Thesis aim: This thesis aims to analyse and critically evaluate the Children Activity Scanning Tool (CAST), which measures children’s physical activity (PA) engagement levels in school playgrounds and similar environments, focusing on CAST reliability and validity using data from four health promotion projects in the Northern Rivers (Move it Groove it (MIGI) project), Illawarra, Central Coast and Dubbo areas in NSW, Australia.
The overall research question is whether physical activity engagement levels of large numbers of children can be validly and reliably measured in school playgrounds and similar environments using CAST.
Methodology: CAST was used to scan school playgrounds and similar environments, in which unstructured rest, play and other physical activities take place. Scanning was undertaken by a team of trained observers who concurrently scanned a pre-defined area of the playground and recorded the number of children engaged in assigned physical activity intensity categories.
Criterion validity was measured in three projects comparing the total number of children counted in separate physical activity categories to an independent count of the number of children in the playground using Pearson correlation coefficients. Data obtained during instrument development were used to determine criterion validity in two projects when compared to videotapes using Pearson correlation, Cronbach alphas and 95% CI around mean engagement levels.
Data obtained during development and training, as well as project field data were used for reliability testing. Cronbach alpha, Pearson correlation, factor analysis and 95% CI around mean engagement levels were used to measure inter-rater reliability.
Data from multi-level regression models obtained in three projects were used to calculate intra class correlation (ICC) and composite reliabilities of scan and break level physical activity engagement.
Results: Criterion validity measures comparing sums of categories counts to independent counts of all children yielded Pearson correlation coefficients of 0.97, 0.81 and 0.96 in the Illawarra, Central Coast and Dubbo projects respectively. Criterion validity measures comparing field and video scores yielded Pearson correlation coefficients of 0.47, 0.52 and 0.56 respectively in the Move it Groove it (MIGI) project, and in the Illawarra study Cronbach alphas of 0.98 and 0.85, and a Pearson correlation of 0.93. There were no significant differences between the mean numbers of children who were engaged in a physical activity level when 95% CI were used except for one category in the MIGI study.
Inter-rater reliability estimates in all projects were high or very high. There were no significant differences between the mean numbers of children engaged in a physical activity level when 95% CI were used. Almost 86% (12/14) of the standardised Cronbach alpha estimates were equal to or greater than 0.93. Principal component analysis values ranged from 76%-98%. ICCs obtained from field data yielded composite reliabilities of 0.77-0.94.
Conclusion: It was concluded that CAST was a valid and reliable instrument to measure physical activity engagement levels of children in school playgrounds.
Recommendations: The two or three categories versions CAST should be used as they produced the highest quality data when used to observe large groups in well-defined playgrounds. Further research is needed to validate other variables (eg equipment), validate a one observer CAST version, and test the feasibility of using children as observers.