AN INVESTIGATION OF ITEM CHARACTERISTICS WHICH ARE PREDICTIVE OF ITEM BIAS (FULL CHI-SQUARE METHOD).
Item
-
Title
-
AN INVESTIGATION OF ITEM CHARACTERISTICS WHICH ARE PREDICTIVE OF ITEM BIAS (FULL CHI-SQUARE METHOD).
-
Identifier
-
AAI8423106
-
identifier
-
8423106
-
Creator
-
STEIN, SUSAN VOLPERT.
-
Contributor
-
Alan Gross
-
Date
-
1984
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Education, Educational Psychology
-
Abstract
-
The primary purpose of this study was to determine if a set of observable item characteristics could be identified a priori which would be predictive of a statistical index of item bias. First, data was analyzed from the standardization sample of the Metropolitan Achievement Test, 1978, Form J (MAT-J) for grades seven, eight and nine to establish the criteria by which the item characteristics would be scaled on the Reading, Math and Science subtests. A set of seven item properties were defined for each subtest. The full chi-square index was used as the measure of item bias and developed for each item as both a continuous and dichotomous value. In addition, four different methods for constructing the majority (i.e. White) and minority (i.e. Nonwhite) groups were employed. Then, prediction equations were derived based upon the multiple correlation of these characteristics with the eight chi-square indices. Cross-validation of these prediction equations were done on Form K of the MAT, 1978 (MAT-K) with a different sample of examinees of the same age, and on the General Education Development Test (GED) a high school equivalency test for examinees aged 17 and older.;The results indicate that item characteristics can be defined which correlate with the measures of item bias. It was found that the prediction equations worked better for the MAT-K, a different form of the same test with the same age group than for the GED, an entirely different test administered to a different age group. The practical implications of the a priori identification of biased items are enumerated as well as suggestions for future research.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
-
Program
-
Educational Psychology
