This study is the first to test the reliability and convergent validity of the National Center for SRTS survey instruments at each grade level using the hand-raising protocols recommended by Center. Student test-retest reliability and parent-student convergent validity of student-report of school travel have been tested in smaller samples , with a written student survey form [14, 15], and with a small time gap between student test-retest . Forman et al  assessed the reliability of parental attitudes to walking to school, but the National Center's parent survey has not been evaluated.
Test-retest reliability for student-report of school travel mode using hand-raising was high, with 93% and 92% agreement and kappa statistics of 0.86 and 0.85 for trips to and from school, respectively. This agreement was slightly lower than that reported by Evenson et al , who found 96% - 100% agreement and kappa scores of 0.88 - 1.00 between student reports of travel modes to and from school, and Mendoza et al , who found 98% agreement and a kappa score of 0.97 between student reports of travel modes to school. Both of those studies used pen and paper to collect student-reports of school travel mode rather than hand-raising.
Parent-student convergent validity of school travel mode was high, with 87 - 88% agreement and kappa statistics of 0.77 - 0.78 for trips to and from school. Convergent validity was in line with that reported by Evenson et al  who reported 88 - 89% agreement and kappa statistics of 0.80 - 0.81, and Mendoza et al , who reported kappa statistics between 0.57 - 0.87 depending on how travel modes were categorized.
The use of hand-raising to collect school travel data displayed reasonable levels of reliability and convergent validity even in the youngest respondents. Kappa statistics for reliability and parent-student convergent validity for trips to or from school were similar across kindergarten through fifth grade, suggesting that young children are able to reliably report school travel mode.
Student test-retest reliability and convergent validity for the student travel tally were similar for trips to school and trips home. One might expect lower agreement for reports of the trip home, as students were asked to predict the afternoon trip on the first day, while on the second day students recalled a trip that had already happened. The similar percentage agreement and kappa for trips to and from school indicates that predicting trips home in advance of the trip was not a major source of disagreement. One might also expect more disagreement on trips to home after school if students stopped at intermediate destinations such as daycare or a friend's house, as reported by Evenson et al . While the patterns of disagreement were slightly different for trips to school and home from school (Tables 3, 4), it was difficult to determine whether travel to intermediate destinations was a source of disagreement.
Researchers observed some confusion among parents, students, and volunteer survey administrators about the definitions of family vehicle, carpool, and other. For example, a student reporting a trip in a family vehicle and a parent reporting a carpool trip was one of the most common disagreements for trips to or from school. This confusion has been noted by other authors . Calculating reliability and validity for five travel modes, with family vehicle and carpool combined, yields higher agreement and kappa statistics for both student test-retest and parent-student agreement (Table 2). Survey designers should consider combining family vehicle and carpool into one option.
Parent surveys showed substantial reliability for travel mode, travel time, education, and income questions. Test-retest reliability was unacceptably low for the subjective attitudinal questions, including grade level at which parents would allow children to walk or bicycle and barriers to walking and biking. Forman et al.  developed a similar parent survey instrument, in which they asked parents to rate the importance of seventeen different attitudinal factors on their decisions to allow their children to walk or bicycle to three types of destinations, including schools. Seven of their items roughly paralleled attitudinal questions on the SRTS parent survey, but had remarkably higher reliability (ICC ranging from 0.6 to 0.8). The difference in reliability might be explained by their use of a 5-point Likert scale, rather than the yes/no check boxes used in the SRTS parent survey. Survey designers should consider restructuring the questions on barriers to walking and biking to school with simplified questions and Likert-scale responses.
Limitations of this study include the relatively low proportion of walking and biking in the sample, variation in student survey administration methods by classroom, variation in parental contact method by school, and a relatively low response rate to parent surveys in the more socioeconomically disadvantaged school. Between 4 and 10% of students in the sample walked or bicycled to or from school in this study. If there are differences in reliability and validity by usual travel mode, then our results mainly reflect motorized travellers. Future research should assess variability by mode. Also, in-class tallies were administered by volunteers, and while all volunteers participated in a brief pre-survey training session, they may not have administered the surveys in exactly the same way. Also, volunteer survey administrators in some classrooms failed to collect data on return trips home (explaining the lower sample size for mode to home in both the 24-hour test-retest reliability assessment and the parent-student convergent validity assessment). Furthermore, some volunteers returned from their classrooms with anecdotes of students experiencing peer pressure to raise their hands inappropriately. However, this may be a more realistic test of what may actually happen in a classroom being surveyed.