Aggregate data are derived from micro-level survey datasets by applying statistical methods such as summing or averaging, then applying additional calculations such as weighting and estimation of sampling error. These procedures are designed to provide reliable inferences about an entire population based on data collected from the set of samples surveyed.
Unemployment is one of the most closely watched measures of the health or otherwise of a nation's economy. But how is it calculated?
The standard measure of unemployment is the unemployment rate which measures the number people who are unemployed as a proportion of the number of people who are economically active. This unemployment rate is derived from survey data. In the UK, the Office for National Statistics uses the Labour Force Survey as a basis for calculating the unemployment rate. The Labour Force Survey is conducted every 3 months and interviews around 101,000 people over the age of 16. Each respondent is classified as either in employment, unemployed or economically inactive. The unemployment rate is then derived from these responses.
So the unemployment rate is an example of how an aggregate indictor is produced from a survey. In fact, it is a legal requirement for every country in the EU to produce a Labour Force Survey, and Eurostat and the OECD then use these surveys to publish a monthly rate of unemployment for each of their member countries. However, basing unemployment rate on a survey raises several methodological issues. Firstly, although the Labour Force Survey is covers a large, representative sample, not 100% of the over-16 population are included in the survey. This means, a different sample would produce a slightly different result. The spread of results from different samples is known as the sampling variability. The sample variability places a confidence interval in the final unemployment rate produced by the survey. Secondly, as it is based on a survey the unemployment rate also needs to be adjusted to reflect the population as a whole, perhaps by adjusting the results for a particular local area by age group and gender. For example, the 2001 UK census revealed that the UK population was 1 million lower than had been previously estimated. This meant a new set of adjustment factors based on the revised population figures had to be applied to all the Labour Force Surveys from 1992 onwards.
A second measure of unemployment is claimant count. This is a headcount of everyone claiming Job Seekers Allowance and National Insurance credits. Since it is a 100% count, the claimant count is unaffected by sampling variability and so can be used as an indicator of those without work at very small levels of geography.
However, as claimant count only measures people claiming benefit, the unemployment rate is considered to be a more accurate measure of actual unemployment and is the standard indicator. In fact, it is a legal requirement for every country in the EU to produce a Labour Force Survey, and Eurostat and the OECD then use these surveys to publish a monthly rate of unemployment for each of their member countries.