Leaderboard

Data points being mutually exclusive Adding weightage to attributes Choosing which attributes should participate in pareto

This quiz was started on Yahoo groups on Thu Jul 31, 2008 Six Sigma is all about making decisions based on facts and data. Hence, it is critical that we ensure that the data used on Six Sigma projects are of good quality. The process of ensuring that the data is of good quality is called Measurements Systems Analysis (MSA). There are several problems that can affect data quality: Accuracy, Stability, Linearity, Repeatability, and Reproducibility. The most common tools used in Six Sigma projects to check MSA is the Gage R&R (Crossed) for continuous data and Attribute Agreement Analysis for discrete data. I have the following discussion questions for the group (assuming the data is continuous): a) Is a Gage R&R study necessary on all Six Sigma projects? For example, what if the data is obtained from an automated system. Is a Gage R&R study sufficient to ensure data validity? _____________ CORRECT RESPONSE IS GIVEN BELOW Dear All, Thanks for your responses to this Quiz. It is clear to all of us that data integrity and validity are very important on Six Sigma projects. Since Six Sigma is based on making decisions based on facts & data, if we don't have the right data, we will of course make the wrong conclusions. Nevertheless, several Six Sigma projects that I had a chance to review clearly did not spend much time validating the data to make sure that the data was good. One reason given was that the data was coming from an automated mechanism, so it had to be good. Even if the data is coming from an automated source, we need to ask several questions. a) Is the system able to detect changes as and when they occur? b.) Do the measurements represent the "reality"? c) Is there a common understanding between different people with respect to what the data is telling them? A good operational definition is critical to interpreting the data correctly. Maybe the measurement system is measuring one thing and we need a different start and stop point based on our project metric. d) Is the measurement system stable over time? e) Is the measurement system linear over the measurement range? f) Are there potential for data entry, or data transfer errors from one application to another? g) Is there sufficient discrimination in the measurement system? Gage R&R only checks for one aspect of the measurement system (repeatability and reproducibility). You will need to perform system audits, calibration studies, linearity studies, run charts, & control charts to answer all the above questions. The amount of resources you invest in checking data of course depends on the criticality of making an error. At a minimum, it is imperative that we perform some rudimentary checks otherwise we run the risk of making incorrect conclusions on our Six Sigma projects. Best Regards, SJ.