Refinamento psicométrico da escala de uso das mídias sociais baseada na Teoria do Comportamento Planejado

Abstract

The use of social media (SM) has grown significantly, becoming an essential part of everyday life. Understanding the factors that influence the intention to use these platforms is crucial, given their impact on social interactions, self-image, and public perception. Applications such as Instagram, YouTube, X (formerly Twitter), and Facebook remain among the most widely used worldwide. In this context, a robust psychometric instrument to assess the motivation for SM use is essential. This dissertation aims to psychometrically refine the EMUMS, based on the Theory of Planned Behavior (TPB), which comprises Attitude (AT), Subjective Norms (SN), and Perceived Control (PC). TPB is widely used to predict intentions and behaviors, making it highly relevant for studying SM use, as it considers both individual aspects and social influences that shape user behavior. Previous studies have demonstrated TPB’s strong predictive power for this phenomenon. However, the original version of EMUMS presented limitations, necessitating its refinement. The study was conducted in two phases. Study 1 applied the Item Response Theory (IRT) to reduce the number of scale items. Exclusion criteria included RMSEA (S_χ²) > 0.05 and low discrimination, resulting in the selection of three items per dimension. The sample consisted of 250 participants (76.7% women, mean age = 24.3 years, SD = 5.56). The items with the best psychometric properties were AT-2 [a = 2.90; b = -0.75; RMSEA(S_X²) = 0.04] and PC-2 [a = 5.80; t4 = 0.57; RMSEA(S_X²) = 0.01]. Study 2 tested the factorial structure and predictive validity of the refined EMUMS-12 using Structural Equation Modeling (SEM) with the WLSMV method, which is appropriate for categorical data. The sample included 664 participants (76.9% women, mean age = 23 years, SD = 5.99). The model demonstrated good fit indices (X²/df = 3.83; CFI = 0.98; TLI = 0.97; RMSEA = 0.06; SRMR = 0.07), supporting its plausibility. Additionally, Intention to Use (IU) significantly predicted Engagement in Social Media (ESM) (B = 0.37; p < 0.01). The findings from Study 1 confirm that IRT effectively identified the best items for each TPB component, resulting in the refined EMUMS-12. In Study 2, the predictive validity of EMUMS-12 was confirmed, indicating that AT, SN, and PC are fundamental in predicting Intention to Use and Engagement in SM. Finally, the new version of EMUMS demonstrated strong psychometric properties; however, future studies should expand the sample to further validate its properties, given that most participants in Study 2 were students