(This policy has been superseded. The new policy is available on our Guidance about Assessment page.)
Experimental statistics are defined in the Code of Practice for Official Statistics as “new official statistics undergoing evaluation. They are published in order to involve users and stakeholders in their development and as a means to build in quality at an early stage.”
Attached to this Statement is a note setting out how the Statistics Authority will approach the assessment and designation as National Statistics of official statistics currently labelled as ‘experimental statistics’.
Assessment and Designation of Experimental Statistics
The Statistics Authority’s approach to assessing and designating as National Statistics official statistics currently labelled as ‘experimental statistics’ is consistent with that for any other official statistics. We will approach the assessment and designation of experimental statistics (ES) in the following way:
- If the producer body had not already conducted, or was not planning, an evaluation of the ES, then we would require them (in relation to the relevant part(s) of Principle 4 or Protocol 1 of the Code of Practice) to do so at an appropriate point;
- If the ES have been evaluated and were considered to be of poor quality and beyond improvement we would not designate them as National Statistics (in relation to Principle 4 Practice 2 of the Code of Practice). In practice, we would expect producer bodies to have withdrawn any such statistics once they were aware of their significant shortcomings (i.e. prior to an assessment), so this scenario is unlikely to arise;
- If an evaluation has been undertaken and users have said that it is too soon for them to judge the merits of the ES definitively, we would either require a producer body to continue the process of evaluation and to commit to a further evaluation (in relation to Principle 4 Practices 2 and 5 of the Code of Practice), or we would endorse its published plan to do so. We may additionally make requirements about quality, documentation, or any other aspects of the Code, as with any other assessment. All of these circumstances would lead to a conditional designation as National Statistics. The timescale to meet the requirements would be determined on a case-by-case basis, as with any other assessment, and would take into account the timing of the producer body’s own evaluation where appropriate; and,
- If users are satisfied that the statistics meet their needs, then (assuming more general compliance with the Code, and assuming that the producer body was committed to publishing a summary of its evaluation of the ES), we would expect to offer an unconditional designation as National Statistics. This does not preclude a producer body from itself deciding, in the light of an evaluation, that the ES label should be removed. Indeed, we would expect this to be the norm, and for this scenario to apply mainly where there is a coincidence of timing of an assessment and the undertaking of an evaluation.
It is for producer bodies, rather than the Statistics Authority, to determine which statistics should be labelled as experimental statistics. It would seem natural for a producer body to cease to use the ‘experimental statistics’ label once the statistics had been unconditionally designated as National Statistics. As at present, we will continue to note in Assessment Reports which of the statistics are, or have been, published as experimental statistics.