Toddler Exuberance as an Influence on Positive Social Behavior in a High-Intensity Context in Middle Childhood

Abstract

Exuberance, a profile of temperament characterized in toddlerhood by high approach motivation, positive affect, and sociability, is associated with both adaptive and maladaptive socioemotional outcomes. The aims of the current study were to introduce a novel approach/avoidance-eliciting social task, as well as longitudinally extend our understanding of toddlerhood exuberance to outcomes in middle childhood. Specifically, affect and social behavior at age seven during a high-intensity game were compared to that observed during a low-intensity freeplay task. As part of a longitudinal study, 291 infants were selected at 4 months for a wide range of reactivity to novelty. The sample was assessed repeatedly across early childhood (9, 24, and 36 months of age) and at 7 years of age. A high exuberance profile was formed with approach, positive, and sociable behaviors observed in the laboratory from 4 to 36 months. At 7 years of age, affect and social behavior were assessed during high- and low-intensity interactions with an unfamiliar peer in the laboratory. Path Analyses using structural equation models demonstrated that a high exuberance profile was associated with greater positive social behavior at age 7 during a high-intensity game, but not negative social behavior or behavior during a low-intensity freeplay task. These results illuminate the need for targeted methodology, such as high-intensity approach/avoidance-eliciting social tasks, in order to clarify the links between early temperament and adaptive or maladaptive socioemotional outcomes across development.

Developmental growth consists of enduring adaptations to the environment, providing children with the skills to navigate the social world successfully. Both environmental factors, such as peer interactions, as well as genetic influences, such as temperament, influence social developmental outcomes (Degnan, 2017; Lamb & Bornstein, 2011). Through peer interactions, children learn to adopt socially acceptable behaviors and limit undesirable behaviors (Berdan et al., 2008). Additionally, one’s temperament can expand or constrain their exposure to social skills, as expressions of temperamental reactivity can modify social experiences in ways that support or hinder the development of socially competent behavior (Coplan & Bullock, 2012; Degnan & Fox, 2007; Rubin et al., 2009). Temperamental exuberance, defined by heightened positive approach toward novel and social contexts (Polak-Toste & Gunnar, 2006; Stifter et al., 2008), is thought to expand exposure to the social world and, in turn, contribute to greater social competence (e.g., Degnan et al., 2011; Dollar et al., 2017). However, exuberance also relates to maladaptive outcomes, such as poor emotion regulation and externalizing behavior problems (e.g., Degnan et al., 2011; Dennis et al., 2010; Dollar & Stifter, 2012). Therefore, trajectories of early temperamental exuberance across development require clarification. These conflicting results may be due to differences between measures (e.g., observational vs. report), contexts (e.g., novel vs. familiar), or individuals (e.g., longitudinal profiles), among other factors. The need for development and comparison of methods for eliciting temperamentally exuberant emotion and behavior across development preempts clarification of these differential effects. Thus, the current study was designed to 1) present a novel approach/avoidance-eliciting social task from which both affect and social behavior may be assessed, and 2) extend and compare the links from early temperamental exuberance to socioemotional behavior in middle childhood across multiple social contexts. Given previous literature reporting a combination of adaptive and maladaptive outcomes for temperamentally exuberant toddlers, it was expected that exuberance would be associated with greater positive and negative social observations at 7 years of age during a high-intensity, approach/avoidance-eliciting game, as compared to a low-intensity freeplay task.

Temperament: Reactivity to Novelty

Temperament includes individual differences in emotionality and behavior that are especially evident from birth through early childhood, consisting of stable levels of reactivity and regulation rooted in biology, with expressions of each influenced across development by environmental factors (Rothbart & Bates, 2006; Sanson et al., 2011). In particular, infant or toddler emotional and behavioral responses to various stimuli and situations can be an early indicator of innate temperament (Garcia-Coll et al., 1984; Goldsmith & Campos, 1990). In reaction to novelty, research from multiple samples supports three early temperament profiles (e.g., Dollar et al., 2017; Fox et al., 2015; Stifter et al., 2008). For instance, at 4 months of age, Fox et al., (2001) reported three infant reactivity groups, based on infants’ average affect and motor reactivity to novel audio and visual stimuli: High Negative reactive (above the means on motor reactivity and negative affect); High Positive reactive (above the means on motor reactivity and positive affect); and Low Reactive (below or at the means on motor reactivity, positive affect, and negative affect). Further, these profiles reflect heterotypic continuity into toddlerhood, with high negative reactivity leading to behavioral inhibition, high positive reactivity leading to exuberance, and low reactivity reflecting less extreme behavior and affect in novel social and non-social situations (Dollar et al., 2017; Fox et al., 2001; Kagan & Snidman, 1991). Additional work shows predictive value of these temperament profiles to later social behavior, as well as symptoms of internalizing or externalizing problems, across childhood and even into adulthood (e.g., Beekman et al., 2015; Degnan et al., 2014; Fox et al., 2001, Lahat et al., 2012; Tang et al., 2020).

In general, infants prone to avoidance and negative reactivity to novelty may express more wariness in novel and social settings, higher levels of effortful control, and lower externalizing behavior problems across development (Degnan & Fox, 2007; Schwartz et al., 1999). Those prone to approach and positive reactivity to novelty may express more sociability, impulsivity in approach-eliciting settings, reward-related risk-taking behavior, and externalizing behavior problems (e.g., Degnan et al., 2011; Morales et al., 2015; Polak-Toste & Gunnar, 2006). However, other studies find more complex behavioral trajectories for those at these temperamental extremes, such as exuberance leading to both internalizing and externalizing behavior problems (Stifter et al., 2008) or behavioral inhibition leading to adolescent substance abuse problems for males (Williams et al., 2010). More research on these complex pathways is necessary to elucidate the patterns and mechanisms inherent to temperamentally reactive children’s adaptive and maladaptive outcomes across development. Thus, the current paper provides a focused investigation on the links from early exuberance to social behavior in middle childhood.

Exuberance

The majority of research on temperamental reactivity has focused on behavioral inhibition (BI) or other negative affectivity traits (e.g., frustration reactivity), as they are seen as more of a concern to parents and teachers (Calkins & Degnan, 2006; Sanson et al., 2011). Conversely, temperamental exuberance consists of positive affectivity, high sociability, and high approach behaviors (Polak-Toste & Gunnar, 2006). Previously viewed as an expression of low inhibition, recent research identifies unique behavioral and biological components of exuberance (Dollar & Stifter, 2012; Dollar et al., 2017). Exuberant children are highly social and show high positive affect and approach in response to novelty in infancy (Stifter et al., 2008). When observed in a laboratory setting in toddlerhood, Fox et al. (2001) found these children were social toward unfamiliar peers and adults, approached novel stimuli quickly and fearlessly, and engaged in more risky behaviors in ambiguous or potentially scary situations. Similarly, Degnan et al. (2011) showed children with a high, stable profile of exuberance from infancy through early childhood displayed less social reticence and more social competence in social interactions with an unfamiliar peer at 5 years of age. As a temperamental construct, exuberance reflects a relatively stable profile across early childhood, but its typical trajectory across childhood and into adolescence remains unclear.

In addition to adaptive outcomes, exuberance profiles also relate to maladaptive developmental outcomes, such as externalizing behavior problems and hyperactivity (e.g., Degnan et al., 2011; Stifter et al., 2008). Parent-report of temperamental surgency, often equated with exuberance, frequently is associated with greater behavior problems (e.g., Berdan et al., 2008; Dollar & Stifter, 2012; Garstein et al., 2012). Surgency includes stable, positive approach to novelty, like exuberance, but also greater impulsivity and higher activity levels (Putnam, et al., 2008). Although these are not inherently maladaptive behaviors, children categorized as impulsive and hyperactive are prone to externalizing behavior problems, such as aggression, and peer rejection (Beekman et al., 2015; Degnan et al., 2011). Thus, exuberant children who reflect traits that are more surgent may be more likely to exhibit maladaptive outcomes, due to higher behavioral reactivity and potentially lower regulation. These more maladaptive behaviors may be particularly evident when a goal is blocked or inaccessible, in that they may exhibit more frustration, as well as aggressive behaviors to remove barriers to their goals (Berdan et al., 2008; Polak-Toste & Gunnar, 2006). It is unclear whether all exuberant toddlers are at risk of maladaptive behavior or whether there are additional factors that could explain these differential outcomes as reflected in the literature.

Both individual factors, such as regulation, and environmental factors, such as context (e.g., home vs lab), type of stimuli (e.g., novel vs. familiar), or socialization practices, may influence whether exuberance is associated with adaptive vs maladaptive behavior across studies. Given the bimodal nature of these outcomes and the extent to which the relations between temperamental exuberance and specific profiles of functioning are still relatively unspecified, nuanced research into the effects of this temperamental profile on social behavior throughout development is necessary. Examining under which conditions exuberant temperament relates to specific behavior patterns would elucidate the development and influence of temperamental dimensions on social and emotional well-being (Coplan & Bullock, 2012; Polak-Toste & Gunnar, 2006). Moreover, a greater understanding of how to support positive, socially competent development in exuberant children may lead to broader adaptive outcomes in late childhood, in turn preventing maladaptive outcomes as they transition into adolescence, early adulthood, and beyond.

Current Study

Understanding when and how temperamental exuberance influences emotional and behavioral outcomes in childhood will enhance an understanding of social developmental processes and contribute to the support of adaptive outcomes across development. The current study adds to the existing literature regarding temperamental effects on development in general and enriches the literature on temperamental exuberance in particular. Given links established between temperamental exuberance and differential outcomes across early childhood, the current study extended these paths into middle childhood.

The first goal of this study was to find a behavioral task that would capture children’s temperamental reactivity in a social and novel context in middle childhood. As children develop, the ability to regulate their reactivity to the environment increases, often masking their biological roots (Degnan & Fox, 2007). An intense approach/avoidance-eliciting game was hypothesized to draw out a child’s temperamental predisposition toward novelty, allowing their true internal motivation to play the game, interact with a peer, and enjoyment of the task to emerge. Although there are many individual tasks designed to elicit positive emotions (e.g., Goldsmith & Campos, 1990), reactivity to novelty (e.g., Fox et al., 2001), and social interactions (e.g., Walker et al., 2015), there is a paucity of tasks able to provide all three dimensions to an observation. Even fewer are available for use with older children. The Crocodile Dentist game (Winning Moves company, 1993) is an intense and exciting game, capable of eliciting high levels of positive affect, approach, and sociability behaviors typical of a temperamentally exuberant child. Examining children’s peer interactions during this task in the laboratory at 7 years of age, variations in frequency and intensity of adaptive and maladaptive social and non-social behaviors and affect are described.

The second goal of this study was to explore how early exuberance across infancy and toddlerhood relates to observed positive and negative behavior and affect during this high-intensity social context, compared to a more neutral peer interaction task. Temperamental exuberance was expected to predict both positive and negative social behaviors at 7 years of age, but especially during the exciting game.

METHOD

Participants

As part of a longitudinal study, 779 infants were brought into the laboratory at 4 months of age for a temperament screening, during which affect (positive and negative) and motor reactivity during the presentation of novel visual and auditory stimuli were observed (for more details, see Hane et al., 2008). Two hundred ninety-one infants (54% female) were selected to continue in the study based on their temperamental reactivity to novelty: high negative/high motor reactive (n = 116), high positive/high motor reactive (n = 106), and control (i.e., average or low reactive, n = 69). Based on parent report (99% response rate) at recruitment, mothers identified as 69.7% Caucasian, 16.6% African American, 7.2% Hispanic, 3.1% Asian, or 3.4% other/multiple ethnicities. Fathers were reported (98.6% response rate) to be 69.7% Caucasian, 18.8% African American, 5.6% Hispanic, 2.8% Asian, or 3.1% other/multiple ethnicities. No information regarding family income was collected; however, of the mothers who reported education level (99% response rate), most (78.2%) had at least a college education, whereas others (16.3%) reported at least a high school education, and some (5.5%) reported ‘other’ education. Similarly, most fathers were reported (98% response rate) to have at least a college education (71.8%), with others (19.9%) reported as having at least a high school education, and some (7%) reported as having ‘other’ education. ‘Other’ education typically included some college, vocational school, or other training programs beyond high school, although a few parents did not specify. In addition, the majority of families spoke English at home and approximately 2% spoke at least one additional language in the home (i.e., Spanish, Russian, or Chinese), but conducted assessments in English. The current study presents data from follow-up assessments with this sample across infancy (9 months), toddlerhood (24 and 36 months), and into middle childhood (age 7 years).

Of the sample selected in infancy, 188 (53% female) participated in a peer dyad assessment at 7 years of age with a same age, same sex, unfamiliar peer, recruited from the local community. Based on parent-report (99% response rate) at recruitment (at 2, 3, 4, 5, or 7 years of age), the majority of parents identified as Caucasian (65.8 – 69.2%), had at least a college education (71.7 – 85.6%), and spoke English at home (98%). A Chi-square analysis showed that the selected and unselected samples were not statistically different on parent ethnicity or education level (all ps > .10). Previous work also has reported that the unselected peer sample did not differ from the selected sample on mean levels of parent-reported shyness (Walker et al., 2015).

Procedure

Following the initial temperament screening at 4 months, participating families were invited to laboratory assessments at 9, 24, 36 months, and 7 years of age. Previous work describes the methodology used during laboratory assessments at 9, 24, and 36 months to derive latent profiles of temperamental exuberance (N = 291; Degnan et al., 2011). At 7 years of age, children returned to the lab paired with a same age, same sex, unfamiliar peer, and each dyad (n = 188) was presented with a variety of games/tasks while being observed and video-recorded by research assistants through a two-way mirror. The first task the peer dyad completed was a Free Play task (FP). In this task, a range of age appropriate toys were scattered across the floor and the children were left alone (i.e., no adults) to play as they wanted to for approximately 6 minutes. Later on in the visit, the dyad played the Crocodile Dentist (CD), a game appropriate for 4–7 year olds and 1–4 players (Winning Moves company, 1993). For this game, the children pretend to be dentists helping a crocodile who ‘has a sore tooth’ by taking turns pressing down the crocodile’s teeth, one at a time. If they press the ‘sore’ tooth, the crocodile will surprise them by snapping his mouth shut. Each time the game is reset (i.e., the mouth is opened) a new tooth is set to be ‘sore,’ at random. After the game was explained to the children, they were left alone to play for approximately 3 minutes. Reasons for attrition at the 7-year assessment included difficulty scheduling laboratory visits, family relocating, and permanent attrition. Families missing data for Free Play and/or Crocodile Dentist data (n = 125) at 7 years were not significantly different from the rest of the sample (n = 166) by 4-month temperament group, gender, maternal ethnicity, maternal education, or any of the key study variables (all ps > .05). Additional information regarding patterns of missing data is provided below in the Planned Statistical Analyses section.

Measures

Exuberance

All data included in the latent profile of temperamental exuberance were observed during laboratory assessments across infancy and toddlerhood (n = 291): Positive Reactivity group membership at 4 months, 0 = not in positive group, 1 = in positive group; Positive Approach score from tasks in the Laboratory Temperament Assessment Battery (Goldsmith & Rothbart, 1996) at 9 months; as well as Positive Affect, Approach, and Sociability scores from an approach/avoidance eliciting paradigm at both 24 and 36 months. A two-profile model (high exuberance and low exuberance) reflected the best fit to the data, as well as a great deal of stability across measures for the high exuberance profile (for more details, see Degnan et al., 2011). For the current study, a continuous score denoting each child’s probability of membership in the high exuberance profile was included as the measure of High Exuberance.

Social Behavior

Observational data from the 7-year visit was accessible for 176 unfamiliar peer dyads. Videos from this visit were coding behaviorally and analyzed. Separate coding teams coded each set of child behavior during the FP and CD tasks.

Free Play (FP).

Of the 176 dyads, one had a video recording error. After watching each of the 175 remaining dyadic interactions in its entirety, global behavioral ratings were recorded for Social Interest, Wariness, Activity level, Unfocused/Disengaged, Positive Affect, Negative Affect, and Aggression on a scale ranging from low (1) to high (7) levels of each behavior (Almas et al., 2011; Walker et al., 2015). Specifically, Social Interest was based on a child’s interest in the peer, with low scores indicating little to no acknowledgement of the peer and high scores indicating repeated initiation and/or engagement with the peer for most, or all, of the episode. Wariness was based on hesitance to play and fearful behavior toward the peer or social situation, with low scores indicating that the child displayed little to no hesitance or fearful behavior and high scores indicating the child displayed hesitance or fearful behavior for most, or all, of the episode. Activity Level was based on children’s movement around the room, with low scores indicating that the child stayed in one place, or only moved as needed, and high scores indicating that the child displayed intense movement for the majority of the episode. Unfocused/Disengaged behavior was based on children’s focus on play or an activity, with low scores indicating that children were very focused on one activity at a time, with little to no distraction, whereas high scores indicated that children displayed minimal to no focus, or engagement, in play. Positive affect was based on displayed positive emotion and facial expressions (e.g., smiling, laughing, excited tones, giggling), and negative affect was based on displayed negative emotion and facial expression (e.g., frowning, pouting, crying, negative tones). Low scores on positive or negative affect indicating minimal to no displays of affect, whereas high scores indicated continued expressions of intense positive or negative affect for the majority of the episode. Aggression was based on the use of threatening language or behaviors displayed with the intention to harm the peer physically or emotionally, with low scores indicating little to no displays of aggression and high scores indicating repeated displays of physical and/or verbal aggression.

Interrater reliability (ICC) amongst 3 coders blind to participant temperament ranged from .71 to .88 across 29% of the videos. The seven global ratings formed three broad variables based on previous work with similar data (e.g., Almas et al., 2011; Walker et al., 2015). FP Social Engagement consisted of the mean of global social interest (M = 4.98, SD = 1.21), activity level (M = 2.99, SD = 1.34), and positive affect (M = 3.74, SD = 1.07) codes, average r = .25. FP Social Wariness consisted of the mean of global wariness (M = 1.13, SD = 0.56) and unfocused/disengaged (M = 2.73, SD = 1.03) codes, average r = .31. FP Dysregulation consisted of the mean of global negative affect (M = 2.58, SD = 0.95) and aggression (M = 1.88, SD = 1.14) codes, average r = .37. FP Social Wariness was expected to show redundant, opposite effects compared to FP Social Engagement. Therefore, only FP Social Engagement and FP Dysregulation are included in the present study due to their theoretical overlap with social behaviors expected from children with exuberant temperament.

The Crocodile Dentist (CD).

Of the 176 dyads, 9 videos did not provide CD data due to differences in protocol (6), child distress (1), or technical errors with the videos (2). Thus, each of 167 dyads were coded separately by two teams: one for affect, and one for social behavior and engagement with the toy.

CD Affect coding.

The duration (0 = no affect seen, 0 seconds; 1 = less than ¼ of an epoch, 1– 6 seconds … 5 = more than approx. ¾ of epoch, 25–30 seconds) and peak intensity (0, no affect, to 3, displaying 3 aspects of affect at a time: facial, body, and verbal) of positive and angry/frustrated affect were rated in 30-second epochs across the task. Examples of positive affect included the presence of excitement and expressions of joy towards the other child or towards the task. Examples of angry/frustrated affect included the presence of angry and/or frustrated expressions, such as scowling, groaning, and shoving, towards the other child or towards the task. Duration and intensity scores for each type of affect were averaged separately across epochs. Interrater reliability (ICC) between two coders blind to participant temperament was .85 for CD Positive Affect intensity, .88 for CD Positive Affect duration, .75 for CD Angry/Frustrated intensity, and .73 for CD Angry/Frustration duration, across 20% of the videos. The final CD Positive Affect and CD Negative Affect variables consisted of the product of the duration and intensity codes of positive and angry/frustrated affect, respectively, in order to compute affect scores that would range from little expression (low duration × low intensity) to intense, frequent affect (high duration × high intensity).

CD Social Behavior coding.

Codes for positive and negative social behaviors reflected how the children socially interacted with each other during the game. Codes for engagement with the toy reflected how each child independently interacted with the toy, or the non-social engagement with the task. These intentional distinctions separately assessed the child’s engagement to social vs. non-social aspects of the task, common indicators of temperamental reactivity to novelty (Fox et al., 2001; Polak-Toste, & Gunnar, 2006). Both the frequency (number of instances) and intensity (0–2; involving one or more types of behavior: body language, verbal) of positive and negative social behaviors were coded in 10-second epochs across the task. Positive social behaviors included smiling, making eye contact, encouraging or giving advice to the peer, having an excited or amused tone of voice, and conversation not regarding the task. Negative social behaviors included avoiding eye contact, pushing or hitting the peer, blocking the peer from taking a turn, mocking or making mean comments to the peer, and using a frustrated or annoyed tone of voice. Engagement with the Toy was coded as present (1) or absent (0) for possession (whether or not the child took a turn, i.e., had possession of the toy), as well as for physical engagement (0 = none, 1 = some) and attention engagement (0 = none, 1 = some) within 10 second epochs across the task. Examples of physical engagement with the toy included holding the mouth open to prevent it from snapping closed, forcing the mouth closed, and hitting/pressing down the teeth with excessive force. Examples of attention engagement included looking at the toy during the other child’s turn and attentional engagement with the toy on their own turn. Scores for frequency and intensity of positive and negative social variables, as well as for possession, physical toy engagement, and attention toy engagement variables, were converted into standardized sums (i.e., summed across number of epochs, divided by number of actual epochs, and multiplied by maximum 16 epochs) in order to equate scores between dyads observed over slightly varying lengths of time. Interrater reliability (Intra-class correlations, ICC) for the raw measures between two coders blind to participant temperament, using 13% of the videos, ranged from .73 to .97 across the measures. The final CD Positive Social Behavior and CD Negative Social Behavior variables consisted of the product of the frequency and intensity codes of positive and negative social behavior, respectively. The CD Total Task Engagement variable consisted of the difference between the product of possession by attention engagement and the product of possession by physical engagement, indicating engagement with the toy that was appropriately on-task and not aggressive toward the toy.

CD Summary Scores.

Five participants were removed from the final analyses due to their scores being greater than 4 SDs above the means of the CD Positive Social Behavior, CD Negative Affect and/or CD Negative Behavior summary scores. Thus, there were 162 participants with CD Summary Scores included in the final analyses. Despite removing these outliers, CD Negative Affect and CD Negative Social Behavior maintained a substantial positive skew. Therefore, these scores were square root transformed (sqrt).

Correlations between the CD summary scores revealed that CD Positive Affect and CD Positive Social Behavior were modestly associated, r = .16, p = .042. Thus, they were standardized and averaged together to form the final CD Positive Behavior composite. CD Negative Affect (sqrt) and CD Negative Social Behavior (sqrt) also were associated moderately, r = .23, p = .003 and were standardized and averaged together to form the final CD Negative Behavior composite. Although CD Total Task Engagement was associated negatively with CD Negative Affect (sqrt), r = −.27, p = .001, and CD Negative Social Behavior (sqrt), r = −.26, p = .001, it was standardized, but remained as an independent measure of positive non-social behavior during the CD task.

Planned Statistical Analyses

First, the CD codes were explored in order to determine whether the task elicited a range of approach/avoidance, illustrated by children’s positive/negative social behaviors and affect. Then, preliminary analyses examined associations between potential covariates (i.e., gender, ethnicity, and education) and the key variables of interest (High Exuberance, FP scores, and CD scores). Covariates associated with Exuberance or at least one of the FP or CD scores were included in the larger models. In addition, missing data patterns were analyzed. The assumption that the patterns of missing data were non-random was not confirmed by a missing data analysis conducted in IBM SPSS Version 26, Little’s MCAR (Missing Completely at Random) χ² (15) = 9.01, p = .877, thus the assumption that data were missing at random (MAR) is reasonable for this data set. In addition, missingness at 7 years was not related to 4-month selection group, gender, maternal ethnicity, maternal education, or any of the key study variables (all ps > .05). Therefore, as is recommended practice (e.g., Little & Rubin, 2019), Full Information Maximum Likelihood (FIML) estimation was used to account for missing data, providing a potential sample size of 291. As noted below, however, 5 participants’ scores were markedly different from the rest of the sample and were removed from all final models. This resulted in a final sample size of 286. Finally, Path Analyses with Structural Equation Modeling (SEM) in IBM SPSS AMOS Version 26 (IBM, 2019) tested whether there were longitudinal links between early temperamental High Exuberance and positive and negative behavior during the CD and FP tasks at 7 years of age. A follow-up analysis compared model fit between a trimmed model and the original model. Additional alternative models including effortful control as a predictor, a moderator, and gender as a moderator were also tested, but proved to result in either significantly worse fit or non-significantly different fit (with less parsimony) compared to the models presented below (see S1).

RESULTS

Descriptive Statistics

Descriptive statistics for the individual 7-year measures are in Table 1. The CD positive affect and behavior, the CD negative behavior, the CD toy engagement and physical behavior, and the FP measures demonstrated relatively normal distributions, whereas the CD negative affect and toy attention measures demonstrated a prominent level of skew (> 2). On average, participants displayed a little over 21 positive behaviors in 16 epochs of the CD game. An even deeper analysis of the positive social behaviors showed that positive verbal behaviors were almost twice as common as positive body language (M =17.63, SD = 7.23 and M = 9.13, SD = 4.52, respectively). In comparison, CD negative affect and behaviors were much less common. On average, participants displayed less than 3 negative behaviors in 16 epochs of the CD game. Specifically, negative verbal behavior was just as common as negative body language (M = 1.39, SD = 1.74 and M = 1.40, SD = 2.49, respectively). With regard to engagement with the CD toy, on average, participants were more attentive and positively engaged with the toy than they were physical, or negatively engaged, with the toy. In an average CD game, participants had possession of the toy for over 13 epochs and were physically harsh with the toy for under 3 epochs, but attentive to the toy during almost all 16 epochs. During FP, participants were more likely to display higher levels (> 3 on a 7 pt scale) of positive affect and social interest, moderate levels (~3 on a 7 pt scale) of activity level, and lower levels (< 3 on a 7 pt scale) of negative affect and aggression (see Table 1).

Table 1.

Descriptive Statistics for Key Crocodile Dentist (CD) and Freeplay (FP) Measures

Variable	Mean	SD	Skewness
CD Positive Frequency	21.27	6.86	0.05
CD Positive Intensity	26.71	9.86	0.18
CD Negative Frequency	2.26	2.23	1.18
CD Negative Intensity	2.59	2.72	1.41
CD Toy Frequency	13.55	3.22	−1.57
CD Toy Physical	2.76	3.37	1.45
CD Toy Attention	15.57	1.15	−3.81
CD Positive Affect Duration	2.23	0.83	0.24
CD Positive Affect Intensity	1.98	0.53	0.19
CD Angry/Frustrated Duration	0.07	0.13	2.60
CD Angry/Frustrated Intensity	0.08	0.17	2.37
FP Social Interest	4.98	1.21	−0.17
FP Positive Affect	3.74	1.07	−0.12
FP Activity Level	2.99	1.34	0.34
FP Negative Affect	2.58	0.95	0.21
FP Aggression	1.88	1.14	1.03

Note. Outliers are removed; CD variables are standardized sums (n = 162); FP variables are global scores (n = 170)

Preliminary Analyses

Descriptives for and preliminary correlations between the key summary variables are in Table 2. The probability of High Exuberance was correlated significantly and positively with CD Positive Behavior, r = .19, p = .014, but none of the other CD or FP measures. Among the CD variables, CD Total Task Engagement was correlated significantly and negatively with CD Negative Behavior, r = − .34, p < .001, but not with CD Positive Behavior. CD negative and positive scores were not significantly correlated. Between the CD variables and FP variables, CD Positive Behavior was significantly, positively correlated with FP Social Engagement, r = .18, p = .027, and FP Dysregulation, r = .22, p =.010. CD Negative Behavior and CD Total Task Engagement were not correlated significantly with CD Positive Behavior or any of the FP measures. Among the FP measures, FP Social Engagement and FP Dysregulation were correlated significantly and positively, r = .44, p < .001.

Table 2.

Descriptive Statistics and Bivariate Correlations among Key Summary Scores

Variables	Mean	SD	1	2	3	4	5
1. High Exuberance (N=286)	0.35	0.34	---
2. CD Positive Behavior (N=162)	-0.00	0.76	0.19*	---
3. CD Negative Behavior (N=162)	-0.00	0.79	0.10	−0.02	---
4. CD Total Task Engagement (N=162)	0.00	1.00	−0.06	0.10	−0.34***	---
5. FP Social Engagement (N=170)	3.90	0.85	−0.02	0.18*	0.09	−0.07	---
6. FP Dysregulation (N=170)	2.25	0.87	0.14	0.22**	0.11	−0.12	0.44***

Note. With outliers removed and transformations computed;

^*p ≤ .05,

^**p ≤ .01,

^***p ≤ .001

Additional preliminary analyses using independent samples t-tests examined mean differences in terms of gender, maternal ethnicity, and maternal education on all key study variables (Table 3). Gender differences were significant for FP Social Engagement, t (168) = 4.62, p < .001, and FP Dysregulation, t (168) = 3.52, p = .001, with males showing greater levels of both types of behaviors. None of the key study variables showed significant differences in terms of maternal ethnicity. Maternal education differences were significant for CD Negative Behavior, t (33.77) = 2.23, p = .032, such that mothers with less than a college education had children with higher CD Negative Behavior. Given these preliminary results involving gender and maternal education, they were included as covariates in all SEM analyses.

Table 3.

Means Differences (SDs) in Key Summary Variables by Demographic Categories at Recruitment

Variables	Gender (Male vs Female)	Maternal Ethnicity (Caucasian vs. Minority)	Maternal Education (< College Degree vs. ≥ College Degree)
High Exuberance (N=286)	−0.03 (0.04)	0.03 (0.04)	−0.07 (0.04)
CD Positive Behavior (N=162)	0.07 (0.12)	0.18 (0.14)	0.23 (0.15)
CD Negative Behavior (N=162)	0.22 (0.13)	0.05 (0.14)	0.42 (0.19)*
CD Total Task Engagement (N=162)	0.01 (0.16)	0.17 (0.18)	−0.09 (0.20)
FP Social Engagement (N=170)	0.57 (0.12)***	0.02 (0.15)	−0.06 (0.14)
FP Dysregulation (N=170)	0.46 (0.13)***	.16 (0.15)	−0.13 (0.15)

Note. With outliers removed and transformations computed;

^*p ≤ .05,

^**p ≤ .01,

^***p ≤ .001

Longitudinal Effects of Early Exuberance

The first SEM analysis examined the relations between the probability of High Exuberance and children’s behavior during the CD game and FP task, controlling for gender (Male vs. Female) and maternal education (< College Degree vs. ≥ College Degree). This model (Figure 1) specified direct paths of High Exuberance, gender, and maternal education on CD Positive Behavior, CD Negative Behavior, and CD Total Task Engagement, as well as on FP Social Engagement and FP Dysregulation. Based on preliminary analyses, the model included a covariance between CD Negative Behavior and CD Total Task Engagement, as well as covariances for CD Positive Behavior with FP Social Engagement and FP Dysregulation.

Figure 1.

Path Analysis Model 1

The resulting model demonstrated a moderately poor fit to the data, χ² (10, N = 286) = 37.38, p < .001, CFI = .69, RMSEA = .10. However, results indicated that High Exuberance significantly predicted CD Positive Behavior, β = .17, p = .024, but did not significantly predict CD Negative Behavior, β = .11, p = .149, CD Total Task Engagement, β = −.06, p = .450, FP Dysregulation, β = .13, p = .071, or FP Social Engagement, β = −.02, p = .805. Gender significantly predicted CD Negative Behavior, β = −.15, p = .049, FP Social Engagement, β = .34, p < .001, and FP Dysregulation, β = −.28, p < .001, but did not significantly predict CD Positive Behavior, β = −.04, p = .609, or CD Total Task Engagement, β = −.01, p = .948. Maternal education level significantly predicted CD Negative Behavior, β = −.24, p = .002, but did not have any significant effects on CD Positive Behavior, β = .11, p = .161, CD Total Task Engagement, β = .04, p = .578, FP Dysregulation, β = .04, p = .598, or FP Social Engagement, β = .02, p = .818.

Given the lack of relations between High Exuberance and the FP variables in the first model, a second SEM model removed the FP summary scores in order to test the effect of these paths on the overall model fit. Thus, this model specified direct paths of High Exuberance, gender, and maternal education on CD Positive Behavior, CD Negative Behavior, and CD Total Task Engagement, as well as a covariance between CD Negative Behavior and CD Total Task Engagement (Figure 2). The resulting model demonstrated a great fit to the data, χ² (5, N = 286) = 4.78, p = .443, CFI = 1.00, RMSEA = .00, and a significantly better fit than the previous model, Δχ² (5, N = 286) = 32.60, p < .001. Therefore, ignoring predictive paths to the FP variables resulted in significantly better model fit. Results maintained significant relations between the probability of membership in the High Exuberance profile and CD Positive Behavior, β = .17, p = .027, between gender and CD Negative Behavior, β = −.15, p = .049, and between maternal education and CD Negative Behavior, β = −.24, p = .002. All other relations in the trimmed model remained non-significant, as in the original model.

Figure 2.

Path Analysis Model 2

DISCUSSION

The overarching aim of this study was to examine how early temperamental exuberance relates to adaptive and maladaptive social behavior in middle childhood, using a novel high-intensity social task. The results of this study support previous reports by finding longitudinal links between early exuberance and positive social behavior (Degnan et al., 2011, Morales et al., 2016), but not negative behavior or non-social behavior. In addition, the current study extends an understanding of the temperamental trajectory of exuberance into middle childhood by presenting a behavioral coding scheme for observations of peer dyads interacting during an intense approach/avoidance-eliciting game.

The first goal of this study was to identify a task that could capture the influence of temperamental exuberance on social behavior with peers beyond early childhood. Although many studies assess behaviorally inhibited temperament and social behavior using laboratory tasks throughout adolescence, there are fewer tasks designed to assess positive reactions to novel and social contexts, especially for older samples. When children enter formal education they often learn to adapt or regulate temperamental predispositions toward socially appropriate behaviors (Berdan et al., 2008; Dollar & Stifter, 2012). By introducing an intense approach/avoidance game like Crocodile Dentist (Winning Moves company, 1999), a wide range of positive and negative reactions to this novel, social context were observed. Overall, participants enjoyed the game and were engaged with the toy, as evidenced by their high levels of non-social attention during the task. Further, the game evoked positive social behavior more often than negative social behavior, suggesting it is a fun game that the peers enjoy playing. The addition of this task to future work regarding positive reactivity should enhance the field’s ability to track the development of these behaviors into middle childhood.

The second goal of this study was to address how a high exuberance profile across infancy and toddlerhood might be predictive of observed social behavior within multiple contexts in middle childhood. The relation between temperament and later outcomes is found to be incredibly complex and more research is needed to disentangle how different aspects of temperament may be associated with behavioral outcomes at different points in development (Dennis et al., 2010). Our data showed a significant effect of exuberance on positive social behavior, but not negative or non-social behavior during the Crocodile Dentist task in middle childhood. Children who were more likely to show a highly exuberant temperament across infancy and toddlerhood exhibited more positive social behavior and affect when playing an exciting approach/avoidance game with an unfamiliar peer. However, exuberance was unrelated to engagement or disruptive behavior during a more standard free play task. In general, these findings support previous work demonstrating relations between toddlerhood exuberance and positive social engagement (e.g., Degnan et al., 2011; Dollar et al., 2017), and suggest these relations may be particularly evident when exploring behavior within high-intensity approach contexts (Putnam & Stifter, 2005).

Although previous research has found relations between exuberance and negative behavioral outcomes, the present results do not support this connection. There are a few possible reasons for this discrepancy. One, the current analysis was predicting behavior in middle childhood, and previous work has primarily focused on toddlerhood and early childhood (Degnan et al., 2011; Dollar & Buss, 2014, Gartstein et al., 2012). Thus, perhaps exuberance relates more strongly to negative behavior in early childhood, prior to the development of increased self-regulation in middle childhood. Second, previous work has found relations between exuberance and maternal report of behavior problems (Degnan et al., 2011; Morales et al., 2016), as well as reported surgency and observed negative behavior with peers (Dollar & Stifter, 2012), whereas the current study included observed measures of exuberance and negative social behavior. This finding suggests a potential discrepancy between maternal report and observations of negative behavior and supports the use of multiple reporters and assessments to bolster the reliability of the results (Dennis et al., 2010; Kagan & Fox, 2006). Other than low reliability of measures, this discrepancy could indicate children’s ability to regulate their behavior in certain contexts (e.g., a lab setting with an unfamiliar peer), as opposed to a more comfortable context (e.g., at home with family), exhibiting less negative behavior in the more public setting. It could also suggest a maternal perceptual lens contributes to the view of their child’s behavior compared to an objective researcher. Further research should examine consistency in outcomes across age, context, and reporter, especially for children who are able to alter their behavior (i.e., regulate) toward what is socially acceptable in various contexts.

The current study adds to the literature on when and how early temperamental profiles relate to social-emotional functioning, including a broader understanding of both adaptive and maladaptive outcomes. Presenting data from a novel approach/avoidance-eliciting social task, the present results expand the available methods for assessing early temperament’s role in trajectories of behavior through middle childhood. Further, a comparison of behavior during social contexts differing in intensity, helped clarify links between early exuberance and social behavior. Additional examples with this type of in-depth study could assist developmentalists, clinicians, parents, and teachers in promoting adaptive outcomes for all children (Calkins, & Degnan, 2006). By understanding the contextual factors that lead different temperament profiles to influence later development, professionals may be able to adapt the environment so that children may grow and develop in ways that are most adaptive given their reactivity to the world around them (Sanson et al., 2011). For exuberant children, in particular, this might involve more instruction and support in self-regulation skills, social information processing, as well as appropriate opportunities for approach-oriented exploration of novel social contexts. In addition, adults who work with exuberant children may benefit from an understanding of the positive approach motivation that supports their social behavior in specific contexts.

Future research should continue to examine the longitudinal patterns in positive social behavior predicted by early temperamental exuberance and the role that other factors, such as regulation and socialization might have on different outcome trajectories. Specifically, differences in how children behave in response to both social and non-social stimuli, as demonstrated during the Crocodile Dentist (Winning Moves company, 1999) game may be indicative of their temperament and regulatory skill, combined. Indeed, alternative models examined with these data suggest that adding effortful control did not help explain the relations between exuberance and social behavior during either Crocodile Dentist or Freeplay tasks (see S1). The development of self-regulation skills and executive functioning, the socialization of these skills, as well as the perceptions of behavior by others, are likely to be important pieces of the process and should be . In addition, understanding differences in exuberant responses to social and non-social situations across development may clarify how certain social-emotional outcomes develop for temperamentally exuberant children (Fox et al., 2001; Polak-Toste, & Gunnar, 2006) and whether those are different for temperamentally surgent children (Putnam et al., 2008).

Limitations of the current study include the skewed nature of the negative social behavioral data, the single tasks used to assess low- and high-intensity social behavior, and the sample demographics. Many of the negative variables showed limited variability and, as a result, may have indicated less problem behavior than found in other studies. This limited variability was likely due to the nature of the tasks as inherently positive. Future studies should examine exuberant children’s social behavior across multiple intense emotion evoking scenarios designed to elicit both positive and negative behavior. In addition, the outcomes examined constituted broad constructs comprised of any number of more specific adaptive and maladaptive behaviors. Additional work should provide a detailed account of social behaviors that are more or less likely during varying contexts and with particular types of temperamentally reactive children. Further, the sample in the present study was selected to represent a wide range of temperamental reactivity to novelty. However, they were not selected for a wide range of demographic characteristics. Therefore, the current results may not generalize to other more diverse samples. Testing the use of the Crocodile Dentist task in other populations will bolster the known effectiveness for the literature. Finally, examining whether these longitudinal effects are sensitive to demographic and/or sample differences is essential to a robust understanding of links between early exuberance and outcomes across development.

Overall, the current study demonstrated a novel method for observing behavioral and emotional variability in a social laboratory setting in middle childhood. Methods such as this one allow us to better explore how temperamental profiles from toddlerhood influence outcomes across multiple contexts later in development. Including behavioral observations such as these robust measures can expand future research questions and investigations, but they should be examined across a variety of samples. This study supports the hypothesis that exuberance in toddlerhood is predictive of positive social behavior in middle childhood, especially in an intense, positive approach/avoidance-eliciting context. Additional research will be necessary to understand the links between exuberance and the negative outcomes found in other studies.