Who gets stopped and searched

Across 36 forces that record ethnicity well enough to report, 1,151,285 searches have a recorded officer-defined ethnicity. Black people account for 18.2% of them, and about 4.2% of the population of England and Wales.

The national picture

Population shares are the 2021 Census for England and Wales, aggregated to police force areas using the ONS local authority to force area lookup. Search shares are calculated only over records where an officer recorded an ethnicity.

Is this what people mean by "two-tier policing"?

The phrase is used in several different arguments, and it is worth separating them, because this data speaks to one and not the others.

These figures show that the people recorded as searched are not distributed as the resident population is. That much is a matter of record and is not seriously disputed by anyone, including the forces themselves.

What the figures do not show is why. A difference in search rates can follow from where officers are deployed, from who is present in those places, from the offences being targeted, from decisions about individuals, or from some combination. This dataset cannot separate those, and any claim that it settles the question in either direction is going beyond what it can support.

The phrase is also used about the policing of protest and of speech, which is a different argument again, about different powers, and one no stop and search dataset can answer.

Every force, ranked

National averages hide the thing worth knowing. Each force is compared to the resident population of its own area, built from the 2021 Census, so the denominator is local rather than national. The ratio is the share of searches divided by the share of population.

Black people are searched at 10.6 times their share of the resident population at City of London Police, and 2.1 times at Northumbria Police.

A dash means the figure is not published for that force, either because it does not record ethnicity well enough or because the group is too small a share of the resident population for a ratio to mean anything.

The recording problem sits underneath all of it

Only 36 of 41 forces record officer-defined ethnicity well enough for their figures to be usable here. Where a force fails to record ethnicity on a large share of its searches, no analysis of that force is possible, and the national picture is missing whatever those searches would have shown.

That is a finding about the state, not a gap in this analysis, and it is the one thing on this page that could be fixed tomorrow by the forces themselves.

The two halves of this ratio are not recorded the same way

This is the most important limitation on the page, and it cuts against the headline figure rather than for it.

The numerator is officer-defined ethnicity: what the officer recorded at the time, without asking. The denominator is the self-defined ethnicity people gave the 2021 Census about themselves. Those are different measurements of different things, and they do not use the same categories in practice.

The clearest evidence is in our own figures above. Nationally, people recorded as Mixed account for 0.5% of searches against 2.9% of residents. That gap is far too large to be real. Officers rarely record "Mixed" at all, so people the Census counts as Mixed are being recorded as something else at the point of search — most often as Black or as White. That inflates whichever category absorbs them, and on these numbers it inflates the Black ratio.

So treat the direction of this finding as solid and the magnitude as uncertain. A disparity exists on any reasonable measure. The exact multiple depends on a recording practice the forces control, not on anything the person searched did.