The purpose of this document is to provide background information on the process that underpins census outputs. Through detailing these processes, users will gain an understanding of the underlying quality of the census outputs.
Statistical processing methodology
The returns from the 2011 Census in Northern Ireland will be subjected to statistical processing with the aim of producing high quality census outputs.
The key elements of the statistical processing will involve:
Edit and imputation
This acknowledges that census returns are sometimes incomplete (that is, respondents do not answer every question) and sometimes inconsistent (for example, in the 2001 Census, some respondents recorded parent-child relationships the 'wrong-way-round', recording the child as the parent). The purpose of edit and imputation processes is the production of a dataset that is internally consistent, and has no relevant missing values for any returned questionnaire.
Coverage assessment and adjustment
This acknowledges that some people will not have been included in a census return, and some people may have been included more than once (for example, the child of a separated couple). This has been addressed primarily through a Census Coverage Survey in which about 14,000 households were visited by trained survey interviewers who conducted a detailed survey that focussed on determining the number of people living at each household on Census Day. Comparison of the responses to the Census and to the Census Coverage Survey have informed adjustments to the census counts that lead to census outputs that are best estimates of the true population, taking account of any people missed or double-counted by the Census.
Statistical Disclosure Control
This acknowledges that while most census outputs take the form of statistical counts, there is a risk that information about an individual person could be deduced from census outputs. For example, if everybody in a particular geographic area was aged under 50 apart from one person of pension age living in a single person household, a cross-tabulation of age and general health would reveal the response of that person to the census question on general health. The census questionnaire gives respondents an assurance that their information will be treated as confidential and statistical disclosure techniques are employed to ensure that the risk of inadvertent disclosure in statistical outputs is minimised.
Specific detail on Statistical Disclosure Control can be found on the Office for National Statistics (ONS) website, describing the principles being applied throughout the UK. The application in Northern Ireland will be similar, except that Local Government Districts (LGDs) will take the place of middle layer Super Output Areas (SOAs) in England and Wales. Similar targeting variables are being used across the UK except, in Northern Ireland, religion is used in place of ethnicity which is used in the rest of Great Britain.
In order to ensure that census outputs across the UK are comparable, similar processes are being applied to each of the Censuses in the UK. ONS is taking the lead in developing the statistical methodology, and accordingly further detail on the statistical methodology used for the 2011 Census in Northern Ireland can be found on the ONS website.
The methodologies described on the ONS website will be applied in Northern Ireland, subject to necessary, but minimal, changes that reflect differences in Northern Ireland. A number of Northern Ireland specific issues are addressed in the additional links below. For example, the geographic base for Northern Ireland census outputs is obviously specific to Northern Ireland and a paper outlining the proposed approach is included below.
‘Geographic units for outputs from the 2011 Census - An Information Paper’ discusses the geographic basis on which statistical aggregates from the Census are based.