A Rasch Analysis of the Astronaut Invented Spelling Test

Abstract

The Astronaut Invented Spelling Test (AIST) was designed as an assessment tool for monitoring the literacy progress of students in the early years of schooling. The test uses an invented spelling format, in which children are encouraged to try to write words that they have probably not been taught how to spell. In the initial development phase of the test, AIST spelling responses were scored to give credit to all reasonable representations of all the phonemes in the test words, with bonus points awarded for use of specific conventional patterns of English orthography. A Rasch analysis was subsequently used to explore the psychometric properties of the test, based on data from two large samples of Australian schoolchildren in their first four years of schooling (N=654 and N=533). Of the original 48 AIST items, 28 were found to provide an excellent fit to the Rasch measurement model. These 28 items, ordered from easy to hard, were calibrated in terms of item difficulty on the same scale as the student measures. The scales for both samples were unidimensional and reliable, with Person Separation Indices of 0.96. The order of item difficulty that emerged from the Rasch analysis provided strong evidence about the early development of phonemic and orthographic awareness. By combining the item difficulty scale and the student measures, the Rasch analysis of the AIST provides early literacy teachers with a classroom-based assessment tool that is not only psychometrically robust for screening purposes, but that also supports the choice of specific instructional targets in the classroom.