|
Activity Number:
|
345
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Quality and Productivity
|
| Abstract - #306721 |
|
Title:
|
Is Average Run Length a Good Measure of Performance?
|
|
Author(s):
|
Jay Schaffer and Chad Eshelman*+
|
|
Companies:
|
University of Northern Colorado and University of Northern Colorado
|
|
Address:
|
Applied Statistics and Research Methods, Greeley, CO, 80639,
|
|
Keywords:
|
average run length ; median run length ; control chart performance ; X Bar ; cusum ; EWMA
|
|
Abstract:
|
Average run length (ARL) has long been the benchmark characteristic used to evaluate a control chart's performance. However, Monte Carlo simulations indicate a control chart's run length distribution is often highly positively skewed. Under these conditions, the ARL tends to be "distorted" and may not provide an accurate measure of what the typical run length might be. This research examines the run length distributions of the X-Bar, CUSUM, and EWMA control charts under various process conditions. Median run length and ARL are compared in detail. The results suggest median run length is a more accurate measure of run length performance.
|
