Teun Remmers

38 | Chapter 3 in elementary school children (8). This study showed good overall reliability (α= 0.87, n= 511) and acceptable internal consistency of the PACES (corrected item-total correlations (CITCs) ranged from 0.26 to 0.71). Confirmatory factor analyses confirmed findings of Motl et al. (2001); showing acceptable fit of the factor (8). The Dutch translated PACES, as used in the present study, was developed, pretested, and validated in a larger sample of 8 to 10 years old children in Maastricht, the Netherlands (30). In this validation study, two items had more than 10% missing values, because some children were not familiar with the used terminology (i.e. “exciting” and “getting a kick out of it”). These items were excluded from further analyses, resulting in 14 remaining items, in which a mean score was calculated. In the exploratory factor analysis most of the items loaded on one factor with acceptable factor loadings. Only one item (0.24) did not meet the threshold of 0.30. For the current study we decided to keep all 14 items. Correlations between the sum scores of the PACES and a validated measure of motivational regulation (e.g. BRePAC) was found to comply with the expected pattern (30). In addition, we evaluated the internal consistency of our own sample by Cronbach’s alpha and CITCs (30). Cronbach's alpha for the PACES scale in the present sample was 0.88. Following the guidelines proposed by Nunnally & Bernstein (1994) to define the minimum levels of item-scale correlations that are acceptable, we used cut-off points of 0.15 and 0.30 (31). Correlations above 0.30 are considered good and correlations below 0.15 are considered unreliable since they would indicate lack of homogeneity of the items within a scale. CITCs ranged from 0.49 to 0.75 for 13 items, indicating homogeneity of the items. One item (“that makes me sad”) fell below the value of 0.30, but still above the critical cut-off point of 0.15. We explicitly instructed the children to complete the questionnaire individually. As the PACES mean scores were negatively skewed, we dichotomized this variable using a median split (≤4.29 versus >4.30). PA behavior We measured PA behavior by accelerometry with the uniaxial ActiGraph 7164 and GT1M accelerometers (ActiGraph, Fort Walton Beach, FL). We instructed participants to wear the ActiGraph during daytime on the right hip for 7 days, and only remove the accelerometer during activities were water was involved, such as swimming. To investigate the influence of swimming, parents were asked to report this in a diary. Although we made all efforts to prevent occurrence of measurements in non-regular weeks, we excluded occasional measurement-days within holidays from further analyses. We initialized the ActiGraph to capture the counts every 15 seconds (epochs). We considered periods of 60 minutes containing no accelerometry counts as non-wearing time and we therefore subtracted these measurements from the registration time. We considered at least 400 minutes of registration time, for a minimum of 2 week days and one weekend day of measurement. We used the thresholds of Evenson (32) to distinguish between sedentary behavior (0-25 counts per epoch), PA at light (26-573 counts per epoch), moderate (574-1002 counts per epoch), and vigorous intensity (≥1003 counts per epoch). Subsequently, we combined the light, moderate and vigorous PA intensities in the ‘active’ category. The moderate and