Found via http://rationalwiki.org/wiki/High_IQ_society
–
Abstract
This field study investigated the effect of retaking identical selection tests on subsequent test scores
of 4,726 candidates for law enforcement positions. For both cognitive ability and oral communication
ability selection tests, candidates produced significant score increases between the 1st and 2nd and the
2nd and 3rd test administrations. Furthermore, the repeat testing relationships with posthire training
performance and turnover were examined in a sample of 1,515 candidates eventually selected into the
organization. As predicted from persistence and continuance commitment rationales, the number of tests
necessary to gain entry into the organization was positively associated with training performance and
negatively associated with turnover probability.
–
Although the coaching studies are informative, test practice
alone is the issue of interest in the present study. Kulik, Kulik, and
Bangert (1984) summarized early research on practice effects
using meta-analysis. The authors drew almost exclusively on stud-
ies with student populations to examine practice effects on aptitude
and achievement test scores. They reported that test score increases
in the second administration were larger when identical tests were
used (0.42 SD) than when parallel forms of the tests were used
(0.23 SD). The authors also found a significant positive relation-
ship between test takers’ ability and size of the practice effect, as
effect sizes over two identical tests were 0.80 SD, 0.40 SD,
and 0.17 SD for subjects of high, middle, and low ability, respec-
tively. Finally, multiple test repetitions resulted in larger practice
effects, with a 0.42-SD mean increase from the first to the second
administration of an identical test (19 studies), a 0.70-SD improve-
ment from the first to the third administration (6 studies), and
a 0.96-SD increase from the first to the fourth administration (5
studies). In the most recent research on practice effects, psychol-
ogists have examined intelligence testing from a clinical perspec-
tive. Studies of the Wechsler Adult Intelligence Scale—Revised
and numerous other neuropsychological measures indicate that
improved scores tend to occur with repeat administrations of most
measures (Rapport, Axelrod, et al., 1997; Rapport, Brines, Axel-
rod, & Theisen, 1997; Watson, Pasteur, Healy, & Hughes, 1994).
in other words, the mathew effect at work. if we let everybody prep for tests, the scores will become more UNEQUAL, not more equal. plainly, smart people get more out of practicing.
–