Relative vs. Absolute percentages

The Wall Street Journal had an article this weekend titled “When Risk is a Red-Meat Issue” by Carl Bialik on March 23, 2012.  This may be behind the WSJ pay wall, but if you can read it, please do.

The article starts with a study that concluded that eating one serving of red meat per day increased your risk of death by 20%, while the author says that it is an increase of 0.2% increase in risk.  The 20% is a relative risk while the 0.2% is the absolute risk.  We may consider changing behavior with a 20% risk while we may not change for a 0.2% risk change.

In reporting on DMAIC projects, this concept of relative vs. absolute percentages is often an issue.  It might be a goal to reduce the defect rate by 50% when you truly reducing the defect rate by 1% (from 2% to 1%).  Both are right, but when communicated, they may not be treated the same by the audience.

Another example is in the current US news.  The US justice department denied a request by the state of Texas to require identification in order to vote.  The US Justice department stated that the Texas Hispanic population is 46.5% less likely to have identification and would be denied voting rights disproportionally.  It sounds like a big deal, but it is a relative percentage.  The absolute percentages are only a 2% difference.  Hispanic voters are projected not have identification 6.3% of the time while white voters are projected to not have identification 4.3% of the time.  What is missing from these percentages is the size of the denominator (the number of voters).  I live in Texas and I find it hard to get exact voter numbers, but another way to look at this issue is that about 22% of Texas voters are Hispanic, or about 2.75 million voters.  6.3% of the 2.75 million is about 170,000 voters without IDs.  Given around 12 Million registered voters in Texas would mean around 9 million are non-Hispanic.  4.3% of the non-Hispanic voters without ID is about 390,000 voters.  So the identification law is estimated to exclude 57% more non-Hispanic voters than Hispanic voters.  Does that change the view of the issue? possibly.

The point of this posting is that percentages can be deceptive and by selecting the numerator and denominator you can take almost any issue and conclude that the percentages prove your point.   Which ever point you choose to support.  This is why Smarter Solutions recommends avoiding percentages as a Lean Six Sigma Define-Measure-Analyze-Improve-Control (DMAIC) project goal (i.e., DMAIC “Measure” drill down to “Baseline Project”).

Use absolute percentage as a metric in DMAIC roadmap

Focus on actual absolute percentages numbers, not the percentage because the value is more difficult to game.

Relative vs Absolute Percentages and Other Metrics in an Enhanced DMAIC roadmap

Organizations benefit when they use the enhanced Integrated Enterprise Excellence (IEE) DMAIC roadmap shown above where Lean and Six Sigma tools are structurally integrated so that the most appropriate tool is used at the correct time to improve a project’s baseline metric.

The free link for a clickable IEE DMAIC roadmap is provided here , where application details of this clickable DMAIC roadmap are described in Management 2.0.

Organizations benefit when they report out the project’s baseline using an  IEE 30,000-foot level format.

A Free Software App for Creating 30,000-foot-level charts

A free software app is available for creating 30,000-foot-level charts for your process output response Excel dataset. This is a standalone software product and not an Excel add-in. 

 

business management system meeting

 

Contact Us through an e-mail or telephone call to set up a time for a discussion on how your organization might gain much from an Integrated Enterprise Excellence Business Process Management system. Or, a Zoom meeting can be schedule directly:

E-mail ([email protected]) or call us (+1.512.918.0280), if you encounter difficulties setting up a Zoom session directly or want to schedule another time that is not available in the Zoom-meeting calendar.