A measurement system analysis (MSA) is a mathematical means of determining the amount of variation present in a measurement process.
Understanding a measurement system analysis
With businesses now reliant on more and more data to make important decisions, the data on which those decisions are based must be as accurate as possible.
The solution is an MSA, a resource-intensive component of the Six Sigma DMAIC process used to reduce defects, increase quality, and control costs.
Measurement system analysis is a formal, statistical method that evaluates measurement systems (devices or people) and assesses their potential to provide reliable data.
In other words, it enables the business to make sure that any variation in measurements is minimal when compared to variation in its processes.
Note that a measurement system is any system of related measures that result in the quantification of certain characteristics.
This also encompasses the validation or assessment of a particular unit of measure that is performed by personnel, software, fixtures, or gauges.
Measurement system analyses consider the accuracy, precision, and stability of the measurement system collecting the data.
Both the process variation and measurement (device) variation are quantified to define the total measurement system variation stems from multiple sources such as:
- Subjective decision-making. For example, one factory worker may consider a machine to be close to failure while another may not.
- The use of an improper tool to provide a numerical reading.
- Systematic errors that result from a poorly calibrated device, such as an industrial scale that is always 2% off.
- Sounding or recording errors that are caused by a person not using enough significant figures or incorrectly recording the number itself.
- Environmental factors such as temperature, heat, humidity, and the like.
The five perspectives of a measurement system analysis
Once the two sources of variation have been examined, it is time to minimize the variation in the management system so that variation in the process can be understood better.
To that end, five perspectives of measurement error must be quantified before process capability can be established and data-based decisions are made. These perspectives relate to measurement precision and accuracy.
Precision describes the degree to which repeated measurements under the same conditions produce the same result. Put differently, it refers to how close two measurements are to each other. Precision-related perspectives include:
- Repeatability – the system is repeatable if the same person who measures the same object multiple times with the same device can obtain identical results.
- Reproducibility – the difference in the average measurements between different people using the identical characteristic or part of the same instrument.
Think of accuracy as the discrepancy between the observed average and the true average.
Inaccurate systems are characterized by an average value that differs from the true average.
Accuracy-related perspectives include:
- Bias – a one-directional tendency or the difference between an average observed value and the true or reference value.
- Linearity – the difference in bias value over the standard operating range of a particular measuring instrument. For example, can a scale that measures an object weighing 1000 kilograms do so as accurately as when it is measuring an object that weighs 50 kilograms?
- Stability – the system is stable if the variation is more or less constant over time.
Measurement systems analysis best practices
To maximize the benefits of an MSA, consider these best practices:
- Larger numbers of parts and repeat readings will produce results with a higher confidence level. But as always in business, the exhaustiveness of tests should be balanced with time, cost, and potential disruption to operations.
- Where possible, those who routinely perform a measurement or are familiar with procedures and equipment should be involved in the analysis.
- Ensure that the measurement procedure is documented and standardized among all MSA appraisers.
- Select parts that best represent the entire process spread. If the process is not properly represented, the extent or severity of the measurement error may be exaggerated.
- A measurement system analysis (MSA) is a mathematical means of determining the amount of variation present in a measurement process.
- Measurement system analyses consider the accuracy, precision, and stability of the measurement system collecting the data. Total measurement system variation is comprised of process and device variation and can be caused by several factors such as poorly calibrated devices or environmental factors such as heat or humidity.
- To minimize variation in the measurement system, five perspectives of measurement error must be quantified. Three relate to the precision of the system, while two relate to its accuracy.
Connected Analysis Frameworks
Related Strategy Concepts: Go-To-Market Strategy, Marketing Strategy, Business Models, Tech Business Models, Jobs-To-Be Done, Design Thinking, Lean Startup Canvas, Value Chain, Value Proposition Canvas, Balanced Scorecard, Business Model Canvas, SWOT Analysis, Growth Hacking, Bundling, Unbundling, Bootstrapping, Venture Capital, Porter’s Five Forces, Porter’s Generic Strategies, Porter’s Five Forces, PESTEL Analysis, SWOT, Porter’s Diamond Model, Ansoff, Technology Adoption Curve, TOWS, SOAR, Balanced Scorecard, OKR, Agile Methodology, Value Proposition, VTDF Framework, BCG Matrix, GE McKinsey Matrix, Kotter’s 8-Step Change Model.