Measurement scientist building fair, valid assessments with modern psychometrics and AI
I have spent more than ten years designing, calibrating and validating tests, from national computerized adaptive exams to test items generated with large language models. My work sits where item response theory meets machine learning and large language models, and it always comes back to one question: does this score mean the same thing for everyone who takes the test?
Rasch and IRT modeling, calibration, scaling, linking and equating, dimensionality and item analysis.
Differential item functioning, measurement invariance, mixture Rasch models, and bias review.
CAT design, item-bank development, blueprints, standard setting and item-writer training.
Language-based performance measurement, machine learning in assessment, LLM-assisted item development and validation.
Construct definition, survey and scale design, reliability and validity evidence, cross-cultural adaptation.
Evaluation designs, measurement plans, performance indicators, technical reports and grant proposals.
Ongoing research on using LLMs to generate test items, and on the evidence needed to show that generated items are valid, fair and ready for operational use. In a randomized controlled trial, I also studied individual and peer-supported ChatGPT use in nursing case analysis and its effects on diagnostic skills and cognitive load, published in Nurse Education in Practice (2026).
Psychometric work on measures of children's language skills: oral reading fluency research shared in several publications and NCME presentations, and partial credit model calibration, scoring and validity evidence for the MELVA-S science vocabulary assessment.
Work on whether tests treat every group the same: DIF detection methods, measurement invariance across many countries in PISA, mixture Rasch models, and a new study testing whether generative AI can flag biased items.
Developing new measurement instruments and adapting existing ones to Turkish culture with colleagues in health, sport and education, from esports orientation and children's digital addiction to rational drug use. Each study builds evidence on factor structure, reliability and measurement invariance.
Built competency frameworks with expert panels, test blueprints and item-writer training, then carried out CAT design, item-bank calibration, linking, scaling and standard setting.
Also lead psychometrician on a National Science Foundation proposal for AI-augmented assessment development, under review in 2026.
I have taught undergraduate and graduate courses in measurement, statistics and research methods, mentored graduate researchers, developed online measurement courses, and review proposals for NCME.
I am open to psychometric, research and evaluation roles, and I take on selected projects with testing programs, universities, EdTech teams and international organizations.
Item and test analysis, IRT calibration, equating and technical reporting.
DIF and invariance studies before an assessment goes to scale.
From construct definition and blueprint to items, pilot and validation.
Evaluating automated scoring and AI-generated items for validity and bias.