Background: In large cohort studies of older children, self-report is the o
nly practical way to assess physical activity. Assessing usual activity ove
r the entire year is desirable, but children and adolescents may overestima
te activities with high seasonal variability. Use of questionnaires in whic
h individuals report each activity by season may improve accuracy.
Methods: A total of 6782 girls and 5110 boys, aged 9-14 years in 1996, comp
leted self-administered activity questionnaires in 1996 and in 1997. In 199
6, participants reported the average time spent in each of 17 activities du
ring the previous 12 months; in 1997, we also asked for the average time sp
ent in the previous year, but within each of the four seasons.
Results: Girls reported a median of 12.8 hours/week total activity in 1996
and 10.4 hours/week in 1997. For boys, the estimates were 15.5 hours/week a
nd 13.4 hours/week, respectively. Girls and boys within 1-year age strata (
e.g., comparison of 10-year olds in 1996 with 10-year olds in 1997) reporte
d an average of 3.7 and 3.1 fewer hours per week, respectively, on the 1997
seasonal format versus the 1996 annual format questionnaire. In longitudin
al analyses, the difference between the annual and the seasonal estimates w
as greater if participants did the activity in fewer seasons in 1997.
Conclusions: In comparison to an annual format questionnaire, a seasonal fo
rmat questionnaire may improve accuracy of self-report of physical activity
by reducing over-reporting of activities in which pre-adolescents and adol
escents engage in fewer seasons.