Measurement System Analysis (MSA) – Explained

While there is a lot of talk about striving for the best product design, there is considerably very less attention on evaluating the variety of quality testing techniques available. While the Quality control Manager learns about an array of choices at school, he applies the method that has been followed in his company for a long time. Similarly, when the report card of whether a product/process/service is up to the mark is determined by the data collected or a sample assessed on a selective basis, it is important to know that the data/sample is collected with accurate systems. Is it not time that we measure the effectiveness of the systems that are used to collect data to judge the quality of a product, information, process or service? Statisticians have always debated on how to reduce the chances of error while applying Sampling techniques, but are they correct?

Measurement System Analysis  (MSA) is a tool or a set of procedures employed to judge the quality of measurement systems. A manufacturing company may have to deal with the following two scenarios if it has a poor measurement system in place. The first case can be that the quality control department reports zero defects, and so the product is declared ‘defect-free’. Later, with a sudden increase in the number of defectives and products returned, the company realizes that the products may have been ‘out of tolerance level’ or wrongly labeled as ‘OK”. The second case can be when due to very stringent quality control mechanisms exercised, good products are rejected as bad/defective ones (look at figure below showing both types of errors).

The role of Measurement System Analysis helps to scientifically answer the following set of questions:
• Is the current measurement system up to the mark? Are the techniques employed reliable, accurate and precise?
• Is the measurement system able to distinguish between good and bad sample data?
• Is the sample itself representative of the population and has the sample been collected on the technique of random sampling (without anyone’s judgment)?
• Is the measurement system stable and reliable over a period of time or are there any inconsistencies?

A faulty or unscientific measurement system leads to poor or inconsistent conclusions. This can jeopardize the entire manufacturing procedure and people may play the ‘blame game’ in order to justify the results. A lot of books and online references suggest that there are some parameters to judge any measurement system: bias, linearity, stability, repeatability and reproducibility. Let us assume that a quality engineer is responsible for examining clevis pins, which have the same dimension, same weight and are produced by the same machine. Now if there is a variation in the average values of the sample, it can be attributed to only a few possible set of causes: either there is something wrong with the sampling system or there is a human error involved (the operator who is testing the pins may be committing an error).

While the measurement system has to be stable (give consistent values over a period of time), accurate (acceptable difference between the average values and the actual values of the sample measured), it must not ignore the human role towards precision and accuracy. An operator is expected to perform the same task repeatedly and yet, there are big chances of error. Any measurement system can be judged on this account to test if the operator is able to test the same part with the same measurement device and get the same result or that different operators test the same part and get the same results. These are called the features of repeatability and reproducibility of measurement systems.

Conclusion: The above discussion gives an overview to the purpose of measurement system analysis and explains how such an effort can prevent a bad/defective product from reaching the consumer’s workplace or home. Statisticians and Quality Assurance engineers often couple the use of MSA with Statistical Control Techniques.

Learn more about our Six Sigma training courses or services or to register, please visit our website or call us at (855) 674-4622.