價值驅動注意力作業之再測信度：反應時間與資料有限正確率

背景與目的：本研究旨在探討價值驅動注意力典範中（Value-driven attentional capture, VDAC），價值刺激物（value stimulus）對注意力控制的影響，而此現象 可能與成癮疾患的病理機制有關。過去以反應時間的測量方式再測信度不佳。因此，本研究探討資料有限正確率的測量能否改善再測信度不佳的問題。方法：本研究包括兩個實驗，實驗一為複製 Anderson 等人（2011）的設計；實驗二為改用資料有限正確率進行測量。每次實驗均包括制約與測驗階段，總共進行兩次測量，且參與者只在制約階段獲得金錢報酬。測驗階段設計有高干擾、低干擾與無干擾三種情境，以 2（時間）*3（干擾情境）完全受試者內變異數分析以及針對高干擾效果進行信度分析。結果：兩個實驗中的有價值刺激物干擾效果，僅在反應時間為主的實驗中呈現顯著，使用資料有限正確率作為測量未能有效反映價值干擾對注意力的影響，因此並未進行信度分析。反應時間的信度分析結果顯示，採差異分數的信度分析結果皆不佳；而改以單一分數則觀察到良好的信度。結論：研究發現使用單一分數（single scores）代替差異分數（difference scores），能有效提升作業的信度。相較之下，單一分數具有較低的測量誤差和更高的個體間變異。然而，為使單一分數的解釋具備意義，必須以VDAC效應顯著作為前提。

Background and Purpose: This study aimed to examine the influence of valuestimuli on attentional control within the framework of the Value-driven Attentional Capture (VDAC) paradigm, a mechanism potentially associated with the pathology of addiction. Previous research has highlighted poor test-retest reliability when using reaction time (RT) as the primary metric. To address this, the study investigated whether accuracy measurements from a limited dataset could improve reliability. Methods: Two experiments were designed to test this hypothesis. Experiment 1 replicated the methodology of Anderson et al. (2011), while Experiment 2 utilized accuracy measurements with limited data as the primary focus. Each experiment comprised a conditioning phase followed by a test phase, repeated across two sessions. Monetary rewards were provided exclusively during the conditioning phase. The test phase included three conditions: high interference, low interference, and no interference. A 2 (time) × 3 (interference condition) within-subjects ANOVA and a reliability analysis of the high interference effect were conducted. Results: The results revealed that the interference effect of value stimuli was significant only in the experiment using reaction time as the primary measurement. Conversely, accuracy data from the limited dataset failed to effectively capture the attention-related effects of value interference. Reliability analysis showed poor outcomes for difference scores but demonstrated strong reliability when single scores were used. Single scores were associated with reduced measurement error and increased between-participant variability. Conclusion: The findings indicate that replacing difference scores with single scores can significantly improve the reliability of the VDAC task. Single scores offer advantages, including lower measurement error and greater variability across participants. However, for the interpretation of a single score to be meaningful, the significance of the VDAC effect must be a prerequisite.