NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED318747
Record Type: Non-Journal
Publication Date: 1990-Apr-12
Pages: 24
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Effects of Representation of Gender Groups in the Examinee Population on the Mantel-Haenszel Procedure.
DeMauro, Gerald E.
Recent comparisons (R. Freedle and I. Kostin, 1988) of differential item difficulty (DIF) have found that items that are in some way easier differentially give advantage to white examinees. In this study, the male/female ratio was manipulated among college students taking a social sciences examination to determine the effect of the male/female ratio on the relationship between item difficulty and the male/female DIF. A total of 1,009 students comprised the male sample; the female sample included 827 students. Mantel-Haenszel delta values and associated standard error values were computed for the first 251 and 110 females in the test file as well as for the total group of females. The relationship between item difficulty and the male/female DIF appears to be very complex. Some easier items were differentially difficult for males, and others were differentially difficult for females. Larger male/female sample ratios were associated with a decrease in item difficulty among items that disadvantage women. It is cautioned that the Mantel-Haenszel statistics, although affected by the examinee variables of this study, are nonetheless sensitive to real performance differences among populations, and that care must be taken to ensure that the samples on which they are based accurately represent the examinee population. Six tables and one graph present study results. (Author/SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A