Unemployment Model Based Estimates (level and rate)
(showing types of area available in these data)
This column appears in 2 spreadsheet views.
|Time series spreadsheet
(ie all possible values for Reference Period dimension)
|Cross section spreadsheet
(ie all possible values for measure type dimension)
Note: These may be large files.
The data have been sorted into coloured groups, each of a similar size. Hover over any area for details.
The model-based estimate produces a single-best estimate for total unemployment for the local area.
The Annual Population Survey (APS), on which these estimates are based is published quarterly, but with each publication including data for one year. The four publications per year cover the periods January to December; April to March of the following year; July to June of the following year; and October to September of the following year. Three quarters of the survey observations included in one publication will therefore be included in the following publication. The model-based unemployment estimates (and the direct APS survey estimates) represent valid estimates for the period to which they refer but because, as explained above, estimates from consecutive quarters are based on common observations they are not independent, and should not be used for comparisons over time.
Comparisons over time for say the latest annual estimate should only be made using estimates for the same period a year earlier, as the data for these two periods will not include any common observations. Estimates of economic activity status are usually taken from surveys of individuals (nationally from the Labour Force Survey, and from the annual Labour Force Survey or Annual Population Survey for smaller areas). However for most local areas, even the Annual Population Survey does not have a sufficiently large sample to provide precise estimates of unemployment. For this reason a statistical model has been developed to improve the annual LFS/APS estimates of unemployment, for small areas, by using supplementary information, mainly the numbers of claimants of Jobseekers' Allowance (the claimant count). The claimant count is not a measure of unemployment as not all unemployed people are eligible for the benefit (and some legitimate claimants would be considered to be employed under the International Labour Organization definition of employment). However, the claimant count does have a high positive correlation with the number of unemployed people in an area, and it is an administrative count so it is known without sampling error.
The model considers unemployment from the annual LFS/APS and claimant count for six age and sex groups, and also includes a socio-economic area indicator and a random area effect. The relationship between claimant count and the number of unemployed may be different in two areas in spite of them sharing the same factors in the model and this last term is included in order to model these random local differences. The inclusion of the random effect term gives the model-based estimates the property that, for sufficiently large samples, they will coincide with the direct survey estimates.
This slice of multidimensional data is not a Linked Data resource in the database: it's a virtual resource (i.e. you can't query it by SPARQL). But does have a permanent unique URL which can be bookmarked.
A linked data-orientated view of dimensions and values
95% Lower Confidence Limit
|(not locked to a value)