Adaptation of the Teachers’ Trust in Artificial Intelligence–based Educational Technologies Scale into Turkish: A validity and reliability study

This study aims to adapt a scale measuring teachers' trust in AI-based educational technology into Turkish and to examine its validity and reliability. Data were collected from 231 teachers across various subject areas, including biology, chemistry, geography, language, mathematics, physics, and science. Following the assessment of univariate and multivariate normality, Confirmatory Factor Analysis (CFA) and reliability analyses were conducted. Mardia's multivariate kurtosis test indicated a violation of multivariate normality. Maximum Likelihood (ML) estimation was retained for the CFA, while bias-corrected bootstrap confidence intervals and the Bollen–Stine bootstrap procedure were used to evaluate the robustness of parameter estimates and model fit under non-normality. The initial 24-item, eight-factor model was revised based on the CFA findings, resulting in a 21-item, six-factor structure. The revised model was supported by most fit indices (χ² = 340.179, df = 174, χ²/df = 1.955, CFI = 0.903, RMSEA = 0.064), although the TLI value (0.883) was below the conventional .90 threshold. The six subdimensions were Perceived Benefits of AI-based EdTech; Lack of Human Characteristics of AI-based EdTech; Anxieties Related to Using AI-based EdTech; Self-efficacy in Using AI-based EdTech; Preferred Means to Increase Trust in AI-based EdTech; and AI-based EdTech vs Human Advice/Recommendation. Reliability evidence was stronger for four of the six subdimensions, whereas F2 and particularly F3 showed comparatively weaker internal consistency and convergent validity evidence. Overall, the findings provide initial support for the six-factor Turkish adaptation while indicating that the weaker subdimensions and the revised factor structure require further examination in independent samples.