A Quantile Regression Analysis of Cognitive Ability and Spelling Predictors of Written Expression: Evidence of Gender, Age, and Skill Level Moderation

Cognitive ability and spelling predictors of written expression were examined between gender groups as well as across age and writing skill levels using quantile regression with the Kaufman Assessment Battery for Children--Second Edition and the Kaufman Test of Educational Achievement--Second Edition conormed standardization sample data (N = 2,117). Quantile regression is an analytic tool that allows for examination of predictors at different levels of the achievement distribution. Spelling predicted differing levels of writing competency for both genders. For girls, retrieval fluency and crystallized ability predicted lower and higher levels of writing competency, respectively. From a bifactor model interpretation, general intelligence was the largest and most consistent predictor of writing competency. Using a higher order model interpretation, learning efficiency predicted lower to average writing competency for girls, whereas for boys, visual processing predicted all levels of writing competency, and retrieval fluency predicted average to higher levels of writing competency. Although results are preliminary, researchers and practitioners are encouraged to consider that predictors of writing competency may differ markedly in strength across the continuum of writing achievement.