Cross-cultural adaptation and initial psychometric evaluation of the fruit and vegetable self-efficacy scales among secondary school adolescents in Selangor, Malaysia.

Fruit and vegetable self-efficacy may influence adolescent dietary behavior, yet culturally adapted measures for Malaysian adolescents remain limited. This study aimed to cross-culturally adapt the 17-item Fruit Self-Efficacy (FSE) and Vegetable Self-Efficacy (VSE) scales into Malay and obtain initial evidence on their measurement properties. The study comprised two sequential phases. Phase 1 involved forward and backward translation, cognitive debriefing with five adolescents, linguistic equivalence assessment by six experts, content validity assessment by nine experts, qualitative and quantitative face validity assessment involving 15 unique adolescents, and pretesting with 37 adolescents. Phase 2 comprised an initial psychometric evaluation among 128 adolescents from four classes in one purposively selected public secondary school in Selangor. The adaptation process included age-appropriate refinements to the linguistic and contextual aspects of fruit and vegetable preparation, snacks, social occasions, serving terminology, and meal contexts. In particular, "pieces of fruit" was adapted to "servings of fruit" and supported by a pictorial serving guide, while "hot meal" was contextualized as "main meal." All source items met the prespecified content-validity criteria. Scale-level content-validity indices were 0.99 for FSE and 0.95 for VSE, while scale-level face-validity indices ranged from 0.95 to 0.97. One FSE item with an initial clarity index below the criterion was revised using concrete preparation examples and reassessed. In the separate psychometric sample, the prespecified one-factor FSE model showed acceptable fit (CFI = 0.974, TLI = 0.970, RMSEA = 0.069, SRMR = 0.061), although pronounced item-level ceiling effects were observed. The prespecified VSE model showed substantial misfit (RMSEA = 0.136, SRMR = 0.099), and its structural validity was not established. A same-sample post hoc specification with six correlated residuals remained hypothesis-generating because its RMSEA was 0.094. The multistage process produced linguistically and culturally adapted FSE and VSE scales that met the prespecified content and face-validity criteria among the expert and adolescent samples assessed. However, evidence supporting the one-factor structure was stronger for FSE than VSE. Further cognitive investigation and adequately powered, independent multisite evaluation across diverse ethnic and age groups are required before broader use.