School and Student Level Correlates of Objectively Measured Physical Activity. A Multilevel Analysis with ISCOLE Portuguese Children

  • Presented on May 31, 2014

Background: Children spend most of their awake time at school, where they are engaged in active and sedentary activities. As such, it is relevant to understand the importance of school environmental characteristics (SEC) related to children’s physical activity (PA). The main aim of the International Study of Childhood, Obesity, Lifestyle, and the Environment (ISCOLE) is to determine the relationship between lifestyle characteristics and obesity in children, and to investigate the influence of behavioural settings and physical, social and policy environments on the observed relationships within and between countries.

Purpose: To identify school and student level correlates accounting for PA variability in Portuguese ISCOLE children.

Methods: PA and SEC were collected in 582 children (mean age 10.5 years; 335 girls, 247 boys) from 23 schools located in Porto, Portugal. Total PA (TPA) was assessed with the GT3X+ ActiGraph accelerometer, and a whole week of valid information was considered. Mean of counts/per minute from the seven days was used. Variables at the child level included body fat (BF), gender, percentage of attained predicted adult height, number of siblings, socioeconomic status (SES), and parental support for TPA. School characteristics included school size (number of students at school), percentage of students engaged in sports or PA clubs, number of breaks, and playground equipment. Multilevel modeling within the framework of HLM 6.0 software was used for all analysis.

Results: School-effects account for 5.3% of the total variance in TPA, and child-level correlates for 94.7%. Having siblings (β=14.568,t=2.783), and being a boy (β=95.701, t=9.248) are positively correlated with TPA; children with higher BF (β=-2.684, t=-4.278) and higher SES (β=-8.489, t=-4.900) had lower TPA levels. At the school level, only number of breaks was negatively associated with children’s TPA (β=-27.992, t=-2.276). School context explained 59% of the 5.3% of the variance in TPA; however, student-level correlates explained only 18% of the 94.7% variance fraction at the individual level in TPA.

Conclusion: Gender, BF, number of siblings, and SES are important correlates of variability in Portuguese children’s PA levels. However, school context explains a small fraction of the TPA variability among school-aged children.


  • Thayse Natacha Gomes 1
  • Daniel V. Santos 1
  • Fernanda K. dos Santos 1
  • Raquel N. Chaves 1
  • Michele C. Souza 1
  • Alessandra Borges 1
  • Sara Pereira 1
  • Peter T. Katzmarzyk, FACSM 2
  • José Maia 1


CIFI2D, Faculty of Sports, University of Porto, Porto, Portugal
Pennington Biomedical Research Center, Baton Rouge, LA

Presented at

ACSM 2014 Annual Meeting


, ,