenting behaviors promotes children’s adaptive emotional functioning and its neural correlates (Luby, Belden, Harms, Tilman & Barch, 2016; Rao et al., 2010). However, we still lack a clear understanding about how parent functioning affects child social-emotional adjustment in the context of adversity. The present study examined risk and protective effects of parents’ own cognitive emotion regulation (ER) strategies and parenting behaviors on internalizing and externalizing symptoms among school-aged children experiencing homelessness.

We focused on children experiencing homelessness because they are more likely than low income and middle-income housed peers to experience other adversities (Narayan et al., 2017; Samuels, Shinn, & Buckner, 2010). Consistent with their elevated risk profiles, childhood homelessness has been linked to increased psychopathology, academic problems, and emotion dysregulation (Bassuk, Richard, & Tsertsvadze, 2015; Evans, Li & Whipple, 2013; Labella, Narayan, McCormick, Desjardins, & Masten, 2017; Masten, Fiat, Labella, & Strack, 2015; Obradović, Bush, Stamperdahl, Adler, & Boyce, 2010; Koss & Gunnar, 2018). These findings motivate the need for research on protective processes (e.g. parenting and parent cognitive ER) during key developmental sensitive periods in families experiencing homelessness.

Child Internalizing and Externalizing Symptoms

Internalizing symptoms include withdrawal, somatic complaints, anxiety, and depressed mood, whereas externalizing symptoms refer to delinquent, reactive, and aggressive behaviors. Internalizing and externalizing symptoms have been linked to a range of concurrent and future adjustment problems, including decreased academic achievement, peer problems, and risk for later psychopathology (Calkins & Perry, 2016; Gazelle & Ladd, 2003; Laird, Jordan, Dodge, Pettit, & Bates, 2001; Patterson, DeBaryshe, & Ramsey, 1990). Adaptive emotional and behavioral regulation is especially important during young children’s transition to school. This developmental period represents a time of rapid emotional development with concurrent increases in self-regulatory demands for children across contexts. Emotional and behavioral functioning at this time may lay the foundation for children’s long-term functioning across developmental domains. Children who begin to display behavioral and emotional difficulties, indicated by internalizing and externalizing symptoms, may be at risk for future social, emotional, behavioral, and academic difficulties (Cicchetti & Toth, 1998). Further, some evidence suggests that experiencing poverty during this developmental period increases risk for continuing to experience internalizing and externalizing symptoms throughout childhood (National Institute of Child Health and Human Development Early Child Care Research Network, 2005).

Internalizing and externalizing problems in children are also influenced by parenting behaviors (Chang, Schwartz, Dodge, & McBride-Chang, 2003; Zahn-Waxler, Klimes-Dougan, & Slattery, 2000). The emotion socialization framework suggests that the repetitive nature and salience of everyday social interactions with caregivers shape children’s formation of affective biases and their subsequent likelihood of affective maladjustment (Eisenberg, Cumberland & Spinard, 1998). Parent’s capacity to understand and effectively address their own emotions contributes to how they respond to children’s emotion and implicit emotion socialization. For example, parents with maladaptive ER may model ineffective ER strategies and interact harshly with their children contributing to the development of child symptomology. This is relevant for families experiencing homelessness because they are more likely to suffer from self-regulation challenges (Bassuk et al., 2015). Parents’ own psychological distress has been related to children’s psychopathology in families experiencing homelessness (Bassuk, Weinreb, Dawson, Perloff, & Buckner, 1997; Vostanis, Grattan, Cumella, & Winchester, 1997; Zima et al., 1999). Given that psychological distress is often accompanied by maladaptive ER strategies (Aldao, Nolen-Hoeksema, & Schweizer, 2010), maladaptive ER strategies in parents may be related to children’s psychopathology symptoms.

Parenting Behaviors

High-quality parenting behaviors (i.e., problem-solving skills, positive involvement, and skill encouragement) among caregivers experiencing homelessness have been related to adaptive child functioning, including increased child self-regulation and fewer behavior problems (Gewirtz, DeGarmo, Plowman, August, & Realmuto, 2009; McEwen & Flouri, 2009; Narayan, Herbers, Plowman, Gewirtz, & Masten, 2012; Narayan, Sapienza, Monn, Lingras, & Masten, 2015). Unfortunately, exposure to high levels of adversity can have detrimental effects on parent functioning. Parents experiencing poverty-related stress have fewer resources available to support effective parenting and are more likely to display sub-optimal parenting behaviors (McLoyd, 1998; Vernon-Feagans & Cox, 2013; Winsper, Zanarini, & Woke, 2012; Zimmer-Gembeck & Skinner, 2016;). Given the protective effects of effective parenting in the context of adversity, it is important to understand what supports parenting behaviors and subsequent child outcomes. One understudied process involves the cognitive strategies employed by parents to regulate their own emotions.

Cognitive Emotion Regulation

Cognitive ER strategies involve the modulation of one’s thoughts in order to manage one’s own emotional reactivity (Garenfski, Kraaij, & Spinhoven, 2001). The use of adaptive cognitive ER strategies (e.g., positive reappraisal, positive refocusing) has been related to fewer internalizing symptoms, more meaning-making in the context of adversity, higher positive affect, and greater well-being within the individual (Aldao, et al., 2010; Garnefski et al., 2001; Gross & John, 2003; Schroevers, Kraaij, & Garnefski, 2007). Given well-established associations between parental well-being and child adjustment (e.g., Conger & Donnellan, 2007), parents’ adaptive cognitive ER strategies are expected to promote healthy child adjustment. Parents who use adaptive regulatory strategies may help their children develop emotion regulation skills by modeling effective strategy use, exhibiting more effective parenting behaviors and interacting more positively with their children (Crandall, Deater-Deckard, & Riley, 2015). In a community sample, parents’ use of positive reappraisal was related to less negative affect and more supportive parenting practices (Hughes & Gullone, 2010).

Maladaptive cognitive ER strategies (e.g., rumination, catastrophizing, self-blame, suppression) have been related to negative outcomes within the individual, including psychopathology (Aldao et al., 2010; Garnefski et al., 2001). One meta-analysis found that use of maladaptive ER strategies was associated with more internalizing psychopathology among adults and children, whereas adaptive ER strategies were only weakly associated with lower internalizing symptoms (Aldao et al., 2010). This suggests that maladaptive strategies are more powerful in predicting psychopathology than adaptive strategies are in preventing it. Maladaptive cognitive ER strategies have also been correlated with harsh parenting behaviors. Hughes and Gullone (2010) found that mothers who suppressed their own emotions tended to punish or minimize their children’s expression of negative emotions. Studying maladaptive cognitive ER may be especially relevant for families experiencing high poverty-related stress, which simultaneously elicits negative emotions while exhausting regulatory resources.

Parent Functioning and Homelessness

More research needs to investigate whether parent functioning (e.g., cognitive ER) and parenting behaviors in high-risk crisis contexts differ from lower risk contexts. In a study of parents residing in emergency homeless shelters, parental psychopathology had a direct effect on children’s emotional adjustment but it was not mediated by observed parenting practices (Gewirtz et al., 2009). Further, a study of families exposed to recent incidents of domestic violence also demonstrated no association between parent psychopathology and parenting practices, despite a main effect of parental mental health on child emotional adjustment (Gewirtz, DeGarmo, & Medhanie, 2011). Within high-risk contexts, parents may actively separate the management of their own distress from their role as a parent, in order to provide high-quality care for their children.

In qualitative interviews conducted with homeless women, several participants reported that they were unconcerned with effectively managing their own distress and often pushed it aside to “act appropriately” for their children (Banyard, 1995; Holtrop, McNeil, & McWey, 2015). Adults in high-risk circumstances may need to balance drawing upon limited resources in their physical and social worlds, maintain adaptive self-regulation and functioning, while also serving as an effective caregiver. Consistent with classic theories, including Maslow’s hierarchy of needs (Maslow, 1987), physical security (i.e., housing and food) and relationship functioning (family/children) will likely be organized before self-fulfilling needs (self-regulation skills). Given the large body of research and interventions focused on parenting skills (e.g., DeGarmo, Patterson, & Forgatch, 2004), there is reason to expect that adaptive parenting behaviors could serve as a protective influence on child adjustment despite the maladaptive regulation strategies of their parents. Parental emotional functioning and parenting behaviors may independently predict child outcomes, and/or moderate each other’s effects on child adjustment (i.e., moderation hypotheses).

However, little research has been conducted on the association between emotion regulation and parenting behaviors in high-risk context. In low-risk contexts, theory and research support a mediation hypothesis that predicts a direct effect of parent ER on parenting behaviors (e.g. Crandal et al., 2015; Hughes & Gullone, 2010; Rutherford, Wallace, Laurent, & Mayes, 2015). In this framework, parents need effective coping skills to regulate challenging emotional experiences and to respond effectively and adaptively with their children (Crandall et al., 2015). From this perspective, parents’ emotional functioning indirectly impacts child outcomes through parenting behaviors (i.e. mediation hypotheses).

The Current Study

This study investigated parental cognitive ER and parenting quality as predictors of internalizing and externalizing symptoms in young children experiencing homelessness. We hypothesized that higher observed parenting quality would be related to fewer symptoms for children in school. We also expected that more frequent use of adaptive cognitive ER strategies by parents and less frequent use of maladaptive strategies would be related to fewer child symptoms.

Theoretical and qualitative studies in high-risk contexts suggested a moderating effect of parenting quality on the association between parent cognitive ER and child symptomology. Thus, we expected parenting quality to have a buffering effect against maladaptive cognitive ER, and to enhance the positive effect of adaptive cognitive ER on child symptoms. However, given prior empirical work with low-risk populations that suggests a mediation pathway from parent ER through parenting quality to child outcomes, we evaluated mediation. If this mediating effect is consistent across risk contexts, then we would expect a partial mediation effect of parenting quality on the relation between cognitive ER strategy use and child adjustment. Greater use of adaptive strategies and fewer maladaptive strategies would predict more adaptive parenting behaviors and thereby fewer symptoms in children.

Methods

Participants

Families with a child expected to enter kindergarten or first grade during the next school year were recruited from two urban homeless shelters in a Midwestern U.S. city during the summer of 2014. To be eligible for the study, families had to reside at the shelter for at least three days, speak English well enough to complete study measures, and children could not have severe developmental delays. Of the 180 eligible families, consent was completed with 110 families and two families withdrew after the consent process. Families experiencing homelessness are often dealing with urgent matters related to benefits, housing, and/or health. The 72 families who did not participate left before they could participate, never learned of the study, were too busy, or perhaps were not interested. The final sample of 108 families represented 60% of all possible eligible families.

Primary caregivers were 89.8% female, and their average age was 30.3 years old (SD = 6.6, Range = 19.0 - 50.0). The majority of parents identified as African American (n = 61, 56.5%), biological mothers (n = 95, 88.0%), and single parents (n = 76, 70.4%). Caregivers varied in education - 26.9% did not finish high school, 38.9% received a high school diploma or GED, and 33.3% had some form of higher education or technical school. Children were 46.3% female and 5.2 years old on average (SD = 0.6). Children were 59.3% African American, 24.1% Multiracial, 7.4% Native American, 3.7% Asian or Pacific Islander, 2.8% White, and 2.8% Other.

Families had two other children on average (Range = 0 – 6). About half of the caregivers reported the birth of their first child when they were 18 years old or younger (Median = 19 years old). The majority of parents had been homeless before (n = 75, 69.1%) as had about half (n = 51, 47.2%) of their children. Families ranged significantly in how long they had been staying at shelter, from 3 nights to 485 days (i.e., 1.5 years). The mean number of nights was 44.2 evenings (Median = 17.5).

Parents reported on their own adverse childhood experiences (ACES; Center for Disease Control and Prevention [CDC], 2013) and the majority of caregivers (60.9%) had four or more ACES, compared to 12% in the original CDC study (CDC, 2013). Parents also reported on their children’s experiences of adversity using the Life Time Events Questionnaire (Masten, Miliotis, Graham-Bermann, Ramirez, & Neeman, 1993) Out of 22 questions, children’s median number of adverse experiences was three (Range = 0 – 11). The most common adversities were living in a home with severe adult relationship problems (49%), witnessing violence happening to another person (42.6%), and having an incarcerated parent (39%). Additional information regarding participant demographics can be found in Table 1.

Table 1.

Bivariate Two Tail Pearson Correlations Between Key Study Variables (N = 108)

	1	2	3	4	5	6	7	8	9	10	11	12
Variables	INT	EXT	Mal CER	Adap CER	PQ	Sex	Child Age	Parent Age	No HS	HS	>HS	Teach Rel
1. Internalizing	---
2. Externalizing	0.38**	---
3. Maladaptive Cognitive ER	0.16	0.14	---
4. Adaptive Cognitive ER	−0.13	0.02	0.04	---
5. Parenting Quality	−0.28**	−0.10	−0.08	0.01	---
6. Child Sex (0 is Female)a	−0.26*	0.13	−0.01	0.01	0.09	---
7. Child Age	−0.16	−0.18	−0.27*	0.09	−0.06	−0.08	---
8. Parent Age	−0.01	−0.06	−0.05	−0.07	−0.16	0.19	0.27*	---
9. No High School Educationb	−0.04	0.12	−0.05	−0.07	−0.23*	0.19*	0.13	−0.04	---
10. High School Educationb	−0.10	−0.17	0.12	−0.10	0.11	−0.13	−0.08	0.06	−0.51**	---
11. Post-Secondary Educationb	0.13	0.05	−0.07	0.17	0.14	−0.07	−0.06	−0.02	−0.54**	−0.45**	---
12. Teacher Relationship (Months)	0.19	0.09	−0.13	0.06	0.12	−0.14	0.11	−0.06	−0.03	0.12	−0.10	---
N	79	79	104	104	105	108	108	108	108	108	108	76
Mean	2.09	2.30	11.10	14.81	2.95		5.21	30.31				6.72
SD	2.38	2.50	3.13	2.99	0.36		0.66	6.56				4.27

Note:

^aFemale (n = 50);

^bLess Then a High School Education (n = 41), High School Education (n = 32), Post-Secondary Education (n = 35);

^*p < .05;

^**p < .01;

^***p < .001

Procedures

Initial study sessions took place in private research rooms located inside emergency shelters. Children participated in a 90-minute assessment of school readiness while parents were interviewed about demographic information, cognitive ER strategy use, and experiences of adversity. Caregivers and children then participated in a 12-minute structured sequence of Family Interaction Tasks (FITs; DeGarmo et al., 2004) which were previously modified and validated for use with homeless families (Gewirtz et al., 2009; Narayan et al., 2012). These tasks consisted of a five-minute problem-solving discussion task and two three- and four-minute teaching-oriented games.

Teachers were contacted the following school year and reported on children’s social-emotional adjustment. Highly mobile children are extremely challenging to follow, but we located 82.4% of participating children in 47 different schools. Teachers returned completed questionnaires for 88.7% of these children (n = 79). Teachers reported on children eight months later, on average (SD = 2.8; Range = 3 - 14 months). In an attempt to allow children to form relationships with their teachers, we waited three months into the school year to contact the teachers. However, given the high mobility of these children, there was considerable variability in how long teachers knew the children, ranging from one to twenty-four months (M = 6.7 months, SD = 4.3).

Measures

Parenting quality.

Parent-child interactions were video-recorded and rated using an observational coding system from the creators of the FITs (DeGarmo et al., 2004; Forgatch & DeGarmo, 1999; Palmer & Plowman, 2017). Parenting behaviors were rated on a 1 to 5-point Likert-type scales that were composited into four validated scales. Four scales were computed by averaging items scores: problem solving (16 items, α = .86), positive involvement (37 items, α = .87), skill encouragement (13 items, α = .89), and coercive discipline (20 items, α = .77). Coders were trained to reliability before they independently rated tapes (20, 30, 24, and 20 tapes). Eleven tapes (10.5%) were rated by all coders to assess inter-rater reliability, and intraclass correlation coefficients (ICCs) for each subscale fell in the “good to excellent range” (Koo & Li, 2016): Problem Solving: .94; Positive Involvement: .91; Skill Encouragement: .85; Coercive Discipline: .90.

The problem-solving scale assesses parents’ abilities to identify and define a problem within the family, find solutions, and develop a plan to solve the problem. Positive involvement describes the amount of warmth, affection, and interest the parent directs towards the child. A higher score indicates both a presence of positive involvement and a lack of disrespect. Skill encouragement measures the extent to which a parent scaffolds new skills for their children by providing direction and support. Coercive discipline refers to parents’ harsh, inconsistent, and/or absent limit setting behaviors. A high score indicates a higher frequency of coercive parent behaviors.

The FITs protocol was abridged from its originally implemented form (DeGarmo et al., 2004) to lessen the burden on families. To assess whether behavioral ratings represented an underlying latent factor of parenting quality, a confirmatory factor analysis model was fitted using the R package lavaan (Rosseel, 2012). All scales loaded significantly and in the predicted directions onto an overarching parenting quality factor (problem solving: .50; positive involvement: .92; skill encouragement: .71; coercive discipline: −.63). Fit indices showed acceptable model fit (CFI = 1.0, TLI = 1.04, RMSEA < .001, p = .89, SRMR = .01), suggesting the coder impressions provide an adequate estimate of a general parenting quality factor. Parenting quality was averaged across the outlined scales and then mean-centered for analyses.

Parent cognitive ER.

Parents reported on cognitive strategies they use to respond to stressful life events using the Cognitive Emotion Regulation Questionnaire-Short Form (CERQ-Short; Garnefski & Kraaij, 2006). The CERQ-Short has five scales, consisting of four items each: positive refocusing (α = .72; e.g., “you think of nicer things than what you have experienced.”); positive reappraisal (α = .72; e.g., “you think that the situation also has its positive sides.”); self-blame (α = .79; e.g., “You feel that you are the one to blame for it.”); rumination (α = .75; e.g., “you dwell upon the feelings the situation has brought up for you.”) and catastrophizing (α = .75; e.g., “You continually think how horrible the situation has been.”). Items are measured on a 5-point Likert scale, ranging from 1 (“almost never”) to 5 (“almost always”). Scales were summed and then averaged across adaptive (positive refocusing and positive reappraisal; α = .80; M = 14.8, SD = 2.99, Range = 8.0 – 20.0) and maladaptive (self-blame, rumination, and catastrophizing; α = .85; M = 11.0, SD = 3.1, Range = 5.0 - 19.3) regulation strategies. A principal component analysis (see supplemental materials) was completed to assess scale structure, and results support a two-factor structure along the derived composite scores. Scores were mean-centered for analyses.

Child internalizing and externalizing symptoms.

Teachers completed the Strengths and Difficulties Questionnaire (SDQ; Goodman, 1997). The emotional symptoms subscale (α = .81; M = 2.1, SD = 2.4, Range = 0 – 10.0) indexed internalizing symptoms and the conduct problems scale (α = .80; M = 2.3, SD =2.5, Range = 0 – 10.0) indexed externalizing symptoms. Each scale has five items rated on a 3-point scale (0 = not true, 1 = somewhat true, and 2 = certainly true), which are summed to yield a composite score. Results of a principal component analysis - completed to verify scale structure - are in supplemental materials. Based on a standardization sample (Goodman, 1997), most children in this sample (86.1%) had emotional problem scores in the normal range, 1.9% had scores in the clinically borderline range, and 12% had scores in the clinical range. In a normative American sample of four- to seven-year-olds, 3% scored in the clinical range of symptoms on the SDQ (Bourdon, Goodman, Rae, Simpson & Koretz, 2005). Regarding conduct problems, the majority of children in this study (58.2%) scored in the normal range, 5.6% in the clinically borderline range, and 36.2% in the clinical range. In contrast, 11% of four-to-seven-year-olds scored in the clinical range in a normative American sample (Bourdon et al., 2005).

Covariates.

Covariates that might modify the association between variables of interest were explored. Candidate covariates included child age, sex, and ethnicity (white versus non-white) of the parent and child, caregiver relationship (biological mother versus other), length of teacher relationship, shelter site (private versus public), and dummy coded parent education (less then high school, high school, or more than high school education). Correlations between potential covariates were first explored (Table 1) and then entered into a preliminary regression. If covariates were not significant predictors in the preliminary step or we did not have a theoretical reason to expect they would influence our variables of interest they were dropped from future analyses.

Plan for Analyses

A set of hierarchical linear regressions were conducted using R version 3.3.2 (R Development Core Team, 2008). Child symptomatology was regressed onto predictors in the following steps: (1) covariates, (2) adaptive cognitive ER strategies, maladaptive cognitive ER strategies, and parenting quality, and (3) interaction effects. Child sex, parent education, and length of teacher relationship were included as covariates. All other covariates (child age, parent sex, caregiver relationship, ethnicity, parent education, and shelter site) were unrelated to the variables of interest and were excluded from analyses. Interaction effects for adaptive cognitive ER strategies and maladaptive cognitive ER strategies were tested in separate models. Given a small amount of overlap in some children attending the same schools we utilized a cluster robust standard error adjustment with the R program clubSandwich (Supplemental Table 4), which corrects for unaccounted-for error dependence in our models (Barnard & Rubin, 1999; Pustejovsky & Tipton, 2018). Given the minimal amount of clustering in schools, and no change in findings once implementing the adjustment, results are presented without the adjustment. To assess mediation, we used the Hayes Process Macro in IBM SPSS Statistics Version 25 with 5,000 bootstrapped samples while including relevant covariates.

Missing data.

The proportion of missing data from shelter sessions was minimal, ranging from 0% (sociodemographic information) to 3.7% (parent cognitive ER). Teacher data was available for 73.1% of the sample. This rate of missingness is expected for studies following highly mobile families. Children with teacher data did not differ from children without teacher data on key demographic variables and we concluded that missingness of teacher data was unlikely to reflect the child’s psychological adjustment, permitting the assumption that the data were missing at random (MAR). We used multiple imputation with the R program MICE to generate unbiased parameter estimates across 20 datasets using fully conditional specifications (van Buuren & Groothuis-Oudshoorn, 2011). Data from all 20 imputed datasets were pooled using Rubin’s rules which incorporates both within and between imputation variability (Rubin, 2004; Schafer & Graham, 2002). Results using listwise deletion were similar. Results using imputed data are reported.

Results

Bivariate Correlations

Descriptive statistics and bivariate correlations are presented in Table 1. Teacher reports of internalizing and externalizing problems were positively correlated. Internalizing problems and child sex were also correlated such that teachers reported more internalizing symptoms in girls than boys. Lower observed parenting quality was related to higher internalizing problems; however, it was not related to externalizing problems. Parents with less than a high school education had lower parenting quality scores. Cognitive ER strategies were not significantly correlated with observed parenting behaviors nor child adjustment.

Child Internalizing Symptoms

Hierarchical linear regression results (Table 2) indicated a main effect of parenting quality and maladaptive cognitive ER on child internalizing symptoms such that higher parenting quality and lower use of maladaptive cognitive ER were related to fewer internalizing symptoms. Contrary to expectations, there was no main effect of parental adaptive cognitive ER strategy use on child internalizing symptoms.

Table 2.

Hierarchical Linear Regression Analyses of Cognitive Emotion Regulation Strategies and Parenting Quality on Teacher Rated Child Internalizing Symptoms Using Multiple-Imputation (N = 108)

	B	SE B	t	R2	ΔR2
Internalizing Symptoms
1. (Intercept)	3.16	.61	5.14***	.11
Child Sex	−1.35	.53	−2.55*
Parent High School Education	−0.82	.66	−1.24
Parent Post-Secondary Education	0.22	.68	0.33
Length of Teacher Relationship	0.10	.07	1.54
2. (Intercept)	2.72	.58	4.69***	.27	.16**
Child Sex	−1.05	.49	−2.14*
Parent High School Education	−0.60	.63	−0.95
Parent Post-Secondary Education	0.84	.67	1.26
Length of Teacher Relationship	0.15	.06	2.57*
Maladaptive Cognitive ER	0.17	.09	2.05*
Adaptive Cognitive ER	−0.15	.08	−1.88†
Parenting Quality	−2.29	.80	−2.52**
3a. (Intercept)	2.79	.58	4.81***	.31	.04*
Child Sex	−1.16	.49	−2.38*
Parent High School Education	−0.70	.63	−1.11
Parent Post-Secondary Education	0.78	.66	1.17
Length of Teacher Relationship	0.14	.06	2.52*
Maladaptive Cognitive ER	0.20	.08	2.40*
Adaptive Cognitive ER	−.17	.09	−1.99†
Parenting Quality	−2.37	.79	−3.02**
Maladaptive Cognitive ER X Parenting Quality	−0.54	.26	−2.10*
3b. (Intercept)	2.69	.58	4.62**	.27	.00
Child Sex	−1.03	.49	−2.09*
Parent High School Education	−0.58	.63	−0.91
Parent Post-Secondary Education	0.87	.66	1.32
Length of Teacher Relationship	0.15	.06	2.63*
Maladaptive Cognitive ER	0.17	.08	2.06*
Adaptive Cognitive ER	−0.17	.09	−1.89†
Parenting Quality	−2.27	.80	−2.83**
Adaptive Cognitive ER X Parenting Quality	0.14	.27	0.52

^*p < .05,

^**p < .01,

^***p < .001,

^†p < .10

Interaction terms were entered individually at step three and evaluated one at a time. Results supported the hypothesized interaction between maladaptive cognitive ER strategies and parenting quality (Table 2). We graphed this interaction effect (Figure 1) and then probed the interaction to understand the direction and magnitude of the association (Figure 2).

Figure 1.

Interaction effect of parenting quality and parent negative cognitive emotion regulation on child internalizing symptoms using centered variables, controlling for the main effect of child sex, parent education, length of teacher relationship, and positive cognitive emotion regulation. Low parenting quality and high parenting quality are one standard deviation from the mean

Figure 2.

The relation between parent maladaptive cognitive emotion regulation and child internalizing problems per observed level of parenting quality. The x axis indicates the standardized level of the moderator and the vertical dashed line (−0.86) indicates the level of the moderator at which the focal variable becomes signficantly associated with the depndent variable. The 95% confidence region is indicated by the shaded area

As depicted in Figure 1, simple slopes analysis suggest that parents’ use of maladaptive cognitive ER strategies was associated with higher child internalizing symptoms when observed parenting quality was low (−1 SD: B = 0.36, SE = 0.11, t(99) = 3.33, p < .05). Parents who demonstrated above average parenting quality had children with lower internalizing symptoms, regardless of maladaptive cognitive ER use (+1 SD: B = −.02, SE = 0.09, t(99) = .18, p = .64). At the mean level of parenting quality, maladaptive cognitive ER was associated with higher children internalizing symptoms (B = 0.19, SE = 0.07, t(99) = 2.78, p < .05). Results were further probed via regions of significance testing. Figure 2 shows that the effect of maladaptive cognitive ER on child internalizing symptoms was significantly higher for caregivers who were average to low in parenting quality. Overall, results suggested that high parenting quality buffered the effects of parent maladaptive cognitive ER strategy use on child internalizing symptoms. No interaction effect was observed for adaptive cognitive ER by parenting quality in predicting child internalizing. There was also no evidence of an effect of parent ER on parenting quality, thus there was no support for a mediation effect.

Child Externalizing Symptoms

Hierarchical linear regression was conducted to evaluate parents’ cognitive ER strategies and parenting quality as predictors of teacher-rated child externalizing symptoms. Analyses proceeded in the same manner as described for child internalizing. Consistent with bivariate correlations, neither demographic covariates nor focal parent variables predicted child externalizing symptoms. See supplemental materials for additional information. Mediation analyses did not yield evidence for parenting quality mediating the association between parent cognitive ER strategies and child social-emotional outcomes.

Discussion

The goal of this study was to examine how parent cognitive ER strategies and parenting quality were associated with child internalizing and externalizing symptoms in the context of homelessness. Parenting and parent emotion regulation abilities have been highlighted as key mechanisms by which child adaptive emotional functioning is fostered. Further, research suggests that children may be particularly sensitive to parent behaviors and functioning in early childhood (Landry et al., 2011; Luby et al., 2016; Rao et al., 2010). The current study adds to this work by evaluating how these constructs are related in the context of homelessness, where individuals are at higher risk for psychopathology symptoms, difficulties with emotion regulation, and ineffective parenting practices. Findings differed from research in low-risk samples, such that there was no association between parent cognitive ER strategies and parenting practices. However, findings did suggest that fewer maladaptive cognitive ER strategies and effective parenting reduced risk for internalizing symptoms in the classroom. Parenting quality also moderated the association between parent cognitive ER with child social-emotional adjustment. Parents who displayed lower parenting quality and reported higher use of maladaptive cognitive ER strategies had children with higher teacher-reported internalizing symptoms. Maladaptive cognitive ER strategies were not related to child outcomes when parenting quality was above the sample mean, which is consistent with the hypothesis that high-quality parenting has a buffering effect on parents’ maladaptive cognitive ER strategies with respect to children’s internalizing symptoms.

It is conceivable that a specific subset of parents who use maladaptive methods of self-regulation can alter their behavior and “rise to the occasion” in the context of parenting in stressful circumstances. This is interesting, given that past research has underscored the function of social relationships as promoters for adaptive self-regulation (Finkel & Fitzsimons, 2011), and parents who are experiencing homelessness report that their children are their biggest social supports (Wood, Valdez, Hayashi, & Shen, 1990). This process may also become particularly salient in the context of parenting under extreme stress. Despite their own struggles, parents may be motivated by their role as caregivers to provide the most supportive environment possible when children are experiencing a number of other risk factors that could threaten their functioning.

Self-regulation and behavioral systems likely function differently in the context of homelessness. Parents in this study endorsed significantly higher use of maladaptive and adaptive cognitive ER strategies when compared to the CERQ-Short validation sample (Garnefski & Kraaij, 2006). Given the high-risk nature of this sample, parents likely experience more stressors and greater demands on coping skills. The study found a direct association between parental maladaptive cognitive ER and child internalizing symptoms, such that those using more maladaptive cognitive ER strategies had children with higher rates of teacher-reported internalizing symptoms. This adds to previous work that has found intraindividual effects and suggests that there could also be interindividual effects of parents’ own cognitive ER on child psychopathology. However, our study did not find a direct association between adaptive cognitive ER strategies and children’s internalizing or externalizing symptoms. In previous research, adaptive cognitive ER strategies have been related to lower psychopathology within individuals (Garnefski et al., 2001). Contrarily, a meta-analysis with children and adults suggested that there were weak and inconsistent associations linking adaptive cognitive ER strategy use and externalizing outcomes within the individual (Aldao et al., 2010). It may be that in high-risk environments such as homeless shelters, strategies involving positive reappraisal and positive refocusing are rendered less effective by the context of current stress. Exploring more neutral reappraisal strategies, similar to mindfulness strategies, is an important area for future research. Shift and persist concepts may also be relevant given that families experiencing homelessness often have less control over their immediate environment (Chen & Miller, 2012). This may render many behaviorally oriented self-regulation strategies useless, and necessitate more acceptance rather than direct volitional management.

Additionally, we should note that there was a main effect of sex on children’s internalizing symptoms in which teachers reported girls having more internalizing symptoms than boys. This extends findings in adult and adolescent populations that suggest that girls are at higher risk for internalizing psychopathology (e.g., Schleider & Weisz, 2016). However, for preschool-aged children, the majority of work finds no difference between sexes (e.g., Carneiro, Dias, & Soares, 2016; Sterba, Prinstein, & Cox, 2007), and if differences do emerge, boys are more symptomatic (Carneiro et al., 2016; Sommers, 2000). If our findings are replicated, the development of risk for internalizing problems in high-risk contexts should be closely examined. This pattern of risk could reflect greater vulnerability to depression in the family and/or accelerated development in brain regions implicated in the development of internalizing psychopathology (e.g. VanTieghem & Tottenham, 2017). If young girls living in the context of high poverty-related stress have a higher risk for early-emerging internalizing problems, it would be important to intervene early.

Further, inconsistent with previous work (e.g., Chang et al., 2003; Hoeve, Dubas, Gerris, van der Laan, & Smeenk, 2011; Webster-Stratton, 1998), we did not find effects of parenting quality on child externalizing symptoms. It is possible that externalizing problems at school could emerge later in development, or parenting may not consistently predict conduct problems under conditions of extreme poverty-related stress because parental influences on child behavior are overwhelmed by other environmental factors. In prior research with preschool-aged children experiencing homelessness, parenting quality predicted lower externalizing behavior only when sociodemographic risk was below the sample mean (Labella et al., 2017).

Strengths and Limitations

A key strength of this study was the use of multiple methods and informants to reduce response bias. Parents reported on their own use of cognitive ER strategies, and parent-child interactions were rated with a rigorous behavioral coding system for measurement of parenting quality (DeGarmo et al., 2004). Further, the inclusion of teacher report after laboratory assessment is a particular strength considering how difficult it is to follow families experiencing homelessness longitudinally. These methods limited bias associated with parent report and illuminated the association between family processes and child functioning across contexts. Finally, this study is unique in that we focus on parent adjustment in the context of homelessness and use a resilience framework to highlight the importance of how parent functioning may cascade to affect children’s adjustment in school contexts.

This study also had limitations. The study did not include some potentially relevant covariates such as parental affective expression, child temperament, parental depression, and child housing status at teacher follow-up. Children with more difficult temperaments may elicit more negative reactions from parents, making emotion regulation demands harder for those parents (Kiff, Lengua, & Zalewski, 2011). Parental depression is known to be related to both maladaptive coping strategies and maladaptive child outcomes (Zimmer-Gembeck & Skinner, 2016). Without controlling for parental depression, it is unclear if parental depression would account for the association between parental cognitive ER strategies and child internalizing outcomes in the context of low parenting quality. The study also was limited by assessment at only two time points because it would have limited the interpretation of mediation effects.

Future Directions and Clinical Implications

The current study is innovative because it suggests that maladaptive ER among parents may have cross-generational implications for child internalizing symptoms, contingent on the broader context of parenting quality. This suggests that it is important to help parents avoid maladaptive emotion management strategies in addition to teaching supportive parenting practices (Havighurst & Kehoe, 2017). Early childhood—a period of rapid self-regulation development—may be a particularly valuable period for interventions that bolster these protective factors. These results emphasize the role of the caregiver in shaping child functioning and conferring risk or protection from psychopathology symptoms in contexts of adversity. However, teaching parents cognitive ER skills may not improve their explicit parenting skills in the context of homelessness. Future work should explore these associations in high-stress contexts and continue to aim for differentiating between the effects of low parental maladaptive cognitive ER versus high adaptive cognitive ER on child adjustment in longitudinal studies. Moreover, it is important to continue research on malleable family processes that support parenting and child resilience in high-risk families, while also implementing policies that mitigate exposure to severe poverty-related adversities, including homelessness.