Early Childhood Risk-Taking Behaviors in Chinese Parenting Environment: The Mediating Role of Parental Supervision

ABSTRACT

Research Findings: Parental supervision, parental risk perception, and parental risk attitudes constitute an important parenting environment for the development of young children. To examine the mechanisms of Chinese parenting environment on young children’s risk-taking behaviors, this study first established a four-factor model and surveyed 497 parents of preschoolers. The results indicated that: (1) Chinese young children’s risk-taking was at moderately low level, and parental risk perceptions, parental attitudes and parental supervision were at moderately high level; (2) parental supervision completely mediated the relationship from parental risk perception to children’s risk-taking behavior (β = -0.139,95%CI [−0.267, −0.062], p < .001); (3) parental supervision partially mediated the relationship from parental risk attitudes to children’s risk-taking behavior (β = -0.068, 95%CI [−0.162, −0.027], p < .001); (4) this mediation model remained stable cross genders of children (∆χ² = 12.022, p = .799). Practice or Policy: The findings indicated that sufficient support and guidance for parenting should be provided to improve parental risk perceptions, attitudes and supervision which in turn create an environment safe and appropriate for risk-taking of Chinese young children.

Risk-taking behavior refers to the different choices individuals make in an uncertain situation. When faced with an approach-avoidance conflict, individuals may engage in risky behaviors in order to achieve a more valuable or beneficial outcome that meets their needs (Chen, 2003). Risk-taking behaviors in early childhood are mainly expressed as physical risk-taking into unknown situations or with uncertainty (Davis & Eppler-Wolff, 2015). The consequences of such behaviors may be beneficial (e.g., feeling happy and gaining learning) or harmful (e.g., physical pain and mental injury) (Gong et al., 2021; Y. Hu, 2015).

Risk-taking behavior in early childhood is a “double-edged sword.” Young children have a natural psychological need to have adventures with danger and excitement during which they take risks, learn how to cope with the unknown, and overcome their fears (Sando et al., 2021). Allowing young children to access appropriately to different risk situations is a very important process of learning to explore the world (Huang et al., 2014; Liang et al., 2019). If risk-taking was removed or constrained in childhood, this can lead to a decline in physical functioning, a lack of independence, and even many psychological problems (depression, lack of self-confidence and empathy, etc.) (Gray, 2014). However, unintentional injuries resulting from inappropriate risk-taking behaviors often cause certain physical and psychological hurts to young children, which is a significant threat to the safety and health of young children (DeGeorge et al., 2020).

Literature shows that three factors in parenting environment have direct relations with young children’s risk-taking behavior: parental supervision, parental risk perceptions, and parental risk attitudes (Lee et al., 2022; B. A. Morrongiello & McArthur, 2018; Napper et al., 2015; Sneed et al., 2015). In addition, Sandseter (2010) proposed a conceptual model that illustrated the role of parental supervision as a filter in young children’s risk-taking behaviors. Other factors (e.g., parents’ perceptions and attitudes, children’s temperament, and their own previous experiences) may influence the children’s propensity to take risks by impacting parental supervision (Sandseter, 2010). Existing empirical studies supported the mediating role of parental supervision in two relationships: (1) the relationship between parental risk perception and children’s risk-taking behaviors (Farizan et al., 2020); and (2) the relationship between parental risk attitudes and children’s risk-taking behaviors (Karaca, 2020). When all the three parenting factors are considered simultaneously in young children’s risk-taking behaviors, there is no empirical research evidence to date to support the mediating role for parental supervision. In China, the situation of unintentional injuries in early childhood is exceptionally serious (Gong et al., 2021). Parents are torn between giving their children the freedom to take risks in order to unleash children’s nature or protecting them from harm (Hao, 2020). However, the majority of studies about the relationships between parenting mechanism of children’s risk-taking behaviors were conducted out of China. Due to the significant differences in risk-taking behaviors of young children and parenting behaviors across cultures (Hao, 2020), the findings of previous studies do not provide good lessons for Chinese parents. Therefore, based on the of model of young children’s risk-taking behavior, this study established a model (Figure 1) to explore the effects of the three parenting factors on Chinese young children’s risk-taking behavior, using parental attitudes and parental perceptions as independent factors, and parental supervision as a mediating factor.

Figure 1. Hypothesized mediation model of young children’s risk-taking behaviors.

Three Parenting Factors in Relation to Children’s Risk-Taking Behaviors

Parental supervision refers to care and control of children by parents in parenting (Saluja et al., 2004). Parental supervision has been found to be a significant predictor of young children’s risk-taking behaviors. Under looser parental supervision, young children obtain more opportunities for freedom and independence and thus present more risk-taking behaviors (B. A. Morrongiello & McArthur, 2018). More failures in parental supervision may lead not only to more injuries to the child, such as fall injuries (Harrell, 2003), choking (Pollack-Nelson & Drago, 2002), poisoning (Alqahtani et al., 2021), and drowning (Anderson et al., 2021), but to more serious injuries (Schwebel et al., 2006). On the contrary, improved parental supervision effectively reduced the likelihood of risk-taking behaviors and unintentional injuries in young children by constant attending to and being within reach of the children across contexts (e.g., water area, road, agricultural settings, and home) (Barton & Huston, 2012; B. Morrongiello & Schell, 2010; B. A. Morrongiello et al., 2012; Peden & Franklin, 2020). By constraining children’s risk-taking, parental supervision is the most effective method of decreasing unintentional injuries in early childhood (B. A. Morrongiello, 2018). However, high levels of parental supervision deprive children of the much fun and growth they deriving from ongoing exploration and challenge (Sandseter, 2010).

Risk perception refers to people’s subjective evaluation about the characteristics and severity of risk (Myhre et al., 2020). Risk perception varies by the specific risk people are facing and on the context in which the assessment is taken (Cohen, 2015). In parenting, parental risk perception consists of two main dimensions: perception of the child’s supervision needs and perception of the likelihood of injury to the child (B. A. Morrongiello et al., 2014). Parental risk perceptions negatively affect children’s risk-taking behaviors. If parents believed the child was likely to be injured and had a high need for supervision, they may have more safety precautions and communication with the child. As a result, the child is less likely to engage in a risk-taking behavior (Napper et al., 2015, Sneed et al., 2015). Alternatively, if parents judged their child to be able to manage injury risk for her/himself, parents often assume that they do not need to closely supervise the child, which elevates children’s risk of injury (B. A. Morrongiello et al., 2014). For example, perceived improvements in swim ability produce parents to become more confident that their child is capable of keeping themselves safe near water, which may increase the likelihood of the children approaching the water and drowning (Keatinge et al., 1969). Thus, raising parents’ risk perception and making them perceive a negative consequence as more likely and serious can reduce the likelihood of the behavior with the perceived risk.

Being different from parental risk perception, parental risk attitude is likely to be persistent across contexts. Parents’ risk attitude is their stable orientation toward risk without specific information in children’s daily life (You et al., 2019). The risk attitudes encompass parents’ general beliefs regarding to the learning value and potential benefits associated with childhood risk-taking and injuries (Lewis et al., 2004). Similar to risk perceptions, parents’ risk attitudes also directly associate with young children’s risk-taking behaviors (Lee et al., 2022). However, the direction of the effect of risk attitudes on children’s risk-taking might be different from that of risk-perceptions. If parents embrace a positive attitude toward risk-taking behavior, seeing it as having value and benefits for their child’s growth, they tend to offer fewer warnings and interventions when faced with some of the child’s adventurous actions. Their responses are relatively slower, and they might even explicitly encourage the child to engage in certain risk-taking activities (Zheng & Zhang, 2009). Children of parents who have positive attitudes toward risk exhibit more risk-taking behaviors and get more injuries (Lu et al., 2023). On the other hand, if parents hold a negative view of risk-taking behavior, they are more likely to provide warnings and protection, emphasizing the potential harm that adventurous actions may cause (Zheng & Zhang, 2009). Divergent parental attitudes toward risk-taking can also predict a child’s attitudes of injuries which, in turn, directly related to their own risk-taking behaviors (B. A. Morrongiello & Dawber, 2000). Parents’ risk perceptions and risk attitudes can both independently play a role in risk-taking of early childhood, but the roles they play are distinct. How these two factors simultaneously interact in shaping children’s risk-taking behaviors remains a question the requires further investigation.

Parental Supervision Plays a Mediating Role in Children’s Risk-Taking Behaviors

According to Sandseter’s conceptual framework (2010) and previous research, parental supervision plays a mediating role in children’s risk-taking behaviors. Sandseter (2010) proposed the first conceptual model applicable to risk-taking behaviors in preschoolers. In this model, adults’ supervisory behavior is presented as a kind of filter to mediate the effect of other factors on children’s risk-taking behaviors. Main caregivers constantly watch children’s actions and intervene if they thought it was necessary. Children’s risk-taking behaviors are allowed to occur or are suppressed largely by parents in supervision. And the extent to which parents devote supervision to responding to their children’s risk-taking, supporting or stopping it, is heavily correlated with their own evaluation of different risks. Therefore, how parents perceive risks and how they then supervise their children collectively affect the level of risk-taking in children. Previous studies also supported the mediating role parental supervision played in the relation between parental risk perceptions and children’s risk-taking behaviors. B. Morrongiello et al. (2010) found that parental risk perceptions had an effect on children’s risk-taking behaviors through parental supervision. When parents perceive that children have a low level of need for supervision (e.g., they do not believe that children need close constant supervision) and a low likelihood of injury, parents will reduce the level of supervision and encourage or allow children’s risk-taking behaviors (Farizan et al., 2020). Conversely, parents will tend to invest more supervision to discourage or reduce the occurrence of risk-taking behaviors if they assessed that children were unable to handle the risk, had a higher likelihood of getting injury, and required more supervision (Truong et al., 2022).

Parental risk attitudes may also have an effect on children’s risk-taking behaviors through parental supervision. According to Sandseter’s theoretical model (2010), parents’ supervision, constraining or enabling children to experience risks and challenges, is also colored by their attitudes toward risks in addition to perceptions. She took Norway as an example, people in this country have a more liberal attitude toward outdoor life and risky play, and they are particularly supportive of risky activities as an important part of children’s learning and growing up. Many studies have shown that if parents had positive attitudes (e.g., “It is beneficial for children to get hurt, there is no pain and no gain”), they are inclined to weaken supervision, provide children’s risky activities with support, and children will take more risks (B. A. Morrongiello et al., 2010). Conversely, those parents who have negative attitudes (e.g., “injuries should be completely avoided in the development of young children”) may elevate their supervision to prevent their young children from potential risks, thus the level of children’s participation in risks decreases (Karaca, 2020). However, there is a small body of research showing that parental supervision would be positively aligned with attitudes or sometimes the association between the two is not significant. For example, in Niehues’s study (2016), parents placed a high value on risk-taking in early childhood while retaining a high level of supervision of their children. Parents said that even if they thought risk-taking was beneficial, they never let their children do many things they had done alone as children (e.g., playing in the bush) without constant supervision both because of their own fear, but also because parents believed that today was so much more dangerous than their time. There are several other studies showed a “attitude-practice gap” existing in caregivers. A recent study found that Lebanese mothers had fair risk attitude but poor supervisory practice to their children (Al-Hajj et al., 2023). The mixed findings on the relationship between parental risk attitudes and supervision all come from different contexts than China’s (e.g., Australia, Canada, Croatia, Lebanon, Norway, U.S.). Therefore, the consistency between parental supervision and risk attitudes, especially in Chinese caregivers, needs to be examined by further research.

Children’s Gender Differences in the Relationships of Risk-Taking Behaviors

The limited research findings on gender differences in the relationships between children’s risk-taking and parenting factors are mixed. For example, girls’ risk-taking engagement was more likely to be influenced by parental perceptions (B. A. Morrongiello & Dawber, 2004). If parents considered a certain activity to be risky and discouraged it, girls tended to avoid such activity. On the other hand, boys were more focused on whether the activity could bring them enjoyment (B. A. Morrongiello & Dawber, 2004). Zheng and Zhang’s study (2009) demonstrated that parents’ attitudes were positively correlated with risk-taking behaviors in both boys and girls. Parents tended to hold a positive attitude toward boys’ adventurous behavior, perceiving it as valuable and beneficial. As a result, boys exhibited more risky behaviors. Conversely, parents tended to adopt a more cautious attitude toward girls’ risk-taking, emphasizing the potential for harm. Consequently, girls engaged in less risk-taking activities. B. A. Morrongiello and McArthur (2018) found that compared to girls, the effect of parental supervision on children’s risk-taking may be more pronounced for boys. When parents loosen their supervisory behaviors (e.g., present to absent), boys engaged in hazard-directed behaviors, whereas girl’s risk behaviors did not vary with parents’ presence. However, Wang’s study (2016) suggested that the association between parental presence and risk-taking behaviors in young children was not influenced by the child’s gender. Another study (Kogan et al., 2011) indicated that family protective processes, encompassing parental supervision and attitude communication, could predict risk-taking behaviors in both boys and girls. To our best knowledge, there is currently a lack of research that comprehensively investigates potential gender differences in the relationships between these three parenting factors and risk-taking behaviors in young children.

The Current Study

In China, unintentional injuries pose as one of the primary threats to child mortality (Hao & Hsueh, 2018). Appropriate supervision could prevent young children from engaging unsafe activities and support them to accomplish challenges successfully. When parents underestimate or overestimate their children’s skills and environmental risks, inappropriate parental supervision emerges. As a result, parents either become overprotective and obstruct children’s development or put them in danger (J. Yang et al., 2019). The exploration of risk-taking in early childhood in China has just begun in recent years (J. Liu, 2020). Until now, there are no studies on how parental supervision and relevant factors (i.e., parental risk perceptions and parental risk attitudes) in relation to risk-taking behaviors of young children in China. Even studies on the relationship between each of the three parenting factors and children’s risk-taking behaviors are overwhelmingly from outside China. It is necessary to examine how early childhood risk-taking varies with the parenting factors. Moreover, it is important and meaningful to further research on how Chinese parents support children for adaptive risk-taking and protect them from harmful injuries.

Based on the model of young children’s risk-taking behaviors (Sandseter, 2010) and previous studies, this study was first to construct and test a mediation model (Figure 1) that examined how three parenting factors (i.e., parental supervision, parental risk perceptions, and parental attitudes) in relation to young children’s risk-taking behaviors. To achieve this purpose, this study had three hypotheses to test:

H1:

Parental supervision mediated the relationship between parental risk perception and children’s risk-taking behaviors;

H2:

Parental supervision mediated the relationship between parental risk attitudes and children’s risk-taking behavior;

H3:

The mediating role of parental supervision varied across different genders of young children.

Methods

Participants

In this study, children and their parents were recruited from two preschools (one public preschool and one private preschool) in Kaifeng, a city in the middle of China, by using a convenience sampling method. Paper questionnaires were distributed and collected after obtaining the consent of the directors and parents. A total of 497 questionnaires were collected. The mean age of the children was 52.46 months (SD = 13.62), and 226 were girls. The mean age of the parents was 31.36 (SD = 4.98), and 288 were mothers. The parents’ education level was 34.01% for “secondary school and below,” 22.54% for “secondary or high school,” 42.86% for “college or bachelor’s degree,” and 0.40% for “graduate and above.”

Measures

Parent Supervision Attributes Profile Questionnaire (PSAPQ)

Parental supervision was assessed using the Parent Supervision Attributes Profile Questionnaire (PSAPQ) (B. A. Morrongiello & House, 2004). PSAPQ was widely used in different studies to examine parental protectiveness and supervision of young children, and all showed good psychometric properties (see Andrade et al., 2013). This scale contained 29 items (e.g., “I try to keep my child away from all hazards”), covering four dimensions: protectiveness, supervision beliefs, tolerance for children’s risk-taking, and belief in fate. A 5-point scale was used, ranging from 0 “never” to 4 “all of the time.” Higher scores indicated higher levels of supervision, that is, parents had more engagement with supervision for their children. This scale achieved good measurement results in Chinese samples, with good reliability (Cronbach’s α = 0.840) and validity (concurrent validity r = 0.170, p < .010) (J. Yang et al., 2019). In this study, the Cronbach’s alpha of this scale was 0.792.

Situational Supervision Questionnaire (SSQ)

Parental risk perception was rated using the Situational Supervision Questionnaire (SSQ) (Damashek et al., 2014). The SSQ was administered by asking parents two questions about each behavior listed in Injury Behavior Checklist (IBC): perception of child supervision needs (e.g., ”How much do you think the child needs adult supervision when the child climbs high on the furniture?‘), and perception of the likelihood of injury to the child’ (e.g., ”How likely do you think the child is to be injured when the child climbs high on the furniture?”). A 4-point scale was used. Higher scores indicated higher levels of risk perception of the child’s risk-taking behaviors. When faced with the same behavior, parents with higher levels of risk perceptions perceived it as more dangerous, were more likely to believe that the child would be injured, and would need more supervision. The SSQ was previously used in a Chinese sample and was shown to have good reliability (Hao & Hsueh, 2019). In this study, the Cronbach’s alpha of the scale was 0.962.

Injury Attitude Questionnaire (IAQ)

Parental risk attitudes were assessed using the Injury Attitude Questionnaire (IAQ) (Lewis et al., 2004). This 14-item scale was widely used to measure parents’ beliefs in two areas: the learning value (e.g., “Some minor injuries may be beneficial to my child as it will help him/her learn to be more cautious”) and “toughening up” characteristics of childhood injuries (e.g., “injury helps my child to be strong-willed”). A 4-point scale was used, ranging from 0 “strongly disagree” to 3 “strongly agree.” Higher scores indicated that parents had more positive attitudes toward their children’s injuries. This scale achieved good measurement results in China, with good reliability (Cronbach’s α = 0.907) and validity (S-CVI = 0.929) (Sun et al., 2021). In this study, the Cronbach’s alpha of this scale was 0.843.

Injury Behavior Checklist (IBC)

Children’s risk-taking behaviors were assessed using the Injury Behavior Checklist (IBC) which was frequently used to assess the risk-taking behaviors of young children (Speltz et al., 1990). The scale listed 24 common risk-taking behaviors (four categories) in preschoolers’ lives: fall injury related (e.g., “jumping on the couch”), burn injury related (e.g., “playing with fire”), sharps injury related (e.g., “using knives”), and others (e.g., “talking to strangers”). Parents checked the frequency of their child’s risk-taking behaviors over the past six months. A 4-point scale was used, ranging from 0 “never” to 3 “at least once a week.” Higher scores indicated more risk-taking behaviors. Cronbach’s α coefficient of Chinese IBC was 0.904 (Tian et al., 2016). In this study, the Cronbach’s alpha of this scale was 0.803.

Procedures

The municipal Bureau of Education and the Institutional Review Board of the second author’s institution approved the study. This study followed the ethical procedures and met the ethical norms and standards. First, researchers contacted the preschool directors and obtained their permission for the data collection. Second, researchers communicated with the supervising teacher of each class. With the assistance of the teachers, the parents were explained about the study. The questionnaires were distributed to the parents who were interested to participate in this study. Then, the teachers collected the completed questionnaires from the parents and handed them to the researchers.

Data Analysis

There was a two-step analysis process. The first step was to apply preliminary analyses to four research factors (i.e., parental supervision, parental risk perceptions, parental attitudes, and children’s risk-taking behaviors), using SPSS 25.0 software. Specifically, preliminary analyses included (1) Harman one-way test for common method bias. (2) descriptive analyses of the four factors; (3) Pearson correlation to analyze correlations between the four factors.

In order to test the three research hypotheses, structural equation modeling was used to test the established mediation model (Figure 1) as the second step, using AMOS software. SEM estimated relationships among the variables, both observed and latent. This model consisted of four latent variables: parental risk perceptions (two observed variables), parental attitudes (two observed variables), parental supervision (four observed variables), and children’s risk-taking behaviors (four observed variables). SEM involved two main steps: measurement model test and structural model test. The examination of the measurement model adopted Confirmatory Factor Analysis (CFA). The goal of CFA was to assess the relationship between latent factors and their observed variables (Figure 2). The examination of the structural model adopted path analysis to examine the relationships between four latent variables. The results of both the measurement model test and the structural model test would use RMR less than 0.050, NFI greater than 0.950, IFI greater than 0.950, TLI greater than 0.950, GFI greater than 0.950, CFI greater than or equal to 0.950, RMSEA less than 0.050, SRMR less than or equal to 0.050 as indicators to assess the goodness of fit of the model (L. Hu & Bentler, 1999; Wu, 2010).

Figure 2. Measurement model of young children’s risk-taking behaviors.

Then, bias-corrected percentile Bootstrap method (5000 replicate samples) was used to test the stability of the mediation effects of the model. At last, a multi-group analysis was used to test equivalence of the model paths between boys and girls. Two models were developed, one of which allowed paths to be different across children’s genders (unconstrained structural paths), while the other restricted path coefficients to be equal for both genders (constrained structural paths). Model fit differences were determined using the likelihood-ratio test, also known as the chi-square difference test (Cheung & Rensvold, 2002). The chi-square difference was calculated as ∆χ² = χ²_c (constrained model) - χ²_uc (unconstrained model). Significance was evaluated with the degree freedom difference (∆df = df_c – df_uc). If the Chi-square difference was significant, unconstrained structural paths were different from constrained structural paths, which suggested that the model paths were different between boys and girls. Then, paths of the constrained model were released, one path at a time, to examine which path had significant difference across genders.

Results

Preliminary Analyses

Results of Harman One-Way Test

The Harman one-way test for common method bias was performed, using a characteristic root greater than 1 (Tang & Wen, 2020). The results found that without rotation the first factor obtained accounted for 21.06% of the loadings, which did not exceed the critical value of 40%, indicating that there was no significant common method bias in this study.

Descriptive Statistics of Four Research Factors

Preschoolers’ risk-taking behavior mean score was 1.03 (SD = 0.53) (see Table 1), with the highest risk-taking behavior score related to falling injuries (M = 1.72, SD = 0.82), the lowest risk-taking behavior score related to burns (M = 0.49, SD = 0.46). Referring to the scoring rules of the Injury Behavior Scale, the current preschoolers’ risk-taking behaviors were generally at a moderate to low level, with a frequency of about once a month, and falling injury-related risk-taking behaviors were the most frequent.

Table 1. Means and standard deviations of research factors and dimensions (N = 497).

	M	SD
Parental risk perceptions	2.264	0.494
Supervision needs	2.537	0.459
Injury likelihood	1.990	0.608
Parental attitudes	2.131	0.527
Learning value	2.110	0.372
Toughening up	2.158	0.692
Parental supervision	3.100	0.566
Protectiveness	3.121	0.621
Supervision beliefs	3.027	0.537
Tolerance	3.131	0.698
Belief in fate	3.069	0.418
Children’s risk-taking behaviors	1.028	0.534
Fall-related behaviors	1.721	0.815
Burn-related behaviors	0.490	0.462
Sharps-related behaviors	1.230	0.881
Other risk-taking behaviors	0.671	0.511

The mean score of parental risk perception was 2.26 (SD = 0.49), of which the mean score of supervision need assessment was 2.54 (SD = 0.46) and the mean score of injury possibility assessment was 1.99 (SD = 0.61). Referring to the scoring rules of the contextualized risk perception scale, parents’ risk assessment of their children’s risk-taking behavior was moderately high: parents generally believed that their children were likely to be injured in the face of their children’s risk-taking behavior, and that adult supervision was highly needed during their children’s risk-taking.

The mean score of parental risk attitude was 2.13 (SD = 0.53). Referring to the scoring details of the Injury Attitude Scale, parents of preschoolers had moderately high risk attitudes and generally showed more positive attitudes toward risk-taking behavior and possible injuries in young children.

The mean score of parental supervision behavior of preschool children were 3.10 (SD = 0.57), including protectiveness (M = 3.12, SD = 0.62), supervision beliefs (M = 3.03, SD = 0.54), tolerance (M = 3.13, SD = 0.70), and belief in fate (M = 3.07, SD = 0.42). Referring to the rules of Parent Supervision Attributes Profile Questionnaire (PSAPQ), parents’ supervisory behaviors were moderately high: parents often took the initiative to remove risk factors, kept an eye on their children’s behavior, and were always around to respond when their children took risks.

The results of correlation coefficients between the four research factors were shown in Table 2. Children’s risk-taking behaviors were negatively correlated with parental risk perceptions and parental supervision, and was positively correlated with parental attitudes. Meanwhile, parental supervision was positively correlated with parental risk perceptions and parental attitudes.

Table 2. Correlation coefficients (N = 497).

	1	2	3	4
1. Parental risk perceptions	–
2. Parental attitudes	−0.098*	–
3. Parental supervision	0.347**	0.099*	–
4. Children’s risk-taking	−0.183**	0.118**	−0.201**	–

*p < .05, **p < .01, ***p < .001.

The Mediating Effect of Parental Supervision

In order to test the three research hypotheses, the structural equation model was tested in two steps; the first step was to test the measurement model. As shown in the Figure 3, there were four latent factors in the measurement model: young children’s risk-taking behavior, parental risk perception, parental risk attitudes, and parental supervision. The results showed that the data fitted well with the measurement model [χ² = 65.946, p = .221, χ²/df < 2, RMR = 0.013, NFI = 0.965, IFI = 0.989, TLI = 0.986, GFI = 0.987, CFI = 0.995, RMSEA = 0.017 (90% CI 0.004, 0.033), SRMR = 0.015]. In addition, the factor loadings were all significant, which indicated that the latent factors could represent the observed variables well.

Figure 3. Standardized estimates of the mediation model of young children’s risk-taking behaviors (N = 497). *p < .05, **p < .01, ***p < .001.

The second step of the structural equation model testing was to examine the significance of the model paths. As shown in Figure 3, model fitted well [χ² = 80.038, p = .036, χ²/df < 2, RMR = 0.016, NFI = 0.954, IFI = 0.988, TLI = 0.983, GFI = 0.977, CFI = 0.987, RMSEA = 0.027 (90% CI 0.007, 0.041), SRMR = 0.034]. All path coefficients of the model were significant (Table 3).

Table 3. Regression weights of structural paths (N = 497).

Paths	B	SE	CR	p value
Perceptions→Supervision	0.584	0.085	6.863	<.001
Perceptions→Risk-taking	−0.157	0.079	−1.998	.046
Attitudes→Supervision	0.285	0.077	3.682	<.001
Attitudes→Risk-taking	0.221	0.080	2.755	.006
Supervision→Risk-taking	−0.238	0.068	−3.484	<.001

B = Regression Coefficient; SE = Standard Error; CR = Critical Ratio.

Bias-corrected percentile Bootstrap method (5000 replicate samples) was used to test the stability of the mediation effects of the model. As Table 4 shows, the results of the direct effect (β = −0.158, 95%CI [−0.324, 0.041], p = .113) and the indirect effect (β = −0.139, 95%CI [−0.267, −0.062], p < .001) of parental risk perceptions on children’s risk-taking behaviors indicated that parental supervision completely mediated the relationship from parental risk perception to children’s risk-taking behavior. The first hypothesis (H1) was accepted.

Table 4. Direct and indirect effects of the Model (N = 497).

Effects	Paths	Standardized Effects	SE	95%CI			p
				Proportion	Upper	Lower
Direct	Perceptions → Risk-taking	−0.158	0.091	53.20%	−0.324	0.041	.113
Indirect	Perceptions → Supervision → Risk-taking	−0.139	0.050	46.80%	−0.267	−0.062	<.001
Total	Perceptions → Risk-taking	−0.297	0.078	100.00%	−0.45	−0.14	<.001
Direct	Attitudes → Risk-taking	0.221	0.080	76.47%	0.090	0.413	.002
Indirect	Attitudes → Supervision → Risk-taking	−0.068	0.030	23.53%	−0.162	−0.027	<.001
Total	Attitudes → Risk-taking	0.154	0.072	100.00%	0.028	0.318	.018

SE = Standard Error; CI = Confidence Interval.

The results of the direct effect (β = 0.221, 95%CI [0.090, 0.413], p = .002) and the indirect effect (β = −0.068, 95%CI [−0.162, −0.027], p < .001) of parental risk attitudes on children’s risk-taking behaviors indicated that parental supervision partially mediated the relationship from parental risk attitudes to children’s risk-taking behavior (as shown in Table 4) The second hypothesis (H2) was accepted.

Comparison of Cross-Gender Models

To test the third hypothesis H3, a multi-group analysis was used. Two models were developed, one of which allowed paths to be different across children’s genders (unconstrained structural paths), while the other restricted path coefficients to be equal for both genders (constrained structural paths). The results (Table 5) showed that there was no significant difference between the two models (∆χ² = 12.022, p = .799). Also, all fit indicators of both models met the criteria for fitness (Cheung & Rensvold, 2002). This indicated that there was no significant difference cross gender models, that is, the mediating role of parental supervision in this model remained the same between boys and girls. The third hypothesis (H3) was rejected.

Table 5. Comparison cross gender models (N = 497).

Model	χ^2	df	CFI	RMSEA	SRMR	AIC	ECVI
Unconstrained	164.073	118	0.974	0.028	0.045	292.073	0.590
Constrained	176.095	135	0.977	0.025	0.052	270.095	0.546
∆χ^2	12.022
p	0.799

Discussion

Allowing young children to access appropriately to different risk situations is a very important process of learning to explore the world (Huang et al., 2014). However, with the continuous increase of unintentional injuries to young children in China, safety is highly concerned, and parents may sacrifice children’s free play and risk-taking opportunities (J. Liu, 2020). The research to understand the relationships between Chinese parenting mechanism of children’s risk-taking behaviors was still rare. The current study first established a model to examine the mediating relationships between three parenting factors and young children’s risk-taking behavior in the context of China.

The results confirmed the direct effects of parental supervision, risk perceptions, and risk attitudes on children’s risk-taking behaviors in China. In addition, this study revealed a mediating effect of parental supervision between parental risk perceptions and children’s risk-taking behaviors, and between parental risk attitudes and children’s risk-taking behaviors, indicating that parents’ risk perceptions and risk attitudes may influence children’s risk-taking behaviors through parental supervision. The findings provide an empirical basis for exploring and intervening risk-taking behaviors in early childhood. This section will discuss these findings in detail.

Parental Supervision Mediated Between Parental Risk Perceptions and Children’s Risk-Taking

This study found that parental risk perceptions had negative effects on children’s risk-taking behaviors, which had been noted previously (W. Liu et al., 2022; B. A. Morrongiello & McArthur, 2018; Sneed et al., 2015). In psychosocial and behavioral models of children, parents’ perceptions become key variables, as how parents think with regard to risk and safety can be essential to predict risk-taking and injury for young children (Lam, 2001). From a subjective perspective, parental perceptions determine the level of risk for different child behaviors or activities. If parents perceive a heightened risk of injury for the child and feel a strong need for supervision, they are likely to implement more safety measures and engage in increased communication with the child. Consequently, the child is less inclined to partake in risk-taking behaviors (Napper et al., 2015, Sneed et al., 2015). For example, parents’ perceptions of high risk for snacking directly decreased children’s proximity to and consumption of unsafe snacks (J. Li & Luo, 2016). As McBride and Mills (1993) found, children of parents who had reduced risk perceptions behaved more playfully and showed greater physical arousal and activity levels than other children.

As the first hypothesis, parental supervision mediated the relationship between parental risk perception and children’s risk-taking behaviors. When parents perceived their child’s behavior as risky, with a high likelihood of injury and a need for close supervision, they were likely to increase their supervision level, which in turn reduced the child’s engagement in risky activities. This finding confirmed the mediating role of parental supervision in the model of young children’s risk-taking behaviors (Sandseter, 2010). During parenting, parental risk perception directly associated with their subsequent supervisory behaviors. As parents promote their perceptions of risks, they have more engagement with supervision in children’s activities. Consequently, children’s propensity to risky activities is diminished. Children’s actions generally occur under the surveillance of parents, constraining where children are allowed to go and what they are allowed to do (B. A. Morrongiello & Dawber, 2010). Many of the decisions in risky situations involving young children are directly made by parents or are associated with parents’ perceptions of specific risks and their decisions to act upon children’s risk-taking in daily life (B. A. Morrongiello et al., 2015). How supervising adults perceive and react to the children’s risk-taking would thus function as a kind of filter for the actions that the young children were able to actualize (Truong et al., 2022).

Parental Supervision Mediated Between Parental Risk Attitudes and Children’s Risk-Taking

This study also found a positive effect that parental risk attitudes had on children’s risk-taking behaviors. When parents viewed experiencing risks and injuries as opportunities for their children to learn and get strong physically and mentally, their children were more likely to present a greater inclination toward risk-taking. Consistent with previous research, children of parents who had positive risk attitudes tended to engage in more risky activities (Lu et al., 2023). Parental attitudes on risk-taking can also affect children’s own attitudes of risks through daily communication, consequently directly shaping inclination for engaging in risk-taking behaviors of the children (B. A. Morrongiello & Dawber, 2000). Whether parents encourage or discourage risk-taking behaviors in young children, and in what ways they respond to them, is a manifestation of supervisory behavior governed by their risk attitudes, which can be either negatively limiting or positively contributing to a young child’s propensity for risk-taking (Y. Hu, 2015).

As the second hypothesis, the findings demonstrated the mediating role of parental supervision between parental risk attitudes and children’s risk-taking behaviors. This result supported the consistency between parental risk attitudes and supervisory behavior. It is just that this consistency goes in the opposite direction of what many of the previous studies suggested. For example, Yue and Ren (2021) found a negative association between parental risk attitudes and their supervision. If parents felt that taking risks and getting hurt will do more good than harm to their child’s development, parents are more likely to reduce their supervision and thus elevate children’s risk-taking behaviors (Karaca, 2020; K. Li & Wang, 2020). However, our study indicated that as parents adopted a more positive attitude toward risk, they tended to enhance their supervisory behavior, leading to a decrease in risk-taking behaviors in young children. Simultaneously, due to the direct positive effect, parents’ positive attitude stimulated the increase of risk-taking behaviors in children. This suggested that the positive effect of parental risk attitudes on children’s risk-taking behaviors was weakened after introducing parental supervision as a mediating factor. This finding reflects a dilemma that has emerged in Chinese caregivers in recent years (Hao, 2020). More and more parents are realizing that this is a “risk society” (Mythen & Ulrich, 2004) where taking risks is not only unavoidable in childhood, but also has positive contributing values to early childhood health and development. Meanwhile, parents may be subject to societal pressure, that is, they wanted others to see their child healthy and safe and to be seen as “good” parents (Hao, 2020). As stated by Niehues et al. (2016), acknowledging the importance of embracing risk and developing a more favorable attitude toward it, parents simultaneously prioritize the safety and well-being of their children. This phenomenon is a response to early childhood risk-taking as a “double-edged sword.” In order to mitigate the potentially harmful consequences of risk-taking behaviors, parents have increased their supervisory efforts to ensure a safe environment for their children. However, at the same time, in order for children to benefit from the positive impacts of such adventurous activities, parents maintain a relatively positive attitude, aspiring for their children to explore and learn physically and allowing them to derive enjoyment from risk-taking experiences (Gong et al., 2021; Y. Hu, 2015).

Gender Invariance in the Relationships of Young Children’s Risk-Taking

As the third hypothesis was rejected, the findings revealed that the mediation model did not vary across genders of young children. As this study was the first to examine the mediating role of parental supervision across genders of children, there had been no empirical evidence to directly echo this finding. However, some previous studies could provide some indirect supporting evidence. Parents’ perceived high level of concerns make it nearly impossible to allow their children to play in nature without supervision. Children’s gender was found to be insignificant on neither the high risk perceptions nor the intimate supervision in the context of natural play (Truong et al., 2022). However, this study did not examine the relation between parental risk perceptions and supervision, nor did it test the influence of the children’s gender on the relation. In another study (B. A. Morrongiello & Dawber, 2010), ANOVA was employed to investigate gender disparities in parental attitudes and supervision practices. Findings revealed that variations in parental supervisory responses to risk-taking behaviors were aligned with differences in their attitudes toward risk-taking across genders. This alignment indicated that parental supervision was adjusted according to changes in attitudes toward risk-taking, regardless of child gender. In addition, the negative effect of parental supervision on children’s risk-taking did not vary by children’s gender. Wang’s study (Wang, 2016) found that the interaction effect of parental presence in supervision and child gender on young children’s propensity to engage in risky activities was not significant. For both boys and girls, tight supervision reduced the occurrence of risk-taking behaviors compared to loosen supervision. With changing perceptions of traditional gender roles and an emphasis on gender equality and individual development in China, parents of girls aspired for their daughters to become confident, independent, and resilient (X. Yang, 2018). Similarly to their counterparts raising boys, parents of girls acknowledged the significance of embracing risk-taking in fostering physical strength and the ability to deal with life risks. With similar positive attitudes toward taking risks, parents are increasingly aligning their approaches to nurturing both girls and boys, which include risk assessment and ensuring safety (Hao, 2020).

The Current State of Risk-Taking and Parenting Factors in Chinese Preschoolers

This study indicated that Chinese preschoolers’ risk-taking behaviors were generally at a moderate to low level, and parental risk perceptions and parental supervision were at moderately high level. Consistently, a previous study (Hao, 2020) showed that Chinese parents had higher levels of parental risk perceptions and parental supervision, and were more likely to have young children with lower levels of risk-taking behaviors compared to American counterparts. This difference is influenced by family parenting and cultural values. A primary goal of American culture is to assist their child’ independence (Tobin et al., 2009). American parenting supports children’s individual activities with this goal, even if they face some risks. They were more willing to believe that children should risks by themselves and thus need loose supervision (Hao, 2020). Correspondingly, American children are found to be more physically active and adventurous (Super & Harkness, 2010). Different from American cultures, Chinese culture values obedience so that Chinese parenting is more likely to direct their child’s attention and closely instruct their children’s activities (Super & Harkness, 2010; Tobin et al., 2009). Chinese children are found to be more physically obedient and less risk-taking than their counterparts (Ahadi et al., 1993).

This study also suggested that Chinese parents still had positive attitudes toward risk-taking behaviors in early childhood, even though they concerned about young children’s risk-taking behaviors (e.g., physical pain and injury). The present study showed that Chinese parental attitudes were at moderately high level. This finding was consistent with previous evidence. X. Yang (2018) noted that Chinese parents affirmed the importance of risky play for young children’s physical and mental development, and believed that risk-taking activities could promote a variety of developmental aspects (e.g., physical energy conservation, personality, risk perception, and social development). Compared to wealthier countries (e.g., the United States and Germany), individuals from fast-developing countries (e.g., China and Thailand) tend to be less risk averse. This distinction can be attributed to the prevalent individualistic culture in many western developed countries, while East Asian countries are characterized as collectivistic societies (Rieger et al., 2014). The “cushion hypothesis,” introduced by Hsee and Weber (1998), sheds light on this phenomenon: the robust social networks in collectivistic societies act as a protective “cushion” against potential financial crises, consequently fostering a greater propensity for risk-taking behavior.

Implications and Limitations

The current study to identify factors and their pathways to childhood risk-taking in China has important implications. This study constructed an operational model with measurable factors built upon Sandseter’s conceptual relations of young children’s risk-taking behaviors (2010). Researchers are encouraged to use this model as a base to begin with future studies. Moreover, this study went beyond individuals’ immediate environments to recognize the importance of the parenting environment in children’s behaviors with risk. Risk-taking was not objective or isolated.

Parents’ early concerns and restrictions on children’s behavior deprived children of the opportunities to learn to manage risk and had the potential for putting their children at increased risk once children gain more independence (Brussoni & Olsen, 2012; Valentine, 1997). This study offered possible guidance for design and implementation of parent education programs for keeping children safe while letting them take and manage risks. On the one hand, harmful risk-taking behaviors may be accompanied by serious and irreparable damages to young children, such as playing on the road and unsupervised swim (Brussoni et al., 2012). Such behaviors could be reduced by weakening parental risk attitudes so that parents are more aware that such behaviors cause more harm than value to young children, as well as by enhancing parents’ perceptions about risks and parental supervision. For example, an intervention program (B. A. Morrongiello et al., 2014) to prevent child drowning focused to make parents more risk aware about drowning hazards and foster more realistic appraisals of their children’ capability to save themselves in water, which elicited greater active parental supervision and reduced risk of child downing. On the other hand, there are also tons of beneficial risk-taking activities children may encounter in daily life that provide developmental rewards that far outweigh the harms, such as climbing on monkey bars and rough-and-tumble play (Brussoni et al., 2012). Education programs could be designed to (1) facilitate parents’ risk attitudes through helping parents see the beneficial outcomes of letting children encounter risks and challenges within a relatively safe setting (e.g., joyful experience and physical exercise); and (2) to reduce their risk perceptions by helping them understand the positive role of minor injuries in child development (e.g., learn to be resilient and to deal with failure), thereby decreasing their level of supervision.

As a first attempt of its kind, this study provided a start in a larger project moving toward two goals: (1) to further understand the complexity of Chinese young children’s risk-taking behavior, including the mediating role of parental supervision and other possible factors; and (2) to devise programs for parents and teachers use with Chinese preschoolers. Attaining these two goals may offer a window through which we can guide preschoolers to engage in adaptive risk-taking behaviors and protect them from maladaptive risk-taking behaviors, that is, keeping children safe while they enjoy the freedom to explore. More research is needed to empirically explore and validate specific intervention strategies.

This study, however, has some limitations. First, this study was conducted in the central area of China. It would be most appropriate to interpret the findings within the areas. Future research should recruit larger samples from different areas to overcome this limitation. Second, this study relied heavily on reports from parents. Although the data collected were adequate to achieve the main purpose of this study, it would be better to also create objective measures by including real-life observation scales, particularly for parental supervision and children’s risk-taking behaviors. Three, it was not a longitudinal study; thus, we could not speculate the potential effects of the parental factors on young children’s risk-taking behaviors from a development perspective. Fourth, this study did not include any control variables in the model. Moving forward, we will further investigate and eliminate the influence of demographic variables (e.g., child age, ethnicity, family socioeconomic status and residence) to this model in order to gain a clearer understanding of the relationships between the three parenting factors and young children’s risk-taking.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research is funded by Zhejiang Provincial Philosophy and Social Science Planning Project [No. 22NDQN224YB].