Examining the Psychometric Properties of Seven Experimental Morphology Measures with Struggling Adult Readers: Item- and Person-Level Characteristics

Examining the Psychometric Properties of Seven Experimental Morphology Measures with Struggling Adult Readers: Item- and Person-Level Characteristics

First Author: Elizabeth Tighe -- Georgia State University
Additional authors/chairs: 
Yaacov Petscher
Keywords: Adult Literacy, Morphological Awareness, Item Response Theory, Explanatory item response models, Measurement
Abstract / Summary: 

Purpose: This study builds on literature supporting the important role of morphological awareness (MA) to the reading skills of struggling adult readers (e.g., Tighe & Fernandes, 2019). Although MA is an important contributor to reading skills, little is known about the reliability and validity of experimental MA measures used with struggling adults. We discuss the psychometric properties of seven experimental morphology measures administered to a sample of struggling adult readers. This information is helping us design a larger MA battery tailored to struggling adult readers.

Method: The battery of seven experimental morphology measures includes a total of 145 items administered to 136 struggling adult readers. Morphological skills across tasks involved using analogies to solve inflected word types, decomposing multimorphemic words, transforming root words into derived words, and identifying morphological endings in pseudowords. We will examine item functioning, consider multidimensionality within some of the tasks, and examine how item- and person-level characteristics may influence responses.

Results: Preliminary categorical CFAs and Rasch modeling suggests that generally items across morphological measures function well, discriminate best for those at lower to average abilities, and exhibit unidimensionality. One task had items that were too easy and did not discriminate well. One analogy-based task was multidimensional, characterized by regular vs. irregular inflected items. We will also use explanatory IRT modeling to examine how item- (e.g., inflected/derived, response format) and person-level characteristics (e.g., race/ethnicity, vocabulary, age) explain performance.

Conclusions: This study has important implications for designing more sensitive morphological items and measures tailored to struggling adult readers.