Development of a Testlet-Based HOTS Assessment in Physics Using Polytomous Scoring Graded Response Model

Abstrak

This study develops a higher-order thinking skills (HOTS) assessment instrument using a testlet model and examines its measurement characteristics in physics learning. A mixed-methods Research and Development (R&D) design was employed following Mardapi’s framework, which includes planning, try-out, and measurement stages. Data were collected through expert validation, student responses, and field testing. Quantitative data were analyzed using the Graded Response Model (GRM), while qualitative analysis was used to evaluate content validity and item clarity. The results indicate that the instrument demonstrates acceptable measurement quality, with all items showing positive discrimination, appropriate difficulty levels, and adequate model fit. The testlet structure supports the assessment of sequential reasoning processes, allowing students’ responses to be represented across different levels of understanding. These findings suggest that the instrument provides a valid and practical approach for assessing higher-order thinking skills in physics learning contexts.