Effortful Control and Health among Triads of Mothers and Twin Children: An Actor-Partner-Interdependence Modeling Approach

Abstract

Associations between effortful control and physical and mental health were examined among triads of mothers and twin children by simultaneously modeling the effects of one’s own effortful control on one’s own health (actor effects) with the effects of the other two family members’ effortful control on one’s health (partner effects). Families (N = 761 individuals; 254 families) included mothers (M_age = 39.98, SD = 5.64) and their twin children (M_age = 8.58, SD = 0.57; 51% female; 54.7% non-Hispanic White, 28% Hispanic). Mothers completed online and in-person questionnaires, and mothers’ and children’s body composition indicators and dominant-hand grip strength were directly assessed during two home visits. Actor effects dominated in families with twin boys and mixed-sex twins, whereas partner effects were more apparent in families with twin girls. Additionally, the effects of children’s effortful control on mothers’ health depended on the sex of the twins. Findings uncovered important family dynamics linking effortful control to health, including that associations may vary by sex-match of children in the family.

Although almost 90% of children are rated as having very good or excellent overall health by their parents, 31% of 10- to 17-year-olds in the U.S. are rated as overweight/obese, and 37% of children have at least one physical or mental health condition (National Survey of Children’s Health [NSCH], 2017). The prevalence of children with a health problem affecting everyday life rose from about 2% in 1960 to greater than 8% in 2010, which includes rises in obesity and mental health problems (Perrin et al., 2014). Moreover, health problems run in families; only 65% of mothers perceive their own overall health as very good/excellent, more than 60% of U.S. adults are overweight/obese, and more than 8%of adults experience debilitating depression (Brody et al., 2018; NSCH, 2017; Wyatt et al., 2006). Identifying factors associated with physical and mental health in children and parents is important for prevention efforts.

Existing literature has tended to focus on evaluating parent- or child-related predictors of child health, without considering the simultaneous effects that may be at play between parents and children or among siblings (e.g., Clark et al., 2007; Graziano et al., 2013). The current study examined associations between family members’ effortful control (EC), a central aspect of temperament-based self-regulation, and their own and one another’s health. EC integrates abilities to focus and shift attention, inhibit behavior in favor of accomplishing a goal, and initiate and complete tasks proactively (Rothbart & Bates, 2006). Most work on EC reflects studies conducted among children with respect to their developmental functioning. In general, well-regulated children are more prosocial, compliant, and academically competent, whereas children low in EC are prone to psychological maladjustment and psychopathology (Eisenberg et al., 2016). Although the self-regulation literature has focused on weight-related health, and particularly associations with body mass index (BMI; e.g., Anderson & Whitaker, 2018; Graziano et al., 2013), other aspects of health have yet to be investigated.

Furthermore, the effects of EC on health among family members are unknown. Theory suggests that both an individual’s own self-regulation and the proximal environment that includes other family members affect health (Bronfenbrenner & Ceci, 1994; Rothbart & Bates, 2006). Individuals in the child’s most immediate environment, including parents and siblings, likely have a substantial impact on the child (and vice versa) even when taking the child’s (or others’) own temperament into consideration. Similarly, health may be dynamically affected both by an individual’s own EC, and by the EC of family members. Importantly, our inquiry involved testing understudied sibling influences and child influences on mothers, given that U.S. children are likelier to grow up with a sibling than a father in the home (McHale et al., 2013).

Effortful Control and Health: Dynamic Associations

To account for the unique role each member has within the group, we employed a novel approach to the Actor-Partner-Interdependence Model (APIM; Cook & Kenny, 2005) with triadic data from mothers and twin siblings. We simultaneously examined associations between mothers’ and each twin’s EC and four physical and mental health outcomes: general health status, body composition (i.e., a composite of BMI, waist circumference, body fat percentage), dominant-hand grip strength (indicator of muscular strength; Wind et al., 2010), and internalizing symptoms. We focused on the more commonly studied subjective indicators of physical (general health) and mental (internalizing symptoms) health, and strengthened the design by supplementing with objectively assessed height, weight, waist circumference, body fat percentage, as well as an index of muscle strength, grip strength. Although APIM analyses have traditionally explored dyadic relationships (Cook & Kenny, 2005), models that include more than two members fit the call for considering diverse interpersonal dynamics and novel approaches to this framework (Ledermann et al., 2017). Within our triadic model, actor effects capture how individuals’ characteristics or behaviors affect their own outcomes, whereas partner effects capture how individuals’ characteristics or behaviors affect others’ outcomes.

Actor effects: Associations linking EC to health within individuals.

Self-regulation, including EC, has been associated with chronic health problems, obesity, and mental health problems in children (e.g., Berg et al., 2014; Lemery-Chalfant et al., 2008). Longitudinally, facets of self-regulation in childhood, including inattention and executive functioning, have been associated with health in adulthood, such that poor self-regulation predicts increased rates of multiple health problems (e.g., cardiovascular, respiratory, metabolic; Moffitt et al., 2011; Schlam et al., 2013). Evidence involving EC is scant; however, one study found college students’ EC was positively associated with reported energy, overall health, and emotional wellbeing (Massey-Abernathy & Byrd-Craven, 2016). Associations may exist partly because EC exerts a protective influence on health behaviors. For example, self-regulation is important for adolescents’ adherence to Type 1 diabetes management (Berg et al., 2014). Additionally, the ability to delay gratification (a component of EC) among four-year-olds has predicted their lower BMI scores 30 years later (Schlam et al., 2013). Findings demonstrate potential long-term ramifications of EC for positive health outcomes through promoting self-regulation in childhood.

Most studies to date involve links between EC, or self-regulation more broadly, and weight-related health, and suggest that self-regulatory competence protects against obesity risk in childhood (Anzman-Frasca et al., 2012). Toddlers’ self-regulation has been related to BMI and pediatric obesity at 5.5 and 10 years of age (Graziano et al., 2013). Cross-sectionally, high levels of EC were associated with decreased odds of nutritional risk at three and five years of age (van den Heuvel et al., 2017). Among overweight/obese adults, low EC was linked to liking and wanting high-fat, sweet foods and trait disinhibition (e.g., disinhibited eating patterns; Mackey et al., 2019), suggesting that EC remains relevant for weight-related health behaviors among adults.

Supporting evidence is limited, but EC may also be important for other health facets, including grip strength, a prospective indicator of general, as well as cardiovascular, health and longevity in both adolescent males and adults (e.g., Leong et al., 2015; Ortega et al., 2012). Among adults, self-regulation traits, including high self-control, have been positively associated with grip strength (Sutin et al., 2018), suggesting a link between EC and physical strength. Muscle strength is critical to promoting physical health across the lifespan (Wolfe, 2006). Because methods of obtaining handgrip strength are inexpensive, portable, and easy to administer, grip strength has been widely used in samples of youth to assess upper body muscle strength. Importantly, evidence also suggests that handgrip strength serves as a reliable indicator of whole-body muscle strength in children and adolescents (rs > .70; Wind et al., 2010). We included grip strength as a health outcome, expanding exploration of the EC – grip strength association from samples of primarily adolescents and men to include women and children.

EC has been associated with mental health across the lifespan. In general, EC is negatively associated with psychopathology, likely because EC represents the efficiency of the executive attention neural network involved in resolving internal emotional and cognitive conflicts and re-orienting attention (Petersen & Posner, 2012). In middle childhood, adolescence, and adulthood, EC has been negatively associated with concurrent or later reports of depressive, anxiety, and internalizing symptoms (e.g., Clements & Bailey, 2010; Lemery-Chalfant et al., 2008; Nelson et al., 2018). Here, we examined associations between EC and internalizing symptoms, in addition to indicators of physical health.

Partner effects: Associations linking EC to health between individuals.

The EC of family members may also influence one’s health outcomes. Research in these areas is scant and primarily focused on parental influence on child health. Mothers’ and fathers’ self-regulation was important for glycemic control in their children with Type 1 diabetes (Healey et al., 2018). Parents’ EC was also positively associated with healthy food preferences among young children (Zhou et al., 2019). Parental EC may impact child health via influences on the child’s EC or the use of parenting practices and responses to child behavior (e.g., Valiente et al., 2007). Parental self-regulation predicts parenting attitudes and behaviors, including discipline strategies and emotional availability (e.g., Barros et al., 2015; Rea-Sandin et al., 2020). Parents must adapt to different parenting situations and regulate their reactions to their children’s emotions and behaviors (e.g., Barros et al., 2015). For example, parents may employ practices including harsh punishment and avoidance, which have been related to children’s behavioral, emotional, and health problems (e.g., Morawska & West, 2013).

Children’s EC also may be associated with their parents’ physical and mental health. Parental hostility was negatively associated with an indicator of physiological stress activity in parents (i.e., cortisol awakening response), but only for parents whose children were low in EC (Merwin et al., 2015). Children’s regulatory competence may elicit particular parenting practices and behaviors, which affect parents’ health. For example, children’s perceived behavioral regulation predicted father-child physical coactivity (Cook et al., 2018), with implications for fathers’ physical health. To date, however, the extent to which children’s EC is associated with parental health outcomes is unclear.

Finally, whether siblings’ EC affects one another’s health is unknown. Siblings may influence one another’s development, health, and behavior through their interactions, attachment to each other, modeling, or competition for resources (McHale et al., 2013). For example, sibling participation has been associated with increased physical activity among younger girls (Cleland et al., 2011). A sibling high in EC may model healthy behaviors including control of eating choices or persistence in challenging physical tasks. Despite theoretical reasons to anticipate these dynamic associations among siblings, they have not been empirically tested. It is imperative to understand the role of siblings’ EC on health within the family unit.

Effortful Control and Health: Differences based on Sibling Sex-Match

Differences between daughters and sons and having a sister versus a brother may affect the nature of associations linking EC to health within a family, because these characteristics alter family dynamics. A meta-analysis found that girls typically have higher EC than boys (Else-Quest et al., 2006). Sex differences in EC and self-regulation are evident in children as young as 36 months, with girls higher in inhibitory control and boys higher in activity level (Gagne et al., 2013). Sex differences in associations between self-regulation at age two and weight at age 5.5 indicate a linear association among boys, with the highest percentage of obesity among those with the lowest self-regulation scores (Anderson & Whitaker, 2018). Among girls, there was a curvilinear association, with the highest percentages of obesity among those in the lowest and highest quartiles of self-regulation scores; thus, under- and over-regulation may influence girls’ weight-related health. Additionally, greater attention-span persistence in infancy predicted a lower odds of obesity risk in childhood (at age 6) only for boys (Faith & Hittner, 2010). All told, findings suggest sex differences in EC and health associations among young children.

Given that girls have higher EC overall than boys, having a sister may be beneficial to one’s own health because, on average, sisters may be more likely to model healthier behavior and manage negative emotionality than brothers. Studies have shown girls have a protective influence over their sisters’ substance use/delinquent behavior (e.g., Samek et al., 2015), and also that younger sisters (as opposed to younger brothers) are more heavily influenced by their siblings’ behavior when it comes to interests in athletics (Whiteman et al., 2007). Because boys tend to be higher in activity level than girls, brothers’ EC may promote actively engaging in physical activity and trying new sports, which might affect a sibling’s (especially a sister’s) health behavior. We aimed to further elucidate sex differences in EC and health associations, including whether differences arise by having a sister versus a brother.

Current Study

This study extends prior research by using an APIM to simultaneously estimate associations between EC and health among triads of mothers and twin offspring. In a single model accounting for twins’ age, race/ethnicity, and socioeconomic status (SES), we investigated whether 1) mothers’ and their eight-year-old twins’ EC predicted their own and each other’s general health, body composition, grip strength, and internalizing symptoms (Figure 1) and 2) associations differed between families as a function of twins’ sex-match. Our comprehensive assessment included both reported and directly assessed measures of physical and mental health.

Figure 1.

Conceptual actor-partner-interdependence model depicting hypothesized paths linking mothers’ and children’s effortful control to health outcomes, accounting for child- and family-level covariates.

We hypothesized that high EC would have positive ramifications on one’s own health, and that these positive effects would extend to others within the family unit. Specifically, we hypothesized that 1) high individual EC would be associated with better health outcomes; better general health, stronger grip strength, lower body composition, and fewer internalizing symptoms (actor effects), 2) higher EC among mothers would be associated with better health among children (partner effects), 3) higher EC among children would be associated with better health among mothers (partner effects), 4) higher child EC would be associated with siblings’ healthy functioning (partner effects), and 5) there would be differences based on the sex of the twins, with more partner effects apparent among families with more girl twins (i.e., same-sex girl families [2 girls], then mixed-sex twin families [1], then same-sex boy families [0]).

Method

Participants

Participants included 254 mothers and their twin children (i.e., 90 girl, 85 boy, 79 mixed-sex twin pairs) from an ongoing longitudinal study (Lemery-Chalfant et al., 2019). The EC and health descriptive statistics of the twins are similar to other studies of singletons (El-Sheikh et al., 2014; Kotelnikova et al., 2017). Families (29% monozygotic [MZ], 37% same-sex dizygotic [DZ], 32% mixed-sex DZ twin pairs) were recruited from birth records. At the time of the current study, 273 families had completed the eight-year assessment. Sixteen families with non-mother primary caregivers and three families with missing data on all variables were excluded. One child was excluded due to severe mental and physical disabilities, but the mother’s and co-twin’s data were retained (N=761 individuals). Children (51% female; 54.7% non-Hispanic White, 28% Hispanic, 4.3% Black/African American, 3.9% Asian/Asian American, 2% Native American, 1.2% Native Hawaiian/Pacific Islander, 5.1% unknown) were approximately eight years of age (M=8.58, SD=0.57). Mothers (66.9% non-Hispanic White, 20.9% Hispanic, 3.1% Black/African American, 2.4% Asian/Asian American, 0.8% Native American, 0.8% Native Hawaiian/Pacific Islander, 5.1% unknown) ranged in age from 27 to 54 years (M=38.98, SD= 5.64). Based on an income-to-needs ratio, 9.3% of families were below the poverty line, 22.9% were near the poverty line, 15.6% were lower-middle class, and 52.2% were middle-to-upper class.

Procedure

Institutional Review Board approval was obtained prior to the start of the study. Informed consent and assent were obtained from mothers and children, respectively. Families participated in an intensive assessment involving online questionnaires and two home visits. Mothers filled out questionnaires on demographics, EC, general health, and internalizing symptoms. During home visits, scheduled about one week apart, trained research assistants assessed mothers’ and children’s height, weight, body fat percentage, waist circumference, and grip strength. Mothers were compensated for participating and children received a small toy.

Measures

Demographics.

Mothers reported on children’s sex, age, and race/ethnicity; household income (prior to taxes); family size; and their own and secondary caregivers’ (if applicable) educational attainment. Mothers could select more than one race/ethnicity. An income-to-needs ratio was calculated for each family based on household income and family size using 2017 (i.e., time of assessment) federal poverty guidelines (U.S. Department of Health and Human Services, 2017). We created a mean composite of SES from the standardized values of the income-to-needs ratio, mothers’ education, and secondary caregivers’ education.

Effortful control.

Mothers responded to items from the Adult Temperament Questionnaire (Evans & Rothbart, 2007) indicating their own EC, using the attentional focusing (7 items), activation control (5 items), and inhibitory control (7 items) scales to represent the EC factor (rs > .36, ps < .001; α = .78), with items rated from 1 (extremely false) to 5 (extremely true). In a separate assessment, mothers used the Temperament in Middle Childhood Questionnaire (Simonds et al., 2007) to report on children’s EC, using the attentional focusing (7 items), activation control (15 items), and inhibitory control (8 items) scales to represent the EC factor (rs > .44, ps < .001; α = .85), with items rated from 1 (almost always untrue of your child) to 5 (almost always true of your child). Higher scores indicate higher EC.

General health.

Mothers rated their own general physical health with the General Health scale of the Medical Outcomes Study 36-Item Short-Form Health Survey (Ware & Sherbourne, 1992) during a home visit. Based on scoring guidelines, we converted the 5-item scores to a 0–100 scale before computing the composite (α = .80). In a separate online assessment, mothers rated twins’ health using the Global Physical Health scale (5 items) of the Health and Behavior Questionnaire (HBQ; Armstrong et al., 2003). A mean composite indexed each child’s general health (α = .75). We used standardized scores, with higher scores indicating better health.

Body composition.

Mothers’ and children’s height in inches were measured twice with a tape measure to the nearest 1/16 inch (once per visit). Weight and body fat percentage were assessed three times at each home visit. Weight was measured using an FDA-approved full body composition scale for mothers (Tanita BF522-W Scale) and children (Tanita IronKids BF-2000 Scale) to the nearest 1/10 pound. BMI was calculated for each trial using the Center for Disease Control’s BMI formula (weight/[height]²×703). Body fat percentage was assessed using wireless Tanita full body composition scales that include bio-electrical impedance analysis and yield values to the nearest 1/10 percentage. BMI and body fat percentage were averaged across trials. Waist circumference was assessed using a spring-weighted Gulick tape measure to the nearest 1/16 inch (once per visit) at the natural waistline approximately two inches below the lowest rib (Davis, 2008). Because BMI, body fat percentage, and waist circumference were significantly correlated for mothers (rs > .85, ps < .001) and children (rs > .80, ps < .001), we computed mean composites with standardized values as an index of body composition. Higher scores indicate greater BMI, body fat percentage, and waist circumference.

Dominant-hand grip strength.

Mothers’ and children’s dominant-hand grip strength was assessed with a digital dynamometer that measures force in pounds (Jamar Plus+ Dynamometer). Across three trials (scores were averaged), mothers and children were instructed first to sit up straight with their dominant-hand arm bent at a 90-degree angle, and then to squeeze the dynamometer with full force. Body size is likely to affect grip strength in growing children, and BMI was significantly correlated with grip strength among children (r = .36, p < .001), but not mothers, in the current sample; therefore, we regressed children’s grip strength on their BMI and used the residual values as the index of each child’s dominant-hand-grip strength.

Internalizing symptoms.

Mothers’ internalizing symptoms were assessed using the 20-item Center for Epidemiologic Studies – Depression Scale (Radloff, 1997) and the 14-item anxiety scale of the Depression Anxiety Stress Scales (Lovibond & Lovibond, 1995). We computed a mean composite of mothers’ standardized depression and anxiety scale scores (r = .58, p < .001; α = .91). Children’s internalizing symptoms were assessed via mothers’ reports on the HBQ (Armstrong et al., 2003), using the depression (7 items), overanxious (13 items), and separation anxiety (10 items) scales, rated 1 (never or not true) to 3 (often or very true). A mean standardized composite of items from correlated scale scores (rs > .42, ps < .001) was used to index child internalizing symptoms (α = .85).

Analytic Strategy

Using Mplus 8 (Muthén & Muthén, 1998–2017), we estimated a multiple-group APIM accounting for children’s age and race/ethnicity and family SES, to test hypotheses of self-effects (actor), mother-child effects (partner), and sibling effects (partner) of EC on health outcomes, by twin-pair type (i.e., girl-girl, boy-boy, girl-boy). Specifically, distinguishable groups within the single model were based on the sex-match of the twins; therefore, the model simultaneously estimated effects for the three separate possible twin pairings. In same-sex twin pairs (girl-girl and boy-boy), a random number generator assigned one of the twins to the position of Twin 1, whereas in mixed-sex twin pairs (girl-boy), Twin 1 was always the boy. A dichotomous variable of child race/ethnicity was used in the analyses (0 = not non-Hispanic White, 1 = non-Hispanic White). To handle missing data, full information maximum likelihood estimation was used.

The nature of the family roles of mothers and twin siblings in this model warranted several constraints. A number of paths were equated within each group of same-sex twin families: 1) mothers’ effects of EC on children’s outcomes, 2) children’s actor effects, and 3) sibling effects. Mothers’ actor-effects were constrained to be equal across all three groups. To test for significant group differences, we compared Chi-squares for models with freed paths to models with fixed paths, for each path separately. When testing for differences involving the mixed-sex twins with another group, we conducted two separate comparisons: one in which the path in the other group was constrained to be the same as the path of the male twin and one where it was constrained to be the same as the female twin. Although the children are twins, this study focused on the phenotypic associations between EC and health among family members and did not focus on estimating heritability. Given that many parameters were estimated within one APIM analysis, we conducted a Monte Carlo simulation of 500 replications to examine post hoc predictive power. For all estimates, at least 90% of replications demonstrated that the 95% confidence interval contained the population parameter estimate (indicated by the 95% cover estimates), supporting high power for testing the model.

Results

Preliminary Analyses

Supplemental Table 1 contains descriptive statistics and zero-order correlations, computed separately for girl and boy children (regardless of twins’ sex-match). Mothers’ EC was positively related to their own general health and negatively related to their own body composition and internalizing symptoms. Girls’ EC was negatively associated with their own internalizing symptoms, whereas boys’ EC was associated with their own general health (positively), body composition (negatively), and internalizing symptoms (negatively). There were no significant correlations linking EC to grip strength. Notably, mothers’ EC was positively associated with girls’, but not boys’ general health, whereas mothers’ EC was negatively related to boys’, but not girls’, body composition. Mothers’ EC was negatively associated with girls’ and boys’ internalizing symptoms. Additionally, boys’ EC was correlated with their mothers’ general health (positively), body composition (negatively), and internalizing symptoms (negatively), but girls’ EC was only associated (negatively) with mothers’ internalizing symptoms.

To test zygosity as a covariate, we computed a one-group model including all families. Zygosity did not emerge as a significant predictor of child or mother health outcomes; therefore, we proceeded with the multi-group APIM structure focused on the sex-match of the twins.

Multiple Group Actor-Partner-Interdependence Model

Among 254 triads (761 individuals), a single multiple-group APIM estimated effects separately for 79 mixed-sex, 85 same-sex boy, and 90 same-sex girl twin families, accounting for children’s age, race/ethnicity, and family SES. The model demonstrated good fit: χ²(100, N=254) =118.60, p =.10; CFI =.98; RMSEA =.05; SRMR =.06. We report actor and partner effects from the single, comprehensive model by group in the following sections (Figures 2, 3, and 4). Although included in the model, path estimates from covariates to outcomes are not depicted in figures.

Figure 2.

Mixed-sex twins’ results only (79 families) from a comprehensive multiple-group actor-partner-interdependence model that included families with same-sex girl and same-sex boy twins. Unstandardized estimates are followed by standardized estimates, which are in parentheses. Solid lines represent significant paths, whereas dotted lines represent nonsignificant paths and dashed lines represent paths approaching significance. Additionally, paths regressing outcomes on covariates (i.e., child age, child race/ethnicity, family socioeconomic status) are not shown but are described in-text. The model fit the data well, χ²(100, N = 254) = 119.99, p = .08; CFI = .98; RMSEA = .05; SRMR = .06. Within-groups, all predictors were permitted to covary, and all outcomes were permitted to covary (not shown). Race/ethnicity was coded 0 = not non-Hispanic White, 1 = non-Hispanic White. EC = Effortful control, SES = Socioeconomic status. ^†p ≤ .10; *p ≤ .05; *p < .01.

Figure 3.

Same-sex boy twins’ results only (85 families) from a comprehensive multiple-group actor-partner-interdependence model that included families with same-sex girls and mixed-sex twins. Unstandardized estimates are followed by standardized estimates, which are in parentheses. Solid lines represent significant paths, whereas dotted lines represent nonsignificant paths and dashed lines represent paths approaching significance. Additionally, paths regressing outcomes on covariates (i.e., child age, child race/ethnicity, family socioeconomic status) are not shown but are described in-text. The model fit the data well, χ²(100, N = 254) = 119.99, p = .08; CFI = .98; RMSEA = .05; SRMR = .06. Within-groups, all predictors were permitted to covary, and all outcomes were permitted to covary (not shown). Race/ethnicity was coded 0 = not non-Hispanic White, 1 = non-Hispanic White. EC = Effortful control, SES = Socioeconomic status. ^†p ≤ .10; *p ≤ .05; *p < .01.

Figure 4.

Same-sex girl twins’ results only (90 families) from a comprehensive multiple-group actor-partner-interdependence model that included families with same-sex boy and mixed-sex twins. Unstandardized estimates are followed by standardized estimates, which are in parentheses. Solid lines represent significant paths, whereas dotted lines represent nonsignificant paths and dashed lines represent paths approaching significance. Additionally, paths regressing outcomes on covariates (i.e., child age, child race/ethnicity, family socioeconomic status) are not shown but are described in-text. The model fit the data well, χ²(100, N = 254) = 119.99, p = .08; CFI = .98; RMSEA = .05; SRMR = .06. Within-groups, all predictors were permitted to covary, and all outcomes were permitted to covary (not shown). Race/ethnicity was coded 0 = not non-Hispanic White, 1 = non-Hispanic White. EC = Effortful control, SES = Socioeconomic status. ^†p ≤ .10; *p ≤ .05; *p < .01.

Families with mixed-sex twins.

Although one multiple-group APIM was estimated, Figure 2 contains only results for mixed-sex twin families. In the model, Twin 1 among mixed-sex twins was the boy, and Twin 2 was the girl twin. There were several actor effects and no significant partner effects among mixed-sex twin families, and all significant findings were consistent with hypotheses. Mothers’ EC was positively associated with their own general health (b = 0.50, 95% CI [0.24, 0.76], p < .001) and negatively associated with their own internalizing symptoms (b = −0.49, 95% CI [−0.71, −0.28], p < .001), such that higher EC was associated with better general health and fewer internalizing symptoms. Boys’, but not girls’, EC was negatively associated with their own body composition (b = −0.32, 95% CI [−0.61, −0.03], p = .03), indicating higher reports of EC were related to lower body composition scores. Girls’, but not boys’, EC was positively associated with grip strength (b = 2.78, 95% CI [0.02, 5.53], p = .048). EC was negatively associated with one’s own internalizing symptoms for boys (b = −0.50, 95% CI [−0.89, −0.13], p = .01) and girls (b = −0.64, 95% CI [−1.16, −.11], p = .02).

Families with same-sex boy twins.

Figure 3 contains results for same-sex-boy twin families. Similar to families with mixed-sex twins, there were several actor effects and all significant findings were consistent with hypotheses. Mothers’ EC was positively associated with their own general health (b = 0.50, 95% CI [0.24, 0.76], p < .001) and negatively associated with their own internalizing symptoms (b = −0.49, 95% CI [−0.71, −0.28], p < .001). Boys’ EC was positively associated with their own general health (b = 0.33, 95% CI [0.06, 0.59], p = .02) and negatively associated with their own body composition (b = −0.37, 95% CI [−0.70, −0.04], p = .03), such that higher EC scores were associated with better general health and lower body composition. Mothers’ and siblings’ EC had no significant effect on same-sex boy twins’ health outcomes. Notably different from mixed-sex twin families, boys’ EC was negatively associated with mothers’ body composition (b = −0.27, 95% CI [−0.53, −0.004], p = .047).

Families with same-sex girl twins.

Figure 4 contains results for families with same-sex-girl twins only. In contrast to families with mixed-sex and same-sex boy twins, there were several partner effects in addition to the actor effects, and several findings were in unexpected directions. Consistent with hypotheses and the other groups, mothers’ EC was positively related to their own general health (b = 0.49, 95% CI [0.24, 0.76], p < .001) and negatively associated with their own internalizing symptoms (b = −0.49, 95% CI [−0.71, −0.28], p < .001). Same-sex girl twins’ EC was negatively associated with their own internalizing symptoms (b = −0.32, 95% CI [−0.63, −0.01], p = .046), with higher EC being related to fewer internalizing symptoms.

Among partner effects, mothers’ EC was positively associated with same-sex twin girls’ general health (b = 0.48, 95% CI [0.07, 0.88], p = .02); however, contrary to hypotheses, higher mothers’ EC was associated with lower grip strength (b = −2.82, 95% CI [−4.93, −0.72], p = .01) and higher body composition among twin girls (b = 0.45, 95% CI [0.03, 0.87], p = .04). Also, contrary to hypotheses, twin girls’ EC was positively associated with mothers’ body composition (b = 0.20, 95% CI [0.01, 0.39], p = .04), but, as expected, was negatively associated with mothers’ internalizing symptoms (b = −0.23, 95% CI [−0.43, −0.03], p = .02). Regarding sibling effects and consistent with predictions, same-sex sisters’ EC was negatively associated with each other’s internalizing symptoms (b = −0.53, 95% CI [−0.86, −0.20], p = .002), such that higher EC of the co-twin was related to lower internalizing symptoms in their sister.

Covariates.

SES was negatively associated with mothers’ body composition across all three groups: families with mixed-sex twins (b = −0.36, 95% CI [−0.58, −0.13], p = .002), same-sex boy twins (b = −0.17, 95% CI [−0.26, −0.09], p < .001), and same-sex girl twins (b = −0.17, 95% CI [−0.26, −0.09], p < .001). Children’s age was negatively associated with mothers’ grip strength in both same-sex boy and same-sex girl families (b = −3.56, 95% CI [6.93, −0.20], p = .04). Children’s age was associated with boys’ and girls’ grip strength (b = 4.05, 95% CI [2.04, 6.06], p < .001) and body composition (b = −0.36, 95% CI [−0.58, −0.13], p = .002) among families with mixed-sex twins. Race/ethnicity (b = 2.56, 95% CI [0.46, 4.66], p = .02) was positively associated with the twins’ grip strength among families with mixed-sex twins. Race/ethnicity was negatively associated with the twins’ body composition (b = −0.48, 95% CI [−0.89, −0.08], p = .02) among families with same-sex boys. SES was positively associated with grip strength among girls in families with same-sex girls (b = 1.12, 95% CI [0.30, 1.95], p = .01).

Path comparisons between groups.

To examine group differences for significant paths in more detail, we compared Chi-square fit statistics for freed-paths models versus models with paths constrained to be equal across relevant comparison groups. Regarding actor effects, the positive effect of children’s EC on their own general health was significantly stronger for same-sex boy twins versus for same-sex girl twins (Δχ²[1] = 5.44, p = .02). There were no significant differences between any of the groups on the effects of child EC on child internalizing symptoms. The effects of mothers’ EC on their own health outcomes were not examined because these paths were constrained to be equal across all groups in the model.

When examining partner effects, the effect of mothers’ EC on children’s grip strength was significantly different for same-sex girl twins versus boys in mixed-sex twins (Δχ²[1] =5.01, p = .03): Mothers’ EC was negatively associated with grip strength only among same-sex twin girls. The positive association between mothers’ EC and same-sex twin girls’ body composition was significantly different than the negative association between mothers’ EC and same-sex boy twins’ body composition (Δχ²[1] = 4.11, p = .04). The unexpected, positive association between same-sex girls’ EC and mothers’ body composition was significantly different than the expected, negative association between same-sex boys’ EC on mothers’ body composition (Δχ²[1] = 8.31, p = .004). The effect of siblings’ EC on children’s internalizing symptoms was significantly stronger for girls in same-sex than mixed-sex twin families (Δχ²[1] = 4.80, p = .03).

Discussion

We offer novel evidence that EC is associated with health across domains, that family members’ EC is associated with one another’s health, and that associations differ by the sex of the twins (i.e., girl-girl, boy-boy, girl-boy twins). The current study used an innovative APIM technique with triads to simultaneously model the effects of one’s own EC on one’s health, while accounting for the effects of two other family members’ EC and covariates. Differences in actor and partner effects emerged among the three family types. For families with mixed-sex twins, all significant effects were actor effects (one’s EC predicted one’s health). Actor effects also dominated in families with same-sex twin boys, but the twins’ EC also predicted mothers’ lower body composition. In contrast, for families with same-sex girls, a mix of actor and partner effects emerged. In families with at least one boy, one’s own EC may be more important for healthy functioning, whereas multiple girls increase partner effects linking EC and health.

Actor Effects

Consistent with hypotheses, there were several actor effects in which higher EC was associated with better health among individuals in the triad. For example, aligned with prior research, mothers’ EC was associated positively with their own general health and negatively with their own internalizing symptoms. Promoting adults’ self-regulation strategies has been related to increases in physical activity and (among women) decreases in depressive symptoms (Liau et al., 2018). Among college students low in EC, negative implicit attitudes toward physical activity were associated with less physical activity during leisure-time (Padin et al., 2017), suggesting that EC may facilitate managing negative emotions and persisting in health behaviors. EC may be important for overall wellbeing, as EC fosters engagement in behaviors that promote good health (e.g., regular exercise, routine doctor visits) and inhibit poor health (e.g., unhealthy food choices). EC aids in regulating negative reactivity and emotionality, likely helping promote positive mental health among mothers (Petersen & Posner, 2012).

For mixed-sex and same-sex girl twins, higher child EC was related to fewer internalizing symptoms. Among same-sex boy twins, higher EC was associated with better general health and lower body composition (e.g., lower BMI). Last, among families with mixed-sex twins, EC was negatively associated with boys’ body composition and positively associated with girls’ grip strength. These findings are consistent with work that positively links children’s EC to healthy functioning, including lower risk for obesity, healthier diet, and fewer depressive symptoms (Graziano et al., 2013; Nelson et al., 2018; van den Heuvel et al., 2017). Therefore, accounting for simultaneous estimation of actor and partner effects, an individual’s EC uniquely predicts multiple facets of their own health, underscoring the importance of EC across development.

Partner Effects

Several key partner effects emerged. Consistent with hypotheses, mothers’ EC was positively associated with children’s general health among same-sex girl twins. Limited existing research has linked parental EC to child health via the promotion of glycemic control and healthy eating habits (Healey et al., 2018; Zhou et al., 2019). Thus, mothers high in EC may promote daughters’ overall wellbeing by attending to their health behaviors, such as routine doctor visits. As expected, among families with same-sex twins, children’s EC was negatively associated with mothers’ internalizing symptoms. Further, among families with same-sex boy twins, children’s EC was also negatively associated with mothers’ body composition. It is possible that high EC among children may diminish, or even buffer, parental stress, lessening potential physiological strain and subsequent poor health (Merwin et al., 2015).

Among siblings, higher sibling EC was associated with fewer internalizing symptoms among same-sex girl families. Relationships among sisters have been rated as more caring and supportive (e.g., Cole & Kerns, 2001); these close relationships may influence health behaviors. Among middle school girls, but not boys, a positive sibling relationship protected against later substance use (Shepard, 2004). Also among girls (but not boys), sibling co-participation in physical activity was related to increases in moderate to vigorous physical activity (Cleland et al., 2011). Although the literature is small, findings to date suggest that health-related influences may be evident among sisters.

Among families with same-sex girl twins, there were partner effects that ran counter to expectations. Mothers’ EC was positively associated with twin girls’ body composition. Correspondingly, same-sex girl twins’ EC was positively associated with their mothers’ body composition. Mothers’ EC was negatively associated with their same-sex twin daughters’ grip strength. A non-linear association may exist between EC and body composition among girls and/or mothers, such that both those with the lowest and highest EC (i.e., overcontrolled) may be most at risk for obesity and other health problems (Anderson & Whitaker, 2018). There also may be environmental influences and societal pressures affecting body composition (or interacting with EC) among females outside the current model (e.g., media exposure; Rodgers et al., 2017). Continued work is needed to both replicate and explicate associations between EC and health among members of a family while considering sex differences.

Children’s self-regulation appears to be a key indicator of health and development; recently, it has been suggested as an identifier of children who may need support services to increase positive developmental and health outcomes (Claussen et al., 2021)—making all the more essential the identification of links between EC and health within the family unit. Dyadic work conducted during the global COVID-19 pandemic suggested that changes in mothers’ emotion regulation and changes in children’s inhibitory self-control were related to each other’s psychological wellbeing (DiGiorgoi et al., 2020). These recent findings in addition to the present study’s results point to the dynamic interplay in families between self-regulation competencies and health. Further, our work takes into consideration another family member, a twin sibling or second daughter/son.

Findings from investigations testing other forms of health functioning and collective biopsychosocial responses in particular contexts and interpersonal dynamics within and beyond families lend support for this line of inquiry. New work has suggested physiological synchrony (adrenocortical) among mother-child dyads as a potential indicator of biobehavioral co-regulation (Borelli et al., 2019). Additionally, among parent-adolescent dyads, suppression of emotions in one member of the dyad was associated with emotional eating in the other member (in addition to actor effects; Ferrer et al., 2017). These findings suggest that close group dynamics may affect biological and psychosocial regulatory processes and ultimately impact health. Family members, in particular, and others in close interpersonal contact, appear to uniquely affect their own as well as others’ health outcomes via self-regulatory competencies, including EC. As this was a first step in identifying associations between EC and health that may be unique, based on the family composition, future research should examine mediating mechanisms to elucidate why such associations exist.

Limitations

The study had several limitations. First, because we used mothers’ reports for children’s and mothers’ EC, general health status, and internalizing symptoms, we cannot dismiss that shared method variance may have biased some estimates. However, we used well-validated instruments, and obtained child and mother assessments on separate occasions to minimize bias. Moreover, several health indicators were directly assessed (i.e., BMI, body fat percentage, waist circumference, grip strength), bolstering confidence in reported associations. Second, family members were restricted to mothers and twins. Fathers and other siblings outside the scope of the investigation could also influence family dynamics and the observed associations. Third, findings may not generalize to children of different ages or other family structures (e.g., grandparent primary caregivers). Finally, causality between EC and health cannot be established with the cross-sectional, correlational design.

Future Directions and Clinical Implications

Future models should incorporate the influences of other family members into the model, including fathers/other caregivers and other siblings (e.g., Cook et al., 2018). Longitudinal research could help elucidate directions of effects between EC and health, including whether reciprocal associations are apparent (e.g., EC is associated with later health, which promotes later regulatory competence), and whether associations remain consistent across development. Extensions of this design could examine whether new associations emerge as children age, such as increasing sibling effects in adolescence. Further, assessing health behaviors (e.g., physical activity, diet) may highlight mediating or moderating mechanisms. For example, EC may play a role in whether or not an individual engages in a specific health behavior, such as structured routine exercise, and this in turn impacts their health outcomes (e.g., weight-related health, cardiovascular health).

Findings from the current study contribute to a small but rapidly growing literature documenting associations between self-regulatory competence and health. Overall, findings suggest that one’s own EC may affect one’s own health among same-sex boy and mixed-sex twins during middle childhood, as well as among mothers, whereas the EC of other family members may be especially important when considering the health of girls. Boys and mothers may benefit most from interventions focused on their own EC. Girls, especially with sisters, may benefit most from inclusion of the family in their interventions.