• cymraeg

• UK Snapshot
• Neighbourhood
• Economy
• Census
• About ONS
• Jobs

Many labour market statistics are derived from survey data. Sample surveys, by definition, only contact a fraction, or sample, of the 'population' they aim to cover. The results estimate features of the entire population on the basis of information collected from this sample. When a survey includes the entire population, it is called a census.

Weighting allows us to estimate population totals. It also compensates for different levels of non-response in different subgroups of the population and reduces the random variability in our estimates.

Surveys can cover the 'population' of people, such as the Labour Force Survey (LFS), or businesses, such as the Annual Business Inquiry (ABI). The LFS, for example, is a continuous household survey, that collects information from a random sample of over 100,000 adults every quarter. The results from the LFS are weighted up to UK population totals. Business survey data are weighted to the total number of businesses or the number of people employed in those businesses.

For example, each respondent in the LFS is given a weight which can be thought of as the number of people that he or she represents in the population. If there were no non-response, each person in the population would have an equal probability of being in the 'achieved' LFS sample, about 1 in 500, and the weight would simply be the inverse of this probability.

However, because of differential non-response, some people are less likely to be in the achieved sample than others. For example, young males in urban areas are more difficult to contact and less likely to agree to take part in surveys. The LFS respondents are divided into groups defined by age, sex and where they live. People from low-response groups are given larger weights. Further information on LFS weighting is given in Volume 1 of the LFS User Guide.

Following the 2001 Population Census, the aggregate LFS series were further adjusted using an interim weighting method. This adjusts the LFS totals to take account of changes in population totals arising from the Census.

Business survey weights comprise two factors: the design weight and the calibration weight. The samples for business surveys are generally divided into groups of businesses with similar characteristics. For the ABI employment data, for example, these subgroups are defined by industry sector and the number of people the businesses employ. Details of the ABI weighting may be found in 'The launch of the Annual Business Inquiry' in Labour Market Trends, May 2001.

Design weights are calculated for each group and are approximately the total number of businesses in a group divided by the number of businesses in that group in the sample. The calibration weight is a correction factor that rescales the estimates to more accurately represent the businesses in the whole group. The Annual Survey of Earnings and Hours (ASHE), uniquely, uses LFS estimates of employee totals in this calibration step.