The UK Faculty of Public Health has recently taken ownership of the Health Knowledge resource. This new, advert-free website is still under development and there may be some issues accessing content. Additionally, the content has not been audited or verified by the Faculty of Public Health as part of an ongoing quality assurance process and as such certain material included maybe out of date. If you have any concerns regarding content you should seek to independently verify this.

Methods of population estimation and projections

Introduction

Learning objectives: You will learn about methods of population estimation and population projections.

Please now read the resource text below.

Resource text

Population estimates

Description

Whilst the census is crucial for resource allocation and planning, because it is carried out only every ten years, other methods are required for planning in the intervening years. Population estimates use the census as a baseline, adding births and subtracting deaths and make allowances for migration. They can be used for national and local planning. Population estimates are produced annually. In the UK, the Office for National Statistics (ONS) estimates are also used as the basis for capitation-based funding of local authorities and Primary Care Trusts. Under-estimation can therefore have effects on local services, and over-estimation can lead to unfair resource distribution.

In the UK, because of the accession of ten new states to the European Union in 2004, there has been a rapid increase in migration into the UK, especially when comparing 2005/2006 with 2003. International migration is now the main driver of population change in the UK (in 2005, the ONS estimated a net inflow of 185,000 people) [1].

Strengths

  • more up to date than census.

Weaknesses

  • questionable accuracy with regard to internal migration.
  • data on migration are difficult to obtain, particularly at local level. Some migrant workers may only stay a few months and it is hard to measure change.

Population projections

Population projections aim to estimate characteristics and sizes of populations in the future. They make assumptions on:

  • mortality (not too difficult to estimate for today's 25 - 45 year olds)
  • fertility (much harder to estimate; often, a range of high, low and medium fertility levels are used in the projections, which are then published as a series)
  • migration (easy to misjudge as has happened in recent years)

In the UK, the Government Actuary's Department produces 'full' population projections by age and sex for the United Kingdom and constituent countries every two years [2]. Special projections can be undertaken in the intervening years, such as estimates by marital status.

ONS uses the national figures to produce sub-national population projections. Data are available, 2 or 3 times in each decade, on the GAD website by age and sex for Government Office Regions, counties, county districts, unitary authorities, and London boroughs [3].

Household projections are undertaken by the Department for Communities and Local Government (CLG).

For London, the Greater London Authority (GLA) creates population projections, which are updated on a yearly basis, both at population and at household levels. In addition to the frequency, GLA projections can differ from ONS versions because they adjust estimates at borough level when ONS mid-year estimates have been found not to be robust.

Example

In addition to the national and regional projections, the GLA calculates London projections by ethnicity and at ward level.

Uses

  • longer term planning e.g. demand for services for care of elderly in 20 years time.

Weaknesses

  • lose accuracy the further ahead the planning is attempted.

Capture - Recapture techniques

It often happens that it is not the total population that is of concern but some fraction of unknown size. For example, we may wish to know the numbers of problematic drug users in order to plan service levels, or the numbers of diabetics in an area to estimate the completeness of a register, or the frail elderly who can stay in their own homes with appropriate community support. In these cases we can draw on techniques developed in the field of ecology.

In field studies a sample of the population is taken (say S1), and 'marked', then released. After enough time has elapsed for the sample to mingle with the rest of the population, a second sample is taken (say S2), and the numbers previously marked (say M) and unmarked (say U) are noted. S2 = M+U. The total population may then be estimated as:

S1 *S2/M or S1*(M+U)/M.

There are two assumptions here that must hold:

(a) the population must be closed, that is to say of fixed size between the two samples, and

(b) the two samples must be genuinely independent.

This estimate can be made on the basis of two samples only. There is a significant body of literature discussing statistical tools appropriate to larger numbers of samples.

In epidemiology, sometimes separate datasets held concurrently are substituted for repeat samplings. However, someone known to one dataset may be more likely to be known to another than a genuinely random member of the population, and this can be a source of significant bias in estimates, and sophisticated statistical techniques are required to compensate for this.

Catchment populations - the multiplier method

We may wish to estimate the catchment population served by a particular facility. Assume we have data on the total number of people using the facility, and survey data that tells us that a set fraction comes from an area whose total population (P) is known. From this we can calculate the proportion of P who use the facility, and on this basis calculate a notional overall catchment.

Worked example: a fitness centre serves 20,000 people in a season. A survey has shown that approximately half its users live in the local district, half from outside. The population of the local district was 92,000 at the last census. Then the notional total catchment can be estimated as (92,000 / 0.5) = 184,000.

This method can also be used epidemiologically, for example to estimate injecting drug use, where appropriate data is available.

Related links