School Entry Detection of Struggling Readers Using Process Data From a Literacy App

• In Event: Tools to Support Decision Making: Multi-Tiered System of Supports (MTSS), Machine Learning, and More?

Abstract

Current methods for risk detection at school entry rely on few behavioural datapoints, and remain error-prone. We present a novel approach, using a fun and pedagogical serious literacy app for “stealth detection”.
The serious literacy app GraphoGame was played in class daily for ten minutes by 1676 Norwegian school starters over a five-week period during fall 2021,
generating rich process data.
Combined with results from the end-of-year national screening test, machine learning models were trained on the process data to
identify learners with risk of developing reading difficulties.
Results suggest that process data from serious apps in combination with machine learning may supply teachers with new tools for early identification of struggling learners.

Authors

• Njål Foldnes, University of Stavanger
  Presenting Author

• Steffen Grønneberg, BI Norwegian Business School
  Non-Presenting Author

• Per Henning Uppstad, University of Stavanger
  Non-Presenting Author

• Jenny Thomson, University of Sheffield
  Non-Presenting Author