Guidance

DSIT quality management approach for official statistics

Updated 12 May 2026

1. Introduction 

This policy documents the approach to quality management for official statistics produced by the Department for Science, Innovation and Technology (DSIT), in compliance with the following standards set out in the Code of Practice for Statistics, Standards for Official Statistics: 

• Standard 5: Producers must support a quality culture that promotes good practice and encourages learning and improvement, under the direction of the Chief Statistician/Head of Profession for Statistics   

• Standard 6: Producers must use suitable data sources and sound methods, and assure the quality of the statistics across the production and release processes   

This policy is designed to ensure that all DSIT official statistics publications use a transparent and consistent approach to quality management. This will be reviewed at least annually. 

This policy applies to official statistics (including accredited official statistics and official statistics in development) produced in DSIT. It does not apply to releases of other forms of data and statistics (although it may be applied voluntarily in these cases). It does not apply to external research outputs, which instead follow the relevant professional guidance for those types of publication. However, this policy is consistent with the approach to quality assurance that is set out in the Aqua Book: guidance on producing quality analysis.

2. QA logs 

Each publication is required to maintain a quality assurance (QA) log. The checklist items on this QA log are to be agreed after examining previous QA logs and error logs, where these logs are available. The checklist items should list all the QA steps that are anticipated to be required in the production of the statistics.  

Producers are encouraged to follow the Government Statistical Service (GSS) guidance on Quality Statistics in Government and the Office for Statistics Regulation (OSR) toolkit on the Quality Assurance of Administrative Data when determining steps to include in their QA logs. Examples of QA logs are made available to help producers with this process. 

QA logs should record who has performed each step on the QA. Where there is follow-up work after a particular QA step, a brief description of this work should be included in the log.  Checklist items are to be agreed in advance so that an appropriate amount of time for QA to be conducted can be accounted for as part of the overall production timeline.   

Where QA steps have been prioritised, which QA steps have and have not been taken place are to be recorded accurately so that an informed decision can be made later as to whether sufficient QA has taken place.  

3. Clearance statement for official statistics  

A clearance statement for Official Statistics publications is to be produced and signed off by the responsible analyst before the statistics can be published.  

The clearance statement should: 

• confirm that QA has taken place and contain (or contain a link to) a summary of the QA checks that have been conducted 

• highlight any key changes e.g. in scope or methodology, and weaknesses in or caveats to the final statistics  

The clearance statement does not need to be a comprehensive description of all the QA that has taken place but should give a fair and honest sense of the quality of the statistics and document key decisions taken.   

An approved clearance statement should then form part of the set of documents considered before final sign-off by a senior statistician. 

4. Error logs 

Each business area is encouraged to maintain an error log, either for each publication or covering the work of the wider team. This log is to document errors identified in releases that have been published on GOV.UK. Where errors arise, these will be handled as described in the DSIT Revisions and Corrections policy. 

Errors are likely to be things which could be avoided with the appropriate QA step. Therefore, when agreeing the QA checklist in preparation for a new edition of a statistical publication, producers are advised to look at the current error log as well as the previous QA logs to determine whether a QA step should be included in the new QA log.  

A key part of maintaining an error log is documenting the learning process resulting from handling these incidents. The error log should therefore include details of the actions taken in order to rectify the error and lessons learnt for future productions.   

An example of data fields that we would include in a DSIT error log is provided below:

• statistical series
• output format (eg Official Statistics release, tables, PQ, open data, published ad hoc, internal ad hoc)
• date error occurred
• date identified
• lead statistician
• description of error
• how was the error identified?
• how significant was the error (reflect scale and impact)?
• what actions were taken to rectify the error?
• did the error require an unscheduled revision?
• if yes, when was the revision published?
• when were other actions to rectify the error completed?
• were the causes of the error investigated?
• main findings
• actions taken to prevent recurrence