Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier

https://orcid.org/0009-0006-0843-1827

AccessType

Open Access Dissertation

Document Type

dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Education

Year Degree Awarded

2023

Month Degree Awarded

September

First Advisor

Lisa Keller

Subject Categories

Educational Assessment, Evaluation, and Research

Abstract

There is increased interest in personalized learning and making e-learning environments more adaptable. Some e-learning systems may use an Item Response Theory (IRT)-based assessment system. An important distinction between assessment and learning contexts is that learner proficiency is expected to remain constant across an assessment, while it is expected to change over time in a learning context. Constant learner proficiency during an assessment enables conventional approaches to estimating person and item parameters using IRT. These IRT-based systems could be abandoned for alternative approaches to modeling learners and system learning content, but assessments may provide more functions than adapting learning material to students. Thus, there is the question, how can e-learning systems with IRT-based assessment components more dynamically adapt their learning content? Is there a solution that leverages IRT for adapting the learning content of the system? A promising solution is the Urnings algorithm. Like other candidate algorithms, it is computationally light, but this algorithm has mechanisms for preventing variance inflation and is suitable for e-learning contexts. It also provides a measure of uncertainty around estimates. It has been studied both through simulations and applications to e-learning systems. Results are promising; however, there has not been an application of the Urnings algorithm to an e-learning context where there are conventionally estimated person parameters to compare the algorithm estimates to. This study addresses this gap by applying the Urnings algorithm to a K–8 reading and mathematics learning platform. In data from this platform, we have person parameter estimates across academic years from an in-system diagnostic assessment. Results from this study will help industry researchers understand the feasibility of the Urnings algorithm for large e-learning systems with IRT-based assessment components.

DOI

https://doi.org/10.7275/35778802

Download

Share

COinS