Rounding bias

A reading that falls between two marks on a ruler has to be rounded somehow, and most people round the same direction without noticing they are doing it.

The pattern

Given a reading that sits between, say, 15.6 and 15.7 cm, there is a small but consistent pull toward calling it 15.5 or 16, toward the nearer round number rather than the nearer millimetre. This is not really about the measurement. It is a general feature of how people round ambiguous quantities, and it shows up here the same way it shows up anywhere someone reports a number that matters to them: rounding tends to drift toward the value they would rather have, more often than it drifts the other way.

Nobody doing this is being dishonest in any deliberate sense. It happens at the moment of deciding what to write down, which is exactly why it is easy to miss in yourself and hard to catch after the fact.

The fix

Write down the raw mark before deciding how to round it. Look at where the reading actually falls - between which two millimetre lines, or closer to which one - and record that position first, in as much detail as you can see it, before converting it to a final figure. Deciding the rounding rule in advance, separately from looking at the ruler, removes the moment where a preference can quietly steer the decision: round to the nearest millimetre, always, and apply that rule after the raw mark is already written down rather than while looking at the shaft.

This is a personal habit fix, not a claim that everyone rounds the same amount or in the same direction - it is a bias worth checking for in your own readings, not a fact about anyone else's.

Where this fits

This sits next to but is different from how population-level self-reports pile up at round numbers, which is a pattern visible across many people's reported figures rather than a single-reading habit; and it is different from the general rule for how finely to round, which is about resolution rather than direction. This post is only about the moment of deciding, for one reading, which way an ambiguous mark goes.

Checking this once is enough to know whether it applies to you. It has no bearing on anything subjective - a rating on Rate Cock or a photo score is a judgement, not a reading between two millimetre marks, and a human judge's opinion does not round the way a ruler does either. Nor does an AI model's output round in this sense - it produces a score from an image directly, with no ruler mark to round in the first place. A ruler reading rounds toward a number. A judgement just is one.