Dear Roger

SCOTTISH HOMELESSNESS STATISTICS

 As you are aware we recently completed our checks of compliance with the Code of Practice for Statistics for two of Scottish Government’s sets of housing statistics:

• Homelessness in Scotland – currently published as National Statistics annually, with a supporting bi-annual update; and
• Housing Options (PREVENT1) Statistics in Scotland (PREVENT1) – currently published as Experimental Statistics. We chose to consider these in the context of considering the value added by each statistic output to the broader topic of homelessness in Scotland.

I am pleased to confirm that the Homelessness statistics should continue to be designated as National Statistics and the PREVENT1 statistics as experimental statistics (subject to your own review of their status) but we would expect to see evidence of ongoing improvements in response to this letter.

We reviewed the statistics against the three pillars – trustworthiness, quality and value – that are fundamental to supporting public confidence in statistics and form the basis of the Code. We recently had a very constructive conversation with the responsible statisticians about our review’s findings and the current plans for these statistics. This letter confirms our review’s findings.

We found a range of positive examples around Scottish Government’s statistics production relating to user engagement, ongoing work to quality assure the data received from local authorities, and recent or planned innovations to develop improved insights:

• The annual Homelessness Statistics Users Group hosted by Scottish Government which has broad membership of data providers and statistics users, and is an effective forum to discuss ongoing or planned developments. Your statisticians told us that at the most recent meeting the group agreed it would be valuable to meet bi-annually in the future, which we think is positive. Your statisticians also told us that they supplement this formal engagement with written updates to users via ScotStat.
• The statisticians’ ongoing work with local authorities to quality assure the quarterly data submissions. This includes developing automated validation checks to identify potential errors for further investigation.
• Assessing the robustness of data received via the new ‘HL3’ case-level temporary accommodation returns compared to ‘HL2’ aggregate temporary accommodation returns, and assessing differences in the proportion of entries that can be linked at a case-level to the ‘HL1’ homeless applications data return. This is with a view to ceasing the ‘HL2’ data collection once the statisticians are content that the ‘HL3’ returns are of sufficient quality and can provide equivalent statistics.
• Scottish Government’s recent published research exploring the relationship between homelessness and health, by linking homelessness and health datasets for the first time at a national level. We think this is a great example of innovation, providing really important insights to understand more about people classed as homeless, the instances of repeat homelessness and their interaction with health services. It provides a strong foundation for developing further insight and intelligence to inform policy development.
• Developing and publishing the PREVENT1 publication as Experimental Statistics since 2010/11 to reflect a new data collection on the operation of Housing Options services made available when households seek assistance for housing-related issues. Your statisticians told us that they were questioning the statistical value of this publication, given different approaches to Housing Options services across Scotland. They would be exploring this further with policy colleagues and external stakeholders, to inform a decision about the future of this data collection and statistics. We note that Experimental statistics should be evaluated to determine whether they should be produced permanently, or withdrawn, under the criteria set out in this guidance.
• Planned collaboration with National Records of Scotland to develop and publish new statistics on homeless deaths in Scotland, seeking to replicate the methodology used by the Office for National Statistics in developing its own experimental statistics on Deaths of homeless people in England and Wales.

We also found a number of areas where improvements to the production and presentation of Homelessness in Scotland would enable you to meet the highest standards expected of National Statistics; and we suggest you could take into account when considering the future of the Experimental PREVENT1 statistics:

• The statisticians told us they have recently undertaken an assessment of the extent to which existing data governance policies continue to meet new data protection legislation. We therefore think there is scope for you to review and update existing published information about data governance, particularly considering any personal information included in case-level data submitted by local authorities. This is particularly important given recent work to link case-level data from health and homelessness datasets. In doing so, we encourage you to consider relevant findings from the Authority’s Joining Up Data for Better Statistics
• Your statisticians also told us that they work closely with local authorities and conduct a range of validations to quality assure the data, and are also considering moving to use a new automated data collection system to further improve the process. An important aspect of offering reassurance around your statistics is to describe your quality assurance arrangements and the implication for the quality of the statistics. Therefore, you should review and update published information about the methods used to produce the statistics, the quality and quality assurance of the statistics including their strengths and limitations to reflect these recent activities. We recommend you draw on the Authority’s Quality Assurance for Administrative Data Regulatory standard to demonstrate how each data source is assured to inform your guidance to users about the robustness of the statistics.
• We found that accessibility between the main statistics publications, supporting guidance and data tables across Scottish Government’s new website and the old website was not intuitive, with some circular loops from hyperlinks to data tables and guidance between the new and old website, making navigation difficult. Your statisticians told us that there are known limitations with the new website at present, which doesn’t offer obvious capabilities to present and link to supporting information for statistics publications. They also told us there is a current project within your corporate team to develop solutions to this limitation and move all relevant content over to the new website in due course. Given the high-profile nature of homelessness in Scotland, we strongly encourage you to prioritise these statistics as part of the planned solutions being considered.
• Increase the insight presented through the presentation of Homelessness in Scotland, taking into account your consideration of, and future decision about whether to cease publication of the PREVENT1 statistics and any useful insights presented within that publication. We suggest that improvements could include:
• Refining the commentary to focus on drawing out the key messages from the statistics, seeking opportunities to simplify the language to increase its clarity for a wider (non-technical) audience.
• Review the suitability of the charts and diagrams presented to ensure these are the most effective ways of aiding users’ interpretation of key messages and trends.
• Seek opportunities to present more insight on the topic of homelessness and to explain trends over time, for example by drawing on recent developments such as Scottish Government’s research based on linking homelessness and health datasets.
• Strengthening guidance about the comparability of the statistics with related statistics across the UK. While tackling homelessness is a devolved matter, some users will want to understand trends in homelessness across the UK. We understand that work in this area is underway, which we welcome.

In doing this, we encourage you to continue engaging users in developing your proposed improvements to ensure that the value of the statistics is maximised.

Our Housing, Planning and Local Services Domain Lead will continue to engage with your team on progress in the coming months. Your statisticians have confirmed they will seek support from the Government Statistical Service Good Practice Team in improving the presentation of these statistics, who would be happy to work with you.

Please do not hesitate to get in touch if you would like to discuss any aspects of this letter further.

I am copying this letter to Susan Carsley and Lee Bunce, the responsible statisticians and Adam Krawczyk, Head of Housing, Homelessness & Regeneration Analysis at the Scottish Government.

Yours sincerely

Mark Pont
Assessment Programme Lead