Comparison of the computerized ACTIVITYGRAM instrument and the previous day physical activity recall for assessing...

This study compared the validity of children's physical activity recall through a computerized assessment (ACTIVITYGRAM) and through a popular paper and pencil assessment (Previous Day Physical Activity Recall; PDPAR). A total of 147 sixth-grade children from two schools completed the two recall instruments. Data from both self-report instruments were obtained over the same 3 days, and these data were matched by time with data from a subsample of children who wore an accelerometry-based activity monitor (n = 28). To confirm previous validity research with the PDPAR, a larger sample (n = 128) with matched data on the PDPAR and the accelerometer were also used. Activity levels recorded from all three instruments were computed for three periods (afternoon: 3-6 pm, evening: 6-11 p.m. and afternoon/evening: 3-11 p.m.). Descriptive and correlational analyses were conducted across days and for the separate time periods to evaluate the two self-report instruments. There were no differences in mean activity bouts reported on the ACTIVITYGRAM and the PDPAR (p > .05). Correlations with the activity monitor were above .50 for both the PDPAR and the ACTIVITYGRAM, and the relationships were consistent across days and periods for both measures. The magnitude of the correlations varied widely in the two schools for the ACTIVITYGRAM comparisons (r = .80 vs. r = .20) but not for the PDPAR, suggesting differential administration or understanding of the ACTIVITYGRAM instrument. Direct correlations between the ACTIVITYGRAM and the PDPAR measures were high (r = .72), and classification agreement exceeded 80% for the school that provided adequate assistance and training on the ACTIVITYGRAM. Results were weaker in the school with less staff support. This study provides convergent and criterion evidence to support the validity of the computerized ACTIVITYGRAM assessment and suggests similar information as the paper and pencil PDPAR assessment in young children. The study also revealed no differences in PDPAR values or associations, when a simpler scoring procedure was used instead of the traditional more complex metabolic equivalent-based approach.