What moderates the accuracy of ease of learning judgments?

Abstract

When people begin to study new material, they may first judge how difficult it will be to learn. Surprisingly, these ease of learning (EOL) judgments have received little attention by metacognitive researchers so far. The aim of this study was to systematically investigate how well EOL judgments can predict actual learning, and what factors may moderate their relative accuracy. In three experiments, undergraduate psychology students made EOL judgments on, then studied, and were tested on, lists of word-pairs (e.g., sun – warm). In Experiment 1, the Goodman-Kruskal gamma (G) correlations showed that EOL judgments were accurate (G = .74) when items varied enough in difficulty to allow for proper discrimination between them, but were less accurate (G = .21) when variation was smaller. Furthermore, in Experiment 1 and 3, we showed that the relative accuracy was reliably higher when the EOL judgments were correlated with a binary criterion (i.e., if an item was recalled or not on a test), compared with a trials-to-learn criterion (i.e., how many study and test trials were needed to recall an item). In addition, Experiments 2 and 3 indicate other factors to be non-influential for EOL accuracy, such as the task used to measure the EOL judgments, and whether items were judged sequentially (i.e., one item at a time in isolation from the other items) or simultaneously (i.e., each item was judged while having access to all other items). To conclude, EOL judgments can be highly accurate (G = .74) and may thus be of strategic importance for learning. Further avenues for research are discussed.