對於少量多樣的產品，由於資料的少量化，導致無法有效估計製程參數，所以傳統管制圖的設計方式並無法有效的監控這類型的製程。另一方面，在製程開始時，若要及早發現錯誤，也是有類似的問題。這種少量資料的製程，其監控方式在文獻中有兩大類作法。第一類是Hiller（1969）提出的兩階段設計方式，作者以兩階段的方式重新計算正確的管制圖參數，以確保在製程是管制中（in control）時，發生錯誤警訊（false alarm）的機會是在控制中。另一類是Quesenberry（1991）所提出的Q統計量管制圖，此類型的管制圖無須收集大量資料估計製程參數便可計算管制圖管制界線以監控製程。這兩類管制圖雖都有其設計的優點，但當製程不在管制中（out of control）時，尚無文獻比較其偵測出異常的機率。本篇文章將比較這兩類管制圖的優劣，並提出適當的建議以監控少量資料的製程（short runs process）。
Many production processes in manufacturing industries have been found with short runs. Because of the deficiency of sample size, the control limits of traditional control charts cannot be established accurately. Two major charting approaches to overcome this difficulty were provided, respectively, by Hiller (1969) and Quesenberry (1991). Control charts developed under both approaches have a false alarm rate close to the nominal one. However, these two approaches do not address the detection power for process shift. In this paper, the strength and weakness of these two approaches are discussed and some useful suggestions are provided for monitoring the short runs processes.
