Childhood temperament types provide specificity over dimensions in predicting early adolescent symptoms of psychopathology: Genetic and environmental contributions

Abstract

Temperament is a heterogenous construct that captures individual differences in reactivity and regulation. Traditionally, dimensions of temperament are used over temperament types that may better capture heterogeneity. Using a racially/ethnically and socioeconomically diverse sample of twins (N=997 twins; 51.4% female; 58.5% white, 24.0% Hispanic; 28.7% at or below the poverty line), we modeled associations between primary-caregiver-reported temperament (M_age = 8.43) and later cross-reporter factors of anxiety, depression, externalizing, and ADHD symptoms (M_age = 10.88) and estimated latent genetic effects using temperament dimensions and types. The dimension of effortful control was negatively associated with all psychopathology symptoms over and above negative emotionality and surgency whereas temperament types yielded more nuanced results. We also uncovered a larger role of the shared environment for temperament and symptoms of psychopathology independently and in their relation than previous research has suggested. Findings provide greater specificity to current understanding of the relation between temperament and psychopathology.

Temperament is a broad construct that incorporates constitutionally-based individual differences in emotional and behavioral reactivity and regulation (Shiner et al., 2012). There are three central dimensions of temperament that emerge across child development: negative emotionality, effortful control, and surgency (Putnam et al., 2001), which are composed of more specific traits (e.g., negative emotionality includes anger, sadness, and lower soothability; effortful control includes activation control, attentional focusing, and inhibitory control; surgency includes impulsivity, activity level, high intensity pleasure, and lower shyness). Although temperament is a well-studied core concept in developmental psychology, most studies tend to measure and analyze it using a variable-centered, or dimensional approach (e.g., Putnam et al., 2024). Though less common, the person-centered analytic approach, or typological approach, captures temperament by identifying patterns of responses across temperament dimensions or traits that come together to make up an individual’s temperament. The current study aimed to investigate how using both the dimensional and typological approaches to temperament help us understand the phenotypic relation between childhood temperament and psychopathology symptoms in early adolescence. Further, we also aimed to use both approaches to investigate the shared genetic and environmental underpinnings of temperament and psychopathology symptoms.

The Complementary Nature of Dimensional and Typological Approaches

The commonly used dimensional approach, which most often utilizes one-to-two dimensions or traits, has improved our knowledge of the depth, complexity, and development of temperament. Gagne and Goldsmith (2020) review studies using the dimensional approach that support temperament as a generally stable construct, with the exception of early childhood when rapid development occurs. Although stability in temperament increases with age, neural system maturation and change in temperament as a result of the environment still occurs (Gagne & Goldsmith 2020; Shiner et al., 2012). Extant findings have enabled parents, teachers, and researchers to draw on an empirically-based understanding of this core aspect of child development (Blackwell, 2004).

In recent years, more advanced analytic methods have been developed that can incorporate all temperament traits in one analysis. Person-centered analyses, such as latent profile analysis (LPA), can be used to uncover different combinations and patterns of temperament (e.g., high negative emotionality, low surgency, and low regulation versus low negative emotionality, average surgency, and high regulation), yielding latent temperament ‘types.’ In contrast, dimensional approaches treat temperament as a homogeneous construct by assuming that an individual’s level of one trait provides insight into their levels of other traits and that the interrelations among traits are constant across the population (Laursen & Hoff, 2006). These assumptions stand in contrast to much of the temperament literature.

Historically, researchers have emphasized the importance of considering temperament wholistically for two reasons: 1) temperament dimensions are synergetic in nature (e.g., effortful control can attenuate the negative consequences of particularly high negative emotionality and surgency; Nielsen et al., 2019; Raines et al., 2023); 2) there is heterogeneity in the population regarding varying combinations of temperament dimensions (e.g., inhibited and uninhibited children; easy, difficult, and slow-to-warmup children; Kagan, 1994; Thomas et al., 1963). By modeling such heterogeneity, the person-centered approach may help fill the gaps left in the dimensional temperament literature. For example, the risk of high negative emotionality may vary depending on whether the child is also high or low on effortful control.

Early researchers taking a typological perspective to temperament using observational studies have suggested that larger samples would yield more types that are less common (Kagan, 1994). Yet, modern cross-sectional or short-term longitudinal studies using sample sizes of 322 to 787 participants have consistently identified only 3–4 temperament types (e.g., Beekman et al., 2015; Murillo et al., 2024; Scott et al., 2016; See Table S1 for a summary). The four temperament types that have been replicated across studies include: Negative Dysregulated (defined by high negative emotionality and low regulation), Well-Regulated (defined by high regulation), Typical Expressive (defined by average negative emotionality, regulation, and surgency), and the less prevalent Exuberant or Surgent types (defined by high surgency and average or low regulation).

Using the same sample as the current study, Murillo and colleagues (2024) consistently identified three of these temperament types at five developmental periods from infancy to early adolescence using LPA with 6 to 10 parent-reported temperament traits. The three types that emerged were termed Negative Dysregulated, Well-Regulated, and Typical Expressive types. Each type had consistent characterizations across ages with the most inconsistencies found in toddlerhood. Latent transition analyses revealed that the number of children transitioning into and out of the different types decreased the most from middle childhood to late childhood and remained consistent from late childhood to early adolescence. Importantly, a fourth type emerged in late childhood that resembled the Surgent or Exuberant types identified in prior studies (e.g., Scott et al., 2016), though this type was not identified in any other developmental period (Murillo et al., 2024). In sum, the Negative Dysregulated, Well-Regulated, Typical Expressive, and Surgent or Exuberant types have been identified across studies, and the first three have emerged across developmental periods. Thus, findings from the typological approach are consistent and may be useful in understanding temperamental risk for developmental psychopathology.

Measuring Psychopathology Symptoms

The most prevalent forms of psychopathology in adolescents include symptoms of anxiety, depression, externalizing (i.e., oppositional defiance, conduct problems, and aggression), and ADHD (Maternal and Child Health Bureau, 2024). Much of the literature to date has utilized a mix of approaches to measuring psychopathology symptoms. Many studies have chosen to use broadband measures of internalizing symptoms, including symptoms of anxiety and depression, and externalizing symptoms, incorporating ADHD with oppositional defiance, conduct problems, and aggression symptoms (Nielsen et al., 2019; Rhee et al., 2015; Wang et al., 2016). However, behavioral genetic studies have shown that anxiety, depression, externalizing and ADHD symptoms have unique genetic and environmental etiologies (Burt, 2009; Faraone & Larsson, 2018; Du Rietz et al., 2021), with the most recent diagnostic manual for mental disorders classifying ADHD as a neurodevelopmental disorder rather than an externalizing disorder (American Psychiatric Association, 2022).

Importantly, prior work has suggested that individuals do not consistently exhibit psychopathology symptoms across contexts (Martel et al., 2017). As a result, studies examining psychopathology symptoms are encouraged to use a multiple-reporter approach. The current study acquired primary caregiver, secondary caregiver, and teacher reports on youth’s symptoms of psychopathology, which were used to create four factors representing symptoms of anxiety, depression, externalizing, and ADHD, providing measures that spanned multiple contexts.

Phenotypic Relation Between Temperament and Psychopathology

With most studies to date using a dimensional approach, the association between various temperament dimensions and psychopathology symptoms is well-established with each dimension of temperament showing unique associations with psychopathology symptoms. For example, traits that make up surgency, such as high positive emotionality and impulsivity, predict lower anxiety and depression symptoms and higher externalizing and ADHD symptoms (Van Beveren et al., 2019). In contrast, higher effortful control is associated with less externalizing and ADHD symptoms while higher negative emotionality is associated with more symptoms across anxiety, depression, externalizing, and ADHD (Kostyrka-Allchorne et al., 2020; Scheper et al., 2017). Higher effortful control may also act as a buffer for the associations between higher negative emotionality and more psychopathology symptoms (Nielsen et al., 2019; Raines et al., 2023). Such heterogeneity may suggest that a person-centered approach to temperament may provide an important lens for fully understanding the association between temperament and psychopathology symptoms (Laursen & Hoff, 2006).

Despite person-centered approaches becoming more common in other areas of psychology, research on temperament using the modern typological approach to examine its relation with psychopathology symptoms has been limited with only one study doing so to our knowledge. After identifying five temperament types using cluster analyses, Janson and Mathiesen (2008) found that the types with the highest emotionality and sociability had the most externalizing symptoms while types with the highest shyness had the most internalizing symptoms. Thus, more research is needed. The first aim of the current study was to investigate the relation between temperament and psychopathology symptoms using both the dimensional and the typological approaches.

Genetic and Environmental Contributions

In addition to examining the phenotypic association between temperament and psychopathology, some research has considered the genetic and environmental underpinnings of the association. Twin studies can disentangle genetic and environmental contributions to traits by leveraging differences in the resemblance of monozygotic (MZ) twins relative to dizygotic (DZ) twins to statistically parse trait variance into genetic, shared environmental, and nonshared environmental components. Genetic influences increase the similarity of MZ twins relative to DZ twins, shared environmental influences make cotwins more similar regardless of genetic relatedness, and nonshared environmental influences contribute to differences between twins regardless of genetic relatedness (Knopik et al., 2017). There are two types of genetic factors, additive and dominant. Additive genetics represent the average main effects of individual genetic variants/alleles across the genome while dominant genetics capture genetic variants/alleles that interact with one another to contribute to the trait of interest (Knopik et al., 2017). Importantly, twin studies can also identify the genetic and environmental contributions to the covariance between two traits, such as temperament and psychopathology.

Twin studies using a dimensional approach have consistently found that temperament is moderately genetically influenced. Overall, estimates of genetic influence on dimensions of temperament tend to range from 20%–60% (Saudino, 2009). Some dimensions (e.g., effortful control) have high heritability and others, such as positive emotionality, are more influenced by the shared environment (Planalp et al., 2017; Rea-Sandin et al., 2023). For example, a twin study evaluating three traits that make up negative emotionality (fear, anger, and sadness) in middle childhood found that all three shared genetic influences, but anger and sadness also had genetic and shared environmental influences that did not overlap with fear (Clifford et al., 2015).

The only two twin studies that have used the typological approach to temperament in middle childhood have found similar heritability estimates to the dimensional literature and a more prominent role of the shared environment. Using factor mixture modeling to identify temperament types while also modeling dimensional variation within each profile, Scott and colleagues (2016) examined mean composites of mother- and father-reported temperament in middle childhood (M_age = 7.4 years). They found that the Negative Dysregulated type showed both moderate genetic and shared environmental influences. The Well-Regulated and Surgent types were the most highly heritable with no shared environmental influence, while the Typical Expressive type was the least heritable with significant shared environmental influences. A different study, using the same sample as the present study, found that the Negative Dysregulated, Well-Regulated, and Typical Expressive types were all moderately heritable, but Well-Regulated was the only type to have shared environmental influences (Murillo et al., 2024). Although these two studies found differing results regarding the role of the shared environment, both studies found a larger role of the shared environment than previous studies using dimensional approaches to temperament.

Turning to psychopathology symptoms, many studies of youth have used genetically informed designs to examine heritability. A meta-analysis of 490 twin and adoption studies found that internalizing (e.g., anxiety, depression) and externalizing (including ADHD) symptoms were moderately heritable (44% - 59%) with generally low, but significantly greater than zero, shared environmental influences (10%–16%) and moderate non-shared environmental influences (26%–42%; Burt, 2009). When anxiety, depression, externalizing, and ADHD symptoms are examined separately, rather than using broadband composites of internalizing and externalizing symptoms, studies have found differences in estimates of heritability. Specifically, when ADHD symptoms were analyzed separately from externalizing symptoms, large dominant genetic influences on ADHD symptoms were revealed explaining 70–74% of the variance (Burt, 2009; Faraone & Larsson, 2018). In contrast, when ADHD symptoms were included in the externalizing symptoms composite, these dominant genetic influences did not appear (Burt, 2009). Genetic theory suggests that the estimates of additive genetic influences become upwardly biased when dominant genetic effects are present and not modelled (Rijsdijk & Sham, 2002). Thus, the current study examined cross-reporter factors of anxiety, depression, externalizing and ADHD symptoms as separate outcomes to capture these etiological differences.

In most studies on the etiology of the relation between temperament and psychopathology symptoms, genetic influences tend to explain the covariance between the constructs, though these studies have only used the dimensional approach to temperament (e.g., Lemery-Chalfant et al., 2008; Martel et al., 2022; Vasin & Lobaskova, 2016). Whether the typological approach would result in different findings remains unclear. Thus, the second aim of the current study was to investigate whether and how the dimensional and typological approaches differed in estimates of shared genetic and environmental influences between temperament and psychopathology symptoms.

Current Study

Both the dimensional and typological approaches to temperament were used to model the relation between temperament and cross-reporter factors of anxiety, depression, externalizing, and ADHD symptoms. Based on prior research using the dimensional approach, we expected negative emotionality to be positively, and effortful control to be negatively, associated with all four factors of psychopathology symptoms. We also expected that surgency would be negatively associated with anxiety and depression symptoms and positively associated with externalizing and ADHD symptoms.

Although there is only one study on the relation between temperament types and psychopathology symptoms, we drew on the extensive literature using the dimensional approach to temperament to inform our hypotheses. Given that the Negative Dysregulated type was characterized by both the highest negative emotionality, lowest effortful control, and moderate surgency, we hypothesized that the probability of being Negative Dysregulated would be positively associated with all factors of psychopathology symptoms. In contrast, the Well-Regulated type is characterized by the lowest negative emotionality, highest effortful control, and the second highest surgency. Thus, we hypothesized that the probability of being Well-Regulated would be negatively associated with all factors of psychopathology symptoms. Finally, given that the Exuberant type was characterized by the highest levels of surgency and moderate to low levels of effortful control, we expected that the probability of being Exuberant would be negatively related to anxiety and depression symptoms and positively related to externalizing and ADHD symptoms. The Typical Expressive type was not included in the phenotypic analyses to address multicollinearity.

A twin design was used to elucidate genetic and environmental underpinnings of associations between temperament and psychopathology symptoms. With limited literature, we did not develop any specific hypotheses for the typological approach that differed from that of the dimensional approach. We expected that the shared covariance between temperament, as measured by both dimensions and types, and factors of psychopathology symptoms would be attributed to moderate to high additive genetic influences. Additionally, the factors of psychopathology symptoms in the current study spanned multiple reporters, and consequently multiple contexts, likely eliminating most context-dependent variability in symptoms of psychopathology providing a more trait-like measure. Therefore, we anticipated the estimates of genetic influence to be higher overall in this study than other studies where psychopathology symptoms were assessed by a single reporter.

Transparency and Openness

Preregistration

This study was not pre-registered.

Data, Materials, Code, and Online Resources

Due to the sensitive nature of genetic information collected in this study, data is not publicly available. De-identified data are available from the study principal investigators upon reasonable request. The code used for all analyses and the materials needed to reproduce all reported methodology are available at OSF: https://osf.io/8qv2f/.

Reporting

Given the intensive, longitudinal nature of the Arizona Twin Project, we have only reported the measures relevant to the present study. Information about other measures collected can be found in prior publications (Lemery-Chalfant et al., 2019). We report how we determined our sample size, all data exclusions and all manipulations in the study.

Ethical Approval

This study was approved by the university Institutional Review Board and was carried out in accordance with the provisions of the World Medical Association Declaration of Helsinki. All adults provided written informed consent (including teachers) and children provided verbal assent. Participants were compensated.

Method

Participants

The Arizona Twin Project is an ongoing longitudinal observational study of twins born across the state of Arizona, with data collected at 11 waves starting when twins were 12 months of age and follow-ups through mid-adolescence (Lemery-Chalfant et al., 2019). Across 499 families (N = 997 twins), temperament was assessed in late childhood (M_age=8.5 years, SD=.71), and psychopathology was measured 2.04 years later (SD=.64) in early adolescence at mean age 10.88 years (SD = 1.13). Families were initially recruited by the Arizona State Department of Health Services using state birth records between July 2007 and July 2008.

In late childhood, twins were approximately evenly split on sex assigned at birth (51.4% female) and primarily Hispanic/Latinx (24.0%) and non-Hispanic/Latinx White (58.5%), with the remainder being biracial with Hispanic/Latinx (4.8%), Asian American (3.4%), Black or African American (3.9%), Native American (2.7%), or multiracial or other (2.8%). Families were socioeconomically diverse, with income-to-needs ratios that fell below (7.4%) or near (22.3%) the poverty line, with the remainder categorized as lower middle class (16.3%), middle class (18.1%) and upper middle-to-upper class (35.8%).

Procedures

Primary caregivers (>93% mothers) completed demographic, temperament, and psychopathology online questionnaires along with 2–3-hour home visits at both ages (see Lemery-Chalfant et al., 2019 for details). Secondary caregivers (73.3% fathers) and teachers completed online psychopathology questionnaires as well.

Covid-19 Pandemic

The onset of the COVID-19 pandemic occurred during early adolescence resulting in 164 twins (21.9%) and their families participating virtually after quarantine was declared in the state of Arizona. Although study procedures had to fully transition online, raters completed psychopathology questionnaires as part of online Qualtrics surveys both pre- and post-pandemic (the temperament data were already collected). Independent samples t-tests, randomly selecting one twin from each pair, suggested that there were no differences between twins who completed the study before or after the onset of the COVID-19 pandemic on psychopathology symptoms regardless of reporter. The results of these analyses can be found in Table S2 of the Supplementary Material.

Measures

We found mixed support for the assumption of essential tau equivalence and unidimensionality for some of the temperament and psychopathology measures. Thus, McDonald’s omega, which is robust to these assumptions (Hayes & Coutts, 2020), was used in addition to Cronbach’s alpha for internal consistency reliability calculations. The sample size for each computation is reported in Table S3.

Zygosity

Zygosity was assessed using the 32-item caregiver-report Zygosity Questionnaire for Young Twins (Goldsmith, 1991), a valid alternative to genotyping with accuracy ranging between 93% and 98% in differentiating MZ and DZ twins (Constantino et al., 2017; Forget-Dubois et al., 2003). If the questionnaire did not yield clear results, we then used expert ratings using photos and videos taken during home visits, hospital birth records, lab placentae reports, and genotyping to determine the zygosity of the twins. In addition, all MZ twins participated in a secondary study in which they were genotyped, with 100% agreement between the questionnaire based zygosity determination and genotyping.

Temperament

In late childhood, primary caregivers reported on twins’ temperament using the Temperament in Middle Childhood Questionnaire (TMCQ; Simonds, 2006). Example items included “likes to be physically active,” “interrupts others when they are talking,” and “tends to become sad if plans do not work out,” all answered on a Likert scale from 1 (almost always untrue) to 5 (almost always true) of your child within the past six months. Parent-reported temperament measures, including the TMCQ, have shown modest to moderate convergence with lab-based observational measures of temperament (Kopala-Sibley et al., 2018).

Initial composites for the temperament trait subscales were computed by taking the average of all items that composed the trait subscale, resulting in ten composites: attentional focusing (7 items; alpha=.90; omega=.83), inhibitory control (8 items; alpha=.68; omega=.67), activation control (15 items; alpha=.77; omega=.78), anger (10 items; alpha=.82; omega=.83), soothability (8 items; alpha=.75; omega=.74), sadness (10 items; alpha=.77; omega=.77), activity level (9 items; alpha=.89; omega=.90), high-intensity pleasure (11 items; alpha=.70; omega=.73), impulsivity (13 items; alpha=.87; omega=.89), and shyness (5 items; alpha=.90; omega=.91). Following standard practice, mean composites were then created for effortful control (attentional focusing, inhibitory control, and activation control), negative emotionality (anger, soothability (reversed), and sadness), and surgency (activity level, high intensity pleasure, impulsivity, and shyness (reversed)) for use in the dimensional approach (Simonds & Rothbart, 2009).

The initial ten temperament traits were used as indicators in LPAs conducted in a prior study on the same sample (Murillo et al., 2024 for information on model fitting). LPA is an exploratory data driven approach that can be sample specific (Nylund-Gibson & Choi, 2018). However, the types identified were consistently identified in prior studies (Beekman et al., 2015; Scott et al., 2016), providing evidence that types used in the current study were not specific to this sample. Model fit statistics for the LPA are provided in Table S4, with the four-profile solution selected as the most parsimonious: Negative Dysregulated, Well-Regulated, Typical Expressive, and Exuberant (see Figure 1 for a visualization of the profile plots for each temperament type).

Figure 1. Profile plots of temperament types in late childhood.

Note. Wald Z-tests showed significant differences in key indicator means across all four profiles with four exceptions: Typical Expressive and Exuberant were not different on anger, Negative Dysregulated and Well-Regulated were not different on high intensity pleasure, Negative Dysregulated and Typical Expressive were not different on shyness, and Exuberant and Well-Regulated were not different on shyness.

The Well-Regulated type (23.91% of sample) was high, and the Negative Dysregulated type (20.72% of sample) was low on soothability and all dimensions tapping self-regulation (activation control, attentional focusing, and inhibitory control). Anger, sadness, and impulsivity were highest for the Negative Dysregulated type and lowest for the Well-Regulated type. However, both Negative Dysregulated and Well-Regulated children displayed high activity level and high intensity pleasure. The third type, Typical Expressive (35.82% of sample), was characterized by intermediate levels of anger, attentional focusing, and inhibitory control, but also the lowest activity level and high intensity pleasure, and levels of activation control, soothability, sadness, and shyness that were similar to the Negative Dysregulated type. Finally, the Exuberant type (19.55% of sample) was characterized by the highest levels of activity and high-intensity pleasure, the lowest levels of shyness, and low-to-moderate attentional focusing, inhibitory control, and anger.

Latent profile probability scores from the most parsimonious four-profile solution were extracted for use in the typological approach in the current study. These scores provide estimates of the probability of profile membership across all four profiles for each person based on their specific levels of each temperament trait compared to each profile’s average level of each dimension. Thus, we extracted posterior probabilities of profile membership across all four profiles for use in the typological approach.

Psychopathology Symptoms

In early adolescence, primary and secondary caregivers and teachers reported on twins’ mental health symptoms using the mental health domain of the Health and Behavior Questionnaire (HBQ-MH; Armstrong et al., 2003). The questionnaire assessed symptoms of depression (7 items; alphas = .68-.81; omegas = .70-.82), overanxiousness (13 items; alphas = .76-.78; omegas = .74-.77), separation anxiety (10 items; alphas = .75-.81; omegas = .76-.81), externalizing (i.e., oppositional defiance, conduct problems, and aggression; 25 items; alphas = .88-.94; omegas = .91-.95), and ADHD (i.e., inattention and impulsivity; 15 items; alphas = .82-.92; omegas = .91-.92). The HBQ is a widely used measure well-validated against gold standard structural clinical interviews and on twin samples (e.g., Lemery-Chalfant et al., 2007). Sample items including, “worries about past behavior,” “sleeps more than most children during the day and/or night,” “has temper tantrums or a hot temper,” “steals; takes things that don’t belong to him/her,” and “tries to get others to dislike a peer,” were all answered on a Likert scale of 0 (never or not true) to 2 (often or very true) of your child within the past six months.

Covariates

Primary caregivers completed demographic questionnaires at each age. Given that early adolescence marks a period of rapid and abrupt changes associated with age, sex, and the development of psychopathology symptoms (i.e., puberty; Graber, 2013), age in years and sex assigned at birth (0=male, 1=female) were included as covariates in all analyses.

Analytic Plan

Full information maximum likelihood (FIML; Enders, 2023) was used to address missing data for all analyses. In Mplus Version 8.8, “Type = Complex” was used to account for the interdependence between cotwins for all phenotypic analyses (Muthén & Muthén, 2017). We assessed assumptions of normality by computing skewness and kurtosis statistics for psychopathology symptoms across all reporters. The maximum likelihood estimator with robust standard errors (i.e., MLR estimator) was used to mitigate potential issues due to non-normal data. We used the OpenMx package in Rstudio (version 2023.06.1 Build 524; Boker et al., 2023) for all twin analyses. Prior to the primary analyses, descriptive statistics, and correlation matrices were calculated.

Phenotypic Analyses

The primary analyses focused on the relation between temperament and symptoms of anxiety, depression, externalizing, and ADHD used structural equation modeling (SEM). A Correlated Trait/Correlated Method-1 (CTCM-1) model was used to form latent factors of anxiety, depression, externalizing, and ADHD symptoms across reporter (i.e., primary caregivers, secondary caregivers, and teacher-reports) covarying for sex and age (see Eid et al., 2008 for more information on the CTCM-1 model). Primary caregiver-report was chosen as the comparison standard, under the assumption that they observe the child across a wider variety of contexts compared to the secondary caregiver or teacher. The resulting cross-reporter factors of psychopathology symptoms represent anxiety, depression, externalizing, and ADHD symptoms in early adolescence including measurement error from primary-caregiver reports and excluding measurement error from secondary-caregiver and teacher-reports.

The first SEM, examining the dimensional approach to temperament, included negative emotionality, effortful control, and surgency in late childhood as independent variables and the factors of psychopathology symptoms as outcome variables, covarying for sex and age. The second SEM, examining the typological approach to temperament, included the extracted posterior probabilities of belonging in the Negative Dysregulated, Well-Regulated, and Exuberant temperament types in middle childhood as independent variables and the factors of psychopathology symptoms as outcome variables, covarying for sex and age. The Typical Expressive type was left out of all phenotypic analyses to guard against multicollinearity. Finally, we accounted for multiple testing by using an adjusted Bonferroni procedure based on the number of parameters of interest estimated that accounted for the correlation between the outcome variables (i.e., factors of psychopathology symptoms) in each model (Anderson, 2023; Smith & Cribbie, 2013). The alpha level for the dimensional approach was set at α=.007 while the alpha level for the typological approach was set at α=.006. All statistical tests conducted on the parameters of interest were corrected (12 tests for each model) using the adjusted Bonferroni procedure. See Figure 2 for a visualization of both structural equation models.

Figure 2. Structural equation model using the dimensional approach to temperament.

Note. PC – Primary caregiver report; SC – Secondary caregiver report; T – Teacher report; ADHD – Attention Deficit Hyperactivity Disorder symptoms; Sep. Anx. – Separation Anxiety symptoms; Overanx. – Overanxious symptoms; Dep. – Depression symptoms; Ext. – Externalizing symptoms. The above figure depicts the first proposed structural equation model for aim 2 with temperament dimensions as independent variables and latent factors of anxiety, depression, externalizing, and ADHD symptoms across reporter as outcomes, controlling for age and sex, where sex is coded 0 = male and 1 = female. The same model was used for the typological approach, replacing negative emotionality, effortful control, and surgency with the Negative Dysregulated type, Well-Regulated type, and Exuberant type.

Genetic and Environmental Contributions to Covariance

To investigate the shared genetic and environmental influences on temperament and symptoms of psychopathology, we utilized a twin design and both dimensional and typological approaches to temperament. Monozygotic (MZ) twins share 100% of their DNA while dizygotic (DZ) twins share about 50% of their DNA on average (Knopik et al., 2017). To estimate the genetic and environmental influences, quantitative genetic approaches such as twin modeling capitalize on genetic theory mentioned above.

Intraclass correlations (ICCs) can be used to descriptively assess genetic, shared environmental, and nonshared environmental influences. When MZ ICCs are greater than DZ ICCs, the role of additive genetics is present, but if DZ ICCs are less than half of MZ ICCs, we would expect the role of dominant genetics to be present (Neale & Maes, 2004). When MZ and DZ ICCs are similar, the role of shared environment is present. Finally, if MZ and DZ ICCs are small, the role of the non-shared environment is present. To get more accurate quantitative estimates for genetic and environmental influences that take into account sample sizes and standard errors, we fit twin models using multigroup SEM.

Univariate ACE twin models decompose the variances in a trait into latent additive genetic (A; cotwin covariance fixed to 1.0 for MZ twin pairs and .50 for DZ pairs), shared environmental (C; fixed to a correlation of 1.0 for both MZ and DZ pairs), and nonshared environmental (E; uncorrelated between cotwins and includes measurement error) components (Neale & Maes, 2004). . Due to model identification, dominant genetic influences (D; cotwin covariance fixed to 1.0 for MZ twin pairs and .25 for DZ pairs) and C cannot be included in the same model. As an alternative to the full ACE model, an ADE model can be estimated which estimates both additive and dominant genetic influences.

For two variables, the Cholesky Decomposition was used in the bivariate twin models (Neale & Maes, 2004) which decomposes the cross-twin cross-trait covariance, in addition to cross-twin covariance of each individual trait, into A, C or D, and E components. The path estimates of the bivariate twin models produce estimates of additive genetic, shared environmental or dominant genetic, and nonshared environmental influences on the variance of trait 1 (e.g. a11, c11, e11), on the covariance between trait 1 and trait 2 (e.g. a21, c21, e21), and on the residual variance of trait 2 (e.g. a22, c22, e22; see Figure S1). In addition to this covariance decomposition, bivariate twin models also yield genetic and environmental correlations between the traits. The genetic correlation indicates the extent to which the genetic influences of trait 1 overlap with the genetic influences of trait 2. The same is true for the shared environmental and nonshared environmental correlations.

Preliminary analyses included computing ICCs, saturated twin models to check assumptions of equal covariances across zygosity and sex, and univariate twin models for each temperament dimension, temperament type, and factors of psychopathology symptoms. For the primary analyses, we fit bivariate twin models (including ADE models when necessary; Rijsdijk & Sham, 2002) estimating genetic and environmental influences on the covariance between temperament dimensions or type and the factors of psychopathology symptoms.

For univariate model fitting, the full ACE model is always fit first (Neale & Maes, 2004). If the ICC of the DZ twins was less than half the ICC of the MZ twins, a full ADE model was fit next. After fitting the full model(s), reduced models dropping the A and C or D parameters were fit. The E parameter was never dropped to consistently account for measurement error.

For bivariate model fitting, we first fit the full ACE (influences on temperament)-ACE (influences on covariance)-ACE (residual influences on psychopathology symptoms) model (Neale & Maes, 2004). If the univariate model suggested that the ADE model was the best fit for at least one of the traits, a full ADE-ADE-ADE model was fit next. Then, paths were systematically dropped to identify the best fitting model. The order of dropping paths for the bivariate models was to first drop superfluous paths that were estimated at or near zero. Then, we used the best fitting model from the univariate twin models as a guide for dropping additional paths. For both univariate and bivariate models, reduced models were compared to the full model and, consistent with recommendations, a reduced model was chosen over the full ACE model with a nonsignificant difference if the −2loglikelihood was not significantly different (p>.05) and the estimated AIC was lower than that of the full model.

Results

Descriptive statistics and bivariate correlations for all temperament dimensions, types, and psychopathology symptoms are summarized in Table S3. All correlations between temperament dimensions, types, and psychopathology symptoms were in the expected directions with a few exceptions. Surgency had significant positive associations with primary caregiver reported symptoms of overanxiousness and separation anxiety in addition to primary caregiver, secondary caregiver, and teacher reported symptoms of externalizing and ADHD. Results from independent-samples t-tests suggested that males had higher primary caregiver, secondary caregiver, and teacher reported symptoms of externalizing and ADHD compared to females. There were no other sex differences (see Table S5). Although primary caregiver and teacher reported depression symptoms and teacher reported externalizing symptoms exceeded recommendations on skewness (>2.0; West et al., 1995) and kurtosis (>7.0; West et al., 1995; see Table S3), the MLR estimator used in all phenotypic analyses is robust to such assumption violations (Muthén & Muthén, 2017). No statistical variable transformations were necessary.

Measurement Model

Upon running the CTCM-1 model, Mplus returned an error that suggested constraining the variance of the primary caregiver report on externalizing symptoms due to a small, negative estimated residual variance. After constraining this parameter, the CTCM-1 model converged without estimation issues. The RMSEA, CFI, and SRMR were above the recommended thresholds for good model fit (see Table 1; Hu & Bentler, 1999). The standardized factor loadings for trait and method factors were above .3 indicating that each scale meaningfully contributed to the appropriate factors. Results from the CTCM-1 model are provided in Figure S2. The factors of psychopathology symptoms were extracted from the results of the CTCM-1 model for later use in twin modeling. The extracted factors were also used, along with the respective independent variables for each analysis, to create P-P plots in SPSS (see Figure S3a–b). P-P plots did not suggest major normality violations, and the MLR estimator was used in each SEM.

Table 1.

Model fit indices for the measurement model and full structural equation models using both dimensional and typological approaches to temperament

Model	# of Parameters	Log-likelihood	χ2	p-value	RMSEA	CFI	SRMR
Measurement Model	73	−1309.21	244.59	<.001	.05	.94	.06
SEM - Dimensions	106	−2305.62	275.30	<.001	.04	.95	.06
SEM - Types	77	572.44	220.97	<.001	.04	.96	.07

Note. SEM=Structural equation model; RMSEA=Root Mean Square Error of Approximation.

CFI=Comparative Fit Index; SRMR=Standardized Root Mean Square Residual.

Temperament and Psychopathology

Model fit statistics for each of the two SEM are provided in Table 1, with parameter estimates provided in Table 2. A visualization of the proposed SEMs can be seen in Figure 2.

Table 2.

Estimates from the structural equation models relating temperament to psychopathology symptoms

Variable	ΔR2	b	SE	95% CI	t	p-value
Anxiety
SEM - Dimensions	Total = .08
Negative Emotionality	.01	.08	.06	[−.04, .20]	1.37	.17
Effortful Control	.04	−.21	.06	[−.33, −.10]	−3.57	<.001
Surgency	.00	.06	.06	[−.05, .18]	1.09	.28
Sex		.16	.04	[.07, .24]	3.55	<.001
Age		.04	.05	[−.07, .14]	0.66	.51
SEM - Types	Total = .14
Negative Dysregulated	.03	.19	.06	[.06, .31]	2.88	.01
Well-Regulated	.05	−.24	.05	[−.34, −.14]	−4.59	<.001
Exuberant	.02	−.12	.05	[−.21, −.03]	−2.58	.01
Sex		.13	.04	[.04, .21]	2.93	.004
Age		.04	.06	[−.07, .14]	0.66	.51
Depression
SEM - Dimensions	Total = .11
Negative Emotionality	.02	.14	.05	[.04, .24]	2.65	.01
Effortful Control	.08	−.31	.06	[−.43, −.19]	−5.07	<.001
Surgency	.01	−.15	.06	[−.26, −.03]	−2.54	.01
Sex		−.02	.05	[−.11, .07]	−0.34	.73
Age		.07	.05	[−.03, .17]	1.42	.16
SEM - Types	Total = .21
Negative Dysregulated	.05	.22	.07	[.09, .36]	3.20	.003
Well-Regulated	.07	−.26	.04	[−.34, −.18]	−6.13	<.001
Exuberant	.05	−.21	.05	[−.30, −.12]	−4.64	<.001
Sex		−.03	.04	[−.12, .05]	−0.75	.45
Age		.09	.05	[−.01, .19]	1.81	.07
Externalizing
SEM - Dimensions	Total = .20
Negative Emotionality	.01	.10	.06	[−.01, .21]	1.84	.07
Effortful Control	.09	−.35	.05	[−.45, −.25]	−6.66	<.001
Surgency	.002	.11	.06	[−.01, .22]	1.74	.08
Sex		−.03	.04	[−.11, .05]	−0.78	.44
Age		.02	.05	[−.07, .10]	0.32	.75
SEM - Types	Total = .24
Negative Dysregulated	.13	.42	.05	[.31, .52]	7.98	<.001
Well-Regulated	.02	−.15	.04	[−.22, −.07]	−3.93	<.001
Exuberant	.01	.12	.05	[.03, .21]	2.50	.01
Sex		−.03	.04	[−.11, .04]	−0.86	.39
Age		.01	.04	[−.07, .09]	0.24	.81
ADHD
SEM - Dimensions	Total = .39
Negative Emotionality	−.002	−.04	.05	[−.13, .06]	−0.79	.43
Effortful Control	.26	−.57	.04	[−.66, −.49]	−13.85	<.001
Surgency	−.01	.07	.06	[−.06, .19]	1.07	.28
Sex		−.07	.04	[−.14, .01]	−1.79	.07
Age		−.06	.05	[−.15, .03]	−1.31	.19
SEM - Types	Total = .33
Negative Dysregulated	.17	.48	.05	[.38, .02]	9.25	<.001
Well-Regulated	.06	−.08	.04	[−.16, −.01]	−2.30	.02
Exuberant	.06	.31	.05	[.21, .41]	5.86	<.001
Sex		−.07	.04	[−.15, .02]	−1.58	.11
Age		−.07	.04	[−.16, .01]	−1.65	.10

Note. Boldface - p < .007 for the dimensional approach and p < .006 for the typological approach. This table contains the results from two independent structural equation models (SEMs) with the same outcomes using temperament dimensions for the first and temperament types for the second.

Dimensional Approach

The SEM for the dimensional approach had good model fit (Hu & Bentler, 1999). The R-squared statistics suggested that the dimensional approach accounted for 8% of the variance in anxiety symptoms, 11% of the variance in depression symptoms, 20% of the variance in externalizing symptoms, and 39% of the variance in ADHD symptoms. As expected, effortful control was negatively related to psychopathology symptoms across all factors. Inconsistent with hypotheses, both negative emotionality and surgency did not have any meaningful associations with anxiety, depression, externalizing, or ADHD symptoms.

Typological Approach

Model fit indices for the SEM using the typological approach suggested good model fit (Hu & Bentler, 1999). Despite the little change in model fit statistics and no change in number of estimated parameters, number of variables included in the model, or sample size, all three temperament types explained descriptively more variation in psychopathology symptoms compared to the dimensional approach except for ADHD symptoms. Specifically, the typological approach accounted for 14% of the variance in anxiety symptoms, 21% of the variance in depression symptoms, 24% of the variance in externalizing symptoms, and 33% of the variance in ADHD symptoms. Findings from the typological approach were only partially consistent with hypotheses. The probability of being in the Well-Regulated type was negatively related to symptoms of anxiety, depression, and externalizing, and not related to ADHD symptoms. Further, the Well-Regulated type explained the most variance in anxiety and depression symptoms. The probability of being in the Negative Dysregulated type was positively associated with depression, externalizing, and ADHD symptoms, but not related to anxiety symptoms. The Negative Dysregulated type also explained the most variance in externalizing and ADHD symptoms. Finally, the Exuberant type was negatively associated with depression symptoms and positively related to ADHD symptoms, but not related to anxiety or externalizing symptoms.

Heritability of Temperament and Psychopathology

Twin models assume equal variances and covariance across zygosity group and sex. To test these assumptions, five group saturated models were fit for all temperament dimensions, types, and factors of psychopathology symptoms (see Table S6). Results suggested that there were differences in the variances and covariances across zygosity and sex for depression and externalizing symptoms. We did not find any other significant differences.

Prior to conducting twin models, intraclass correlations were run within zygosity (see Table S7). MZ correlations were greater than DZ correlations across all temperament dimensions, types, and factors of psychopathology symptoms, suggesting a role of genetics. Notably, the DZ correlation was less than half of the MZ correlation for the Typical Expressive and Exuberant types, so the ADE model was fit for both. Although the ICCs for ADHD symptoms did not suggest fitting an ADE model, the ADE model has previously been shown to fit ADHD symptoms (Burt, 2009; Faraone & Larsson, 2018). Therefore, the ADE model was also fit for ADHD symptoms. Model fit indices and parameter estimates of the univariate twin models are in Table 3.

Table 3.

Genetic and environmental influences on temperament dimensions, temperament types, and factors of psychopathology symptoms

		−2LL	df	AIC	Δ-2LL	Δdf	p	A	C or D	E
Dimensions
Negative	ACE	154.85	576	162.85	-	-	-	−.19/.30	−.27/.57	.13/.13
Emotionality	AE	190.88	577	196.88	36.04	1	<.001
	CE	172.06	577	178.06	17.22	1	<.001
	E	435.30	578	439.30	280.45	2	<.001
Effortful	ACE	830.61	576	838.61	-	-	-	.51/.81	0/0	.24/.19
Control	AE	830.61	577	836.61	0	1	>.99	.51/.81	-	.24/.19
	CE	868.85	577	874.85	38.24	1	<.001
	E	959.85	578	963.85	129.24	2	<.001
Surgency	ACE	264.20	576	272.20	-	-	-	.31/.80	0/0	.15/.20
	AE	264.20	577	270.20	0	1	>.99	.31/.80	-	.15/.20
	CE	301.42	577	307.42	37.22	1	<.001
	E	394.23	578	398.23	130.03	2	<.001
Types
Negative	ACE	354.58	576	362.58	-	-	-	−.32/.76	0/0	.18/.24
Dysregulated	AE	354.58	577	360.58	0	1	>.99	−.32/.76	-	.18/.24
	CE	384.85	577	390.85	30.27	1	<.001
	E	461.07	578	465.07	106.49	2	<.001
Well-	ACE	468.78	576	476.78	-	-	-	−.26/.41	−.24/.34	.21/.25
Regulated	AE	477.60	577	483.60	8.82	1	.003
	CE	480.19	577	486.19	11.41	1	<.001
	E	617.76	578	621.76	148.98	2	<.001
Exuberant	ACE	347.50	576	355.50	-	-	-	−.28/.63	0/0	.21/.37
	ADE	344.97	576	353.97	-	-	-	−.13/.15	.24/.50	.20/.35
	AE	347.50	577	353.50	2.53	1	.11	.28/.63	-	.21/.37
	E	427.76	578	431.76	82.79	2	<.001
Typical	ACE	571.61	576	579.61	-	-	-	.40/.80	0/0	.20/.20
Expressive	ADE	562.74	576	570.74	-	-	-	−.10/.05	.38/.76	.19/.19
	AE	571.61	577	577.61	8.87	1	<.001
	E	682.04	578	686.04	119.30	2	<.001
Psychopathology
Anxiety	ACE	1447.93	617	1455.93	-	-	-	.68/.57	.45/.25	.39/.19
	AE	1453.18	618	1459.18	5.26	1	.02
	CE	1476.42	618	1482.42	28.49	1	<.001
	E	1643.20	619	1638.20	186.27	2	<.001
Depression	ACE	1401.61	617	1409.61	-	-	-	.75/.74	.29/.11	.34/.15
	AE	1402.71	618	1408.71	1.10	1	.29	.80/.85	-	.33/.15
	CE	1450.33	618	1456.33	48.72	1	<.001
	E	1591.20	619	1595.20	189.59	2	<.001
Externalizing	ACE	1468.11	617	1476.11	-	-	-	.71/.56	.51/.29	.36/.15
	AE	1475.96	618	1481.96	7.85	1	.005
	CE	1504.50	618	1510.51	36.40	1	<.001
	E	1692.82	619	1696.82	224.71	2	<.001
ADHD	ACE	1425.72	617	1433.72	-	-	-	.91/.92	−.07/.01	.25/.07
	ADE	1425.72	617	1433.72	-	-	-	.91/.93	0/0	−.25/.07
	AE	1425.72	618	1431.72	0	1	.96	.91/.93	-	.25/.07
	CE	1536.06	618	1542.06	110.35	1	<.001
	E	1695.10	619	1699.10	269.39	2	<.001

Note. Boldface – most parsimonious model. −2LL=−2 log likelihood; df=degrees of freedom; Δ=change; p=p-value; AIC=Akaike’s Information Criterion; A, C, D, and E are variance components representing additive genetic, shared environmental, dominant genetic, and nonshared environmental variance, respectively, with unstandardized variance reported first, followed by standardized variance.

Heritability of Temperament Dimensions and Types.

Results from the univariate twin models showed moderate to large additive genetic influences across all temperament dimensions. The final models for effortful control and surgency estimated an absent C factor with 80% of variance in effortful control and 81% of variance in surgency explained by genetic factors. Only 13% and 19% of the variance was explained by nonshared environmental factors for effortful control and surgency, respectively. Negative emotionality was less heritable, with only 30% of the variance explained by genetic factors, 57% by shared environmental factors, and 13% by nonshared environmental factors.

Univariate twin models showed moderate to large genetic influences across all four temperament types. Only the Well-Regulated type had a C factor estimated greater than zero with 34% of the variance explained by shared environmental factors, 41% of variance explained by genetic factors, and 25% explained by non-shared environmental factors. The Negative Dysregulated, Exuberant, and Typical Expressive types were highly heritable with no influence of the shared environment. Specifically, 76% of the variance in the Negative Dysregulated type, 63% in the Exuberant type, and 81% (A = 5%; D = 76%) in the Typical Expressive type was explained by genetic factors. These types also had a small nonshared environmental influence with 24% of the variance in the Negative Dysregulated type, 35% in the Exuberant type, and 19% in the Typical Expressive type explained by E.

Heritability of Psychopathology Symptoms.

Consistent with hypotheses, results from the univariate twin models conducted on the factors of psychopathology symptoms showed moderate to large additive genetic influences. While small, all factors of psychopathology symptoms had estimated C paths greater than zero in the full models. Depression and ADHD symptoms were the most heritable with 85% of the variance in depression symptoms and 93% of the variance in ADHD symptoms explained by genetic factors. The remainder of the variance in depression and ADHD symptoms was explained by E, 15% and 7% respectively. Although the ADE model was tested for ADHD symptoms, dominant genetic influences on ADHD symptoms were nonsignificant and dropped from the final univariate model. Anxiety and externalizing symptoms were moderately heritable with 57% of the variance in anxiety symptoms and 56% of the variance in externalizing symptoms explained by genetic factors. Anxiety and externalizing symptoms also had moderate shared environmental influences with 25% and 29% of the variance explained by C respectively. The remainder of the variance, 19% and 15%, in anxiety and externalizing symptoms was explained by E.

Shared Heritability of Temperament and Psychopathology.

Table 4 provides model fit indices for all bivariate twin models tested in the study and genetic and environmental correlations for the full and best-fitting bivariate models. Bivariate models were not assessed for any dimensions or types that were not significantly related to anxiety, depression, externalizing, or ADHD symptoms in the SEMs (e.g., for the dimensional approach, only effortful control was included in the bivariate twin models) or did not have significant correlations with psychopathology symptoms (e.g., the Typical Expressive type was not included in the bivariate twin models for symptoms of anxiety or depression).

Table 4.

Genetic and environmental influences on covariance between temperament dimensions and types with factors of psychopathology symptoms

	Model	−2LL	df	AIC	Δ-2LL	Δ df	p	rA	rC	rE
Anxiety
Effortful Control	ACE- ACE-ACE	2238.97	1190	2260.97	-	-	-	−.25	−1.00	.30
	AE-AE-ACE	2245.64	1192	2263.64	6.67	2	.04	−.45	-	.35
	AE-E-ACE	2276.61	1193	2292.61	37.64	3	<.001
	AE-A-ACE	2257.55	1193	2273.55	18.57	3	<.001
Well-Regulated type	ACE-ACE-ACE	1888.45	1190	1910.45	-	-	-	−.13	−.71	.004
	ACE-CE-ACE	1888.92	1191	1908.92	0.47	1	.49
	ACE-AE-ACE	1894.13	1191	1914.13	5.68	1	.02
	ACE-AC-ACE	1888.45	1191	1908.45	0.002	1	.97
	ACE-C-ACE	1889.03	1192	1907.03	0.58	2	.75	-	−.88	-
Depression
Effortful Control	ACE-ACE-ACE	2198.71	1190	2220.71	-	-	-	−.27	−1.00	.26
	AE-AE-ACE	2201.46	1192	2219.46	2.75	2	.25
	AE-AE-AE	2201.91	1193	2217.91	3.20	3	.36	−.36	-	.27
	AE-E-ACE	2232.36	1194	2246.36	33.65	4	<.001
	AE-A-ACE	2209.31	1194	2223.31	10.60	4	.03
Negative Dysregulated type	ACE-ACE-ACE	1720.92	1190	1742.92	-	-	-	.26	1.00	.08
	AE-AE-ACE	1723.35	1192	1741.35	2.43	2	.30
	AE-AE-AE	1723.47	1193	1739.47	2.56	3	.47
	AE-E-AE	1750.07	1194	1764.07	29.16	4	<.001
	AE-A-AE	1723.91	1194	1737.91	2.99	4	.56	.37	-	-
Well-Regulated type	ACE-ACE-ACE	1839.48	1190	1861.48	-	-	-	−.30	−.52	−.06
	ACE-AE-ACE	1840.74	1191	1860.74	1.26	1	.26
	ACE-ACE-AE	1840.04	1191	1860.04	.57	1	.45
	ACE-AE-AE	1841.17	1192	1859.17	1.70	2	.43	−.42	-	−.04
	ACE-AC-ACE	1839.88	1191	1859.88	.40	1	.53
Exuberant type	ACE-ACE-ACE	1740.34	1190	1762.35	-	-	-	−.14	1.00	−.16
	ADE-ADE-ADE	1739.04	1190	1761.04	-	-	-	.09	−1.00	−.16
	ADE-AE-AE	1739.54	1192	1757.54	0.05	2	.78
	AE-AE-AE	1741.77	1193	1757.77	2.73	3	.44
	AE-E-AE	1743.84	1194	1757.84	4.80	4	.31	-	-	−.21
	AE-A-AE	1745.32	1194	1759.32	6.28	4	.18
	ADE-E-AE	1959.85	1193	1975.85	3.28	3	.35
	ADE-A-AE	1960.36	1193	1976.36	3.79	3	.28
Externalizing
Effortful Control	ACE-ACE-ACE	2236.57	1190	2258.57	-	-	-	−.44	−1.00	.28
	AE-AE-ACE	2240.88	1192	2258.88	4.31	2	.12	−.59	-	.32
	AE-E-ACE	2298.63	1193	2314.63	62.07	3	<.001
	AE-A-ACE	2250.90	1193	2266.90	14.33	3	.002
Negative Dysregulated type	ACE-ACE-ACE	1755.07	1190	1777.07	-	-	-	.36	1.00	.22
	AE-AE-ACE	1758.81	1192	1776.81	3.73	2	.15	.50	-	.18
	AE-E-ACE	1799.42	1193	1815.42	44.34	3	<.001
	AE-A-ACE	1761.71	1193	1777.70	6.63	3	.08
Well-Regulated type	ACE-ACE-ACE	1898.72	1190	1920.72	-	-	-	−.04	−.83	−.16
	ACE-CE-ACE	1898.77	1191	1918.77	0.05	1	.82	-	−.87	−.17
	ACE-AE-ACE	1908.66	1191	1928.66	9.34	1	.002
	ACE-AC-ACE	1901.12	1191	1921.12	2.40	1	.12
Typical Expressive type*	ACE-ACE-ACE	2034.55	1190	2056.55	-	-	-	−.23	1.00	.10
	ADE-ADE-ADE	2034.55	1190	2056.55	-	-	-	.62	−1.00	.09
	ADE-AE-AE	2036.22	1192	2054.22	1.68	2	.43	−.30	-	.07
	AE-AE-AE	2044.30	1193	2060.30	9.75	3	.02
	ADE-E-AE	2038.62	1193	2054.62	4.07	3	.25
	ADE-A-AE	2044.77	1194	2058.77	10.22	4	.04
ADHD
Effortful Control	ACE-ACE-ACE	2105.49	1190	2127.49	-	-	-	−.57	1.00	−.40
	AE-AE-AE	2105.54	1193	2121.54	0.04	3	>.99	−.56	-	-.41
	AE-A-AE	2121.71	1194	2135.71	16.22	4	.003
	AE-E-AE	2206.82	1194	2220.82	101.32	4	<.001
Negative Dysregulated type	ACE-ACE-ACE	1698.79	1190	1720.79	-	-	-	.56	−1.00	−.06
	AE-AE-AE	1699	1193	1715.00	0.21	3	.98
	AE-E-AE	1776.36	1194	1790.36	77.58	4	<.001
	AE-A-AE	1699.29	1194	1713.29	0.51	4	.97	.52	-	-
Exuberant type	ACE-ACE-ACE	1734.24	1190	1756.24	-	-	-	.26	−1.00	.36
	ADE-ADE-ADE	1731.53	1190	1753.53	-	-	-	−.12	1.00	.35
	ADE-AE-AE	1733.02	1192	1751.02	1.49	2	.48
	AE-AE-AE	1734.45	1193	1750.45	2.92	3	.40	.24	-	.37
	AE-E-AE	1746.34	1194	1760.34	14.81	4	.01
	AE-A-AE	1750.80	1194	1764.80	19.27	4	<.001
Typical Expressive type*	ACE-ACE-ACE	1966.57	1190	1988.57	-	-	-	−.40	1.00	−.02
	ADE-ADE-ADE	1956.95	1190	1978.95	-	-	-	1.00	−1.00	−.01
	ADE-ADE-E	1956.95	1192	1974.95	0	2	>.99
	ADE-AD-E	1956.96	1193	1972.96	0.01	3	>.99	1.00	−1.00
	ADE-AE-AE	1967.60	1192	1985.60	10.65	2	.004
	AE-AE-AE	1988.46	1193	2004.46	31.51	3	.003

Note.

^*- the typical expressive type was significantly correlated with externalizing (r = −.23) and ADHD (r = −.12) symptoms at p < .05. Boldface – most parsimonious model. −2LL=−2 log likelihood; df=degrees of freedom; Δ=change; p=p-value; AIC=Akaike’s Information Criterion; rA, rC, rD, and rE represent the correlation between the additive genetic, shared environmental, dominant genetic, and nonshared environmental variance components. Effortful control was the only temperament dimension included in the bivariate twin models because it was the only dimension significantly related to anxiety, depression, externalizing, and ADHD symptoms.

Symptoms of Anxiety.

The covariance between effortful control and anxiety symptoms was best explained by the AE-AE-ACE model, suggesting that the relation was explained by genetic (rA = −.45) and non-shared environmental factors (rE = .35). The covariance between the Well-Regulated type and anxiety was best explained by the ACE-C-ACE model, suggesting that the relation was explained by only shared environmental factors (rC = −.88).

Symptoms of Depression.

Across all bivariate models for depression symptoms, all relations with temperament were at least partially explained by genetic factors with the exception of the Exuberant type. The covariance between effortful control and depression symptoms was best explained by the AE-AE-AE model, suggesting that genetic (rA = −.36) and non-shared environmental factors (rE = .27) explained the relation. The AE-A-AE model fit best for the Negative Dysregulated type and depression symptoms, suggesting that the covariance was only explained by genetic factors (rA=.37). The ACE-AE-AE model fit best for the Well-Regulated type and depression symptoms, suggesting that the covariance was explained by genetic (rA = −.42) and nonshared environmental factors (rE = −.04). The AE-E-AE model fit best for the Exuberant type and depression symptoms suggesting that the relation was only explained by nonshared environmental factors (rE=−.21).

Symptoms of Externalizing.

With the exception of the Well-Regulated type, the covariance between temperament and externalizing symptoms was at least partially explained by genetic factors. The AE-AE-ACE model fit best for effortful control and externalizing symptoms, suggesting that the covariance was primarily explained by genetic (rA=−.59) and the non-shared environmental (rE=.32) influences. Similarly, the covariance between the Negative Dysregulated type and externalizing symptoms was best explained by the AE-AE-ACE model, suggesting that the relation was explained by genetic (rA=.50) and non-shared environmental (rE=.18) factors. The ACE-CE-ACE model fit best for the Well-Regulated type and externalizing symptoms where the covariance was explained by shared environmental (rC=−.88) and nonshared environmental (rE=−.17) influences. Finally, the ADE-AE-AE model fit best for the Typical expressive type and externalizing symptoms, suggesting that the covariance was partially due to genetic factors with a moderate negative correlation between the genetic factors (rA=−.30) and a small positive correlation between the nonshared environmental factors (rE=.07).

Symptoms of ADHD.

All bivariate twin models showed that the covariance between temperament and ADHD symptoms was at least partially explained by genetic factors. The covariance between effortful control and ADHD symptoms was explained by the AE-AE-AE model, suggesting that both genetic (rA=−.56) and nonshared environmental (rE=−.41) influences explained the relation. The AE-A-AE model fit best for the Negative Dysregulated type and ADHD symptoms, suggesting that the covariance was explained by genetic factors (rA=.52) only. The AE-AE-AE fitting best for the Exuberant type, suggesting that the covariance was primarily explained by both genetic (rA=.24) and nonshared environmental (rE=.37) influences.

Finally, the covariance between the Typical Expressive type and ADHD symptoms was explained by genetic influences only with the ADE-AD-E model fitting best. The genetic factors for the Typical Expressive type and ADHD symptoms were completely correlated. Specifically, the additive genetic factors were positively correlated (rA=1.00), while the dominant genetic factors were negatively correlated (rD=−1.00), suggesting that only one A and one D factor was needed in the model to explain the additive and dominant genetic influences on the variability in the Typical Expressive type, in ADHD symptoms, and their covariance.

Discussion

Temperament is a well-studied concept in child development with implications for socioemotional development, adaptation, and adjustment (Shiner et al., 2012). Some contemporary researchers are extending the dimensional approach to temperament by using advanced statistical methods, such as LPA, to examine multiple components of temperament simultaneously, identifying common heterogeneous patterns or types as a result. The current study used both the dimensional and typological approaches to examine the relation between temperament and common forms of child psychopathology symptoms – anxiety, depression, externalizing, and ADHD. We found considerable differences between the dimensional and typological approaches with the results only partially consistent with our predictions. Most notably, effortful control was negatively associated with psychopathology symptoms over and above negative emotionality and surgency. The typological approach yielded more nuanced findings that are discussed in detail in the following sections. The association between temperament and psychopathology symptoms was explained at least in part by genetic influences in nearly all cases with two exceptions providing new contributions to the literature: 1) the shared variation between Well-Regulated type and anxiety symptoms was explained by only shared environmental factors; 2) the shared variation between the Well-Regulated type and externalizing symptoms was only explained by shared and non-shared environmental factors.

The Relation Between Dimensional Temperament and Psychopathology

The strong, consistent negative associations between effortful control and psychopathology symptoms echo previous work suggesting that effortful control is a common promotive factor (Berger, 2011; Kostyrka-Allchorne, et al., 2020; Scheper et al., 2017). Thus, effortful control may be an ideal target mechanism in prevention programs to reduce psychopathology symptoms (e.g., Family Check-Up prevention program increases children’s effortful control and, in turn, decreases internalizing and externalizing symptoms; Hentges et al., 2020). Contrary to previous work, we did not find any associations between negative emotionality or surgency and later psychopathology symptoms (Campagna et al., 2023). One reason for this difference may be that prior work testing associations between negative emotionality or surgency and psychopathology symptoms tend to not include other temperament dimensions in the model (Kostyrka-Allchorne et al., 2020). We conducted three additional SEMs excluding one dimension at a time to examine whether including all three dimensions of temperament explained these differences in results as post-hoc analyses (See Table S8). In sum, the results suggest that associations between negative emotionality and depression symptoms and between surgency and externalizing symptoms identified in prior studies may be accounted for by effortful control. By including all three temperament dimensions in the same model, we examined the extent to which each temperament dimension is individually associated with later psychopathology symptoms.

In stark contrast to the dimensional approach where effortful control alone was associated with all four factors of psychopathology symptoms, all three types were associated with at least two psychopathology factors. The positive associations identified in the current study between the Negative Dysregulated and Exuberant types with externalizing and ADHD symptoms were consistent with the only other study that has examined the relation between temperament types and symptoms of psychopathology (Janson & Mathiesen, 2008). Given that the Well-Regulated and Typical Expressive types were not comparable to any of the five types identified by Janson and Mathiesen (2008), more research using the typological approach to understanding temperamental risk for psychopathology symptoms is needed.

Integrating Dimensions and Types

The most interesting findings emerged when considering the dimensional and typological models together. The dimensional approach provides insight into the associations between temperament and psychopathology symptoms on average while the typological approach examines how temperament patterns are associated with psychopathology symptoms. Thus, integrating findings from both approaches for each outcome may help to clarify the relationship between temperament and psychopathology symptoms. This section synthesizes results across both approaches rather than focusing on any single finding in isolation.

Anxiety Symptoms.

Results from the dimensional and typological approaches supported high effortful control as a key protective factor for anxiety symptoms, while also providing a contrasting perspective compared to prior work. Much of the literature emphasizes the importance of negative emotionality and behavioral inhibition in the development of anxiety symptoms, often overlooking the direct role of effortful control by only including the construct as a moderator, if at all (Nielsen et al., 2019; Kostya-Alchorne et al., 2020; Raines et al., 2023). Yet, effortful control and the Well-Regulated type emerged with the strongest associations with anxiety while negative emotionality and the Negative Dysregulated type yielded weak associations with anxiety. One reason for these contrasting findings may be that earlier studies investigating associations between effortful control and internalizing rarely parsed internalizing symptoms into its more specialized facets of anxiety and depression symptoms (e.g., Eisenberg et al., 2009).

Depression Symptoms.

Effortful control remained a key factor in the association between temperament dimensions and depression, while the typological approach uncovered a more nuanced relationship. Although the Negative Dysregulated and Well-Regulated types had the lowest and highest effortful control, respectively, all three of the temperament types tested were associated with later depression symptoms. This may suggest that effortful control is not the only important facet of temperament in relation to depression symptoms. The negative association found between the Exuberant type (i.e., moderate negative emotionality and effortful control with high surgency) and depression symptoms suggests that surgency also plays a role in risk for depression, reiterating the protective role of surgency in the development of depression symptoms found in prior work (Van Beveren et al., 2019). Given the lack of association found for surgency as a dimension, high surgency may only be protective for symptoms of depression when paired with moderate effortful control and moderate negative emotionality. A study investigating interactions between surgency and effortful control predicting depression symptoms over time yielded a similar pattern (Kotelnikova et al., 2014).

Externalizing Symptoms.

Despite the emphasis on negative emotionality and surgency as risk factors for externalizing symptoms in prior work, effortful control was the only dimension associated with externalizing symptoms in the current study (Campagna et al., 2023; Scheper et al., 2017). This discrepancy may be further clarified by taking a closer look at the typological approach. The Negative Dysregulated type was positively related with externalizing symptoms while the Well-Regulated type was negatively associated with externalizing symptoms. Although the Exuberant type, which has the highest surgency, was not associated with externalizing symptoms, this does not necessarily imply that surgency is not important. In fact, the Negative Dysregulated type, which had the highest impulsivity (i.e., one facet of surgency), had the largest effect in predicting later externalizing symptoms. Thus, findings may suggest that effortful control may be important for externalizing symptoms more broadly (i.e., at both high and low levels), but the combination of low effortful control, high negative emotionality, and high impulsivity may be a notable risk factor for externalizing symptoms.

ADHD Symptoms.

In the case of ADHD symptoms, effortful control had stronger associations than any other dimension or type suggesting that low effortful control is a risk factor for ADHD symptoms regardless of other temperament dimensions, consistent with prior work (Kostyrka-Allchorne et al., 2020). These findings are reiterated further when considering the results from the typological approach, where the Negative Dysregulated and Exuberant types were both positively associated with ADHD symptoms. Interestingly, despite being particularly high on effortful control and low on impulsivity, the Well-Regulated type was not associated with ADHD symptoms, which may suggest that high effortful control may not necessarily be a protective factor for ADHD symptoms. Results also suggest that high surgency may be a risk factor for ADHD symptoms, echoing previous work (Kostyrka-Allchorne et al., 2020), but only when combined with low-to-moderate effortful control and negative emotionality.

Heritability of Temperament

Regardless of the method used to capture temperament, both dimensions and types were at least partly influenced by genetic factors, consistent with the commonly held notion that temperament is biologically-based (Shiner et al., 2012). Like prior twin studies examining the heritability of temperament (Saudino, 2009), negative emotionality, effortful control, and surgency had moderate to large genetic influences. Notably, negative emotionality was the only dimension found to have significant shared environmental influences which is consistent with prior work (Clifford et al., 2015).

Like the dimensional approach, the typological approach illustrated a consistent role of genetics, with all four types having moderate to large genetic influences and a small to moderate role of the nonshared environment. These results complement the one other study with an independent sample that has examined the heritability of temperament types (Scott et al., 2016); however, the role of the shared environment was largely inconsistent across studies. For example, Scott and colleagues (2016) identified a moderate to large role of the shared environment for their Negative Dysregulated and Typical Expressive types and no role of the shared environment for the Well-Regulated and Surgent types. In contrast, our study only identified shared environmental influences for the Well-Regulated type. Although all four profiles had similar characteristics across the two studies, the samples were different, with our sample more racially/ethnically diverse and lower income. These inconsistencies between studies support sociocultural influences on temperament (Putnam et al., 2024) that should be studied in the future within the context of genetically-informed designs such as twin studies.

The typological approach uncovered a moderate role of the shared environment in influencing the Well-Regulated type despite the small to nonexistent role of the shared environment in temperament found in other studies when using the dimensional approach. Out of the three dimensions of temperament evaluated in this study, only negative emotionality was influenced by the shared environment. Even though both Negative Dysregulated and Well-Regulated types have particularly high and low negative emotionality respectively, only the Well-Regulated type was also influenced by the shared environment. Thus, low negative emotionality may have shared environmental influences while high negative emotionality may not, a potential indication of gene-environment interplay (Gjone et al., 1996). However, we did not fully isolate high and low negative emotionality in our genetic analyses and this was the first study to consider the results of genetic analyses using dimensional and typological temperament together. Therefore, these results should be interpreted with caution and future studies should attempt replication.

Heritability of Psychopathology

While our results suggested that all factors of psychopathology symptoms were primarily explained by genetic factors, consistent with prior work (Burt, 2009; Faraone & Larsson, 2018), our estimates on the role of the shared environment differed from the previous literature. A meta-analysis that spanned 490 twin and adoption studies on the heritability of a single reporter on symptoms of psychopathology found that 11–31%, 10–14%, and 15% of the variance in symptoms of anxiety, depression, and externalizing was explained by shared environmental factors respectively (Burt, 2009). In the current study, anxiety and externalizing symptoms were the only two factors explained by shared environmental factors, and those estimates were above the range estimated in the meta-analysis. Importantly, the broader sociocultural context and sample characteristics play a major role in the results of twin studies (Shanahan & Hofer, 2005). When environments that are relevant to the trait being studied have less variability, estimates for the role of the environment can be low. Compared to previous work, our sample was particularly socioeconomically and racially/ethnically diverse, which may explain the larger estimates of shared environmental influences on anxiety symptoms, externalizing symptoms, negative emotionality, and the Well-Regulated type. As such, current findings emphasize the need for genetically informed studies on temperament and symptoms of psychopathology using a range of methods (e.g., person- and variable-centered approaches, primary caregiver and teacher reported measurements) and more diverse samples across differing contexts.

Inconsistent with prior work where large dominant genetic effects were identified for symptoms of ADHD, dominant genetic influences were not present for ADHD symptoms (Burt, 2009; Faraone & Larsson, 2018). This may be because the factors of psychopathology symptoms in this study were created by using a CTCM-1 model, which removed measurement variance unique to secondary caregivers and teachers. Further research is required to investigate why estimates of dominant genetic influences on symptoms of ADHD are not consistently observed using different methodological and estimation approaches.

Genetic Etiology of the Relation Between Temperament and Psychopathology

The final aim of the current study was to elucidate the etiological mechanisms underlying the relation between temperament and symptoms of psychopathology using the dimensional and typological approaches to temperament. We found evidence for both genetic and environmental influences depending on the analytic method used with shared etiological mechanisms to some degree. Overall, our findings suggest that genetics play a large role in explaining the relation between temperament and symptoms of psychopathology while the results from the typological approach also highlight the importance of the shared environment.

Beginning with anxiety symptoms, shared genetic and nonshared environmental influences were found for the negative association with effortful control and only shared environmental influences explained the negative association with the Well-Regulated type. Only one study examined the genetic etiology of the association between effortful control and anxiety symptoms, capturing anxiety along with broader internalizing symptoms, which found small to moderate genetic influences similar to the findings of the current study (Lemery-Chalfant et al., 2008). Interestingly, common shared environmental influences between temperament and internalizing symptoms have not been consistently identified in prior work (Lemery-Chalfant et al., 2008; Wang & Saudino, 2015). Yet, our results suggested a large, common influence of the shared environment between the Well-Regulated type and anxiety symptoms. Future work will want to replicate these findings and unpack possible shared environmental factors that could explain this association.

Turning to depression symptoms, its negative associations with temperament were primarily a result of genetic factors, except in the case of the Exuberant type, which is consistent with the literature (Lemery-Chalfant et al., 2008; Wang & Saudino, 2015). Only nonshared environmental influences accounted for the negative association between the Exuberant type and depression symptoms. These results may suggest that high surgency paired with moderate negative emotionality and effortful control is associated with depression symptoms due to nonshared environmental reasons. Behavioral genetic work examining surgency, a key feature of the Exuberant type, and depression symptoms has been limited (Wang & Saudino, 2015). Thus, more work needs to be conducted in this area to better understand when, why, and how surgency is linked to symptoms of depression.

For externalizing symptoms, common genetic influences explained negative associations across effortful control and the Well-Regulated type and the positive association with the Negative Dysregulated type, which is consistent with prior work using the dimensional approach (e.g., Lemery-Chalfant et al., 2008), with one exception. Specifically, in the case of the Well-Regulated type, shared environmental influences overlapped substantially with that of externalizing symptoms despite being genetically independent of one another. This finding is particularly notable given the lack of consistency for the role of the shared environment in explaining covariance between temperament and externalizing symptoms (e.g., Vasin & Lobaskova, 2016). Thus, results suggest that the shared environment may only explain covariance between temperament and externalizing symptoms when high effortful control is paired with low negative emotionality and moderate surgency. Compared to the other types, the Negative Dysregulated type shared the most genetic influences with externalizing symptoms suggesting that while individuals with low effortful control have a higher genetic risk for externalizing symptoms, this effect may be exacerbated when low effortful control is coupled with high reactivity and high impulsivity (Berger, 2011).

Finally, ADHD symptoms shared common genetic influences across effortful control and temperament types, which is consistent with prior work (Berger, 2011; Martel et al., 2022). Some studies have suggested that effortful control is linked to ADHD symptoms for genetic reasons while surgency is linked to ADHD symptoms for shared environmental reasons (Martel et al., 2022). This stands in contrast to the current study’s findings that only effortful control was negatively associated with ADHD symptoms for genetic reasons and surgency was not related to ADHD. The results from the typological approach may provide a more nuanced understanding of this inconsistency. Whereas the Well-Regulated type was not associated with ADHD symptoms, the Exuberant type was positively associated with ADHD symptoms for genetic reasons. Together, these findings may provide initial evidence that surgency is only linked to symptoms of ADHD for genetic reasons only when high surgency is coupled with moderate effortful control and negative emotionality.

Across all temperament types and psychopathology outcomes, the Negative Dysregulated type was linked with depression, externalizing, and ADHD symptoms and the Typical Expressive type was linked with externalizing and ADHD symptoms primarily for genetic reasons. Further, the genetic factors explaining the variation in ADHD symptoms were completely shared with the Typical Expressive type. These findings may suggest that children who exhibit the Negative Dysregulated or Typical Expressive types have higher genetic risk for psychopathology symptoms compared to the Well-Regulated or Exuberant types where the environment plays a larger role. Future research should investigate possible unmodelled gene-environment interactions that could inflate the role of genetics and minimize the role of the environment (Purcell, 2002).

Contributions of a Multiple Reporter Approach

The current study also added to the literature by using a unique methodological approach to measure psychopathology symptoms across primary caregiver, secondary caregiver, and teacher-reports. By utilizing a multiple-reporter approach to assess symptoms of anxiety, depression, externalizing, and ADHD separately, we found that the genetic etiology of the relation between temperament and symptoms of internalizing psychopathology changed depending on whether symptoms of anxiety or depression were examined. The same pattern held true for externalizing and ADHD symptoms. These findings are consistent with prior work that has found differences in the genetic and environmental underpinnings of psychopathology symptoms (Burt, 2009; Faraone & Larsson, 2018; Du Rietz et al., 2021). Thus, assessing different facets of psychopathology symptoms (i.e., anxiety, depression, externalizing, and ADHD) can add specificity to our understanding of the associations between temperament and symptoms of psychopathology.

Study Limitations and Future Directions

The results of the current study should be considered in the context of the limitations associated with the statistical methods. For example, a multiple-reporter approach was used to measure psychopathology symptoms, but the same approach was not used for temperament. In fact, the most recent studies examining temperament types with advanced statistical methods to date have only used parent or observer reports (e.g., Planalp & Goldsmith, 2020; Scott et al., 2016). Future studies should examine whether and how temperament types and their relation to symptoms of psychopathology differ based on reporter. To create multiple-reporter factors of psychopathology, we used an innovative method, the CTCM-1 model, which prevented issues with overfitting and model identification. Primary caregiver report was used as the comparison standard in the model and, thus, measurement error solely related to primary caregiver report was still incorporated into the resulting factors. Using a different comparison standard than the one chosen may cause the results of the current study to change due to the asymmetrical nature of the CTCM-1 method (Eid et al., 2008).

Although it is the most common method used in other studies that have identified temperament types, LPA is only one of many mixture modeling techniques that can be used to identify common heterogeneous patterns among several correlated and uncorrelated indicators, (Beekman et al., 2015; Gartstein et al. 2017). The number of types identified by LPA can change depending on model-fit selection criteria, a study’s sample size, and chosen measures of temperament (Spurk et al., 2020). Notably, all four of the temperament types used in the current study have been replicated in other studies (Beekman et al., 2015; Scott et al., 2016).

While the twin study design served as a major strength of the present study, prior research has shown that the genetic and environmental influences that explain the variation in temperament can vary depending on the environment (e.g., chaotic homes; Lemery-Chalfant et al., 2013). The univariate twin model and Cholesky Decomposition do not account for gene-environment interplay, which could inflate or underestimate genetic and environmental influences in particular contexts. Estimates of genetic and environmental influences from both types of biometric models can also change depending on sample characteristics (Shanahan & Hofer, 2005). As such, it will be important for future work to investigate whether different aspects of the environment moderate the heritability of children’s temperament and to replicate the unique findings of the current study in different samples.

Conclusion

The present study examined the dimensional and typological approaches to temperament under the overarching aim of investigating temperamental risk for common types of developmental psychopathology symptoms. The classic twin method was used to uncover the genetic and environmental etiology of these relations. In doing so, we uncovered a larger role of the shared environment for temperament and symptoms of psychopathology independently and in their covariance than previous research has suggested. Our results highlight the need to identify aspects of the shared environment that increase the probability of belonging to the Well-Regulated type and those that may affect the heritability of effortful control and high negative emotionality for future interventions aiming to increase resilience and reduce psychopathology symptoms.

Supplementary Material

Supplemental material: All relevant code and materials are available in at https://osf.io/8qv2f/.