Department for Transport v The Information Commissioner & Anor

In relation to statistical information s 36(2)(c), read with s 36(4), provides:

S 36(4) has the effect of removing the need for the reasonable opinion of a qualified person to engage the exemption where statistical information is concerned.

In relation to ‘chilling effect’ arguments, the tribunal is assisted by the following paragraphs from the Upper Tribunal decision in Davies v IC and The Cabinet Office[2019] UKUT 185 (AAC):

The exemption is prejudice based. ‘Would or would be likely to’ means that the prejudice is more probable than not or that there is a real and significant risk of prejudice. The public authority must show that there is some causative link between the potential disclosure and the prejudice and that the prejudice is real, actual or of substance. The harm must relate to the interests protected by the exemption.

S 36 is a qualified exemption, so that the public interest test has to be applied.

The tribunal’s remit is governed by s.58 FOIA. This requires the tribunal to consider whether the decision made by the Commissioner is in accordance with the law or, where the Commissioner’s decision involved exercising discretion, whether she should have exercised it differently. The Tribunal may receive evidence that was not before the Commissioner, and may make different findings of fact from the Commissioner.

The issues for the tribunal to determine are:

Lisa Jordan is the Cabinet Chief Economist and head of the Analysis and Insight Team. She has a bachelor’s degree in economics and a masters’ degree in economics and social policy analysis. Ms Jordan gave evidence on the consequences of ordering the release of the requested information and on the public interest.

Her evidence, in summary, and in so far as is relevant to the issues we have to determine, is as follows.

Ms Jordan’s evidence was that the following changes would need to be made if the appeal was upheld because of the ‘precedent’ set by this release of information.

Team level reports contain the number of responses and percentage response rate which means the total number of staff in a team can be calculated. This is sensitive information for certain teams (e.g. immigration enforcement). It is not sensitive for AAIB.

Ms Jordan asked us to take into account the precedent set by this release. She states that one effect is that the Commissioner and the FTT can be expected to order release of other People Surveys at sub-organisational level in the future.

A consistent ‘neither confirm nor deny type approach’ is necessary in relation to the release of people surveys at sub-organisational level rather than an approach which looks at the specific harm that will be caused by release, because the mere fact of non-disclosure would indicate that the relevant unit had a problem. Ms Jordan gave examples in her closed evidence of particular units where release would lead to particular harm.

In order to pre-empt, or in response to, large volumes of requests for publication, organisations may review their reporting hierarchies and start to exclude specific units or only have large reporting units. The vast majority of teams and units will not be covered by the National Security exemption. Ms Jordan anticipated that some organisations would be likely to reconsider their continued participation in the survey.

Ms Jordan submitted that these concerns were not speculative because certain Civil Service units already do not take part in the survey or restrict the circulation of their results internally because of FOI concerns.

Ms Jordan stated that there is a justified concern that by enabling the publication of sub-organisational reporting unit results, teams and areas could be subject to undue attention or inquiry, targeting, for example, the small unit that is the AAIB.

Ms Jordan stated that this clearly could have consequences for participation in the survey either by individuals themselves or as a result of pressure by their line manager.

Ms Jordan stated that the fear of this has already been expressed to her team in relation to the AAIB:

“A member of my team recently received a call from the [job title within the AAIB] to say that staff in the AAIB were aware of the FOI request and should the results be published are concerned about negative commentary or mischaracterisation by Dr Alexander and would not participate in the future if the AAIB results were published”

The statement is dated April 2019, and therefore the call must have taken place at some point not too long before then. Ms Jordan, understandably, no longer recalled the details of the call and stated that her witness statement reflected her knowledge at the time.

Ms Jordan gave evidence in closed which she says is evidence of the chilling effect already taking pace within AAIB. Any closed evidence which follows is extracted from the full gist of the closed session, other than a small amount of additional evidence that does not appear in the gist but which has been checked with the DfT.

Ms Jordan provided the response rates of the AAIB and the DfT to the People Survey 2021 and drew a comparison between those rates (77% against 85%). Ms Jordan explained that the results were considerably lower than those for peers in the Rail and Marine Accident Investigation Branches, which had a similar profile (95%).

Ms Jordan was asked questions about how the AAIB response rate compared to its historic response rates as shown in the disputed information. Ms Jordan accepted that 77% was broadly in line with the survey response rates for AAIB in the closed bundle. She did not have information on historic response rates by the DfT, so a comparison between AAIB and DfT over time could not be done. Ms Jordan said that 77% was still good, but lower and less robust than higher response rates would be.

As well as response rates, Ms Jordan was asked about the quality of the responses to the People Survey since the request. She was unable to comment on whether the quality of the data in the 2021 survey had been affected by Dr Alexander’s request, but commented that this would be almost impossible to detect in the survey results. Ms Jordan said that it was impossible to police responses to the survey. One cannot see from the data if people have been unduly influenced, as their response has to be taken to be their view.

Ms Jordan explained that in addition to the annual People Survey, in-year surveys, known as ‘pulse surveys’ may be conducted. These are optional, and if conducted they might be run either by the Cabinet Office or by the Departments concerned.

Pulse surveys are different as the People Survey has 100+ questions and a large amount of promotion work to get the highest response rates so as to obtain as robust data as possible. Although not recorded in the gist, Ms Jordan also stated that it was therefore not appropriate to compare response rates to the pulse survey to response rates to the People Survey because they were ‘completely different surveys’ and that you ‘simply would not compare the two’.

Ms Jordan gave evidence as to the response rate to an AAIB pulse survey in 2019, and stated her view that it was low because it was around the time of Dr Alexander’s request. She said that this was evidence that showed that the FOI request was having a negative impact on response rates. Ms Jordan did not have any information on pulse surveys by AAIB in any other years. There is further closed evidence on this in para 38 of Ms Jordan’s witness statement which is dealt within the closed annex.

Ms Jordan was asked why, if the pulse survey and People Survey were not comparable, she thought that the low response rate to the pulse survey was evidence of the risk of a chilling effect on responses to the People Survey. Ms Jordan stated that it showed the impact that the FOI request is having on people’s confidence in responding to surveys of this type.

Ms Jordan expressed the view that the combination of the low response rate to the pulse survey, and the views of individuals within the AAIB (set out in part above and in part in closed) shows that this FOI request is leading to a lower likelihood of responding to the surveys. Ms Jordan said that this was only within AAIB as people do not know about the FOI request outside of AAIB.

Ms Jordan stated that the figures and report do not on their own provide the wider contextual factors and information that internal parties know or can easily ascertain. The results of the People Survey are not intended to be used in isolation. Consideration has to be given to the particular circumstances in the unit over the relevant period. The manager of a unit may have limited control over these circumstances and their impact on staff. These circumstances are not conveyed in the reports of the results.

In oral evidence she referred to additional contextual information being required in order for the public to understand what the results were ‘really showing’ because within the organisation there is understanding about, for example, whether there has been a high turnover or organisational change. When asked by the Commissioner whether there was anything of that nature in the AAIB with reference to the requested information, she was not able to say.

Ms Jordan was not able to say how difficult or burdensome it would for the AAIB, or any particular team, to provide contextualising information but was able to say that in general there are a lot of questions within the reports and it would be incredibly difficult to provide contextualising information on each one.

Ms Jordan stated that media coverage of the organisational level People Survey results rarely provides balanced coverage, concentrating largely on negative scores and focussing particularly on bulling and harassment. There is a concern that enabling the publication of sub-organisational reporting unit results could subject teams to undue attention or inquiry.

The evidence set out under ‘effect on staff participation’ above is also relevant to this heading.

Ms Jordan stated that she was not aware of any particular media interest or history of media interest in how the AAIB works internally.

Ms Jordan explained that within survey research there is a well understood concept of social desirability bias: individuals provide answers they think interviewers want to hear rather than an individual’s true belief or experiences. The People Survey uses self-completion methods which minimise this effect.

Ms Jordan stated:

Ms Jordan’s oral evidence was that this was wider than just a concern about selective press coverage. She said that individuals could respond in a way to align with the public portrayal of a particular unit, so that if there were concerns around bullying and harassment in a particular unit, responses might be made to align more with that concern. Individuals might respond to try and please a line manager, or whoever they saw as being the audience for the results. The concern is whether the results would be accurately portraying the true experience and opinion of the civil servants who were responding and therefore if it would provide robust data.

Ms Jordan’s evidence was that in permitting the external release for sub-organisation reporting units, prospective employers might choose to request the People Survey results of managers that apply for jobs outside the Civil Service, or prospective employees from outside the Civil Service might request the People Survey results of a team to which they are considering applying. Ms Jordan set out a number of reasons why this might be undesirable.

Ms Jordan stated that only limited elements of the survey are directly and specifically related to Dr Alexander’s particular focus and identifies those questions at para 51 of her witness statement. Further she states that these questions focus on perceptions of the DfT rather than AAIB. Insofar as the request is for the purposes of examining aspects of Mr. Conradi’s tenure as Chief Investigator it is unclear why Dr Alexander has requested survey results outside that period.

Ms Jordan did not accept that there was a general public interest in the results of a specific sub-unit within an organisation. Permanent Secretaries and Chief Executives are publicly accountable and held to account through the publication of organisation wide surveys. In turn, Permanent Secretaries and Chief Executives hold their other senior officials and managers to account for the results of the organisation’s sub-units.

The public has a right to an effective, efficient and well managed Civil Service and the People Survey is one mechanism for assessing the effectiveness of the Civil Service and identifying and targeting improvements. However there is a legitimate and real concern that the consequences stemming from a decision to release sub-unit results would impact on the quality and granularity of information available from the People Survey in the future thereby hindering the ability of the Civil Service to identify and target continued improvements in its effectiveness.

Ms Jordan believed that the publication of organisational level results provides a sufficient balance between the public interest in understanding the performance and effectiveness of the Civil Service and the public interest in ensuring that the People Survey remains an effective internal management tool.

Ms Jordan was asked by the tribunal if requests for unit-level results were dealt with on a case by case basis or whether a ‘blanket’ approach was adopted. Her answer was that it was ‘a little bit of both’. She explained that having considered the interests of transparency and the need to protect the robustness of the survey they had considered at what level the information should be released. However a survey manager would review a particular request individually and consider which FOIA exemptions might apply. Each request is considered individually, having already considered at what level this information should be released.

The Commissioner makes the following points:

The public interest in favour of withholding the information is very limited. The public interest favours disclosure.

In oral submissions Ms Johns submitted that, in accordance with Davies, we should treat the evidence of a ‘chilling effect’ with some caution.

Ms Johns submitted that the key question was the causal link between disclosure and prejudice, and that in this appeal there was no real evidence of how disclosure would cause the consequences.

The evidence from Ms Jordan was that the disputed information did not contain contextual information, which leaves it open to being misunderstood or potentially misrepresented, and if the media misrepresents things there is a risk of social desirability bias in future surveys. Although this works in theory, in practice there was nothing to tie it to the disputed information. There is nothing to show how any of those concerns are likely to arise from the disclosure of this particular information.

In terms of the impact on the handling of future requests this concern is misplaced. Every request has to be considered on its own merits.

A blanket approach is categorically not appropriate. If information is not within an exemption, it cannot be shoehorned in using arguments about information in hypothetical future cases. This is not a case about where the dividing line should be drawn between what elements of the survey should be published and what should not.

The administrative concerns about the restructuring of the entire survey are overblown. If a manager’s name is included, or there is sensitivity in relation to the number of responses, this information could be redacted when responding to any future request. There is no need to restructure the entire system.

The information is not personal data and there is no rational basis for concerns that individuals might be identified.

In so far as any assurance has been given that there would not be any publication other than in line with the Cabinet Office policy (of publishing only department level reports), that was not appropriate and cannot be relied on as a reason for not disclosing information.

Where results are published, people still do participate and the results are still considered useful.

None of the evidence adds up to a causative link between disclosure and the consequences the DfT is concerned about.

The tribunal is invited to consider the specialist evidence of Ms Jordan, which is candid, analytical and comprehensive and provides a clear and evidenced analysis of likely behavioural changes that would result from an order for disclosure in this case.

The DfT does not advance a ‘chilling effect’ defence as that phrase is used in Davies i.e. by impacting on the ‘safe space’ for deliberation. The overarching concern is that disclosure will have a real and prejudicial impact on the quality of data collected. The cohort surveyed is the entirety of the 500,000+ Civil Service who participate voluntarily and therefore the comments about ‘robustness’ of senior civil servants do not apply.

The principal basis for the opposition to disclosure is that release will bring about behavioural changes that will impact on the quality of the People Survey and there is no parallel with Davies or any other authority concerning a ‘safe space’ for policy formulation.

In relation to the relevant public interest exemption (the applicable interests), the DfT points to the following features of the requested information and People Survey:

In relation to the nature of the prejudice and a causal connection, the principal effect of disclosure would be to reduce participation and to require detrimental changes in the information collected and presented, undermining the utility of the information and its role in achieving the public interest objectives:

There is an obvious causal connection between release of the information, including the threat of future release, and the prejudice, which is clearly more than de minimis.

Although it is probable that a large proportion of civil servants will understand the purpose and desirability of the People Survey, and will be unfazed by publication, others will (legitimately or capriciously) simply take the view that disclosure tips the balance against participation. The tribunal should consider the realistic effects of disclosure, bearing in mind that the vast majority of the civil servants are in operational roles and are not high-level officers formulating policy with strong familiarity with the principles of freedom of information.

Participants can decide not to take part independently and without reasons and could be subject to covert and improper pressure not to participate or manipulate responses.

Even a modest reduction in utility will be detrimental to the People Survey’s objectives.

In terms of the public interest balance, the DfT recognises that there is some public interest in disclosure but it is outweighed by the damage that would be caused to the People Survey.

The public interest in disclosure is already served by the publication of information from the People Survey.

The Civil Service has given serious consideration to what extent it is appropriate, and compatible with the utility of the People Survey, to make results public in the interests of transparency. In part this depends on whether the person to whom it relates is accountable up the line or to the public and Parliament. The size of the reporting organisation is important, seen in its broader structure and operating environment. Certain case studies are published with the consent of the unit involved.

It is relevant that there has already been consideration of disclosure and the Government has balanced the public interest in publication against the need to maintain the fidelity of the People Survey. It is a decision for the tribunal but the assessment is conducted in circumstances where there has already been this careful consideration.

It is not clear what additional public interest would be served by the publication of the requested information.

In oral submissions Mr Brown submitted as follows.

It is acknowledged that the workings of the AAIB are a matter of public interest. The People Survey is hugely important as a management tool. It is one of the few pieces of data that it connected and consistent across the Civil Service.

This dispute takes place in circumstances where regard has already been had to the clear and obvious public interest in freedom of information. The issue is whether the dividing line between disclosure and non-disclosure should be set in a different place from the one set by the Commissioner. The evidence adduced goes to the appropriateness of the dividing line being set in the place where it is currently set.

‘Prejudice to the effective conduct of public affairs’ is broad language.

This is not a case which necessarily turns on the specific detail of the information relevant to the AAIB. The reason why disclosure would prejudice the effective conduct of public affairs is because what was previously perceived to be a risk or a possibility will now have eventuated. This would cross the line from something that was effectively managed by way of assurances to individuals, knowing that FOIA was in place and that anything was susceptible to disclosure, to an actual eventuated situation in which those responsible for the administration of the survey can no longer say confidently to participants that their responses will remain confidential, because the tribunal will have determined the matter against that proposition.

The appeal is concerned more with likely behavioural changes as a result of disclosure as opposed to any concrete concern relating to the information in question.

Ms Jordan’s evidence was measured. Her credentials in the field are impeccable. Her evidence was that, based on her professional judgment, there will be these sorts of behavioural impacts. Once response rates are reduced the overall quality of the data is reduced and the utility of the tool is blunted.

The prejudice relied on is the impact on the fidelity of the information because of a reduction in participation.

Ms Jordan, in her professional judgment, says, as a matter of conjecture, combining the data and what was said by officials in AAIB, that it is having that effect.

In relation to, for example, adverse media coverage, the issue is not really adverse media coverage of the work of the AAIB. This is about what happens if there is a sea change in disclosure of People Survey information. In truth the decision of the tribunal would be changing the position, because the tribunal would be practically declaring that from now on assurances cannot be made in respect of confidentiality. From this point on the risk would have eventuated.

In relation to the extent to which the Survey results serve the particular public interest, Dr Alexander submits:

The tribunal accepts that the People Survey is a fundamentally important and extremely valuable tool for the Civil Service. We accept that there is a clear and weighty public interest in maintaining the integrity of the People Survey.

Mr Brown submits that the ‘chilling effect’ case law is not relevant to this appeal, because this appeal is not concerned with the likelihood of senior civil servants changing their behaviour as a result of fears of disclosure.

First, we observe that this is clearly not a ‘safe space’ appeal, in the sense that it is not about preserving a space to explore radical options without the threat of lurid headlines.

Second, we accept that the entire Civil Service cannot realistically be expected to behave in the manner described in Davies. The reference (taken from DfES v ICO) to ‘highly educated and politically sophisticated public servants who well understand the importance of their impartial role as counsellors to ministers of conflicting convictions’ is clearly not apt to describe all respondents to the People Survey.

We do not accept, however, that concerns about the ‘chilling effect’ argument expressed, inter alia, in Davies are entirely irrelevant to this appeal. The argument in this case is, fundamentally, a ‘chilling effect’ argument in the sense that, as per Mr Brown in his closing submissions, the concerns are about future behavioural changes in response to the perceived likelihood of disclosure of similar information as opposed to any concrete concern relating to the disclosure of the information in question.

It therefore potentially suffers from the flaw identified in the Upper Tribunal decision in Lewis, cited with approval in Davies:

“27.

...The lack of a right guaranteeing non-disclosure of information ...means that that information is at risk of disclosure in the overall public interest ... As soon as this qualification is factored into the candour argument (or the relevant parts of the safe space or chilling effect arguments), it is immediately apparent that it highlights a weakness in it. This is because the argument cannot be founded on an expectation that the relevant communications will not be so disclosed. It follows that ... a person taking part in the discussions will appreciate that the greater the public interest in the disclosure of confidential, candid and frank exchanges, the more likely it is that they will be disclosed...

28.

...any properly informed person will know that information held by a public authority is at risk of disclosure in the public interest.

29.

... In my view, evidence or reasoning in support of the safe space or chilling effect argument in respect of a FOIA request that does not address in a properly reasoned, balanced and objective way:

i)

this weakness, ... is flawed.”

We do not expect all civil servants to be ‘highly educated and politically sophisticated’. We do not expect frontline staff to have freedom of information principles ‘at their fingertips’, as Mr Brown put it. However, we would expect at least a basic understanding by all civil servants of the fact that all information held by a public authority is potentially subject to disclosure in response to a freedom of information request. Similarly we would expect at least a basic appreciation that the lines drawn by the Government/the Civil Service as to what information it publishes voluntarily are not determinative of the position under FOIA.

The argument of the DfT, as we understand it, is that an order that the information be disclosed in this case will mean that what was previously perceived to be a risk or a possibility of disclosure will now have eventuated. In effect, the DfT argues that the always present and understood risk of disclosure will have been ‘brought home’ by our decision.

For the reasons set out above we accept thatthe authorities on ‘chilling effect’ are relevant to the extent that a degree of caution or circumspection is justified in relation to assertions of a ‘chilling effect’. As noted by the Upper Tribunal at para 138 of the appeal against the August decision, this does not mean that the threshold can never be discharged, particularly given the low degree of likelihood required.

In reaching our assessment of whether or not there is a causative link and on the degree of likelihood of prejudice, we are assisted by the observations of Charles J in Lewis in relation to the approach to deciding whether disclosure is likely to have a chilling effect. Although the observations were concerned with the public interest balance, like the Upper Tribunal in Davies, we consider that they are also relevant to whether or not the section is engaged.

We repeat here the summary from Davies:

“30.

Charles J said at [para 69] that the First-tier Tribunal’s decision should include matters such as identification of the relevant facts, and consideration of “the adequacy of the evidence base for the arguments founding expressions of opinion”. He took into account (see [para 68]) that the assessment must have regard to the expertise of the relevant witnesses or authors of reports, much as the qualified person’s opinion is to be afforded a measure of respect given their seniority and the fact that they will be well placed to make the judgment under section 36(2) – as to which see Malnick at [para 29]. … We acknowledge that the application of this guidance will depend on the particular factual context and the particular factual context of the Lewis case, but that does not detract from the value of the approach identified there.”

In our view the applicable interest under s36(2)(c) is clear on the face of the section: the public interest in public affairs being conducted effectively. We accept that the claimed prejudice, i.e. an impact which undermines the fidelity or integrity of the People Survey to the extent that its use as a tool by the Civil Service is blunted, relates to this interest.

We do not accept that any impact, however slight, on the levels or manner of participation in the People Survey in itself amounts to more than de minimis prejudice to the effective conduct of public affairs. It would only do so if the impact was of such an extent that it led to a real and significant risk of undermining the fidelity or integrity of the People Survey to a more than de minimis extent.

We have had regard to the experience and expertise of Ms Jordan. She is the Cabinet Chief Economist and head of the Analysis and Insight Team. She has a bachelor’s degree in economics and a masters’ degree in economics and social policy analysis.

We accept that she has formed the view that there is a risk that some people in AAIB and elsewhere would be deterred from answering the people survey by the disclosure of the requested information. She has reached that view, in the light of her experience and expertise, on the basis of:

Whilst we have regard to Ms Jordan’s experience and expertise, we must consider the adequacy of the evidence base for her opinion.

We accept that the pulse survey response rate in 2019 appears to be low, both when compared to the overall DfT response rate in 2019 and when compared to the, albeit limited, information about response rates received in previous pulse surveys within AAIB set out in closed in para 38 of Ms Jordan’s witness statement.

Ms Jordan has inferred that the pulse survey response rates were low because of knowledge of the FOI request. For the following reasons we disagree that there is sufficient evidence on which to base that inference.

Our consideration of the item of closed evidence is in the closed annex. We conclude that it does not provide any support for an inference that the low response to the pulse survey was caused by knowledge of the FOI request.

The comments reported to Ms Jordan’s colleague do not provide any evidence to support a conclusion that participation in the pulse survey was affected. Those comments were made ‘recently’ in April 2019. They state that staff had stated that they would not participate ‘in the future’ if the AAIB results were published, as a result of certain concerns about Ms Alexander’s future conduct. This cannot be evidence as to why staff had not participated in a pulse survey that had already taken place in March 2019.

There was no attempt to find out why the response rate to the pulse survey was low in March 2019. We accept that the survey was anonymous but in our view it would not have been difficult to ask all AAIB staff anonymously whether they had participated in the recent pulse survey and if not, why not. Ms Jordan does not work within AAIB, and would not have been aware if there were other factors which might have led to a low response rate in March 2019.

Ms Jordan relied on the fact that the pulse survey was around the time of the request as support for her conclusion that the FOI was the reason for the low response rates. She stated that the survey took place ‘soon after’ the request. The request was in March 2018 and the pulse survey was in March 2019. We do not think there is sufficient coincidence of timing to enable an inference that the reason for low participation in a pulse survey in March 2019 was the request made 12 months earlier.

In any event, we note Ms Jordan’s evidence that there are significant differences between a pulse survey and the People Survey. Further, at the time the pulse survey took place the previous FTT had not yet reached a decision.

In our view, a better indication than the pulse survey as to whether our decision to order disclosure would have an impact on response rates to the People Surveys, would be the response rates to the People Surveys which took place after the previous FTT had ordered disclosure of this information in August 2019.

Unfortunately we do not have before us evidence on the response rates to the People Survey in AAIB in 2019 or 2020 (or 2018). However the 2021 response rates (dealt within in more detail below) to the AAIB People Survey were 77%. These are broadly in line with the historical response rates in AAIB.

We acknowledge that by the time the 2021 survey took place, the Upper Tribunal had overturned the previous First-tier Tribunal’s decision which does limit its evidential value to some extent.

Despite this, in our view, the 2021 People Survey participation levels within AAIB remain a better indication of the potential impact of our decision than the participation levels in the 2019 pulse survey because:

We find that the weight of the comments made to Ms Jordan’s colleague in about April 2019 is limited for the following reasons:

Against the limitations of that evidence must be set the participation rates in 2021 which suggest that, despite the impact of the FOI request and the subsequent eventuation of the risk of disclosure of unit level reports, participation levels within AAIB remain broadly in line with what they have always been.

Further we note that some organisations have their results reports published despite being of a comparable size to the AAIB unit. The People Survey still produces useful data in relation to these organisations. This suggests to us that even if Ms Jordan is right to conclude that there was a real and significant risk of some reduction of participation in the AAIB, there is not a risk of a reduction of such an extent that there would, as a consequence, be a real and significant risk of more than de minimis prejudice to the effective conduct of public affairs.

Participation rates in 2021 in AAIB remained broadly in line with historical participation rates, even after the risk of publication had been ‘brought home’ by the August decision. There is no suggestion that participation rates in organisations whose results are published have reduced or are at a level which impacts on the utility and fidelity of the People Survey. These factors suggest to us there is not a real and significant risk that any dip would be extensive enough to impact, more than de minimis, on the effective conduct of public affairs.

Ms Jordan stated in evidence that a 77% response rate was considerably lower than those for peers in the Rail and Marine Accident Investigation Branches, which had a similar profile (95%). She also stated that it was lower than the overall DfT response rates. She relied on this as evidence that the response rates in AAIB were lower than expected, presumably as a result of knowledge of this FOI request.

Ms Jordan accepted in evidence that she was not aware of the difference between the response rate in AAIB and the Rail and Marine Accident Investigation Branches before the FOI request was made, nor was she aware of the historical difference between AAIB response rates and DfT response rates. Without this comparative data we do not think that the difference in response rates provides an adequate basis for Ms Jordan’s conclusion that the AAIB response rates in 2021 were lower than expected.

Further Ms Jordan did not appear to be aware, until taken to it in evidence, that 77% was broadly similar to the historical response rates within AAIB over the period covered by the request. Further we find, on the basis of Ms Jordan’s evidence, that 77% is a good response rate. The fact that a higher response rate would be better does not mean that a 77% response rate risks undermining the integrity and fidelity of the People Survey. We do not accept that any drop in participation, of whatever extent, would lead to a real and significant risk of undermining the utility and fidelity of the People Survey.

Ms Jordan was not aware of any history of the AAIB being misrepresented in the media, or of the media being at all interested in the workings of the AAIB. She could not say whether there was any specific contextual information that AAIB would want to supply to ensure the public would understand the information. There is no reason to think that Dr Alexander will be anything less than responsible in her campaigning.

On this basis we find that there is not a real and significant risk of AAIB resources being diverted such that there would be a more than de minimis impact on the conduct of public affairs as a result of a disclosure of this particular information.

In terms of the burden of providing contextual information, there is no evidence before us to suggest that this would be necessary in relation to the particular information requested, and certainly not that it would be necessary to the extent that there would be a real and significant risk of more than de minimis prejudice to the effective conduct of public affairs.

In terms of any burden in future of providing contextual information or of dealing with adverse press reaction as a result of future disclosures of sub-organisational level reports, we do not accept that this is causatively linked to ordering disclosure in this appeal for the reasons set out elsewhere.

We accept Ms Jordan’s evidence that within survey research there is a well understood concept of social desirability bias: individuals provide answers they think interviewers want to hear rather than an individual’s true belief or experiences. The People Survey uses self-completion methods which minimise this effect.

Ms Jordan extrapolates from this a risk that individuals will not only provide answers they think interviewers want to hear, but will also provide answers they think a future audience might want to read.

The first element relates to what might happen in the event of selective press coverage of unit level results that individuals view as unfair. We do not accept that this future risk, dependent upon unfair press coverage of AAIB results, of which we have found there is no real and significant risk, is causatively linked to this disclosure.

Ms Jordan’s evidence was that this was wider than just a concern about selective press coverage. She said that individuals could respond in a way to align with the public portrayal of a particular unit, so that if there were concerns around bullying and harassment in a particular unit, responses might be made to align more with that concern. There is no evidence of any particular portrayal of the AAIB or any particular concerns relating to AAIB. We do not accept that there is a real and significant risk of this occurring in AAIB.

Ms Jordan stated that individuals might respond to try and please a line manager or whoever they saw as being the audience for the results. As the results are already published within the organisation, we do not accept that this risk flows from external publication.

The results that are already published still provide useful data. We are not satisfied that there is a real and significant risk of this happening to the extent that it constituted a real and significant risk of prejudice to the effective conduct of public affairs.

Many of the risks identified by Ms Jordan flow from the fact that release in this case would ‘set a precedent’ or from the fact that ordering disclosure in this case would, in effect, ‘bring home’ the risk of disclosure of other unit level surveys under FOIA.

We do not accept that this risk flows from our decision. Our decision is not binding on the Commissioner or other First-tier Tribunals. It is based on the particular information requested in this particular case. It does not change the law. The tribunal is not lifting a blanket exemption to the publication of unit or team-level reports. If the Civil Service have been operating on the basis that under FOIA, they were allowed to draw a bright line between organisational level reports and sub-organisational level reports, there was no basis for that view. That is and clearly was never the position under FOIA. It cannot realistically be maintained that the Civil Service have not been aware of this until this FTT pointed it out.

The risk that a tribunal or the Commissioner might order disclosure of another survey at this level is already present. The fact that the Civil Service, or organisations or units within the Civil Service, might not, for whatever reason, have foreseen the risk of disclosure under FOIA for any results below organisational level does not mean that there is any causative link between the disclosure of this particular information and any steps that the Civil Service might choose to take to deal with that risk.

Any reluctance to participate arising simply because there is a risk of disclosure under FOIA, is not in our view caused by our decision, but by the existence of FOIA. The decision to disclose the information in this case does not become the cause of that reluctance simply because our decision might remind people about the existence of FOIA.

Our decision does not represent a ‘step change’. We are not determining where the line should be drawn between reports that should be published and reports that should not be published. We accept that the Civil Service has made a policy decision to voluntarily publish survey results above departmental level, and in cases where the head is directly accountable to parliament. This is relevant to the public interest in disclosure. It cannot alter the fact that FOIA requires a consideration of whether or not prejudice would be caused by disclosure of this particular survey.

Any action that is expected to be taken ‘in order to pre-empt, or in response to large volumes of publication requests’ is not causatively linked to the disclosure of information in this case. There is no basis for a conclusion that disclosure in this case would lead to large volumes of publication requests. The tribunal is not lifting a blanket exemption to the publication of team-level reports.

Ms Jordan is wrong to assume that the Commissioner and the FTT can be expected to order release of other People Surveys at sub-organisational level in the future. Each decision will have to be made on the facts and on the basis of the individual exemptions relied on. They are no more likely to order release as a result of our decision in this appeal.

Any matters that are said to flow from the fact that team level reports ‘cannot be withheld using FOIA exemptions’ are not causatively linked to the disclosure of information in this appeal. Our decision does not change the law. It does not set any precedent. Other team level reports do not face any increased risk of disclosure as a result of our decision. The teams where the total number of staff is sensitive information do not face an increased risk of disclosure.

Where the total number of staff in a team can be calculated from team level reports and where this is sensitive information, this can be dealt with through redaction or the application of exemptions. The risk of disclosure remains the same as it did before our decision. Exemptions that are currently available to resist disclosure of sensitive results or to resist vexatious requests remain available.

Ms Jordan in her statement states that a consistent ‘neither confirm nor deny type approach’ is necessary in relation to the release of People Survey results at sub-organisational level rather than an approach which looks at the specific harm that will be caused by release, because the mere fact of the outcome would indicate that the relevant unit had a problem. Section 36(2)(c) is prejudice based. It entails a public interest balancing exercise. It cannot be applied as a blanket exemption to a particular class of information. The fact that release of other People Survey results at sub-organisational level would cause the particular harms identified in Ms Jordan’s closed evidence is not relevant to whether release of the information requested in this appeal would cause harm.

The fact that some Civil Service units already do not take part in the People Survey or restrict the circulation of their results internally because of FOI concerns illustrates the point: the risk of disclosure of unit level reports already exists under FOIA. The example given in closed evidence was of a decision, taken before this request, by a particular organisation not to include particular reporting units because of concerns about the release of the same type of aggregated survey reports sought by Dr Alexander. That is caused by the existence of FOIA, not by disclosure in this appeal.

For this reason, we do not accept that there is a causative link between ordering disclosure in this case and any decision by the Cabinet Office to mandate that manager’s names cannot appear in the names of reporting units. Further, even if this were the case, we do not accept that such a step would be so disruptive that it would be likely to cause more than de minimis prejudice to the effective conduct of public affairs. If it was going to cause significant problems, an alternative would be to redact the names of managers as and when any particular survey was being provided in response to a FOI request.

In relation to the additional change set out in the closed annex, we do not accept that this is causatively linked to disclosure in this appeal for the same reason. Ms Jordan begins her sentence with, ‘If team level reports cannot be withheld using FOIA exemptions…’. That is not the effect of our decision. FOIA exemptions apply in the same way that they applied before our decision.

In terms of requests from prospective employers or employees we do not accept that there is a causative link between the release of information in this case and future requests from prospective employers or employees for unit-level surveys. We are not making a decision ‘permitting the release of results for sub-organisation reporting units’ in general. Whatever we had decided, there is no permissible bright line under FOIA between the results of the surveys in sub-organisation units and the results of organisation level surveys. This has always been the position and prospective employers or employees could always have chosen to request particular surveys.

For the above reasons, although we acknowledge and give respect to the evidence of Ms Jordan, and take full account of her experience and expertise, we disagree with her conclusion that there is a real and significant risk of prejudice to the effective conduct of public affairs as a result of the disclosure of this information. We find that the exemption is not engaged. The Commissioner did not err and the appeal is dismissed.

We have gone on to consider what our conclusions would have been on the public interest balance if we had concluded that the exemption was engaged, on the basis that the threshold was discharged. For this alternative scenario we assume that we had been satisfied that there was a causative link and that we had concluded that there was a low degree of likelihood, but that it passed the threshold.

If we had decided that the section was engaged, we would have concluded, for the reasons set out above, that there was a low risk of more than de minimis prejudice to the effective conduct of public affairs and that any prejudice was not likely to be severe. We would have taken account of the importance and value of the People Survey, and the fact that there was a risk, albeit small, of at least some reduction in its utility to the Civil Service. Further, we would have taken account of the fact that the Government had considered the need for transparency and had decided to publish only organisational level results. All this would have carried weight in the public interest balance.

In terms of the public interest in disclosure, we would have taken account of the following:

Taking all those factors into account, if we had had to carry out the public interest balance we would have concluded that the public interest in disclosure outweighed the public interest in maintaining the exemption. We therefore would have dismissed the appeal even if we had concluded that s 36(2)(c) were engaged.

In the closed session on 26 April 2022 the Appellant’s witness, Ms Lisa Jordan, was asked additional questions following on from those that were asked in the open session.

Ms Jordan provided the response rates of the AAIB and the DfT to the People Survey 2021, and drew a comparison between those rates (77% against 85%). Ms Jordan explained that the results were considerably lower than those for peers in the Rail and Marine Accident Investigation Branches, which had a similar profile (95%). She was asked questions about how the AAIB response rate compared to its historic response rates as shown in the disputed information. She did not have information on historic response rates by the DfT, so a comparison between AAIB and DfT over time could not be done. Ms Jordan said that 77% was still good, but lower and less robust than higher response rates would be.

As well as response rates, Ms Jordan was asked about the quality of the responses to the People Survey since the request. She was unable to comment on whether the quality of the data in the 2021 survey had been affected by Dr Alexander’s request, but commented that this would be almost impossible to detect in the survey results. Ms Jordan said that it was impossible to police responses to the survey. One cannot see from the data if people have been unduly influenced, as their response has to be taken to be their view.

Ms Jordan also explained that in-year surveys, known as ‘pulse surveys’ may be conducted. These are optional, and if conducted they might be run either by the Cabinet Office or by the Departments concerned. They are different as the People Survey has 100+ questions and a large amount of promotion work to get the highest response rates so as to obtain as robust data as possible. She gave evidence as to the response rate to an AAIB pulse survey in 2019, and stated her view that it was low because it was around the time of Dr Alexander’s request. She said that this was evidence that showed that the FOI request was having a negative impact on response rates. She did not have any information on pulse surveys by AAIB in any other years. Ms Jordan expressed the view that the combination of the low response rate, combined with the views of individuals within the AAIB show that this FOI request is leading to a lower likelihood of responding to the surveys. Ms Jordan says that this was only within AAIB as people do not know about the FOI request outside of AAIB.

Ms Jordan confirmed that:

Ms Jordan explained why she is concerned that publication may lead to a defensive culture in addressing management issues. Where there are non-performing units, or where there are more sensitive results, results might instead be aggregated in future, which would reduce the usefulness of the data rather than resolving problems within teams. Ms Jordan explained that the information is ultimately a tool for the effective running of the team. She said that there should be some scrutiny, which is why data is published at departmental level. She said that to publish small team results could be a distraction from the matter at hand and not be a reflection of what is happening in that team. Ms Jordan gave a specific example about the problems that might arise from publication of protected characteristics information, combined with bullying and harassment information.

Ms Jordan said that the evidence in paragraph 42 was contemporaneous with the telephone call (April 2019). She is not aware of a contemporaneous note having been made of the telephone call to her team from the AAIB, and in view of the passage of time since the events in question she does not remember anything more than what is stated in her witness statement. She did not know how AAIB staff came to be aware of Dr Alexander’s request.