NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED318791
Record Type: Non-Journal
Publication Date: 1990-Apr
Pages: 15
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Reconsideration of the ANOVA Method of Detecting Item Bias.
Seong, Tae-Je
The validity of the analysis of variance (ANOVA) of a split plot factorial design was investigated using a complex interaction contrast with matched two-group ability data for detecting biased items. The definition of a biased item by this method is: an item is biased if there is an item-by-group interaction when there is no group difference in achievement levels. Data were drawn from a study by M. J. Subkoviak et al. (1984), consisting of the responses of 1,022 white and 1,008 black college students to 50 multiple-choice vocabulary test items. In this data set, bias was deliberately manipulated by including black vocabulary items. Sixty matched pairs of subjects were selected to determine the ANOVA method with a small data group. The ANOVA method with the matched-pair data had a large interaction for an item if that item was biased. The ANOVA method with small data set is still valid for detecting biased items after matching the abilities of group members. Four tables are included and two tables are appended. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A