Line to take - LTT52 - Statistical Information

• FOI/EIR: FOI
• Section/Regulation: s35
• Issue: Statistical Information
• Source: Information Tribunal
• Details: DWP / ICO (5 March 2007)
• Related Lines to Take: n/a
• Related Documents: FS50083103 (DWP), FS50105898 (HMT), EA/2006/0040 (DWP), Awareness Guidance 24, MOJ Guidance FOI s35
• Contact: RM
• Date: 28/09/2009
• Policy Reference: LTT52
• © Copyright Information Commissioner's Office, re-used with permission
• Original source linked from here: LTT

Line to take

Statistical information is the product of some form of mathematical or scientific analysis and will include the facts/data which are fed into the analytical models. It is not simply a view or opinion that happens to be expressed numerically.

Further Information

Background

In the Tribunal case DWP v the In formation Commissioner, the applicant had requested a feasibility study on the impact that the introduction of ID cards would have on the DWP’s business. It was accepted that the information fell within the class of information relating to the formulation and development of government policy. The Tribunal also found that the government had taken the policy decision (i.e. to introduce ID cards) by the time the request was made. It therefore became important in this case to determine whether or not it was statistical information.

This is because under s35(2) statistical information cannot be exempt under 35(1)(a) - formulation & development of government policy, or 35(1)(b) — ministerial communications, once the policy decision has been taken.

Statistical Information

Statistical information is broader than the term statistics. Statistics refer to the facts and figures whereas statistical information also includes the scientific or mathematical analysis of those facts (see para 77). The Tribunal adopted the broad definition of what constitutes statistical information contained in the MoJ’s guidance on s35.

“Statistical information used to provide an informed background to the government policy and decision... will usually be founded upon the outcomes of mathematical operations performed on a sample of observations or some other factual information. The scientific study of facts and other observations allows descriptive approximations, estimates, summaries, projections, descriptions of relationships between observations, or outcomes of mathematical models, etc to be derived.

A distinguishing feature of statistical information is that it is founded to at least some degree on accepted scientific or mathematical principles. Statistical information is therefore distinguished by being;

i) derived from some recorded or repeatable methodology, and

ii) qualified by some explicit or implied measures of quality, integrity and relevance.

This should not imply that the term ‘statistical information’ only applies to where standards of methodology and relevant measures are particularly high. What distinguishes statistical information is that the limitations of methodology, and the relevant measures of quality etc., allow for a rational assessment of the validity of the information used as an informed background to the formulation and development of government policy”

It should be noted that this definition originates from the Office of National Statistics and had already been adopted our Awareness Guidance on this exemption (No. 24).

The Information

In this particular case the information requested consisted of working assumptions; estimates of such factors as how many citizens would take up ID cards over a number of years, together with a whole range of scenarios on how the cards could/would be used by individuals etc. These working assumptions were then used to predict the impact ID cards would have on DWP business. Much of this information was expressed numerically. Tribunal found that these values had not been generated by some mathematical analysis. The Tribunal concluded that it did not constitute statistical information. It follows that it accepted that these predictions were based on judgements, views and opinions of officials.

Statistical Information and the factual information on which it is based.

From the MoJ’s definition of statistical information it is clear that statistical information is the product of some form of mathematical or scientific analysis of facts and figures. A distinction could be drawn between the analysis and the actual facts upon which it is based, but if this approach was accepted then, it’s conceivable that, although a public authority was required to provide the analysis, it could still claim that the data on which that analysis was based was exempt. However the we should adopt a more pragmatic approach which includes both the facts that are fed into the scientific model and the model itself, as well as the product of that analysis, as all being statistical information. It should be noted that although this is the line to take, it is not a matter that was addressed by the Tribunal.

Accuracy of the Statistical Information.

It is noted that at para 77 the Tribunal commented that the Mr Pitt-Payne “helpfully contends that statistical information is information that is put forward with a high degree of confidence...” which suggest it had sympathy with this view. On the face of it this seems at odds with the final paragraph of the definition set out above. In essence it is not that the statistical analysis has to produce highly accurate predictions, rather it's that its margins of error have to be understood so that it can be used with confidence.

Application of S35(2)

The Tribunal explained (at para 79) that since it had determined that the policy decision to which the information related had already been taken then, had the information been statistical information, it would have been obliged to go onto consider whether the information had been used to provide an informed background to that policy decision.

This illustrates the importance of checking that all 3 elements of s35(2) are in place when determining whether statistical information is exempt or not;

1) Is it statistical information?

2) Has the policy decision to which it relates been taken?

3) Did the statistical information provide an informed background to that decision?

Consistency with other DNs

In DN FS50105898 the complainant had requested the information on the variables that were fed into an economic model. The information in question was a range of predictions on factors effecting the economy, e.g., performance of the housing market. These predictions were expressed as numerical values.

The DN acknowledged the ONS’s definition of statistical information and in particular the reference to ‘projections’. However the DN also presented a dictionary definition of a statistic being a ‘quantitative fact or statement’, to argue that a projection into the future could not be a fact & therefore not statistical information. Although ultimately the DN took other factors into account when reaching its decisions, this particular approach is no longer a line we can follow in light of the Tribunal decision in the DWP appeal.