Construct Validation and Design Relevance of Neurophysiological Measures in TEL: A Scoping Review Dataset
Description
This is an extracted table of design relevance and construct validity of 241 studies reviewed as a part of a scoping review paper titled "From Signal to Design: Evaluating How Neurophysiological Measures Inform Technology-Enhanced Learning Environments" and submitted to T4E conference 2026 to be held at IIT Jodhpur from December 7-9, 2026.
Files
Steps to reproduce
This dataset was generated through a transparent and reproducible systematic literature review conducted in accordance with the PRISMA 2020 guidelines. The review protocol was prospectively registered on the Open Science Framework (OSF) on 23 June 2026 before database searching. A comprehensive search was performed in Scopus and Web of Science (WoS) using database-specific Boolean search strings combining neurophysiological measures (e.g., EEG, eye tracking, fNIRS, biosensors), technology-enhanced learning (TEL) contexts, and learning-related constructs (e.g., cognitive load, attention, engagement, learning outcomes). Searches were limited to peer-reviewed English-language journal articles published between 2010 and 2025. A pilot search validated the search strategy before the final search was executed on 23 June 2026. Eligibility criteria required empirical studies that collected neurophysiological data during authentic learner interaction with a TEL environment. Studies outside formal educational settings, non-peer-reviewed publications, non-English records, and studies where TEL was only background context were excluded. After database filtering, 2,449 records were exported as BibTeX (.bib) files. Bibliographic data were converted to Excel format using Biblioshiny via the Bibliometrix package in R (v4.5.2). Records from both databases were merged and deduplicated using an R workflow, producing 1,646 unique records for screening. Title and abstract screening was conducted in ASReview LAB, an open-source active-learning screening platform. A manually selected seed set of relevant and irrelevant studies initialized the machine learning model, which continuously reprioritized records based on reviewer decisions. Screening followed predefined stopping criteria and included independent verification of a 10% sample by a second reviewer. This process resulted in 342 studies for full data extraction. A standardized extraction form was piloted before full extraction. Variables included bibliographic information, TEL context, participant population, neurophysiological measures, targeted learning constructs, construct validity evidence, and design impact evidence. Construct validity and design impact were coded using predefined four-level ordinal frameworks developed for this review. Data were synthesized narratively. Studies were categorized by TEL environment, neurophysiological measure, and learning construct to address RQ1. For RQ2 and RQ3, coded evidence levels were summarized using frequencies and percentages overall and stratified by measure type and TEL context. This workflow ensures transparency, reproducibility, and consistent coding across the dataset.
Institutions
- Indian Institute of Technology KharagpurWest Bengal, Kharagpur
- University of HelsinkiUusimaa, Helsinki