Journal article
Semi-supervised learning method for adjusting biased item difficulty estimates caused by nonignorable missingness under 2PL-IRT model
2020
Proceedings of the 13th International Conference on Educational Data Mining (EDM 2020), pp. 715–719
Abstract
In data collected from virtual learning environments (VLEs), item response theory (IRT) models can be used to guide the ongoing measurement of student ability. However, such applications of IRT rely on unbiased item parameter estimates associated with test items in the VLE. Without formal piloting of the items, one can expect a large amount of non-ignorable missing data in the VLE log file data, and this is expected to negatively impact IRT item parameter estimation accuracy, which then negatively impacts any future ability estimates utilized in the VLE. In the psychometric literature , methods for handling missing data are mostly centered around conditions in which the data and the amount of missing data are not as large as those that come from VLEs. In this paper, we introduce a semi-supervised learning method to deal with a large proportion of missingness contained in VLE data from which one needs to obtain un-biased item parameter estimates. The proposed framework showed its potential for obtaining unbiased item parameter estimates that can then be fixed in the VLE in order to obtain ongoing ability estimates for operational purposes.
See MoreThis article was published outside of NWEA. The full text can be found at the link above.
Topics: Measurement & scaling
Related Topics
Predicting Amira Reading Mastery Based on NWEA MAP Reading Fluency Benchmark Assessment Scores
This document presents results from a linking study conducted by NWEA in May 2024 to statistically connect the grades 1–5 English Amira Reading Mastery (ARM) scores with the Scaled-Words-Correct-Per-Minute (SWCPM) scores from the MAP Reading Fluency benchmark assessment taken during Fall and Winter 2023–2024.
Products: MAP Reading Fluency
Topics: Computer adaptive testing, Early learning, Measurement & scaling, Reading & language arts
Achievement and Growth Norms for Course-Specific MAP Growth Tests
This report documents the procedure used to produce the achievement and growth user norms for a series of the course-specific MAP® Growth™ subject tests, including Algebra 1, Geometry, Algebra 2, Integrated Math I, Integrated Math II, Integrated Math III, and Biology/Life Science. Among these tests, Integrated Math I, Integrated Math II, Integrated Math III, and Biology/Life Science were the first time to have their norms available. The remaining tests, i.e., Algebra 1, Geometry, and Algebra 2, had their norms updated including receiving more between-term growth norms by using more recent test events. Procedure for norm sample selection and a model-based approach using the multivariate true score model (Thum & He, 2019) that factors out known imprecision of scores to generate the norms are also provided in detail, along with the snapshots of the achievement and growth norms for each test.
Products: MAP Growth
Topics: Measurement & scaling
Longitudinal models of reading and mathematics achievement in deaf and hard of hearing students
New research using longitudinal data provides evidence that deaf and hard of hearing (DHH) students continue to build skills in math and reading throughout grades 2 to 8, challenging assumptions that DHH students’ skills plataeu in elementary grades.
Topics: Equity, Measurement & scaling
Achievement and growth norms for English MAP Reading Fluency Foundational Skills
This report documents the norming study procedure used to produce the achievement and growth user norms for English MAP Reading Fluency Foundational Skills.
Products: MAP Reading Fluency
Topics: Measurement & scaling
The MAP Growth theory of action describes key features of MAP Growth and its position in a comprehensive assessment system.
Products: MAP Growth
Topics: Equity, Measurement & scaling, Test design
Bayesian uncertainty estimation for Gaussian graphical models and centrality indices
This study compares estimation of symptom networks with Bayesian GLASSO- and Horseshoe priors to estimation using the frequentist GLASSO using extensive simulations.
Topics: Measurement & scaling
School officials regularly use school-aggregate test scores to monitor school performance and make policy decisions. In this report, RAND researchers investigate one specific issue that may contaminate utilization of COVID-19–era school-aggregate scores and result in faulty comparisons with historical and other proximal aggregate scores: changes in school composition over time. To investigate this issue, they examine data from NWEA’s MAP Growth assessments, interim assessments used by states and districts during the 2020–2021 school year.
Topics: Measurement & scaling, COVID-19 & schools
