Electronic portfolio assessment for metacognitive development in chemistry: needs analysis and conceptual framework for AI-enhanced implementation

Abstract

Grounded in self-regulated learning theory and formative assessment frameworks, this study investigates the theoretical basis for integrating electronic portfolios with artificial intelligence to support metacognitive assessment in secondary school chemistry, specifically in learning buffer solutions. A qualitative descriptive design was employed as the define stage of the 4D development model (Define, Design, Develop, Disseminate). Data were collected through semi-structured interviews with two chemistry teachers selected via purposive sampling to represent contrasting instructional contexts: a public senior high school implementing the Kurikulum Merdeka and a boarding school with restricted digital infrastructure. Interviews were supplemented by a systematic literature review. Data were analyzed using thematic analysis, involving familiarization, open coding, theme development, and synthesis. Trustworthiness was established through member checking and data triangulation between interview findings and literature. The findings reveal four interrelated barriers to effective metacognitive assessment: a persistent feedback bottleneck arising from high teacher workload; disparities in digital access across school contexts; students’ overreliance on algorithmic procedures at the expense of conceptual understanding; and a gap between authentic assessment ideals and operational classroom realities. These findings establish a theoretically grounded rationale for developing an AI-enhanced electronic portfolio that positions artificial intelligence as an automated scaffolding mechanism for formative feedback, rather than merely a technological add-on. The proposed conceptual framework offers design principles for subsequent development and empirical validation of the assessment system.