Interpreting our data

Data on this website comes from either:

  • survey data – figures we show are estimates which are checked to make sure they are reliable and representative
  • administrative data – figures are based on actual numbers

Whichever type of data is used, you should always keep in mind:

  • how group size affects the reliability of data
  • the risks of making generalisations about any ethnic group as a whole

How group size affects the reliability of data

We usually only comment on reliable (‘statistically significant’) findings from survey data, otherwise we say if data is not reliable.

Data is reliable when we can be confident it reflects the total population. This means we would get similar findings 19 times out of 20 if we did the same survey with different random groups of people.

Ethnic groups and reliability

The larger the number of people in an ethnic group, the more reliable the data.

Estimates for the white British ethnic group tend to be the most reliable as they’re usually based on larger numbers of people. At the time of the 2021 Census, 74.4% of people in England and Wales identified as white British.

Estimates for ethnic minority groups (including white minorities) tend to be less reliable as they’re usually based on smaller numbers of people.

Estimates based on a small number of people are more likely to change from year to year – what can look like changes over time might not reflect real differences.

Where the data is available, we include the number of people from each ethnic group in our tables or data downloads. This shows how many people each estimate is based on (for survey data) or the actual number of people (for administrative data).

Find out how we use confidence intervals to determine the reliability of survey data.

Problems with making generalisations

It is important to consider the risks in making generalisations about large groups of people. People within any one ethnic group are likely to have very different experiences from each other.

Generalisations are harder to make when ethnic groups are combined.

Example:

The aggregated ‘black’ ethnic group might include a range of people, from black people born in Britain to British parents, to recent migrants. The aggregated ‘white’ ethnic group usually includes people from white British backgrounds as well as white minorities such as Gypsy, Roma and Irish Traveller groups.

Making better generalisations

To help make generalisations more accurate we use:

  • weighting to make survey data more representative
  • administrative data (where it's available) based on real numbers rather than survey estimates

We also use confidence intervals to determine the reliability of data.

We use additional factors like age, gender and area to analyse differences within and between ethnic groups, where data is available and reliable.

We ‘suppress’ data (make it unavailable) if:

  • the figures are too unreliable to use
  • it is possible to identify individual respondents