Psychometric properties of the Exercise Dependence Scale –Revised (EDS-R) in Colombian gyms users

Abstract

The article presents a psychometric study of the Physical Exercise Dependence Scale-Revised (EDS-R) in Colombian gym users with the aim of analyzing its factorial structure, internal consistency, construct reliability and its convergent, discriminant and criterion validity. An instrumental research was carried out with the participation of 515 gym users (44.3 % men and 55.7 % women) aged between 18 and 58 years (Md = 25; RIQ = 21-30) from the city of Medellín. The Spanish version (Sicilia & González-Cutre, 2011) of the Physical Exercise Dependence Scale-Revised (EDS-R) (Downs, Hausenblas & Nigg, 2004) was analyzed. Confirmatory and exploratory factor analyses were performed. Results: The original structure of the EDS-R showed adequate internal consistency and factorial validity, confirmed through the goodness-of-fit indices for the seven-dimensional model. Cronbach's alpha was .917 and the Guttman two-half coefficient was .924. The goodness-of-fit indices for the original version were: PCMIN/DF = 2.291; RMR = .091; GFI = .917; AGFI = .886; CFI = .951; RMSEA = .056 (90% CI = .049 - .063); PCLOSE = .093. However, the values for mean variance extracted, maximum shared variance and maximum reliability revealed difficulties with construct reliability, convergent validity and discriminant validity. There were also no-good results when analyzing criterion validity. To analyze whether the data in the sample fit better to a different factor structure, a complementary exploratory factor analysis was performed that arrived at a five-factor model, with the same 21 items. However, the assessment of composite reliability, convergent validity and discriminant validity yielded better results than the seven-factor model.