電腦化數字警醒度測驗（ComputerizedDigit VigilanceTest, C-DVT)可迅速且 有效地評估個案之持續性注意力，具臨床常態使用之潛力。然而，C-DVT之隨機 測量誤差仍大，尚待改良。本研究目的為比較二種施測C-DVT方式（「增加練習 時間」與「二次施測取平均」）之隨機測量誤差是否低於原施測方式，以探索有效 降低隨機測量誤差的方法。同時，我們檢驗上述二種施測方式對C-DVT再測信度 與練習效應的影響。共16位中風病患參與本研究，個案以隨機分配到增加練習時 間組或二次施測取平均組（每組8人）。我們另抽取二次施測取平均組（即個案 於初、再測之第一次評估結果）的資料組成原始施測組。隨機測量誤差之結果（以 最小可彳貞測差異值百分率（percentage of minimal detectable change, MDC°%))顯示： 增加練習時間（19.4°%)與二次施測取平均（21.4°%)都比原始施測組（22.9°%)低。 再測信度方面，二次施測取平均組的組内相關係數（intraclass correlation coefficient =0.74)高於增加練習時間組（0.69)與原始施測組（0.70);練習效應部分則發現 二次施測取平均組的效應值（Cohen's d= 0.34)和增加練習時間組（0.37)都比 原始施測組低（0.39)。本研究結果顯示：正式施測C-DVT前增加5分鐘的練習， 具潛力可降低較多隨機測量誤差。而二次施測取平均較有助於提升再測信度與降 低練習效應。未來研究宜增加樣本數，以確認增加練習時間對降低隨機測量誤差 的效果。

The Computerized Digit Vigilance Test (C-DVT) is an efficient and valid measure of sustained attention that has a great potential to be used routinely in clinical settings. However, the random measurement error of the C-DVT remains large. The purpose of this study was to investigate potential methods of administration to minimize random measurement errors of the C-DVT. Test-retest reliability and practice effects of these methods were also compared. Sixteen patients with stroke were randomly assigned to two groups (i.e., "increased practice time" and "averaging two assessments") with each of 8 patients. The results of the original C-DVT group were extracted from the "averaging two assessments" group (i.e., the first testing at the pre- and post-assessments). Regarding the random measurement error, we found that both "increased practice time" (19.4%) and "averaging two assessments" (21.4%) groups had smaller percentages of minimal detectable change than that of the "original C-DVT group" (22.9%). Considering test-retest reliability, only the "averaging two assessments" group had a higher intraclass correlation coefficient (0.74) than that (0.69~0.70) of the others. About practice effect, both "increased practice time" (d = 0.34) and “averaging two assessment sets" (0.37) groups had smaller effect sizes than that (0.39) of the "original C-DVT" group. Our findings support that increased practice time has the potential to reduce random measurement errors of the CDVT; whilst averaging two assessments may be a better choice for reducing test-retest reliability and practice effect. Further study can verify effect of increased practice time on random measurement error with a larger sample size.