Publishing official statistics on polling days
This policy was developed following a consultation that took place during November 2017. A summary of responses to the consultation is provided after the policy statement.
The policy was approved by the National Statistician on 1 March 2018. It was updated in line with accessibility guidelines in August 2022.
|Publication date:||14 March 2018|
|Owner:||UK Statistics Authority|
|Who this is for:||Producers of statistics|
Official statistics are published according to a pre-announced schedule and continue to be published during pre-election period of sensitivity. Publishing on a national polling day itself can however cause difficulties in communicating data clearly and fully, in particular due to reporting restrictions. Because of these difficulties, publication procedures should ensure that statistical bulletins are not published on national polling days.
When enough notice of a national polling day has been given, the release of statistics and related materials should be scheduled to avoid this date.
If an election or referendum is called at short notice, any statistical releases which had been pre-announced to be published on the polling day should be rescheduled to the next day.
Decisions on pre-release access remain with the relevant person responsible as defined by legislation. The National Statistician advises there should be no pre-release access to statistics published close to a polling day.
What this policy covers
This policy applies to:
- UK general elections
- UK referendums
- devolved elections
- devolved referendums
This policy does not apply in the case of local elections.
The Devolved Administrations have asked that the policy apply when elections are held in their domain. For devolved elections and devolved referendums, the policy only applies to the relevant devolved statistics, as defined in Section 66 of the Statistics and Registration Service Act 2007. There should be proportionality to ensure that UK statistics could still be published on a devolved polling day.
Summary of responses to consultation
The consultation received responses from:
- six UK government bodies
- two devolved administrations
- a broadcast journalist
- a national charity
- a professional body
- one privacy group
Flexibility and predictability
The responses show that the reasons for avoiding publishing politically sensitive statistics on polling days were generally accepted. There were mixed opinions on whether to allow flexibility to decide whether to reschedule statistics.
About half of respondents suggested flexibility was the right compromise. These respondents suggested there should be clear criteria to help producers of statistics decide what constitutes ‘significant public interest’. One response suggested these decisions should be made solely by the National Statistician.
The rest of the respondents suggested that the policy should be changed to provide a ‘blanket’ approach where the announcement of an election or referendum would affect statistics in a consistent and predictable way. These responses emphasized the need to protect statisticians from being perceived as taking selective decisions.
Respondents did not think it was possible to define ‘significant public interest’ clearly enough to remove the risk of controversy. We have revised the policy to a blanket approach because of these consultation responses.
Advance or delay
A few respondents supported the idea of bringing forward statistical releases and publishing them the day before a polling day. It would be possible to do this for some statistics, but there are other statistics where it would not be possible to bring forward the release date without causing unacceptable risks to quality. This means that the only practical blanket approach is to delay all statistics.
Length of delay
Most people that supported a delay thought that statistics should be released the day after a polling day. Two respondents thought there should be a longer delay. The reasons for a longer delay included:
- statistics being overshadowed by media coverage of the election result
- avoiding pre-release access issues across an election
We have considered the arguments for a delay of longer than one day, including the fact that statistics are often published on days where the media is dominated by predictable high-profile events. The concern about pre-release access is addressed below. We have therefore applied a delay of 24 hours for the release of statistics.
By moving to a blanket delay there is a risk that departments might give pre-release access to individuals on polling day. A public breach of this access could have an effect on the poll.
By law this is at the discretion of the ‘person responsible’. The new policy includes advice from the National Statistician that there should be no pre-release access to statistics close to a polling day.