Communities and Local Government

• Communities and neighbourhoods
• Indices of deprivation 2004 - Population Denominators

Communities and neighbourhoods

Indices of deprivation 2004 - Population Denominators

For a considerable number of indicators in the ID2004, estimates of 'at-risk' population ¹ were constructed using a combination of the 2001 Census and the 2001 Mid-Year Estimates (MYEs) published by the Office for National Statistics (ONS).

Defining 'at-risk' population

The population estimates employed as denominators in a considerable number of ID 2004 indicators included resident population and communal establishment population but excluded prison population. Prisoners ² were not included as they are not at risk of many forms of deprivation captured in the ID2004. Other types of communal establishment population (e.g. students; persons in care establishments; children in local authority homes) are at risk of experiencing these forms of deprivation (age/sex restrictions allowing) and so are included in the denominator.

The final population estimates can be summarised as:

[1] a_ij = r_ij + c_ij - p_ij

where: a represents the 'at-risk' population in area i at time j

r represents the resident population in area i at time j

c represents the communal establishment population in area i at time j

p represents the prison population in area i at time j

Geography and Time Point

The ID 2004 have been produced at the level of 2001 Census 'Super Output Area' for the whole of England. Super Output Areas (SOAs) are a continuous geography of areas of approximately equal population size (the mean SOA population in England according to the 2001 Census was 1500). This geography was overwhelmingly supported by the consultation exercise. As the majority of indicators employed within the ID2004 use numerator data for 2001, the denominators required to produce the indicators were also constructed for 2001.

Data

The 2001 Census contains a wealth of demographic and social information on the population of the UK, as at Census night on 29^th April 2001. It is the most comprehensive survey undertaken, with every individual person being legally required to register at the address they occupied on 29^th April 2001.

The 2001 Census contains data at a variety of spatial scales. The smallest scale, and the building block from which all other higher level geographies are formed, is Output Area (OA) level. OAs contain a mean population of 300 persons in England. OAs aggregate to both SOA level and statistical ward level; and both SOAs and wards aggregate to local and unitary authority level. Although the majority of SOAs nest within wards, a number do not adhere to this norm.

In addition to providing details of the total population living in each OA, the 2001 Census also provides a detailed breakdown of population subgroups, often by age and sex. For the construction of the ID 2004 denominators it was important to have a measure of prison population at small area level that could be subtracted from the total population to create the 'at-risk' population. The Census includes this information relating to prison populations, although a degree of data recalibration was required to produce the necessary estimates of prison population by age and sex.

The 2001 Census website www.statistics.gov.uk/census2001 has links to a substantial literature explaining the different geographies and variables within the 2001 Census.

In the intercensal years, ONS publish annual Mid Year Estimates (MYEs) broken down by age and sex for each local and unitary authority in England and Wales. These estimates are derived by 'ageing' the previous Census estimates, adding in births, subtracting deaths, and adjusting for migration. These estimates are the 'gold standard'. While MYEs have the benefit over the Census of being annually updated, they lack the desired level of spatial disaggregation (lowest level being local and unitary authority level) and in the detail which they contain (simple population counts by age and sex).

It is now widely recognised that the 2001 Census suffered from a population undercount in particular demographic groups and in particular geographical areas. ONS subsequently recalibrated the Census results and published revised figures for Local and Unitary Authority areas as the 2001 MYEs. It was agreed that the 2001 MYEs should be regarded as the most reliable estimate of population at local and unitary authority level for the purposes of the ID2004, and so these recalibrated estimates were used.

Methodology

Although the final ID 2004 population estimates were required at SOA level, the construction procedure was undertaken at OA level and then aggregated to SOA level once all the steps were complete. This approach was necessary as, at the time of producing the ID 2004, data were only available at OA or ward level - not SOA level.

The methodology employed in the construction of ID 2004 population estimates involved calculating an estimate of the total 2001 Census population by age and sex at OA level; subtracting estimated prison population by age and sex at OA level; constraining the resulting OA level age/sex estimates to the 2001 MYEs; and finally aggregating from OA to SOA level. The steps involved are as follows:

1. Obtain total [resident + communal establishment] population by age and sex for each OA in England from the 2001 Census

OA level counts of 2001 Census 'resident' population and 'communal establishment' population - by sex and single or 5 year age groups - were extracted from SASPAC (provided by the Greater London Authority).

2. Calculate estimate of prison population by age and sex for each OA in England from the 2001 Census

OA level counts of 2001 Census 'prison establishment' population - by sex - were extracted from SASPAC. Ward level counts of 2001 Census 'prison establishment' population - by sex and single or banded age groups - were also extracted from SASPAC.

The ward level age/sex distribution of prison population was expanded to form single or 5 year age groups, based on the assumption of equal probability of age within the broader age groups. For example, if a ward contained 9 male prisoners in the broad age group 20-34, the assumption of equal probability resulted in 3 prisoners being allocated to the 20-24 age group, 3 prisoners to the 25-29 age group, and 3 prisoners to the 30-34 age group.

This expanded age/sex distribution at ward level was then applied to the OA level sex distribution of prison population to give an estimated age/sex distribution by single or 5 year age groups at OA level.

Due to the assumptions of equal probability within the broad age groups, a small number of OAs were found to have estimated prison populations in certain age/sex groups that were larger than the total communal establishment population in those age groups. As the prison population is a constituent part of the wider communal establishment population, this disparity had to be readjusted. For the age/sex groups in those OAs that suffered from this inconsistency, the prison population was 'capped' at the total number of communal establishment population in the age/sex group in the OA. The assumption was therefore made that all communal establishment population in that particular age/sex group in that OA were prisoners.

The disparities between the expected prison population in each age/sex group and the capped prison population were then summed over the entire age distribution (separately by sex) to give a total number of prisoners who had not been allocated an age group. This 'floating' prison population was then re-distributed over the remaining OA level age/sex differential between prison population and communal establishment population (in practice this refinement affected the numbers only negligibly). This adjustment ensured that no age/sex group in any OA had a greater prison population than communal establishment population, thus preserving the internal integrity of the estimates.

3. Subtract estimate of prison population from total population by age and sex for each OA in England.

The final OA level age/sex estimates of prison population were subtracted from the previously extracted OA level counts of total [resident + communal establishment] population to give an initial unconstrained estimate of 'at-risk' population by age and sex.

4. Constrain resulting estimates of [resident + communal establishment - prison] population by age and sex to the 2001 MYEs

The expanded (i.e. single or 5 year age groups) ward level age/sex prison population counts were aggregated to local/unitary authority level. These prison population counts were then subtracted from the 2001 overall MYEs (as provided by ONS) to give an estimated count of 'at-risk' population at local/unitary authority level that excludes prisoners. This process is based on the assumption that the 2001 MYEs offer a more reliable estimate of total population at local/unitary authority level than the 2001 Census itself.

The OA level counts of 'at-risk' population were then constrained to the local/unitary authority level counts of 'at-risk' population to give the final set of OA level population estimates of 'at-risk' population by age and sex.

5. Aggregate from OA level to SOA level

The OA level population estimates were aggregated to SOA level to produce the complete set of ID2004 denominators.

Results

The resultant population estimates were rigorously checked and quality assured by the Team.

The majority of the indicators do use these population estimates as denominators, either using all ages (e.g. for all indicators in the Income Domain) or a sub-set of ages (e.g. for all indicators in the Employment Domain).

For most of the remaining indicators, denominators were drawn from the numerator's dataset (e.g. PLASC data for the three Key Stage indicators in the Education, Skills and Training Domain). When Census numerators were used, Census denominators were also used and these were not constrained to 2001 MYEs (this was also the case for Census households and dwellings).

This refers to the population 'at risk' of experiencing a given type of deprivation.

² Prisoners are defined as individuals serving 6 months or longer of a custodial sentence.

¹ This refers to the population 'at risk' of experiencing a given type of deprivation.

² Prisoners are defined as individuals serving 6 months or longer of a custodial sentence.