With part-time administrative support, a walking school bus program was implemented in a low-income, ethnically diverse, urban, public elementary school. Intervention and control schools had statistically similar baseline proportions of students walking to school and these estimates were comparable to previously published national estimates [10
]. The WSB program was associated with significantly higher proportions of students who walked to school at short and long-term follow-up, as compared to the control schools. Comparing baseline to 12-month follow up results, the WSB intervention was associated with an increased number (n = 19) of students who walked to school in contrast to the controls which had a decline in students (n = 30) walking to school. A previous intervention study reported increases in the number of students walking to school, although that study lacked control schools for comparison [22
]. A short-term (10-week), quasi-experimental trial reported significant increases from baseline to immediate post-intervention in the mean distance walked to school (+555 meters, P < 0.001) and decreases in the mean distance traveled by car to school (-850.5 meters, P < 0.001) by intervention (Traveling Green) versus control students [21
]. Our study builds upon these studies by reporting long-term results specifically for a WSB program with a study design that included control schools. Taken together, WSB programs may improve the numbers of children walking to school, and may improve their physical activity, which are both objectives of Healthy People 2010 [3
]. While no long-term differences in school travel patterns were detected as a result of a previous randomized controlled trial [20
], that study's school travel coordinator(s) offered only 16 hours of expert assistance over one school year to each school, which may not provide sufficient time to develop, implement, and sustain a long-term school-wide travel plan. In comparison, the school coordinator for this study's WSB program spent about the same amount of time each week at the intervention school.
These results may underestimate the change in proportions of students who walked to school since they reflected days without scheduled walking school buses. Alternatively, the results suggest that WSB programs may not need to operate WSB routes every school day to have an impact on school travel patterns. Moreover, since data collection was relatively comprehensive for each school on assessment days, the changes in school travel patterns reflect the WSB program's school-wide impact, not just its impact on students who regularly used the WSB program.
This pilot study has a number of limitations. First, the evaluation used a non-randomized design. However, the control schools were comparable to the intervention school and served predominantly disadvantaged, minority populations from the Central District of Seattle, Washington, which should minimize selection threats to internal validity. Second, method of transport to school was assessed publicly in the classroom by self-report from elementary school students similar to a previous study [22
], which may limit validity. However, the baseline percentage of students who walked to school at both the intervention and control schools were consistent with previous national estimates [10
], which suggests that the transportation measurement method was comparable to previous methods. While few active transport studies have reported validity of students' self report for method of transport to school [4
], previous studies examining students' self reported school travel have demonstrated acceptable test-retest reliability (kappa coefficient 0.96) and validity (kappa coefficient = 0.80) compared to parental report in a sample of children aged 8–11 years [30
] and high concordance for test-retest reliability (97%) and validity (97.5%) compared to parental report in a sample of children aged 9–11 years [31
]. These studies suggest that child-assessed measures are reasonably valid and reliable. Third, method of transport to school was assessed by school teachers rather than research staff. Teachers were not specially trained nor informed of the study's goals or a priori
hypotheses. Using teachers for the evaluation allowed us to efficiently utilize an experienced group of professionals. It seems unlikely that teachers would knowingly or unknowingly bias results, given the small nature of the WSB program, the infrequent and brief transport assessments, and the multitude of academic teaching demands that they faced. Fourth, we did not have repeated measures on individual subjects nor socio-demographic data. Instead, we conducted cross sectional surveys with relatively high student participation (>78%) at each assessment. Given the cross-sectional assessments, we cannot determine if new students enrolled at the intervention school were already more likely to walk to school (or vice versa at the control schools) as a competing explanation for the results, but this appears unlikely. Fifth, the intervention occurred at a single urban, public elementary school, which limits external validity. Sixth, the study involved only three schools with a small sample size, especially when the unit of analysis was the school group level, which limits the ability to detect differences in the study's outcomes. Finally, the measurements were taken on only one day per assessment point, due to constraints inherent with pilot studies and natural experiments. Ideally, transport would be assessed over multiple days, to better estimate habitual school transport and account for day to day variation.
This pilot evaluation was designed to efficiently provide useful preliminary information from a natural experiment within the constraints of a limited budget and rapid timeline, to inform more methodologically rigorous studies. Studies are needed to examine the impact of WSB programs on child pedestrian safety behaviors. Studies ideally should be long-term group randomized controlled trials, which longitudinally assess individual students and their socio-demographics; use objective and validated measures for transport and physical activity; assess changes in psychosocial constructs related to physical activity; and consider the role of the built environment in moderating the effects of WSB programs.