Construct validity of a revised Physical Activity Scale and testing by cognitive interviewing

Aim: To validate the construct validity of a new version of a Physical Activity Scale (PAS 2) for measuring average weekly physical activity of sleep, work, and leisure time to determine whether a further criterion validation is justified. Methods: The validity of responses to the questionnaire was evaluated by cognitive interviewing in 16 Danish men and women aged 21-70 years. Construct validity was validated in 342 men and women aged 35-66 years by assessing agreement between 24-h MET-scores obtained from average weekly physical activity measured by PAS 2 and a 24-h Physical Activity Scale (PAS 1), previously found to overestimate physical activity. Results: Cognitive interviewing revealed few problems in the questions on physical activity in different domains. No problems regarding the structure of the questionnaire were identified. The agreement between PAS 1 and PAS 2 MET-scores was high among participants with a PAS 1 24-h MET-score <45 MET-hours. Among participants with a PAS 1 24-h MET-score >= 45 MET-hours, the weekly-based scale, PAS 2, systematically estimated fewer MET-hours compared to the 24-h based scale, PAS 1. The difference increased proportionally with the average of the two MET-scores. Conclusions: Few, small, lexical revisions were implemented into the new scale to improve the validity. As hypothesized, PAS 2 produced lower estimates of energy expenditure compared to PAS 1, indicating that the new scale may provide more valid measurements and that further validation against an objective criterion is justified.