ERIC Number: ED104949
Record Type: RIE
Publication Date: 1975-Apr
Reference Count: 0
A Reassessment of Standard Error of Measurement.
Klaas, Alan C.
Current usage and theory of standard error of measurement calls for one standard error of measurement figure to be used across all levels of scoring. The study revealed that scoring variance across scoring levels is not constant. As scoring ability increases scoring variance decreases. The assertion that low and high scoring subjects will correctly guess the same number of unknown questions is incorrect. Low and high scoring subjects correctly guess the same percentage of unknown questions. Clearly, low scoring subjects are guessing at a greater number of unknown questions and, therefore, will have a larger amount of scoring variance. Post hoc analysis of response option frequencies and changes of option choice was undertaken. Low scoring subjects apparently employ random guessing when answering questions where they do not know the correct answer. High scoring subjects seem to employ eeucated guessing when answering questions where they do not know the correct answer. Item difficulty appeared to be unrelated to the change from random guessing to educated guessing; however, the large number of omissions by low scoring subjects made the item difficulty effect difficult to interpret. (Author)
Publication Type: Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Annual Meeting of the American Educational Research Association (Washington, D.C., March 30-April 3, 1975). Document not available in hard copy due to marginal legibility of original document