Child and Household Regulation: Influences on the Bidirectional Link Between Harsh Parenting and Behavior Problems in Middle to Late Childhood

Abstract

The current longitudinal study examined bidirectional links between harsh parenting (HP) and child externalizing behaviors (EXT) in middle childhood (ages 6–8, 55.7% female, 92% White), with self-regulation (effortful control, working memory, attention regulation) and household chaos as moderators. Data were collected from 174 families across three annual waves, using both maternal reports and observational assessments. Cross-lagged panel analyses revealed bidirectional effects based on observer-reported HP, while maternal reports showed only a child-driven effect (EXT → HP). Moderation analyses indicated that distinct aspects of child-regulation influenced these dynamics in nuanced ways. Effortful control unexpectedly amplified the parent-driven effect (HP → EXT), while working memory buffered the parent effect. Attention regulation strengthened the child-driven effect. Household chaos did not moderate these links. These findings underscore the dual role of self-regulation as both a risk and protective factor, depending on the context and regulatory domain. Results highlight the value of multimethod, multi-informant designs in understanding complex parent-child interactions and suggest that interventions promoting both supportive parenting and child regulatory skills may help mitigate behavioral difficulties during this key developmental stage.

Child and Household Regulation: Influences on the Bidirectional Link Between Harsh Parenting and Behavior Problems in Middle to Late Childhood

Harsh parenting (HP) involves coercive behaviors and negative emotional expressions, often characterized by psychologically and sometimes physically controlling, reactive, and punitive. HP can lead to children’s aggressive behavior through mechanisms like modeling and social information processing (Pinquart, 2017). While much of the existing literature focuses on the unidirectional effects of parenting on children’s externalizing behaviors, or EXT (i.e., impulsivity, and aggressiveness), parent-child interactions are inherently bidirectional. Children’s behaviors influence parenting and vice versa, creating reciprocal effects over time (Yan et al., 2021). For instance, a child’s behavioral problems may increase parents’ reliance on harsh discipline, further exacerbating the child’s issues (Pinquart, 2017; Yan et al., 2021).

Bidirectional dynamics are shaped by broader individual and household contexts that influence parenting and child maladjustment (Coldwell et al., 2006; Hong et al., 2021, 2024). However, limited research examines how distinct aspects of child self-regulation (e.g., effortful control, working memory, and attention regulation) and household regulation (e.g., routines, predictability) may amplify or buffer the bidirectional associations between HP and child EXT over time. This study addresses this gap by integrating maternal reports and observational data, using cross-lagged panel analyses to explore these relationships. This approach incorporates various moderators, including task-based assessments of working memory, survey-based measures of effortful control, observer ratings of attention regulation, as well as household regulation (household chaos). The study focuses on middle to late childhood (ages 6 to 8), a critical developmental period when self-regulation skills are consolidating, and parent-child interactions play a critical role in shaping behavioral and emotional outcomes.

Bidirectionality

From the perspectives of the transactional system and family system theory, the parent-child relationship is bidirectional and reciprocal (Belsky, 1984; Lansford et al., 2018; Yan et al., 2021). For example, children exhibiting high levels of EXT often elicit increased hostility and reduced warmth from parents, while positive child behaviors typically evoke warmer, less harsh parenting (Williams & Steinberg, 2011). These interactions influence parenting quality and shape developmental trajectories over time (Yan et al., 2021). However, research on the bidirectional relationship between HP and child EXT shows mixed findings, with some studies identifying parent-driven effects (e.g., Coley et al., 2014; Serbin et al., 2015), others child-driven effects (e.g., Burke et al., 2008; Lansford et al., 2018), and some showing bidirectional effects between parents and children (Bauer et al., 2021; for meta-analysis Yan et al., 2021). Understanding bidirectional mechanisms requires robust research designs, such as using multiple informants to reduce bias and employing longitudinal methods to track changes over time (Singer & Willett, 2003). Advanced analyses, like cross-lagged panel analyses or latent growth modeling with three or more time points, help infer temporal and potentially causal directional effects between parent and child behaviors (Little et al., 2007). These methods facilitate the examination of dynamic changes and patterns over time, offering insights that single-time-point studies cannot provide (White & Arzi, 2005).

Middle childhood is a critical period for examining EXT, as behavior problems often emerge or intensify alongside HP practices. This developmental phase, marked by the transition to school, can shift bidirectional parent-child relationships and introduce external influences like teachers (Sameroff & Mackenzie, 2003). However, despite significant developmental changes occurring during the transition from middle to late childhood, the impact of these changes on the bidirectional relationship between HP and child EXT is not well examined. To address gaps in existing research, the current study incorporates multiple reporters (i.e., mothers and observers) and longitudinal data collected across three time points to examine the bidirectional relationship between HP and child EXT during middle childhood.

Regulation as Moderator of Bidirectional Effects

The bidirectional relationship between HP and child EXT does not occur in isolation but is shaped by broader individual and environmental factors, (Coldwell et al., 2006; Hong et al., 2021, 2024). In particular, regulatory processes at both the child level (effortful control, working memory, attention regulation) and the household level (e.g., consistency of routines and predictability) may influence the strength and direction of these bidirectional parent-child effects. Although often studied separately, both individual regulatory capacities and environmental regulation likely shape the dynamic interplay between parenting and child behavior. Next, we consider how child and household regulation function as moderators in the longitudinal, bidirectional associations between HP and EXT. We begin by exploring the role of child self-regulation in shaping these effects.

Child Self-regulation

Self-regulation is a multifaceted construct encompassing physiological, social-emotional, cognitive, and behavioral domains (McClelland et al., 2014). It involves the ability to plan and adapt behavior to achieve goals (Gross & Thompson, 2007), which is critical for children’s development and overall well-being (McClelland et al., 2014). This study focuses on three key distinct components of self-regulation: effortful control, working memory, and attention regulation. While they are modestly correlated, they are recognized as conceptually and empirically distinct and may differentially impact parent-child interactions depending on children’s emotional and cognitive regulatory abilities. However, limited research has examined how these specific self-regulatory components may moderate the bidirectional associations between HP and child EXT.

Effortful control, a temperament-based self-regulatory system (Posner &Rothbart, 2000), includes the dimensions of inhibitory control, attentional focusing, and attentional shifting (Eisenberg et al., 2001). It plays a significant role in the regulation of emotions and behaviors and reflects the ability to inhibit reactive dominant responses in order to perform subdominant responses (e.g., detecting errors and engaging in planning) (Posner & Rothbart, 2000).

Working memory refers to the ability to retain and manipulate information over short periods, which is crucial for completing tasks as it enables individuals to retain key details, adjust plans, and integrate new information (Garon et al., 2008). This capacity is integral to self-regulation, allowing individuals to maintain focus on goals, resist distractions, and adapt flexibly to environmental changes (Friedman et al., 2006). As a core component of executive function, working memory helps regulate thoughts and emotions (Ochsner & Gross, 2008). Children with low working memory often struggle to process and retain information, leading to impulsive responses and, in some cases, aggressive behaviors (McQuade et al., 2013). Research also links working memory to cognitive reappraisal, a strategy for interpreting situations and modulating emotional responses effectively (De France & Hollenstein, 2022). Thus, working memory plays a critical role in the cognitive control of emotions, enabling reflective decision-making over reactive behavior (De France & Hollenstein, 2022).

Attention regulation refers to the ability to sustain focus, filter distractions, remain alert to goals, and coordinate attention effectively (Ruff & Rothbart, 1996). These skills are critical for helping children manage emotions, inhibit impulsive behaviors, and navigate social environments (Harris et al., 2007). The capacity to refocus attention is particularly important for regulating anger and mitigating antisocial behavior (Wang et al., 2012). Furthermore, difficulties in attention and impulse regulation are linked to higher child EXT (Perry et al., 2018).

Previous literature links lower self-regulation capacities to higher levels of EXT and other aspects of psychopathology. For example, children with lower effortful control, working memory, or attention regulation are more likely to display EXT (e.g., Wang et al., 2015, Huang-Pollock et al., 2016; Zhou et al., 2007). Although few studies have examined self-regulation as a moderator in the bidirectional link (a child and parent effect) between HP and child EXT, preliminary evidence suggests that it plays a role. The diathesis-stress model suggests that children with lower levels of effortful control are more vulnerable to adverse parenting practices (Monroe & Simons, 1991; Morris et al., 2002). Likewise, stronger executive functions like working memory and attention regulation may buffer against the negative effects of HP by helping children manage stress and regulate behavior (Hong et al., 2024; Horn et al., 2018).

Specifically, child effortful control was found to function as a moderator in the parent effects on child EXT, such that the effects of HP on child EXT (HP ◊ EXT) were buffered for children with high effortful control (e.g., Lee et al., 2022; Morris et al., 2002; Reuben et al., 2016), with one study finding a risk-inducing effect of effortful control (Hong et al., 2025). However, some studies found no evidence for this effect (e.g., Hong, 2021; Zubizarreta et al., 2019). Regarding whether child effortful control moderates the child-driven effect (where EXT predicts HP), research is limited. Two studies found that child effortful control significantly moderated this child effect, with EXT predicting HP (Hong et al., 2024, 2025). Interestingly, Hong et al. reported that higher levels of child effortful control amplified risk rather than buffering it, showing a stronger association between EXT and HP—contrary to the expected protective effect and some prior findings about the parent-driven effect linking HP to subsequent EXT. These findings highlight the complexity of effortful control in parent-child interactions and the need for further investigation.

To our knowledge, no studies have specifically examined the moderating role of working memory in the bidirectional relationship (parent and child effects) between HP and child EXT. However, evidence suggests that executive function, including working memory, can serve as a protective factor in stressful environments. For example, one study found that strong executive function at age 9 mitigated the longitudinal pathway from EXT at age 6 to HP at age 7.5 and subsequent EXT at age 9, demonstrating its protective factor (Hong et al., 2024). Similarly, children with low executive function exposed to parental polyvictimization exhibited severe EXT, whereas those with high executive function did not (Horn et al., 2018).

Similarly, to our knowledge, no studies to date have specifically examined attention regulation as a moderator in the parent and child effects linking HP and child EXT. However, research has shown that attention skills moderate the link between the intensity of maltreatment and academic performance, indicating their potential significance in resilience processes (Slade & Wissow, 2007). These findings underscore the need for further investigation into both working memory and attention regulation as moderators in the dynamic interactions between parenting practices and child behavior.

What could explain the buffering or risk-enhancing effects of child self-regulation? When examining how self-regulation moderates the influence of HP on child EXT—a “parent effect”—stronger regulation skills, such as higher effortful control and working memory, often help children cope with stress, manage emotions, and mitigate the negative impact of HP (Eisenberg et al., 2005; Hong, 2021). Conversely, children with weaker self-regulation may struggle to meet environmental demands, which may intensify negative outcomes in stressful contexts. Regarding the “child effect” (EXT → HP), poorly regulated children with high EXT may increase parenting stress, eliciting harsher reactions from parents and thereby escalating EXT, which reinforces the negative cycle over time (Perry et al., 2018). Although EXT is generally linked with lower self-regulation, this association is not particularly strong. Significant variability exists in self-regulation capacities, even among children with pronounced externalizing tendencies. Some children with higher EXT demonstrate specific strengths in self-regulation skills, which can either buffer or amplify the effects of HP (Hong et al., 2024). By exploring child self-regulation in both parent and child effects, we aim to understand how individual differences in regulatory abilities influence the dynamic relationship between HP and EXT over time. This aligns with evidence emphasizing the importance of self-regulatory skills in shaping the developmental trajectory of EXT, despite substantial variability across individuals (Kuhn et al., 2018).

Thus, self-regulation is key to managing frustration, anger, and fear in the presence of HP, potentially moderating both parent effect (HP → EXT) and child effect (EXT → HP). This study examines distinct aspects of self-regulation—effortful control, working memory, and attention regulation—to explore how they uniquely buffer or exacerbate the bidirectional link between HP and EXT.

Household regulation

In addition to child self-regulation, the broader family environment—particularly the level of household organization and predictability—can also moderate the bidirectional relationship between HP and child EXT. Household chaos, defined by disorganization, unpredictability, and instability, reflects low household regulation (Evans & Wachs, 2010). Prior research links high household chaos to increased levels of EXT (Jaffee et al., 2012) and harsher parenting practices (Coldwell et al., 2006; Hong et al., 2021). Chaotic environments can disrupt children’s ability to regulate attention and arousal, as they may divert their focus away from overstimulating stimuli to reduce over-arousal (Berry et al., 2016; Evans, 2006). While these adaptations may help children cope in the short term, they can limit opportunities for learning socialization skills from parents, ultimately hindering socio-emotional development. Furthermore, high-quality parent-child interactions are essential for fostering children’s ability to maintain and control attention (Conway & Stifter, 2012), but chaotic households often reduce interaction quality due to unpredictability and overstimulation (Berry et al., 2016). These disrupted interactions can negatively affect a child’s emotional and behavioral adjustment, perpetuating maladaptive cycles within the parent-child relationship.

Additionally, an unpredictable and overstimulating environment may adversely affect a child’s attention and emotional regulation (Evans, 2006). Overstimulation can place a burden on the child’s attentional and executive systems, hindering their ability to process and interpret information effectively, thereby impairing their behavioral and emotional regulation capacities (Berry et al., 2016). As a form of environmental “dysregulation”, household chaos may be a critical environmental factor that moderates the bidirectional relationship between HP and child EXT. Research indicates that household chaos strengthens the link between negative parenting and child problem behaviors, especially in high-chaos homes (Chen et al., 2014; Coldwell et al., 2006). However, one study found a contrasting result, finding that the effect of EXT on subsequent HP was stronger in low-chaos households (Hong et al., 2025). Importantly, this moderating effect of chaos was observed specifically in the child-driven effect (EXT → HP), which had not been examined in other studies, such as those by Coldwell and colleagues. Based on prior research, household chaos may influence the bidirectional dynamics by taxing the emotional and behavioral regulation of parents and children (Andrews et al., 2021).

The Present Study

This study examined the bidirectional relationship between parents’ use of HP and children’s EXT from middle childhood to late childhood (approximately ages 6 to 8). Additionally, it examined how distinct aspects of child self-regulation and household chaos may moderate these longitudinal bidirectional relationships, potentially intensifying or mitigating the effects of HP and EXT over time. We hypothesized that (1) there is a longitudinal and bidirectional relationship between HP and child EXT, such that HP will predict subsequent child EXT, and vice versa; (2) the bidirectional relationship between HP and child EXT is stronger among children with lower regulation skills (i.e., lower effortful control, working memory, and attention regulation) compared to those with better regulation skills; (3) the bidirectional relationship between HP and child EXT is stronger in households characterized by higher levels of dysregulation (i.e., chaos) compared to those with lower levels of dysregulation.

Method

Participants

The study included 174 same-sex dizygotic (DZ) twin pairs (55.7% female; mean age at wave 1 = 6.12 years, SD = 0.69 years, range = 4.32–7.92) who were assessed over three annual in-home visits or ‘waves’. Each visit lasted about 2.5 h. The majority of participants were recruited through school nominations, with 273 schools representing over 80% of enrolled families. Schools were asked to distribute study materials to parents of twins who had started kindergarten but had not yet completed first grade. These materials included an introductory letter, a study brochure, and a pre-addressed, stamped postcard for families to express interest. Parents who returned the postcard were contacted by phone, and those who expressed continued interest were sent a brief demographic questionnaire. Families who returned the completed questionnaire were officially enrolled in the study. Additional participants were recruited through Ohio state birth records, mother of twin clubs, and media advertisements. For the current study, one twin from each family was randomly selected. A subsample of families (n = 111) completed temperament surveys near the third annual wave. Depending on the measures used, the final sample included 168–174 families, analyzed using full information maximum likelihood (FIML). Parental education varied: 12–17% had a high school diploma or less, 23–29% had some college or an associate degree, 30–31% held a bachelor’s degree, 4–6% had some post-graduate education, and 5% had completed a postgraduate degree. In prior research with data on short term and working memory among the mothers of the children in the current study (Deater-Deckard, 2010), we noted that this and other correlated cognitive skills among the mothers (e.g., oral vocabulary, spatial reasoning) showed a typical population distribution of scores. The sample was predominantly White (92%), with 6% of families led by single mothers.

Procedure

Families completed three annual in-home assessments lasting approximately three hours each. All procedures for the study were approved by the Ohio State University Institutional Review Board. Parents provided consent, and children assented. Parents completed questionnaires, and children participated in behavioral and cognitive tasks. Mothers engaged in two structured, mildly frustrating 10-minute games with each twin. These games involved collaborative tasks such as drawing pictures on an Etch-A-Sketch (e.g., drawing a house together) and moving a marble through a tilting maze box. For each task, the mother and child controlled separate knobs and were instructed not to touch each other’s controls. These interactions were videotaped and later coded by trained research assistants who were not involved in the home visits. At the end of the home visit, trained testers evaluated children’s behaviors during cognitive assessments (see Petrill et al., 2006 for further details). Families received $100 per wave. Around the time of the third wave (approximately one month prior to the visit), parents received a supplemental questionnaire to assess child temperament. A subgroup of families (n = 111) returned the completed questionnaire.

Measures

Harsh Parenting (HP)

During annual home visits, mothers and children participated in two structured tasks. Trained research assistants reviewed recorded interactions and rated mothers’ behaviors using the Parent-Child Interaction System (PARCHISY; see Deater-Deckard et al., 1997) on 7-point Likert-type scales (1 = no occurrence of the behavior, to 7 = continual occurrence of the behavior). Coders maintained reliability levels of Cronbach’s α > 0.75 throughout data collection. Specific PARCHISY items—negative control (e.g., physical control of dials or child’s hand/arm/body, criticism), negative affect-rejection (e.g., frowning, cold/harsh voice), and conflict (e.g., minor or major disagreement, shared negative affect, arguing)—were combined to create a PARCHISY harsh parenting score.

Mothers also completed the Parent Feelings Questionnaire, PFQ (Deater-Deckard, 2000) at child ages 6, 7, and 8 years, which included a 16-item Negativity scale (α = 0.90) to assess negative feelings and behaviors toward their child (e.g., “Sometimes I am not happy about my relationship with this child”, “This child and I fight or argue more than I would like to.”). Test-retest reliability for negativity scores across the three waves ranged from r =.58 to 0.83.

The observer ratings from PARCHISY and mother-reported PFQ negatively scores were not sufficiently correlated (r range = −0.16 ~ 0.03) to justify creating a single cross-informant HP composite score. Consequently, these measures were analyzed separately.

Child Externalizing Behaviors (EXT)

Mothers completed the Child Behavior Checklist (CBCL; AChenbach, 1991) at child ages 6, 7, and 8 years to assess the frequency of specific behaviors and emotions over the past six months. The Externalizing Syndrome subscale (e.g., aggressive and non-aggressive conduct problems) included 33 items rated on a 3-point scale (0 = not true, 1 = somewhat sometimes true, 2 = very true or often true). The items were averaged to create the Externalizing Syndrome score (e.g., n = 30; α = 0.70 to 0.80). The Externalizing scores showed moderate to high stability over time (r =.52 to 0.79). Between 6.7 and 8.7% of children in our sample (depending on the ages) met the threshold for subclinical levels of EXT (T-scores ≥ 65). A smaller proportion of these children, 2.7–3.9% of all children in the sample, met the criteria for clinically significant EXT (T-scores ≥ 70).

Child Effortful Control

At the age 8 visit, mothers rated their child’s effortful control using the Behavior Questionnaire Short Form (CBQ-SF, Putnam & Rothbart, 2006), which measures attentional and inhibitory control. Example items include “Can wait before entering into new activities if s/he is asked to,” “Is good at following instructions,” and “When building or putting something together, becomes very involved in what s/he is doing, and works for long periods.” The scale consisted of 26 items (α = 0.60 to 0.87). Items were rated on a 7-point Likert-type scale (e.g., 1 = “extremely untrue” to 7 = “extremely true” for the child).

Working memory

Children completed a shortened version of the Stanford-Binet Intelligence Scale IV (SBIV; THorndike et al., 1986) at ages 6, 7, and 8. The Digit Span task assessed working memory using raw scores from Forward and Backward Digit Span. The standard total score was used for analysis.

Attention Regulation

At the end of each annual home visit, trained testers assessed the child’s attention regulation using Bayley’s Behavior Record (BBR; Bayley, 1969), a 5-point Likert scale. The BBR is an observational tool designed to evaluate child behaviors, encompassing affective and behavioral dimensions such as attention/persistence and positive and negative affect. It has been used in previous research to assess infant and child attention regulation (Wang et al., 2012). In this study, two highly correlated items were used to evaluate attention regulation: on-task behavior (1 = constantly off task, does not attend; 5 = constantly attends) and persistence in attempting to complete tasks (1 = consistently lacks persistence; 5 = consistently persistent). Also, independent observers rated children’s on-task behavior during the videotaped parent-child interactions using the Parent-Child Interaction System or PACHISY (see Deater-Deckard et al., 2005; see above explanation about PARCHISY). Specifically, the on-task behavior item (1 = not evident to 7 = constant) was used from PARCHISY.

To compute a reliable overall measure of attention regulation, principal component analyses were conducted using the on-task and persistence scores from BBR and the on-task score from PARCHISY. The first component explained 55–60% of the variance, with loadings from 0.54 to 0.88 at wave 1, 0.50 to 0.87 at wave 2, 0.38 to 0.88 at wave 3. The scores were standardized, averaged, and standardized again to calculate an overall attention regulation z-score (where higher scores indicate better attention regulation).

Household Chaos

Parents completed a short version of the Chaos, Hubbub, and Order Scale (CHAOS; Matheny et al., 1995) at each wave. The short version, used in prior studies (e.g., Coldwell et al., 2006; Hong et al., 2021), includes six items designed to assess household chaos (e.g., “You can’t hear yourself think in our home,” “It’s a real zoo in our home,” “We are usually able to stay on top of things” (reverse scored), “There is usually a television turned on somewhere in our home,” and “The atmosphere in our house is calm” (reverse scored). The items were rated on a 5-point Likert-type scale (1 = definitely untrue, 5 = definitely true) and averaged to compute an overall household chaos score (α = 0.65 to 0.68).

Covariates

Gender differences in EXT may arise from variations in how poor parenting practices impact boys and girls (Pitzer et al., 2009). Boys generally exhibit higher levels of EXT than girls (Wang et al., 2021). Furthermore, greater parental education is linked to reduced HP practices (Berthelon et al., 2020). Therefore, we statistically controlled for the child’s gender (coded as female = 1, male = 2) and the average years of education of both parents.

Data Analysis

First, descriptive statistics and bivariate Pearson correlations were computed using SPSS version 25 (IBM Corporation). To test study hypothesis 1, cross-lagged panel analyses were conducted in Mplus version 8 to examine bidirectional relations between HP and child EXT at ages 6, 7, and 8, separately for maternal and observer perspectives. Missing data were handled using full information maximum likelihood estimation (FIML). The model estimated autoregressive effects (e.g., EXT at age 6 predicting EXT at age 7), concurrent covariances (e.g., links between EXT and HP within the same wave), and bidirectional cross-lagged paths (e.g., EXT at age 6 predicting HP at age 7 and vice versa) (see Fig. 1). This path model controlled for construct stability and simultaneous associations while estimating cross-lagged pathways.

Fig. 1.

Diagram of cross-lagged panel model. Notes. HP = harsh parenting; EXT = externalizing behaviors. The dotted border represents testing of moderation effect

To test study hypotheses 2 and 3, predictors were standardized, and moderating variables (effortful control, working memory, attention regulation, and household chaos) with interaction terms were added to the cross-lagged panel equations. Analyses were conducted in Mplus 8 using FIML, with significance evaluated via 95% bias-corrected confidence intervals (CIs) from 10,000 bootstrap samples. For each informant (mother and observer), four moderation models were estimated: one including effortful control as a moderator, another including working memory as a moderator, a third including attention regulation as a moderator, and a fourth including household chaos as a moderator. Age 6 measures (Working memory, attention regulation, and chaos) moderated links between age 6 EXT/HP and age 7 HP/EXT, and age 7 measures moderated links between age 7 and age 8. Since effortful control was measured only at age 8, it was used accordingly. We first ran each equation with covariates included (i.e., parental education; child sex), then dropped nonsignificant covariates, re-estimated the equation, and interpreted the equation with only significant covariates included.

Results

Descriptive statistics and correlations for study variables are shown in Table 1. All variables showed some skewness, but most approximated a normal distribution spanning the entire range of each scale. Bivariate correlations between study variables showed that in general there was covariation between higher household chaos, harsher parenting, and greater child EXT. Child effortful control and HP were negatively correlated at all waves. Overall, higher child working memory was associated with lower household chaos.

Table 1.

Correlations and descriptive statistics

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18
1.Sex	1
2. Educ	0.10	1
3. EC	−0.20*	−0.08	1
4. WM1	−0.04	0.18*	0.05	1
5. WM2	−0.04	0.10	0.04	0.50**	1
6. AR1_z	−0.06	0.07	0.16	0.16*	0.09	1
7. AR2_z	−0.04	−0.04	0.13	0.07	0.05	0.36**	1
8. Chaos1	−0.08	−0.11	−0.12	−0.21**	−0.18*	0.06	−0.09	1
9. Chaos2	−0.06	−0.18*	−0.16	−0.14	−0.19*	0.04	−0.09	0.76**	1
10. HP1	−0.03	0.04	−0.31**	0.00	−0.03	−0.07	−0.06	0.21**	0.18*	1
11. HP2	0.02	0.12	−0.31**	−0.05	−0.08	−0.11	−0.07	0.24**	0.26**	0.77**	1
12. HP3	−0.02	0.11	−0.25*	−0.04	0.00	−0.07	−0.06	0.22*	0.23*	0.66**	0.72**	1
13. PAR1	0.05	−0.05	−0.09	−0.10	−0.21**	−0.12	−0.03	−0.04	−0.08	0.01	0.01	−0.02	1
14. PAR2	0.15	0.10	−0.09	−0.25**	−0.11	−0.20*	−0.10	0.15	0.12	0.21*	0.29**	0.32**	0.22**	1
15. PAR3	−0.08	0.17	0.04	−0.18	−0.01	−0.10	0.02	0.13	0.19*	0.17	0.28**	0.26**	0.04	.36**	1
16. Ext1	0.02	−0.05	−0.16	0.05	−0.10	0.01	−0.16	0.24**	0.24**	0.63**	0.58**	0.52**	−0.03	.16	.11	1
17. Ext2	0.14	−0.03	−0.21*	0.01	−0.10	−0.02	−0.10	0.19*	0.25**	0.42**	0.61**	0.44**	−0.05	.22**	.23*	.74**	1
18. Ext3	0.04	−0.01	−0.13	−0.07	−0.04	−0.12	−0.19*	0.16	0.18*	0.44**	0.51**	0.61**	0.03	.41**	.37**	.69**	.73**	1
M	0.44	5.88	5.34	52.37	52.55	0.00	0.00	2.36	2.31	32.36	29.37	29.35	1.23	1.17	1.20	.24	.20	.21
SD	0.50	1.64	0.60	4.64	5.25	1.00	1.00	0.64	0.62	12.03	11.54	11.65	0.38	.34	.36	.19	.17	.18

Sex child sex, Educ parental education, EC effortful control, WM working memory, AR attention regulation, HP harsh parenting, PAR PARCHISY observed harsh parenting, Ext externalizing behaviors, 1, wave 1, 2 wave 2, 3 wave 3, z standardized score

Direct Effects

The regression coefficients for the bidirectional model between HP and EXT from each informant are shown in Table 2.

Table 2.

Unstandardized regression coefficients for bidirectional pre-dictions of harsh parenting and child externalizing problems

	Maternal perception		Observer perception
Parameter	B	SE	B	SE
Cross-lagged paths
HP1→EXT2	− 0.01	0.01	− 0.13	0.27
HP2→EXT3	0.02	0.01	1.24**	0.31
EXT1→HP2	1.14**	0.40	0.03**	0.02
EXT2→HP3	0.08	0.54	0.03	0.02
Within-construct paths
HP1→HP2	0.64**	0.06	0.20**	0.07
HP2→HP3	0.71**	0.08	0.33**	0.09
EXT1→EXT2	0.69**	0.06	0.67**	0.05
EXT2→EXT3	0.67**	0.08	0.69**	0.06
Within-wave covariances
HP1—EXT1	14.24**	2.07	−0.02	0.06
HP2—EXT2	3.89**	0.75	0.04	0.03
HP3—EXT3	4.74**	0.98	0.07	0.04

^*p<.05,

^**p<.01 (all two-tailed tests). Note. N = 168. Statistical sig.; statistical significance (p value). All results were computed by Mplus. HP harsh parenting, Ext externalizing behaviors, 1 wave 1, 2 wave 2, 3 wave 3

Maternal perception

Maternal reports of HP and child EXT showed significant stability over time and significant concurrent covariances at each time point. Significant cross-lag effects were observed between EXT and HP. Specifically, EXT at wave 1 predicted HP at wave 2, and HP at wave 2 marginally predicted HP at wave 3. However, HP at wave 1 did not predict subsequent EXT at wave 2, nor did EXT at wave 2 did not predict subsequent HP at wave 3.

Observer perception

The observers’ ratings of maternal HP and maternal reports of child EXT showed significant stability across time points. However, no significant concurrent covariances were found between HP and EXT within each time point. Significant cross-lag effects were observed between HP and EXT. Specifically, EXT at wave 1 predicted HP at wave 2, and HP at wave 2 predicted EXT at wave 3. Notably, HP at wave 1 did not predict EXT at wave 2, and EXT at wave 2 did not predict HP at wave 3.

Moderation Effects

Results from the moderated cross-lagged panel analysis for maternal perception are shown in Table 3, and for observer perception in Table 4.

Table 3.

Unstandardized regression coefficients for the moderating effects predicting externalizing problems from maternal perception

	Moderators
	Effortful control		Working memory		Attention regulation		Household chaos
Parameter	B	SE	B	SE	B	SE	B	SE
Predicting HP2
HP1	0.60**	0.07	0.64**	0.06	0.64**	0.06	0.64**	0.06
Ext1	1.19**	0.40	1.14**	0.40	0.93*	0.42	1.04*	0.42
Moderator	−1.95	1.18	− 0.92	1.33	− 0.86	0.67	0.48	0.98
Ext1 x moderator	0.16	0.60	0.04	0.62	0.82*	0.42	0.25	0.44
R 2	0.63		0.62		0.62		0.62
Predicting HP3
HP2	0.69**	0.08	0.72**	0.08	0.72**	0.08	0.72**	0.08
Ext2	0.26	0.54	0.08	0.53	0.05	0.54	0.03	0.56
Moderator	− 0.21	1.42	1.18	1.44	− 0.42	0.72	− 0.38	1.23
Ext2 x moderator	1.38	0.73	−1.46	0.80	− 0.29	0.39	0.44	0.66
R2	0.52		0.53		0.52		0.51
Predicting EXT2
Ext1	0.69**	0.06	0.69**	0.06	0.69**	0.06	0.69**	0.06
HP1	− 0.01	0.01	− 0.01	0.01	− 0.004	0.01	− 0.004	0.01
Moderator	− 0.40*	0.19	0.02	0.21	− 0.04	0.11	− 0.09	0.16
HP1 x moderator	0.01	0.01	0.02	0.02	0.003	0.01	− 0.001	0.01
R2	0.57		0.56		0.54		0.56
Predicting EXT3
Ext2	0.65**	0.08	0.69**	0.08	0.68**	0.08	0.68**	0.08
HP2	0.03*	0.01	0.02	0.01	0.02*	0.01	0.02*	0.01
Moderator	0.21	0.21	− 0.04	0.22	− 0.23	0.11	− 0.14	0.18
HP2 x moderator	0.03*	0.02	− 0.03*	0.02	− 0.005	0.02	− 0.01	0.02
R2	0.56		0.57		0.57		0.55

^*p<.05,

^**p<.01 (all two-tailed tests). Note. N = 168. Statistical sig.; statistical significance (p value). All results were computed by Mplus. HP = harsh parenting; Ext = externalizing behaviors; 1=wave 1; 2=wave 2; 3=wave 3

Table 4.

Unstandardized regression coefficients for the moderating effects predicting externalizing problems from observer perception

	Moderators
	Effortful control		Working memory		Attention regulation		Household chaos
Parameter	B	SE	B	SE	B	SE	B	SE
Predicting HP2
HP1	0.19*	0.07	0.18*	0.07	0.18*	0.07	0.21**	0.07
Ext1	0.03*	0.02	0.04*	0.01	0.04**	0.02	0.03*	0.02
Moderator	− 0.05	0.06	− 0.02**	0.01	− 0.07*	0.03	0.06	0.04
Ext1 x moderator	− 0.03	0.03	− 0.001	0.003	− 0.02	0.02	− 0.004	0.02
R 2	0.10		0.14		0.13		0.10
Predicting HP3
HP2	0.33**	0.09	0.34**	0.09	0.34**	0.09	0.34**	0.09
Ext2	0.03	0.02	0.03	0.02	0.03	0.02	0.02	0.02
Moderator	0.07	0.06	0.002	0.01	0.01	0.03	0.06	0.05
Ext2 x moderator	− 0.01	0.03	0.001	0.004	− 0.03	0.02	0.04	0.03
R2	0.14		0.14		0.17		0.15
Predicting EXT2
Ext1	0.66**	0.05	0.67**	0.05	0.67**	0.05	0.68**	0.05
HP1	− 0.16	0.27	− 0.09	0.27	− 0.09	0.28	− 0.32	0.31
Moderator	− 0.33	0.18	0.01	0.02	− 0.04	0.11	− 0.10	0.16
HP1 x moderator	0.49	0.50	0.03	0.07	− 0.03	0.35	− 0.36	0.38
R2	0.57		0.56		0.55		0.56
Predicting EXT3
Ext2	0.70**	0.06	0.70**	0.06	0.70**	0.06	0.70**	0.06
HP2	1.26**	0.31	1.15**	0.32	1.16**	0.31	1.49**	0.33
Moderator	0.17	0.20	− 0.01	0.20	− 0.19	0.10	− 0.19	0.17
HP2 x moderator	− 0.12	0.44	− 0.03	0.06	− 0.07	0.24	− 0.90	0.65
R2	0.58		0.57		0.59		0.60

^*p<.05,

^**p<.01 (all two-tailed tests). Note.N = 168. Statistical sig.; statistical significance (p value). All results were computed by Mplus. HP = harsh parenting; Ext = externalizing behaviors; 1=wave 1; 2=wave 2; 3=wave 3

Effortful Control as a Moderator

For maternal perceptions of their own HP, child effortful control significantly moderated the bidirectional relationship between HP and EXT. Specifically, the pathway from HP at wave 2 to EXT at wave 3 was significant only for children with average or high effortful control (see Table 3). However, no significant moderation effects were found for observer perceptions (see Table 4).

Working Memory as a Moderator

For maternal perceptions, the pathway from HP at wave 2 to EXT at wave 3 was significant only for children with low levels of working memory (see Table 3). However, no significant moderation effects by working memory were observed in the analysis of observer perceptions (see Table 4).

Attention Regulation as a Moderator

For maternal perceptions, the pathway from EXT at wave 1 to HP at wave 2 was significant only for children with average or high attention regulation (see Table 3). However, observer perceptions showed no significant moderation effects by attention regulation (see Table 4).

Household Chaos as a Moderator.

No significant moderated cross-lagged effects were found in any of the tested models (see Tables 3 and 4).

Discussion

The present study examined how multiple indicators of child regulation and household chaos moderate the bidirectional relationship between HP and child EXT in middle childhood. This work addressed key gaps in the literature by: (1) considering distinct aspects of child regulation as moderators, (2) exploring the largely overlooked role of household chaos, and (3) integrating both maternal reports and observational data to assess these longitudinal dynamics.

To address these gaps, we utilized multiple reporters (mothers, observers) and methods (tasks, surveys, and observations) and conducted cross-lagged panel analyses to examine the bidirectional relationship between HP and child EXT from ages 6 to 8. Specifically, we tested surveyed child effortful control, task-based working memory, observed attention regulation, as well as surveyed household regulation as potential moderators in the bidirectional parent and child effects. To our knowledge, this is the first study to examine multiple distinct indicators of child regulation and household chaos as moderators using longitudinal data and both maternal and observer perceptions of HP during middle childhood.

Regarding our first hypothesis, bidirectional effects emerged from observers’ perceptions of maternal HP, supporting the reciprocal nature of parent-child interactions. This aligns with coercion theory, transactional, and family systems frameworks, which emphasize that both parents and children influence each other’s behaviors. For example, children with high EXT may elicit harsher parenting, which in turn exacerbates EXT and negatively affects their socio-emotional development (Williams & Steinberg, 2011; Yan et al., 2021). Prior work also shows that children’s behaviors can elicit specific parental responses (Belsky, 1984; Lansford et al., 2018). A meta-analysis by Yan et al. (2021) further supports these findings, demonstrating that child EXT significantly drives changes in parenting practices over time. These child-driven effects were comparable in magnitude to parent-driven effects, highlighting the importance of considering both directions to fully understand the reciprocal nature of parent-child interactions.

In contrast, maternal perceptions of HP indicated only a “child effect”. Mothers may view their parenting as reactive to their children’s actions, potentially underestimating their own role in these interactions. The discrepancy with observer perceptions underscores the importance of integrating multiple perspectives, such as observational and self-reported data, to more comprehensively capture the bidirectional parent-child dynamics. These findings support the idea that children actively influence parenting behaviors, which can lead to more hostile and coercive parenting. Together, they emphasize the reciprocal nature of parent-child interactions and highlight the value of using diverse methods to understand the dynamics fully.

Turning to the second hypothesis, we examined whether child self-regulation moderated the bidirectional parent and child effects. First, we examined child effortful control as a moderator. Surprisingly, effortful control acted as a risk enhancer rather than a buffer for the parent effect. Based on maternal perceptions of HP, HP at wave 2 predicted EXT at wave 3 only for children with average or high effortful control, suggesting that high effortful control exacerbated, rather than mitigated, the impact of HP on subsequent EXT. One possibility is that these children may be more attuned to negative parenting, making them more sensitive to HP and increasing stress internalization. Alternatively, these children may suppress their emotions in response to HP, inadvertently reinforcing negative interactions. A similar pattern was observed in prior research on children’s perceptions of paternal parenting, where effortful control amplified the effects of HP on subsequent EXT (Hong et al., 2025). While these results contradict expectations and currently lack a clear theoretical framework, they underscore the importance of exploring these dynamics further. Future research could investigate potential mediators such as emotional sensitivity, coping strategies, or parent-child relationship quality to better understand effortful control’s role in parent-child dynamics.

Regarding child working memory as a moderator, results showed that, based on maternal perceptions of HP, the pathway from HP at wave 2 to child EXT at wave 3 was significant only for children with low working memory. This suggests that high working memory may protect against the effects of HP. Children with strong working memory are better equipped to retain and process information, enabling them to reflect thoughtfully on social situations and respond in a measured, non-impulsive manner (McQuade et al., 2013). These findings align with previous research by Hong et al. (2024), where child executive function (including working memory) weakened the indirect pathway from EXT to HP and subsequent EXT. Similarly, in the current study, high working memory likely helped children interpret challenging situations accurately and regulate their emotional responses effectively. This regulatory capacity enables children to respond to HP with greater deliberation and reflection, reducing the likelihood of escalating into EXT. By contrast, children with low working memory may struggle to regulate emotions, increasing vulnerability to HP and reinforcing maladaptive interactions. These findings underscore the protective role of cognitive processes like working memory in shaping children’s resilience against negative environmental influences. Importantly, previous work using this sample has also shown that maternal working memory moderates the association between challenging child behavior and maternal harsh reactive parenting; mothers with lower working memory scores were more likely to exhibit reactive negativity in response to child difficult behaviors such as noncompliance and inattention (Deater-Deckard et al., 2010). Given the intergeneration transmission (including heritability) of executive functioning, it is plausible that children with executive function difficulties are more likely to have parents with similar challenges, which may contribute to the parent-child dynamics observed in this study. Future research should explore interventions aimed at strengthening working memory and related capacities to improve outcomes for children at risk of adverse parenting practices and behavioral challenges.

Regarding attention regulation as a moderator, results showed that, based on maternal perceptions of HP, the pathway from child EXT at wave 1 to HP at wave 2 was significant only for children with high attention regulation. Contrary to our expectations, attention regulation appeared to act as a risk enhancer rather than a buffer, amplifying the child effect and strengthening the link between EXT and subsequent HP. To our knowledge, previous studies have not specifically examined how child attention regulation interacts with EXT to predict HP. However, prior research on effortful control as a moderator of the child effect has shown similar patterns, with higher regulatory capacities such as effortful control sometimes increasing the likelihood of HP behaviors (Hong et al., 2024; Hong et al., 2025).

One possible explanation is that parents may perceive children with high attention regulation and high EXT as capable of better control but resistant to complying with expectations or guidance. This perceived mismatch between the child’s regulatory abilities and their behavioral challenges might lead parents to adopt harsher or more coercive parenting practices, potentially in an effort to enforce compliance. Parents might assume that such children should be capable of more appropriate behaviors, and when these expectations are unmet, they may escalate their responses, resulting in increased HP (Hong et al., 2024). While this interpretation is speculative and lacks a specific theoretical framework, it highlights the need for further research to understand how children’s regulatory skills influence parenting behaviors, particularly in contexts of behavioral challenges. Interventions could focus on helping parents better understand and adapt their responses to children’s regulatory abilities, promoting more constructive and supportive parenting practices that align with the child’s individual needs and capacities.

Turning to the third hypothesis, household chaos did not significantly moderate any bidirectional links between HP and child EXT, contrary to expectations. This contrasts with prior research on early adolescents showing child effect (EXT ◊ HP) emerged only in less chaotic homes, suggesting high chaos may dampen parental responsiveness to child behavior (Hong et al., 2025). Alternatively, parents in more regulated household environments may be more responsive to children’s EXT, potentially increasing HP to address such behaviors. Discrepancies in findings may arise from differences in sample characteristics, child age, or how HP is measured. It is also possible that household chaos may function differently across developmental stages, with children responding variably to chaotic environments as they age. Our results suggest that household chaos may not serve as a universal moderator of the well-established bidirectional relationships between parenting and child behavior. Future studies should incorporate more nuanced, multidimensional measures of household chaos and explore its interaction with parenting strategies and child behavior at different developmental stages. Such research could inform interventions aimed at helping families manage disorganization and create supportive home environments.

From observer perceptions of HP, evidence showed bidirectional relations between HP and child EXT, but no significant moderation effects in any models. Observer perceptions may capture immediate and overt dynamics of parent-child interactions, which might reflect a generalized pattern of reciprocal influence that may not be sensitive to individual or contextual differences. Alternatively, the measures tested as moderators may not significantly influence the bidirectional links observed in real-time interactions. Other unmeasured factors, such as parent-child communication quality or emotional tone, may play a more critical role. These findings highlight the need for future research to explore additional contextual and relational factors that could influence the observed bidirectional effects in parent-child interactions.

Overall, the results suggest that child regulation can moderate both child and parent effects, although the moderating influences varied across the examined paths and different aspects of regulation. These variations may reflect developmental changes, suggesting that different indicators of child regulation may act as moderators in distinct ways depending on the child’s age and developmental stage. Alternatively, the inconsistent patterns observed may suggest that these regulatory indicators do not systematically moderate the well-established bidirectional links between HP and EXT during middle to late childhood.

Limited research on how distinct child regulatory indicators and household regulation moderate bidirectional parent-child processes highlights a critical gap. Our findings suggest that child regulation interacts with both parent and child effects, emphasizing the importance of exploring multiple dimensions of regulation. Although the moderating effects differed between self-reported and observed parenting, the results support the idea that child regulation significantly influences the relationship between parenting behaviors and child adjustment (Lee et al., 2022; Reuben et al., 2016). More importantly, different regulatory indicators may uniquely moderate specific pathways (Eisenberg & Zhou, 2016). Future studies should further investigate these factors across developmental stages using multiple informants (Singer & Willett, 2003) to better understand family dynamics and inform targeted interventions.

Caveats and Conclusions

This study has several limitations. First, effortful control was measured only at age 8 with a subset of parents, limiting developmental tracking. However, age 8 data likely approximate age 7, allowing us to infer its influence across the study. Using proximal measures is valid in longitudinal research when exact temporal alignment is not feasible (Kochanska & Knaack, 2003). Second, while the cross-lagged panel models are useful for examining bidirectional relationships, they do not account for within-person changes over time, which may bias estimates (Lucas, 2023). Future research could address this using dynamic structural equation modeling to capture within-person changes and improve the precision of estimates. Third, the predominantly White, middle-class sample limits generalizability, highlighting the need for replication in more diverse populations. Fourth, our study’s results may not generalize to older youth or clinical populations. Future research should explore whether similar patterns are observed in older children, adolescents, and clinically referred and diagnosed samples to assess whether our findings apply more broadly. Only a small proportion of the current sample met clinical thresholds for EXT. If future research with clinical samples revealed different patterns of parent-child dynamics than those reported in our study, this would have important implications for intervention development and clinical practice. Finally, although the current study focused on bidirectional processes as direct effects, the longitudinal structure of the data also could allow for the investigation of potential indirect (i.e., mediating) pathways. For example, early child regulatory abilities may influence later parenting behaviors indirectly through intermediate factors such as EXT, parental stress, or the quality of the parent-child relationship. Future studies could extend the current work by modeling such indirect mediated pathways, to further elucidate the developmental processes that shape parent-child dynamics over time.

Despite its limitations, the current study offers important insights into the bidirectional link between parenting and child adjustment from middle childhood to late childhood. By examining longitudinal patterns of HP and child EXT, and testing the moderating roles of effortful control, working memory, attention regulation, and household regulation, the study highlights how child regulation shapes parent-child dynamics. Using both maternal and observer perceptions of HP provided a more comprehensive perspective. The findings confirm bidirectional relationships between HP and child EXT, with child regulation serving as both a buffer and, in some cases, a risk enhancer. These findings underscore the need to consider child regulation as a pivotal aspect of the parent-child dynamic and a key target for intervention efforts. This study also emphasizes the importance of multimodal assessments, as different measurement approaches can yield distinct insights (Eisenberg & Zhou, 2016). Future research can build on these findings to deepen understanding of parent-child dynamics and develop evidence-based interventions to enhance child and family well-being.

Availability of data and materials:

The data and data analysis programming code used in the current study are available from the corresponding author upon reasonable request.