Cross-sectional and longitudinal comparisons of self-reported and device-assessed physical activity and sedentary behaviour

Abstract

Objectives: To examine the longitudinal associations and differences between self-reported and device-assessed physical activity (PA) and sedentary behaviour (SB), using a multifaceted statistical approach.

Design: Longitudinal measurement burst.

Methods: In total, 52 university students (78% female) aged 18-38 years (mean=21.94±4.57 years) participated. The study consisted of three blocks of six days of measurement, during which participants wore an accelerometer on their wrist for the entire block, and self-reported their PA over the 6 days at the end of each block.

Results: Meaningful latent differences between methods were observed for moderate PA and SB across all three assessment periods, such that participants underreported the time spent in each activity. Bland-Altman plots revealed a positive mean difference for vigorous PA, with over-reporting increasing as mean levels increased. Negative mean differences were observed for all other intensities. Underreporting of moderate PA increased as the mean level increased, whereas for light PA and SB, underreporting decreased at high levels. Repeated measures correlations revealed a meaningful association for vigorous PA only, suggesting that as self-reported minutes increase so too do device-measured minutes.

Conclusions: We found evidence of cross-sectional and longitudinal differences and weak associations between self-reported and device-assessed PA and SB. Future work is needed to enhance the quality of self-reported methods to assess PA and SB (e.g., face and content validity), and consider improvements to the processing of device-based data.

Keywords: Accelerometers; Bias; Discrepancies; Exercise; Sedentary Behaviour.