| Summary: | 碩士 === 中原大學 === 心理學系 === 88 === The present study empirically compared several monitoring-effectiveness indexes with three stability criterions (stability with respect to item difficulty, stability within a single domain, and stability across domains). The G index was evaluated theoretically by Nelson (1984) as the best monitoring-effectiveness index. Consequently, it was widely adopted by many researchers in the field of metacognition. However, some researchers recently found that the value of the G index could not accurately reflect the subject's monitoring-effectiveness; it varied with item difficulty; and it was unstable within a single domain. In these years, some scholars adopted the PSbar, Bias, CI, DI, and ANDI indexes to evaluate the subject's monitoring-effectiveness. The purpose of the present study, thus, is to empirically compare these existing monitoring-effectiveness indexes for the most stable one.
Three experiments, each with a single factor design (item difficulty: easy/normal/difficult) were conducted in this study. Although these experiments belonged to different domains (the word recognition test, the face recognition test, the general knowledge test), they all adopted the confidence-judgment paradigm to measure the subject's monitoring-effectiveness. Results from the three experiments showed that the values of the ANDI and G indexes did not change with item difficulty. The PSbar, Bias, CI, DI, and ANDI indexes are stable within a domain. The PSbar, Bias, CI, and ANDI indexes could reflect the monitoring-effectiveness stability across domains. In conclusion, either from the perspective of discrimination indexes (DI, ANDI, and G) or from other accuracy indexes (PSbar, Bias, and CI), ANDI is the most stable monitoring-effectiveness index.
|
|---|
