Accuracy of reporting – not just for journalists

Here at The Transparency Project, we often politely challenge inaccurate reporting by the press (sometimes we are more polite than others). Everyone gets it wrong sometimes (including us), and some of the errors we’ve flagged have been a matter of simple human error and swiftly corrected when pointed out. Others have met with a more terse response and a brick wall.

Our recent event, Reporting The Family Courts – Are we doing it justice? has produced a reminder that it is also important to be accurate when we are discussing transparency itself, and the evidence base in this area. And that it is not just journalists who need to take care with their words.

There have been a number of write-ups of the event, which we’ve gathered together here. Two of them state that research by Dr Brophy led to the identification of children in 7 of 21 judgments.

The Solicitors Journal says :

A group of techie-minded researchers looked at 21 anonymised rulings and managed to identify the parties in seven of them.

This Week in Fostering say :

…one study that found that the children who featured in seven out of 21 judgements examined could be identified through jigsaw identification despite the judgment being anonymised.

We haven’t asked the authors of these reports to correct these sections of their post because they accurately reflect what was said about the research study in question by a speaker at the event, Martha Cover, Co-Chair of the ALC.

Martha said this (taken from the video recording of the event) :

“…The fact is that we have to adopt a child centred approach and the rule must be do no more harm. And the difficulty of actually managing that is demonstrated very graphically by the the young people from the national youth advocacy service who took 21 judgments, anonymised judgments from BaILII in family cases and in 7 of those judgments within a period of hours with levels of computer skills and social media skills that I haven’t – frankly – put me completely in the shade – they were able to identify those children and those families from jigsaw identification, from cross referencing to criminal reports – don’t forget that a lot of the child protection cases that you most often want to report and which I agree the public ought to know about are concurrent with criminal proceedings. So cross referencing all of that material they could identify those children by their family make up, by their religion or ethnicity, by the area in which they lived, by some detail about the school which they attended, or about the particular shape and size of their sibling group. They identified 7 out of those 21 carefully anonymised cases…” [our emphasis]

But that isn’t actually quite what the research being cited said. So we want to set that right.

The research being referred to is : A review of anonymised judgments on Bailii: Children, privacy and ‘jigsaw identification, by Dr Julia Brophy, October 2015. It was jointly commissioned by the ALC and NYAS and can be read on their website here, for those who want to check. Our Alice Twaite posted an analysis shortly afterwards here.

In March 2016, this research was also considered in a piece in Family Law Journal : Maintaining privacy – Alice Twaite discusses issues arising from new research on jigsaw identification, and wider concerns as to transparency in children cases. You can read that here (reproduced with kind permission). In that article Alice correctly identified that

Alice’s account refers to a subset of 13, whilst Martha refers to 7. There are references in the research to various subsets of 7 of the judgments, but what is said is that 7 of the 21 judgments were referenced on social media, and 7 included allegations of sexual abuse, 7 included criticisms of professionals.

From the report it appears that one or possibly two (unclear) of the judgments were identifiable as a result of family members posting the judgment or identifying detail on social media, i.e. as a consequence of family actions rather than the anonymisation itself. See separate references to :

which may or may not represent the same judgment / case. There is nothing else in the report which supports the proposition that 7 sets of children were identified, and indeed the tasks set for the investigators were to answer the hypothetical question :

The questionnaire the investigators were given does not ask them to record whether the investigator could or had identified the individuals involved, but asked them for an opinion as to whether others may be able to.

We think there IS an important distinction between “would permit identification” and “were identified”. The research did not lead to identification (apart from the grandmother examples above), and of course did not reflect real life in that the child investigators in the study were given an objective of trying to see if they could identify the children in question. The fact that a number of inadequately anonymised judgments have remained on BAILII for lengthy periods without being noticed (see the Cardiff research here – page 46) is an indicator that in practice judgments published on BAILII are not being pored over by the public trying to see if anyone they know is being discussed.

Our Lucy Reed recently appeared on Radio 4 alongside Dr Brophy to discuss the Cardiff research and the publication of judgments. In the course of that discussion it was asserted that there was evidence of actual harm being caused to children as a result of the publication of judgments / poor anonymisation. Lucy Reed took issue with that assertion in an article on The Justice Gap : Transparency in the Family Courts evaluated – could do better. She said :

There is plenty of evidence of anonymisation failures, and therefore of a risk of harm (which should not be minimised), but it is important to be clear that no individual examples of actual catastrophe have been provided. The Cardiff researchers asked a range of individuals and organisations with experience in this field if they had evidence of such harm, and none of those who responded were aware of any individual cases of actual harm to children.  Organisations included the National Youth Advocacy Service, the Children and Families Court Advisory and Support Service (CAFCASS), the Family Justice Young People’s Board, and the family lawyers’ group Resolution. Two judges each reported a single incident of local identification of a parent from an anonymised judgment / local press coverage. None reported identification of children.

Whilst Dr Brophy herself, through the Association of Lawyers for Children, has carried out a number of studies on the topic of children’s views about these issues, those studies evidence the views and fears of small numbers of children when presented with a particular scenario (here and here), which I wrote about at the time (here). They are not studies about children who had actually been identified or harmed. There may be such cases, but to date they have not been publicly identified nor formed the basis of any published research study.

It is unfortunate that the findings of this valuable research have been imprecisely described and that an inaccurate understanding of the evidence base has been propagated as a consequence. It is important to understand the risks of greater transparency, but equally important not to overstate them or to place more weight on the limited evidence base than it can properly bear.

It is possible that some lack of clarity has crept into discussion of this research through the elision of two discrete points in one bullet point in the Executive Summary of the report, under the heading “Coverage in mainstream media and social networking sites”, here :

they identified…coverage on social networking sites for 33% of judgments (7/21). Materials on social networking sites (e.g. Facebook pages etc.) identified children and other family members; some also contained photographs of children.

Feature Pic courtesy of Michell Zappa on Flickr (Creative Commons licence) – thanks!