Business
Experimental Statistics FAQs on expanding range of statistics in development

What are experimental statistics?
These are statistics that are in the testing phase and are not fully developed.

How should users interpret such series?
It is important that users clearly understand the status of these series and the cautions that apply (which will vary for each statistic). The particular limitations of each series is explained in the information that accompanies it.

Why the increasing number of experimental series?
There is a greater emphasis within ONS on consulting users during methodological reviews and the development stage of methodological changes. In particular there is a strong desire to make experimental series available during a development period, to assist in the quality assurance process, and to help familiarise potential users with any changes. There have also been various initiatives to increase the scope of economic statistics, particularly within the service sector.

When are statistics experimental?
Defining what is experimental and non-experimental is largely a matter of statistical judgement, but typically experimental series arise when:

- they are being produced part way through a well defined development programme - whether these statistics are new or changed versions of existing statistics;

- statistics are new but still subject to testing in terms of their volatility and ability to meet customer needs;

- the statistics do not yet meet the rigorous quality standards of National Statistics; or

- a rich variety of new measures is available from a new set of statistics, with components that have considerable immediate value to users. These users are aware of the statistics' theoretical quality and can make use before ONS has completed all operational testing. The testing is designed to fully validate the measures to the standard expected of National Statistics.

How do experimental statistics arise
The experimental nature of such statistics may reflect:

- new methods which are being tested and are still subject to modification;

- partial coverage (eg; of industries) at that stage of the development programme; or

- potential modification in the light of user feedback, in terms of usefulness and credibility versus other sources.

Why publish experimental statistics?
The aims of their publication include:

- Consultation: ONS would like to get informed feedback from potential users.

- Acclimatisation: where the experimental series are alternative versions of the format of existing series, ONS may wish to help users become accustomed to new presentations.

- Use: experimental series can provide very useful information for users as long as their nature is well explained and understood.

When do experimental statistics become non-experimental?
This will be a matter of statistical judgement, but typically will take into account factors such as:

- When it is judged that statistical methods have settled down.

- When coverage reaches a good level.

- When user feedback indicates that these statistics are useful and credible.

- When the defined development phase has ended.

- When it is judged that the statistics meet the rigorous quality standards of National Statistics.

What is the policy on dissemination of experimental statistics
ONS' policy on dissemination of experimental statistics is being developed, but currently states that:

Experimental series will be introduced by an article on the NS Website and in the appropriate "Trends" publication, to explain their concepts and construction etc.

Experimental series will then be updated on the NS Website. They will be accessible via a dedicated area of the Latest Releases section. This will ensure that the status of the series is clear. Articles and/or tables can be made available for downloading.

Whilst updates for experimental series will be disseminated via the Website, the information may also appear in paper publications, such as the next edition of the appropriate "Trends" publication. It is important that in all cases these statistics are clearly marked as experimental.