Improving youth physical activity is an important public health goal for addressing the childhood obesity epidemic. In the US, physical activity declined across age groups among a nationally representative sample [1] and longitudinally among a large US cohort [2]. Since adolescent physical activity was associated with adult physical activity [3], engaging youth in lifelong physical activity behaviors should promote health throughout the lifespan.

Active commuting to school, i.e. walking or bicycling to and from school, shows promise for improving youth physical activity. Reports from Australia [4], Germany [5], Scotland [6], Canada [7], England [8] and the US [9, 10] have linked active commuting to school with higher levels of physical activity [11–13]. Active commuting to school is an integral part of the US Safe Routes to School (SRTS) Program [14], which promotes walking and bicycling to school safely by primary and middle school students. The concept for SRTS programs originated in Denmark in the 1970s and spread internationally [14]. The first US programs started in New York City and Florida in 1997. SRTS programs consist of up to four elements: education, encouragement, enforcement, and engineering to promote safe walking and bicycling to school. Since most US youth attend organized school, SRTS could have large public health impact [15]. Increasing walking and bicycling to school were objectives of US Healthy People 2010 [16]. Evaluations of walk to school programs in Scotland [17] and the US [18–21] have reported success in improving active commuting to school among students from a range of socioeconomic levels and settings.

SRTS programs are widely available and gaining in popularity internationally [22, 23]. The US SRTS Program received substantial federal funding - $612 million from 2005-2009; however, few validated measures exist for assessing outcomes of Safe Routes to Schools (SRTS) Programs [24], such as school travel and pedestrian safety. Previous school travel surveys were too lengthy for focused evaluations of active commuting to school [25], were developed among a rural sample only [26], had unclear validity for children who walked or biked to school [27], or have not been formally validated [23]. We therefore sought to validate a SRTS travel survey in an urban population (Objective 1), with the goal of providing an easy to use school travel survey for general program audits, but also rigorously validated and appropriate for researchers studying SRTS programs.

While most studies on SRTS have focused on students' mode of transport to school, an equally important outcome is pedestrian safety behavior. Increasing the numbers of children walking and bicycling to school also increases children's exposure to pedestrian hazards. SRTS programs emphasize safety through its four program elements. Since linking reductions in child pedestrian injuries to behavioral interventions would require very large sample sizes, assessing children's pedestrian safety behavior provides a direct assessment of what was learned. Videotaped observations of children's pedestrian safety behaviors has been described [28], but may not be feasible for some investigators due to cost. Previous investigators have described observational checklists for pedestrian safety behaviors assessed in the field among children in Israel [29], Scotland [30] and the US [31], although they only reported inter-rater reliability without comparison to an expert. We sought to fill this gap by assessing the reliability of a pedestrian safety behavior checklist designed for field use (Objective 2), which would be useful for SRTS program researchers.

The written SRTS travel survey was simple to administer and had high reliability and validity among elementary schoolchildren in the Houston-metro area. Child reported active commuting to school was 21%, a figure higher than previous national surveys, which reported prevalences of 13 to 14% [34, 35], likely due to differences in participant demographics and the wording of the questions used to assess active commuting to school. While other surveys have been developed or validated among middle-/high-income and rural or suburban samples of children [20, 23, 26, 27], this report is the first to validate a school travel survey in English and Spanish among a low income sample consisting of mainly Hispanic and non-Hispanic black children. This survey is suitable for use among both English and Spanish speaking urban elementary students, 9 years of age and older. Building upon the school travel survey used from the Marin County evaluation [20] and the survey from the National Center for SRTS [23], which asked students to raise their hand in response to their school travel mode, this written SRTS travel survey minimized potential peer influences on responses. This study also builds upon the study that lacked children who walked or biked to school in the validity sample [27]. In contrast to the CLASS instrument, which was more comprehensive and assessed 30 types of physical activities including active commuting to school [25], this survey focused solely on active commuting thereby providing a targeted, efficient, and valid, method to assess SRTS programs.

The pedestrian safety behavior checklist showed high overall agreement at the item level between raters compared to the PI (91%) and acceptable sensitivity (85%) and specificity (83%) in field testing. This high level of agreement is similar to a previous study that used videotapes of children crossing the road and coded for specific pedestrian safety behavior [28]. The checklist had moderate correlation compared to the PI and moderate reliability across raters. Checklist item #6, "kept looking," proved to be the most difficult to assess in the field and is reflected by having the lowest percent agreement (68%). This finding was not surprising, since item #6 required raters to observe whether or not the child was watching for traffic while crossing the street, which can be difficult due to raters having different vantage points during the observation. Thus, although this behavior is important, we recommend eliminating it from the checklist to improve reliability. Since reliability increased substantially from an ICC = 0.48 with one rater to an ICC = 0.65 with two raters, we recommend that two raters simultaneously use the pedestrian safety behavior checklist to maximize reliability. Reliability for this checklist was lower than a previous 4-item instrument (0.90) that assessed whether participants walked in the sidewalk/shoulder, stopped before entering the street, looked left-right-left before crossing, and kept looking for traffic while crossing [31]. This difference may be due to the greater number of items on our checklist (8-items), which potentially increased the difficulty of the instrument for the raters. Generalizability theory analysis revealed that the majority of the variance (78.1%) of the checklist responses was due to the items themselves, which is desirable and indicates that the checklist successfully distinguished among pedestrian behaviors. Overall, the checklist appears to be an acceptable instrument to assess pedestrian safety behavior in the field without the need for video recording equipment. This instrument builds on previously described observational checklists for pedestrian safety behaviors that reported high reliability [29–31] by also providing comparison to the PI, the developer of the checklist.

This study has several limitations. First, the SRTS travel survey assessed only school travel on one day. However, it can be administered on several different days to assess school travel patterns over longer periods of time. The travel survey also forced students to choose only one method of travel to school that best described their mode of transport. While this made the survey easier for the students to complete, some information may be lost for children who used multiple modes of transport for substantial parts of their commute (e.g. walk to the bus stop and then ride the bus). The travel survey did not assess travel from school to home, which may differ from the journey from home to school. Given the high convergent validity and reliability of the SRTS travel survey for assessing the journey from home to school (i.e. how did you get to school today?), it seems likely that children would also give valid and reliable results for the journey from school to home (e.g. how will you/did you get home today?)- this requires confirmation. Finally, convergent validity was determined compared to parent report, which is subjective and may be prone to error. However, no valid method for objectively assessing individual students' mode of travel to school exists and comparison to parent report has been previously used by others for validating school travel surveys [26, 27].

Limitations of the pedestrian safety behavior checklist include the moderate correlations and the lack of studies relating these behaviors to injury outcomes. However, these pedestrian behaviors were drawn from previous pedestrian injury prevention studies and were generally considered fundamental injury prevention behaviors for children to learn [28–31]. Further formative work may increase the checklist's reliability, such as improving training via standardized videotaped examples and focusing on problematic items (item #6 "kept looking" and item #4 "stopped at the curb"). If percent agreement, sensitivity, and specificity do not improve despite this formative work, those items will be removed. Another limitation was the use of the PI, a subjective rater, as a comparison. An objective criterion standard would have been ideal; however, no such standard exists for observations of pedestrians behaviors. While others have previously used videotaped observations, these were beyond the scope of this study, but may be useful for future validity studies.

The written SRTS travel survey is a valid and reliable instrument for assessing school travel outcomes among English and Spanish speaking 9-11 year old elementary school students. It should prove valuable to both researchers and program evaluators seeking an efficient and rigorous method for assessing travel outcomes associated with SRTS programs. The pedestrian safety behavior checklist provides a reasonable method for assessing children's pedestrian behavior in the field, and should be used simultaneously by at least two raters to maximize reliability. Future studies should seek to improve the checklist's reliability and examine the relationship between the pedestrian safety behaviors and childhood pedestrian injuries.