Guidance

Social housing lettings 2020 to 2021: quality report

Published 28 July 2022

Applies to England

Introduction

This Quality Report provides a brief summary of key issues relating to the quality of DLUHC’s annual Social Housing Lettings, England statistical release.

Social Housing Lettings carries the National Statistics badge, the quality mark of government statistics, after assessment by the UK Statistics Authority in 2014.

We collect data and present the statistical release and associated products in line with the principles of the Code of Practice for Statistics to engender trust in our data and encourage using this report as a reliable source of social housing data.

This Quality Report covers the fourteen principles under the three pillars of the Code: trustworthiness, quality and value.

1. Trustworthiness

T1: Honesty and integrity

1.1.1 Data is collected, processed and quality assured by data collection officers and professional statisticians. The team applies rigorous quality assurance procedures and works with users to ensure data are used appropriately, abiding by the Code of Practice for Statistics.

T2: Independent decision making and leadership

1.2.1 DLUHC statisticians abide by the Code of Practice, keeping pre-publication access to the data to those involved in the report’s creation and ensuring the statistical integrity of content. The Department’s Head of Profession for Statistics is engaged as necessary.

1.2.2 The report meets GSS standards of statistical impartiality, separating statistical commentary from any political, press or ministerial statements. Subsequent statements by the government using data from this report quote this source and non-governmental users are encouraged to do the same.

T3: Orderly release

1.3.1 The precise publication date for Social Housing Lettings is pre-announced on the GOV.UK statistical release calendar well in advance of the 4 weeks required by the Code of Practice.

1.3.2 The report and accompanying products, including this Quality Report, is published on GOV.UK at 9:30am on publication day and are equally accessible to all.

1.3.3 Pre-publication access to the report, underlying data and messaging during development is restricted to those involved in the report’s creation and publication and only at the point at which their input is required.

1.3.4 In accordance with the Pre-release Access to Official Statistics Order 2008, eligible members of DLUHC staff are granted pre-release access during the 24-hour period before publication. This is for the sole purpose of enabling those who could reasonably be expected to be held to account and answer questions immediately upon release of the statistics, as stated in the Order. The pre-release access list is kept to the minimum number of people possible and is published at the same time as the report .

1.3.5 Social Housing Lettings follows the Department’s Revision Policy which adheres to the Code of Practice.

1.3.6 Unscheduled revisions are made when a substantial error has occurred as a result of the compilation, imputation or dissemination process. Revised products are released as soon as is practicable, alongside an explanatory note on both the cause and impact of the error. Data are clearly indicated as being ‘revised’ (R) or ‘provisional’ (P).

1.3.7 The only scheduled revision for Social Housing Lettings is the annual revision to the weights once final LAHS data are published. Revisions are made in the subsequent release in the historic tables. For example, 2018/19 weights were revised after the publication of LAHS 2018/19 in January 2020. The summary tables published as part of the 2019/20 Social Housing Lettings use the revised weights for the 2018/19 figures in the time series tables.

T4: Transparent processes and management

1.4.1 Substantial financial and administrative resources are employed to enable this data collection and effective quality assurance, including for further development of the report each year in light of new user requirements or new methodology / collection possibilities.

1.4.2 We are transparent about our methodology and approach to quality, as evidenced by this Quality Report.

T5: Professional capability

1.5.1 Data is collected, processed and quality assured by a team of statisticians and data collection officers. Each individual has appropriate CORE-specific expertise, analytic capabilities and data protection awareness for their roles. All statisticians are professional GSS-badged statisticians.

1.5.2 The process of weighting and imputation of data is done by statisticians in the department. In 2013 the ONS Methodology Advisory Service were commissioned to advise on improving the methodology to derive the Social Housing Lettings statistics, with funding from the UK Statistics Authority Quality Improvement Fund. The main methodology changes were to estimate population totals for social housing lettings by applying improved weighting methods, and address item non-response for the household characteristics (such as age, ethnicity and economic status) by imputing for missing values.

T6: Data governance

1.6.1 All professionals involved in the creation, storage and use of the underlying data for Social Housing Lettings are well-versed in data protection and operate in compliance with data protection legislation.

1.6.2 The CORE website through which the data is collected has been security accredited in line with government standards. As the data is collected at record level (i.e. individual tenancies) and contains personal data, appropriate steps are taken to protect the data.

1.6.3 Upon the introduction of GDPR in May 2018, we made changes to the website, to improve the accountability and auditability of data providers and DLUHC statisticians using the data.

1.6.4 All data providing organisations must sign a data sharing agreement with DLUHC and set up a Data Protection Officer on the CORE website who must confirm they have signed, and abide by, the data sharing agreement.

1.6.5 Also, before each log can be submitted the data provider must tick a box to confirm that the tenant has either been shown the DLUHC CORE privacy notice6 or has been shown the data provider’s privacy notice which contains a link to the DLUHC version. The privacy notice informs tenants why their data is being collected, how it used and how their confidentiality is protected.

1.6.6 We worked with our in-house data protection team and legal colleagues to develop these processes and will continue to monitor how the legislation is being applied more broadly, to ensure we meet our legal obligations whilst minimising the reporting burden for our data providers.

1.6.7 Onward sharing of personal data by DLUHC is restricted to the organisations and purposes listed in the DLUHC CORE privacy notice which is made available to tenants. The small number of organisations receiving CORE data must sign a data sharing agreement with DLUHC setting out the terms of their access and use.

1.6.8 As detailed in our disclosure policy we publish the maximum amount of data available in our Social Housing Lettings report and associated products, whilst abiding by data protection legislation and utilising GSS best practice for statistical disclosure control (e.g. banding sensitive variables).

1.6.9 Record level information for each tenancy is deposited with the UK Data Archive. The UK Data Archive follows strict security protocols to store and access the datasets. Three licences are available:

• End User Licence is for access to datasets that are not private information as defined by the Code of Practice for Official Statistics.
• Special Licence and Secure Access are for access to micro-data with ‘private’ information as defined by the Code of Practice for Official Statistics.

2. Quality

Q1: Suitable data sources

2.1.1 CORE data is collected from housing providers – both local authorities (commonly called ‘council housing’) and private registered providers (commonly called ‘housing associations’).

2.1.2 CORE data comprises three key themes: information about the tenants, tenancies, and properties.

2.1.3 Housing providers are the direct creators (and holders) of the information about tenancies and properties. Often, they need this information themselves for day-to-day management of their stock and lettings. By collecting this data from housing providers, we are collecting data directly from the source.

2.1.4 Some information about the tenant can only be answered by the tenant themselves, e.g. self-reported reason for leaving last settled home. Whilst ideally we would collect this information directly from the tenant, central government has no existing interface with tenants when they get a new social housing let. Since housing providers have at least one face to face interaction with tenants and housing providers also hold the information we require on tenancies as properties, the most efficient method to collect the CORE-specific tenant questions is for housing providers to ask tenants at sign-up, i.e. when tenants move into the property.

2.1.5 To ensure data quality, the data submitted by housing providers is subject to in-form validations at the point of submission and is subject to further quality assurance checks later in the statistical processing.

2.1.6 No other existing data source, either published or held internally by governmental or other central bodies, contains the breadth of information collected by CORE. It is the only census of all new lettings in England, and the only data source across the social housing sector collected at individual tenancy level.

Q2: Sound methods

2.2.1 CORE is designed to be a complete census of new social housing lettings provided by local authorities and private registered providers that own social housing stock. Whilst data providers should submit data for all new social housing tenancies, this does not always happen in practice. In addition, not all questions are compulsory so information may be incomplete for some tenancy records. This introduces bias into statistics using the data which we minimise by applying weighting and imputation processes.

2.2.2 Weighting is applied to adjust for record level non-response by local authorities. Imputation is applied to compensate for item non-response for both local authorities and private registered providers. Weighting

2.2.3 CORE has been a mandatory data collection on the Single Data List for private registered providers in England since its inception, however local authorities have participated in CORE since 2004/5 on a voluntary basis. In the first year, only 24% of stock-holding local authorities participated, but this has increased to 86% in 2020/21.

2.2.4 Weighting is applied to adjust for record level non-response by local authorities. Local authority weights are calculated by reference to the total number of lettings reported to the Local Authority Housing Statistics (LAHS) return. Generally, for the most recent year of data, provisional estimates are used as final LAHS data will not be published until later in the year.

2.2.5 Weights are calculated for groups of similar local authorities, where these groups are defined by the ONS UK area classifications. For the publication of 2020/21 Social Housing Lettings the area classifications based on data from the 2011 Census were used (the classifications have not yet been updated to use the 2021 Census). Prior to 2014/15 the clusters were based on 2001 Census data. The variables used to create the area classifications are directly relevant to the factors of interest for social housing (e.g. age, nationality, employment status).

2.2.6 For each of the 10 geographical clusters, the weight is calculated for all responding local authorities as the ratio of the number of lettings judged to be the most reliable (either reported to LAHS, reported to CORE or a projection if neither are available), relative to those reported to CORE.

2.2.7 Currently weights are not calculated for private rented providers as response is high and there is no alternative data source robust enough to be used for CORE weighting purposes. For instance, whilst the Statistical Data Return (SDR) published by the Regulator for Social Housing holds includes the number of lettings by private registered providers, the Regulator’s coverage is different to that of CORE. We will continue to review this assumption over time.

2.2.8 Weights are also not calculated for Affordable Rents as there were too few lettings of this type to create robust weights.

2.2.9 Weights are only suitable for use when conducting analysis at the national level. When conducting analysis on sub-national geographies such as local authority areas the weights should not be used. Imputation

2.2.10 Some of the questions on the CORE questionnaire are not compulsory. In particular, data on household characteristics may not be available to the housing officer or may be refused by the tenant. In these cases, we impute responses.

2.2.11 Imputation levels are low – in 2020/21 we imputed around 4% for age and gender of the lead tenant and around 9% for economic status, ethnicity and nationality of lead tenant. We impute slightly more for private registered providers than local authorities. We continually work with data providers to improve completion rates.

2.2.12 Our imputation methods are based on the recommendations of the ONS Methodology Advisory review in 2013. We use the Canadian Census Editing and Imputation System (CanCEIS) software to impute data using a ‘donor imputation’ approach. This involves identifying records that are similar to the records with missing data. The software then randomly chooses a record (the ‘donor record’) from the set of records with the closest data for the non-missing variables and copies the data for the missing variables from the donor record.

2.2.13 As with the weighting process, the imputation process uses ONS UK area classifications to group local authorities into clusters – one of the variables used to judge records’ similarity. The imputation is carried out separately for general needs and for supported housing, to reflect the different demographic profiles of their tenants.

2.2.14 Imputation is done in two steps, imputing first age, gender and economic status, and then nationality and ethnicity, on the missing variables for the lead tenant. In most cases the imputation methodology prevents using donor data that would normally fail the usual validation process.

2.2.15 We introduced imputation for Affordable Rent lettings for the first time in 2017/18 once there was a high enough number of complete Affordable Rent records to act as donor records.

2.2.16 Income data is not imputed. Income has a higher non-response rate, particularly amongst the elderly when it was originally reviewed in 2013. Therefore, the data was deemed to be too incomplete and unreliable as imputed results may lead to biased estimates. Instead, to improve completeness of income data the department is working with data providers to encourage a higher response rate.

Q3: Assured quality

The diagram below summarises our quality assurance processes:

1. START: new lettings form published on the CORE website in April

2. Supporting documentation published (guidance on form changes etc)

3. Information collected from tenants and individual CORE logs completed and uploaded to website

• Logs automatically validated on submission
• Data quality issues (errors) raised during process queried to data providers
• CORE helpdesk resolves queries

4. Monitoring of response rates based on number of logs submitted last year

5. End of reporting year 31 March 2020

6. Imputation and weighting carried out

7. END: publication 28 July 2022

Sampling error

2.3.1 CORE is an administrative data source so there is no sampling error in statistics based on CORE data as there is no sample.

Coverage error and non-response bias

2.3.2 Coverage error is where the actual population studied does not match the required population for which you want to draw conclusions.

2.3.3 Non-response bias is where non-respondents differ from respondents meaning conclusions drawn from the data are not representative of the whole population.

2.3.4 See ‘Q2 – Sound methods’ for information on coverage of CORE and steps to mitigate record level non-response (weighting) and variable level non-response (imputation).

2.3.5 On-going monitoring of response rates at organisation level based on number of logs submitted in previous years takes place throughout the year, with targeted communications to those submitting an unexpected volume of data.

Measurement error

2.3.6 Measurement error is where errors may arise due to inherent difficulties in observing, identifying and recording required information. It may occur randomly or may reflect systemic problems experienced by data providers.

2.3.7 Most data captured by CORE is collected via a face-to-face tenant questionnaire by a housing officer at the point they start their tenancy. This aims to minimise the risk of poor quality or incomplete data by ensuring the most appropriate person completes the data, and to minimise the risk of poor memory recall.

2.3.8 To minimise the risk of misinterpretation, guidance materials are available on the CORE system and are well-publicised through newsletters. These include notes within the form as well as the overarching manual and online user guides on how to submit data. The CORE helpdesk assists data providers with more bespoke queries. Periodic face-to-face training sessions are also available to data providers to help them improve their understanding of the questions and use of the data.

Processing error

2.3.9 Processing error is where errors arise in the submitting, coding and internal processing of the data.

2.3.10 Numerous validation checks are built into the data collection programme to reduce the number of submission errors. These range from incompatibility checks, e.g. male-only households containing a pregnant woman, to testing for extreme values, e.g. high rents based on number of bedrooms and local authority area.

2.3.11 Rigorous internal user testing is conducted on the in-form validations and calculation of derived variables in the system before each new form goes live on 1 April. Any errors identified following this are rectified as soon as is practically possible, with any records submitted before this point re-tested under the new validation checks and queried with data providers as necessary.

2.3.12 Further quality assurance checks are run by statisticians within the team each quarter with erroneous values queried with data providers.

3. Value

V1: Relevance to users

3.1.1 CORE is a unique source of information in providing detailed information about individual tenancies. It is the only source of social housing lettings data that allows detailed multivariate analysis of households and dwellings and is a valued source by a variety of users for wide-ranging purposes.

3.1.2 Social housing lettings statistics and the underlying CORE datasets inform national housing policy in areas such as social housing allocation policy, homelessness reduction and as part of the evidence base for the Domestic Abuse Bill. Data on housing costs, housing-related benefits and affordability is used to inform modelling on the value for money and societal impact of investment into new affordable housing, e.g. via the Affordable Homes Programme. The Department uses CORE data on demographics of households in new social lettings to inform equality impact assessments, e.g. for changes to eligibility requirements following the UK leaving the European Union.

3.1.3 CORE data is also used to fulfil the Department’s parliamentary scrutiny obligations through Parliamentary Questions (PQs) and to meet government transparency aims through Freedom of Information requests.

3.1.4 The Department’s Arms’ Length Bodies use CORE data in order to carry out their functions. For example, Homes England use the data to monitor government investments and schemes, and the Regulator of Social Housing use CORE data in their regulatory activities.

3.1.5 Other government departments also use CORE data. For example, CORE data is a component of the key government inflation measure – the Retail Price Index (RPI) published by the Office for National Statistics (ONS). Another example is our collaboration with the Ministry of Defence (MoD) on social housing provision for armed forces personnel to inform their Armed Forces Covenant which aims to ensure members of the Armed Forces are not disadvantaged in their access to government-provided services.

3.1.6 Social housing providers use the data to understand the market for social housing and benchmark their own performance. Local government also use these data to inform their Strategic Housing Market Assessments that form part of the National Planning Policy Framework.

3.1.7 Academics, researchers, charities and the wider public used these statistics to understand social housing issues. Some of these users also use the underlying administrative data available at the UK Data Archive.

V2: Accessibility

3.2.1 Data is free and equally available to all, published on GOV.UK with no restrictions to access. Underlying data, including additional maps and an interactive dashboard for subnational and housing provider-level analysis, is published alongside the National Statistics release. Data is published up to the extent data protection legislation allows, with banded variables where we cannot provide exact figures. Commentary is objective, and a range of graphical visualisations are used to aid comprehension.

3.3.1 DLUHC offers a facility to obtain data in alternative formats (e.g. Braille, large print or audio). For this, users should to quote the title of the publication, their address and telephone number and send to alternativeformats@levellingup.gov.uk.

V3: Clarity and insight

3.3.1 Commentary on the current size and shape of the life sciences sector is objective, focussing on impartial statistical messages. Charts, visualisations and maps are used to illustrate these. Headlines and key charts are on the front page, with a summary graphic published both in the release and as a separate document to increase reuse by users.

3.3.2 Statistics from other sources such as the Local Authority Housing Statistics are throughout the release to put Social Housing Lettings in context.

3.3.3 Time series are used where possible to provide further insight on how social housing lettings have changed over time. If changes in question design or methodology prevent this, it is footnoted in the tables.

V4: Innovation and improvement

3.4.1 Each year we undertake a ‘log review’ where through an open and transparent user engagement we determine any changes to the questions, response categories and guidance notes. Some changes may be made because of new operational needs of registered providers. Decisions are made by balancing user needs against the additional reporting burden for housing providers.

3.4.2 We invite feedback on our publication and liaise with CORE training providers to gather any learning from training sessions about housing providers’ needs from the data. Throughout the year, the department also maintains a list of any issues raised by the users and suppliers of the data and feeds those into the ‘log review’.

3.4.3 Changes for the following reporting year are usually agreed around 3 months in advance of the new data collection going live in April. This allows data providers to implement the required changes in their internal management systems to provide the required data.

3.4.4 In 2017/18 we considerably revamped the format of the Social Housing Lettings 2017/18 publication to improve users understanding of the relevance of this set of social housing statistics. We also incorporated additional content on stock, utilising other data sources, to make the context in which Social Housing Lettings sits clearer.

3.4.5 In 2018/19 we brought in additional analysis on waiting lists and vacancy rates, incorporating further material from the Local Authorities Housing Statistics (all LA-owned social lettings) and the Statistical Data Return (all PRP-owned social lettings).

3.4.6 In 2019/20 we replaced our suite of subnational tables (Local authority area level, housing provider level, and housing provider by local authority area level) with an interactive Power BI dashboard to enable users to undertake their own analysis.

3.4.7 In 2020/21 we released an interim report covering the first 6 months of the year (April to September 2020) to give early insight into the effect of COVID-19 pandemic on new lettings. This included new data series on lettings by month and average re-let times by month. The annual 2020/21 National Statistics release continues this narrative.

V5: Efficiency and proportionality

3.5.1 Other key sources for information relating to social housing lettings in England include the annual releases of:

• Local Authorities Housing Statistics (LAHS), by the Department
• Statistical Data Return (SDR) of private registered providers, published by the Regulator of Social Housing
• English Housing Survey (EHS), by the Department

3.5.2 CORE meets user needs that cannot be filled by these publications and so the public good of a robust evidence base of local level information on new social housing lettings is deemed of suitable value to warrant the CORE collection and publication of Social Housing Lettings.

3.5.3 CORE is a ‘flow’ measure which records data on new social housing lettings at record level, whereas data in LAHS is a ‘stock’ measure of all social housing stock in local authorities, at aggregate level.

3.5.4 The SDR annual return is only completed by private registered providers and information collected includes data on size and type of home, location and rents over the year. Data on lettings in also capture for providers with 1,000 homes or more, unlike CORE which captures all private registered providers. The SDR focuses on the property itself and does not collect information on the socio-demographic characteristics or housing circumstances of the household, which is captured in CORE.

3.5.5 The EHS provides social housing lettings data, but the sample size of the survey leads to results which are not robust enough to be used at the local authority level, unlike CORE.