How low can you go?

At some point you can get no lower.....

Freedom of Information officers often come up against the problem of when and if statistical data can be seen on specific occasions to be a breach of S.40 (Personal Information).

Most commonly this problem has been seen in tables of data – often about medical conditions – where the public authority will put a dash “-“ in any cell where the value is less than five.

The Department of Health (DoH) regularly uses this technique when providing tables showing the number of women who have had multiple abortions.

You will see from this Abortion data table that there are a handful of under-18 girls who are on their third abortion or more – yet the DoH refuses to give us the exact figure.

In Scotland there was a long-running dispute over whether low cell values in a table of the prevalence of childhood cancer, could lead to those children being identified.

I have to say that I still have difficulty with the concept and certainly do not accept that any cell value under 5 or ten automatically has to be anonymised. In the abortion table example how would knowing that six women aged under 18 had their third abortion that year led to their identity being exposed?

The key to these questions, in my opinion, is the total pool from which the information is drawn. So in the abortion data case your pool of identifiable candidates is all the women aged under 18 in the whole country.

But this vexed problem of low cell values has taken a different turn in the case of Beresford Lane in the Sussex village of Plumpton Green.

A company that wants to put together data for prospective house buyers asked the local police force for data about anti-social behaviour in the lane for a six month period.

Sussex Police refused to either confirm or deny (NCND) it held any such information as it claimed that to state it held such information could lead to the identity of either the victim or the culprit. This use of S.40 was rejected by the Commissioner and has now also been thrown out by the Tribunal.

But the key element in this case was in essence the “total pool”/“cell value” which in this case was the number of people who lived in the lane  against the number of anti-social behaviour complaints in that lane.

The Tribunal came to the following conclusion: “The Appellant says that if it confirms or denies that it holds the requested information, that could lead to the identification of either those who have made complaints about ASB, or those about whom such complaints have been made. Clearly, if it does lead to such identification, that would amount to disclosure of personal data. The question is - would it lead to such identification?

“The facts before us are notably brief. We have been told that Beresford Lane, Plumpton Green, BN8, is a rural road with approximately 17 residential dwellings. Presumably, many, if not most of the 17 dwellings, will have multiple occupiers which means there are likely to be many more than 17 people living in the area covered by the request. The request covers a period of 6 months. We have no other facts of any significance. On these simple facts, we have great difficulty in seeing how, if the Appellant confirms or denies that it holds information on the number of ASB complaints reported, and the descriptions of any such ASB complained of, in that period and for that location, that would enable a living individual to be identified as having either made a complaint or having been the subject of such a complaint, or indeed to be identified in any other way. The Appellant has simply not demonstrated how that connection could be made.

“The position might be quite different on different facts. For example, if there was a known incident concerning one or more particular dwellings or individuals and if the time period covered by the request was relatively specific to that incident, a confirmation or denial could well lead to an individual being identified. That, however, is not the situation in the present case.”

So it would appear that in this case the Tribunal thinks that to disclose information about a cohort – sadly we don’t have an exactly figure but the number of people who live in 17 houses (34?) – will not breach S.40. But as the Tribunal has said different circumstances and different figures might lead to a different ruling.

The key question is when does that group number get so low that S.40 would be breached. And it appears that the Tribunal has neatly ducked out of answering that question.

The Tribunal ruling can be seen here. Tribunal decision.