Psychometric Validation of the Hispanic Bilingual Gifted Screening Instrument: An Item Response Theory Approach

Demographics in the United States continue to shift with a rapidly growing Hispanic population. On the other hand, a mismatch still exists between Hispanic students? enrollment in gifted and talented (G/T) programs and general programs. The under-representation of Hispanic students in G/T programs has been attributed to a lack of proper instrument to identify gifted students who are linguistically and culturally diverse; insufficient preparation of teacher in the initial teacher referral phases; and ambiguous definitions of intelligence and giftedness.

In this study I investigated psychometric properties of the Hispanic Bilingual Gifted Screening Instrument (HBGSI) within an item response theory (IRT) framework. The HBGSI was developed with social-cultural context in mind and has been recommended for use in the first phase of teacher referral process. Participants in this study were Hispanic bilingual students in first to third grade, who participated in a large-scale longitudinal randomized study carried out in a Texas urban school district. The purpose of this study was to further validate HBGSI within the framework of IRT, exploring the factor structure and dimensionality of the instrument at the item level. I further tested the possibility of constructing an abbreviated version of HBGSI with fewer items for ease of administration, which would potentially lower the demand on the teacher?s time, enhance accessibility and facilitate increased usage of the instrument.

Results revealed a bifactor structure with a strong general factor corresponding to overall giftedness among Hispanic bilingual students, and five domain factors regarding social responsibility, academic achievement, creative performance, problem solving, and native language proficiency. The multidimensional bifactor IRT model provided information related to each item concerning its discriminating power, thresholds and information regarding the latent constructs. Best items were selected while preserving the integrity of the original HBGSI, and cutting the length to almost half. Thus an abbreviated version of HBGSI was feasible and the adaptation is presented. Overall, this study further validated that the HBGSI holds promise in screening potential Hispanic bilingual students in elementary grades.