NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED531171
Record Type: Non-Journal
Publication Date: 2012-Apr
Pages: 23
Abstractor: As Provided
Reference Count: 12
ISBN: N/A
ISSN: N/A
A Comparison of Scoring Options for Omitted and Not-Reached Items through the Recovery of IRT Parameters When Utilizing the Rasch Model and Joint Maximum Likelihood Estimation
Custer, Michael; Sharairi, Sid; Swift, David
Online Submission, Paper presented at the Annual Meeting of the National Council on Measurement in Education (Vancouver, British Columbia, Apr 12-16, 2012)
This paper utilized the Rasch model and Joint Maximum Likelihood Estimation to study different scoring options for omitted and not-reached items. Three scoring treatments were studied. The first method treated omitted and not-reached items as "ignorable/blank". The second treatment, scored omits as incorrect with "0" and left not-reached as blank and the third treatment scored both omitted and not-reached items as incorrect with "0". These scoring treatments were studied across two levels of missing data: "0.81% omit rate/10% not-reached" and "1.62% omit rate/20% not-reached" and two levels of not reaching the end of the test: independent of ability and dependent on ability. In terms of Bias, RMSD [Root Mean Squared Deviation], and the number of outlier differences between estimated and "true" parameters, performance was best when omitted and not-reached items were ignored or left blank. Performance was poorest when omitted and not-reached items were scored as incorrect with "0". Additional tables are appended. (Contains 5 tables.)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A