The Scottish Government’s Quality Assessment of the Office for National Statistics Labour Force Survey and Annual Population Survey data for Scotland

Annex 3: Detailed Quality Assessment Results for Scotland's Labour Market: People, Places, and Regions

The “Scotland’s Labour Market: People, Places and Regions” publication is sourced from the ONS Annual Population Survey (APS). This is an annual publication that provides labour market data breakdowns by local authority level as well as many of the more detailed Scottish breakdowns that are important to our users. This is a detailed publication that goes into more granular aspects of the labour market than is possible in our other publications, providing valuable insights for the employment status of the people of Scotland.

A complete list of all outputs included in this publication would be extensive, as many are cross tabulations or intersectional breakdowns. However, many quality issues are seen in the data at much higher levels than these detailed breakdowns. Therefore, this section is confined to headline Scotland estimates and breakdowns by age, sex, disability, ethnicity, industry, highest qualification, and inactivity reason.

Annex 3.1 Estimates by Local Authority

This subsection will focus primarily on employment estimates by local authority but will include summaries of economic inactivity within the sections on the current data quality rules and data confidence ratings.

Model-Based unemployment estimates are the main source used unemployment at local authority level. Given they are not derived exclusively using the APS, we have decided to not include them as part of this quality assessment. More information on the model-based unemployment estimates can be found in Concepts and Definitions.

Sample sizes (Employment)

When considering the change in sample size for employment in Scotland by local authority area between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -61.6%
• the largest decrease was for Argyll and Bute which fell by -76.2%
• the smallest decrease was for Fife which fell by -37.0%

In the Jan-Dec 2023 data, the three Scottish Local Authorities with the smallest achieved sample sizes were Orkney Islands, Shetland Islands and Na h-Eileanan Siar. Orkney Islands sample size was not large enough to be considered “robust”, and the Shetland Island’s sample size was borderline.

Figure A7: Since 2019, sample sizes for employed people aged 16 to 64 in Scotland have decreased in all local authority areas

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for employed 16 to 64 year olds by local authority area

Coefficients of Variation (Employment)

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by local authority area:

• all CVs have increased
• the average increase was 1.61
• the largest increase was for Orkney Islands which rose by 3.58 from 3.85 to 7.43
• the smallest increase was for the City of Edinburgh which rose by 0.30 from 2.01 to 2.31

Figure A8: Since 2019, coefficients of variation for estimates for employed people aged 16 to 64 in Scotland by Local Authority have increased

Boxplot of Annual Population Survey estimates of coefficients of variation for employed 16 to 64 year olds by local authority area, Jan-Dec 2019 and Jan-Dec 2023. More detail on understanding boxplots is provided in Concepts and Definitions.

Current Quality Assurance Rules

In Jan-Dec 2019, all employment estimates at local authority level were considered “robust” under current data quality rules. By Jan-Dec 2023, employment estimates remained “robust” with the exception of Orkney Islands which was considered “less robust”.

In Jan-Dec 2023, inactivity estimates by local authority also remain “robust” with the exception of Orkney Islands and Shetland Islands which were “not robust” and Na h-Eileanan Siar which is considered “less robust”. This is in contrast to 2019 when no estimates were “not robust”, and Orkney Islands and Shetland Islands were considered “less robust”.

Table A9: Summary of robustness for headline APS estimates by local authority in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Robustness	2019	2023
Employment	Robust	32	31
Employment	Less Robust	0	1
Inactivity	Robust	30	29
Inactivity	Less Robust	2	1
Inactivity	Not Robust	0	2

Data Confidence Ratings

From the Scottish APS estimates by Local Authority, five of the 32 Local Authority estimates for employment have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023. Out of the 32 Local Authority estimates for economically inactive, 23 Local Authorities decreased in confidence ratings between Jan-Dec 2019 and Jan-Dec 2023.

We have “high confidence” in the majority of Local Authority estimates for employment, indicating we can draw accurate conclusions from these estimates as they provide a good reflection of reality. However, we have “no confidence” and “low confidence” in five Local Authority estimates for economic inactivity and we therefore recommend that these estimates are not used as principal measures of labour market information for Scotland. Similarly, we have “limited confidence” in 20 Local Authority estimates for economic inactivity indicating these estimates provide a restricted view on reality and caution is required whilst using these estimates and that these estimates should be used alongside the suite of other indicators.

Table A10: Summary of confidence ratings for headline APS estimates by local authority in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Confidence Rating	2019	2023
Employment	High Confidence	31	26
Employment	Moderate Confidence	1	6
Inactivity	High Confidence	1	0
Inactivity	Moderate Confidence	27	7
Inactivity	Limited Confidence	2	20
Inactivity	Low Confidence	0	2
Inactivity	No Confidence	2	3

Annex 3.2 Estimates by Sex

Sample sizes

When considering the change in sample size for Scotland split by sex between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -59.6%
• the largest decrease was for “employed men” which fell by -61.6%
• the smallest decrease was for “economically inactive men” which fell by -54.4%

Figure A9: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators and for both sexes

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for headline labour market indicators by sex

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by sex:

• all CVs have increased
• the average increase was 1.46
• the largest increase was for “unemployed women” which rose by 3.75 from 6.13 to 9.89
• the smallest increase was for “employed people” which rose by 0.24 from 0.41 to 0.65

Figure A10: Since 2019, coefficients of variation for the labour market estimates in Scotland have increased for all indicators assessed

Annual Population Survey estimates of coefficient of variation for headline labour market indicators by sex, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, the nine estimates (three estimates for sex and three estimates for economic status) were considered “robust” under current data quality rules and this is still the case in Jan-Dec 2023.

Table A11: Summary of robustness for headline APS estimates by sex in Jan-Dec 2019 and Jan-Dec 2023 showing how many out of the three sex estimates were robust for each of the employment statuses.

Economic Status	Robustness	2019	2023
Employment	Robust	3	3
Unemployment	Robust	3	3
Inactivity	Robust	3	3

Data Confidence Ratings

For the Scottish APS estimates by sex, one of the nine estimates had decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023.

Based on Jan-Dec 2023 APS data for Scotland, we have six estimates with “high confidence” and three estimates with “moderate confidence”.

Table A12: Summary of confidence rating for headline APS estimates by sex in Jan-Dec 2019 and Jan-Dec 2023 showing how many out of the three sex estimates were robust for each of the economic statuses.

Economic Status	Sex	2019	2023
Employed (16 to 64)	People	High Confidence	High Confidence
Employed (16 to 64)	Men	High Confidence	High Confidence
Employed (16 to 64)	Women	High Confidence	High Confidence
Unemployed (16 and over)	People	High Confidence	Moderate Confidence
Unemployed (16 and over)	Men	Moderate Confidence	Moderate Confidence
Unemployed (16 and over)	Women	Moderate Confidence	Moderate Confidence
Inactive (16 to 64)	People	High Confidence	High Confidence
Inactive (16 to 64)	Men	High Confidence	High Confidence
Inactive (16 to 64)	Women	High Confidence	High Confidence

Annex 3.3 Estimates by Age

Sample sizes

When considering the change in sample size for Scotland split by age between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -62.1%
• the largest decrease was for “unemployed 25 to 34 year olds” which fell by -69.5%
• the smallest decrease was for “unemployed people aged 50 and over” which fell by -46.0%

Figure A11: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators irrespective of age

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for headline labour market indicators by broad age groups

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by age:

• all CVs have increased
• the average increase was 2.84
• the largest increase was for “unemployed 25 to 34 year olds” which rose by 9.41 from 9.47 to 18.88
• the smallest increase was for “employed 50 to 64 year olds” which rose by 0.34 from 0.72 to 1.06

Figure A12: Since 2019, coefficients of variation for labour market in Scotland have increased irrespective of age

Annual Population Survey estimates of coefficients of variation for headline labour market indicators by broad age groups, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, the 12 estimates (four estimates for age and three estimates for economic status) were considered “robust” under current data quality rules, and this was still the case in Jan-Dec 2023.

While the estimates remain “robust” under the current rules, the coefficients of variation for the unemployment estimates for all four of the age groups has increased from between 8.11 and 9.47 to between 11.16 and 18.88 over this time period.

It should be noted that the oldest age group for employment and inactivity is for those aged 50 to 64 whereas the oldest age group for unemployment is for those aged 50 and over.

Table A13: Summary of robustness for headline APS estimates by age in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Robustness	2019	2023
Employment	Robust	4	4
Unemployment	Robust	4	4
Inactivity	Robust	4	4

Data Confidence Ratings

From the Scottish APS estimates by age, six of the 12 estimates have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023.

Based on Jan-Dec 2023 APS data for Scotland, we have six estimates with “high confidence”, two estimates with “moderate confidence”, three estimates with “limited confidence”, and one estimate with “low confidence”.

Table A14: Summary of confidence ratings for headline APS estimates by age in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Age Group	2019	2023
Employed	16 to 24	High Confidence	High Confidence
Employed	25 to 34	High Confidence	High Confidence
Employed	35 to 49	High Confidence	High Confidence
Employed	50 to 64	High Confidence	High Confidence
Unemployed	16 to 24	Moderate Confidence	Limited Confidence
Unemployed	25 to 34	Moderate Confidence	Low Confidence
Unemployed	35 to 49	Moderate Confidence	Limited Confidence
Unemployed	50 and over	Moderate Confidence	Limited Confidence
Inactive	16 to 24	High Confidence	High Confidence
Inactive	25 to 34	High Confidence	Moderate Confidence
Inactive	35 to 49	High Confidence	Moderate Confidence
Inactive	50 to 64	High Confidence	High Confidence

Annex 3.4 Estimates by Age and Sex

Sample sizes

When considering the change in sample size in Scotland split by age and sex between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -61.3%
• the largest decrease was for “unemployed men aged 25 to 34” which fell by -79.7%
• the smallest decrease was for “unemployed men aged 50 and over” which fell by -45.3%

Figure A13: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators for all age and sex breakdowns

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for headline labour market indicators by broad age groups and sex

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by age and sex:

• all CVs have increased
• the average increase was 4.05
• the largest increase was for “unemployed men aged 25 to 34” which rose by 15.41 from 12.72 to 28.13
• the smallest increase was for “employed women aged 50 to 64” which rose by 0.47 from 1.08 to 1.56

Figure A14: Since 2019, coefficients of variation for the labour market in Scotland have increased irrespective of age and sex

Annual Population Survey estimates of coefficient of variation for headline labour market indicators by broad age groups and sex, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, all employment estimates by age and sex were “robust” under current data quality rules. However, by Jan-Dec 2023 there were four estimates considered “less robust” under the data quality rules; the other estimates were considered “robust”.

Table A15: Summary of robustness for headline APS estimates by age and sex in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Sex	Age Group	2019	2023
Employment	Men	16 to 24	Robust	Robust
Employment	Men	25 to 34	Robust	Robust
Employment	Men	35 to 49	Robust	Robust
Employment	Men	50 to 64	Robust	Robust
Employment	Women	16 to 24	Robust	Robust
Employment	Women	25 to 34	Robust	Robust
Employment	Women	35 to 49	Robust	Robust
Employment	Women	50 to 64	Robust	Robust
Unemployment	Men	16 to 24	Robust	Robust
Unemployment	Men	25 to 34	Robust	Less Robust
Unemployment	Men	35 to 49	Robust	Robust
Unemployment	Men	50 and over	Robust	Robust
Unemployment	Women	16 to 24	Robust	Less Robust
Unemployment	Women	25 to 34	Robust	Less Robust
Unemployment	Women	35 to 49	Robust	Less Robust
Unemployment	Women	50 and over	Robust	Robust
Inactivity	Men	16 to 24	Robust	Robust
Inactivity	Men	25 to 34	Robust	Robust
Inactivity	Men	35 to 49	Robust	Robust
Inactivity	Men	50 to 64	Robust	Robust
Inactivity	Women	16 to 24	Robust	Robust
Inactivity	Women	25 to 34	Robust	Robust
Inactivity	Women	35 to 49	Robust	Robust
Inactivity	Women	50 to 64	Robust	Robust

Data Confidence Ratings

From the Scottish APS estimates by age and sex, 12 of the 24 estimates have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023.

Based on Jan-Dec 2023 APS data for Scotland, we have “no confidence” or “low confidence” in seven of the 24 estimates. We have “limited confidence” in two of the 24 estimates. There are five estimates which we have “moderate confidence” in and 10 estimates we have “high confidence” in.

Table A16: Summary of confidence rating for headline APS estimates by age and sex in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Sex	Age Group	2019	2023
Employed	Men	16 to 24	High Confidence	High Confidence
Employed	Men	25 to 34	High Confidence	High Confidence
Employed	Men	35 to 49	High Confidence	High Confidence
Employed	Men	50 to 64	High Confidence	High Confidence
Employed	Women	16 to 24	High Confidence	High Confidence
Employed	Women	25 to 34	High Confidence	High Confidence
Employed	Women	35 to 49	High Confidence	High Confidence
Employed	Women	50 to 64	High Confidence	High Confidence
Unemployed	Men	16 to 24	Limited Confidence	Low Confidence
Unemployed	Men	25 to 34	Limited Confidence	No Confidence
Unemployed	Men	35 to 49	Limited Confidence	Low Confidence
Unemployed	Men	50 and over	Limited Confidence	Limited Confidence
Unemployed	Women	16 to 24	Limited Confidence	No Confidence
Unemployed	Women	25 to 34	Limited Confidence	No Confidence
Unemployed	Women	35 to 49	Limited Confidence	No Confidence
Unemployed	Women	50 and over	Limited Confidence	Low Confidence
Inactive	Men	16 to 24	High Confidence	Moderate Confidence
Inactive	Men	25 to 34	Moderate Confidence	Limited Confidence
Inactive	Men	35 to 49	Moderate Confidence	Moderate Confidence
Inactive	Men	50 to 64	High Confidence	High Confidence
Inactive	Women	16 to 24	High Confidence	Moderate Confidence
Inactive	Women	25 to 34	High Confidence	Moderate Confidence
Inactive	Women	35 to 49	High Confidence	Moderate Confidence
Inactive	Women	50 to 64	High Confidence	High Confidence

Annex 3.5 Estimates by Disability

For details of the definition of disability used in this section see Concepts and Definitions.

Sample sizes

When considering the change in sample size in Scotland split by disability between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -57.5%
• the largest decrease was for “unemployed non-disabled people” which fell by -65.6%
• the smallest decrease was for “employed disabled people” which fell by -48.7%

Figure A15: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators irrespective of disability

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for headline labour market indicators by disability

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by disability:

• all CVs have increased
• the average increase was 1.71
• the largest increase was for “unemployed disabled people” which rose by 4.05 from 7.12 to 11.18
• the smallest increase was for “employed non-disabled people” which rose by 0.22 from 0.38 to 0.60

Figure A16: Since 2019, coefficients of variation for the labour market in Scotland have increased irrespective of disability

Annual Population Survey estimates of coefficient of variation for headline labour market indicators by disability, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, the six estimates (two estimates for disability and three estimates for economic status) were considered “robust” under current data quality rules and this is still the case in Jan-Dec 2023.

Table A17: Summary of robustness for headline APS estimates by disability in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Robustness	2019	2023
Employment	Robust	2	2
Unemployment	Robust	2	2
Inactivity	Robust	2	2

Data Confidence Ratings

From the Scottish APS estimates by disability, one of the six estimates had decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023.

Based on Jan-Dec 2023 APS data for Scotland, we have “high confidence” in four estimates, “moderate confidence” in one estimate, and “limited confidence” in one estimate.

Table A18: Summary of confidence rating for headline APS estimates by disability in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Disability Status	2019	2023
Employed	Non-Disabled	High Confidence	High Confidence
Employed	Disabled	High Confidence	High Confidence
Unemployed	Non-Disabled	Moderate Confidence	Moderate Confidence
Unemployed	Disabled	Moderate Confidence	Limited Confidence
Inactive	Non-Disabled	High Confidence	High Confidence
Inactive	Disabled	High Confidence	High Confidence

Annex 3.6 Estimates by Ethnicity

For details of the definition of ethnicity used in this section see Concepts and Definitions.

Sample sizes

When considering the change in sample size in Scotland split by ethnicity between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably for all breakdowns except one
• the sample for “unemployed people with mixed and/or multiple ethnicities” rose by 33.3%
• the average decrease excluding the ethnicity which increased was
  -51.2%
• the largest decrease was for “economically inactive Asian people” which fell by -71.7%
• the smallest decrease was for “economically inactive people with mixed and/or multiple ethnicities” which fell by -30.0%

The only measure to show an increase in sample size was unemployed people with mixed and/or multiple ethnicities. This is a result of the sample size in both 2019 and 2023 being less than five people and highlights why changes in sample size should not be relied on alone when assessing sample quality.

In 2023, the following estimates were below the current data quality rules for sample sizes meaning we could not publish this data:

• unemployed people with mixed and/or multiple ethnicities,
• unemployed Black, Caribbean, and African people,
• unemployed people with other ethnicities

Figure A17: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators irrespective of ethnicity with the exception of unemployed people from mixed and/or multiple groups.

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for headline labour market indicators by ethnicity

Note: The increase for unemployed people from mixed and/or multiple groups is a result of the sample size in both 2019 and 2023 being less than 5 people and highlights why changes in sample size should not be relied on alone when assessing sample quality.

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by ethnicities:

• all CVs have increased
• the average increase was 3.44
• the largest increase was for “employed people with other ethnicities” which rose by 7.59 from 7.22 to 14.81
• the smallest increase was for “employed white people” which rose by 0.24 from 0.41 to 0.65

Five estimates by ethnicity in Jan-Dec 2023 were above 20, indicating a level of volatility in the data which would mean we would not publish these estimates in our publications. The highest CV in Jan-Dec 2023 was 51.08 for the “unemployed people with mixed and/or multiple ethnicities” estimate. A further five estimates had CVs of more than 10 which causes concern.

Figure A18: Since 2019, coefficients of variation for labour market estimates in Scotland have increased irrespective of ethnicity

Annual Population Survey estimates of coefficient of variation for headline labour market indicators by ethnicity, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

Of the 15 estimates (five estimates for each ethnicity and three estimates for economic status), all estimates of employment were “robust” in Jan-Dec 2019 under current data quality rules, and remained “robust” in Jan-Dec 2023.

For unemployment in Jan-Dec 2019, there were three estimates considered “not robust” and one considered “less robust” and this had remained the same in Jan-Dec 2023. Where there has been a slight difference is in inactivity where in Jan-Dec 2019 one estimate was considered “less robust” but in Jan-Dec 2023, two estimates were considered “less robust” under current data quality rules.

Table A19: Summary of robustness for headline APS estimates by ethnicity in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Robustness	2019	2023
Employment	Robust	5	5
Unemployment	Robust	1	1
Unemployment	Less Robust	1	1
Unemployment	Not Robust	3	3
Inactivity	Robust	4	3
Inactivity	Less Robust	1	2

Data Confidence Ratings

From the Scottish APS estimates by ethnicity, seven of the 15 estimates had decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023, whilst four remained as “no confidence”.

Based on Jan-Dec 2023 APS data for Scotland, we have “no confidence” in five of the 15 estimates. We have “limited confidence” in five of the 15 estimates. There were three estimates with “moderate confidence” and we had “high confidence” in two of the fifteen estimates.

Table A20: Summary of confidence rating for headline APS estimates by ethnicity in Jan-Dec 2019 and Jan-Dec 2023

Economic Status	Ethnicity	2019	2023
Employed (16 to 64)	White	High Confidence	High Confidence
Employed (16 to 64)	Mixed and/or Multiple	Moderate Confidence	Limited Confidence
Employed (16 to 64)	Asian	High Confidence	Moderate Confidence
Employed (16 to 64)	Black, Caribbean, and African	Moderate Confidence	Moderate Confidence
Employed (16 to 64)	Other	Moderate Confidence	Limited Confidence
Unemployed (16 and over)	White	High Confidence	Moderate Confidence
Unemployed (16 and over)	Mixed and/or Multiple	No Confidence	No Confidence
Unemployed (16 and over)	Asian	Low Confidence	No Confidence
Unemployed (16 and over)	Black, Caribbean, and African	No Confidence	No Confidence
Unemployed (16 and over)	Other	No Confidence	No Confidence
Inactive (16 to 64)	White	High Confidence	High Confidence
Inactive (16 to 64)	Mixed and/or Multiple	No Confidence	No Confidence
Inactive (16 to 64)	Asian	Moderate Confidence	Limited Confidence
Inactive (16 to 64)	Black, Caribbean, and African	Limited Confidence	Limited Confidence
Inactive (16 to 64)	Other	Moderate Confidence	Limited Confidence

Based on the above assessment, it is not possible to publish estimates for all of the 5-category ethnicity groups by economic status. Therefore, we have also considered the binary categories. Although this is not ideal as it may hide inequalities that occur between separate ethnic groups, it would allow us to provide some information that cannot be provided for the more disaggregated groups.

The binary classification showed similar trends and although the quality rules and our confidence ratings were better for minority ethnic groups combined, it had still decreased since 2019.

Annex 3.7 Employment estimates by Industry

There are 21 industrial categories ranging from A to U. However, we only regularly report on 19 of them (A to S) and exclude T and U.

Sample sizes

When considering the change in sample size in Scotland split by industry of employment between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -58.2%
• the largest decrease was for the “C: Manufacturing” sector which fell by -68.6%
• the smallest decrease was for the “J: Information and communication” sector which fell by -46.6%

Figure A19: Since 2019, sample sizes for employed people in Scotland have decreased in all industries

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for employed people aged 16 and over by industry

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by industry of employment:

• all CVs have increased
• the average increase was 2.36
• the largest increase was for “A: Agriculture, forestry and fishing” which rose by 4.04 from 6.57 to 10.60
• the smallest increase was for “O: Public admin. and defence” which rose by 0.94 from 2.79 to 3.73

In 2023, four of the Industry of employment estimates had CVs of more than 10 which causes concern.

Figure A20: Since 2019, coefficients of variation for labour market estimates for employed people in Scotland increased across all industries

Boxplot of Annual Population Survey estimates of coefficient of variation for employed people aged 16 and over by industry, Jan-Dec 2019 and Jan-Dec 2023. More detail on understanding boxplots is provided in Concepts and Definitions.

Current Data Quality Rules

When considering the estimates in Jan-Dec 2019, the 19 estimates we normally publish were considered “robust” under current data quality rules and this is still the case in Jan-Dec 2023. For the remaining two industry estimates that we normally suppress, in 2019 they would be considered “less robust” and in 2023 they were “not robust”.

Table A21: Summary of robustness for headline APS estimates of employed people by industry of employment in Jan-Dec 2019 and Jan-Dec 2023

Robustness	2019	2023
Robust	19	19
Less Robust	2	0
Not Robust	0	2

Data Confidence Ratings

From the Scottish APS estimates by Industry, 12 of the 21 estimates have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023, whilst two have remained as “no confidence”.

Based on Jan-Dec 2023 APS data for Scotland, we have “no confidence” in two of the 21 estimates. We have “limited confidence” in four estimates. There were nine estimates we had “moderate confidence” in, and we had “high confidence” in six of the 21 estimates.

Table A22: Summary of confidence rating for headline APS estimates of employed people by industry of employment in Jan-Dec 2019 and Jan-Dec 2023

Industry	2019	2023
A: Agriculture, forestry and fishing	Moderate Confidence	Limited Confidence
B: Mining and quarrying	Moderate Confidence	Moderate Confidence
C: Manufacturing	High Confidence	Moderate Confidence
D: Electricity, gas, air conditioning supply	Moderate Confidence	Limited Confidence
E: Water supply, sewerage, waste	Moderate Confidence	Limited Confidence
F: Construction	High Confidence	Moderate Confidence
G: Wholesale, retail, repair of vehicles	High Confidence	High Confidence
H: Transport and storage	High Confidence	Moderate Confidence
I: Accommodation and food services	High Confidence	High Confidence
J: Information and communication	High Confidence	Moderate Confidence
K: Financial and insurance activities	High Confidence	Moderate Confidence
L: Real estate activities	Moderate Confidence	Limited Confidence
M: Prof, scientific, technical activities	High Confidence	High Confidence
N: Admin and support services	High Confidence	Moderate Confidence
O: Public admin and defence	High Confidence	High Confidence
P: Education	High Confidence	High Confidence
Q: Health and social work	High Confidence	High Confidence
R: Arts, entertainment and recreation	High Confidence	Moderate Confidence
S: Other service activities	High Confidence	Moderate Confidence
T: Households as employers	No Confidence	No Confidence
U: Extraterritorial organisations	No Confidence	No Confidence

Annex 3.8 Estimates by Highest Qualification

Sample sizes

When considering the change in sample size in Scotland split by highest qualification between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -63.6%
• the largest decrease was people with “no qualifications” which fell by -69.5%
• the smallest decrease was for people with “degrees and/or professional qualifications” which fell by -52.1%

Figure A21: Since 2019, sample sizes for people in Scotland have decreased for all headline labour market indicators irrespective of highest qualification attained

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for people aged 16 to 64 by highest qualification

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by highest qualification:

• all CVs have increased
• the average increase was 1.80
• the largest increase was for people with “National 4 grade or equivalent” which increased by 4.81 from 5.72 to 10.52
• the smallest increase was for people with “degree and/or professional qualifications” which increased by 0.42 from 1.06 to 1.49

In 2023, one of the Highest Qualification estimates had a CV of more than 10 which is cause for concern.

Figure A22: Since 2019, coefficients of variation for the labour market estimates for people in Scotland have increased for all indicators irrespective of highest qualification attained

Annual Population Survey estimates of coefficient of variation for people aged 16 to 64 by highest qualification, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, the seven estimates for the Scottish data for Highest Qualification were considered “robust” under current data quality rules and this was still the case in Jan-Dec 2023.

Table A23: Summary of robustness for headline APS estimates of people by highest qualification in Jan-Dec 2019 and Jan-Dec 2023

Robustness	2019	2023
Robust	7	7
Less Robust	0	0
Not Robust	0	0

Data Confidence Ratings

From the Scottish APS estimates by Highest Qualification, two of the seven estimates have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023.

Based on Jan-Dec 2023 APS data for Scotland, we have “limited confidence” in the estimate for people with “National 4 grade or equivalent qualifications”, “moderate confidence” in the estimate for people with “other qualifications”, and “high confidence” in all other estimates.

However, due to the “limited confidence” for one of the qualification estimates, it would not be possible to publish any estimates by Qualification for the seven categories at this time unless some of the categories were grouped.

Table A24: Summary of confidence rating for headline APS estimates of people by highest qualification in Jan-Dec 2019 and Jan-Dec 2023

Qualification	2019	2023
Degree, Professional Qualification	High Confidence	High Confidence
HNC/HND or equivalent	High Confidence	High Confidence
Higher, A-Level or equivalent	High Confidence	High Confidence
National 5 grade or equivalent	High Confidence	High Confidence
National 4 grade or equivalent	Moderate Confidence	Limited Confidence
Other Qualifications	High Confidence	Moderate Confidence
No Qualifications	High Confidence	High Confidence

Annex 3.9 Economic Inactivity estimates by Inactivity Reason and Whether Want to Work or Not

Sample sizes

When considering the change in sample size in Scotland split by economic inactivity reason between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• the achieved sample size for all estimates has dropped considerably
• the average decrease was -55.1%
• the largest decrease was for “looking after home/family” which fell by -66.9%
• the smallest decrease was for “retired” which fell by -43.0%

When considering the change in sample size in Scotland split by want to work/do not want to work between Jan-Dec 2019 and Jan-Dec 2023 from the APS:

• “inactive people who want to work” fell by -67.5%
• “inactive people who do not want to work” fell by -55.3%

Figure A23: Since 2019, sample sizes for economically inactive people in Scotland have decreased for all inactivity reasons and whether a person wants to work or not

Annual Population Survey estimates of percentage change in sample size between Jan-Dec 2019 and Jan-Dec 2023 for economically inactive people aged 16 to 64 by inactivity reason and whether a person wants to work or not

Coefficient of Variation

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by economic inactivity reason:

• all CVs have increased
• the average increase was 3.31
• the largest increase was for “discouraged” which rose by 9.80 from 24.46 to 34.26

• the smallest increase was for “long-term sick” which rose by 0.98 from 2.35 to 3.33

When considering the change in CVs between Jan-Dec 2019 and Jan-Dec 2023 for the APS estimates in Scotland split by want to work/do not want to work:

• “inactive people who want to work” increased by 2.18 from 2.91 to 5.09
• “inactive people who do not want to work” increased by 0.26 from 0.75 to 1.01

One estimate of inactivity reason in Jan-Dec 2023 had a CV above 20, indicating a level of volatility in the data which would mean we would not publish this figure in our publications. A further inactivity reason had a CV greater than 10 which causes concern.

Figure A24: Since 2019, coefficients of variation for economically inactive estimates for people in Scotland have increased for all inactivity reasons and whether a person wants to work or not

Annual Population Survey estimates of coefficient of variation for economically inactive people aged 16 to 64 by inactivity reason and whether a person wants to work or not, Jan-Dec 2019 and Jan-Dec 2023

Current Data Quality Rules

In Jan-Dec 2019, six of the seven estimates were considered “robust” under current data quality rules and this was still the case in Jan-Dec 2023. However, the seventh estimate (people inactive because they were discouraged) was considered “less robust” in Jan-Dec 2019 and ”not robust” in Jan-Dec 2023.

Table A25: Summary of robustness for headline APS estimates of inactive people by reason for inactivity in Jan-Dec 2019 and Jan-Dec 2023

Robustness	2019	2023
Robust	6	6
Less Robust	1	0
Not Robust	0	1

In addition to inactivity reason, we can also look at inactive people and whether they want to work or not. In Jan-Dec 2019, both the estimates for inactive people who want to work and inactive people who do not want to work were considered “robust” under current data quality rules. This remained the case in Jan-Dec 2023.

Data Confidence Ratings

From the Scottish APS estimates by inactivity reasons and whether want to work, five of the nine estimates have decreased in confidence rating between Jan-Dec 2019 and Jan-Dec 2023, while one remained as “no confidence”.

Based on Jan-Dec 2023 APS data for Scotland, we have “no confidence” in one of the nine estimates and “limited confidence” in one estimate.

There were four estimates which we have “moderate confidence” in and we have “high confidence” in three of the nine estimates.

Table A26: Summary of confidence rating for headline APS estimates of inactive people by reason for inactivity and whether want to work in Jan-Dec 2019 and Jan-Dec 2023

Topic	Reason	2019	2023
Inactivity Reason	Student	High Confidence	High Confidence
Inactivity Reason	Looking after home/family	High Confidence	Moderate Confidence
Inactivity Reason	Temporary sick	Moderate Confidence	Limited Confidence
Inactivity Reason	Long-Term sick	High Confidence	High Confidence
Inactivity Reason	Discouraged	No Confidence	No Confidence
Inactivity Reason	Retired	High Confidence	Moderate Confidence
Inactivity Reason	Other	High Confidence	Moderate Confidence
Whether want to work	Want to work	High Confidence	Moderate Confidence
Whether want to work	Does not want to work	High Confidence	High Confidence