Doctoral candidate Nnamdi Ezike’s research is helping ensure that test takers’ scores on assessments like Advanced Placement, the GRE, Praxis, and TOEFL are more accurate.
Nnamdi is the first student in the U of A Educational Statistics and Research Methods graduate program to secure an internship with Educational Testing Service, the nonprofit that develops and administers these tests. He’s currently collaborating with other ETS scientists on projects related to performance assessments.
“Human raters who rate open-ended assessments are fallible despite training and calibration. Working with ETS this summer, I performed Monte Carlo simulations to evaluate different methods of adjudicating essays when raters disagree,” he said. “Specifically, I explored the impact of four different adjudication methods in accurately determining students’ true scores.”
Nnamdi is in his element.
“ETS is an excellent research shop with outstanding assessment and measurement scientists,” he said. “I was most excited to learn from experts in the field of psychometrics.”
Statistical methods are widely used in psychometrics, or psychological testing. Nnamdi’s research will help improve the accuracy of test scores that require the input of human raters.
Nnamdi is working with ETS on other research to improve large-scale testing efforts. Advances in computer technology mean essays can be graded using Natural Language Processing algorithms. Collaborating with other ETS scientists, he’s exploring how the presence of rater biases influences the performance of these automated scoring engines.
“Large-scale assessments with open-ended questions will benefit from this research as it will provide more confidence in the scores produced by automated engines,” he said.
Nnamdi has been fascinated by the field of measurement since his first exposure to survey research and statistics. Nnamdi worked as a data entry clerk at a consulting firm in Nigeria after graduating high school. The firm conducted research spanning from health, education, and agriculture. Soon after, Nnamdi went on to earn an undergraduate degree in statistics. He immediately applied to the U of A doctoral program after earning a master’s degree in statistics from Montana State University.
The graduate program’s coursework is focused on theoretical and applied statistics in an educational context. Still, it can also apply to other academic fields. Nnamdi is most interested in working in higher education so he can teach students how to use statistical concepts in their research and projects.
“We live in a data-driven society, and it’s important to provide key stakeholders and administrators with accurate results,” he said.