Bias in dyslexia screening in a Dutch multicultural population

Area of Science:

• Developmental Psychology
• Educational Psychology
• Linguistics

Background:

• Dyslexia screening tools can be culturally biased.
• The Dutch Dyslexia Screening Test (DST-NL) and protocol are used for dyslexia assessment in the Netherlands.
• Non-western immigrant children may be disproportionately affected by cultural bias in screening.

Purpose of the Study:

• To evaluate the adequacy of the DST-NL and Dutch dyslexia protocol for screening immigrant children.
• To identify subtests with less cultural bias and better convergence between measures.
• To investigate predictors of subtest performance, including cultural background and language knowledge.

Main Methods:

• Receiver Operating Characteristics (ROC) analysis comparing DST-NL and protocol outcomes.
• Cross-sectional and longitudinal design across third, fifth, and seventh graders.
• Multilevel analysis incorporating cultural background, dyslexia diagnosis, parental education, and Word Lexicon.

Main Results:

• Sensitivity and specificity improved with grade but were non-significant in the lowest grade.
• One-Minute Reading, Non-Word Reading, and Nonsense Passage Reading showed good convergence.
• Cultural background significantly predicted subtest performance after controlling for Word Lexicon, indicating bias.
• Technical literacy subtests demonstrated better convergence, while Naming Pictures and fluency tasks were less effective.

Conclusions:

• The DST-NL and standard protocol exhibit cultural bias, particularly in lower grades and multilingual groups.
• Dyslexia and literacy difficulties are challenging to disentangle in screening assessments.
• Technical literacy subtests are more reliable for screening in diverse student populations.