Alphabet knowledge in preschool: A Rasch model analysis

In this study, we used Rasch model analyses to examine (1) the unidimensionality of the alphabet knowledge construct and (2) the relative difficulty of different alphabet knowledge tasks (uppercase letter recognition, names, and sounds, and lowercase letter names) within a sample of preschoolers (n = 335). Rasch analysis showed that the four components of alphabet knowledge did work together as a unidimensional construct, indicating all alphabet tasks administered were measuring the same underlying skill. With regard to difficulty of tasks, letter recognition was easier than letter naming, which in turn was easier than letter sounds, and uppercase letter names were easier than lowercase letter names. Most notably, most of the alphabet tasks overlapped, and the Rasch models for the single tasks were no more reliable than the combined measure. This suggests that these alphabetic tasks do not measure distinct skills but are instead indicators of a single ability. Consequently, we support the conceptualization of alphabet knowledge as a unitary construct, and suggest that those assessing and teaching alphabet knowledge in preschool use tests and methods that combine the various alphabetic tasks rather than separating them. These combined assessments will be more likely to capture the range of abilities within a preschool sample and avoid the floor and ceiling effects that have so often complicated early literacy research. (C) 2012 Elsevier Inc. All rights reserved.