2021 Fiscal Year Final Research Report
Devising a new test analysis method constructed from methods of network analysis and item response theory
| Project/Area Number |
20K14152
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 10020:Educational psychology-related
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
TODO Naoya 筑波大学, 人間系, 助教 (70773711)
|
|---|
| Project Period (FY) |
2020-04-01 – 2022-03-31
|
| Keywords | 項目反応理論 / 局所依存性 / ネットワーク分析 |
|---|
| Outline of Final Research Achievements |
This study aimed and achieved to establish a new method of analyzing test data by combining the methods of network analysis, in which dynamics between variables are explored, and item response theory (IRT), in which item responses are modeled, to examine whether there are local dependencies among item responses in an exploratory manner and to enable test data analysis similar to that conducted in IRT.
Specifically, we first devised a method to achieve this goal based solely on models commonly used in IRT-based analyses and simple computational procedures. Then, we verified its effectiveness through simulations. Finally, we examined a method for computing statistics that contribute to test creation, evaluation, and administration.
|
| Free Research Field |
教育測定学、心理統計学、統計学
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究で得られた成果は、主にa)局所依存性を生じさせる要因やb)局所依存性の生じているデータに局所独立性を仮定して分析を行った場合の影響、c)局所依存性の検出方法の開発、d)局所依存性を考慮したモデルの開発といった観点から検討が行われてきた局所依存性に関する先行研究に新たな視座を与えるものとなっている。
また、本研究で提案された手続きは先行研究で用いられてきた複雑なモデルや計算方法を用いることなく実行できるものとなっており、実際にテストの作成・評価・運用を行うという観点からも非常に有用性の高い研究成果になっている。
|
