3.4 PER MILLION
Improve data accuracy for managing, improving processes
by T. M. Kubiak
AS QUALITY or Six Sigma professionals, we have been taught to address the issue of data accuracy and integrity from the statistical viewpoint.
Both the Certified Six Sigma Black Belt (CSSBB) exam and the Certified Quality Engineer (CQE) exam test these concepts, although they are described somewhat differently in their respective bodies of knowledge (BoKs).
The data accuracy and integrity description in the CQE BoK (section VI.A.4) states:
Describe the characteristics or properties of data (for example, source/resource issues, flexibility and versatility) and various types of data errors or poor quality such as low accuracy, inconsistency, interpretation of data values, and redundancy. Identify factors that can influence data accuracy, and apply techniques for error detection and correction. (Apply)
This column will address most of the elements of this subtopic by discussing why data accuracy is important, the numerous causes of poor data accuracy, various techniques for improvement and the placement of effective data collection points.
Importance of data accuracy
Poor data accuracy undermines two key precepts in a quality-focused or Six Sigma organization: fact based and data-driven decision making.
When faced with data that has been rendered useless because of data accuracy issues, decision makers must revert to relying on intuition.
This might provide the naysayer with reason and justification to ask why "this Six Sigma approach won’t work here." Remember, Six Sigma drives cultural changes in organizations, and accurate data are necessary to drive and sustain real improvement.
In addition, some organizations might be required to submit highly accurate data because of regulatory or reporting requirements, or because of contractual requirements. This is particularly true in industries such as aerospace in which data, in addition to product, are a deliverable.
In fact, most organizations have processes that deliver only data to other organizations.
Minimizing poor data accuracy
There are many causes of poor data accuracy. Table 1 includes a few of the most significant causes. Consider how these issues relate to your own experience and create your own list so you can minimize the impact of poor data accuracy in your organization.
Table 2 lists techniques used to minimize the occurrence of poor data accuracy or its impact.
Remember, processes are measurable and generate data. Treat poor data accuracy as you would any other quality defect: Use the plethora of tools available to you as a quality or Six Sigma professional to ferret out root cause. Chart data defects as you would product defects. Make them available for all to see.
Common data collection points
Five common data collection points are summarized in Table 3 and are as follows:
End of process: Traditionally, organizations have collected data to measure process performance at the very end when everything is said and done and it’s usually too late to take corrective action.
In process: Over time, more progressive organizations have realized that data should be collected upstream and during the process. Data collection at critical subprocesses allows corrective action to be taken earlier when change is still an option. This move facilitates more effective process management and allows the organizations to better gauge the overall health of their processes.
Points of convergence: This occurs when multiple streams of processes come together. Collecting data at this point makes sense because it helps organizations understand how each input stream affects the downstream subprocesses.
Decision points involving "yes" or "no" decisions are useful data collection points, particularly when the "yes" path is the desired path and the "no" path is the undesired path.
Understanding the percentage of products or services taking the "no" path becomes an important piece of data necessary for driving process improvement.
Across functional boundaries: These are nothing more than process hand-off points. However, responsibilities and accountability usually change across functional boundaries.
As such, collecting data at these points helps organizations understand how each functional department supports or detracts from process efficiency and effectiveness. It also helps minimize organizational finger pointing with data and fact.
Accuracy is essential
This column covers the numerous causes of poor data accuracy, a variety of techniques available to help minimize or eliminate these causes and the five common data collection points in processes in which the capture of data can have a significant impact on understanding process performance.
The delivery of data might be a regulatory or contractual requirement. Consequently, poor data accuracy can adversely impact an organization. Data defects like product defects can and should be charted and improved using the variety of tools at the disposal of the quality or Six Sigma professional.
While the current CSSBB BoK does not identify data accuracy in the same way as the CQE BoK, data accuracy is essential for managing and improving processes with sustained results. We look forward to seeing the description of data accuracy updated in the next revision of the CSSBB BoK to reflect the key elements of this important topic.
- Benbow, Donald W., and T.M. Kubiak, The Certified Six Sigma Black Belt Handbook, ASQ Quality Press, 2005.
- Certified Six Sigma Black Belt Body of Knowledge, ASQ, 2001.
- Certified Six Sigma Black Belt Body of Knowledge, ASQ, 2007.
- Certified Quality Engineering Body of Knowledge, ASQ, 2001.
- Certified Quality Engineering Body of Knowledge, ASQ, 2006.
- CNN.com, "Metric Mishap Caused Loss of NASA Orbiter," Sept. 30, 1999. www.cnn.com/tech/space/9909/30/mars. metric.02.
T.M. Kubiak is a consultant based in Charlotte, NC. He is a co-author of The Certified Six Sigma Black Belt Handbook. Kubiak, a senior member of ASQ, serves on many ASQ boards and is a past chair of ASQ’s Publication Management Board.