Scottish Employer Skills Survey 2020: Technical Report

Weighting

Survey data were weighted and grossed up to the total population of establishments and total population of employees, according to the 2020 Inter-Departmental Business Register (IDBR) ‒ the latest available business population statistics published by ONS at the time that weighting was carried out.

Weights were created in pairs: a 'unit-based' weight and an 'employment-based' weight. The unit-based weight was designed for analyses by the number or proportion of establishments; the employment-based weight was designed for use when analysing by number or proportion of employees (including volume measures of vacancies, skills gaps and numbers trained). Data dictionary files were created listing each variable with notes and guidance on the correct weight to use.

The following weighting strategy was used:

• Grossing weights were applied on a 13 broad Standard Industrial Classification (SIC) sector and six sizeband grid (i.e. 78 cells). The employment sizebands used were: 2-4, 5-9, 10-24, 25-49, 50-99, 100+.^[7]
• Overlaying these grids, RIM (random iterative method) weights were imposed for according to a combination of Regional Outcome Agreement (ROA) region and local authority (LA) in Scotland. It was not possible to group solely by ROA region for weighting purposes due to overlapping ROAs for some postcodes (i.e. it was possible for a sample record to have more than one ROA):
  • 'Glasgow' ROA was disaggregated into its component local authorities of East Dunbartonshire, East Renfrewshire and Glasgow City.
  • 'Lanarkshire' ROA was disaggregated into its component local authorities of East Dunbartonshire, North Lanarkshire and South Lanarkshire.
  • 'West' ROA was disaggregated into its component local authorities of East Renfrewshire, Inverclyde, Renfrewshire and West Dunbartonshire.
• The regional RIM weights ensured the survey population of each local geography matched the employer population without further correction for size and sector at this level. The RIM weights were calculated by the data processing software which used this method to find a 'best fit' between the data and the local level targets.
• Cell merging was applied in instances where no interviews had been conducted in cells where the IDBR indicated that establishments existed, and, conversely, ones where interviews had been carried out in cells with a reported 'zero population' according to the IDBR. Cell merging was also conducted in instances where a very small number of interviews had been conducted in that cell (typically in the 250+ employment band). In each of these instances, cells were merged within broad SIC sector (i.e. merging sizebands).^[8]