Research on the role of children's temperament in the development of behavior problems, and more generally, adjustment outcomes, has a long and rich history. Since the early Thomas and Chess’ New York Longitudinal Study (Thomas & Chess, Reference Thomas and Chess1977), rapidly growing bodies of literature have delineated complex influences of early temperament on future adjustment. Rothbart and Bates (Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006) presented a comprehensive overview of processes that may link early temperament to future adjustment, encompassing four main classes of effects: direct and indirect linear effects, Temperament × Environment interactions, and Temperament ×Temperament interactions (as well as miscellaneous effects). Enormous bodies of research that have since emerged can be largely mapped onto those classes of effects, although doing so would greatly exceed the scope of the current article.

Rothbart and Bates (Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006) noted that whereas research pertaining to the first three classes of processes had grown exponentially, especially on Temperament × Environment interactions, the last category – Temperament × Temperament interactions – had been accumulating much more slowly. This assessment is generally still accurate today. Consequently, the study of such effects is a useful scientific enterprise, particularly in developmental psychopathology. One possible type of those interactions is particularly relevant, when “One temperament trait can protect against risk consequences of another temperament-based trait” (Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006, p. 137). We would add that one temperament trait can also exacerbate risks due to another trait.

Of note, the idea of an interplay among personality traits has a long history. Already in 1980, Block and Block, inspired by the Lewinian perspective, presented a then-innovative and since very influential approach to the development of individual differences that emphasized two broad personality parameters: ego control and ego resiliency (Block & Block, Reference Block, Block and Collins1980). They assessed those parameters using observations of children's behavior in large behavioral batteries and teachers’ reports in the California Q-set (CCQ; Block & Block, Reference Block and Block1969; a set of 100 diverse personality traits). The construct of ego control emphasized characteristics such as ability to delay gratification, to inhibit impulses, to plan and think ahead, or to resist temptation. Over-control and under-control represented the high and low ends of this dimension, respectively. Both very high and very low scores were considered maladaptive, with mid-range scores seen as optimal. The construct of ego resiliency incorporated a number of personality characteristics. Flexible and adaptive emotion regulation, particularly when frustrated or stressed, appeared key. Resourceful adaptation to changing circumstances and environmental contingencies, including a capacity to regulate flexibly ego control as required in a specific context, represented high ego resiliency. “Brittleness” or “going to pieces under stress” represented the low end. Higher levels of ego resiliency were considered adaptive.

Certain characteristics reflected both ego control and ego resiliency; for example, planfulness or ability to delay gratification (Block & Block, Reference Block, Block and Collins1980). Importantly, the Blocks depicted the interplay of the two constructs in a 2 × 2 matrix illustrating the “personality profiles” that corresponded to the combinations of high and low levels of each trait (Block & Block, Reference Block, Block and Collins1980, p. 89). Ever since, many scholars of children's temperament and personality development have referred to those two concepts, particularly when studying regulatory processes. Both ego control and ego resiliency are often cited in the context of discussion of regulatory mechanisms of temperament (e.g., Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006), although various scholars’ interpretations are not always fully consistent.

In the current article, we focus on two temperament-based inhibition systems. One broad system is associated with reactive inhibition, or behavioral inhibition (BI). Kagan and his colleagues have conducted extensive classic research on this trait, documenting its early origins, stability over time, and biological underpinnings and correlates (e.g., Kagan & Fox, Reference Kagan, Fox, Damon, Lerner and Eisenberg2006; Kagan, Reznick, Clarke, Snidman, & Garcia Coll, Reference Kagan, Reznick, Clarke, Snidman and Garcia Coll1984; Kagan & Snidman, Reference Kagan and Snidman2004). BI is a robust temperament characteristic, typically studied in paradigms involving unfamiliar persons and events, and mildly stressful or challenging behavioral probes (“Risk Room” and “risky acts”). Inhibited children show a consistent pattern of caution, fearfulness, reticence, and shyness; in contrast, uninhibited children readily explore and eagerly engage in “risky” behaviors (e.g., play with an odd toy, crawl through a tunnel, approach a person in a costume).

The other broad system is associated with active inhibition, or ability deliberately to suppress a dominant behavior – often hedonically rewarding and much desired – and execute instead a subdominant behavior, even if mundane or unappealing. Several inter-related massive literatures refer to this system as effortful control (EC), inhibitory control, executive function, self-control, self-regulation, or emotion regulation, to mention only a few (Duckworth & Kern, Reference Duckworth and Kern2011; Eisenberg, Smith, Sadovsky, & Spinrad, Reference Eisenberg, Smith, Sadovsky, Spinrad, Baumeister and Vohs2004; Kochanska, Murray, & Harlan, Reference Kochanska, Murray and Harlan2000; Moffitt et al., Reference Moffitt, Arseneault, Belsky, Dickson, Hancox, Harrington and Caspi2011; Nigg, Reference Nigg2017; Posner & Rothbart, Reference Posner and Rothbart2000; Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006; Rueda, Reference Rueda, Zentner and Shiner2012). Often, researchers have distinguished “hot” and “cool” aspects of EC (e.g., Brock, Rimm-Kaufman, Nathanson, & Grimm, Reference Brock, Rimm-Kaufman, Nathanson and Grimm2009; Kim, Nordling, Yoon, Boldt, & Kochanska, Reference Kim, Nordling, Yoon, Boldt and Kochanska2013; Willoughby, Kupersmidt, Voegler-Lee, & Bryant, Reference Willoughby, Kupersmidt, Voegler-Lee and Bryant2011; Zelazo & Carlson, Reference Zelazo and Carlson2012). Hot EC has been typically assessed in batteries of tasks that call for suppressing a dominant behavior (e.g., delaying a highly appealing, desired action), whereas cool EC has been often measured in more cognitive, executive function tasks.

In developmental psychopathology, both BI and EC in childhood have been extensively studied in the context of risk for future disorders (Nigg, Reference Nigg2000). Kagan's group and many other researchers have demonstrated that children at the high end of BI are at risk for internalizing disorders, mostly anxiety (e.g., Biederman et al., Reference Biederman, Hirshfeld-Becker, Rosenbaum, Hérot, Friedman, Snidman and Faraone2001; Chen & Schmidt, Reference Chen, Schmidt, Lerner and Lamb2015; Degnan, Almas, & Fox, Reference Degnan, Almas and Fox2010; Fox, Henderson, Marshall, Nichols, & Ghera, Reference Fox, Henderson, Marshall, Nichols and Ghera2005; Kagan & Fox, Reference Kagan, Fox, Damon, Lerner and Eisenberg2006; Klein, Dyson, Kujawa, & Kotov, Reference Klein, Dyson, Kujawa, Kotov, Zentner and Shiner2012; Klein & Mumper, Reference Klein, Mumper, Perez-Edgar and Fox2018; Liu & Pérez-Edgar, Reference Liu and Pérez-Edgar2019; Sandstrom, Uher, & Pavlova, Reference Sandstrom, Uher and Pavlova2020).

Very low levels of BI, however, may also pose developmental risks. Fearlessness, or low BI, has been robustly established as a risk for a range of future disruptive problems, including aggression, disregard for rules and others’ welfare, callous unemotional traits, and oppositional and conduct problems (e.g., Biederman et al., Reference Biederman, Hirshfeld-Becker, Rosenbaum, Hérot, Friedman, Snidman and Faraone2001; Calkins, Blandon, Williford, & Keane, Reference Calkins, Blandon, Williford and Keane2007; Gao, Raine, Venables, Dawson, & Mednick, Reference Gao, Raine, Venables, Dawson and Mednick2010; Lykken, Reference Lykken1995; Patrick et al., Reference Patrick, Kramer, Vaidyanathan, Benning, Hicks and Lilienfeld2019; Raine, Reynolds, Venables, Mednick, & Farrington, Reference Raine, Reynolds, Venables, Mednick and Farrington1998; Shaw, Gilliom, Ingoldsby, & Nagin, Reference Shaw, Gilliom, Ingoldsby and Nagin2003; Tackett, Martel, & Kushner, Reference Tackett, Martel, Kushner, Zentner and Shiner2012).

Several researchers have studied characteristics that overlap with the low end of BI, although they considered those characteristics as reflecting a separate temperament system rather than low BI. Fox and colleagues (e.g., Fox, Henderson, Rubin, Calkins, & Schmidt, Reference Fox, Henderson, Rubin, Calkins and Schmidt2001) introduced a temperament dimension labeled “exuberance.” Although exuberant children share common characteristics with children low on BI (high sociability, approach, low fear), Degnan et al. (Reference Degnan, Hane, Henderson, Moas, Reeb-Sutherland and Fox2011) emphasized that BI and exuberance are distinct temperamental dimensions, with the latter described as a combination of high approach and positive affect. Empirically, exuberance is, in part, assessed in the classic Risk Room episodes (coded for approach vs. inhibition, and positive and negative affect, Putnam & Stifter, Reference Putnam and Stifter2005; Stifter, Putnam, & Jahromi, Reference Stifter, Putnam and Jahromi2008).

Buss, Kiel, Morales, and Robinson (Reference Buss, Kiel, Morales and Robinson2014) introduced a concept of “bold approach.” Bold approach was based on observations of young children's behavior in several episodes, including Risk Room episodes; high bold approach represented essentially the low end of BI. Honomichl and Donnellan (Reference Honomichl and Donnellan2012) examined a dimension of Surgency extracted from mothers’ reports of their preschool children's temperament (high approach, high activity, low shyness). Across those studies, the researchers generally reported that high exuberance, bold approach, or Surgency predicted observed and informant- and self-reported measures of externalizing, disruptive behavior problems. In this context, two additional points should be made. One, although the concepts of low BI, high exuberance, bold approach, and Surgency do overlap, they are not the same, and distinctions among them need to be acknowledged. Two, high scores on those dimensions may be associated with positive social behaviors as well (e.g., sociability, friendliness).

With regard to EC, massive literature has linked its low levels with risks for externalizing psychopathology – the development of disregard for rules and externalizing problems (Eisenberg et al., Reference Eisenberg, Smith, Sadovsky, Spinrad, Baumeister and Vohs2004; Morales et al., Reference Morales, Miller, Troller-Renfree, White, Degnan, Henderson and Fox2019; Morales, Pérez-Edgar, & Buss, Reference Morales, Pérez-Edgar and Buss2016; Nigg, Reference Nigg2017; Olson, Sameroff, Kerr, & Lopez, Reference Olson, Sameroff, Kerr and Lopez2005; Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006). EC generally predicts positive adjustment (Kochanska & Knaack, Reference Kochanska and Knaack2003; Kochanska et al., Reference Kochanska, Murray and Harlan2000; Moffitt et al., Reference Moffitt, Arseneault, Belsky, Dickson, Hancox, Harrington and Caspi2011; Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006).

Several bodies of research have examined interactions between the two inhibition systems, including mutually compensatory interactions proposed by Rothbart and Bates (Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006), in which a (high or low) score on one temperament trait can offset a child's high risk due to a (high or low) score on another trait. Several studies have examined interplay of BI and EC with regard to the risk for internalizing problems, particularly anxiety. Anxiety is associated with a high level of BI, but such risk can be offset by a high level of EC (Degnan & Fox, Reference Degnan and Fox2007; Oldehinkel, Hartman, Ferdinand, Verhulst, & Ormel, Reference Oldehinkel, Hartman, Ferdinand, Verhulst and Ormel2007; Nystrom & Bengtsson, Reference Nystrom and Bengtsson2016). Of note, the effects have often been complex, and depended on the type of EC processes that were studied. Certain forms of EC, such as the ability to control and shift attention, seem to attenuate the risk for internalizing problems among children with high BI, whereas some other forms, such as inhibitory control, can exacerbate the risk (Pérez-Edgar, Taber-Thomas, Auday, & Morales, Reference Pérez-Edgar, Taber-Thomas, Auday, Morales and Lagattuta2014; Sportel, Nauta, de Hullu, de Jong, & Hartman, Reference Sportel, Nauta, de Hullu, de Jong and Hartman2011; Thorell, Bohlin, & Rydell, Reference Thorell, Bohlin and Rydell2004; White, McDermott, Degnan, Henderson, & Fox, Reference White, McDermott, Degnan, Henderson and Fox2011). Overall, the findings consistently point to an adaptive role of EC, as long as the child is able to recruit it for a proactive regulation of fear (rather than becoming over-inhibited and inflexible; Buzzell, Troller-Renfree, Morales, & Fox, Reference Buzzell, Troller-Renfree, Morales, Fox, Pérez-Edgar and Fox2018; Hassan, Poole, & Schmidt, Reference Hassan, Poole and Schmidt2020; Troller-Renfree et al., Reference Troller-Renfree, Buzzell, Bowers, Salo, Forman-Alberti, Smith and Fox2019; Troller-Renfree, Buzzell, Pine, Henderson, & Fox, Reference Troller-Renfree, Buzzell, Pine, Henderson and Fox2019).

Some studies have demonstrated that EC had similar protective effects for children with heightened negative emotionality (Rothbart & Bates, Reference Rothbart, Bates, Damon, Lerner and Eisenberg2006). Negative emotionality, sometimes described as difficult temperament, is a broader trait, which, in addition to fearfulness or BI, often includes proneness to other negative emotions, such as sadness or anger (Lonigan & Vasey, Reference Lonigan and Vasey2009; Muris & Ollendick, Reference Muris and Ollendick2005). Those studies have often focused on children's externalizing problems as outcomes, including future aggression, disregard for rules, or disruptive conduct. A large body of research, using a broad range of measures, has persuasively demonstrated that EC can interact with negative emotionality in predicting future disruptive outcomes, such that high EC can offset risks due to the high proneness to various negative emotions (e.g., Colder & Stice, Reference Colder and Stice1998; Eisenberg & Fabes, Reference Eisenberg, Fabes and Clark1992; Eisenberg et al., Reference Eisenberg, Guthrie, Fabes, Shepard, Losoya, Murphy and Reiser2000, Reference Eisenberg, Valiente, Spinrad, Cumberland, Liew and Reiser2009; Nielsen, Olino, Dyson, & Klein, Reference Nielsen, Olino, Dyson and Klein2019). A typical interpretation of such effects has generally emphasized that children can recruit EC to help them regulate negative emotions – mostly anger. EC can help prevent those emotions from becoming disruptive and dysfunctional, as ultimately reflected in conduct problems.

Multiple comprehensive models have explicitly identified low levels of both BI-related traits (low reactive inhibition, fearlessness, boldness, risk-taking) and EC-related traits (low active inhibition, poor capacity for self-regulation of behavior or emotion) as triggering two key pathways to antisocial or disruptive problems for children with low scores (Eisenberg & Fabes, Reference Eisenberg, Fabes and Clark1992; Fowles & Dindo, Reference Fowles and Dindo2009; Frick & Morris, Reference Frick and Morris2004; Nielsen et al., Reference Nielsen, Olino, Dyson and Klein2019; Nigg, Reference Nigg2006; Reference Nigg2017). Conversely, both high BI and high EC have been implicated as temperamental substrates of adaptive pathways to children's mature conscience and sociomoral competence, including internalization of standards and values (Kochanska, Reference Kochanska1993; Kochanska, Coy, & Murray, Reference Kochanska, Coy and Murray2001).

Fewer studies have examined whether EC and BI interact in predicting externalizing conduct problems such that high EC could potentially buffer the risks conferred by low BI, or fearlessness, and whether risks due to deficient EC, amply implicated in the development of disregard for rules and externalizing symptoms, can be buffered by high levels of BI. As well, little research has examined mutually exacerbating effects: Whether low EC can further increase the risk for externalizing problems for children with low BI, and vice versa.

However, this literature has been growing. Thorell et al. (Reference Thorell, Bohlin and Rydell2004) found that the combination of low EC and low-to-average BI was associated with heightened levels of hyperactivity, but the risk for hyperactivity was mitigated among children who were high on at least one of the dimensions (i.e., low EC and high BI, or high EC and low BI).

Researchers studying exuberance and the related characteristics have documented interactions between exuberance or bold approach and various forms of EC with regard to the development of externalizing problems. Stifter et al. (Reference Stifter, Putnam and Jahromi2008) found that emotion regulatory capacities, assessed in a disappointment paradigm, decreased the presence of preschool problem behaviors for exuberant children, although no such protective effects were found for classic EC tasks involving delay. Dollar, Stifter, and Buss (Reference Dollar, Stifter and Buss2017) applied latent group analysis to observations of toddlers in the Risk Room procedure, and identified inhibited, exuberant, and average groups. Mothers rated toddlers’ inhibitory control and social adjustment (in peer relations) 2 years later. Their findings were consistent with Stifter et al. (Reference Stifter, Putnam and Jahromi2008): As inhibitory control abilities increased, exuberant children were rated by their mothers as higher in peer adjustment. Buss et al. (2014) examined interactions between inhibitory control (essentially equivalent of EC) and bold approach and reported a relatively consistent pattern of results across several measures of externalizing behavior problems, such that low inhibitory control and high approach put children at risk for developing externalizing behavior problems. They stated that “Bold approach was a significant moderator in every analysis such that being low in bold approach was a protective factor for children low in inhibitory control” (p. 243).

We continue to examine the interplay of the two temperament inhibitory systems, reactive (BI) and active (EC) in the current work. We present two longitudinal studies of typically developing young children, both examining BI, EC, and their interaction as predictors of future outcomes. Both studies employed fully parallel measures of BI and EC. In the first study, we observed disregard for parental rule in a naturalistic prohibition context as the outcome. In the second study, we examined mothers’ and fathers’ reports of externalizing problems.

To strengthen the potential for replication, both studies involved two-parent community families and typically developing children, drawn from the same population, with the two cohorts of children separated by approximately six years. Study 1 included 112 mother–child dyads. Study 2 included 102 mothers, fathers, and children. In both studies, families entered when children were infants, and were followed longitudinally. In Study 1, we obtained observational measures of BI and EC at age 2.5 years and observational outcome measures (disregard for maternal rules) at age 4. In Study 2, we assessed BI and EC at three assessments, at ages 2, 3, and 4.5, and examined their respective effects on parents’ ratings of children's externalizing problems (opposition and defiance, conduct problems, aggression) obtained at ages 5.5 and 6.5. In addition, given the large literature on the role of BI in the development of internalizing problems, and the known overlap between internalizing and externalizing problems, we controlled for the former. This approach allowed us to rule out the potential confounding effects of internalizing problems when examining the associations between inhibition and externalizing problems.

Study 1

Participants

The participants (two-parent community families) responded to ads and flyers distributed in an area that included a college town, a small city, small towns, and rural areas in US Midwest. Only mothers and their typically developing children (almost all born in 1995–1996) participated (N = 112, 56 girls). They entered the study when children were 8–10 months old. The families were mostly White (97% of mothers, 92% of fathers), but represented a range of SES. Approximately half of the parents had college or postgraduate education (59% of mothers, 57% of fathers); 15% of mothers and 10% of fathers had some college; 26% of mothers and 31% of fathers had high school education only. Most mothers worked outside of home (19% were homemakers): 38% were professionals; 31%, technicians or in sales; 5%, in service jobs; 2% laborers. Among fathers, 40% were professionals; 27%, technicians or in sales; 9%, in service jobs; 1%, farmers; and 13%, craftsmen, repairmen, laborers. Family income varied: under $20,000 (7%), $20,001–$30,000 (13%), $30,001–$40,000 (20%), $40,001–$50,001 (17%), $50,001–$60,000 (l6%), and more than $60,001 (25%). The study was approved by the University of Iowa IRB, and mothers signed informed consents.

Overview

All data were based on observations during 2–3-hour laboratory sessions (two at 33 months, 2.5 years, and two at 45 months, 4 years), conducted by female experimenters (Es) and videotaped for future coding. We present data on BI and EC, observed at 2.5 years (N = 104, 52 girls) and on children's disregard for maternal rule at 4 years (N = 101, 49 girls; in addition, analogous behavior reflecting scores, were obtained at 22 months, N = 106, 53 girls, and used as a covariate). Multiple teams of coders coded behavioral data. Approximately 15%–20% of cases were used for reliability, followed by frequent realignments. Note that the best practices in reliability have evolved over the last 25 years. Consequently, most statistics included kappas, alphas, and in one case of coding latencies, percent agreement.

Measures

Children's BI, 2.5 years

Paradigm

In a 10–12-min “Risk Room” paradigm (Kagan, Reznick, & Gibbons, Reference Kagan, Reznick and Gibbons1989; Kochanska, Aksan, & Joy, Reference Kochanska, Aksan and Joy2007; Kochanska et al., Reference Kochanska, Coy and Murray2001), the child faced mildly threatening events in an unfamiliar environment. The laboratory room was decorated with many odd-looking and slightly frightening objects (e.g., Halloween masks) and objects that could potentially involve mildly challenging physical activities (e.g., balance beam, a strange big black box, trampoline, ladder). The child was first observed while exploring the room (3 min), and then in seven “Risky Acts” –interactions with a stranger who encouraged the child to perform mildly threatening acts (e.g., ride an odd tricycle, slide down a strangely decorated slide, insert a hand into an odd box, touch a strange robot, put on scary masks, approach a person in a costume).

Coding and data aggregation

We coded proximity to mother for every 30 s as touching or within arm's length, hovering close but further than arm's length, or more than 1/3 room away from mother (reliability, kappa, .98), latency to explore (100% within 1 s), presence or absence of exploration for every 30 s, including extra credits for exploring particularly threatening objects (kappas, 1.00), and a fearful response to each Risky Act (0 = act performed immediately after the demonstration and before the first prompt, to 5 = never performed), plus an extra point if the child showed distress (kappa, .96). The number of segments in direct proximity to mother, latency to explore, reversed exploration score, and the scores for each Risky Act were all standardized and averaged into one BI score (Cronbach's alpha = .85).

Children's EC, 2.5 years

Delay tasks

Children performed several highly scripted tasks (Snack Delay, Tongue, Wrapped Gift, and Gift in Bag). The tasks are widely used, and have been included in several publications (e.g., Kochanska, Aksan, Penney, & Doobay, Reference Kochanska, Aksan, Penney and Doobay2007; Kochanska, Barry, Jimenez, Hollatz, & Woodard, Reference Kochanska, Barry, Jimenez, Hollatz and Woodard2009; Kim & Kochanska, Reference Kim and Kochanska2020; Kochanska & Knaack, Reference Kochanska and Knaack2003), and therefore are described briefly, with details available from the authors. All called for delaying highly desirable, hedonically attractive acts. Snack Delay (four trials) required waiting to reach for an M&M placed under a cup until E rang the bell. In Tongue (three trials), the child was asked to keep an M&M on his or her tongue without eating it. Wrapped Gift required first, waiting, facing away from E, who was noisily wrapping the gift, without peeking (Gift Wrap), and second, waiting in the seat, without touching, to unwrap the gift while E want to get a bow (Gift Bow). The Gift in Bag required waiting to retrieve a gift from a colorful bag until E brought a bow.

Coding and data aggregation

In Snack Delay, each trial was coded from 0 = eats the candy before E lifts the bell, to 4 = waits until bell is rung. The scores were averaged. In Tongue, in each trial, the latency to eat the M&M was coded, and the scores were averaged. Wrapped Gift produced two final scores, Gift Wrap and Gift Bow. Gift in Bag produced one final score. Each represented a composite of several (standardized) coded behaviors that cohered; for example, peeking, staying in seat, touching/opening the gift, as well as latencies to peek, to open, to leave seat, etc. Reliability of coding, average kappas, were .94 for Snack Delay, .83 for Wrapped Gift, and .94 for Gift in Bag. For Tongue, 90% of latencies were within a 2-sec difference.

The four scores (Snack Delay, Tongue, Gift Wrap, Gift Bow, Gift in Bag) were aggregated into the final EC score, following standardization of the first two. That score was internally coherent, Cronbach's alpha = .71.

Children's disregard for maternal rule, 4 years

Paradigm and coding

Mothers were asked to prohibit their children from touching very attractive objects on a low shelf in the laboratory, and to enforce the prohibition throughout both laboratory sessions. At the end of the second session, the mother reminded the child about the prohibition, asked him or her to engage in a dull sorting task, and left to the adjoining room, leaving the child alone for 8 min (details are in Kochanska et al., Reference Kochanska, Coy and Murray2001).

Child behavior was coded for every 5-s segment, employing several codes. Here, we use the measure of deviation or disregard for rules (taking objects off the shelf, unconstrained play with the objects). Reliability, kappa, was .92. The deviation codes were tallied and divided by the number of segments (excluding the relatively rare segments when the child was out of room, trying to get into the mother's lap, etc.). A fully analogous score was also obtained at 22 months and served as a covariate in the analyses. All descriptive data are presented in Table 1.

Table 1.

Descriptive statistics and correlations in study 1

Note. BI = behavioral inhibition; EC = effortful control.

^a Composite of standardized scores for behavioral inhibition in the Risk Room paradigm.

^b Composite of standardized scores for delay tasks.

^c,dPercent of observed segments in which child engaged in rule violation.

^e Gender was coded 0 = girl, 1 = boy.

^f Family income was coded on a 7-point scale.

Correlations involving gender are point-biserial. *p < .05. **p < .01. ***p < .001.

Results

Children who dropped out of the study at age 4 years and those who remained did not differ on any variables, with one exception: The former were from families with lower incomes compared to the latter, t (107) = −3.08, p = .003. Little's missing completely at random (MCAR) test indicated that data were missing completely at random, χ² (18) = 16.522, p = .56.

We first inspected the descriptive statistics and correlations (see Table 1). BI and EC at age 2.5 years were uncorrelated. Children's disregard for maternal rule was moderately stable from 22 months to 4 years and both measures correlated negatively with EC, but not with BI. Compared with girls, boys exhibited lower EC and more disregard for rules.

Next, we examined the anticipated interaction of EC and BI in predicting disregard for rules. Specifically, we modeled EC, BI, and their interaction at 2.5 years as the predictors, and disregard for maternal rule at 4 years as the outcome variable. BI was modeled as a moderator of the relation between EC and the outcome. Child gender and family income were included as covariates. In addition, we statistically controlled for disregard for rules at 22 months. We conducted the analysis using Mplus (Muthén & Muthén, Reference Muthén and Muthén1998–2020), which allows for unbiased missing data treatment with the full information maximum likelihood (FIML) method. BI and EC were computed from standardized variables and thus already mean-centered; in addition, we mean-centered disregard for rules at 22 months and family income.

Results of the regression model (Table 2) showed that early EC, but not BI, was associated negatively with disregard for rules at 4 years. However, the association between EC and disregard for rules was qualified by a significant interaction between BI and EC. The follow-up analysis, using simple slopes (Aiken & West, Reference Aiken and West1991), is depicted in Figure 1. For children with low (−1 SD) and average (0 SD) BI (relatively fearless or uninhibited), EC at age 2.5 years was associated negatively with disregard for rules at 4 years, B = −0.22, SE = 0.04, p < .001, and B = −0.15, SE = 0.04, p < .001, respectively. By contrast, for children with high BI (+1 SD, relatively fearful and inhibited), the relation between EC and disregard for rules was not significant, B = −0.07, SE = 0.05, p = .14. Of note, the effect of EC remained significant with all the predictors in the equation.

Figure 1.

Study 1. Simple slopes of EC predicting disregard for rules at low (−1 SD), mean (0 SD), and high (+1 SD) values of BI. Solid lines represent significant simple slopes, and dashed line represents nonsignificant simple slope. The shaded area represents the region of significance. BI = behavioral inhibition. EC = effortful control.

Table 2.

Study 1: Prediction of disregard for rules at age 4 years from BI, EC, and their interaction, controlling for disregard for rules at 22 months

Note. β = standardized regression coefficient using the STDXY (for continuous predictors) and STDY (for categorical predictors) options in Mplus; B = unstandardized regression coefficient; BI = behavioral inhibition; EC = effortful control.

⁺ p < .10. *p < .05. **p < .01. ***p < .001.

Further analysis using the Johnson–Neyman region-of-significance (RoS) technique (Johnson & Fay, Reference Johnson and Fay1950) suggested that compared with children with high BI scores, children with low BI scores showed more disregard for rules when their EC was low (<−0.46 SD). When the levels of EC were moderate to high, children engaged in few rule violations regardless of their BI scores.

Results

Attrition analysis revealed no significant differences on any variables between children who remained versus those who dropped out from the study. Little's MCAR test revealed that data were missing completely at random, χ² (25) = 25.583, p = .43. Descriptive statistics and correlations are listed in Table 3. BI and EC at ages 2, 3, and 4.5 were modestly to moderately correlated with each other. Both traits were moderately stable across the three assessments. EC, but not BI, was correlated negatively with externalizing problems at ages 5.5–6.5. Parents’ ratings of children's externalizing and internalizing problems were positively correlated.

Table 3.

Descriptive statistics and correlations in Study 2

Note. BI = behavioral inhibition; EC = effortful control.

^a Composite of standardized scores for behavioral inhibition in the Risk Room paradigms, two at each age.

^b Composite of standardized scores for delay tasks.

^c Composite of standardized scores for mother- and father-rated externalizing problems and aggression.

^d Composite of mother- and father-rated internalizing problems.

⁺ p < .10. *p < .05. **p < .01. ***p < .001.

Similar to the analytical approach adopted in Study 1, we used BI, EC, and their interaction to predict externalizing problems at 5.5–6.5 years, with BI modeled as the moderator of the association between EC and externalizing problems. Because BI and EC develop rapidly from 2 to 4.5 years, to capture the potential developmental timing effects, we estimated separate effects of BI and EC, and their interaction at each age: at 2, 3, and 4.5 years. Child internalizing problems were covaried. Because child gender and family income did not produce significant effects in the final equation, to reduce the number of predictors those variables were not included.Footnote ¹ All continuous predictors were mean-centered. Analysis was conducted in Mplus, and missing data were handled using FIML.

The findings in Study 2 partially replicated those in Study 1, but in addition revealed the role of developmental timing. Detailed results of the models are in Table 4. At age 2, neither BI nor EC, nor their interaction predicted externalizing behavior at ages 5.5–6.5.

Table 4.

Study 2: Prediction of externalizing problems at age 5.5–6.5 years from BI, EC, and their interactions at ages 2, 3, and 4.5 years

Note. β = standardized regression coefficient using the STDXY option in Mplus. B = unstandardized regression coefficient. BI = behavioral inhibition. EC = effortful control.

*p < .05. **p < .01. ***p < .001.

However, at age 3, EC, but not BI, was associated negatively with externalizing problems at ages 5.5–6.5. As in Study 1, the effect of EC was qualified by a significant interaction between BI and EC. Simple slopes of the interaction are depicted in Figure 2, Panel a. For children with low (−1 SD) and average (0 SD) BI (relatively fearless), EC at age 3 was associated negatively with externalizing problems at 5.5–6.5 years, B = −0.60, SE = 0.17, p < .001, and B = −0.38, SE = 0.14, p = .006, respectively. By contrast, for children with high (+1 SD) BI (relatively fearful), EC was unrelated to externalizing problems, B = −0.16, SE = 0.17, p = .34. Mirroring the findings in Study 1, the Johnson–Neyman RoS test suggested that children with average or low BI scores, and who were also low in EC (<−1.10 SD) displayed more externalizing problems than their fearful peers.

Figure 2.

Study 2. (Panel a): Age 3. Simple slopes of EC predicting externalizing problems at low (−1 SD), mean (0 SD), and high (+1 SD) values of BI. (Panel b): Age 4.5. Simple slopes of EC predicting externalizing problems at low (−1 SD), mean (0 SD), and high (+1 SD) values of BI. Solid lines represent significant simple slopes, and dashed lines represent nonsignificant simple slopes. The shaded area represents the region of significance. Results based on the interaction effects in the final equation, with all predictors entered. BI = behavioral inhibition; EC = effortful control.

At age 4.5, neither BI nor EC was related to externalizing problems. However, as in the model for age 3, the interaction between BI and EC was significant. Simple slopes of the interaction are depicted in Figure 2, Panel b. As at age 3, for children with low BI (−1 SD, relatively fearless), EC was associated negatively with externalizing problems, B = −0.49, SE = 0.18, p = .006; whereas for children with average (0 SD) and high (+1 SD) BI, EC was unrelated to externalizing problems, B = −0.04, SE = 0.13, p = .76, and B = 0.40, SE = 0.30, p = .17, respectively. The Johnson–Neyman RoS test suggested that children who were fearless at age 4.5, and who were also low in EC (<−0.93 SD) displayed more externalizing problems than their fearful peers.