James Coombs v Information Commissioner & Anor

References in what follows to

This was an appeal against the FTT decision, which found that the IC decision notice was in accordance with the law.

The IC decision notice related to certain information requested by Mr Coombs from TBGS on 13 October 2019, namely

The background to this request was, as set out at [4], that there were defects in an 11+ exam (the “2019 exam”) managed and administered by TBGS, and held on 12 September 2019; as a result, TBGS and GL Assessment Limited (“GLA”), a third party contracted by TBGS to design and supply test material and provide other services relating to the 11+ exam, agreed on a “statistics-based solution”; they both put out public statements on 1 October 2019 explaining and defending their solution as a measure based on detailed statistical analysis and ensuring fairness for all children involved.

TBGS did not provide items 1, 2, 3 and 4b of Mr Coombs’ information request (it disclosed the information at item 4a), citing the exemptions in sections 41 (information provided in confidence) and 43 (commercial interests).

The IC decision notice, in the words of [9], “purported to hold” that TBGS correctly applied s41 and s43(2) to the information it withheld (the FTT decision put it that way because, it said, in the accompanying reasons, the IC decision notice had dealt only with s41, and said that s43(2) had not been considered).

The FTT decision recorded at [19] that the parties reached agreement that the information which was the subject of the appeal (and to which I shall refer as the “undisclosed information”) was limited to

The FTT decision followed a hearing on 27-28 November 2023, at which both Mr Coombs and TBGS were represented by counsel (Mr Coombs by David Lawson; TBGS by Felicity McMahon and Hannah Gilliland). The first respondent was not represented at the FTT hearing and, per [3], was content to rely on his written case, which corresponded closely with that of TBGS.

The backdrop to the FTT decision was that a previous decision by the FTT on the same appeal was set aside by the Upper Tribunal in Coombs v Information Commissioner and TBGS [2023] UKUT 157 (AAC) (a decision of mine), and the case was remitted to a freshly constituted panel of the FTT for reconsideration at an oral hearing. The Upper Tribunal directed that there was to be a complete re-hearing of the appeal in all respects except that it would be taken as a finding of fact that the statistician’s report referred to in the requested information was not held by TBGS (or by another person on its behalf) at times relevant to the appeal.

The FTT decision was not unanimous: the majority (the two specialist tribunal members) decided to dismiss the appeal; the minority (the tribunal judge) would have allowed it. Applying regulation 8 of the First-tier Tribunal and Upper Tribunal (Composition of Tribunal) Order 2008 SI 2008/2835, the decision of the majority became the FTT decision. In what follows I refer to the reasons of the majority for dismissing the appeal, as the FTT decision reasons, and to the reasons of the minority for preferring to allow the appeal, as the dissenting reasons.

The evidence before the FTT was noted at [16-17]: oral evidence from Mr Coombs (who had an academic background in statistics) and his two witnesses, Alan Parker (a retired former director of education for a London borough and subsequently, in quasi-judicial roles in bodies including for the Office of the Schools Adjudicator, Ofqual and Qualifications Wales; he had also been a trustee of the National Foundation for Educational Research) and Luke Knightly-Jones (conducting PhD research into perceptions and impact of private tuition for 11+ exams in England); and from Sue Walton (consultant at TBGS) and David Hilton (head of admissions testing at GLA) on behalf of TBGS; and the hearing bundles.

[31-62] were under the heading “Key Facts.” As the sub-headings indicated, this consisted of findings about the arrangements for grammar school admissions in Buckinghamshire (i.e. the context for the 2019 exam); about the exam itself; about past papers, practice materials and a “familiarisation booklet” published by GLA; about tutoring; about the business and contractual relationship between TBGS and GLA; about the errors in the 2019 exam, the solution devised by TBGS and GLA, and related communications with parents and carers for the children who sat the exam, and the wider public; about the concept of “reliability”, in the context of statistics; and about what the PowerPoint material contained.

Having summarised the rival arguments, the FTT decision reasons were presented at [70-84]

Considering s41(1)(b) (whether disclosure of the undisclosed information to the public by TBGS would constitute a breach of confidence actionable by it or any other person) – being the element of s41 that was in dispute – the three sub-questions identified by Megarry J in Coco v AN Clark (Engineers) Ltd (cited at [24]) were looked at. It was decided, on the facts as found, that the undisclosed information (1) had the necessary quality of confidence and (2) was communicated to TBGS in circumstances importing an obligation of confidence (the first two sub-questions).

The third Coco v Clark sub-question – whether unauthorised use of the undisclosed information would be detrimental to the party communicating it (which, the FTT decision said, was GLA) – was considered by looking, in turn, at two arguments made by TBGS: the “tutor advantage argument” and the “competitor advantage argument”.

These two arguments are first mentioned in the FTT decision in a footnote to [64], which introduces them by saying that

The FTT decision reasons found (at [76]) the “competitor advantage argument” to have force because, in overview, the undisclosed information, it found (1) was commercially sensitive and (2) would give a competitor an unfair, one-sided competitive advantage.

The FTT decision reasons saw some, limited force in the “tutor advantage argument”: disclosure of the undisclosed information might be of some “marginal” (my word, not the FTT decision’s) advantage to tutors and their pupils; GLA in turn might suffer reputational damage as a consequence of the fairness of the exam being (or appearing to be) undermined.

Having thus concluded that disclosure of the undisclosed information would constitute a breach of confidence actionable by GLA, the FTT decision reasons turned (at [79]) to whether an action for breach of confidence could withstand a public interest defence. The reasoning referred to the “critical importance of education” and a “powerful public interest” in fostering well-informed debate on education; but considered this outweighed by (and here I reorder the reasons somewhat, to bring out the substance of the reasoning) (1) the public interest in protecting commercially confidential information (see [79(5)]) (2) the fact that neither TBGS nor GLA misled, or attempted to mislead, in their public pronouncements about the errors in the exam or the fairness of their solution to these, together with the fact that TBGS (with GLA’s permission) had, reasonably, disclosed some of the information requested by Mr Coombs (it being reasonable to do so, given the need to explain TBGS’s decision-making and reassure those affected, and the wider public).

The FTT decision reasons concluded (at [80]) that s41 was satisfied; this was enough to dismiss the appeal; however, the FTT decision reasons went on to consider s43, and found (at [82]) that “an appreciable risk of prejudice to the commercial interests of both GLA and TBGS would arise” if the undisclosed information were made public; in respect of GLA, this was for the same reasons as the FTT decision reasons found “detriment” to GLA in its s41 analysis; in respect of TBGS, this was because disclosure could result in TBGS changing the structure of the exam more frequently (and associated costs). As to the public interest balancing test under s2(1)(b), the FTT decision reasons referred to its s41 public interest defence reasoning (which it did not think was displaced by the “mild presumption in favour” of disclosure, when applying that test). It concluded that s43(2) applied.

The dissenting reasons reached a different view on third Coco v Clark sub-question. The “competitor advantage arguments” were said to be “hugely overstated”, due to the undisclosed information being “very narrow in scope” and relating to just one year’s exam; disclosure of the undisclosed information would not tell the competition “anything of real significance” – the exam content is new every year; there is mention at [87(4)] of paucity of evidence for certain matters. The “tutor advantage arguments” were said to rest on “no evidential basis” and to depend on assertion and speculation. It criticised certain “theories” (about tutoring strategies) of Mr Hilton in evidence (indeed, the FTT decision reasons also rejected some such “theories”, at [75(5)]; it is said that certain of TBGS’s witness’ evidence was “mere assertion wholly unsupported by any scientific or empirical evidence”; it praised the evidence of Mr Knightley-Jones, on the question of “tutors’ advantage”, as coming from someone “with conspicuous learning and an impressive command of the subject matter” ([86(7)]. The dissenting reasons saw no advantage to tutors as a result of disclosure of the undisclosed information.

The dissenting reasons went on to say that, even if the Coco v Clark test had been satisfied, an action for breach of confidence would have been defeated by a public interest defence. In the dissenting reasoning, the “strong public interest in protecting information communicated in confidence”, is outweighed by the public interest in “testing”, and fostering an informed debate about the fairness and appropriateness of, the solution devised by TBGS and GLA (and their public statements about it), given that it was speedily devised, and the exam errors were highly embarrassing to GLA. There was also public interest in facilitating an informed debate as to whether TBGS issued misleading information about the fairness of its solution to the errors in the exam; particularly where TBGS did not disclose that the “independent statistician” to which it referred at that time was someone who had worked for GLA between 1986 and 2010. These matters could not be assessed in a transparent and fair debate, without disclosure of the undisclosed information.

The dissenting reasons went on to say that, for similar reasons, the exemption in s43 was not made out: it had not been shown that disclosure of the undisclosed information would, or could well, give rise to risk of substantive harm to TBGS or to GLA; and the public interest balancing test favoured disclosure.

The FTT granted permission to appeal on all grounds set out in an application dated 8 January 2024 (and drafted by Mr Lawson on Mr Coombs’ behalf).

It was said that the FTT decision reasons did not engage with or answer the factors raised in the dissenting reasons.

Under the above sub-heading, the grounds made these points

Regarding s41 and the public interest defence to actionable breach of confidence, the grounds said that the “minor” detriment to GLA (as found in the FTT decision reasons) could not sustain such a defence, on a proper application of the law as set out at [25].

Regarding s43 and the test in Hogan and Oxford CC v Information Commissioner (see [27]), it is said that “nothing identified in the [FTT] decision is substantial or likely to occur”. It is also said that the FTT decision reasons do not apply what was said in APPGER v Information Commissioner (cited at [28]) about the public interest balancing test requiring “an appropriately detailed identification, proof, explanation and examination”” of both harm/prejudice and benefits of the proposed disclosure.

It is said that it is not reasonably possible to balance the factors as the FTT decision reasons do: the impact of the errors on the public and the incorrect statements to the public support disclosure, as does the FTT decision’s finding that, without disclosure, the public cannot know if the distribution of school places was fair; on the other side, the undisclosed information is narrow; there was a possibility that a possible future competitor might try to make a more reliable test; this might impose an unknown cost on GLA. There was no answer to the point in dissenting reasons at [89(6)], discussing the public interest defence: the detriment-based grounds on which TBGS resists disclosure have, at best, very little substance, and “offer no material counterweight to the compelling public interests favouring disclosure”.

The FTT decision’s reasons as regards s43 and prejudice to commercial interests were challenged on much the same grounds as had already been argued:

Montague v Information Commissioner [2022] UKUT 104 at [24] is cited to the effect that the statutory language “cutting down” the right to information under s1 needs to be “carefully construed. The language of [FOIA] should, where possible, be construed broadly and liberally in the context of FOIA’s statutory purpose to make provision for the disclosure of information held by public authorities in the interests of greater openness and transparency …”. The grounds state that the FTT decision reasons fail to follow such a construction of FOIA. Why should residents of Buckinghamshire not know about the exam through which their children are admitted to schools they maintain and fund?

The respondents were directed to make a response to the appeal – TBGS did, but the first respondent did not – and Mr Coombs was invited to provide a reply to that response, which he did.

The parties were directed to indicate if they sought an oral hearing of the appeal. Neither TBGS, nor Mr Coombs, so indicated. Given this, and that I had fulsome written submissions in the form of the grounds, the response, and the reply, I decided it was fair and just to determine this appeal without a hearing.

I am grateful to both Mr Coombs and TBGS for their clear and helpful written submissions.

It is helpful, in my view, to set out a recent summary by a three-judge “Presidential” panel of the Upper Tribunal, in Information Commissioner v Experian [2024] UKUT 105 (AAC), on an appeal from the FTT General Regulatory Chamber, of the law regarding the Upper Tribunal’s “error of law” jurisdiction, and adequacy of reasons, as follows:

I would add that the standard of reasoning required, to avoid error of law, in a decision upon reconsideration on remittal to the FTT by the Upper Tribunal following the setting aside of an earlier decision (as happened in this appeal), is as described in the preceding authorities – no more, no less.

When considering whether the exemption in s41 applied, the issue in contention before the FTT was whether disclosure of the undisclosed information would cause an actionable breach of confidence (i.e. whether s41(1)(b) applied), and, specifically

Resolving both of the above involved the FTT making evaluative judgements i.e. (in the words of the Supreme Court in Lifestyle v Amazon) “multifactorial assessment[s] of the documents, the evidence and the submissions made by the parties”.

The FTT approached the question of “detriment” through analysing the “tutor advantage argument” (disclosure would advantage private tutors, and so undermine the fairness of the 11+ exam, to GLA’s detriment) and the “competitor advantage argument” (disclosure would present a current or future competitor with an unfair (to GLA) advantage in the market).

It is clear enough that the FTT decision reached the view that disclosure would be detrimental to GLA principally on the basis of finding the “competitor advantage argument” persuasive: this is the clear inference from the FTT decision reasons expressly finding “limited” force in the “tutor advantage argument”: see [75] opening sentence, and sub-paragraphs (4) and (5). I note what is said at [77] about the FTT decision reasons concluding on detriment “taking its analysis of the Tutor Advantage and Competitor Advantage arguments together”; but this does not detract from, or change, what is clearly and expressly said at [75] about the limitations of the “tutor advantage argument”. As I put it in my summary of the FTT decision reasons above, any advantage accruing to private tutors as a result of disclosure was marginal; it follows that, even if it had been found that no material advantage would have accrued to tutors on disclosure of the undisclosed information, the conclusion of the FTT decision reasons would not have changed.

From this it follows that there is no need to deal with the aspects of the grounds of appeal that allege errors of law specifically with regard to the FTT decision reasoning on the advantage that would accrue to private tutors if the undisclosed information were disclosed; any legal errors in that reasoning would be immaterial.

Turning to the FTT decision’s (material) finding that the “competitor advantage argument” was made out, such that disclosure would be detrimental to GLA by giving a potential competitor an unfair advantage – the grounds challenged this evaluative judgement on the FTT’s part as inadequately explained, and so in error in law. In particular, the FTT decision reasons, it is said, inadequately explain why the countervailing submissions and evidence put before the FTT – that the nature of the 11+ exam was such that the undisclosed information would not give a competitor any meaningful advantage (over GLA) – was rejected. Further, related points were made in Mr Coombs’ “reply” to TBGS’ “response” to this appeal, about the evidence of Mr Hilton, whose witness statement spelled out the “unfair advantage” consequences of disclosure which the FTT decision found persuasive; it was said that the FTT decision had fallen into error because

It seems to me the FTT decision reasoning on the “competitor advantage” point was clear and straightforward: its essence was (1) that the FTT had found the undisclosed information to be “important and sensitive” (see [72] – with whose analysis even the dissenting reasons agreed); and (2) that to hand such information to a competitor would be to give it a head-start or leg-up (my words; a “benchmark”, is what the FTT decision calls it); that would be an unfair advantage. Reasons are given at [76(4)-(6)] as to why disclosing specific components of the undisclosed information – the number of questions in the 2019 exam, and the “reliability” analysis – would help a competitor, to GLA’s detriment.

The grounds of appeal (in part by reference to the dissenting reasons) point to arguments (presented either in evidence or in submissions to the FTT) as to why the undisclosed information would be of no great value to a competitor; it was “narrow” information, about the exam for single year; new exams were set each year; GLA held “banks” of questions; knowing the number of questions couldn’t possibly convey any material advantage.

In my view, it is clear enough, by obvious inference if not expressly, that the FTT decision simply did not accept that the undisclosed information was as trivial (in commercial terms) as these submissions, and witness evidence, portrayed it; moreover, the FTT decision reasons did adequately explain, in substance, this evaluative judgement on its part: the undisclosed information was important, sensitive and gave a competitor a head-start (commercially) that had not been available to GLA.

As to whether there was an error of law in the FTT decision reaching the evaluative judgement that it did, that, as the authorities illustrate, is a relatively high hurdle. As is graphically illustrated by the split panel in this case, it is clear that different tribunal may have reached a different conclusion on the submissions and evidence that were presented; but, equally clearly, that is no indicator of the evaluative judgement reached being in error of law. The authorities make clear that, on an appeal against a first-instance tribunal’s evaluative judgment, it is not the role of the appellate tribunal to “island hop” amid the evidence presented at first instance (or revisit the first instance panel’s evaluation of the witnesses, including, here, Mr Hilton). The question is whether the evaluative judgement – about the undisclosed information being of material commercial value to a competitor, to GLA’s detriment – was one no reasonable tribunal could have reached on the evidence before it; or whether some material factor was not taken into account. I am not persuaded. The judgement was based on Mr Hilton’s evidence; the reasoning behind it was clear and straightforward, and dealt with the gist of the countervailing argument; it is not a judgement that can be said to be perverse or irrational. Moreover, it was an evaluative judgement by the specialist information rights tribunal, on a matter (the application of the exemption for confidential information) that arises for determination not infrequently.

For completeness, I add the following about some points made in the grounds of appeal and Mr Coombs’ “reply”:

Turning now to the FTT decision’s finding of no “public interest defence”, the FTT decision reasons, in essence, found that the public interest in a fuller “inquest” (my word; I use it colloquially, to allude to the kind of “comprehensive, independent, statistics-based assessment of the fairness of the exam (post-solution)” as is posited at [62]) into the episode of the errors in the 2019 exam, and the solution devised by TBGS with GLA, was outweighed by the public interest in protecting confidential (commercial) information. Importantly, the FTT decision reasons found that TBGS and GLA did not mislead, or attempt to mislead, in their public pronouncements associated with the episode: see [79(3)]. In my view, it is adequately clear, in context, that this latter finding materially countervailed, in the FTT decision reasoning, public interest concerns about actual or suspected “failings” (again, my word) by TBGS and/or GAL as found in the FTT decision (and which the grounds of appeal complain were not adequately dealt with in the FTT decisioning reasoning) e.g. the occurrence of the errors in the 2019 exam, and the upset they caused; and the statements made about the “independent” statistician, as against Dr Hutchison’s past employment with GLA.

The FTT decision reasoning is therefore, in my view and in this regard, adequate. As to whether there is an error of law in the conclusion reached – another “classic” evaluative judgment by the specialist tribunal – I am again not persuaded that this judgement was one no reasonable tribunal could have reached on the evidence before it, or that some material factor was not taken into account. As before, the circumstances of the FTT decision illustrate that another tribunal may have weighted the various factors differently, and reached a different verdict; but this is nothing to the point. I conclude that no error of law is made out.

In my view this also deals with the somewhat whimsical point at the end of the grounds about openness, transparency, and why the “residents of Buckinghamshire” should not see the undisclosed information: the analysis above is not based on any unduly narrow construction of FOIA, but rather on whether an first-instance tribunal erred in law in its evaluative judgement following a full and fair hearing of the issue – to which I have found the answer to be “no”.

I have dealt in substance with all the grounds of appeal that could indicate error of law in the FTT decision as regards the application of s41; none have been made out. It is unnecessary for me to go on to consider how the grounds might indicate an error of law as regards the application of s43, as this would not make any difference to the conclusion that the FTT decision made no material legal error in dismissing Mr Coombs’ appeal.