Multilevel Modeling of Educational Assessment Data

My Education

• B.A. in Primary Mathematics Edu., Kocaeli , 2006
• CESL,UA, Tucson, 2007
• M.S and Ph.D., Research and Evaluation Methodology, UF, Gainesville, 2014
• Ph.D. Minor in Applied Statistics, UF, 2013

Job Talk Outline

• Early Career
• Current Research and Impact
• Contributions at the National Level
• Future Research Interests
• Why Boğaziçi?

Early Career - Foundations

• The Groundwork: Mathematics teacher in Sarıkamış; CESL student in Tucson.
• The Shift to Complexity: Demystifying complex educational data structures at UF (theory, data simulation, and application).
• Early Application: Academic trainee at Camber, Orlando.
• Initial Impact: Established a strong publication and presentation record focusing on methodology prior to 2015.

Early Career - Building the Core

• First 5 Years Post-Ph.D. (RTEÜ): Published 15 peer-reviewed articles.
  • Focused heavily on the theory and application of multilevel modeling across Special Education and Higher Education.
• Tenure (2019): Achieved through foundational theoretical work on latent mean decomposition in three-level data.
• 2019-2022: Administrative work (Ankara and İzmir).

Current Research - Baseline

• The Illusion of Independence
  • Education is inherently a nested system (Students \(\rightarrow\) Classrooms \(\rightarrow\) Schools \(\rightarrow\) Countries).
  • Traditional inferential statistics assume independent observations.
  • Ignoring hierarchical structures leads to biased standard errors and inflated Type I error rates.
• The Shift in the Field
  • Big Data: The massive scale of modern assessments (PISA, TIMSS).
  • Artificial Intelligence: The rise of automated test generation and scoring.

Current Research 1 Scaling with Big Data

• Multilevel Modeling of Large-Scale Data
• The Leuphana Collaboration:
  • Partnering with Educational Science Department.
  • Applying advanced solutions like multilevel decision trees to global educational student data.
  • Ensuring highly rigorous, reproducible research frameworks.

Current Research 2 The AI for Assessment

• Multilevel Modeling in Automated Scoring
  • Bringing psychometric validity to computer science engineering.
• Cross-Domain Applications:
  • Automated Turkish Essay Scoring
  • Evaluating high-stakes medical examinations.
  • Automated scoring in mathematics

Current Research 3 Mentorship

• Advising the Next Generation of Researchers:
  • Theory: Advancing the complex theory of multilevel models.
  • Methodological innovation: Split/analyze/meta-analyze methodologies.
  • Field Studies: Field applications and data gathering for automated scoring.

Current Impact

Contributions at the National Level

• ALTE Accreditation: Yunus Emre Enstitüsü (YEE) - TYS.
• Academic Literature: Eğitimde Ölçme ve Değerlendirme, ANI Publishing, 13th Edition.
• Tool Development: Architecting an R Platform for Social Scientists.
• Policy & Frameworks: UNICEF-MEB Assessment Framework for Special Education.
• Editorial: Section Editor for Education and Science; Special Issue Editor for EPOD.

Future Research Interests

• Medical Assessment: Automated scoring for high- and low-stakes medical examinations (4 works currently in progress).
• Language Acquisition: Automated scoring for Turkish as a Second Language.
• AI and Multilevel: Combining AI capabilities with the multilevel modeling of large-scale educational data.
• Student-centric AI: Harvesting AI capabilities to help students actively discover and map their own abilities.

Why BOĞAZİÇİ University?

• Scientific Synergy: To collaborate!.
• Mentorship at the Highest Level: Access to brilliant student minds to grow together and sha(r/p)e the future.
• Institutional Contribution: Contribute to the existing network
• Personal Commitment: As a family of four, we are fully committed to rooting our future in İstanbul.