How to get really smart: Modeling retest and training effects in ability testing using computer-generated figural matrix items

Retest effects in matrix test performance: Differential impact of predictors at different hierarchy levels in an educational setting

“…Apparently, the above-mentioned abstract nature of the task type helps decrease potential memory and/or eureka!-effects on specific items. This supports and extends the findings reported by Freund and Holling (2011).…”

“…Second, parallel items can possess the same item structure, but occupy different positions in the test, and the position of their solutions among the distractors can be varied. Freund and Holling (2011) found no evidence for differential retest effects for identical vs. isomorphic items. In this study, we went beyond this design and investigated the differential impact of the two different degrees of parallelism.…”

“…Clearly, the majority of participants used close to the maximum time for the first test. In contrast, the participants in the study by Freund and Holling (2011) had no time limit at all, and consequently used significantly more time. A third explanation could be fatigue and diminishing motivation.…”

“…In a recent study, Freund and Holling (2011) investigated retest and training effects for MatrixDeveloper items in a sample of university and high school students, using a single retest design. They found a significant retest effect, which was even augmented with task-specific training.…”

“…We also controlled for possible gender effect and age in our analyses. Because Freund and Holling (2011) convincingly show that the psychometric properties of initial and retest scores are not invariant, we conduct our analyses at the level of the manifest scores only. By choosing the aggregated raw scores, we therefore focus on the primary information test administrators and applicants consider when using ability tests.…”