Is Problematic Physical Activity (PPA) a Uni- or Multi-Dimensional Concept? A Bifactor Analysis of the Exercise Dependence Scale-Revised (EDS-R)

• Published in: Perceptual and motor skills Vol. 131; no. 1; pp. 106 - 134
• Main Authors: Kern, Laurence, Mattar, Lama, Kotbagi, Gayatri, Romo, Lucia, Morvan, Yannick
• Format: Journal Article
• Language: English
• Published: Los Angeles, CA SAGE Publications 01-02-2024; SAGE PUBLICATIONS, INC; Ammons Scientific
• Subjects: Cognitive science; Exercise; Factor Analysis, Statistical; Human health and pathology; Humanities and Social Sciences; Humans; Life Sciences; Neurons and Cognition; Physical fitness; Psychiatrics and mental health; Psychology; Psychology and behavior; Psychometrics - methods; Reproducibility of Results; Students; Surveys and Questionnaires; evaluation; bifactor analysis; exercise dependence; problematic physical activity
• ISSN: 0031-5125; 1558-688X
• DOI: 10.1177/00315125231218745

The study of exercise dependence, or as we prefer, problematic physical activity (PPA), faces both theoretical and methodological challenges. Different factorial solutions were obtained for the widely used Exercise Dependence Scale-Revised (EDS-R), leading us to question both its latent underlying construct and the interpretation of its factor solutions. Through confirmatory factor analysis (CFA) and the use of a bifactor model (BCFA), we assessed the dimensionality of a French version of the EDS-R. We recruited 745 students from Paris Nanterre University, but we removed 88 responses (16.8%) to retain only those who (a) engaged in a moderate to high levels of physical activity, according to the Global Physical Activity Questionnaire (GPAQ; N = 494; 9.5%), and (b) completed the EDS-R. We used a final sample of 435 students (58.4% of the total population) for a factorial analysis of the EDS-R. We conducted a three-step CFA in Mplus, producing three models: (a) unidimensional, (b) second-order, and (c) bi-factor (BCFA). The results of the BCFA indicated that most EDS-R items were better indicators of a general factor than their respective group factors, except for the second factor measuring a withdrawal construct. These results add to an ongoing debate in the field of behavioral addiction as to how to better conceptualize and measure exercise dependence or PPA. Withdrawal appears to have a special position in this debate, since our BCFA suggested that it is the only specific sub-dimension of the EDS-R scale.