ARTICLES

Attributes Measurement System Analysis

BIS.Net Team

Attributes measurement system analysis are a class of systems where outcomes are discrete, not continuous such as a weighing scale. Arguably most attributes gauges are the go-no-go type where a decision on a part is either accept or not accept.

Analysis of the performance of these systems varies and depends on the gauge and application. There is no singe method one can use.

However, since the biggest risk is at the limits (grey area) where accept or reject decisions are made it makes sense to quantify the measurement system with a Gauge Performance System in that region.

BIS.Net Analyst uses the Analytic Attributes Gauge R&R method, based on the MEASUREMENT SYSTEMS ANALYSIS Reference Manual Fourth Edition, AIAG. Manuals can be ordered from www.aiag.org. As this is a copyrighted manual the user needs to refer to this manual to obtain further information on the steps of conducting such a study.

In summary, the study involves evaluating parts, whose reference values are known, 20 times and recording the total number of accepts for each part. This information is then used following the steps outlined in the above-mentioned manual to provide the following output

A Performance Chart to assess the risk of accepting parts outside and near the limits.

A probability plot to evaluate the Normality Assumption. It can also be used to show the unadjusted repeatability which is the range in reference values bounded by the two lines at the .995 and .005 acceptance probabilities.

The most important information is provided in tabular form as shown below.

If the t value for bias exceeds the critical value, then the bias is significant.

Alternatively, if the p value falls below the specified level of significance the bias is significant.