Overall Methodology

Ongoing SITSS testing includes bringing multiples aged 1–5 years to a campus laboratory every year within approximately a month of their birthday. Table 1 presents a summary of measure constructs from all SITSS testing paradigms. Buccal cell data collection began in 2003, and DNA data are now available for 500 children (233 twin-pairs, 10 triplet sets and 1 quadruplet set). These data have been used to confirm zygosity status (we have a 96.4% agreement between DNA and parent and tester ratings of twin similarity for same-sex pairs), and genotyping is available for dopamine receptor D4 gene (DRD4), DRD2, 5HTTLPR, and CHRM2-rs36210735 and CHRM2-rs1824024 single nucleotide polymorphisms (SNPs) for many of the children. Data are stored for future analyses but are not publicly accessible.

Table 1. Summary of SITSS data collected

Note: MO/LT = mailed out, then collected when families visit the laboratory for testing; LA = lab assessment; MS = mail survey; LQ = lab questionnaire; TI = telephone interview; OB = observed assessment.

Various follow-up studies have utilized mailed-out questionnaires and telephone interviews or have involved bringing families back to the lab when the children are older. Studies have examined victimization, bullying, psychopathology symptoms, health behaviors and parenting. Each follow-up study also contains extensive information on temperament, family environment, and child behaviors. A detailed summary of some of the measures collected as part of these follow-up projects can be found in DiLalla et al. (Reference DiLalla, Gheyara and Bersted2013) and are briefly summarized in Table 1. Our most recently completed follow-up studies of 6- to 13-year-old twins and triplets involved in-depth interviews on emotion skills, theory of mind, and interpersonal functioning, including sleep problems, anxiety during peer interactions, victimization and bullying, and health behaviors. Constructs that we have tested that were not included in the earlier table (DiLalla et al., Reference DiLalla, Gheyara and Bersted2013) are summarized in Table 2.

Table 2. Recent SITSS follow-up measures administered

Note: CR = child report; PR = parent report; TO = tester observed/rated; LQ = lab questionnaire; LA = lab assessment; TI = telephone interview; OB = observed assessment.

Study Goals

The initial goal of the SITSS was to examine genetic and environmental causes of early aggression in preschoolers. At the time, we were beginning to understand something about delinquency and criminality, but we knew very little about causes of early aggression and rule-breaking. From the original peer-play paradigm, the SITSS expanded to include early preschool cognitive and social development. More recently, the focus of SITSS has been on emotional skills from preschool through adolescence, as well as implications for bullying and victimization behaviors. We have also begun exploring related behaviors, such as anxiety and physical health behaviors. Genetic factors have been implicated in some early childhood social behaviors, but less is known about the etiology of emotional intelligence, social perspective-taking, and prosocial and aggressive behaviors in preschoolers (DiLalla, Reference DiLalla2002b), although this field is beginning to grow.

The important focus of twin studies now is to examine how genes and behaviors coact. Particular genes have been identified via molecular genetics studies that are related to a large number of behaviors and disorders, although concerns about replication must be addressed because each gene can only account for a tiny fraction of the variance for any given trait (Dick et al., Reference Dick, Agrawal, Keller, Adkins, Aliev, Monroe and Sher2015). The SITSS integrates biological and environmental perspectives in investigating the causes of young children’s social development. The early focus was on heritability but has grown to include genotype and a number of developmental and environmental variables, such as daycare experiences, parent–child interaction styles and cognitive ability. We have used a variety of methodologies to investigate gene–environment correlations (rGE) as well as gene–environment interactions (G×E). Genetic likelihood indices have been calculated using methodology described in Jaffee et al. (Reference Jaffee, Caspi, Moffitt, Dodge, Rutter, Taylor and Tully2005) to help examine rGE. Multilevel modeling (MLM) techniques have been utilized, allowing inclusion of multiple children within each family in a single–multiple regression analysis. MLM analyses have also included measures of zygosity and its interactions with individual twins’ deviance scores from the family mean, using methodology described by Turkheimer and Harden (Reference Turkheimer, Harden, Reis and Judd2014), providing estimates of causality and G×E.

Recent Research with the SITSS Sample

Recent studies utilizing children from the SITSS have focused on both internalizing and externalizing problem behaviors, examining data on children from ages 5 to 18 years. Much of our work has included data from our original peer-play study, described above. The most important aspect of our peer-play paradigm was that we controlled for passive rGE, because the parents had no control over their children’s randomly chosen peer partners, as well as active rGE, because the children themselves also had no control over their play partners. Thus, correlations between the children had to result from evocative rGE (DiLalla & John, Reference DiLalla and John2014).

We took advantage of this peer-play paradigm in a recent study presented initially to the International Society for Twin Studies in Dr. Irving I. Gottesman’s honor. (He is the academic grandfather of the SITSS.) We utilized a genetic risk index for rule-breaking behaviors at age 5, created as a function of parent ratings of early rule-breaking and co-twin relatedness (MZ vs. DZ) in a manner similar to Jaffee et al. (Reference Jaffee, Caspi, Moffitt, Dodge, Rutter, Taylor and Tully2005). We showed that children at genetic risk for rule breaking were more likely to elicit aggression from novel peer-play partners during the peer-play situation, suggesting the importance of evocative rGE in young children’s play scenarios (DiLalla & DiLalla, Reference DiLalla and DiLalla2018). Similarly, children at genetic risk for behaving aggressively with a peer, coded during the 5-year-old peer play, were more likely to elicit aggression from their play partner (DiLalla & John, Reference DiLalla and John2014). We also demonstrated evocative rGE for prosocial behaviors using this same genetic likelihood index method (DiLalla et al., Reference DiLalla, Bersted and John2015a). Children with a genetic likelihood for behaving in prosocial ways during the peer play and who are also temperamentally outgoing, as rated by parents, were more likely to evoke more prosocial and easy-going behaviors from an unfamiliar play partner. Thus, these data have been instrumental in demonstrating that young children’s behaviors elicit similar behaviors from same-age peers and that these eliciting behaviors are related to their genetic make-up (i.e., evocative rGE).

Other measures of externalizing behaviors have also been considered. In one study, we showed that parent ratings of 5-year-old children’s aggressive behaviors were related to the amount of physical aggression displayed in television shows and computer games that those children preferred, as measured by self-report, and this was partially mediated by shared genes (Jamnik & DiLalla, Reference Jamnik and DiLalla2018). Importantly, this study examined children’s preferences rather than their responses to manipulations of exposure to shows or games, and in this way is more naturalistic than manipulation studies.

Finally, two recent studies of externalizing behaviors in the SITSS sample have extended earlier findings regarding DRD4 effects. We found evidence for differential susceptibility when examining child-reported victimization and externalizing behaviors in 6- to 10-year-olds (DiLalla et al., Reference DiLalla, Bersted and John2015b). Children with the 7-repeat allele (DRD4-7r) who did not report having been victimized had the lowest levels of externalizing behaviors, but DRD4-7r children who did report being victimized reported the highest levels of externalizing behaviors. Interestingly, we also demonstrated that 5-year-old children without the 7-repeat allele for DRD4 (DRD4-no7r) showed differential susceptibility to prenatal environmental stressors. DRD4-no7r children with few moderate birth complications showed low-negative emotionality, but DRD4-no7r children with many moderate birth complications were rated as the highest on negative emotionality (Bersted & DiLalla, Reference Bersted and DiLalla2016). Because prenatal birth complications are different from transactional child–parent or peer interactions, the role of dopamine may be quite different (less related to the brain’s reward pathways during interactions). Therefore, this study may highlight some important new avenues of research for the role of dopamine on the developing brain.

Other aspects of child development have also been examined with the SITSS. One study of children’s perceived competence (Klaver et al., Reference Klaver, Palo and DiLalla2014) showed that children who overreported their own self-competence at age 5, as compared to objective ratings, were more likely to have more parent-reported externalizing and internalizing problems. Another study examined parent sensitivity and 4-year-old children’s responsiveness minute by minute during a 10-min interaction in the lab (Elam & DiLalla, Reference Elam and DiLalla2018). Using latent growth curve models, we found that low parent sensitivity was predictive of child unresponsiveness only in DRD4-7r children, thus demonstrating G×E as well as the importance of understanding minute-by-minute changes in parent and child interactional behaviors. Finally, we showed that age 5 parent-reported internalizing problems predicted youth-reported physical health problems as well as BMI when youth were between the ages of 12 and 20 years, thus demonstrating the importance of including preschool temperament and internalizing problems when predicting adolescent health behaviors (Jamnik & DiLalla, Reference Jamnik and DiLalla2019).