Here is the move that makes statistics portable. Take one observation, ask how far it sits from the centre, then measure that distance in standard-deviation-sized steps: $z = \dfrac{x - \mu}{\sigma}$. Read aloud: z equals x minus mu, over sigma. Four letters, four jobs — $x$ is the one observation you are asking about, in the data's own unit; $\mu$ (mu) is the mean of the whole set, in that same unit; $\sigma$ (sigma) is the standard deviation, also in that unit; and $z$ is the standard score, which wears no unit at all. Whichever one the question leaves blank is the one you solve for.

Run the cancellation check and it explains itself: marks over marks, millilitres over millilitres — the unit divides out and $z$ walks away naked. That is exactly why a z-score can compare a mark on a chemistry test with a time in a swimming heat. If a rearrangement of yours leaves a unit stuck to $z$, it is wrong, no appeal. Units that do cancel never prove you right — the check runs one way only.

And a negative z is not an error. It says the observation sits BELOW the mean, which half of any data set does. The wound this lesson exists to prevent is subtracting in the comfortable order: $\mu - x$ reports every strong performance as negative and every weak one as positive, and the arithmetic will not warn you. It is always the value first, the mean second.