Acceso usuarios
Julio César Huamani Cahua, Yuliana Ordoño Ticona, Miriam Yisela Viamonte Quispe, Nanyery Belén Vega Carpio, Estefany Ojeda Flores
Background: Negative attitudes toward Indigenous people remain a persistent expression of prejudice and discrimination in Peru, reinforcing symbolic hierarchies and structural inequities.
Objective: To evaluate the psychometric properties of the Attitudes Toward Indigenous People Scale among Peruvian university students, integrating evidence from content validity, psychometric network analysis, confirmatory factor analysis, measurement invariance, and differential item functioning.
Methods: An instrumental study was conducted with a non-probabilistic sample of 306 university students from Arequipa, Peru. Content validity was assessed using Aiken’s V coefficient with 90% confidence intervals. Item redundancy was examined through Weighted Topological Overlap, and dimensionality was explored using Exploratory Graph Analysis with EBICglasso, the Louvain algorithm, and 1000 bootstrap replications. Confirmatory factor analysis was performed using the WLSMV estimator to compare the original two-factor model with a network-derived unidimensional model. Measurement invariance by sex was tested. Differential item functioning was examined using ordinal regression and likelihood ratio tests with Benjamini–Hochberg correction.
Results: All items showed acceptable content validity and low overall redundancy. Exploratory Graph Analysis supported a stable unidimensional structure, with all items consistently assigned to a single community across bootstrap replications. The original two-factor model showed better global fit than the unidimensional model (CFI = .976; TLI = .968; SRMR = .032; RMSEA = .082), although the correlation between factors was very high (r = .929), indicating substantial overlap and limited discriminant validity. The unidimensional model also showed adequate fit in incremental and residual indices (CFI = .963; TLI = .952; SRMR = .036), strong internal consistency (ω = .923; α = .924), and acceptable convergent validity. Measurement invariance by sex was supported up to the strict level. Uniform differential item functioning was identified in two items, suggesting localized sex-related response bias, whereas no non-uniform DIF was observed.
Conclusion: The Attitudes Toward Indigenous People Scale showed adequate evidence of validity and reliability in Peruvian university students. Although the original two-factor model demonstrated superior statistical fit, the high interfactor correlation and network evidence support a parsimonious unidimensional interpretation.
