• Stephen Biss

Reliability v. Accuracy

Is there a relationship between maintenance and reliability? Is regular maintenance a condition precedent to reliability? Are business records and data essential to an assessment of reliability? These questions will need to be addressed by Canadian Courts over the next few years.

But what is reliability? Too many lawyers confuse reliability with accuracy in their post St-Onge facta.

Accuracy has to do with hitting the centre of the target or getting your arrows to average the centre of the target. If you shoot multiple arrows you want their centremost average to be close to the centre of the target. Over several attempts, accuracy means that your arrows are centred aroud the bullseye. Accuracy is easy to contemplate when you have a fixed target. It is more difficult when you are using an alcohol or other standard that has its own possible sources of error. Alcohol standard in a simulator is subject to all sorts of possible measurement error. Introducing and analyzing the breath of a human being of any gender, weight, health, or age is even more difficult to contemplate. You are literally dealing with a moving target. Although we use Criminal Code presumptions as shortcuts, a defence based on reliability needs a good understanding of these concepts.

Precision has to do with hitting the target in close to the same place on each attempt. Your arrows may be far from the centre of the target but they are very close together. A set of cal. check results may be 117, 118, and 119 and still be precise. Breath test results of 139 and 111 may have acceptable precision with a subject whose true BAC is not in the 110 to 130 range.

Reliability is something different from accuracy and precision.

Specificity is also something different and very important. Are you measuring just ethyl alcohol or are you measuring something else as well?

Here's a link to the Wikipedia explanation of accuracy and precision. Note the target diagrams and references to error calculation. If evidentiary breath testing in Canada is scientific why don't we ever talk about calculation of error from systemic problems? That would mean we would need to know what the systemic problems are. There's yet another concept called trueness. There are also ISO standards of fair measurement.

Everything I can find on reliability seems to suggest repeatability, many times, in many places. It seems to have to do with different experimenters in different places using different but similar equipment getting similar results close to the real target over time.

If one St-Onge focus (the other is the operator) is the reliability of the approved instrument, in its system, with the alcohol standard, simulator, and simulator thermometers, being used by different people on different subjects or on different test benches over time , then don't we need to look at business records and data over time? Don't we need to know history?

Here's a video that I found on the Internet that describes accuracy, precision, and reliability from a science student's perspective.

#reliability #accuracy

5 views0 comments