Fig. 5: Split-half reliability of metacognitive scores.
From: A comprehensive assessment of current methods for measuring metacognition
Correlations between each measure were computed based on odd vs. even trials for sample sizes of 50, 100, 200, and 400 trials. The figure shows that split-half correlations are high when at least 100 trials are used for computations but become unacceptably low when only 50 trials are used. The x-axis shows the results for three different datasets: Hadda (Haddara), Shekh (Shekhar), and Manis (Maniscalco).
