Attribute measurement system analysis (MSA) is a statistical method used to evaluate the reliability and validity of attribute data. Attribute data refers to data that can be classified into discrete categories or attributes. Examples of attribute data include the number of defective products, the presence or absence of a feature, or a pass/fail test result. Attribute MSA is used to determine the accuracy and precision of the measurement system for attribute data.

The importance of attribute MSA

Measurement systems are used to collect data for decision-making, process control, and quality assurance. The reliability and validity of the data collected depend on the accuracy and precision of the measurement system. If the measurement system is not reliable or valid, the data collected will not accurately reflect the true condition of the process, and decisions based on the data may be incorrect.

Attribute MSA is important in ensuring that the measurement system is reliable and valid for attribute data. It helps to identify the sources of variation in the measurement system and to determine the magnitude of the variation. This information can be used to improve the measurement system, which will result in more accurate and precise data.

The attribute MSA process

The attribute MSA process involves several steps:

  1. Define the attributes to be measured: The first step is to define the attributes to be measured. Attributes are characteristics that can be classified into discrete categories or states. For example, the attribute “color” can be classified as “red,” “blue,” or “green.”
  2. Define the measurement method: The next step is to define the measurement method. The measurement method should be standardized, consistent, and reliable. The measurement method should be clearly defined in a standard operating procedure (SOP) and should be followed consistently by all operators.
  3. Conduct a Gauge R&R study: The Gauge R&R study is used to evaluate the reliability and validity of the measurement system. The study involves measuring the same attributes using different operators and different parts to evaluate the reproducibility and repeatability of the measurement system.
  4. Analyze the data: The data collected from the Gauge R&R study are analyzed to determine the sources of variation in the measurement system. The variation can be attributed to the measurement equipment, the operator, or the part being measured. The results of the analysis are used to identify the causes of the variation and to develop a plan to improve the measurement system.
  5. Implement improvements: The final step is to implement improvements to the measurement system. The improvements may include recalibration of the measurement equipment, training of the operators, or improvement of the measurement method. The improvements should be evaluated to ensure that they have resulted in a more reliable and valid measurement system.

Types of attribute MSA

There are two types of attribute MSA: variable data and attribute data.

Variable data attribute MSA involves the measurement of continuous variables, such as length or weight, that can be expressed as numerical values. In variable data attribute MSA, the accuracy and precision of the measurement system are evaluated using statistical techniques such as the Gauge R&R study.

Attribute data attribute MSA involves the measurement of discrete variables, such as the presence or absence of a feature. In attribute data attribute MSA, the accuracy and precision of the measurement system are evaluated using techniques such as the attribute agreement analysis and the Kappa statistic.

Attribute agreement analysis

The attribute agreement analysis is used to evaluate the agreement between two or more operators when they classify the same attributes into different categories. The analysis involves the comparison of the classifications made by each operator with the “true” classification, which is determined by a reference or standard. The level of agreement is then calculated using statistical techniques such as the percent agreement, Cohen’s Kappa, or Fleiss’ Kappa.

The attribute agreement analysis is useful in identifying the sources of variation in the classification of attributes. If the level of agreement between operators is low, it may be due to differences in interpretation of the standard or differences in training. The analysis can be used to identify the cause of the variation and to develop a plan to improve the classification process.

Kappa statistic

The Kappa statistic is a statistical measure of agreement between two or more observers. It is used to measure the agreement between observers when they are classifying attributes into different categories. The Kappa statistic takes into account the level of agreement that would be expected by chance and provides a measure of the agreement that is above chance.

The Kappa statistic ranges from -1 to 1, with values less than 0 indicating no agreement, values close to 0 indicating chance agreement, and values close to 1 indicating perfect agreement. The Kappa statistic can be used to evaluate the reliability and validity of the classification process and to identify sources of variation in the classification process.

Case study

A manufacturer of medical devices was experiencing high levels of rejection due to a lack of reliability in the measurement system for the presence or absence of a feature. The manufacturer conducted an attribute MSA study to evaluate the measurement system.

The first step was to define the attribute to be measured, which was the presence or absence of a feature on the device. The measurement method was defined as a visual inspection of the device.

A Gauge R&R study was conducted with three operators measuring ten devices each. The study revealed a high level of variation in the measurements, with most of the variation attributed to the operator. The attribute agreement analysis was used to evaluate the agreement between the operators. The analysis revealed a low level of agreement between the operators, indicating that the measurement system was not reliable.

The manufacturer implemented a plan to improve the measurement system. The plan included the development of a standard operating procedure for the measurement method, the training of the operators in the use of the measurement method, and the calibration of the measurement equipment.

After the improvements were implemented, a second Gauge R&R study was conducted, and the results showed a significant reduction in the variation in the measurements. The attribute agreement analysis was used to evaluate the agreement between the operators, and the results showed a significant improvement in the level of agreement between the operators.

Conclusion

Attribute MSA is an important process in ensuring the reliability and validity of the measurement system for attribute data. The process involves the evaluation of the accuracy and precision of the measurement system and the identification of sources of variation in the measurement system. The results of the analysis can be used to develop a plan to improve the measurement system, which will result in more accurate and precise data.

There are two types of attribute MSA: variable data and attribute data. The accuracy and precision of the measurement system for variable data are evaluated using statistical techniques such as the Gauge R&R study, while the accuracy and precision of the measurement system for attribute data are evaluated using techniques such as the attribute agreement analysis and the Kappa statistic.

Attribute MSA is particularly important in industries where the quality of the product is critical, such as the medical device industry. A lack of reliability in the measurement system can result in defects or failures that can have serious consequences for the user.

In conclusion, attribute MSA is an essential part of the quality assurance process in any industry that uses attribute data. The process provides a systematic approach to evaluating the reliability and validity of the measurement system, which is critical to the accuracy and precision of the data collected.