METHOD

Participants were 48 parent-child dyads (46% female children, 54% male children; 71% mothers, 29% fathers) randomly selected from seven local parks that were strategically sampled from those parks available in order to ensure considerable variation in the education and socioeconomic status of participants. For education level: 55% completed some or graduated high school, 24% completed some or graduated university, 7% completed some graduate studies, and the remaining chose not to answer the question. For annual family income: 2% reported less than $20 000, 8% reported between $20 000 and $39 999, 15% reported between $40 000 and $59 999, 29% reported between $60 000 and $79 999; 10% reported incomes over $80 000, and the remainder chose not to answer the question. The only constraint on selecting a dyad was that the child had to appear to be between 2 and 5 years of age (n  =  30 children were 2–3 years, n  =  18 children were 4–5 years) and the parent had to be the sole supervisor.

Two types of measures were taken, observational measures and questionnaire measures. Based on a pilot study, a 20 minute observation period was selected. Parent-child dyads were unobtrusively observed, with measures coded after every two minute period (a vibrating alarm watch tracked two minute periods). Subsequently, parents were approached, asked to confirm the age of their child, and asked to complete questionnaires, with $10 paid for participating. Refusal rate was 15%.

Observational measures

Parent supervision comprised five behaviors that were coded at the end of each two minute period: visual supervision (3  =  watching child continuously, 2  =  intermittently, 1  =  not at all), auditory supervision (3  =  able to hear the child continuously, 2  =  intermittently, 1  =  not at all), physical proximity (5  =  constant physical contact with child, 4  =  intermittent physical contact, 3  =  constantly within arm’s length reach, 2  =  intermittently within reach, 1  =  beyond reach of the child), parent distraction (5  =  parent is completely focused on child, 4  =  parent is mostly focused on child with occasional distractions or interruptions, 3  =  equal time is spent with distractions as attending to the child, 2  =  more time on distractions than attending to child, 1  =  all the parent’s time is spent on distraction activities), and parent engagement with child (4  =  all the parent’s time was spent actively playing with child, 3  =  occasional participant in child’s activities, 2  =  parent was not participating in the child’s activities but was engaged nonetheless, for example, smiled at child’s activities, 1  =  completely uninvolved and inattentive to the child’s play). Collapsing scores across the 10 two minute periods provided a total supervision score for each behavior.

Children’s risk taking was measured by keeping a running total of the number of injury risk activities (that is, engaging in an activity that could lead to injury and/or demonstrated inappropriate use of playground equipment in a way that could lead to injury; for example, jumping off a moving swing; standing as one came down the slide) during each two minute period. Collapsing scores across the 10 two minute periods provided a total risk taking score.

Reliability estimates were computed for each behavior and exceeded 0.80 (κ statistic). Coders completed their observations independently (that is, from different locations within the park). Only data from the designated “primary” coder (randomly determined) were included in the analyses.

Questionnaire measures

Questionnaires were distributed to the parent in random order. The Family Information Questionnaire, which was developed by the first author, provided information about family income level and education level. The Injury History Questionnaire, which was developed by the first author, provided several different measures of the child’s injury history, including medically attended injuries and an index of the frequency with which the child had sustained 17 different types of non-minor injuries (for example, fall down stairs or from a substantial height, crushing injury) since birth; non-minor injury was defined as tissue damage (for example, burn, bump, cut) that lasted for longer than one day. Prior research indicates that mothers accurately recollect child injury information for more serious injuries.²⁹ Because of the young ages of the children we reasoned there might be too few medically attended injuries, thereby constraining our ability to identify significant relations between injuries and our measures. Hence, we also included a measure of non-minor injuries.

The Injury Behavior Checklist (IBC) provides a standardized measure of the child’s typical level of risk taking.³⁰ The parent is asked to rate the occurrence of 24 risky activities (for example, jumps down stairs). Higher scores indicate greater frequency of risk taking behaviors. The Parent Supervision Attributes Profile Questionnaire (PSAPQ), developed by the authors, is an evolving measure that seeks to identify aspects of supervision that relate to child injury risk. In the current version of the questionnaire there are four subscales that show adequate internal consistency, including: protectiveness (21 items, α  =  0.70; for example, I make my child keep away from anything that could be dangerous), worry about safety (10 items, α  =  0.69; for example, I spend much of my time worrying that he/she will get hurt), vigilance in supervision (11 items, α  =  0.68; for example, I keep an eye on my child’s face to see if he/she needs my help), and confidence in his/her ability to keep his/her son/daughter safe (eight items, α  =  0.65; for example, I feel confident I know what my child can do). In addition, parents completed the Fate subscale from the Parent Health Locus Of Control (PHLOC) which provides an index of the extent to which parents believe that the health and injury status of their child is predominantly a matter of luck or fate (for example, Whether my child avoids injury is mostly a matter of luck).³¹ Parents also completed two subscales, neuroticism and conscientiousness, from a standardized personality measure, the Big Five Inventory (BFI).³²

RESULTS

Table 1 shows the means and standard deviation scores obtained for all measures. Tables 2 and 3 show the intercorrelations between different measures.

As shown in table 2, visual, auditory, and physical supervision measures were highly intercorrelated. Parents seemed to have certain levels of supervision with which they were comfortable, and this was reflected in various indices of supervision behavior. Additionally, level of parental engagement and distraction were correlated with auditory supervision, visual supervision, and physical proximity. Hence, specific indices of different types of supervision behaviors were reflected also in two general measures of parental attention to the child.

Interestingly, proximity to the child (physical supervision) had the most positive relation to children’s risk taking behavior during observations. As can be seen in table 2, greater physical proximity to the child was associated with less risk taking during the observations. Visual and auditory supervision, which typically occur at greater distance from the child, were not associated with decreases in children’s risk taking during observations. Thus, although supervision behaviors were highly intercorrelated and tended to co-occur, physical proximity was the only supervision behavior associated with decreased risk taking in the immediate situation. Interestingly, none of the supervision behaviors related directly to either child injury history measure (see table 2). However, parent engagement with the child during our observations was negatively correlated with children’s non-minor injuries. Parents who showed more engagement with their child, had children who had a history of having experienced fewer non-minor injuries.

As shown in table 2, there was no correlation between the observational measure of children’s risk taking and the questionnaire measure of risk taking (IBC). This is not surprising given that the IBC taps children’s risk taking in a broad range of situations, whereas our observation of risk taking was limited to a single environment, the playground. As expected, however, children who usually engage in risky activities also had a history of having experienced more injuries: IBC scores and history of non-minor injuries were highly positively correlated (r  =  0.71, p<0.001).

Table 3 provides correlations showing how questionnaire measures of parent attributes related to parent supervision, children’s risk taking, and children’s injury history; a composite measure of supervision was used because individual behaviors were so highly intercorrelated. As can be seen, neuroticism did not relate to parent supervision, children’s risk taking or their injury risk. However, parents high on conscientiousness had children who usually engaged in less risk taking (IBC) and who had a history of fewer non-minor injuries.

Several subscales of the PSAPQ related to supervision, children’s risk taking, and child injury history (see table 3). Specifically, parents who scored high either in protectiveness or worry about their child’s safety had children who had a history of fewer non-minor injuries. That these two subscales did not relate to supervision behavior may indicate that parents use non-supervision strategies to protect their child and act on their worry about their child’s safety. In contrast, subscales that tapped vigilance in supervision and confidence in a parent’s ability to keep their child safe both related directly to supervision behaviors. Parents who scored high in vigilance showed closer supervision during our observations, and they also had children with a history of fewer non-minor injuries. Parents who had a high degree of confidence in their ability to keep their child safe showed closer supervision during our observations, and also had children who typically engaged in less risk taking and experienced fewer injuries. Overall, the PSAPQ demonstrated good convergent validity with related constructs in this initial test, with differential subscales relating to parent supervision, children’s risk taking, and child injury risk.

Additional measures that tapped attributes of parents not in the PSAPQ also provided important insights into children’s injury risk. Parents who believed they had little control over their child’s health and injury status showed decreased supervision of their children and had children with a history of more non-minor injuries (see last column in table 3). Hence, parents’ beliefs about the extent to which their child’s health status is attributable to luck had implications for both how they supervised and the number of injuries their child experienced.

DISCUSSION

This study addresses an important methodological issue regarding measurement approaches in studying how parents’ attributes and behaviors, including supervision, relate to child injury risk. The findings from this study reveal a variety of parent attributes that can be measured reliably by questionnaires and that relate to parent supervision, children’s risk taking, and/or children’s injury histories.

Measures of parent personality attributes revealed the relevance of some attributes, and not others, for understanding child injury risk. Although parent neuroticism previously was found to be associated with increased risk of child injury, this was not the case in this sample.²⁶ Parent conscientiousness, however, related to child injury risk. Parents high on conscientiousness had children who usually engaged in less risk taking and had a history of fewer injuries. Of course, what remains to be determined is whether this decreased risk taking is actually attributable to parenting practices of conscientious parents and, if it is, what it is that conscientious parents actually do in the course of their interactions with their children that result in decreased risk taking and child injury risk. Conscientiousness was not associated with increased supervision behaviors per se (see table 3). Possibly, these parents are somehow very effective in teaching their children about rules restricting injury risk behaviors or in structuring the environment to minimize risk taking, resulting in little need for close supervision in order to moderate children’s risk behaviors. Suffice it to say, while it is unclear why children of conscientious parents engage in less risk taking and experience fewer injuries, the present findings indicate that parent conscientiousness serves a protective function for child injury, and this attribute can be reliably measured with a questionnaire.

Previous studies revealed that many parents attribute injuries to bad luck and do not believe in the preventability of child injuries.^27,28 The present study extends these findings to show that parents who believed that their child’s health status was attributable largely to luck showed decreased supervision and had children who had experienced more injuries. The fact that such beliefs can be easily measured with a questionnaire provides the means to further explore how parents’ beliefs about fate relate to their injury prevention and safety promotion practices.

Finally, the results from this preliminary test of the newly developed PSAPQ suggest that this questionnaire holds promise for the future study of child injury risk. A number of subscales that tapped different parent attributes showed adequate internal consistency and related to parent supervision, children’s risk taking, and/or children’s injury history. Parents scoring high on either protectiveness, worry about safety, or vigilance in supervision, had children with a history of decreased injuries. Moreover, as one would expect, vigilance about safety and confidence in one’s ability to keep one’s child safe were both directly associated with the level of supervision shown by parents at the playground. In ongoing research, we are examining relations between self reports by parents on the PSAPQ and actual home supervision practices. Supervision behaviors we observed on the playground did not relate to children’s injury history. However, this may reflect the fact that most injuries to young children happen in and around the home, rather than on playgrounds. Measuring home supervision, therefore, will likely have greater relevance for indexing child injury risk at these young ages.