單一受試者研究分析中應用C統計量之適切性探討

The Inappropriateness for Applying C Statis-tic in the Analysis of Single-subject Research

C 統計量是目前在分析單一受試者研究資料時，使用較多的統計分析方法。由理論探討與虛擬案例分析可知，C 統計量主要在衡量小樣本序列資料的一階自相關趨勢，但仍需注意首尾兩筆資料的偏離對C統計量的影響。如果研究者不只是想要知道，單一受試者觀察資料是否具有一階自相關趨勢時，還想要知道某種介入是否會造成水準或斜率的改變，此時應用C統計量便不適切。因此，建議研究者日後在單一受試者研究中，應避免採用C統計量，而改採其他適當分析方法，諸如斷裂時間系列分析，或在能捕捉誤差系列相關之模式中執行分段迴歸法等，才能更適切與深入地探討單一受試者的觀察資料。

Currently， C statistic is the statistical analysis method often used for analyzing the single-subject research data. C statistic mainly measures the first-order autocorrelation trend of the small-sample series of data by theory study and simulation case analysis， but it still needs to pay attention to the influence on C statistic due to deviation of the two data in the initial and the last points. If in a single-subject study the researchers want to know not only whether the first-order autocorrelation trend of the observed data exists， but also whether the intervention causes a change of a level or slope， applying C statistic will be inappropriate. Therefore it suggested that in future single-subject studies the researchers use other appropriate methods， such as interrupted time series analysis or segment regression analysis executed in a model that can catch the serial correlation of errors， rather than C statistic to more appropriately and deeply explore the observed data of a single-subject.