TY - JOUR
T1 - An Examination of Item Context Effects, DIF, and Gender DIF
AU - Ryan, Katherine E.
AU - Chiu, Shuwan
PY - 2001
Y1 - 2001
N2 - This investigation examined whether the patterns of gender differential item functioning (DIF) present in parcels of items are influenced by changes in item position. Items were studied collectively to detect differential bundle functioning (DBF) on 2 forms of a test of mathematics for college freshmen. To investigate order effects, 2 forms of the test were assembled: Form 1 (random) and Form 2 (easy to difficult within content area; n=3,932 for Form 1; n=1,074 for Form 2). Based on an attribute analysis (Harnisch, Tatsuoka, & Wilkins, 1995; Tatsuoka, 1993), item categories were constructed to examine the relation between DBF and specific item characteristics identified in previous work (e.g., word problems, items requiring higher order thinking skills, and items containing figures or graphs). Findings suggest that the amount of gender DIF and DIF present in item parcels tends not to be influenced by changes in item position. The gender DIF findings for the word problem category are an issue; the β value was .46 for Form 1 and .39 for Form 2.
AB - This investigation examined whether the patterns of gender differential item functioning (DIF) present in parcels of items are influenced by changes in item position. Items were studied collectively to detect differential bundle functioning (DBF) on 2 forms of a test of mathematics for college freshmen. To investigate order effects, 2 forms of the test were assembled: Form 1 (random) and Form 2 (easy to difficult within content area; n=3,932 for Form 1; n=1,074 for Form 2). Based on an attribute analysis (Harnisch, Tatsuoka, & Wilkins, 1995; Tatsuoka, 1993), item categories were constructed to examine the relation between DBF and specific item characteristics identified in previous work (e.g., word problems, items requiring higher order thinking skills, and items containing figures or graphs). Findings suggest that the amount of gender DIF and DIF present in item parcels tends not to be influenced by changes in item position. The gender DIF findings for the word problem category are an issue; the β value was .46 for Form 1 and .39 for Form 2.
UR - http://www.scopus.com/inward/record.url?scp=0035584836&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0035584836&partnerID=8YFLogxK
U2 - 10.1207/S15324818AME1401_06
DO - 10.1207/S15324818AME1401_06
M3 - Article
AN - SCOPUS:0035584836
SN - 0895-7347
VL - 14
SP - 73
EP - 90
JO - Applied Measurement in Education
JF - Applied Measurement in Education
IS - 1
ER -