Scottish Employer Perspectives Survey 2019: technical report

Sample design

This chapter outlines the sample design adopted for the Scottish EPS, detailing the method by which employers were selected to take part in the research, including the setting of quota targets and sample volumes.

Sampling population and survey sampling unit

The sampling population used for the Scottish EPS encompassed establishments across the full geographical spread of Scotland, in all sectors of the economy (across the commercial, public and charitable spheres). All Scotland establishments with two or more people working at them (including partners and working proprietors) were eligible for the survey – i.e. establishments with a single person on the payroll were excluded. The 2018 Inter-Departmental Business Register (IDBR) was used as the universe data defining the population of employers in Scotland.

Establishments were used as the sampling unit for the survey, as opposed to an organisation-based approach. The term "establishments" denotes specific individual sites or premises; thus if an organisation had several sites it was possible that more than one of these sites were in the sample. This approach was chosen because it is at the establishment level where respondents are most likely to be able to provide a detailed and accurate picture of how employers go about meeting their skills needs. Decisions around training staff are often based on the training offer available in the local area, while recruitment also tends to occur at a more localised level. This approach has been used in the EPS series at a UK level, and its sister survey, the Employer Skills Survey (ESS), since each series began.

Since 2012, both the UK EPS and ESS surveys have considered establishments with two or more people working at their site in scope. The rationale for this is based on both practical and conceptual considerations, discussed below.

From a conceptual viewpoint, the focus of the skills surveys is on the workforce, and as such any establishment covered logically needs to have staff (or the desire / potential to employ staff in the future). In this Scotland specific iteration of the survey, the survey covered employers' approaches to recruiting (particularly young people and education leavers), their engagement with apprenticeships and their facilitation of training staff to Vocational Qualifications. Findings from the survey can then be used to enhance employers' engagement with, and experience of, various recruitment and people development initiatives. Accordingly, the survey was aimed at those for whom such initiatives carry relevance (i.e. establishments with at least one employee).

On a more practical level, however, it tends to be much easier for survey respondents to think in terms of the overall 'headcount' for their site – including both working proprietors and employees – than to split out the two groups (particularly when the lines between the two are not clear-cut). For example, it is typically easier for employers to answer about recruitment channels for all managers / staff, rather than 'only those managers who are not working proprietors'.

The survey also excluded the self-employed (with no employees), as the question approach / context for this group would need to be somewhat different, since they are by definition not "employers". Additionally, there is an absence of robust population figures for this group, providing obstacles for robust and representative sampling and weighting.

Sampling approach and setting quotas

The sampling strategy adopted for the Scottish EPS was based on the approach taken for the most recent UK EPS in 2016.

The sampling approach taken for Scottish EPS was as follows:

• Target interviews were stratified against a two-dimensional sector by size grid (12 sectors and seven sizebands) on an interlocking basis;
• The initial allocation of interviews was done according to employer sizebands, using a set of ratios that deliberately over-sampled larger employers;
• Interviews were then allocated to sector within each sizeband in proportion to their representation within the business population; and
• The starting sample was then drawn from the commercial data supplier, Market Location.

Keeping the approach consistent with UK EPS 2016 preserves the ability to carry out time series analyses and maintain comparability with other nations of the UK should they commission their own versions of EPS.

Using business population figures from the IDBR, the sample was stratified by establishment size and industrial sector on an interlocking basis in order to ensure robust coverage of the full range of business types in all parts of Scotland.

Quotas by size band

Seven employment size bands were used: 2-4, 5-9, 10-24, 25-49, 50-99, 100-249 and 250+.

A purposive approach was taken to sampling by size band, as shown in Table 1, in order to maximise the yield of interviews among larger establishments while seeking to limit the level of skew away from the true population.

This ensured that findings would be representative of large establishments which employ a large proportion of the workforce, whilst balancing this with a need to ensure that the level of weighting needed across the smaller size bands did not impact the robustness of the data.

To have allocated interviews by size band in proportion to the true population of establishments would have produced an interview profile dominated by smaller establishments, with a very small proportion of large establishments.

Table 1: Interview targets by size band

Size Band	Scottish population*	Proportional distribution	Targeted interviews	% of total interview target
2-4	76,000	50%	714	27%
5-9	34,300	23%	610	23%
10-24	24,800	16%	583	22%
25-49	8,500	6%	345	13%
50-99	4,200	3%	159	6%
100-249	2,300	2%	159	6%
250+	1,000	1%	80	3%

*Source: Inter-Departmental Business Register, 2018. Rounded to the nearest 100.

Quotas by sector

After setting interview targets by size band, interviews were then allocated to sectors within each size band in proportion to their representation within the business population, as shown in Table 2.

The only exception to the setting of sector quotas in proportion to the business population was the Financial Services sector. The quota target for the Financial Services sector was boosted from 56 to 100 (a similar number of interviews achieved for Scotland in UK EPS 2016), due to the sector's relative importance to Scotland's economy. To apply this boost, the additional interviews were distributed in line with the size population of the Financial Services sector (e.g. establishments employing 2-4 staff accounted for 49% of the Financial Services population according to the IDBR, hence 49% of the additional Financial Services interviews were assigned to this size band). To preserve the overall size targets these additional interviews were subtracted from the interview targets originally set for Business Services given its similarity to the Financial Services sector (i.e. proximity in SIC codes) and given that it is one of the largest sectors therefore having fewer interviews would have minimal impact on its confidence interval.

Table 2: Final interview targets by sector within size band

	2-4	5-9	10-24	25-49	50-99	100-249	250+	Total
Primary Sector & Utilities	58	28	17	7	4	5	4	123
Manufacturing	29	26	31	21	14	17	9	147
Construction	77	46	35	19	9	8	4	198
Wholesale & Retail	117	154	130	56	21	23	9	510
Hotels & Restaurants	58	79	91	49	13	9	2	301
Transport, Storage & Comms	50	30	31	23	13	13	7	167
Financial Services	34	26	18	9	3	4	6	100
Business Services	161	87	76	40	21	23	10	418
Public Administration	8	10	17	16	9	13	10	83
Education	9	13	28	41	17	17	6	131
Health & Social Work	35	52	72	45	28	21	11	264
Arts & Other Services	78	59	37	19	7	6	2	208
Total	714	610	583	345	159	159	80	2,650

Quotas by region

Interviews were allowed to fall out 'naturally' by geography. For the analysis and reporting Regional Outcome Agreement (ROA) areas have been used (how the ROAs are defined in terms of Local Authority is presented in Appendix E). The expected distribution by ROA is shown in Table 3 (the 'expected 'fall out' of interviews' column, these numbers calculated by multiplying the ROA distribution within the population by the target overall sample size). This expected distribution was felt to provide relatively robust base sizes for regional analyses. The final column shows the actual (unweighted) distribution of interviews by ROA, which is reasonably close to the expected distribution.

Table 3: Expected (and actual) interview distribution by region^[2]

Region	Scottish population*	Proportional distribution	Expected 'fall out' of interviews	Actual achieved interviews (unweighted)
Aberdeen & Aberdeenshire	17,200	11%	297	313
Ayrshire	9,200	6%	159	167
Borders	4,300	3%	72	103
Dumfries & Galloway	5,900	4%	95	111
Edinburgh & Lothians	19,800	13%	360	334
Fife	8,700	6%	154	150
Forth Valley	7,500	5%	133	137
Glasgow	21,000	14%	382	275
Highlands & Islands	19,300	13%	323	432
Lanarkshire	16,800	11%	299	245
Tayside	12,100	8%	209	233
West	9,400	6%	170	143
West Lothian	4,100	3%	72	61

*Source: Inter-Departmental Business Register, 2018. Rounded to the nearest 100.

Sample source

Market Location – a commercial data supplier − was used as the sample source for the Scottish EPS. Market Location was the main sample source for the most recent UK-wide EPS and ESS surveys. Sample was ordered from Market Location at an average sample-to-target ratio of 8:1 against each target cell. Due to the availability of sample (i.e. in certain cells it was not possible to order at an 8:1 ratio) this varied between quota cells; from 4:1 in Constrution 250+ to 11:1 in Financial Services 50-99.

A total of 21,500 records ordered from Market location were loaded for fieldwork.

All sample records were postcode-validated to ensure that geographical regions had been correctly assigned.

Checks were also undertaken in instances where duplicate telephone numbers existed within the sample. In certain sectors, such as retail and finance, it is common for different establishments to appear under the same centralised telephone number. Such establishments were marked up on the sample – with the address of the sampled establishment displayed on-screen – so that interviewers would be aware that the telephone number they were calling was a centralised switchboard and thus they would need to request to be transferred to a particular site.