Differences between Test Accuracy Ratio (TAR) and Test Uncertainty Ratio (TUR)

Can anybody explain the differences between Test Accuracy Ratio (TAR) and Test Uncertainty Ratio (TUR)?

ralutz:

I am under the impression they are the same, used to evaluate and express measurement uncertainty, however I am not positive about this.

The website for the National Institute of Standards and Technology, NIST, might provide you with a more informative answer.

awk

[Jerry Eldred]

They are pretty close, but I think there is a little difference between them.

Uncertainty is the combined inaccuracy of a measurement. In a calibration context, if there are multiple instruments used to make a calibration measurement, if you used the Roo Sum Squares (RSS) method to calculate a total uncertainty, that would fall under the 'uncertainty' umbrella. The total uncertainty of a measurement may simply be the accuracy tolerance of a measurement standard. Or in the case of multiple instruments, and perhaps if environmental variation is a contributing factor, that amount of variation would also be factored in. Uncertainty refers to making a measurement.

Accuracy on the other hand refers to the plus and minus tolerance of an instrument.

TAR is a simpler quantity, referring to the ratios of the accuracies of two pieces of equipment. In calibration, uncertainty is prabably a more correct term. Where as in some other simpler applications, accuracy may be applicable. If you are using a hand held multimeter with a stated accuracy of +/- x% DC volts to adjust a power supply that has a stated accuracy of +/-y%, TAR would be appropriate. A complicated temperature calibration system with multiple instruments and opportunities for error would need to all be calculated together and a measurement uncertainty calculated.

------------------

Another way to approach this is to remember the subtle difference between accuracy and uncertainty.
Accuracy is usually a design specification of a product feature. Uncertainty is always a performance characteristic of a measurement process.

I usually explain the TAR/TUR difference like this:

When you are using an instrument to measure a product, the specified accuracy (uncertainty) of the measuring instrument must be X times better than the accuracy tolerance of the feature being measured. That is a Test Accuracy Ratio.

When you are calibrating a measuring instrument, the uncertainty performance of your calibration measurement process must be X times better than the measurement uncertainty (accuracy) of the instrument being calibrated. That is the Test Uncertainty Ratio.

The difference is subtle, real, and usually does not make much difference.

Graeme C. Payne
ASQ Certified Quality Engineer
Graeme@asqnet.org