Why do the census counts differ from the Council tax register?
There are a number of reasons why a Census could identify a different number of household spaces from those identified on a different source collected for other purposes. For example:

• Timeliness of the data: One measure of the quality of any data source is how up-to-date it is. In particular, how quickly are demolished / derelict properties removed from the Council Tax database and how quickly are new or converted properties added? Many data sources are prone to inflation because 'old' records are not removed.
• Definitional differences: Any two data sources will almost always provide different numbers because of definitional differences. In this case the major sources of definitional difference are likely to be multiply-occupied properties and communal establishments. These could appear on one list as a number of single person households and on another as a single dwelling containing multiple people. This would lead to differences in the household count but not the person count.
  
  In an area where Council Tax is low, and where it is feasible that a large number of people have family homes elsewhere, it could presumably be financially beneficial for some people to declare their room in shared accommodation as a single person household and their family home elsewhere as a second home. However, for Census purposes these people would almost certainly be recorded at their family home and it is possible that the shared accommodation would appear as one household on the Census database.
• Duplication: Another common difficulty with data sources is duplication. Special procedures were used on the Census to search for and remove duplicate households. Council Tax records have been used by a number of areas within the ONS and, in general, problems have been found with double counting households.

Why do the census counts differ from the Electoral roll?
There are definitional differences between the electoral roll and the Census:

• Students may be registered at both their term-time and vacation addresses, although they may only vote at one of these in any election. The published Census figures only count students at their term-time address.
• People with two homes may be registered at both addresses, although again they may only vote at one of them. The Census only counts people at the address which they consider their usual residence.
• British citizens who have left the country within the last 15 years may still register to vote in the UK. As they are no longer resident in the UK, they will not be included in the Census figures.
• Members of the British armed forces and their families are registered to vote in the UK, even when stationed overseas.
• Citizens of countries outside the EU and Commonwealth are not eligible to vote in the UK. These persons are included in the published Census figures if they are resident in the UK.
• A number of other groups are not eligible to vote, in particular most prisoners. Such groups are included in the Census figures.

In addition, residents who have died or moved out of the area may not immediately be removed from the register. How quickly this happens will depend on the electoral registration officer in question. Due to this, and the definitional differences above which cause a difference in the population base between the Census and electoral roll statistics, this administrative source was not used in the Census quality assurance process.

Why do the census counts differ from the Patient register?
NHS patient registers are prone to list inflation, caused by delays in people who have died or moved out of a Health Authority being removed from the register, duplicate entries, and people having more than one NHS number. The degree of list inflation is extremely variable geographically - for instance, it tends to be higher in areas with large numbers of students, who tend to be slow in re-registering with a GP when they leave higher education. In addition certain groups, such as Armed Forces personnel, are excluded from patient registers.

For England & Wales as a whole, total GP patient registrations in 2001 were 5% higher than the 2001 Census figure for the total population. However the differences as a percentage of the patient registration data vary widely by local authority, from 23% lower to 19% higher.

Why do the census counts differ from the Child benefit data?
The Benefits Agency administer the Child Benefit Claimant Register which holds information on all persons claiming child benefit in the UK and the children for whom the benefit is claimed. For the purposes of comparator data, ONS liaised with DWP and Oxford University to obtain 'clean' child benefit data relating to August 2000 (some postcodes were originally missing or needed correcting, and some new born children did not appear for up to three months after they were born). The Census counts those children normally resident in the UK, some of whom are not eligible for child benefit (for example children of foreign armed forces). Differences between the number of children claiming benefit and the Census figure could be explained by poor address data or this difference in definitional base. Also, there can often be a lag in updating records when a claimant moves, particularly now that payments are made directly into bank accounts. There are also some issues regarding benefit fraud, although this is difficult to quantify.

For England and Wales as a whole, the Census figure is 0.9% higher than the number of children claiming child benefit. However the differences as a percentage of the child benefit data vary widely by local authority, from 7% lower to 51% higher.

Why do the census counts differ from the Pensions data?
The Benefits Agency administers the Retirement Pension Register that holds information on all persons claiming a state pension in the UK. Almost all persons aged 65 or over are entitled to claim some form of state retirement pension. For the purposes of comparator data, ONS liaised with DSS and Oxford University to obtain 'clean' retirement pension data for May 2000 (some postcodes were originally missing or incorrect and some records were duplicated). Differences between those claiming pensions at a national level occur because not everyone is eligible. In addition, some expatriates resident overseas are eligible to claim a state pension. At a local level, differences could be explained by poor address data. Also, there can often be a lag in updating records when a claimant moves, particularly now that payments are made directly into bank accounts.

The Census figure for pensioners in England and Wales was 1.3% higher than the corresponding pensions data. The difference between Census figures and pension data as a percentage of the pensions data for local authorities ranged from 3% lower to 30% higher.

Why do the census counts differ from the Schools Census?
The School Census is an annual count of all children attending educational establishments, including schools which are privately funded. In England, information relating to January 1st is collected from local authorities by DfES each year, while in Wales, Scotland and Northern Ireland this role is carried out by the appropriate devolved government authority and has a different reference date. The main problem with this data is that it relates to place of study, which may be in a difference local authority to where the child lives. For this reason, less weight was given to these figures in the quality assurance procedure.

For England and Wales, the Census figure was 0.5% lower than the Schools Census. Differences at local authority level as a percentage of the school census data ranged from the Census being 27% lower to 43% higher. In one exceptional case, the Census figure was 72% below the Schools Census due to large numbers of pupils living outside the area and attending schools inside.

Why do the census counts differ from the Birth registration data?
The civil registration system records all new births in the population, and this was therefore a key source of data when quality assuring ONC estimates for children aged under one year. The numbers will differ where children have died, or moved in or out of the area, before their first birthday.

The Census figure for under 1s in England & Wales is 1.8% lower than the birth registration data. By local authority, the Census figure ranges from 20% below the birth registrations figure to 22% above. Many of the areas where the Census was furthest below the registration data were areas with high concentrations of ethnic minorities or asylum seekers, suggesting that the 'missing' babies were no longer in the country.

This page last updated 01 November 2005