Current Topic...

HBOK 1-066

Title Using Graphical Diagnostics to Deal With Bad Data
Author(s) Mark J. Anderson and Patrick J. Whitcomb
Source Quality Engineering
Topic Data Analysis
Abstract The problem of how to deal with individual results that do not appear to fit with the rest of the data is considered. Graphical tools are provided that aid in the diagnosis of this response data while maintaining a balance between deleting data that vary due to common causes and not detecting true outliers that occur due to special causes. The review of two real life data sets demonstrates that diagnostic plots found in readily available statistical software aid in determining the course of action needed to draw proper conclusions from experiments that produce bad data.
Access Restrictions ASQ members and journal subscribers
Link for .PDF
Link for HTML None
Reference Code 1-066

Add a Comment