Accuracy vs precision

[BrickBiscuit]

On one of those 'Dull People' threads, a best before date of '2026JN01 23:35' was praised for its accuracy (although I suspect the 23.35 was actually a production machine code and not part of the date). Someone commented 'You mean precision. Accuracy is if they go off at that minute.' I was inordinately impressed.

Ooo, burn!

I'd argue that it's not really precision, surely you need more than one value to determine if they are precise or not.

I'd take the state of deterioration of the product as one value and the date as the other. However, I still think how well those values correspond is 'accuracy'. How precisely they are described is 'precision'.

Science teacher here. This annoys me all the time as even the exam boards cant agree on their definitions.

Accuracy - how close to the true value your measurement or result is.
Precision - how close together your measurements are. But can also mean how small an interval a piece of apparatus can measure to.

The expiry date could be both accurate and precise. But you would need to know the exact time the food expired for real to know if the date and time was accurate, and that would be hard to determine as food doesn't just go off from one minute to the next.

A long time ago I did a Physics degree, and I totally agree with this. We did a lot about accuracy and precision in our instrumentation topic. So say for a stopwatch - accuracy = how well it refkects a runner’s speed compared to the time they actually ran. Precision - the error in actually recording it. Eg: stop the watch at time x, what does it display? X? X+- 5%, X +- 20%? Is it always a bit over, a bit under, how consistent is the error?

My background is chemistry and I tend to think about it in terms of quantitative analysis, accuracy being how close to the true value is your test result e.g % of active ingredient and precision being how scattered are your replicate measurements and how many should you take, this is dredging up some memories. We used to use the picture of an archery target to make sure new trainees understood. But agree precision is also about what your equipment can measure eg decimal places on a lab balance.

Depends on the context then. For the best before date in question, accuracy is how close it is to the food going off (or whatever measure is selected). Precision is the fineness of the figures. Going down to the minute in several months time was ridiculous (although I still think it's a production code, not part of the date).