The Influence of Children’s Effortful Control on Parent-Child Behavioral Synchrony

Abstract

Child temperament appears to evoke specific parenting behaviors that contribute to child development. However, questions remain about whether individual differences in children’s temperamental self-regulation, namely effortful control (EC), shape moment-to-moment parent-child interaction dynamics. Accordingly, we examined whether differences in children’s EC were related to dynamic synchrony of parent and child behaviors during a challenging problem-solving task. We also tested whether these relations varied by parents’ expressions of positive and negative behaviors that might differentially support or undermine children’s regulatory efforts. State-trait multilevel models demonstrated that parent-child dyads engaged in dynamic, real-time behavioral concordance while parents engaged in positive but not negative behaviors. Further, dynamic concordance during parents’ expressions of both positive and negative behaviors was moderated such that dyads with children higher in EC showed greater concordance. Additionally, when child behavior was more negative on average, parent behavior was also more negative on average. Results suggest parents’ positive (compared to negative) behaviors are more likely to facilitate real-time synchrony and that children with higher EC may experience or foster greater behavioral synchrony with parents. Discussion centers on the importance of children’s individual differences in shaping parent-child synchrony and potential implications for children’s developing self-regulation.

Research indicates that parent-child synchrony of positive behaviors contributes to children’s adaptive socioemotional development, including improved self-regulation, behavioral adjustment, and cognitive abilities (Davis et al., 2017; Kochanska et al., 2008). Behavioral synchrony can be defined as the moment-to-moment coordination of goal-directed behavior (e.g., parental discipline, parental support of children’s efforts, child compliance, and persistence) in face-to-face interactions (Lunkenheimer, Kemp, et al., 2017). These behaviors may be synchronous when they couple closely in time and change in a coordinated fashion, for example, as parents display supportive behaviors and children enact adaptive responses to that support. Accordingly, we propose synchrony of goal-directed behavior can be conceptualized as concordant or discordant in terms of whether parent and child behaviors are shifting in in the same or opposite direction along a continuum from more adaptive to maladaptive behavior.

It is not yet clear how such dynamic behavioral processes operate in dyadic parent-child interactions and what functions they serve for child development. We know that concordant negative behaviors, such as the combination of parental harsh discipline or intrusion with child noncompliance (e.g., the coercive cycle; Patterson, 2002) is related to children’s compromised self-regulation, whereas concordant positive behaviors such as parental support coupled with child compliance are related to better self-regulation (Dumas et al., 2001; Gardner, 1989; Lunkenheimer, Kemp, et al., 2017). Some patterns of discordance may be adaptive, for example, parents shifting into supportive behavior while the child is misbehaving in order to get the child back on task (Diercks et al., 2020; Lunkenheimer et al., 2016). However, other patterns of discordance may be maladaptive, for example, when parents engage in negative behaviors and young children take on the burden of repairing that conflict, which may be particularly taxing and detrimental for children (Lunkenheimer et al., 2019).

In early childhood, as children navigate new regulatory demands and attempt new behavioral strategies in response to parental directives (Eisenberg et al., 2010; Klein et al., 2018), individual differences in their temperamental self-regulation may shape parent-child behavioral synchrony. Effortful control (EC) reflects the child’s ability to inhibit a dominant response and perform a subdominant action (e.g., an action requested by caregivers), thus reflecting children’s regulatory capacity and the likelihood of compliance with parental agendas during face-to-face interaction. Differences in EC are related to differential evocation of positive and negative parent behaviors (Eisenberg et al., 2010) and to the emergence of differences in parent-child interaction patterns (Harris et al., 2007; Kochanska & Kim, 2014; Wilson & Durbin, 2012). Thus, when exploring how dynamic parent-child behavioral synchrony operates in early childhood, it may be important to consider how individual differences in children’s EC shape synchrony.

In the present study, we examined real-time dyadic behavioral concordance/discordance during challenging interactions and its moderation by child EC. Parent-child interactions involve frequent, dynamic fluctuations among behavioral states (Dishion et al., 2017). Real-time, microlevel analytic approaches can provide insight into how parents and children not only coordinate their behavior in the moment but how an individual’s changing behaviors affect the dyad (Bolger & Laurenceau, 2013). Thus, we applied state-trait multilevel modeling to examine real-time dyadic concordance/discordance in behavioral states, accounting for an individual’s average (reflecting baseline or “trait-like”) behavioral levels during the interaction. We also accounted for differences in parents’ positive versus negative behavioral expressions, given the differential potential for these behaviors to support or not support the child’s task-related efforts.

Parent-Child Behavioral Synchrony

Understanding child factors that shape parent-child interaction dynamics may be crucial in early childhood as children begin to take a more active role, gain internalized regulatory skills, and as interactions become increasingly bidirectional and complex (rapid, numerous behavioral shifts; Davis et al., 2017). One way to study these dynamics is through dyadic synchrony, which reflects how interactions are coordinated between partners in real-time. Feldman (2007) broadly defined dyadic synchrony as the “timed relationship, whether concurrent, sequential, or organized in an ongoing patterned format, between two or more events that cohere into a single process” (p. 329). Research on dyadic synchrony in early childhood may include concurrent behaviors measured at the same time or epoch and temporally sequential or contingent response patterns in a short time window (Lunkenheimer et al., 2013; Pasiak & Menna, 2015; Rocissano et al., 1987), considering that young children’s expressions of affect or behavior may not be immediate given their neurocognitive developmental status (Feldman, 2007).

Research examining sequential behaviors shows that contingent patterns of positive behaviors, such as parental autonomy support followed by child autonomous behavior, are associated with better outcomes such as fewer child behavior problems (Lunkenheimer, Ram, et al., 2017). For parent-child dyads with clinically-elevated aggression, greater synchrony of positive affect is also associated with children’s fewer behavior problems (Im-Bolter et al., 2015; Pasiak & Menna, 2015). Further, in families at risk for maltreatment and experiencing higher life stress, compromised parent-child behavioral synchrony is associated with children’s poorer self-regulation (Lunkenheimer et al., 2019; Diercks et al., 2020). These results suggest a meaningful role of parent-child behavioral synchrony in child development, and suggest that parent-child positive behavioral synchrony may be an important marker of developmental health in families at heightened risk.

To better understand the adaptive or maladaptive nature of behavioral concordance and discordance, we may need to pay close attention to the content, positive versus negative, of the interaction (Davis et al., 2017; Field et al., 1990). Positive behavioral concordance is often investigated as mutual cooperation, reciprocity, and harmonious interactions (Harrist & Waugh, 2002). Studies using global coding approaches have demonstrated in both higher-risk clinical and typically developing samples that more concordance around positive content is related to fewer child behavior problems (Deater-Deckard & Petrill, 2004; Leclère et al., 2014), increased compliance (Kochanska et al., 2008; Rocissano et al., 1987), and better self-regulation (Lindsey et al., 2009). Using a micro-dynamic analytic approach, Lunkenheimer et al. (2013) found that moment-to-moment coupling of parent directives and child compliance, as well as the coupling of parent teaching and child compliance, was related to better child self-regulation. Additionally, lower levels of synchrony around positive content have been linked to maladaptive outcomes such as higher behavior problems (Im-Bolter et al., 2015; Scholtes et al., 2020).

Comparatively, when parents and children display concordance around negative content, including parental harsh discipline, parental intrusion, child misbehavior, and noncompliance, it may be maladaptive. Dumas and colleagues (2001) found that in mother-child dyads with children at risk for behavior problems, dyads were more likely to engage in repeated, contingent real-time cycles of maternal control and noncompliance; other studies have also shown parent-child synchrony around negative content is related to children’s higher behavior problems (Smith et al., 2014). It may be that coordinated negative parent and child behaviors are representative of a coercive cycle, i.e., interactions involving increasingly adverse parent and child behaviors linked to the development of children’s behavior problems (Patterson, 2002). These mutually negative interactions may not provide the scaffolding children need to develop needed regulatory abilities and may lead to maladaptive outcomes.

The meaning of behavioral discordance, when changes in interaction partners’ behaviors are related but changing in opposite directions (e.g., one becomes more positive while the other becomes more negative), may be more ambiguous. Discordance is complicated by the question of which member of the dyad is negative and which is positive. When the parent’s behavior moves into positive states (e.g., autonomy support) while the child’s behavior shifts into negative states (e.g., noncompliance), this may be normative and adaptive (Larzelere et al., 2018) as child misbehavior during early childhood is typical (Wakschlag et al., 2007) and parental use of positive strategies to gain child compliance is associated with better child outcomes (Lunkenheimer et al., 2013). Conversely, if the parent displays negative behaviors while the child displays positive ones, this may indicate parents are not supporting their children or responding in expected ways. Previous research with both lower- and higher-risk dyads has linked greater discordance to children’s increased behavior problems (Kemp et al., 2016; Tronick & Beeghly, 2011), though questions remain about parent-led versus child-led discordance.

Behavioral Synchrony and Children’s Effortful Control

Children’s EC may be a salient temperament characteristic for parent-child behavioral synchrony. EC, a temperamental form of self-regulation, reflects the child’s ability to inhibit a prepotent response and engage in a subdominant response, for example, a response to a parental directive (Kochanska & Knaack, 2003; Rothbart et al., 2003). As children develop, EC may act as a promotive factor that is supportive of positive parent-child interaction patterns (Kiff et al., 2011). For example, parents may experience reduced stress during frustrating tasks and respond more positively to children if children with higher EC comply more easily with parental agendas or goals ( Karreman et al., 2006; Kim & Kochanska, 2012; Kochanska & Knaack, 2003; Tiberio et al., 2016). Additionally, children with higher EC may regulate their behavior more effectively and thus have an easier time completing difficult tasks; thus, parents may have to provide less scaffolding and exert less effort (Eisenberg et al., 2010). Research supports the notion that higher child EC generally benefits parent-child interactions (Lunkenheimer et al., 2011; Tiberio et al., 2016). For example, in dyads with typically developing children, higher child EC positively predicts mothers’ teaching statements (Eisenberg et al., 2010), whereas lower EC predicts mothers’ intrusive behaviors (Eisenberg et al., 2010). These results suggest that EC may serve as a promotive factor by evoking more positive parent behaviors, but questions remain about whether and how EC is a promotive factor for real-time parent-child concordance/discordance.

One possibility is that higher EC only serves as a promotive factor during concordance around positive content. For example, if children with higher EC are able to better regulate their frustration, they may not outwardly appear to be frustrated (Dennis et al., 2010) and parents may respond by engaging in positive, scaffolding behaviors that promote children’s problem-solving skills (Hammond et al., 2012). If children with poorer EC struggle to control frustration, parents may choose to engage in more directive and less autonomy-supportive behaviors with the goal of getting the task done quickly to resolve the frustration (Diercks et al., 2020). During moments of concordance around negative content, children may feel overwhelmed and may require assistance to regulate, or it may become increasingly taxing on their system as they attempt to regulate without parental guidance, especially in dyads with parents at higher risk for psychopathology or child maltreatment (Lunkenheimer et al., 2019). Therefore, investigating the relations between child EC and real-time dynamic behavioral concordance/discordance while considering differences by interaction content (i.e., positive versus negative) may provide greater insight into whether and when EC serves a promotive function in parent-child interactions.

Measuring Parent-Child Synchrony

Parent-child synchrony has been measured in multiple ways, including both global measures and microlevel dynamic methods. Dynamic methods have included the use of autoregressive integrated moving average modeling to extract concordance/discordance values and have demonstrated that more concordant interactions were predictive of children’s increased self-regulation abilities (Feldman et al., 1999). Lunkenheimer and colleagues (2017) used hidden Markov modeling to calculate real-time positive behavior coupling as transition probabilities of children’s autonomous responses to maternal autonomy support and mother’s supportive responses to autonomous behaviors in typically developing children. They showed that stronger coupling predicted lower levels of externalizing and internalizing behaviors, as well as lower levels of harsh parenting. These dynamic approaches have provided a more nuanced view of parent-child interactions and aided in elucidating the relations between dyadic behavioral concordance/discordance and children’s regulatory development.

One way to investigate real-time concordance/discordance is by employing multilevel intradyad dynamics modeling, also called state-trait multilevel modeling, with time-series data. This analysis allows researchers to examine how individuals’ behaviors within a dyadic interaction may be interrelated (intradyad coupling) as well as examine interdyadic (between-dyad) differences in the processes of interest (Bolger & Laurenceau, 2013). Although no known research has used this method to study real-time concordance of observed parent-child goal-directed behavior, we can draw upon other studies that have used this method to examine dyadic parent-child physiology. For example, previous research has used this methodology to demonstrate the presence of real-time dynamic concordance/discordance in mother and child respiratory sinus arrhythmia (RSA), where dyads showed between-dyad variations in RSA synchrony by individual and dyadic factors (Lunkenheimer et al., 2021).

Present Study

The aims of this study were to: (1) describe how real-time dynamic parent-child behavioral concordance and discordance manifests during a challenging task through the application of dyadic multilevel models to intensive time-series behavioral data; (2) investigate how children’s EC may influence real-time dynamic parent-child behavioral concordance and discordance in early childhood. Given the absence of prior work using this method to study parent-child behavioral concordance, the first aim was descriptive and exploratory. However, we did have an a priori expectation that mothers’ expressions of positive and negative behaviors could be differentially supportive and thus inform the adaptive vs. maladaptive nature of concordance and discordance; thus, mothers’ positive and negative behaviors were modeled separately. Using dyadic multilevel models, we were able to directly examine how a between-dyad factor such as EC moderated the within-dyad process of real-time dynamic behavioral concordance/discordance, hypothesizing that children higher in EC would experience more dynamic concordance with mothers in the context of mothers’ positive behaviors. Children’s EC may be particularly salient for positive behavioral concordance, given that children with higher EC may be better able to comply with parental goals and thus experience greater concordance in goal-directed tasks (Kim & Kochanska, 2012; Kochanska & Knaack, 2003; Tiberio et al., 2016), with lower EC compromising this concordance. For negative parent behavior, analyses were more exploratory. Lower EC could be linked with more concordant negative behaviors if parent-child interactions are more challenging when children have lower regulatory abilities; alternatively, lower EC may not be associated with concordance if it simply interferes with the overall process of parent and children coordinating their goal-directed behaviors.

Methods

Participants

Participating families were part of a larger study (N = 150, 80 female children) on parent-child coregulation. Families were oversampled for lower-income, higher parent stress, and child maltreatment risk and were recruited through community agencies, preschools, and Child Protective Services. They were assessed on three occasions when the child was age 2½ years, 3 years, and 4 years. Before the initial visit, families were screened for eligibility based on income (less than 200% of the federal poverty level), government assistance (e.g., food stamps), self-reported life stress (5 or more major life changes in the past year on the Life Stress Inventory; Holmes & Rahe, 1967), and/or CPS involvement, as well as lack of physical or mental health disorder diagnosis or cardiac problems that could interfere with the collection or interpretation of heart rate data. The present analysis makes use of data from 118 dyads that completed the dyadic interaction and EC tasks at Age 3. Little’s Missing Completely at Random test (Little & Rubin, 1989) revealed data was missing at random, χ²(6, n = 118) = 8.55, p = 0.20. Data from the larger study has been used in previous publications and presented at conferences.

Families were sociodemographically representative of a medium-sized Western university town. Mothers’ self-reported race/ethnicity was 73% Caucasian, 14% Latinx, 3% Multi-Racial, 3% African American, 3% Native American, 1% Asian, and 3% unknown. Children’s maternal-reported race/ethnicity was 64% Caucasian, 22% Latinx, 7% Multi-Racial, 3% African American, 1% Native American, and 3% unknown. At the age 2 ½ visit, the majority of parents were married (66.7%), with 12.7% living together, 11.3% single, 8.7% separated or divorced, and 0.6% unknown/unreported. The annual family income ranged from less than $5,000 to over $90,000 (average of ‘$30,000 to $39,000’), and maternal education ranged from junior high school to graduate level (median of Associate’s degree).

Procedures

At the first assessment (Age 2 ½), researchers visited participants in their homes, where mothers provided consent for themselves and their children and completed surveys. At the second (Age 3) and third (Age 4) assessments, participants came to the lab and completed a 2-hour session that included several child and dyadic tasks. This included an EC task battery conducted while mothers were in an adjacent room completing surveys about parenting, child behavior, and family characteristics. It also included a mother-child problem-solving puzzle task that was above the child’s cognitive ability level in order to win a prize. All procedures were approved by the Penn State IRB, #STUDY00009005. This study was not preregistered; the analysis code may be available on request to the first author. The data are not publicly available.

Parent-Child Challenge Task (PCCT).

Dyads completed the PCCT (Lunkenheimer, Kemp, et al., 2017), a 10-minute task with baseline, challenge, and recovery conditions designed to assess interaction patterns during a challenging problem-solving task. Mothers and children were asked to complete three 3-D wooden puzzles for the child to win a prize. The puzzles, which were of increasing difficulty and beyond the child’s cognitive ability level, were selected so that completion required guidance from the parent. Mothers were asked to use only their words and not to physically handle the puzzle pieces. During the baseline condition, mothers and children worked at their leisure for four minutes. Then, the experimenter briefly entered the room and stated that the dyad only had two minutes to complete the task. During this challenge phase, mothers continued to help their children (with words only). After three minutes, the experimenter entered the room, gave the child a prize (regardless of how many puzzles the dyad finished), and asked the parent and child to play with the prize (this recovery condition lasted three more minutes). The prize was either a coloring book with markers or play dough. The PCCT was videotaped using Noldus Observer 10.0 for later coding and analysis.

Observational Coding.

Age 3 PCCT videos were subsequently coded offline by trained and reliable graduate and undergraduate student coders using a validated coding system for parent-child interactions (Lunkenheimer, 2009). Behaviors were coded on a continuous, second-by-second time scale. All coders were trained until they reached at least 70% agreement in content, duration, and timing (average = 74%; range = 70–78%). Reliability estimates based on 20% of the total number of videos coded were calculated for content, timing, and duration of behavior across the entire task using the standard 3-second window in Noldus Observer.

Measures

Parenting behavior.

Parent behavior was coded continuously such that its presence and timing were represented by nine mutually exclusive codes: positive reinforcement - praise, e.g., “Good job,” proactive structure – effortful, child-centered attempts to keep the child on task, teaching statements - statements providing instruction and explaining the task, directive statements - direct commands, e.g., “Place the red block there,” engagement - watching and attending to the child and task without offering specific direction, disengagement - ignoring the child and task, negative discipline - warnings, threats, or directives with negative consequences, intrusion - physically taking over the task for the child, and emotional support – supporting children’s emotional expression. Because few mothers engaged in emotional support during the task, this code was dropped and treated as missing data.

Parent behavior was put on a quasi-ordinal scale from negative to positive and recoded as numeric, ranging from −3 to 4, with −3 = negative discipline, −2 = disengagement, −1 = intrusion, 1 = engagement, 2 = directives, 3 = teaching, 4 = proactive structure, and 4 = positive reinforcement. Proactive structure and positive reinforcement were given the same value of 4 to indicate that they are both highly positive behaviors. Parent behaviors were also separated into negative (values below 0) and positive behaviors (values above 0) for subsequent analysis.

Child behavior.

Child behavior was coded continuously with seven mutually exclusive codes: behavioral dysregulation - temper tantrums, disengagement – undirected, disengaged off-task behavior, noncompliance - not complying with a parental request, solitary play - off-task playing, social conversation – off-task conversation with the parent, compliance - complying with a parental request, and persistence – focused effort on the task without parental prompting. Child behaviors were placed on a quasi-ordinal scale from negative to positive and recoded as numeric. Behaviors ranged from −3 to 4 with −3 = behavioral dysregulation, −2 = disengagement, −1 = noncompliance, 1 = solitary play, 1= social conversation, 4 = compliance, and 4 = persistence. This coding was specified such that parent and child behaviors that were of similar valence would be numerically matched. For example, parent proactive structure and child persistence were coded as 4 to be aligned as two of the most positive behaviors. Additionally, child persistence and compliance were coded at the same value because they were positive, adaptive responses to parental behaviors (e.g., directives, teaching) that could be related to the parental behavior expressed, and children alternated among these two behaviors frequently. For example, when children were compliant in response to parents and then showed continued persistence at the task in the absence of parental prompts, we did not want those shifts to reflect discordance, given that the child was still responding appropriately to the parent.

Effortful control (EC).

EC was assessed at Age 3 using two tasks from an established battery (Kochanska et al., 1996). The Tower Task and Gift Delay task were introduced as rule-based games (rules were repeated about halfway through each task). In the Tower Task, children were asked to place blocks on a tower one at a time, alternating turns with the experimenter. A task score was calculated as the proportion of turns where the child correctly waited for the experimenter to place their block before placing their own.

In the Gift Delay, children faced the opposite direction and asked not to look until the experimenter finished wrapping a gift for them. After wrapping, it was placed in front of them. The experimenter then left to get a bow and asked the child to wait without touching the gift. The experimenter recorded the degree and latency of peeking and touching for the wrapping and waiting phases. For the wrapping phase, the degree of peeking was coded into five categories; 1 = child turns around, doesn’t return fully forward, 2 = turns around but turns back forward, 3 = peeks over shoulder far enough to see wrapping, 4 = turns head to the side but less than 90 degrees, and 5 = does not try to peek. For the waiting phase, children’s behavior was coded for the degree of touches and remaining seated. Child touches were coded into four categories: 1 = opens gift, 2 = lifts/picks up gift, 3 = touches but doesn’t lift gift up, and 4 = never touches gift. Child seating was coded in four categories: 1 = child is in seat for a total time of less than 30 sec, 2 = is in seat 30 sec or more but less than 1 min, 3 = is in seat 1 min or more but less than 2 min, and 4 = is in seat more than 2 min. The average latency to peek, average latency to touch, and degree of peeks and touches, were standardized and averaged to form within-task totals (i.e., a total gift delay score and a total tower task score). To align with previous literature and the original battery, within-task totals were then averaged to compute a total EC score (Kochanska et al., 1996). Scores on the two tasks were internally consistent (Cronbach’s alpha ranging from 0.71–0.81) and they were marginally significantly correlated, r = 0.20, p < 0.10.

Data Analysis

Real-time dyadic behavioral concordance/discordance and its relation to between-child differences in overall task behavior and EC were examined using multilevel models that accommodated the nested nature of the data (seconds nested within dyad) and provided for operationalization of both within-dyad and between-dyad associations. As part of data preparation, the time-varying predictor variables were separated into trait-like (within-person task average) and state components (deviations from the within-person task average within each time unit; Bolger & Laurenceau, 2013). Children’s behavior was separated such that their trait-like behavior was quantified as the within-child average behavior or the average of all behaviors across the interaction. Children’s state behavior was then quantified as deviation from this within-child average within each unit of time. This separation reduces bias created by the effects of the average level of behavior of a member of the dyad (Bolger & Laurenceau, 2013). For instance, if the mother moves into more positive behavior and the child did not shift into a more positive behavior but was, on average, more positive, the dyad may seem more coordinated since the child was already more positive and did not have to change to demonstrate concordance.

Differences in dyadic behavioral concordance/discordance and their relations to children’s trait-like behavior and EC were operationalized and examined using the model

$$PBstat{e}_{it}={\beta }_{0i}+{\beta }_{1i}CBstat{e}_{it}+e_{it}$$

where parental behavior at segment t in dyad i (PB_it) is a function of a dyad-specific intercept (β_0i ), a dyad-specific slope indicating the extent of behavioral concordance/discordance between the child’s state behavior and the parent’s state behavior (β_1i), and residual error (ε_it). In turn, the dyad-specific intercepts and slopes were modeled as

$${\beta }_{0i}={\gamma }_{00}+{\gamma }_{01}MomEd{u}_i+{\gamma }_{02}CBtrai{t}_i+{\gamma }_{03}E{C}_i+{\gamma }_{04}CBtrai{t}_i*E{C}_i+u_{0i}$$

$${\beta }_{1i}={\gamma }_{10}+{\gamma }_{11}MomEd{u}_i+{\gamma }_{12}CBtrai{t}_i+{\gamma }_{13}E{C}_i+{\gamma }_{14}CBtrai{t}_i*E{C}_i+u_{1i}$$

where γ₀₀ is an overall (sample-level) intercept, γ₀₁ indicates how maternal education level (a covariate) is related to overall level of parent behavior; γ₀₂ indicates how between-dyad differences in children’s trait-like behavior are related to differences in overall level of Parent Behavior; γ₀₃ indicates how between-child differences in EC are related to differences in overall level of parent behavior; γ₁₀ is the level of behavioral concordance/discordance for the prototypical dyad in the sample; γ₁₁ indicates how maternal education level (a covariate) is related to behavioral concordance/discordance; γ₁₂ indicates how between-dyad differences in children’s trait-like behavior are related to differences in behavioral concordance/discordance; γ₁₃ indicates how between-child differences in EC are related to differences in behavioral concordance/discordance, and u_0i and u_1i are unexplained between-dyad differences that are assumed multivariate normally distributed with standard deviations σ_u0 and σ_u1 and correlation r_u0u1. Of particular interest is the γ₁₀ parameter, which operationalizes bidirectional parent-child relations (as opposed to unidirectional from parent to child or vice versa) along the concordance/discordance continuum for the prototypical dyad. A positive γ₁₀ was interpreted as an indication of concordance, where mothers and children were changing in the same direction (e.g., both members becoming more positive or more negative). A negative γ₁₀ was interpreted as an indication of discordance, where mothers and children were changing in opposite directions (e.g., one member becoming more positive while the other became less positive).

Behavioral concordance/discordance of mothers’ positive versus negative behaviors was examined in separate models. The positive parent behavior model examined concordance/discordance between positive parent behaviors and child behaviors while excluding any moments of negative parent behavior. The negative parent behavior model examined concordance/discordance between negative parent behavior and child behavior, excluding any moments of positive parent behavior. All 118 mothers displayed positive parenting behaviors. However, not all mothers displayed negative parent behaviors. Thus, the model of negative parent behavior only includes the n = 81 dyads where mothers engaged in at least one negative parent behavior. The lmer package in R was used for all models (Bates et al., 2015).

Results

Preliminary Analysis

Descriptive statistics and correlations are shown in Table 1. Between-child differences in EC and child trait-like behavior during the PCCT were normally distributed. Sociodemographic factors were explored for potential inclusion as covariates. Maternal education was related to EC, r = 0.26, p = 0.01, and was thus included as a covariate in subsequent analyses. Child trait-like behavior was unrelated to child sex, r = −0.11, p = 0.24, annual family income, r = −0.02, p = 0.87, and maternal race/ethnicity, r = −0.02, p = 0.82. Child EC was also unrelated to child sex, r = −0.15, p = 0.16, annual family income, r = 0.19, p = 0.08 and maternal race/ethnicity, r = 0.15, p = .16. Figures 1 and 2 demonstrate descriptive illustrations of the present real-time dyadic data. Table 2 presents the average frequencies and duration of each parent and child behavior.

Table 1.

Means, standard deviations, and correlations with confidence intervals

Variable	M	SD	1	2	3
1. Maternal Education	5.97	1.47
2. Effortful Control	0.31	0.40	.26* [.06, .45]
3. Child Trait-like Behavior	2.84	0.87	−.02 [−.20, .16]	.24* [.03, .43]
4. Parent Trait-like Behavior	1.85	0.25	.13 [−.05, .31]	.20 [−.01, .39]	.55** [.41, .66]

Note. M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval for each correlation. The confidence interval is a plausible range of population correlations that could have caused the sample correlation (Cumming, 2014).

^*indicates p < .05.

^**indicates p < .01.

Figure 1.

Example of moment-to-moment parent-child dyadic behavior across the entire interaction.

Figure 2.

All seconds for all 118 dyads are included in this graphic, with each dot representing a second in the interaction. Note. Green box = Parent Positive, Child positive (Positive Synchrony) Red box = Parent Negative, Child Negative (Positive Synchrony). Blue box = Parent Positive, Child Negative (Negative Synchrony). Orange box = Parent Negative, Child Positive (Negative Synchrony)

Table 2.

Average frequency and duration per instance of parent and child behaviors

Parent Behavior	ND	Dis	Int	Eng	Dir	Teach	PS	PR
Average Frequency	3.96	2.25	4.39	30.47	31.58	19.60	4.93	13.58
Average Duration Per Instance (seconds)	3.14	12.00	2.76	9.56	5.16	4.25	4.97	2.31
n	24	16	66	118	118	118	106	115
Child Behavior	BD	Dis	NC	SP	SC	Com	Per
Average Frequency	3.19	5.07	4.82	5.62	6.92	20.49	16.92
Average Duration Per Instance (seconds)	8.01	10.29	5.69	15.69	12.98	12.87	10.40
n	16	68	92	84	108	118	118

Note. ND = Negative Discipline. Dis = Disengagement. Int = Intrusion. Eng = Engagement. Dir = Directive. Teach = Teaching. PS = Proactive Structure. PR = Positive Reinforcement. BD = Behavioral Dysregulation. Dis = Disengagement. NC = Noncompliance. SP = Solitary Play. SC = Social Conversation. Com = Compliance. Per = Persistence.

Concordance/Discordance of Child Behavior and Positive Parent Behavior

There was evidence of overall concordance of child behavior and positive parent behaviors, γ₁₀ = 0.12, p < 0.001, see Table 3. In the sample, shifts in children’s state behaviors (i.e., deviations from their trait-like behavior) were positively associated with shifts in parent positive behaviors. As shown in Figure 3a, parents’ and children’s behavioral shifts were coordinated such that shifts toward more positive or less positive behaviors were coordinated in real-time. Differences in children’s trait-like behavior (i.e., average behavior level across the entire interaction) was also associated with differences in parents’ overall average positive behavior, γ₀₂ = 0.21, p < 0.001. Children whose trait-like behavior was more positive had mothers with higher overall level of positive behaviors (and children whose trait-like behavior was less positive had mothers with lower overall level of positive behaviors).

Table 3.

Behavioral Synchrony Models

	Positive Parent Behavior			Negative Parent Behavior
	Estimate	CI	p–value	Estimate	CI	p–value
Fixed Effects
Intercept, γ00	1.59***	(1.39, 1.79)	<0.001	−2.12***	(−2.91, −1.34)	<0.001
Maternal Education, γ01	0.04*	(0.005, 0.74)	0.02	0.10	(−0.04, 0.23)	0.16
Child State Behavior, γ10	0.12***	(0.10, 0.16)	<0.001	−0.004	(−0.07, 0.07)	0.91
Child Trait-like Behavior, γ02	0.21***	(0.12, 0.30)	<0.001	0.61**	(0.24, 0.98)	0.002
Effortful Control, γ03	0.02	(−0.11, 0.13)	0.86	0.08	(−0.45, 0.61)	0.76
Child State x Effortful Control, γ13	0.06*	(0.001, 0.11)	0.04	0.25**	(0.10, 0.41)	0.001
Child Trait-like x Effortful Control, γ14	−0.01	(−0.21, 0.18)	0.93	−0.81	(−1.66, 0.05)	0.06
Random Effects	Estimate	CI	Corr	Estimate	CI	Corr
Intercept, σu0	0.21	(0.18, 0.24)		0.67	(0.54, 0.83)
Child State Behavior, σu0	0.10	(0.08, 0.11)	ru0u1 = −0.30	0.16	(0.11, 0.22)	ru0u1 = 0.55
Residual, σe	0.93	(−0.47, −0.09)		0.43	(0.21, 0.76)

Note. Estimate = unstandardized beta coefficient; 51,201 observations nested within 118 dyads for positive parenting behavior and a total of 1,056 observations across 81 dyads for negative parenting behaviors.

^***p < 0.001

^**p < 0.01

^*p < 0.05

Figure 3a.

Within dyad synchrony of positive parent behavior and child state behavior for a prototypical dyad.

Note. PS = Proactive Structure. PR = Positive Reinforcement. Teach = Teaching. Dir = Directive. Eng = Engagement. Int = Intrusion. ND = Negative Discipline. Dis = Disengagement. BD = Behavioral Dysregulation. Dis = Disengagement. NC = Noncompliance. SP = Solitary Play. SC = Social Conversation. Com = Compliance. Per = Persistence.

As hypothesized, differences in child EC were associated with differences in parent-child behavioral concordance, γ₁₃ = 0.06, p = 0.04. As shown in Figure 3b, children with higher EC exhibited stronger concordance with their mothers’ positive behavior (i.e., coordinated changes in the same direction). Simple slopes analysis revealed that EC was a significant moderator at lower EC (− 1 SD), b = 0.13 (0.02), p = <0.001, mean EC, b = 0.15 (0.01), p < 0.001, and higher EC (+ 1 SD), b = 0.17 (0.02), p = <0.001. The association between children’s trait-like behavior and positive parent behavior was not moderated by EC, γ₁₄ = −0.01, p = 0.93.

Figure 3b.

The effect of children’s effortful control on parent-child synchrony of positive parent behavior.

Note. PS = Proactive Structure. PR = Positive Reinforcement. Teach = Teaching. Dir = Directive. Eng = Engagement. Int = Intrusion. ND = Negative Discipline. Dis = Disengagement. BD = Behavioral Dysregulation. Dis = Disengagement. NC = Noncompliance. SP = Solitary Play. SC = Social Conversation. Com = Compliance. Per = Persistence.

Concordance/Discordance of Child Behavior and Negative Parent Behavior

There was no evidence of overall concordance or discordance of child behavior and negative parent behaviors, γ₁₀ = −0.004, p = 0.91, see Table 3. For the prototypical parent-child dyad, changes in the child’s moment-to-moment state behaviors were unrelated to changes in the parent’s negative behavior. However, differences in children’s trait-like behavior were related to differences in mothers’ overall level of negative parent behavior, γ₀₂ = 0.61, p = 0.002. Children with less positive trait-like behavior had mothers with higher overall negative behavior.

Differences in EC were related to the concordance/discordance of negative parent behavior and child behavior, γ₁₃ = 0.25, p = 0.001. As shown in Figure 3c, at average and high levels of EC, when children’s state behavior was more negative, mothers’ behavior was, on average, more negative or vice versa. Simple slopes testing supported that the association between children’s state behavior and negative parent behaviors was only significant at mean EC, b = 0.07 (0.03), p = 0.01, and higher EC, b = 0.17 (0.04), p < 0.001, but was unrelated at lower levels of EC. EC did not significantly moderate the association between children’s trait-like behavior and parents’ overall level of negative behaviors, γ₁₄ = −0.81, p = 0.06.

Figure 3c.

The effect of effortful control on the relation between children’s state behavior and parent’s negative behavior. Simple slopes test revealed that the relation between children’s state behavior and mothers’ behavior was only significant at high and average levels of EC.

Note: Int = Intrusion. ND = Negative Discipline. Dis = Disengagement. BD = Behavioral Dysregulation. Dis = Disengagement. NC = Noncompliance. SP = Solitary Play. SC = Social Conversation. Com = Compliance. Per = Persistence.

Discussion

The present study investigated complex patterns of dynamic parent-child concordance and discordance in goal-directed behavior and how individual child characteristics influenced these processes using an intradyad dynamics (or state-trait) multilevel model. This analytic approach allowed us to gain novel insights into complex parent-child interaction dynamics during early childhood, when behavioral states are rapidly changing and children increasingly bring their characteristics to bear on interaction dynamics (Eisenberg et al., 2010; Harrist & Waugh, 2002; Klein et al., 2018). Results demonstrated that dynamic concordance was observed during positive but not negative maternal behavior, suggesting mothers’ positive behaviors may be more likely than negative behaviors to facilitate real-time concordance of goal-directed behavior (e.g., discipline and compliance) with children. Additionally, higher child EC was related to greater real-time concordance regardless of whether mothers were engaged in positive or negative behavior. This finding aligns with prior research suggesting better EC facilitates coordination around positive behaviors by allowing children to more easily comply with parental goals (Karreman et al., 2006), and extends it to negative behaviors, suggesting children’s higher EC may foster greater concordance with mothers regardless of how the parent is behaving.

Positive Behaviors Facilitate Concordance

Findings of dynamic concordance around mothers’ positive behaviors are in line with previous research suggesting that parent-child synchrony of positive behavior is present in early childhood and may be adaptive for child outcomes (Field et al., 1990; Lindsey et al., 2009). Positive behaviors may facilitate synchrony through positive reinforcement of desired behaviors, broadening emotional thought-action repertoires with which to coordinate with the interaction partner (Fredrickson, 2004), and begetting positive emotions and warmth that support secure attachment (Dozier et al., 2018). In contrast, we did not find dynamic concordance overall with respect to mothers’ expression of negative behaviors. Negative behaviors may disrupt synchrony; for example, negative parenting is associated with mothers’ indiscriminate responses to children, and this reduced predictability may hinder behavioral concordance (Oldershaw et al., 1986). Though mutual negative behaviors have been demonstrated in higher-risk or clinical populations (Dumas et al., 2001), it is possible that the laboratory setting (Hendriks et al., 2018) precluded the opportunity to observe sufficient levels of negativity for these analyses. Future research should continue to explore differences between observed negative and positive behaviors and the potential disruptive effects of negative behaviors at various levels of familial risk.

Intradyad Modeling of Parent-Child Concordance

The ability to examine both within- and between-dyad effects using the novel method of intradyad modeling provided a more comprehensive picture of dyadic parent-child synchrony processes compared to overarching summary indicators of interaction dynamics such as global codes (Brinberg et al., 2018). For example, summarizing a time series into a single value to examine between-dyad differences in interaction patterns may obscure data on within-dyad dynamic variation and the potential effects of individual differences on within-dyad dynamics (Brinberg et al., 2018). This understanding of real-time patterns allows for a more complete picture of how parent-child interactions shape child development (Davis et al., 2017). Further, by including moderators, this study provides insights into potentially salient child characteristics that shape parent-child interactions. For example, knowing that EC plays a meaningful role in parent-child interaction dynamics, there may be unique strategies that parents or practitioners can take for children with lower EC to boost coordination of parent and child behavior.

Children’s Effortful Control Moderates Parent-Child Behavioral Concordance

We found that parent-child concordance was influenced by EC such that children with higher EC experienced greater dynamic concordance. This aligns with research demonstrating that EC shapes parent-child interactions (Kochanska & Kim, 2014). It may be that children with higher EC not only found it easier to comply with parental demands (Karreman et al., 2006), but that their greater self-regulation allowed the dyad to better coordinate their behavior as children were less dysregulated by the task. These results begin to demonstrate that higher EC may act as a promotive factor not just for longitudinal effects of parent-child interaction (Eisenberg et al., 2010) but also in real-time parent-child interaction dynamics. If children with higher EC experience more concordance around positive behaviors, they may receive increased parental scaffolding, thus fostering more adaptive long-term trajectories of regulatory development.

Surprisingly, although dyads did not display overall dynamic concordance when mothers engaged in negative behaviors, we did find dynamic concordance around negative parent behaviors only when children’s EC was average or high. Thus, children’s higher EC facilitated greater behavioral concordance with mothers regardless of the specific content or valence of parental behavior. EC may facilitate greater concordance overall given that children with higher EC are more easily able to comply with parental demands (Karreman et al., 2006). Although the potential risks of greater concordance around negative behavior should be considered, in the current study, we observed mostly positive and neutral behaviors with occasional negative behaviors, which is typical for a community sample in this age range (Lunkenheimer, Kemp, et al., 2017). Thus, it may be that children with greater self-regulatory capacities were less overwhelmed by occasional parental displays of negative behavior, and the dyad was able to maintain concordance through brief negative episodes and repair quickly to a mutually positive state (Kemp et al., 2016). Conversely, children with lower EC may be more dysregulated by mothers’ expressions of negative behaviors, especially during challenging tasks (Kochanska & Knaack, 2003), which may contribute to lower concordance.

A strength of the present study is that we oversampled for families with lower income, higher stress, and higher child maltreatment risk, factors that can pose challenges for parenting behavior (de Maat et al., 2021; Keim et al., 2011) and child self-regulation (Lunkenheimer, Ram, et al., 2017) and have been shown to alter parent-child coregulation patterns (Diercks et al., 2020). Thus, findings suggest EC is promotive of parent-child behavioral concordance in at-risk community populations, adding to a body of literature on the importance of child self-regulation as a key mechanism of adaptation in the face of risk (Tiberio et al., 2016), as well as specific studies on its role in relation to harsh parenting (Dumas et al., 2001; Lunkenheimer, Ram, et al., 2017; Patterson, 2002). The effects of children’s individual differences in EC on dyadic processes may vary by the level and nature of familial, parent, or child risk present; accordingly, future research will be needed to explore whether these relations generalize to other populations, for example, maltreating families or parents and children with diagnosed clinical disorders.

Limitations and Future Directions

Although this study has numerous strengths, including a novel analytic application and rigorous exploration of real-time interaction dynamics, there are limitations to consider. The study sample was comprised of mostly lower-income families at heightened risk for child maltreatment, which may limit generalizability. Approximately one-third of children in the sample were of ethnic minority background; therefore, additional racial, ethnic, or cultural diversity in future work could strengthen the generalizability of the findings and allow for a greater understanding of the role of cultural context in parent-child interaction dynamics.

Additionally, in this study, there was a lower occurrence of negative behaviors compared to positive behaviors, thus limiting the analytic power to investigate negative behaviors. Previous research has demonstrated that parents are less likely to display negative behaviors in the laboratory setting (Hendriks et al., 2018). However, given that parents often do not express negative affect in lab environments (Hendriks et al., 2018; Lunkenheimer, Kemp, et al., 2017), when they do, it may be extremely meaningful given that children have trouble regulating themselves in the presence of parental negativity (Cummings et al., 1985). Additionally, it is important to note that in the current study, dyads only needed to display one occurrence of negative behavior to be included in the analysis. Thus, we were unable to examine the effects of multiple repeated negative parent behaviors over the course of interaction. Given the weighty role of parents’ negative behaviors in child development, future research should examine parent-child concordance/discordance with higher rates of negative parent behaviors.

Lastly, this analysis was cross-sectional; as such, we are limited in understanding the directionality of effects between child EC and parent-child behavioral synchrony. Given previous research demonstrating the potential longitudinal effects of EC on parent behaviors (Eisenberg et al., 2010), future research could look at these factors longitudinally to see if EC predicts later synchrony. This could aid in understanding how concordance/discordance and EC work together to either promote or attenuate the adaptive development of children’s self-regulation.

Overall, this study provides an increased understanding of real-time behavioral synchrony and identifies EC as a salient child self-regulation factor influencing parent-child synchrony of goal-directed behaviors. By providing a more comprehensive understanding of influences on real-time dynamics, researchers may be able to strengthen intervention programs already targeting supportive parental responses to children (e.g., Dozier et al., 2018).