Accuracy v Precision

We live in an age dominated by IT and its ability to crunch data. Without doubt this can yield significant benefits, but do we sometimes lose sight of reality as a result?

We were recently asked by a regulator to supply CBR results to two decimal places. Now, we can calculate data to any number of decimal places using the IT tools available to us, but just because we can doesn’t mean we should. Anyone familaiar with CBR testing will know that the test just isn’t that accurate. Quoting a CBR result to the nearest 0.1% might (just) be realistic for soft/loose materials. But quoting a result to the nearest 0.01% implies an accuracy that just isn’t warranted. The precision of the answer isn’t justified by the accuracy of the test.

So this is a plea for regulators (and anyone else for that matter) to apply a little common sense, and to remember the difference between accuracy and precision.