With item response data, systematic variation within nested groups of generated items may negatively impact the estimation of item and person parameters. This paper studies a model that can capture the multilevel structure of the data and explain within-template systematic variability. The goal of this model is twofold. First, explaining and removing non-random error may improve ability and item parameter estimates. And second, finding systematic variation can bring insights into the educational process. Simulation results are discussed at length.
The Impact of Within-Template Systematic Variation on Response ModelsMaster's Thesis
|Author||Quinn Nathaniel Lathrop|
|Contributor||Scott Maxwell, Committee Member|
|Contributor||Lijuan Wang, Committee Member|
|Contributor||Ying Cheng, Committee Chair|
|Degree Level||Master's Thesis|
|Departments and Units|