Screening for Dyslexia in University Students: a Standardized Procedure Based on Conditional Inference Trees

Area of Science:

• Neuroscience
• Educational Psychology
• Psychometrics

Background:

• Dyslexia screening in university students is crucial for academic success.
• Existing screening methods may lack efficiency or optimal diagnostic properties.
• There is a need for validated tools to identify dyslexia in higher education.

Purpose of the Study:

• To validate seven tests for adult dyslexia screening.
• To develop a standardized, efficient screening procedure using a subset of tests.
• To identify optimal tests, cut-off scores, and administration order via machine learning.

Main Methods:

• Conditional inference tree analysis, a supervised machine learning approach.
• Validation of seven diagnostic tests including reading fluency, phonemic awareness, and self-report questionnaires.
• Data collected from 60 university students with dyslexia and 65 without.

Main Results:

• A four-test combination (Adult Reading History Questionnaire, text reading fluency, phonemic awareness, pseudoword reading) demonstrated high discriminatory power.
• Optimal conditional cut-off scores were identified for the selected tests.
• The screening procedure achieved an overall classification accuracy of approximately 90%.

Conclusions:

• Conditional inference trees enable efficient dyslexia screening using a reduced test battery.
• The findings support a more streamlined approach to identifying dyslexia in university students.
• The study offers practical tools for researchers and practitioners in dyslexia assessment.