Scottish House Condition Survey: Methodology Notes 2023

1 Methodological and Technical Notes

1.1 Survey Estimation

From 2012 the Scottish House Condition Survey (SHCS) has been a module of the Scottish Household Survey (SHS). In 2012, around a third (36%) of respondents to the SHS were invited to participate in a follow-up inspection by SHCS surveyors. This proportion has had to increase over time as the conversion rate from the social interview to the physical survey has decreased. Over half of respondents (58%) to the 2023 SHS were invited to participate in the 2023 SHCS to ensure that the required number of physical surveys were achieved.

1.1.1 Sample Sizes and Gross Dwelling Numbers

In Table 1.1 we provide the achieved sample sizes in the social interview and physical dwelling inspection follow-up for all years of the annual SHCS to 2023.

Table 1.1: Achieved sample for the social interview and physical survey and the number of occupied dwellings by survey year, 2003/04 to 2023

Survey year	Social interview (achieved sample)	Physical survey (achieved sample)	Number of occupied dwellings (thousands)
2003/4	3,870	3,090	2,269
2004/5	3,783	3,093	2,301
2005/6	3,679	3,147	2,315
2007	3,867	3,033	2,314
2008	3,763	3,015	2,331
2009	4,153	3,346	2,344
2010	3,853	3,115	2,357
2011	3,949	3,219	2,368
2012	3,813	2,787	2,383
2013	3,780	2,725	2,393
2014	3,787	2,682	2,406
2015	4,083	2,754	2,417
2016	4,220	2,850	2,431
2017	5,049	3,002	2,445
2018	4,843	2,964	2,458
2019	4,843	2,997	2,473
2020	[z]	[z]	[z]
2021	3,980	3,174	2,498
2022	5,501	2,983	2,515
2023	6,059	3,151	2,535

Table 1.1 also shows the estimated total number of households (occupied dwellings) in Scotland for each survey year which provides the basis for grossing up the estimates of households and dwellings in this report. These figures are produced annually by the National Records of Scotland as part of their inter-censal household estimates publication[1].

The SHCS is a sample survey. All survey results are estimates of the true prevalence within the population and will contain some error associated with sampling variability. The likely size of such variability can be identified, by taking account of the size and design of the sample, as described in the subsections on confidence intervals, design effects and statistical significance.

In addition to sampling variability, there are other sources of uncertainty, such as those arising from incomplete responses or failure to secure participation in the survey from each sampled household. Where non-response is not random, i.e., some types of households are less likely to participate than others, bias is introduced into the survey data. Such errors have not been quantified in this report.

In general, the smaller the sample size, the greater the likelihood the estimate may not be reflective of the true value in the population or housing stock, so more care must be taken when using smaller subsets of the survey sample for analysis. In this report estimates representing 2 or fewer cases, or where the base sample is below 30 have been suppressed.

Different types of estimates are subject to different levels of uncertainty associated with sampling and design. For example, estimates of change (i.e., figures relating to comparisons across survey years) are generally subject to greater sampling error than point-in-time estimates (i.e., figures relating to one survey year only) and such errors would be understated by the confidence intervals in Table 1.2. There is more uncertainty associated with complex measures, such as the fuel poverty rate and this is not quantified in this report or reflected by the confidence intervals in Table 1.2.

1.1.2 Confidence Intervals

By convention, a 95% confidence interval is used to quantify the variability of a sample estimate, under which there is a 1 in 20 chance that the true value will fall outside the given confidence interval.

Table 1.2 shows the 95% confidence limits for estimates of proportions based on sub-samples of various sizes before design effects are considered.

Table 1.2: 95% confidence limits for estimates of proportions based on sub-samples of various sizes (excluding design effects)

Sub-sample size	Estimate of 10% or 90%	Estimate of 20% or 80%	Estimate of 30% or 70%	Estimate of 40% or 60%	Estimate of 50%
100	5.9%	7.8%	9.0%	9.6%	9.8%
250	3.7%	5.0%	5.7%	6.1%	6.2%
500	2.6%	3.5%	4.0%	4.3%	4.4%
1,000	1.9%	2.5%	2.8%	3.0%	3.1%
1,500	1.5%	2.0%	2.3%	2.5%	2.5%
2,000	1.3%	1.8%	2.0%	2.1%	2.2%
3,000	1.1%	1.4%	1.6%	1.8%	1.8%

1.1.3 Design Effects

The design effect is the ratio between the variance (average squared deviation of a set of data points from their mean value) of a variable under the actual sampling method used and the variance computed under the assumption of simple random sampling. In short, a design effect of 2 would mean doubling the size of the sample used to obtain the same level of precision as with a simple random sample; a design effect of 0.5 implies the reverse. Design effect adjustments are necessary where standard errors (and confidence intervals) are affected by the design and complexity of the survey.

Disproportionate stratification and sampling with non-equal probabilities tends to increase standard errors, giving a design effect greater than 1. However, this can be controlled by deliberately over-sampling in stratum where the item of interest is either very rare or variable. The impact of non-response weighting on standard errors tends to be, although with exceptions, comparatively limited. The sampling design of the SHCS meets the criteria above in that disproportionate stratification is applied across the 32 local authority areas with over-sampling of remote rural areas - for example in Orkney Islands and Shetland Islands. As a result, one would expect the design effect to be above 1 although only modestly so.

Table 1.3 shows the design factors (the design factor is the square root of the design effect) for all the SHCS waves since 2003/04. When using a mixture of the physical and social survey data, the physical survey design factor must be used. The physical survey design factor for the 2022 SHCS is 1.13. Since 2021 it is not possible to produce social weights as summary surveys have not been undertaken for all households in the SHCS subsample of the SHS. Therefore, it is not possible to produce a social survey design factor.

Table 1.3: Design factors for the SHCS by survey year, 2003/04 to 2023

Survey year	Physical survey design factor	Social survey design factor
2003/04	1.14	1.13
2004/05	1.18	1.17
2005/06	1.14	1.14
2007	1.13	1.11
2008	1.11	1.11
2009	1.09	1.08
2010	1.11	1.10
2011	1.12	1.11
2012	1.09	1.08
2013	1.09	1.08
2014	1.09	1.08
2015	1.10	1.08
2016	1.10	1.08
2017	1.10	1.08
2018	1.11	1.08
2019	1.12	1.08
2020	[z]	[z]
2021	1.12	[z]
2022	1.13	[z]
2023	1.11	[z]

In general, when producing estimates at a local authority level from the SHCS, no design effect adjustment of standard errors is necessary because simple (equal interval) random sampling is carried out within each local authority.

1.1.4 Weighting

Up to and including the 2019 SHCS, results had been weighted in broadly the same way. However, the shift from face-to-face to remote social interviewing, and external only physical inspections in 2021, as a response to Covid-19 was associated with changes in the profile of the achieved sample that were unlikely to reflect real changes in Scotland’s population. As a result, alternative weighting methods to those used pre-2021 were developed that incorporated SIMD quintiles and tenure as part of the calibration process. The methodology for this is detailed in full in the 2021 report.

In 2022 face-to-face interviews and in home surveying resumed, and the SHCS resumed the historic weighting approach from 2019 and earlier. The 2023 results are weighted in line with this method which is discussed below.

Weighting for the SHCS is done in two stages: a selection weighting stage to address the unequal selection probabilities followed by calibration weighting to correct for non-response bias.

Selection Weighting

Selection weights are inverse to the probability of being selected to participate in a survey. The SHCS sample is stratified by local authority, with smaller local authorities and those with historically lower response rates having higher sampling rates. The selection weight for each stratum is calculated as the proportion of Scottish households (from NRS estimates) in the stratum divided by the proportion of eligible selected addresses in the stratum.

Calibration Weighting

Calibration weighting corrects for non-response bias in surveys by weighting the achieved sample so that it is consistent with known external totals.

For the SHCS the stratum selection weight was applied to the survey data to act as entry weights for the calibration. The execution of the calibration step then modified the entry weights so that the weighted total of responding households matched:

Table 1.4 Calibration characteristics and sources

Calibration characteristic	Geography	Categories	Source
Number of households	Local authority	N/A	NRS household estimates
Dwelling age	Scotland	Pre-1919; 1919-1944; 1945-1964; 1965-1982; 1983-2002; and post 2002	SHCS
Dwelling type	Scotland	Detached; semi-detached; terraced; tenement; and other flats	SHCS
Urban-rural classification (2020)	Scotland	Six-fold classification	SHCS

The Scotland-level calibration targets were generated from the weighted sample for dwelling age, dwelling type, and urban-rural classification.

As noted in the 2022 Methodology Notes a potential bias in tenure was identified in the 2022 sample. However as the results of the 2022 Scottish census were not yet available we were unable to get a complete picture of the tenure of Scottish households from published sources other than the SHS. As the census results have now been published we discuss this below.

1.1.5 Comparison to previous SHCS waves and 2022 census

The results of the 2022 survey were published as Accredited Official Statistics and were determined to be broadly comparable to 2019 and earlier years.

When the SHCS 2022 results were published, most key measures that we would expect to remain broadly stable were in line with 2019. As discussed in the 2022 methodology report, weighted results for tenure were slightly different than expected, but this could reflect genuine changes in the population.

The tenure results from Scotland’s Census 2022 have now been published. Owner occupation and socially rented dwellings appear to be slightly over-represented in SHCS 2022 compared to the 2022 Census, and private rented dwelling slightly under-represented. The same was true, to a similar extent for rented properties, although slightly lower for owner occupied properties, in the SHCS 2011 compared to the 2011 Census. Of note the 2022 SHCS does estimate a lower percentage of rent free dwellings than the census by a wider margin than in 2011, however as rent free dwellings are a small part of the sample there is greater uncertainty with this estimate. Collectively the similarity in differences between the 2011 census and SHCS and the 2022 census and SHCS suggests that the SHCS 2022 tenure results are comparable to those from SHCS 2019 and earlier.

Table 1.5: Tenure in Scottish House Condition Survey and Scotland’s Census

	SHCS 2011	Census 2011	SHCS 2022	Census 2022
Owned	62.00%	61.57%	64.64%	63.20%
Social rented	25.64%	24.29%	23.62%	22.50%
Private rented	11.11%	12.43%	10.95%	12.87%
Rent Free	1.2%	1.28%	0.79%	1.43%

It should also be noted that the SHCS (and SHS) are annual surveys based on a sample of the general population in private residences in Scotland and are not designed to capture tenure as comprehensively as other formal surveys of tenure, e.g. the census. Therefore, figures in this publication may not align with National Statistics on household tenure.

For estimates of the total number of dwellings by tenure, readers are referred to the Scottish Government Housing Statistics for Scotland publication which uses information from social landlords’ returns which comprehensively cover the social housing sector and therefore provides more accurate estimates of the total stock by tenure

1.1.6 NRS estimates of occupied dwellings

As discussed above the SHCS uses the NRS inter-censal household estimates to calibrate the weights used in the summary report. These figures are produced annually by NRS and are calculated using the Council Tax Base form available from the Scottish Government website. The household estimates for Scotland and for each council area are calculated by subtracting vacant dwellings and second homes from the total dwellings figures.

The resulting number of occupied dwellings is then adjusted from September back to June, as the National Records of Scotland (NRS) mid-year population estimates and household projections are both based at 30 June each year.

These estimates are then adjusted to account for differences in the number of households estimated from Council Tax data and recorded in Scotland’s Census 2001, 2011 and 2022[2].

Following the 2022 census NRS revised their household estimates for the 2012 to 2022 period. For Scotland as a whole there was a downward revision which increased over time, from 0.2% in 2012 (-3,590 households) to 1.4% (-34,478 housheolds) in 2022.

This revision means that the 2023 estimated household total which the SHCS used to calibrate the 2023 weights is lower than the figure which was used to calibrate the 2022 weights for the annual report. As such when comparing this publication to previous publications it may appear that there has been a decrease in the dwelling stock, when in fact there has not been.

In order to account for this we have updated the time series estimates in our tables in this publication to incorporate the revised 2012 to 2022 household estimates. This has been applied in tables only where household totals were presented, with revisions based on the previous estimated percentages applied to the new household total. For example: in 2019 we estimated 24.6%, around 613,000 households, were in fuel poverty out of an estimated 2,495,000 occupied dwellings. Applying this percentage to the revised 2019 household estimate of 2,473,000 gives us an updated estimate of 608,000 households in fuel poverty, equal to 24.6% of the revised household figures.

Due to this household level figures from 2012-2022 in the 2023 report will be inconsistent with earlier SHCS publications.

​​​​​​​1.1.7 Statistical Significance

Because the survey’s estimates may be affected by sampling errors, apparent differences may not reflect real differences in the population. A difference is significant if it is so large that a difference of that size is unlikely to have occurred purely by chance.

Comparisons in this publication are tested at the 5 per cent level as described in the subsection on confidence intervals. Testing significance involves comparing the difference between two statistics (for example, the percentage of households rated as EPC band C or better for the social sector compared to the private sector) with the 95 per cent confidence limits for each of the two estimates considered.

Our approach to testing statistical significance follows that described in the Scottish Household Survey 2023 methodology and fieldwork outcomes.

In the example above (see Table EE9 in the supporting energy efficiency tables), the percentage of social sector households rated as EPC band C or better is 69% with a 95 per cent confidence interval of +/- 4 percentage points, having accounted for the design factor of 1.11 in Table 1.3. The percentage of private sector households rated as EPC band C or better is 51% with a 95 per cent confidence interval of +/- 2 percentage points. As the absolute difference between the estimates (18 percentage points) is greater than the square root of the sum of the squared confidence intervals (4 percentage points), we conclude that the difference between the estimates is statistically significant at the 5 per cent level.

1.1.8 ​​​​​​​Table Conventions

The following conventions are used in tables:

• [low] indicates a value is less than 0.5% or 500 households
• [w] indicates there are no sample cases
• [c] indicates that the base sample is too small to report (below 30 cases) or the estimate represents 2 or fewer sampled households
• [x] for not available, i.e. the data was not collected in the survey 
• [z] indicates that a value is unavailable as it is not applicable

These conventions are consistent with the guidance on using symbols and shorthand when publishing data tables on public sector websites.

All numbers are rounded to the nearest 1,000 and percentages are generally rounded to the nearest whole number. Because of rounding, figures in tables and charts may not always add exactly​​​​​​​.

1.1.9 Households missing income

Although some level of item non-response is inevitable across all aspects of the social and physical surveys (e.g. where a householder refused to answer a particular question, or a surveyor could not get into a loft), in most situations this does not affect the power of the survey to produce valid and useful estimates. The exception to this is the assessment of income, where there is generally a higher proportion of item refusals.

In order for the survey to be able to produce income estimates, a statistical process known as imputation is carried out. Imputation involves replacing missing values with the values associated with other households which have the same characteristics, defined according to the nature of the missing item.

Hot Deck imputation was used for all missing income items. In Hot Deck imputation, the sample is divided into imputation classes based on the relevant characteristics of cases and these classes contain potential donor cases. A donor case is selected at random from the imputation class and the item value for that case is assigned to the case with the missing item value. The relevant characteristics were chosen using regression analysis.

The imputation of missing income data has been carried out by the survey contractor, Ipsos.

Nevertheless, some households do not provide a complete enough response to income questions, and as such have no income data recorded. These households therefore are excluded from any analysis which reports on income./In 2023 there were 47 such households, around 1.5% of the achieved sample. 

​​​​​​​1.2 Missing Tenure Information

Because of a routing error tenure information is not available for a small number of cases in the 2012 and 2013 surveys (46 in 2012, 42 in 2013). Unlike previous years, respondents who reported living in their property rent free were not asked from whom they rent their property. Answers to that question are required to assign respondents to one of four tenure groups and then into social or private sector categories.

This was rectified for the 2014 fieldwork and the full sample has been used when reporting on tenure for subsequent years. This introduces some discontinuities in comparing statistics for the social (or the private) sector for 2014 onwards, on the one hand, and previous years, on the other. However, these are expected to be small as the proportion of households who reported living “rent free”, and were thus excluded, in the years leading up to 2012 and 2013 ranged from 1.3% to 2.6%. For further details please refer to the respective earlier key findings reports. Tables in key findings reports from the SHCS are clear whether data for 2011 and earlier are presented including or excluding rent free cases.

​​​​​​​1.3 Energy Models

Two domestic energy models, summarised in Table 1.6, are used to produce the energy outputs in this report. They are based on the same core methodology but have some different assumptions and calculations affecting the output values.

Table 1.6: Summary of domestic energy models used on SHCS data

Model	SAP	BREDEM 2012
Version	SAP 2009 SAP 2012 (RdSAP 9.92) for 2014 to 2019 SAP 2012 (RdSAP 9.93) from 2018 onwards	Version 1.0 for data up to 2013 Version 1.1 for data from 2014 onwards
Outputs	Energy efficiency ratings Environmental impact ratings	Fuel poverty energy use and running costs Carbon emissions
Fuel prices	SAP standard	Based on a range of sources. For more details see Table 1 in the section on Measuring Fuel Poverty in SHCS methodology notes 2019
Occupancy	Number of occupants derived based on total floor area of the dwelling	Actual number of occupants in the dwelling
Heating regime	21°C in the main living area and 18°C elsewhere; 9 hours per weekday and 16 hours at the weekend	For carbon emissions, as SAP For fuel poverty energy use/running costs for 2021 onwards, as described in paragraph 50 of the SHCS methodology notes 2019 For fuel poverty energy use/running costs for 2012 to 2019, as described in paragraph 48 of the SHCS methodology notes 2019
Climate	East Pennines	Based on geographical location. For fuel poverty energy use/running costs postcode district-level weather data is being used for 2021 onwards
Energy end-use included	• space heating • water heating • fixed lighting • gains from renewable energy technologies	As SAP but also energy used for: • cooking • running appliances

All energy efficiency and environmental impact rating related statistics from and post 2022 presented in this report are based on SAP 2012 (RdSAP 9.93).

Carbon emissions are calculated based on the standard heating regime, applying carbon intensity values to each fuel type used. Emissions factors for the BREDEM 2012 model come from SAP 2012 and are provided in Table 1.7.

Table 1.7: Carbon intensity of common heating fuels, SAP 2012

Fuel	Kilograms (kg) of carbon dioxide (CO2) per kilowatt hour (kWh)
Mains gas	0.216
LPG	0.241
Oil	0.298
Coal	0.394
Anthracite	0.394
Smokeless fuel	0.433
Wood - logs	0.019
Wood - pellets	0.039
Wood - chips	0.016
Electricity	0.519

From 2018 to 2019 SAP based energy variables under SAP 2012 RdSAP v9.92 and v9.93 were reported. For 2021 onwards energy variables under SAP 2012 RdSAP v9.92 are not available. Compared to v9.92, U-values for solid, insulated stone and uninsulated cavity walls improved, whereas they declined for insulated cavity walls. As a result, the mean SAP rating under v9.93 was 0.16 SAP points less than under v9.92 in 2019 and 0.17 points less in 2018.

Over the years improvements have been made to how the BREDEM 2012 model is used to produce energy outputs from the SHCS.

From 2016 the SHCS has collected information about the presence of pre-payment meters in dwellings which allows more accurate fuel prices to be assigned to these dwellings.

From 2019 more detailed information on combi boilers has been included to improve the accuracy of calculations surrounding hot water losses. As a result, the mean BREDEM 2012 modelled energy consumption is expected to increase by around 33 kWh per year.

Furthermore, from 2019 a household’s lights and appliances are assigned as using an off-peak tariff if an off-peak electricity meter is present, even if there is no form of electric heating in the dwelling. Previously, where a household did not have a form of electric heating, the lights and appliances were assumed to use standard electricity. This change does not affect the energy consumption of a dwelling, only the fuel prices applied to the energy associated with lighting and appliance use.

From 2022 a minor bug fix, which had already been fixed retrospectively for the 2021 data, was applied to improve the modelling of non-mains gas fuelled cookers. This bug fix had no impact on SAP ratings, as cooking energy use is not a component of the SAP methodology. It is however included in the BREDEM energy consumption which is used for fuel poverty calculations. As data is not collected on the type of non-mains gas fuel used by the cooker, it is now assumed that if bulk LPG is used for space heating then it is also being used for cooking, otherwise it is assumed bottled gas is being used. The error affected 146 cases in the 2021 data and resulted in a decrease in total fuel costs of between £22 and £75 for households with normal size cookers. Households with range cookers on the other hand, had an increase in total fuel costs of approximately £32 for oil fuel ranges, £272 for solid fuel ranges and £1,383 for gas ranges.

Climate factors such as external temperature, wind speed, latitude, mean global solar irradiance and height above sea level are determined by the area in which the dwelling is located. Prior to 2021, weather data for the nine Scottish regions in Table U6 in SAP 2012 was used. From 2021 more detailed postcode district weather data is being used from Table 172 of the Product Characteristic Database (PCDB).

The impact of using postcode district weather data has been measured using data from the 2015, 2016, 2017 and 2021 Scottish House Condition Surveys. It was found that many dwellings in the achieved sample were clustered in postcode districts where the average external temperature was higher and wind speed was lower than the regional averages previously used. As expected, a decrease in wind speed (which is most likely to be affected by local geography) combined with an increase in external temperature resulted in a decrease in mean energy consumption and mean annual running costs.

​​​​​​​1.4 Boilers

Testing compliance of boilers with current Scottish Building Standards for domestic properties is carried out by comparing the boiler efficiency to minimum requirements. Data on the efficiency of household heating systems was first produced for the 2012 SHCS. However, there was a change to the methodology for the 2014 and 2015 SHCS which made an adjustment to the modelling to allow for the assumption that a poorly controlled system is, in effect, less efficient.

In the 2016 SHCS report, the full boiler efficiency dataset was revised to ensure it was on a consistent basis across years and represented the efficiency of the heating system before any adjustments for lack of controls. Efficiencies are taken directly from the Product Characteristics Database whenever possible and from the SAP default efficiencies for that system otherwise. This is therefore more representative of the actual boiler efficiency.

The thresholds used to test compliance for oil condensing boilers were also updated in 2016 to reflect current minimum standards. The full time series presented from 2017 onwards continues to reflect these changes.

Furthermore, from 2022 an improvement has been made to the boiler model which results in a higher proportion of boilers being matched to the Product Characteristics Database (PCDB), thereby providing greater accuracy in boiler efficiencies. The default efficiencies which are used for unmatched cases are based on tables published in SAP 2012, but boilers in general have become more efficient since then. The impact of using the improved boiler model on SAP was tested on a national survey dataset (English Housing Survey 2021). There was no significant difference in the mean overall SAP rating between models which gives confidence that the change in methodology has not impacted the historic timeseries.

1.5 Fuel Poverty

1.5.1 Changes to modelling from 2021

The 2021 key findings report was the first to include fuel poverty estimates which fully met the definition of fuel poverty as laid out in the Fuel Poverty (Targets, Definition and Strategy)(Scotland) Act 2019, the Fuel Poverty (Enhanced Heating) (Scotland) Regulations 2020 and Fuel Poverty (Additional Amount in respect of Remote Rural Area, Remote Small Town and Island Area) (Scotland) Regulations 2020.

However, due to the change in the mode of approach and data collection, as well as bias in the sample in 2021 (for further details see Chapter 6 of the 2021 SHCS key findings report), the rates of fuel poverty in 2021 are not comparable to previous or current waves of the survey, and are therefore not included in this report.

The key differences between the previous fuel poverty methodology and the updated methodology, used since 2021, are presented below in Table 1.8. For a full description of the updated fuel poverty calculation methodology see the 2022 Methodology Notes.

Table 1.8 differences in methodology between previous fuel poverty calculations  and updated methodology[3]

Component of definition	2012-2019	2021 - present
Income	Highest income householder (HIH) and Spouse	HIH, Spouse, and up to three other adults
Housing costs	Only those reported	Imputed if missing
Heating regime	2 regimes	4 regimes
Childcare costs	No	Included
Minimum income standard (MIS)	11 publicly available	Full 107

​​​​​​​1.5.2 Income after housing costs including council tax

For the 2017 Scottish Household Survey (SHS), an updated set of questions collecting council tax information were incorporated and accounted for in fuel poverty analysis. Previously respondents were only asked to provide what they paid in council tax whether or not they received any deductions or reductions. The survey now distinguishes between reported council tax after any deductions or reductions, and full council tax. This reduces the risk of double counting Council Tax Reduction in household income in the former case.

​​​​​​​1.5.3 Income including Cost of Living Payments

For 2023 household income used in the calculation of fuel poverty also includes an adjustment to account for the cost of living payments as paid in the 2023 calendar year. This includes the £900 Cost of Living Payment for households on means tested benefits, the £150 Disability Cost of Living Payment for persons in receipt of select disability benefits, and the £300 Pensioner Cost of Living Payment for households in receipt of Winter Fuel Payment.

Unlike other forms of income such as earnings or benefits, income received by households for any cost of living payments was imputed based on household eligibility[4]. This was done through cross referencing household benefit data gathered through the SHS with any household that met the eligibility criteria of a cost of living payment having that amount added to their income[5].

These adjusted incomes where then used for all calculations in the fuel poverty chapter including: the overall fuel poverty rate, the fuel poverty gap, and household income statistics.

​​​​​​​1.6 Energy Bill Support Scheme (EBSS)

Between October 2022 and March 2023  the Energy Bill Support Scheme provided a £400 discount to each house with a domestic energy connection. In order to account for this while modelling fuel poverty all households had their modelled energy bill reduced by £201, (the amount received for the EBSS in the 2023 calendar year).

These adjusted energy bills where then used for all calculations in the fuel poverty chapter including: the overall fuel poverty rate, the fuel poverty gap. Importantly however, while this methodology reduced fuel bills it did not reduce energy consumption. Therefore any reference to energy consumption reflects the typical approach to modelling fuel bills as set out in section 1.3.

​​​​​​​1.7 Warm Home discount (WHD)

The Warm Homes Discount (WHD) scheme was launched in April 2011. Energy suppliers are mandated to provide support in the form of discounts and rebates, as well as advice and assistance, to fuel poor customers.

The SHCS does not collect information on whether respondents receive direct financial support under this scheme. In fact, it would be difficult to collect such information as many people are not aware that they are benefiting from a rebate. However, unless this is accounted for in the survey, the modelled fuel bill and therefore fuel poverty would be overestimated.

The publication of the 2014 SHCS Key Findings report introduced an allowance for the WHD rebate in the estimation of the number of fuel poor households in Scotland and the 2014 Methodology Notes contain a detailed description of the methodology. This was based on modelling households’ eligibility for the scheme. This method has been used in all subsequent Key Findings reports.

The approach consists of the following stages:

• Details of the number of households in receipt of each component of the WHD are provided by Ofgem for GB as a whole. It is assumed that the number of recipients in Scotland is proportional to Scotland’s share of households in GB (9.2%).
• Details of eligibility for each element of the WHD provided by Ofgem, are used to flag all households in the SHCS dataset who meet these criteria. Because of limitations in the available survey information, some approximations are necessary.
• A series of runs are made, where a sample of likely recipients is drawn at random from the pool of all eligible households. For each sample the WHD rebate amount (£150 from 2022) is subtracted from the modelled household fuel bill. The estimated number of households in receipt of the Core and Broader Group element of the WHD in Scotland is used to constrain the size of the sample which is selected.
• A representative iteration based on the number of fuel poor households among modelled recipients is selected from all runs as the best estimate of the set of household in the survey who benefit from the Core or Broader Group element of the WHD scheme.

Elements of the WHD scheme and eligibility criteria

For the 2023 SHCS we have continued to use the WHD eligibility criteria as set out in Warm Home Discount Scotland: Guidance for Suppliers 2022 – 2026. The eligibility criteria for the WHD Scotland scheme is divided into two groups: Core and Broad.

Each uses different criteria for eligibility and the corresponding information is not always collected in the SHS interview and in some cases, it has been necessary to simplify the criteria or make certain assumptions in order to determine if a household would be eligible for support under the scheme.

Under the Core Group (CG) element households receive an electricity bill rebate, currently worth £150. Eligibility criteria have changed in the period the scheme has been in operation. This is summarized below:

• Year 1 (2011-2012): Recipients of Pension Credit Guaranteed Credit only
• Year 2 (2012 – 2013): Recipients of Pension Credit Guaranteed Credit only and households where someone receives both Guaranteed and Savings Credit and is over 80 years old
• Year 3 (2013- 2014): Recipients of Pension Credit Guaranteed Credit only and households where someone receives both Guaranteed and Savings Credit and is over 75 years old
• Year 4 (2014- Current): Recipients of the Guaranteed Credit element of Pension Credit irrespective of whether they receive the savings elements of pension credit.

The SHCS collects information on whether the household reference person (HRP) or their partner receives Pension Credit, however no detailed information on the particular elements of Pension Credit is collected. For this reason, all Pension Credit recipients in the survey have been assumed to be eligible for the Core Group of WHD. This is a larger and slightly better-off group of households than the households who would in reality be eligible for the Core Group element. This means that WHD recipients in the survey are selected from a broader pool and there is a risk that the effectiveness of WHD in targeting fuel poor households is understated in the modelling. This also means that the impact WHD has in helping reduce the level of fuel poverty may be understated.

Broader Group

The Broader Group forms part of a compulsory supplier’s non-core obligation. It obligates suppliers to identify Scotland domestic customers to provide a rebate to. The guidance states that customers should be in fuel poverty or a fuel poverty risk group and not captured under the Core Group. Generally, the criteria for the broad group is one of:

• A range of benefits covering disability, low incomes and job-seeking; and
• A range of vulnerability characteristics such as having young children, elderly and infirm household members or low incomes.

The 7 Broad group criteria are set out below in Table 1.9

Table 1.9: Warm Home Discount Compulsory Broader Group qualifying benefits

Mandatory Criteria	Plus	Additional Criteria
A Person who receives Income Support	And	(a) has parental responsibility for a child under the age of 5 who ordinarily resides with that person. Or (b) receives any one of the following in addition to Income Support: - Child tax credit which includes a disability element; - A disabled child premium; - A disability premium, enhanced disability premium or severe disability premium; - A pensioner premium or higher pensioner premium
A person who receives Income related Employment and Support Allowance (IR ESA) which includes a support component	And	(a) has parental responsibility for a child under the age of 5 who ordinarily resides with that person. Or (b) receives any one of the following in addition to Income related Employment and Support Allowance: - Child tax credit which includes a disability element; - A disabled child premium; - A disability premium, enhanced disability premium or severe disability premium; - A pensioner premium or higher pensioner premium
A person who receives IR ESA and is a member of the work-related activity group	And	(a) has parental responsibility for a child under the age of 5 who ordinarily resides with that person. Or   (b) receives any one of the following in addition to Income based Jobseeker’s Allowance:   - Child tax credit which includes a disability element;    - A disabled child premium;    - A disability premium, enhanced disability premium or severe disability premium;    - A pensioner premium or higher pensioner premium
A person who is in receipt of Income based Jobseeker’s Allowance	And	(a) has parental responsibility for a child under the age of 5 who ordinarily resides with that person.   Or   (b) receives any one of the following in addition to Income based Jobseeker’s Allowance:   - Child tax credit which includes a disability element;   - A disabled child premium;   - A disability premium, enhanced disability premium or severe disability premium;   - A pensioner premium or higher pensioner premium
A person who is in receipt of Housing Benefit	And	(a) has parental responsibility for a child under the age of 5 who ordinarily resides with that person. Or (b) receives any one of the following in addition to Housing Benefit:  - Child tax credit which includes a disability element; - A disabled child premium; - A disability premium, enhanced disability premium or severe disability premium;  - A pensioner premium or higher pensioner premium
A person who is in receipt of Universal Credit, has an earned income not exceeding the relevant periodic amount[6] in at least one relevant assessment period	And	(a) has limited capability for work or limited capability for work and work-related activity; Or (b) is in receipt of the disability child element Or (c) has parental responsibilities for a child under the age of 5 who ordinarily resides with that person.
A person who is in receipt of Child Tax Credit by virtue of an award which is based on an annual income not exceeding the relevant annual amount	And	(a) has parental responsibilities for a child under the age of 5 who ordinarily resides with that person; Or (b) is in receipt of child tax credit which includes a disability element, Or (c) is in receipt of a disabled child premium

Not all eligibility criteria used in the Broad Group element are available in the SHCS.

For example, the following have not been included in the modelling: Child Tax Credit which includes a disability or severe disability element, disability premium, disabled child premium, pensioner premium, or if a person has limited capability for work or limited capability for work and work-related activity[7].

For Child Tax Credit which includes a disability or severe disability element, disability premium, and disabled child premium we have used if a household has reported a long-term health condition (in addition to receipt of child tax credit) as proxy in these cases. Similarly, in the case of limited capability for work we have used if a respondent has reported a long-term health condition and is not in work as a proxy.