Skill level: Advanced
The gauge linearity and bias study is used extensively in quality control in manufacturing but also extends beyond this practical application. It is used to determine the accuracy of measurements through the expected measurement range. In other words, the “linearity study” helps you determine if your gauge (or the tool used to take the measurement) has the same accuracy (precision) when measuring objects of different sizes, and “bias study” helps you to determine if your gauge is biased when compared to a master value.
As a regular occurrence in businesses, and especially in service/transactional processes, key measurement systems that are relied upon to judge whether an entity is within specification may not be very reliable. Unless you perform a detailed study of the gauge, you cannot be confident that the data you are getting is reliable, therefore adding risk to the decision making process. Ensuring that the gauge used to take the measurement is linear and has no bias is key and critical in any process.
- Leads to reliable data with little variation
- Offers increased confidence in the data and the decision making process
- Can be easily performed with available statistical application or method
How to Use
- Step 1. Linearity study: Set up an experiment with different outcomes over the range of the measurement.
- Step 2. Conduct the measurements and collect the data. Analyze the data using an appropriate statistical application or method.
- Step 3. Analyze the results. If the gauge proves to be non-linear over the range, it may need to be repaired, replaced, or calibrated.
- Step 4. Bias study: Identify a “master” or “reference” to check the gauge against to detect or measure any bias.
- Step 5. If the gauge has a bias, it can be corrected or repaired depending on the complexity of the gauge. A known bias can also be subtracted for the reading to get the true value.
A clinic serving patients diagnosed with diabetes uses blood pressure as its first assessment of a patient’s situation as either critical or non-critical. Patients are referred from either their family doctors or family practice clinics. If their blood diastolic (low) and systolic (high) numbers are outside of the standard, they are considered as critical and will be treated as high priority. Otherwise they will be on the waiting list to see a specialist.
The clinic nurse must check the medical files from the family doctors to make sure that patients have been pre-screened for diabetes. She notices that her readings over several days are very different than the ones from the other clinics, and she reports the issue to the clinic’s principal.
The clinic starts to investigate its blood pressure instrument by using a “reference” pressure simulator provided by the instrument manufacturer. The results are below:
These data show that the blood pressure instrument is unable to produce an accurate measurement in the full range of the specification, therefore leading the nurse to make an erroneous diagnostic. Patients with very high blood pressure were assessed as normal or not in critical condition in several cases.
The clinic sends the instrument back to the manufacturer for repair, calibration, and adjustment. As a control measure, the clinic begins to use on a random basis two different instruments to measure blood pressure and also conducts a monthly equipment calibration procedure to ensure that the data they are collecting is accurate and reliable.
Note: Gauge linearity and bias study is a complex topic, and many books and other resources are available. Only those specialized and well trained in quality control and instrumentation calibration can address this delicate and complicated issue.
The same concept applies when “humans” are the measurement instruments, as in, for example, an interview situation when candidate value and fit is evaluated. Much qualitative research addresses this issue of “human gauge linearity and bias,” which is beyond the scope of this discussion.