Optimistic Children Engage in More Constructive Risk-Taking Behaviors

Children continuously encounter opportunities to explore and learn in their physical environment. Some opportunities present favorable challenges that help children develop motor and socioemotional skills and confidence; others present greater risks with potentially harmful consequences including injuries. In the United States, unintentional injuries are the leading cause of child death, and playground injuries account for over 200,000 pediatric emergency department visits annually (Centers for Disease Control and Prevention, 2015). Risk-taking that results in injuries often occurs when children’s motor skills do not align with the activities they engage in, with decreasing levels of injuries with age reported from toddlerhood through middle childhood (Bradbury et al., 1999; Schwebel & Plumert, 1999). Injuries also occur with greater exposure to risky activities, with a general trend of boys being more prone to injuries than girls (Mytton et al., 2009). However, given the established benefits of risk-taking on children’s health and development (see Brussoni et al., 2015 for a review), it is important to understand how children assess and decide to engage in potentially risky activities in order to create environments that support children’s learning and development while minimizing injuries.

Risk is defined as the uncertainty of achieving a desired outcome (Kopfstein, 1973), and how children respond to and learn from challenges is likely shaped by their expectations about these uncertain future outcomes. Approximately 80% of adults are optimistic – that is, they tend to overestimate the likelihood of positive outcomes and underestimate negative ones (Sharot, 2011). Optimism is linked to positive attributes in adults including greater persistence during challenging tasks (Tenney et al., 2015), more positive beliefs about one’s abilities (Yu & Luo, 2018), and more effective use of coping strategies (Nes & Segerstrom, 2006). As cognitive biases such as optimism appear to be present early in development (e.g., Bamford & Lagattuta, 2020; Hennefield & Markson, 2022; Wente et al., 2020), optimism likely confers similar benefits in children. However, in situations with potentially serious consequences, such as high-risk physical activities, being overly optimistic may jeopardize children’s well-being. Therefore, understanding contexts in which optimism may be either advantageous or disadvantageous has both theoretical and real-world implications. To investigate the role of optimism in risk-taking, the present study assessed how children’s general expectations about event outcomes relate to their (1) perceptions of risk, and (2) willingness to engage in risky physical activities.

Optimism

Broadly positive assessments of abilities and outcomes may promote trial and error learning and enhance perseverance in young children (e.g., Bjorklund, 1997). Numerous studies demonstrate that children as young as 3 years of age hold positive perceptions of themselves and others, and that these perceptions peak and persist through middle childhood (see Boseovski, 2010, for a review). For example, among samples of 3- to 10-year-old children, there is evidence that younger children are more positive about their academic competence (Benenson & Dweck, 1986), physical abilities (Mantzicopoulos, 2006), event outcomes (Bamford & Lagattuta, 2020), and others’ future success (Boseovski et al., 2009) than older children. These positive perceptions may be motivational in that they encourage children to seek out learning opportunities and persist in challenging situations. However, positivity might not always be beneficial – such as in situations involving danger and high likelihood of physical harm. A recent study found that 10- and 11-year-old children expected harmful accidents were more likely to happen to other children than themselves and justified these expectations by overestimating their abilities or degree of control over risky events (Joshi et al., 2018). Thus optimism, a specific type of positivity based on overestimating event outcomes, could be disadvantageous in certain contexts, underscoring the need for studies that consider both positive and negative aspects of optimism.

Risk Constructs

Risk is a multifaceted construct influenced by various factors including environmental factors and individual child characteristics (Morrongiello & Lasenby-Lessard, 2007). Risk has traditionally been assessed in two distinct ways – children’s perceptions of how dangerous a given activity or situation is, and their risk-taking behaviors measured via intentions or observations of risk-taking actions. Studies find that child-reported intentions to take risks closely match their observed risk-taking (Morrongiello, 2004; Morrongiello et al., 2009), suggesting both are valid measures of risk-taking. The present study measures risk perception and risk-taking for the same activities to systematically investigate (1) children’s perception of danger and willingness to take risks as a function of degree of risk, and (2) links between optimism, risk perception, and risk-taking.

Risk perception is argued to be a cognitive process that drives risk-taking (Morrongiello & Lessard-Lasenby, 2007). However, few studies have directly compared both constructs. In one exception, risk perception was measured via 4- to 7-year-old children’s ratings of photos of dangerous household situations and contrasted with observed risk-taking behaviors in a simulated household environment (Boles et al., 2005). Overall, children who rated the situations as less dangerous engaged in more risky behaviors. However, although older children rated the situations as more dangerous than younger children, older children engaged in more risky behaviors. Despite all being household activities, the constructs were not assessed using the same situations making direct comparisons difficult. Thus, although there is some evidence that these constructs are related, given the lack of direct comparison, how children’s perceptions of danger translate to their behaviors remains an open question.

A key factor that may influence children’s risk perception or risk-taking is the degree of danger present. One study revealed that 4- and 5- year-old children acknowledged differences in injury severity between low- and moderate-risk but not moderate- and high-risk activities (Little & Wyver, 2010). Another found that 6- to 11-year-old children rated riskier physical activities as more dangerous, but with increasing risk, girls rated activities as more dangerous than boys (Morrongiello & Rennie, 1998). These findings broadly highlight the importance of interpreting risk findings as a function of degree of risk. Critically, participating in moderately risky activities is likely advantageous as those activities provide opportunities to develop new skills, experiment with boundaries, and foster persistence (Stephenson, 2003). Conversely, participating in high-risk activities may be disadvantageous as the likelihood of achieving a successful outcome is diminished, and potential costs, including severe injury or death, may not outweigh the benefits. By distinguishing between degrees of risk, we investigate how optimism may be associated with children’s risk-taking behaviors when a successful outcome is more or less likely.

Age and gender are often identified in the risk literature as individual characteristics that influence children’s risk behaviors. However, findings from the extant literature lack consistency. Some studies report no age differences in 6- to 12-year-old children’s risk perception (Hillier & Morrongiello, 1998; Morrongiello & Rennie, 1998) or risk-taking (Morrongiello & Matheis, 2007; Morrongiello et al., 2008). Others have found that among 3- to 11-year-olds, older children perceive some activities as more dangerous (Boles et al., 2005) and engage in more dangerous activities than younger children (Boles et al.; Ginsburg & Miller, 1982). When gender differences are reported, most studies find that boys perceive activities as less dangerous (Hillier & Morrongiello; Morrongiello & Rennie) and engage in more dangerous activities than girls (Boles et al.; Ginsburg & Miller, 1982). Yet many studies do not find gender differences in risk perception (Boles et al.) or risk-taking (Morrongiello & Matheis; Morrongiello et al., 2009). To further this issue, age and gender were identified as a priori factors of interest to investigate as a function of risk perception, risk-taking, and optimism.

Present Research

The present research assessed links between risk and optimism in 4- to 8-year-old children. We first created a task to capture distinctions between physical activities that varied in degree of risk, then tested children’s perceptions of danger (risk perception) and willingness to engage in each activity (risk-taking). Next, we tested whether children who made more optimistic predictions in an unrelated task were more (or less) likely to perceive the activities as dangerous and be willing to engage in them. We predicted that, overall, children would perceive activities with more risky elements as more dangerous and be less willing to engage in them. Further, if optimism is linked to motivation, children who are more optimistic should perceive lower risk activities to be less dangerous, and therefore be more willing to engage in them compared to children who are less optimistic. If optimism becomes disadvantageous in certain contexts, then more optimistic children should also perceive higher-risk activities as less dangerous and be more willing to engage in them than less optimistic children.

Method

Participants

Participants were 121 4- to 8-year-old children (M=6;6, 58 girls) who were recruited and tested at a science museum in the Midwestern United States. Child racial demographics were as follows: White=74%, Black=7%, Asian=5%, Bi/Multiracial=9%; 3% identified as Hispanic. Children were primarily from middle to upper-middle class families (household incomes: <$50,000=11.7%, $50,000-$74,999=24.8%, $75,000-$149,000=40.5%, >$150,000=17.4%). Three additional children were excluded due to response biases suggesting lack of attention or understanding (e.g., giving the same response 10+ consecutive times). The decision to target 24 children/age was made a priori. Recruitment stopped after the session in which 120 children enrolled, resulting in ~24 children/age group with a roughly equal distribution of boys and girls (Table 1). An additional 35 4- to 8-year-olds (M=6;4, 20 girls) participated in a preliminary stimuli norming task. Study procedures were approved by the Washington University Institutional Review Board (#201805004). Written consent from caregivers and verbal assent from children was obtained.

Table 1.

Sample characteristics by age and gender

Age	Boys (n)	Girls (n)	Total
4	13	11	24
5	13	11	24
6	12	12	24
7	12	13	25
8	13	11	24
Total	63	58	121

Materials & Design

Each child completed three tasks in a fixed order on a laptop computer: risk perception, risk-taking, and optimism. Consistent with recommended best practices in individual differences research, a fixed order was used to minimize random error variance across tasks (see Goodhew & Edwards, 2019, for a review).

Risk Materials.

To measure risk perception and risk-taking, we created a task that captured distinctions between novel physical activities that varied in degree of risk. These novel activities were loosely based on common activities but adapted expressly for this study to ensure children would not have prior experience with the activities. Increasing risk was achieved by adding features including height, speed, hazardous elements, and chance of getting lost (Sandseter, 2007). Lower-risk activities captured typical physical activities most children engage in with only slight risk of minor injuries. Medium-risk activities were more challenging with several moderately risky elements and increased risk of minor to moderate injuries. Higher-risk activities were extremely difficult with multiple hazards that could cause severe injury.

A norming procedure was first used to generate eight novel outdoor activities (from an initial set of 12), each with three scenarios that children reliably categorized into three risk-levels (low, medium, high). Brief descriptions were displayed below each scenario and included the name of the novel activity, an explanation of how the activity is performed, and a description of the risky elements (Figure 1; Supplement 1). Children rated each scenario on a 4-point scale from not dangerous to very dangerous. To select the eight best activities, we first eliminated two in which the scenarios did not receive increasing ratings of danger with increasing risky elements. Of the remaining ten, the eight with the largest difference between the lowest and highest ratings for the three scenarios were selected, resulting in a final set of 24 scenarios.

Figure 1.

Sample Outdoor Activity Scenario

Risk-Task Design.

One order for the 24 risk scenarios was created such that no more than two scenarios of the same risk-level were presented consecutively and 2+ trials separated scenarios of the same activity to inhibit children from directly comparing scenarios between risk-levels or activities. The same order was used for risk perception and risk-taking.

Optimism Task.

Optimism was assessed using a child-friendly version of a probabilistic reasoning task modeled on an adult card game (Lench & Ditto, 2008) and adapted for young children (Hennefield & Markson, 2022). Children played a computer-based game with sets that vary in ratio of “winning” to “losing” tokens. For each set, children predict whether they will receive a winning or losing token. Their responses are then modeled as a function of the number of winning tokens in the set, with responses that were more positive than the true likelihood considered optimistic.

The task was programmed using PsychoPy (Pierce et al., 2019) and included two warm-up and five test trials. Test trials involved winning (green squares) and losing (red triangles) tokens presented in five ratios (5:1, 4:2, 3:3, 2:4, 1:5). Each child received the same pseudorandomized order. Shape stickers replaced two laptop keys and recorded children’s responses.

Parent Reports.

The Injury Behavior Checklist (IBC) assessed children’s risk-taking frequencies (Speltz et al., 1990). Parents reported how often their child engaged in behaviors that could result in injury over the past six months, with frequencies summed to create a composite score. The IBC was included to assess how parent-reported risk-taking compared to the present risk measures.

Procedure

For risk perception, the experimenter presented a scenario, read the description aloud, and asked the child how dangerous they thought the scenario was (e.g., How dangerous do you think it is to ride this wheelie?). For risk-taking, the experimenter presented each scenario again and asked if they would try the activity (e.g., Imagine you are here. Do you think you will or will not try to ride this wheelie?).

The optimism task began with two warm-up trials to familiarize the child with the task mechanics. The child was then taught that green squares win 10 points, red triangles lose 5 points, and the goal is to win points. For each target trial, the child viewed six tokens. The tokens were then hidden inside capsules, and the capsules mixed in a spinning wheel (Figure 2). To provide a sense of agency over the outcome, the child pressed a key to stop the wheel and receive a capsule. Then they predicted which shape was inside the capsule by pressing the corresponding key. Capsule contents were only revealed upon task completion to ensure that children’s responses were not affected by feedback from previous predictions.

Figure 2.

Optimism Task Sample Trial

Analysis Approach

Data Coding.

Risk-perception ratings were coded as 1 = not dangerous¸ 2 = a little dangerous, 3 = somewhat dangerous, 4 = very dangerous. Based on the mean risk-perception ratings for each activity, three data-driven risk levels were created by categorized scenarios with ratings between 1–2 as low-risk (8 scenarios), 2–3 as medium-risk (6 scenarios), and 3–4 as high-risk (10 scenarios; Figure 3). Risk-taking responses were coded as 0 if children stated that they will not try the activity and 1 if they will try the activity. In the optimism task, children’s predictions were coded as 0=losing token and 1=winning token.

Figure 3.

Children’s Risk Perception as a Function of Risk Level

Note. Children’s ratings of perceived risk increased with increasing number of dangerous elements (indicated above with an icon for each element). Based on these ratings, the scenarios were categorized as low, medium, or high-risk. N=121

Analysis Approach.

We first examined each measure independently to provide descriptive information and examine age and gender effects and then tested our a priori hypotheses concerning the relationships between risk perception, risk-taking, and optimism in a single model.

To examine risk perception, a RMANOVA was conducted with risk-level (low, medium, high) for the risk perception scores as the within-subjects factor, and gender and age as between-subjects factors. A parallel RMANOVA was then conducted with risk-level (low, medium, high) for the risk-taking scores as the within-subjects factor. Pairwise comparisons were used to probe significant effects. Finally, Pearson correlations between the IBC score and risk-taking were conducted to assess overlap between parent-reports of children’s risk-taking behaviors and children’s willingness to engage in the activities.

To assess optimism, children’s predictions of receiving a losing or winning token were used in Hierarchical Generalized Linear Models (HGLM) with binomial distribution and logit link function to model the probability of children predicting the winning token as a function of number of winning tokens present within the set (i.e., “winning tokens possible”, a continuous variable centered at 3). HGLMs were used to account for multiple trials per child and allowed modeling intercept and number of winning tokens as both fixed and random effects. A compound symmetry covariance structure was used, as that was the covariance structure with the best-fitting model (i.e., lowest Bayesian information criterion; BIC). Model 1 (optimism) modeled children’s prediction of selecting a winning token as a function of winning tokens possible with age, gender, and income as covariates, and 2-way interaction terms for each covariate with winning tokens possible. Model 2 (optimism and risk) included the covariates from Model 1, risk perception and risk-taking scores at each risk-level, and 2-way interaction terms for each significant covariate from Model 1 with each risk measure with winning tokens possible. There were no significant interactions between winning tokens possible and any risk measures; thus, these interaction terms were not included in the final model (Supplement 2).

Results

Risk Measures

Overall, children rated scenarios as more dangerous as the number of risky elements increased. Children were also less likely to state they would try activities perceived as higher risk (Table 2)

Table 2.

Average rating of danger (risk perception) and percent of children willing to engage in each activity (risk-taking) across the three levels of risk

	Level of Risk
Task	Low	Medium	High
Risk perception (M; Range 1–4)	1.46	2.59	3.49
Risk-taking (%)	86	46	13

Note: N=121

For risk-perception, the RMANOVA demonstrated a significant main effect of risk-level, F(2, 222)=878.557, p<.001, ηp²=.89, which was expected based on how the variables were created. Pairwise comparisons revealed that children rated low-risk scenarios as least dangerous, followed by medium-risk, then high-risk (ps.<.001; Figure 4). There was also a significant interaction between risk-level and age, F(8, 222)=7.687, p<.001, ηp²=.22, which appears driven by older children’s ratings of low-risk scenarios as less dangerous and high-risk scenarios as more dangerous relative to younger children. Pearson correlations between risk-level and age support this interpretation: low-risk r=−.39, p<.001, medium-risk r=−.03, p=.977, high-risk r=.41, p<.001.

Figure 4.

Children’s Risk Perception and Risk-Taking

Note. Children’s perception of degree of danger (top) and willingness to engage in the scenarios (bottom) as a function of risk-level and age. Asterisks (*) indicate significant group differences, p < .001. N=121

For risk-taking, the RMANOVA similarly demonstrated a significant main effect of risk-level, F(2, 222)=482.560, p<.001, ηp²=.82. Pairwise comparisons revealed that children were most willing to engage in low-risk activities, followed by medium-risk, then high-risk (ps.<.001). There was also a significant interaction between risk-level and age F(8,222)=5.078, p<.001, ηp²=.18, which appears driven by older children’s increasing willingness to engage in low-risk activities and decreasing willingness to engage in high-risk activities relative to younger children. Pearson correlations between risk-level and age support this interpretation: low-risk r=.37, p<.001, medium-risk r=.05, p=.593, high-risk r=−.27, p=.001. A trend-level interaction between risk-level and gender, F(2,222)=3.00, p=.052, provides marginal support to suggest that boys were more likely than girls to engage in higher-risk activities.

The IBC was significantly correlated with children’s overall willingness to engage in the activities, r=.23, p=.016, and more specifically with the low- (r=.27, p=.018) and medium-risk (r=.25, p=.009), but not high-risk (r=.04, p=.704) activities. These findings provide support for the validity of our risk-taking measure to assess risk-taking behaviors. Moreover, items queried in the IBC assess a wide range of potentially risky behaviors (e.g., “stands on chairs”, “plays with sharp objects”) which may align best with our low- and medium-risk activities, but are notably distinct as they are commonplace activities.

As perceptions of risk increased, children demonstrated less willingness to engage in those activities, supporting the validity of this task to measure risk. Results also suggest developmental differences. Whereas children across all ages acknowledged that activities with minimal hazards were relatively safe, older children were more likely to try them. However, older children perceived high-risk activities as more dangerous, and were less willing to engage in them.

Optimism.

The estimated probability of predicting a winning token for the 1:1 ratio of winning to losing tokens was .88. This is greater than the .5 expected if children responded solely based on probability, indicating that, overall, children’s responses were more optimistic than realistic. Model 1, which assessed optimism, demonstrated an increase in the likelihood of children predicting they would receive a winning token as winning tokens within a set increased (B=1.19, SE=.26, t=4.53, p<.001). A significant interaction between age and winning tokens possible (B=.33, SE=.12, t=2.79, p=.006) revealed that older children were less likely than younger children to predict a winning token as winning tokens decreased (Table 3). These findings, which mirror those of 3- to 6-year-olds in prior work (Hennefield & Markson, 2022), suggest that children are optimistic while taking into account the true likelihood of the event happening, and with age are better able to integrate probabilistic information into their predictions.

Table 3.

Detailed statistics for Model 2

Variable	Est.	95% CI	t	p
Age	−0.31	−0.63, −0.01	−2.01	.045
Male gender	−0.27	−0.89, 0.34	−0.87	.383
Income	−0.16	−0.46, 0.13	−1.09	.278
Number of winning tokens possible	1.33	0.89, 1.77	6.04	<.001
Low risk perception	0.19	−0.73, 1.11	0.40	.689
Medium risk perception	0.34	−0.38, 1.07	0.93	.353
High risk perception	−1.44	−2.47, −0.42	−2.77	.006
Low risk taking	0.04	−0.17, 0.25	0.40	.690
Medium risk taking	0.26	0.01, 0.51	2.07	.039
High risk taking	−0.24	−0.46, −0.04	−2.30	.022
Age X winning tokens possible	0.33	0.10, 0.57	2.79	.006

Est. = Estimate

Note. This table presents results from a hierarchical generalized linear model of predicting a winning token by risk perception, risk-taking, and number of winning tokens possible covarying for age, gender, and the interaction of age and number of winning tokens possible. N=121

Risk and Optimism

Model 2, which assessed the relationship between risk and optimism, revealed that children who were willing to try more medium-risk activities were more likely to predict winning tokens (B=.26, SE=.13, t=2.07, p=.039). As predicting more winning tokens indicates greater optimism, this supports our hypothesis that more optimistic children take more risks in situations that are most conducive to exploration and learning. In contrast, children who were willing to try more high-risk activities were less likely to predict winning tokens (i.e., were less optimistic; B=−.25, SE=.11, t=−2.30, p=.022; Figure 5). Moreover, although there was no significant relationship between children’s perceptions of medium-risk activities and predicting winning tokens (B=.34, SE=.37, t=.93, p=.353), children who perceived high-risk activities as less risky were more likely to predict winning tokens (B=−1.44, SE=.52, t=−2.77, p=.006). Thus, more optimistic children perceived high-risk activities to be less dangerous – yet they were less willing to try those activities than less optimistic children.

Figure 5.

Children’s Risk-Taking as a Function of Optimism and Risk Level (Model 2)

Note. ^aIncreased willingness to take risks is indicated via the dashed lines which reflect risk-taking scores +1 standard deviation above the mean. ^bLower willingness to take risks is indicated via the dotted lines which reflect scores −1 standard deviation below the mean. N=121

General Discussion

The present study investigated how optimism relates to children’s perception of and willingness to engage in risk-laden physical activities. Our findings illuminate distinct roles that optimism might play in how children evaluate and decide to try moderately challenging or extremely high-risk activities. Specifically, more optimistic children were more likely to try moderately risky activities than less optimistic children. This finding aligns with the notion that optimism is motivational (e.g., Bjorklund, 1997), as it is precisely these types of moderate-risk activities that provide ideal conditions for learning – that is, children are stretched beyond their comfort zone but are not likely at risk for serious injuries. Further, whereas more optimistic children perceived the higher-risk activities as less dangerous than less optimistic children, they were less likely to try them. Thus, optimism might serve a motivational role for engaging in activities that provide moderate challenges conducive to learning and growth, but a protective role against engaging in activities that involve great hazards and may cause severe injuries.

Contrary to our prediction that optimism might be advantageous in some contexts and disadvantageous in others, the present findings instead suggest that optimism is broadly advantageous – although how these advantages manifest differ across contexts. We posit this is because the likelihood of achieving a successful outcome is low for high-risk activities, but more attainable for moderate-risk activities. Optimistic biases are most apparent in contexts where the likelihood of a positive outcome is uncertain (e.g., Lench & Ditto, 2008). Thus, when facing moderately challenging activities, children higher in optimism might attempt those activities as there is a reasonable chance of a successful outcome. In contrast, children higher in optimism might be less inclined to attempt higher-risk activities with a low chance of success.

Overall, children’s perceptions of how dangerous the activities were aligned with our objective delineation of risk (i.e., number of risky elements) and their willingness to try the activity. That is, children were less willing to try activities they perceived as more dangerous. There was also marginal evidence that boys may engage more in higher-risk activities than girls, which aligns with findings from prior studies (e.g., Boles et al., 2005; Mytton et al., 2009). However, in contrast to several studies which have found 6- to 11-year-old boys perceive typical play activities as less dangerous that girls (Hillier & Morrongiello, 1998; Morrongiello & Rennie, 1998), our study did not find gender differences in risk perception. This may be due to the younger age of our sample (4- to 8- years), raising the possibility that gender differences in risk perception may not emerge until later in childhood. Additionally, our use of novel physical activities instead of typical play activities may have prompted children to more directly evaluate the risks associated with each activity rather than basing their evaluations on their own past experiences and capabilities.

Within our sample of 4- to 8-year-old children, older children perceived higher-risk activities as more dangerous and were less willing to try them than younger children. This contrasts with research demonstrating that across ages 3- to 11 years, older children engage in more risky behaviors (Ginsberg & Miller, 1982) and that across ages 4- to 7-years, older children perceive dangerous activities to be more risky but do not show lower engagement in dangerous activities than younger children (Boles et al., 2005). These conflicting findings likely reflect the extreme danger depicted in high-risk scenarios in the present study, along with older children’s stronger abilities to assess hazards, weigh possible outcomes, and/or inhibit risky behaviors. Moreover, it is also possible that children older than those tested in the present study (i.e., older than 8 years) might be more willing to try the higher-risk activities. Future research is needed to assess this possibility along with the implications of children’s increased physical competencies with age, and prior experiences with related activities, on risk perception and risk-taking. Interestingly, the present study did not find any interactions between optimism, risk, and age. This suggests that relationships between optimism and risk are present from early childhood and highlights a potential role for optimism to impact the development of risk across childhood – or, conversely, for risk behaviors to impact the development of optimism.

One study limitation thus concerns the directionality of the relationship between risk and optimism. Whereas our primary interpretations of the findings are based on the assumption that optimism is a generally stable bias that underlies decisions pertaining to risk, it is possible that by taking more moderate risks – that potentially lead to successes – children become more optimistic; or conversely, by taking more high risks – that potentially lead to failures – children become less optimistic. Future studies could use longitudinal or experimental designs to untangle these possibilities.

A second limitation concerns whether children’s expressions of willingness to take risks reflects actual risk-taking behaviors. There is evidence that children’s intentions to take moderate risks accords with actual risk-taking (e.g., Morrongiello, 2004); however, whether this holds for all degrees of risk is unknown. The small-to-moderate correlations between higher frequencies of children’s actual day-to-day risk-taking behaviors measured via parent-report and children’s stated willingness to engage in the novel outdoor activities lends support and validity for the task aligning with actual risk-taking behaviors. Furthermore, the present study focused on physical risk-taking as a concrete and familiar form of risk. As consequences in the physical domain differ from other domains (e.g., academic, social), factors influencing risk-taking might also differ across domains (Morrongiello et al., 2009). There is increasing recognition of the importance of examining risk in domains other than physical activity (Cooke et al., 2019), and the present methods provide a template through which risk can be investigated more broadly.

A third limitation concerns the assumption that achieving a successful outcome is the primary goal of participating in physical activities. This is particularly problematic with higher-risk activities where other factors, such as the thrill of trying something dangerous or the feeling of accomplishment from being successful at a higher-risk activity, might drive some children’s decisions. Accounting for these possibilities will be an important direction for future studies.

Finally, it is important to interpret our findings with reference to our study sample. Our sample was what is considered predominantly WEIRD (White, educated, industrialized, rich, and demographic; Henrich et al., 2010). The extent to which processes driving risk perception and risk-taking functions in similar ways in other populations may differ. Additionally, the types of risks present in the environment of other populations may differ, and the cultural norms and values across different populations may vary which may drive differences in what is considered risky and in children’s perception of and inclinations to take risks (Lam, 2005).

From a practical perspective, the present findings offer some directions for injury prevention in outdoor play. Whereas equipment safety modifications and adult supervision are commonly recommended injury prevention strategies (Morrongiello & Schell, 2010), they primarily rely on external environmental factors to (ostensibly) enhance safety. There is also increasing recognition of the long-term negative impact of highly restrictive play on children’s mental and physical health (Wyver et al., 2020). Our finding that, among 4- to 8-year-olds, older children were better able than younger children to appropriately judge the highest risk activities as most dangerous and were less willing to engage in those activities, suggests that younger children may need greater support to align their risk-taking to their cognitive and physical capabilities. Thus, attunement to young children’s developing motor and cognitive skills, and subsequent modifications to equipment or supervision, may be used to create optimal environments for children to play safely and explore. Furthermore, the finding that children lower in optimism perceived high-risk activities as more risky yet were still inclined to engage with them suggests that interventions might target children’s abilities to align their behaviors with their (fairly accurate) perceptions of risk, rather than altering perceptions. In this vein, intervention studies indicate that exposure to “risky” play activities under a controlled setting can help children better evaluate and identify ways to reduce potential risks (Lavrysen et al., 2017). Similarly, strategies such as targeting children lower in optimism for preventative intervention, may be preferable to placing broad external restrictions on children’s play.

To recap, the present findings indicate two seemingly beneficial relationships between optimism and risk in childhood. First, children higher in optimism are more willing to engage in moderately challenging activities relative to children lower in optimism. Second, children higher in optimism are less willing to engage in higher-risk activities. These findings support the possibility that optimism motivates children to engage in beneficial moderately challenging activities and protects them from engaging in severe injury-inflicting activities. These broadly positive benefits of optimism also underscore the importance of understanding how optimism develops and operates across early childhood.