Need clear and simple clarification on the difference between accuracy and precision

Could someone suggest a better explanation of the difference between "accuracy and precision concept" other than this very common approach which is not clear to me.

EDIT

For example: Why is picture in top rights corner is "precise"? Is it because the dots are within the outer circle? Or is it because the dots are clustered together? What about the distance between the clustered dots center and the radius of the inner circle, this could be huge, would it still be precise?

Thank you.

• What about this explanation confuses you? I can’t imagine there is another way to explain the difference besides just trying to describe this picture. Nov 2, 2021 at 13:19
• It depends on how you define "accurate": if it is saying you are pointing in the right direction in general, but there is unbiased error - possibly substantial - then you get the top-left and bottom-left outcomes described as accurate; if it is saying each individual case must be close then you restrict yourself to the top-left and you need a different word for the bottom-left. Similarly for "precise" and the top-right outcome Nov 2, 2021 at 13:24
• According to the pictures , "precise" means "small standard deviation" and "accurate" means "close to the actual mean". But I do not think that the terminology is lucky in this case. Nov 2, 2021 at 13:30
• In this answer (and the comments underneath it), I characterise your given diagram as a statistics/business/science perspective of accuracy-versus-precision, and talk about how the concept/distinction seems to have an orthogonal meaning in numerical analysis. Hopefully it's not too rambly. Nov 2, 2021 at 13:56
• @ryang, Thanks for you help Nov 2, 2021 at 14:10

1 Answer

A numerical estimate is accurate if it's close to the true value. It's precise if you know it to many decimal places.

A clumsy but well conceived laboratory experiment could produce an accurate measurement that wasn't precise.

A well executed but faultily designed experiment could find lots of correct decimal places for a result nowhere near what you thought you were measuring.