Open AccessMEASUREMENT ERROR AND SAT SCORE CHANGE
Author(s) -
Alderman Donald L.
Publication year - 1981
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.2333-8504.1981.tb01266.x
Subject(s) - test (biology) , test score , selection (genetic algorithm) , affect (linguistics) , statistics , psychology , aptitude , contrast (vision) , correlation , regression , score test , mathematics , statistical hypothesis testing , standardized test , computer science , artificial intelligence , geometry , paleontology , communication , biology
ABSTRACT Score changes on admissions tests such as the Scholastic Aptitude Test (SAT) can vary widely among individuals repeating the test. To a large extent these score changes probably reflect the influence of errors of measurement as test candidates with low initial scores usually experience score gains upon retesting while test candidates with high initial scores often show score losses‐ Besides this phenomenon of scores regressing toward the mean upon test repetition, student self‐selection may affect score change as test candidates who perceive their initial scores as underestimates of their true abilities decide to repeat the test. This study applies a procedure which yields estimates of true score change on the SAT adjusted for regression effects and student self‐selection. It is shown that student self‐selection in deciding to repeat an admissions test probably involves factors in addition to the measurement error attributable to variations in aspects of test specifications and to variations in responses of test candidates across forms, and that estimated true score change remains nearly constant across initial score levels in contrast to the negative slope of observed score change across initial score levels.