Dissertation
Innovative approaches to analyzing multiple choice distractors: exploring patterns and relationships in educational assessment
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Autumn 2025
DOI: 10.25820/etd.008195
Abstract
This dissertation introduces and evaluates two innovative approaches for analyzing multiple-choice distractor data to uncover patterns of student misconceptions. The first approach applies latent class analysis (LCA) to nominal response data to classify students into latent groups based on shared response patterns. The second approach, nominal response modeling (NNM), builds on existing network frameworks by adapting them for nominal data and introducing a new way to map relationships among item options. Through simulation studies and empirical analysis using the Algebra Concept Inventory, this work examines the performance, interpretability, and instructional value of each method.
Results from simulation studies revealed that while LCA identified broad group patterns among students, it struggled to correctly classify all individuals, particularly when misconceptions were highly related. In contrast, NNM effectively highlighted connections between item options, segmenting portions of the network into clusters corresponding to specific misconceptions. When applied to real data, both methods provided complementary perspectives on student understanding, illustrating how certain item options group together, potentially reflecting common misconceptions and influencing reasoning across items.
Together, these findings advance contemporary distractor analysis beyond traditional binary scoring approaches. By offering new tools to model nominal response data, this work contributes to psychometric research and provides actionable insights for educators and assessment developers seeking to design items and interpret results in ways that reveal patterns in students’ reasoning and misconceptions.
Details
- Title: Subtitle
- Innovative approaches to analyzing multiple choice distractors: exploring patterns and relationships in educational assessment
- Creators
- Jacinta L Olson
- Contributors
- Jonathan Templin (Advisor)Lesa Hoffman (Committee Member)Won-Chan Lee (Committee Member)Claire Wladis (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Psychological and Quantitative Foundations (Educational Measurement and Statistics)
- Date degree season
- Autumn 2025
- DOI
- 10.25820/etd.008195
- Publisher
- University of Iowa
- Number of pages
- ix, 124 pages
- Copyright
- Copyright 2025 Jacinta L Olson
- Language
- English
- Date submitted
- 11/20/2025
- Description illustrations
- illustrations, tables, graphs
- Description bibliographic
- Includes bibliographical references (pages 92-94).
- Public Abstract (ETD)
- When students take multiple-choice tests, their wrong answers called distractors can reveal a lot about what they misunderstand. This dissertation explores new ways to study those responses to better understand how students think about math concepts. Two methods were used. The first, latent class analysis, grouped students based on similar response patterns. The second method, a new approach called nominal network modeling, mapped how different answers choices are related, showing which misunderstandings often occur together.
By studying both simulated data and real responses from the Algebra Concept Inventory, these methods showed how teachers and test developers can look beyond simply right and wrong answers. For example, the network approach identified key misconceptions that, if addressed, could help students improve across several items on an assessment.
This research provides a pathway to more meaningful assessments that capture student thinking rather than just performance. The findings can guide teachers in targeting common errors and support test developers in writing questions that provide richer information about learning.
- Academic Unit
- Psychological and Quantitative Foundations
- Record Identifier
- 9985135147002771
Metrics
1 Record Views