The International Passenger Survey (IPS) measures migration flows to and from the UK. It forms the basis of immigration, emigration and net migration estimates in the ONS Migration Statistics Quarterly Report. Separate estimates are produced for long-term (12 months or more) and short-term (1–3 months and 3–12 months) migration. To produce estimates at the local authority level ONS uses the published estimates of long-term immigration totals and then uses administrative data to distribute the total across local authorities. Similarly, for short-term immigration estimates, ONS uses published short-term immigration totals derived from the IPS data for non-British workers and students staying for at least 3 but less than 12 months (UN definition). Worker and student totals are then distributed by mapping the IPS definitions as closely as possible to the relevant data within the administrative sources provided by other government departments. In both cases, the national-level IPS estimates come with margins of error, and the local-level numbers are also estimates rather than precise figures.
IPS estimates are not sufficient to measure migration to Northern Ireland, because they do not account for movements across the Irish border. The Northern Ireland Statistics Agency (NISRA) supplements ONS quarterly migration figures with GP registrations data. One result of this is that short-term migration estimates are not available for Northern Ireland. Supplementary analysis of the data for Northern Ireland, which is also available through ONS, is provided by NISRA.
The Annual Population Survey (APS), which is used for the ONS dataset, Population of the UK by Country of Birth and Nationality, is a survey of households in the UK. The National Records of Scotland (NRS) also uses the APS to provide estimates specific to Scottish local authorities, in a similar publication called Population by Country of Birth and Nationality. APS estimates are not available for local authorities in Northern Ireland. The APS does not cover all migrants evenly. Estimates exclude people living in communal establishments – e.g., hotels and caravan parks – and may undercount subgroups of people with lower-than-average response rates to the survey, the recently arrived and short-term migrants. Response rates have declined over the past 20 years.
Margins of error in survey data
Caution must be used when examining estimates of small groups because of margins of error. Estimates represent what is considered to be the most likely figure, but the true number could be higher or lower. APS and IPS data are reported with 95% confidence intervals which indicate the margins of error. For example, the estimated non-UK born population residing in the UK in 2018 was 9,342,000, with a confidence interval of +/- 146,000. One effect of this is that it may not be sensible to make too much of the difference between two estimates – small differences may result from statistical noise rather than real changes in the population. Users should look at the trend over several years as in some cases the apparent year-on-year change may be misleading.
In some local areas, confidence intervals are quite wide relative to estimates of the population of interest. For example, the estimated non-EU born population living in Sunderland was 8,000 in 2017 and 2018, with a confidence interval of +/-4,000. The corresponding estimate for the year 2016 was 6,000. In other words, the data do not show with certainty that the non-EU born population of Sunderland changed between 2016 and 2018.