Page 366 - Six Sigma Demystified

346 Six SigMa DemystifieD

Interpreting a Reproducibility Control Chart
The X chart for reproducibility analyzes whether each appraiser’s average for
the selected parts is within the expected variability of these averages. In other
words, it detects whether the differences between appraiser averages are an
indication of appraiser bias or an expected variation owing to equipment.
If the equipment error, as represented on the combined appraiser repeat-
ability range chart, is in control, then σ is an appropriate estimate of the
repeat
variation from trial to trial. As long as there are not outside influences affecting
the trial measurements, such as appraiser bias, trial averages should fall within
the variability reflected by

σ no. of trials
repeat

This is the basis for the reproducibility X control chart.
The range chart for reproducibility analyzes the variability in appraiser aver-
ages for each part. As such, it may be used to identify whether the reproduc-
ibility is in a state of statistical control. If control is not exhibited, then the
estimate of reproducibility may not be valid. Out-of-control points indicate
that the variability in average measurements between appraisers is affected by
part bias.

regression Analysis

Regression analysis is used to determine if a dependent variable can be predicted
adequately by a set of one or more independent variables.

When to Use

Measure Stage

• Assess the linearity of a measurement system

Analyze Stage
• Investigate the relationship of process factors to metrics

Improve Stage

• Verify the relationship of process factors to metrics after improvement