Work

Combining ratings with traditional assessment data using a mixed fixed/random psychometric model

Year: 2024

Type: article

Abstract: We present a new statistical and psychometric model for combining ratings with traditional assessment data, focusing on how fixed and random elements can coexist within a single model to provide a more holistic view of assessments. We highlight practical applications, such as using teacher ratings along with traditional tests to assess student performance over time. The newly proposed mixed item model is particularly useful in environments where both predictable and variable elements influence learning outcomes. By combining fixed items from standardized tests with random items derived from teacher observations, these models offer a more nuanced understanding of student performance and learning dynamics. Such integration enhances the interpretability of in-class observations, augmenting the reliability of assessments and providing a richer, more detailed picture of student progress throughout the academic year.

Source: Measurement Sensors

DOI: 10.1016/j.measen.2024.101506

Authors: Perman Gochyyev, Mark Wilson

Institution: University of California, Berkeley

Language: English

Topic: Psychometric Methodologies and Testing

Subfield: Management Science and Operations Research

Field: Decision Sciences

Domain: Social Sciences

SDG: Quality education

Open Access status: gold

Estimated APC paid: $1,800