Please use this identifier to cite or link to this item:
An Analysis of Reliability and Item Fuction of the ELIPT Dictation Test
|Title:||An Analysis of Reliability and Item Fuction of the ELIPT Dictation Test|
|Abstract:||Dictation is commonly used as an integrative test of listening skills, yet relatively little has been published about the psychometric qualities of this format. Moreover, dictation test scores are problematic to analyze because typically one point is awarded for each correctly written word, which may result in violation of the assumption of local independence required by many statistical analyses. This study examined the psychometric properties of an English dictation test currently used for placement purposes at the University of Hawai`i. While initial classical and Rasch analyses suggested that the test was reliable and the items were generally functioning well, subsequent factor and other analyses indicated that the assumption of independence of items (words) did not hold. Re-scoring the test by pooling words within bursts (segments of text separated by pauses during reading) may provide a method for avoiding the local independence problem. Classical analysis suggested that scoring by burst produced good results, but Rasch analysis uncovered large amounts of person misfit and overfit. Potential causes of this Rasch misfit is discussed at the end of the paper.|
|Appears in Collections:||Student Work - ELI|
Items in ScholarSpace are protected by copyright, with all rights reserved, unless otherwise indicated.