心理學家的最愛與最恨：自我報告

THE ACCURACY AND STABILITY OF SELF-REPORTS

本報告認為「自我報告」需要相當的精確性及穩定性，才可作為基本的資料供研究用。本報告提出一對的係數，Cp及Cr'，分別用來測量順向和逆向自我監測的精確性。Cp是在知感的情境下發展成功的。一個由知感評定與再認所組成的數列首先解離為一組2×2的數列，每一數列由兩排知感程度的再認分數組成，其知感精確性是由兩知感程度間的再認率提升量接近其最大提升量的程度來表示。然後，整個數列的知感精確性是所有2×2數列的知感精確性的加權平均值。實驗結果指出，Cp能顯示一個高精確又穩定的知感預測力。Cp也是一個能免於「限制級距」及「細微性／粗糙性」效果的計量法。

In this report, a pair of coefficients is proposed for measuring the accuracy of prospective and retrospective metacognitive monitoring. The coefficient for prospective monitoring is developed in the context of feeling-of-knowing (FOK) prediction. The degree of FOK predictive accuracy for a data array formed by recognition performance and FOK rating is a weighted average of all degrees of FOK accuracy for the data tables formed by two rating points of the array. The degree of FOK accuracy for each fourfold table is indicated by the extent to which the observed increase in percentage recognition from one FOK rating to another approaches its maximal increase. It is a measure that is able to indicate a high and stable FOK accuracy across different portions of the FOK range. It is also a measure free from restricted-range and fineness/coarseness effects. Its counterpart applicable to retrospective monitoring is also described in this paper.