Sigma Metric Analysis
Scoring for Six Sigma metrics
In an effort to improve the Sigma-metric studies presented on this website, we're establishing a ranking scale for the quality of a Sigma-metric. This is meant to make the utility of a metric more transparent, as well as an encouragement to others to improve how they collect data and assess it by Sigma-metrics
A Six-Star Scale for Sigma-metrics
January 2024
Sten Westgard, MS
In an accompanying essay, we discussed the challenges of standardizing or standardising the Six Sigma metric. In the face of daunting challenges, the very least step we can make is to assess how reliable a particular Sigma-metric calculation is, and communicate that transparently.
Points will be awarded for better metric practices, with the higher the score indicating a more reliable metric.
Points for Imprecision Determination
- (+2) From reference material
- Or, if using something other than a reference material:
(+0.5) From 3rd party control source - (+0.5) Control levels measured near the critical decision levels
- (+0.5) Commutable materials
- (+0.5) intermediate precision conditions (>100 pts, 3-6 months)
- ( 0 ) From manufacturer controls
- ( 0 ) Under short-term/ within-run precision conditions
Points for Bias Determination
- (+2) Comparison vs Reference method/material
- (+1) Comparison vs EQA/PT survey
- (+1) Comparison vs peer group program
- (+0.5) Under intermediate precisions conditions
- (>100 data points, ideally 3-6 months)
- (+0.5) using commutable materials or actual patient samples
- ( 0 ) Comparison vs. field method
- ( 0 ) Comparison vs. IQC target mean
- ( 0 ) Under short-term / within-run conditions
Points for Performance Specification Selection
- (+2) from Clinical Use
- (+1) from Biological Variation derived specification (EFLM database, Ricos goals)
- ( 0 ) From CLIA/CAP/national EQA/PT programs
Points for method of calculation
- (+0.5) calculated with units
- (0) calculated with % values
Final Scoring
| Points | Usefulness of Sigma-metric |
| 0-1 | MISLEADING |
| 2 | UNHELPFUL |
| 3 | MODERATE |
| 4 | GOOD |
| 5 | EXCELLENT |
| 6-7 | GOLD STANDARD |
You will note that it is possible to get better than 6 Stars on a Six Sigma metric, but this is highly unlikely to occur. It would require the use of reference methods and materials to determine imprecision and bias. It would be a marvelous accomplishment, but very few laboratories or manufacturers have the resources to conduct such a study. We are much more likely to see 3 star Sigma metrics than 6. In this way, the scoring of the Sigma-metric mimics the scale itself.