Introduction. There is controversy regarding the structure of subjective well-being (SWB) and the possibility of calculating a SWB total score. Objective. To test and compare five models proposed for the description of SWB. Method. The study was implemented with a cross-sectional, ex-post-facto design using an incidental sampling method. The Positive and Negative Experience Scale and the Satisfaction with Life Scale were applied to a sample composed of 600 students of health sciences from two universities of Nuevo Leon, Mexico. Data were analyzed through a structural equation modeling, using Maximum Likelihood and Corrected-Bias Percentile methods. Results. The bifactor model comprising three specific factors, vis-à-vis the model composed of three correlated factors, had the best data fit (¿x2/Adf = 8.166 > 5, ¿NFI =. 018, ¿NNFI =. 015, and ¿CFI =. 016 >. 01), and all its fit indices were close; however, the specific factor related to positive affect had a poor contribution. Nevertheless, the model composed of three correlated factors had the greatest parsimony (PR =. 853, PNFI =. 804, PNNFI =. 813, PCFI =. 819, and PGFI =. 706) and its three factors showed convergent validity, discriminant validity, and internal consistency reliability. Discussion and conclusion. The two models with the best properties justify the use of a composite score of SWB based on the scores of positive affect, negative affect, and satisfaction with life, as well as scores for these three specific domains of content. From a psychometric perspective, the model composed of three correlated factors yielded the best result.
