Multidimensionality of leisure-time physical activity behaviour in adults with spinal cord injury: implications for measurement.

The Physical Activity Recall Assessment for People with SCI (PARA-SCI) has demonstrated the best test-retest reliability and validity of self-report measures of leisure-time physical activity (LTPA) for adults with SCI. However, the absence of internal consistency reliability data has been raised as a concern. Internal consistency is relevant only for unidimensional measures. In other populations, LTPA is known to be a multidimensional construct, evaluated with multidimensional measures. We tested for unidimensionality through secondary analysis of PARA-SCI data (n = 703). Cronbach's α was 0.227. Principal components analysis showed two components/dimensions ('Moderate and Heavy Intensity LTPA' and 'Mild Intensity LTPA') explained 73% of the variance. Binary logic crosstabulation produced no discernible patterns of behavioural interrelatedness across LTPA intensities. Together, these data demonstrate the multidimensionality of LTPA and the PARA-SCI is not unidimensional. Internal consistency should not be a criterion for evaluating LTPA questionnaires for use in studies of people with SCI.