Application of S-1 Bifactor Model to Evaluate the Structural Validity of TMMS-24

Analysis of the Structural Validity of the Reduced Version of Metacognitive Awareness of Reading Strategies Inventory

The application of metacognitive strategies is considered a basic skill of the student at any educational level. In the present study, we evaluate the reduced version of the Metacognitive Awareness of Reading Strategies Inventory (MARSI-R) in Spanish, a self-report instrument designed to measure the metacognitive awareness of students and their perception of the strategies that they use while they are reading school materials. MARSI-R is formed by three subscales: (a) global reading strategies (GRS), (b) problem-solving strategies, and (c) strategies to support reading. We conducted a confirmatory factor analysis (CFA) in a Spanish student sample (N = 570) and the results shown relative inadequate fit for the proposed theoretical three-factor model. More important, the three subscales presented a high level of inter-correlation, which raises the need to assess to what extent the construct should be considered as unidimensional. We conducted two additional CFA models: a unidimensional model and a bifactor S-1 model, and the results indicated the presence of a strong general factor related to the GRS subscale. These results have important implications, since they imply that it is more appropriate to use the total score of the instrument derived of the S-1 model instead of the scores derived from each subscale. The bifactor S-1 model has allowed us to develop a closer approximation between the psychometric model and the theoretical model.

Structural and Predictive Validity of the Spanish Short Version of Utrecht Work Engagement Scale in Educational Institutions Workers

In this paper, we evaluate the factorial validity of the Spanish short version of the Utrecht Work Engagement Scale (UWES-9) and assess its predictive validity with respect to self-assessed work performance. A total of 229 employees from educational institutions in Ecuador participated. Using a model comparison analysis, the unidimensional model exhibited an excellent goodness of fit, χ2 = 26.176 (24), p = .344; CFI =1.000; TLI = 1.000; RMSEA = .020; SRMR = .034; it was not improved by more complex models, Three-factor model: χ2 = 22.148 (21), p = .391; CFI =1.000; TLI = 1.000; RMSEA = .016; SRMR = .033. Two-factor model: χ2 = 26.080 (23), p = .297; CFI = 1.000; TLI = 1.000; RMSEA = .025; SRMR = .034). Therefore, it is justified as a unidimensional instrument of work engagement. However, upon analyzing the correlation patterns of the overall score and the work engagement dimensions in relation to the task performance, contextual performance, and counterproductive behaviors, we conclude that, while the unidimensional model exhibits a good fit, the three-factor theoretical approach is substantively superior in that it maintains differential predictive validity for each theoretical dimension.