V. Ponsoda et al., The effects of test difficulty manipulation in computerized adaptive testing and self-adapted testing, APPL MEAS E, 12(2), 1999, pp. 167-184
Easy and difficult versions of self-adapted tests (SATs) and computerized a
daptive tests (CATs) were compared. No significant differences were found a
mong the 4 tests (easy SAT, difficult SAT, easy CAT and difficult CAT) for
either estimated ability or posttest state anxiety. Significant differences
were found for the number of correct responses, testing time, anxiety chan
ge (pretest minus posttest anxiety difference), and standard error of abili
ty. The difficulty manipulation was successful as easy and difficult tests
differed in the number of items passed. Conditions with high percentage of
items passed produced less posttest than pretest anxiety (this was so for b
oth SATs and the easy-CATs). In the difficult-CAT condition, the number of
correct responses was lower and posttest anxiety exceeded pretest anxiety.
Results suggest that SAT research should continue to take into account the
variable "number of items passed" as it may hide the effects on anxiety whe
n CATs and SATs are compared. The 2 easy conditions show good psychometric
and motivational characteristics. However, the easy-CAT gave a higher preci
sion than the easy-SAT.