Repeated testing sessions and scholastic aptitude in college students’ metacognitive accuracy

Abstract

We performed three experiments to examine the effects of repeated study–judgement–test sessions on metacognitive monitoring, and to see if better students (those with higher Scholastic Aptitude Test or SAT scores) outperform low SAT students. In all experiments, mean metacognitive accuracy (bias scores and Gamma correlations) did improve with practice. Most improvement involved students’ ability to predict which items would not be recalled later. In addition, students with high SAT scores recalled more items, were less overconfident, and adjusted their predictions more effectively. Thus, high SAT students may be able to adjust their metacognitive monitoring effectively without feedback, but low SAT students appear unlikely to do so. Educators may need to devise more explicit techniques to help low SAT students improve their metacognitive monitoring during the course of a semester.

Acknowledgements

Portions of Experiments 1 and 2 were presented at the 40th annual meeting of the Psychonomic Society in November 1999. The authors thank Rita Massey for retrieving students’ SAT scores.

Notes

¹Consistent with previous research (e.g., Frey & Detterman, 2004), we used SAT scores as a proxy for general intelligence test scores. The claim that SAT scores can be converted accurately to IQ scores is controversial, however (Bridgman, 2005).

²Some Gs could not be computed for two participants due to a lack of variability in recall. In Sessions 1 and 4, one participant failed to correctly recall any stimuli; in Session 5, one participant correctly recalled all 20 stimuli. Data from these participants were excluded from the following ANOVA and post hoc test.

³Bias scores are used frequently to assess absolute metacognitive accuracy, although statistical limitations of under- and overconfidence measures have been noted (Juslin, Winman, & Olsson, 2000).

⁴SAT scores were unavailable for 19 of the 135 participants. For 12 of these students, scores from the American College Test (ACT) were obtained and converted to equivalent SAT scores using a conversion table provided by the Institutional Research and Testing department at Baylor University. Data from the remaining seven students were omitted.

⁵Before computing the ANOVAs, a chi-square test for independence was performed on condition and SAT group. No reliable effect was observed, χ²(2) = 3.69, p>.15, so ANOVAs were computed using data summed across all three conditions.