From Digital Reading Traces to Adaptive Instruction an Evidence Centered Diagnostic Model for Primary Education
Nurjamal Abdurazakova 
Abstract. This conceptual study develops an evidence centered model for using digital reading traces in the adaptive pedagogical diagnosis of primary school pupils. The problem addressed is that response accuracy, reading time, rereading, navigation, hint use, and answer revision are often treated as direct measures of comprehension although each trace is open to several interpretations. The study synthesizes evidence centered assessment design, formative assessment, process data research, digital reading studies, cognitive diagnostic adaptive testing, and ethical learning analytics. The resulting model connects six elements: a map of the reading construct, purposefully designed tasks, trace capture, evidence rules, teacher mediated instructional decisions, and continuous validation. It distinguishes observable events from inferred reading processes and requires triangulation before a learner is assigned a diagnostic profile. A construct to action matrix is proposed for word recognition, fluency, vocabulary in context, literal retrieval, inferential integration, evaluation, and metacognitive monitoring. The model also introduces safeguards for interface effects, language and cultural bias, unstable response time, privacy, opaque automation, and premature labeling. Its central claim is that adaptation should modify the next learning opportunity rather than define the child. Digital traces become educationally valuable only when they are linked to an explicit claim, generated by a suitable task, interpreted with uncertainty, and reviewed by a teacher. The framework is intended for future piloting in Karakalpak medium and other multilingual primary classrooms.
Keywords: reading literacy, primary education, digital traces, adaptive diagnosis, evidence centered design, formative assessment, multilingual education