Open AccessESTIMATING RELIABILITY FOR TESTS COMPOSED OF ITEM SETS
Author(s) -
Lawrence Ida M.
Publication year - 1995
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2333-8504.1995.tb01653.x
Subject(s) - statistics , consistency (knowledge bases) , reliability (semiconductor) , internal consistency , mathematics , econometrics , test (biology) , set (abstract data type) , standard deviation , estimation , item analysis , computer science , psychometrics , paleontology , power (physics) , physics , geometry , management , quantum mechanics , economics , biology , programming language
ABSTRACT To what extent, if any, are estimates of reliability for a multiple choice test affected by the presence of large item sets where each set shares common reading material? The purpose of this research was to assess the effect of local item dependence on estimates of reliability for verbal portions of the old and new SAT, where the new test contains larger item sets associated with reading passages. Estimates based on a single administration of the test (estimates based on internal consistency and estimates based on covariances among parts) were compared to estimates based on two administrations of the test. When adjusted for a fixed standard deviation, estimates based on covariances among parts tended to be similar to estimates based on parallel forms. Both types of estimates were lower than the internal consistency estimates.