This study compares how the generative AI (GenAI) large language model (LLM) ChatGPT performs in grading university exams compared to human teachers. Aspects investigated include consistency, large discrepancies and length of answer. Implications for higher education, including the role of teachers and ethics, are also discussed. Three…

This document describes a set of best practices for developing, implementing, and maintaining the critical process of scoring constructed-response tasks. These practices address both the use of human raters and automated scoring systems as part of the scoring process and cover the scoring of written, spoken, performance, or multimodal responses.…

Scoring constructed-response items can be highly difficult, time-consuming, and costly in practice. Improvements in computer technology have enabled automated scoring of constructed-response items. However, the application of automated scoring without an investigation of test equating can lead to serious problems. The goal of this study was to…

The study covers the development and testing of the alternative mechanics survey (AMS), a modified force concept inventory (FCI), which used automatically marked free-response questions. Data were collected over a period of three academic years from 611 participants who were taking physics classes at high school and university level. A total of…

Studies on automatic scoring systems in writing assessments have also evaluated the relationship between human and machine scores for the reliability of automated essay scoring systems. This study investigated the magnitudes of indices for inter-rater agreement and discrepancy, especially regarding human and machine scoring, in writing assessment.…

The advent of artificial intelligence (AI), particularly large language models (LLMs) like ChatGPT and Gemini, has significantly impacted the educational landscape, offering unique opportunities for learning and assessment. In the realm of written assessment grading, traditionally viewed as a laborious and subjective process, this study sought to…

Purpose: Our aim was to evaluate the psychometric properties of the online administered format of the Test of Narrative Language--Second Edition (TNL-2; Gillam & Pearson, 2017), given the importance of assessing children's narrative ability and considerable absence of psychometric studies of spoken language assessments administered online.…

Automated essay scoring engines (AESEs) are becoming increasingly popular as an efficient method for performance assessments in writing, including many language assessments that are used worldwide. Before they can be used operationally, AESEs must be "trained" using machine-learning techniques that incorporate human ratings. However, the…

Purpose: Our aim was to evaluate the psychometric properties of the online administered format of the Test of Narrative Language--Second Edition (TNL-2; Gillam & Pearson, 2017), given the importance of assessing children's narrative ability and considerable absence of psychometric studies of spoken language assessments administered online.…

This study employed a mixed-design approach and the Many-Facet Rasch Measurement (MFRM) framework to investigate whether rater bias occurred between the onscreen scoring (OSS) mode and the paper-based scoring (PBS) mode. Nine human raters analytically marked scanned scripts and paper scripts using a six-category (i.e., six-criterion) rating…

In the current study, two pools of 250 essays, all written as a response to the same prompt, were rated by two groups of raters (14 or 15 raters per group), thereby providing an approximation to the essay's true score. An automated essay scoring (AES) system was trained on the datasets and then scored the essays using a cross-validation scheme. By…

In this research report, we describe the design and empirical findings for a large-scale study of essay writing ability with approximately 2,500 high school students in Germany and Switzerland on the basis of 2 tasks with 2 associated prompts, each from a standardized writing assessment whose scoring involved both human and automated components.…

Through Many-Facet Rasch analysis, this study explores the rating differences between 1 computer automatic rater and 5 expert teacher raters on scoring 119 students in a computerized English listening-speaking test. Results indicate that both automatic and the teacher raters demonstrate good inter-rater reliability, though the automatic rater…

Approximately a third of the Programme for International Student Assessment (PISA) items in the core domains (math, reading, and science) are constructed-response items and require human coding (scoring). This process is time-consuming, expensive, and prone to error as often (a) humans code inconsistently, and (b) coding reliability in…

To provide an opportunity for teachers to better understand the automated scoring process used by the state of West Virginia on our annual West Virginia Educational Standards Test 2 (WESTEST 2) Online Writing Assessment, the West Virginia Department of Education (WVDE) Office of Assessment and Accountability and the Office of Research conduct an…