In this cross-sectional study, we examined the associations of parental concerns related to safety on walking to school among 4th grade students who lived within a 2-mile network buffer of selected elementary schools across the state of Texas. A series of single-factor regression analyses were conducted to investigate safety concerns (road safety and personal safety) across three spatial domains (home neighborhood, en-route to school, and near the school). These analyses showed that, in general, children’s walking to school depended on parental perceptions of the following factors related to road safety: sidewalks and safe road crossings in the neighborhood; sidewalks, speed and amount of traffic, and intersections along school route; and sidewalks, crossing guards, and availability of trees along streets near the school. In terms of personal safety, parents were concerned about general neighborhood safety, stray or dangerous animals, and availability of adults with whom their child can walk en-route.
Our findings expand upon prior studies that suggest that parental safety concerns are related to walking to school among children. For instance, parental perception of the presence of sidewalks was found to be associated with walking to school in all three spatial domains studied. Two prior studies using children’s perspectives of the neighborhood did not find a significant association between the presence of sidewalks and walking to school [42, 43]. Of three studies that used parent perceptions, two found a significant association [44, 45] while one did not . The parent’s perception of sidewalk availability may be more influential on children’s walking to school than the perception of the child. This may be particularly true for younger children; Trapp and colleagues studied children in grades 5-7 , while the students in the current study were in grade 4.
In the current study, we found more consistent associations between WTS and the road safety factors than the personal safety factors examined. The potential salience of road safety is highlighted when observed relationships with WTS are assessed in the home and school spatial domains. In the home neighborhood environment, three in four road safety items maintained significant relationships with WTS in the adjusted models, while one in four stayed significant for personal safety. A similar trend was observed for adjusted models in the school domain, with three in eight for traffic safety and zero in five for personal safety. This finding is in line with a nationally-representative study that found a greater proportion of parents felt that it was too dangerous for their 5-11 year old child to walk to school because of traffic than because of crime (37.0% vs. 14.2%) , as well as a prior review on attributes of the physical environment and children’s physical activity levels . In this review, parental concerns about road hazards (street crossings and traffic) were more consistently associated with children’s physical activity levels than were perceptions of safety from crime.
Our findings suggest that the en-route environment may be the most critical environment to parents for both traffic safety and personal safety. All but one of the 8 items that were assessed in the en-route environment maintained significant relationships with walking to school in the expected direction, i.e. more safety concern associated with less walking to school. Comparatively, 4 of 8 and 3 of 13 items remained significant in adjusted models at the home neighborhood and school environment respectively. Further, the largest measures of effect were seen in the en-route domain. These findings suggest that parents may weigh the safety of the specific route a child will travel over the safety of the neighborhood or school environment when deciding whether to allow their child to walk to school. This finding lends further support to the call for specificity when defining the spatial domain of a behavior of interest .
Our assessment of the relationships between the selected covariates and WTS confirmed previous findings in some cases, and offered some additional insights. We saw a negative relationship between indicators of socio-economic status and walking to school, as has been generally, but not consistently, noted in other studies. A 2009 systematic review of determinants of children’s active travel reported negative associations with children’s active travel in six of seven studies that considered household income, nine of twelve studies considering car ownership, and four of twelve that considered parental education . We also found that student perception of teacher support and parent perception of school support for active commuting had a positive association with students’ walking to school. A similar finding has been reported in at least one prior study . Considering the low prevalence of this perception among students (16.6%) and parents (26.2%) in this study, school policy may be a practical target for interventions. For instance, schools may consider adopting an official policy statement to support active commuting to school and making this statement of support known to all families and the larger community. Also, we saw negative associations between several measures of civic engagement (voted in an election, attended a school board meeting, and volunteered in child’s school) and WTS, significant at the p < 0.05 level, although these were not significant with the Bonferroni correction. Taken together, these results suggest that children from higher SES families and those who are civically-engaged may be less likely to walk to school than their counterparts. Any relationships between these variables are likely complex, but do suggest that social norms may be involved. Further work in this area may be warranted.
Several potential limitations can be noted. The cross-sectional design precludes causal inference, and our findings were based on self-reported information, which may lead to recall bias. Respondent burden might have played some role in the general response rate of the parents (31.6%), and possibly influenced the reliability of reported study variables. However, other researchers and governmental organizations rely on self-reported information for their analyses, and evidence of a systematic bias due to self-reporting of mode choice to school is largely absent in the literature. Importantly, given that perceived safety was the primary exposure of interest in the current set of analyses, the use of survey was therefore an appropriate means of measuring participants’ perceptions. Another issue that is related to the assessment of perceived safety concerns and WTS is the potential for a mismatch between perceptions of safety and “actual safety”. Others have reported differing findings on the concordance between environmental perceptions and objective measures –. Therefore, if safety perceptions do not correspond well to actual risk in the home-to-school journey, attempts to improve traffic or personal safety “on the ground” might not increase WTS. Essentially, it may be that it is the perceptions of risk that need changing as much, or even more, than the actual environment. This point is being highlighted in the current paper, as an important theme in this subfield. Nonetheless, an in-depth critique is beyond the scope of the current study.
Despite the acknowledged limitations, our findings have relevance to the behavioral medicine field in a variety of ways. First of all, the current study asked participants about specific safety concerns, rather than using general safety questions, which provides evidence that road safety may be more relevant than personal safety to parents, as far as walking to school is concerned. However, despite this more robust assessment, the full range of parental perceptions around safety for their child may not be fully captured. Future research would benefit from the use of qualitative data gathering in communities (e.g. focus group discussions and interviews) to improve the operationalization of safety concern constructs. Secondly, a major contribution to existing knowledge is the level of spatial specificity offered by T-COPPE data that previous studies have lacked. This study provides the ability to examine relevant safety concerns across different spatial domains (i.e., home neighborhood, en-route, and school environments) going beyond previous single domain studies. Consequently, we were able to examine the differential effects in the exposure-outcome relationships across these spatially-distinct domains.
There are other prominent aspects of the T-COPPE study. T-COPPE participants were selected from both urban and rural schools across Texas; therefore, our findings may be generally applicable to Texas 4th grade students and their parents. Notably, since the current analyses included participants that live within a 2-mile distance from their school, this inclusion criterion addressed potential rural-urban distance-based differences. The T-COPPE population was more diverse and low income than previously reported data, and our sample is fairly large when compared to other similar studies. The methods used for construct development, data sourcing, and analyses can be replicated in most, if not all, settings.