Zafer Özen, 2024 Early Career Mini-Grant winner

The most common gifted identification methods in the U.S. involve a multiple-level process, with teacher and parental referral usually serving as the first level and intelligence or IQ tests as the second. However, both methods have been criticized for low reliability, validity, and underrepresentation.

Zafer Özen, a Purdue University doctoral candidate in educational studies with a concentration in gifted education, wants to develop a first-level gifted education tool using machine learning. His study will use data from the 2019 Trends in International Mathematics and Science Study (TIMSS) dataset of eighth-grade students.

Özen’s machine learning model will be trained on a subset of TIMSS data to identify patterns and features indicative of giftedness. The model will then be tested on a separate subset to evaluate its accuracy in identifying gifted students.

The development of a first-level gifted identification tool has the potential to provide more equitable and cost-efficient identification of gifted students. Özen hopes to improve the identification and support of gifted students from diverse backgrounds while contributing to the growing body of research on using machine learning in education.

With a successful model, he would then make an open-source tool schools can use to help with identification. Özen said, “Each school [could] potentially administer the publicly available TIMSS questionnaires and be able to insert their data into our website with pseudonyms and see which of their students should be considered for the second step of gifted education.”

The Foundation’s Early Career Mini-Grant was created to assist post-doctoral researchers or early career faculty for research related to intelligence, creativity, or gifted education.