Validity and Reliability of Modified Physical Activity Questionnaire for Children

Background/Purpose: Parent proxy Physical Activity (PA) reports are used to measure young children’s PA participation. However such instruments are subject to further evaluation because parents may not stay with their children for entire day and therefore may not be able to accurately report their children’s PA participation. This study aimed to examine the psychometric properties of a modified physical activity questionnaire for children (MPAQ-C) that was used by parents to report their children’s PA participation outside school time in Chinese young children. 

Method: A total of 595 parents (441 mothers & 134 fathers) of primary school students aged 6-9 year old in Hong Kong completed the MPAQ-C in the Spring of 2011. The original physical activity questionnaire for children (PAQ-C) consists of 9 items for assessing children physical activity in the past seven days. The PAQ-C was modified by deleting three items that were used to measure children’s PA at lunch time, at recess time, and during physical education classes because parents were not present and could not observe their children’s PA participation during these time periods. Model fit of the MPAQ-C was examined using confirmatory factor analysis (CFA) with maximum likelihood estimation in LISREL 9.1. A sequence of increasingly restrictive CFA models was conducted to examine the factorial invariance of the MPAQ-C across mothers and fathers.

Analysis/Results: The CFA results showed that a single factor model of the MPAQ-C fit the data well (χ² (8) = 26.24, p < .001; comparative fit index [CFI] = .987; Non-normed Fit Index [NNFI] = .976; root mean squared error of approximation [RMSEA] = .062 [90% confidence interval = .037 to .089]), with a composite reliability of.80 and items’ squared multiple correlation ranging .28-.48. In addition, the CFI had no change (ΔCFI = 0) and the chi-square difference tests were insignificant (χ² (6) = 4.12, p >.05) from the baseline to the increasingly restrictive models, indicating invariance in factor structure, factor loadings and error variance of the MPAQ-C across fathers and mothers. 

Conclusions: The MPAQ-C is suitable parent proxy PA measure for assessing younger children’s PA. The MPAQ-C of a single factor model has demonstrated construct validity and good reliability with factorial invariance across gender of parents in the current sample of Chinese young children.