Files in this item
|(no description provided)|
|Title:||Differential Item Functioning and Opportunity to Learn: Adjusting the Mantel-Haenszel Chi-Square Procedure|
|Author(s):||Switzer, Deborah Marie|
|Doctoral Committee Chair(s):||Harnisch, Delwyn L.|
|Department / Program:||Education|
|Degree Granting Institution:||University of Illinois at Urbana-Champaign|
|Subject(s):||Education, Tests and Measurements
Education, Educational Psychology
|Abstract:||Test fairness is an omnipresent educational and political issue. The validity of a test is in question whenever any group feels its members are not being treated exactly like everyone else. Differential item functioning (DIF) refers to a test item that does not perform the same way for two groups of examinees after controlling for ability.
Instructional differences may cause observed DIF. If two equal-ability students have had varying amounts of instruction on a topic, it would be expected that one might perform better than the other on items measuring that topic. Instruction on specific information covered by a test item is referred to as opportunity-to-learn (OTL). When OTL differs between groups, group membership may be confounded with this difference in OTL, which therefore can weaken an item bias study.
A critical consideration in DIF research is that only persons with the same level of ability are compared. The Mantel-Haenszel chi-square procedure controls for ability by comparing only those test-takers who have the same estimated ability as measured by total test score. Adjusting this procedure to incorporate OTL information is the purpose of this study.
In this study the Mantel-Haenszel chi-square procedure was adjusted by (a) creating an ability estimate from the percent of correct items from the subset of items for which the student had OTL; (b) separating the 2 x 2 chi-square tables to separate students who had had OTL and those who had not; (c) combining (a) and (b).
The data set contained item information from the 1985 Second International Mathematics Study. The 40-item core test was used. The students were United States eighth graders.
No significant changes in the DIF indices were observed in the adjusted procedures. The results suggest that the Mantel-Haenszel chi-square procedure is quite robust to differences in OTL. However, four items were flagged as biased against males. Four items were found to be biased against females, three of which were items testing knowledge of fractions. Further study is needed into possible gender differences in cognitive strategies, prior knowledge, and motivation.
Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 1993.
|Date Available in IDEALS:||2014-12-16|