on an application for JUDICIAL REVIEW
At Field House | Decision signed: 07.06.2017 |
on 30.05.2017 | Handed down: 20.06.2017 |
The Queen on the application of Mohd. NAWAZ
Applicant
v
Secretary of State for the Home Department
Respondent
Before Upper Tribunal Judge John FREEMAN
Counsel for the applicant: Barnabas Lams
Counsel for the respondent: Natasha Barnes
(a) Deception in ETS cases is not a question of precedent fact, except in particular circumstances, for example those in Abbas [2017] EWHC 78 (Admin).
(b) There is no fundamental right to study in a foreign country; nor for children to be there with their would-be student parents; nor can a different standard of review fairly be applied in these cases to applicants with and without children.
(c) It follows that the standard of review in all such cases is on ordinary judicial review principles, requiring fair consideration, bearing in mind both the potentially serious effects of deception findings in general, and the requirements of effective administration.
(d) Oral or other evidence of an applicant’s English-language skills or attainments is unlikely to have any decisive effect in judicial review proceedings on the fairness of the decision under challenge, for the reasons given in Habib (JR/1260/2016) [20],and those at[21].
(e) Evidence obtained by use of the Look-up Tool, and subject to the human verification procedure, is an adequate basis for the Secretary of State’s deception finding in these cases, in the light of Flynn & another [2008] EWCA Crim 970 [24 – 27], and the evidence of both Dr Harrison and Professor French.
(f) The lack of visible note-taking by the human verifiers does not provide any ground of challenge to the decision as insufficiently transparent, where there has been an offer (whether accepted or not) to provide a copy of a voice recording for analysis.
DECISION
This is an application for judicial review of the decision of the respondent on 18 March 2016, to curtail leave to remain as a student for a citizen of Pakistan, born there on 25 November 1981. The respondent took the view that the applicant had engaged in deception by getting a certificate from a TOEIC [Test of English for International Communication], administered by ETS [Educational Testing Services], and later reported by them as having been taken for him by a proxy. This certificate was not used in any application for leave to remain. The decision letter did not give details of when or where the test had been taken; but the supplementary decision letter of 5 September 2016 says it was on 19 September 2012 at Elizabeth College (in London).
As it happened, this case came on straight after I had handed down my judgment in Ranjit Kaur (JR 8997-15), a copy of which is attached to this. I had provided copies for counsel in this case too, and made it clear to them that I was not prepared to go over the same ground again; but I would, and did consider any new evidence or arguments put forward.
The material put forward by Mr Lams fell into two main parts:
proposed oral evidence by the applicant himself, and an argument based on Manchester City Council v Pinnock [2010] UKSC 45 as to why I should admit it; and
challenges to the reliability of the respondent’s evidence, based on what Mr Lams said were unresolved contradictions between Dr Harrison and Professor French (see Ranjit Kaur for an introduction), and on Flynn & another [2008] EWCA Crim 970.
The application to file a witness statement by the applicant (8 May 2017) was very late indeed (26 May), and Mr Lams did not apply to call him to give oral evidence till the start of the hearing; but I made it clear that I should not refuse to consider the proposed contents, so far as they seemed reasonably capable of being decisive. At the same time I allowed Miss Barnes to put in a statement by Adam Sewell, which provides an explanation for one of the points made by the applicant. Neither counsel opposed this course.
The applicant’s further statement gave details of how he had passed an IELTS English-language test in 2008: however he had failed a TOEIC test in speaking and writing in December 2011, but passed in January 2012. On 19 September that year he had sat both speaking and writing tests at Elizabeth College, and he gives details of how he got there; on 23 October he had sat the listening and reading tests at Colwell College in Leicester. He says he passed the speaking and writing at Elizabeth College on 19 December, presumably meaning that was when he got the results, and also passed the listening and reading.
However, the applicant said, he had never sat the TOEIC test at Colwell College on 16 November 2011, still less given his date of birth for that as 7 May 1978, as the Home Office claimed. These details did not come from the decision letter, or the further letter of 5 September, but from a statement previously filed by Mr Sewell. His further statement explains that the 19 September 2012 tests at Elizabeth College had provided a date of birth exactly matching the applicant’s; but the 16 November test at Colwell College had not, and had been included in his evidence by a mistake on his own part.
There was more to Mr Sewell’s further evidence; but, as this point was not pursued before me by Mr Lams, realistically in the light of it, I need not go into it any further. What Mr Lams did pursue was the applicant’s suggestion that, in the perhaps not entirely idiomatic words of his statement:
The Home Office allegations are false. I am a highly education individual and very competent in the English language. Before I came to the UK I had passed IELTS exam which is more tough than the TOEIC examination.
The applicant says he had completed an MBA at the University of Cardiff, and was studying for a PhD at Bolton when his leave was curtailed. Similar scenarios have featured in a number of previous ETS cases, including Habib (JR/1260/2016), briefly referred to in Ranjit Kaur. I had not dealt with that decision in any detail at all, partly because it could not help the applicant in that case, and partly because it had not been reported, or permission to cite it obtained.
However, though there is nothing so far to show that Habib has been reported, I note the terms of the Presidential Guidance Note of 2011, no 2, as updated from time to time:
Judgments of the Chamber on applications for judicial review (following grant of permission) in immigration judicial review proceedings3 are automatically given a neutral citation (which includes the letters “IJR”) and reported, without keywords or italicised summary.
Though this does not seem to have happened yet with Habib, clearly it will; so Miss Barnes did not require permission to cite it. I shall return to the relevant parts in due course.
Mr Lams’s argument on Pinnock is that the circumstances of this applicant’s case required a higher standard of review, in terms of the spectrum discussed in Ranjit Kaur, because, as the decision letter itself had made clear, the decision involved the best interests of his child or children. The applicant’s son is named in the letter before claim, though no date of birth or further details have been given there or since; nor was permission granted on anything to do with this point. However, the duty under s. 55 of the Children Act to consider his best interests is acknowledged in the decision letter, and so I shall consider Pinnock in the light of that.
Mr Lams took me straight to paragraph 74 of Pinnock:
In summary. Where it is required in order to give effect to an occupier’s article 8 Convention rights, the court’s powers of review can, in an appropriate case, extend to reconsidering for itself the facts found by a local authority, or indeed to considering facts which have arisen since the issue of proceedings, by hearing evidence and forming its own view.
However, as so often, the background to that statement of principle needs to be considered. Mr Pinnock was the occupier of a house let him on a secure tenancy by the local authority, where he lived with his partner and some or all of their five children. Because of his family’s behaviour, previous County Court proceedings had resulted in a ‘demotion of tenancy order’ under the Housing Act 1996. Since further proceedings were taken by the local authority within a year of this order, that provision entitled them to serve a notice seeking a possession order, subject to a statutory review by them. Unless this procedure had not been followed, the Court was then obliged to make the order sought.
That was the background against which the County Court judge in the current proceedings had decided (see paragraph 19 of the Supreme Court judgment) that his rôle was “limited to conducting a conventional judicial review”, and did not extend to resolving factual disputes. The local authority’s decision to serve a notice within the time limited by the demotion order might be regarded as a condition precedent to the court’s obligation to grant possession.
That was the legal basis on which Mr Lams argued that I should consider the facts of this case and hear evidence for myself. He did not go so far as to suggest that I should consider the best interests of the applicant’s son for myself, and, in view of the basis on which judicial review was sought and permission granted, could not have done so; but he did argue that, because those interests were likely to be seriously affected by the decision under challenge, that decision needed to be subject to a higher standard of review.
As I made clear in Ranjit Kaur at paragraphs 18 – 20, any decision of this kind is one with potentially serious effects, in terms of the time bar on future visa applications; but it is not in itself one which engages the applicant’s fundamental rights; or those of any children he may have (which the applicant in that case did not). No-one has a fundamental right to study in a foreign country; nor, consequently, to be with their parents when exercising any such right as they claimed. What this applicant was entitled to was fair consideration on the evidence of the means by which he had got his current leave, and judicial review of its curtailment, bearing in mind the consequences.
As for the argument that the existence of children required a higher standard of review of the basis for the curtailment decision itself, it is useful to set out the rule for which Mr Lams was in effect contending: someone with children is entitled to set his own evidence against the evidence before the decision-maker, in a way in which a childless person is not. That proposition cannot be right or fair, and I do not accept it. What the best interests of children might require, once the curtailment decision had been made, is a separate question, and, as explained at 14, not one before me.
It follows that I do not accept Mr Lams’s argument on Pinnock that I should consider the facts behind the curtailment for myself, or hear the applicant’s evidence in order to do so. As for the argument that the question of deception was one of precedent fact, so that this was necessary for that reason, it has consistently been rejected by this Tribunal and the courts, except where there was reason for some exception.
One example of that was Abbas [2017] EWHC 78 (Admin), also referred to in Ranjit Kaur. Here the validity of the applicant’s grant of leave was a condition precedent to the success of his application for naturalization as a British citizen. That was why Wm Davis J had to treat the decision on it as a matter of precedent fact, for him to re-examine on that basis. In this case the finding of deception was simply the factual basis for the decision under challenge itself, which has to be considered on ordinary judicial review principles, as set out at 15.
For completeness, I should say that I saw nothing in a final argument by Mr Lams on the standard of review, based on the writer of the supplementary decision letter having set the standard of proof of deception as being on the balance of probabilities. Miss Barnes accepted that this was the correct standard: whether or not it set a higher hurdle than appears in other cases, the question for me is whether the respondent reached a fair decision on all the evidence.
I can now return to Habib. The passage relied on by Miss Barnes is at paragraph 104:
I remind myself that the applicant’s actual English-language competence, whether at the date of the decision or now, is not determinative of this application. It is not for the Upper Tribunal to seek to establish for itself the level of competence which the applicant has in the English language, or what other reason he might have had for choosing to pay someone else to take his test for him, as the respondent and ETS say that he did. As stated in Abbas, there are many reasons why a person might choose a dishonest, rather than an honest, way to obtain an English language qualification. …
To those considerations, which are in no way diminished by being referred to as common sense, has to be added the possibility of language colleges themselves taking the initiative in encouraging the use of proxies. I pointed this out to the parties as a very real consideration in cases such as these, where the whole pattern of a college’s results led to that conclusion.
The only counter-argument put forward by Mr Lams referred to paragraph 94 of Habib. Here Judge Gleeson noted a previous decision by another Upper Tribunal judge, with a contrary result. However, it is clear that this (Kadivar IA 42545-14) was not only an unreported decision on a statutory appeal, but one where the presenting officer had accepted that the appellant had given a credible account and was a reliable witness. On a statutory appeal, he had a right to give his evidence and have the judge consider it for herself on its merits, and it follows that this decision could have no bearing on the question of the standard of review to be applied in this case.
The result is that I am not prepared to accept this applicant’s proposed evidence as capable of having any decisive effect in the present case, which I have to consider on Mr Lams’s remaining points. Before returning to the differences between Dr Harrison and Professor French, I will deal with the effect of Flynn.
Flynn, as the citation will have made clear, was a criminal decision, inevitably depending to a considerable extent on the standard of proof in such cases, where the jury had to be sure that the defendants were guilty before finding them so; and to the judge’s power under s. 78 (1) of the Police and Criminal Evidence Act 1984 to exclude evidence whose admission would be unfair in all the circumstances.
However Flynn was referred to by Dr Harrison: at p 10 of his report he cited this passage:
The increasing use sought to be made of lay listener evidence from police officers must, in our opinion, be treated with great caution and great care. In our view where the prosecution seek to rely on such evidence it is desirable that an expert should be instructed to give an independent opinion on the validity of such evidence. In addition, as outlined above, great care should be taken by police officers to record the procedures taken by them which form the basis for their evidence. Whether the evidence is sufficiently probative to be admitted will depend very much on the facts of each case.
It goes without saying that in all cases in which the prosecution rely on voice recognition evidence, whether lay listener, or expert, or both, the judge must give a very careful direction to the jury warning it of the danger of mistakes in such cases.
This ties in with Dr Harrison’s main contention, which was that only a recognized expert could reliably pronounce on voice recognition evidence. It is interesting to see, in that context, that the Court of Appeal had just made it clear (at paragraph 62) that they considered it:
… neither possible nor desirable to go as far as the Northern Ireland Court of Criminal Appeal in O’Doherty which ruled that auditory analysis evidence given by experts in this field was inadmissible unless supported by expert evidence of acoustic analysis. So far as lay listener evidence is concerned, in our opinion, the key to admissibility is the degree of familiarity of the witness with the suspect’s voice. Even then the dangers of a mis-identification remain; the more so where the recording of the voice to be identified is poor.
Clearly the Court of Appeal regarded lay listener evidence as admissible, even in a criminal case, where the question was one of recognition, and subject to the warning in the last sentence. The present case, and others like it, involve not recognition, but matching of unfamiliar voices, and there is typically no information, one way or the other, about the quality of the recording itself.
Turning to Mr Lams’s argument on the differences between Dr Harrison and Professor French, I need to start by remembering that it could not be based on any suggestion that the decision-maker could not have validly reached the decision under challenge in the face of conflicting expert evidence as such, because neither report was before him. The purpose of both of them, as I made clear in Ranjit Kaur, was to enable me to carry out an informed review of the Look-up Tool evidence on which the decision had been reached.
It is not surprising, especially as both experts come from the same stable (JP French Associates) that there is a good deal of common ground between them. Dr Harrison lists, at paragraphs 4.1 – 4.5.4 of his report, a number of components of the auditory-phonetic analysis to be expected of an expert in the field, which Mr Lams accepted were broadly the equivalent of the 13 features listed by Professor French at paragraph 3.3.1 of his.
At paragraph 4.7 Dr Harrison gives the generally accepted training required for such an expert as at least a master’s degree in phonetics or linguistics, with “… substantial components of phonetics, socio-phonetics and speech acoustics”. That should ideally be followed by one to two years’ specific forensic training, involving shadowing the work of an established expert, and doing independent work checked and scrutinized by a mentor, all over a large number of different forensic cases. Professor French does not dissent from that.
However Professor French also notes at paragraph 3.3.1:
One would expect phoneticians to have a heightened awareness of the features. However, that is not to say that only phoneticians can perceive them. Non-phoneticians are regularly heard to comment on, for example, accent, grammatical errors, distinctive voices or speech impediments. [The approach of ETS’s trained listeners – see Ranjit Kaur for details] was analytical rather than merely undifferentiated holistic listening and … the components and features attended to in comparing the voices matched by the ASR [Automatic Speaker Recognition system] stood in at least a rough correspondence with those outlined above and by Dr Harrison.
Mr Lams pointed out, however, that Professor French had accepted at 3.3.2 that listeners were typically available for two or three days’ training, followed by a mentoring period of varied duration. The trainers themselves might have had no university education in phonetics or speech science. Professor French nevertheless goes on to say that, while this was far from ideal, the areas of training corresponded with those in university forensic speech science courses and “… the descriptions [by ETS] of those areas indicate knowledge of significant issues in speaker comparison testing”.
Another criticism of Dr Harrison’s related to the lack of note-taking by the listeners, which would have provided some transparency and the possibility of checking their results. Professor French accepted at 3.3.5 that this was the case, while giving reasons for not accepting Dr Harrison’s estimate of the time required for the analysis as a whole. Mr Lams criticized ETS’s verification process as ‘completely opaque’, in the light of what Mr Millington [see Ranjit Kaur for introduction] had said at paragraph 29 of his statement [AB 103 in this case] about the lack of details for the biometric technology used in the ASR process, owing to a confidentiality agreement between ETS and the suppliers of it.
Next Mr Lams referred to the points made by Dr Harrison at 5.3: while he did not enlarge on those about duration [5.3.1] or quality [5.3.2] of samples, he noted Professor French’s main conclusions at 4.
The conditions used for trained listener pair confirmation, in conjunction with the (albeit unspecified) conservative thresholds set for ASR match identification … would, in my view, have resulted in substantially more false rejections than false positives.
Even though there is still material missing from the body of information called for by Dr Harrison, I am not convinced that the provision of such information could be used to establish a closely specified percentage of false positives.
If the 2% error established for the TOEFL pilot recordings were to apply to the TOEIC recordings, then I would estimate the rate of false positives to be very substantially less than 1% after the process of trained listeners had been applied. This is because: [Professor French goes on to give reasons for his conclusions].
As Mr Lams pointed out, Dr Harrison had given his view at 5.3.3 on the reference population used in systems such as the ASR:
For the reference populations to do the job correctly, it must match various characteristics of the materials being testing [sic], including recording quality and language. If there is a mismatch with these characteristics then the system is likely to make more errors.
Mr Lams suggested that Professor French’s final conclusion (3) had been based on no more than an assumption about the populations used in the TOEFL and TOEIC tests. I asked him whether there were anything to suggest any significant difference between the two populations, on which he was only able to refer to the general lack of any data at all beyond the spread-sheets obtained through the Look-up Tool.
Finally Mr Lams referred to the difficulties inherent in comparing non-native English-speakers, noted by Dr Harrison at 6.3.7. These were dealt with by Professor French at 3.3.4 on the basis that the listeners
… were in a more favourable position than most lay-listeners, in that they had gained familiarity with foreign accented English through their involvement in the TOEFL program from ‘reviewing speaking responses from candidates around the world’.
Mr Lams described that as highlighting the tentative nature of Professor French’s conclusions.
I need not deal to any extent with Miss Barnes’s submissions, which were largely anticipated in argument, as reflected above. The only significant point to which I have not so far referred was on the question of the contrasting background to analysis of voice recognition evidence, as dealt with in the consideration of individual serious criminal cases in Flynn; and in dealing with what appeared from the ‘Panorama’ programme [see Ranjit Kaur] to be evidence of deception on a massive scale.
While it is clear that the consequences of deception findings in individual cases might be very serious for the subject, I put it to Mr Lams in the course of argument that the duty of the decision-maker must be to give a decision which was fair in all the circumstances. Mr Lams accepted that the answer to the question of what was fair was specific to the circumstances, but suggested that the scale of the suspected fraud should make no difference in an individual case. I cannot accept this: the respondent had a duty to the public, as well as to individuals affected, and needed to balance the interests of effective administration with those of fairness.
CONCLUSIONS
Dealing with Mr Lams’s points on the evidence, the basis of expert analysis of voice comparison evidence is agreed by Dr Harrison and Professor French, in terms of what are essentially common factors. Structured and reasoned analysis of those factors is what can be expected from those who hold themselves out as experts in the field; but not even in criminal cases, at least in England and Wales, is such evidence regarded as essential: see Flynn.
Professor French goes on to make what I consider a very important point: see 31. Although technical analysis of voice comparison evidence is possible, and no doubt the ideal, this is not an exclusively technical subject (as for example might be analysis of blood for DNA or other features). It is one of which all human beings have some knowledge, including knowledge of and ability to comment on the features identified by Dr Harrison and Professor French.
While the training of a recognized expert ideally involves both a master’s degree and what might be described as a practical pupillage to follow, and the ETS training for listeners fell well short of that, they were trained in areas corresponding to the factors identified by the experts. The lack of note-taking, at least in available form, is a deficiency which will have to be taken account of.
So far as the comparability of TOEIC and TOEFL populations is concerned, while there is no direct evidence to support it, neither is there any going the other way. Without at least some reason to the contrary, it might well be supposed that two groups taking English-language tests for non-native speakers, wishing to study or establish themselves in foreign countries, might at least be broadly comparable.
As for the difficulties posed by voice comparison evidence of non-native speakers, Professor French accepts that they do exist; but once again (see 36) he notes the advantages given by practical experience. It is reasonable to suppose that the scale of ETS’s operations provided a good deal of this, from candidates around the world.
Clearly the original ASR system on its own was not capable of independent verification, owing to the confidentiality agreement between ETS and their supplier; but, even in the light of Dr Harrison’s strictures on it, the evidence was held in SM and Qadir (ETS - Evidence - Burden of Proof) [2016] UKUT 229 (IAC) to satisfy the evidential burden on the Home Office. It is not too surprising that it should not have outweighed the evidence given by Dr Harrison and the appellants themselves in their statutory appeals in that case.
Looking at the evidence before the decision-maker in this case, with the benefit of the balanced hindsight provided, not only by Dr Harrison, but now also Professor French, a number of things become clear. Dr Harrison sets what might be called the gold standard for the kind of independent expert analysis of voice comparison evidence which would ideally be required in a criminal case such as Flynn, where one or a relatively small number of speakers will be under consideration.
However Professor French confirms, from the point of view of an at least equally recognized expert, that natural ability, training, even of a fairly basic kind, and experience all play a valuable part. With the hindsight provided by his evidence, as well as Dr Harrison’s, into the system of ASR, together with human verification, operated by ETS, I do not think the respondent can be regarded as having acted unfairly, in this and very many other cases of the same kind, in taking it as the basis for findings of deception.
While the lack of visible note-taking means that direct independent checking of results obtained in an individual case is not possible, this applicant, like Ranjit Kaur, was offered the chance to get copies of the recordings made in his case, so he could have them analysed: see supplementary decision letter paragraph 22. The system as a whole is not unfair, in the context in which it had to be operated.
Application
(a judge of the Upper Tribunal)
[Please see next page for appendix]
[appendix]
Upper Tribunal
Immigration and Asylum Chamber
Application no: JR8997-15
on an application for JUDICIAL REVIEW
At Field House | Decision signed: 25.05.2017 |
on 12.05.2017 | Handed down: 30.05.2017 |
The Queen on the application of RANJIT KAUR
Applicant
v
Secretary of State for the Home Department
Respondent
Before Upper Tribunal Judge John FREEMAN
Counsel for the applicant: Hafsah Masood
Counsel for the respondent: Rory Dunlop
DECISION
This is an application for judicial review of the decision of the respondent on 11 June 2015, confirmed on administrative review on the 26th, to refuse an application for further leave to remain as a student by a citizen of India, born there in 1980.
The first decision refused the application on these grounds:
(Immigration Rules paragraph 322 (2)): the applicant had made false representations (involving proxy taking of an ETS English-language test) to get leave to remain in 2013; and
she had not provided a valid CAS [Confirmation of Acceptance for Studies], which would have required satisfactory evidence of funds.
The decision on administrative review confirmed the decision on (a), and said nothing more about (b). Although the applicant needed to succeed on both points to get leave to remain, it is agreed that she is entitled to challenge the decision on (a), in view of its consequences for the future.
CASE HISTORY
I shall need to go into this, to set the scene for an application with which the hearing before me began.
judicial review claim filed
Upper Tribunal Judge Ward
refuses an extension of time for acknowledgement of service [AoS]
grants permission, with standard directions
UT Judge Allen extends time for detailed grounds of defence
04.11 detailed grounds of defence filed and served
respondent applies for adjournment to await result of lead cases
Mohibullah [2016] UKUT 561 (IAC)
Saha & another [2017] UKUT 17 (IAC)
09.06 UT Judge Blum gives applicant till 28 days after publication of decisions in all lead cases to notify Upper Tribunal whether [s]he intends to pursue this claim, and directs that
if the answer is no, it will be treated as withdrawn;
if yes, then further directions were to follow, beginning with the applicant serving amended grounds, if so advised.
final decision in Saha handed down
time for applicant to comply with Judge Blum’s order expires
hearing listed for 14 May
GLD (Mr Hayes) speaks to applicant’s previous solicitors: they have withdrawn
respondent
offers access to voice recording used on applicant’s ETS test; and
applies for permission to file further evidence
05.05 respondent’s skeleton argument
09.05 applicant’s skeleton argument
respondent replies, with witness statement from Mr Hayes
Since the respondent’s application of 26 April was opposed, the hearing began with Mr Dunlop addressing me on it. Referring to his reply (paragraph 8), he made these points:
the question was whether it was in the interests of justice, in terms of the overriding objective, to admit the evidence: see Nottinghamshire and City of Nottingham Fire Authority v Gladman Commercial Properties [2011] EWHC 1918 (Ch);
the late stage at which it had been put forward was the applicant’s fault, for not complying with Judge Blum’s order;
the applicant’s present solicitors had in any case only been instructed at about the same time as the evidence had been put forward; and (not in the reply)
the evidence was vital to the issues before me; and the applicant had not taken up the offer to examine the voice recording;
it would, even if the applicant succeeded in having the decision under challenge set aside, be relevant to the relief to be granted; and finally
similar evidence had been consistently considered by the Upper Tribunal, in Saha, Mohibullah and elsewhere.
On (4), Miss Masood explained that the offer about the voice recording had only been made to the applicant’s previous solicitors: she hadn’t heard of it herself till the morning of the hearing. I invited Mr Dunlop to confirm that it remained open, which on instructions he did; but Miss Masood said the applicant preferred to go on with the hearing.
Miss Masood argued that the applicant had been entitled to know in good time what the case against her was; and, out of the further evidence put forward, only the statement of Jacqueline John-Rose (signed 25 April 2017) had not been already in existence before Judge Blum’s order came into effect on 10 February. The reports by Professor French and Mr Heighway were over a year old: in fact, though Professor French’s is dated 20 April 2016, Mr Heighway’s did not appear till last December. Documents they referred to had not been disclosed, under the respondent’s duty of candour.
My conclusions on the respondent’s application are as follows: the overriding objective is to deal with cases fairly and justly; the evidence put forward is obviously highly relevant, for reasons which will become clear when I discuss it; so unless it cannot be admitted, fairly to the applicant in terms of the case history, and of her ability, through Miss Masood, to deal with it, then in the interests of justice I ought to let it in. While, as Miss Masood pointed out, it had not been in existence at the date of the decision under challenge, so could not be relied on by the decision-maker, it was clearly relevant to the reliability of the evidence which had been relied on.
Dealing first with the history, as I put it to Miss Masood, the case was on ice till 10 February this year: it might with hindsight be regretted that Judge Blum’s order, perhaps in standard terms, given the large number of cases like this before the Tribunal at the time, did not say anything about the consequences of the applicant not saying anything at all before the deadline about what she wanted to do. However, he had very clearly directed her to do so, and, under r. 2 of the Upper Tribunal Procedure Rules (if authority were required for this)
Parties must
help the Upper Tribunal to further the overriding objective; and
co-operate with the Upper Tribunal generally.
The applicant failed to do anything at all to communicate with the Tribunal till she put in her skeleton argument on 9 May, only five days before the hearing; and then only in response to the respondent’s. Mr Hayes explains in his statement why no action had been taken on the respondent’s side before 26 April, and I do not blame him, or her in the least for that. Putting it in ordinary language, if mixing metaphors: till 10 February this case had been on ice, and since then the ball had been very much in the applicant’s court. The fact that she and her present solicitors had not received the further evidence till 26 April, a fortnight before the hearing, was in reality her fault, whatever the terms set out in the standard directions, given back in September 2015.
So much for the case history: turning to the applicant’s situation on the date of the hearing before me, it remained open to her, whenever the opportunity had originally been offered, to examine the actual evidence on which the decision under challenge had been based, and to test Professor French and Mr Heighway’s expert evidence against the thing itself. On advice, she chose not to take this opportunity; but to adopt the tactical position of seeking to exclude the evidence altogether.
I am not blaming Miss Masood for that, since she had clearly been put in a difficult position by the applicant’s very late instruction of her present solicitors, and no doubt Miss Masood herself, coupled with the applicant’s previous complete inaction from 10 February onwards. However, litigation is not a game, but a serious business whose object is fairness and justice. On both heads I had no hesitation in ruling that the further evidence should be admitted.
APPLICANT’S CASE
Miss Masood put her case on two main grounds: first, the decision under challenge had been based on evidence of insufficient quality; next, the applicant had not been given an opportunity to answer the allegations against her. While what I have already said is of some relevance on these substantive issues, as well as the procedural ones discussed, Miss Masood developed her submissions at some length, and I will deal with them.
On Miss Masood’s first ground, she began by pointing out that the decision under challenge, as made, had been based on nothing more than an ‘invalid’ match on the Look-up Tool. Since these and other terms used, though well known to those involved in litigation of this kind, are very far from plain, or even legal English, I had better do what I can to explain them, from the decided cases, and then go on to discuss the standard of review, before turning to assess the evidence itself.
SM and Qadir (ETS – Evidence – Burden of Proof)[2016] UKUT 229 sets the scene: TOEIC [Test of English for International Communication] refers to the test taken by those involved in these cases, so as to satisfy the requirements of the Rules by getting a certificate establishing their English-language skills; ETS [Educational Testing Services] is ‘the international entity which provided the Home Office with services relating to English language testing in the United Kingdom’; the Look-up Tool is the Home Office system under which ETS testing analyses were matched to the person with the name, date of birth and nationality of the certificate holder. Matches were graded either valid, invalid, or questionable.
Standard of review Miss Masood referred to the test used by the Tribunal (McCloskey P and UTJ Rintoul) at paragraph 63 of Mohibullah: whether the decision under challenge “… lay within the spectrum of decisions open to the hypothetical reasonable decision-maker”. She suggested that, despite the Tribunal’s reference to Keyu & others [2015] UKSC 69, this represented an incorrect reading of that authority. (The same panel in Saha nevertheless described it, at paragraph 61, as an ‘elevated hurdle’).
This was rather a bold submission to make at the present stage of these proceedings, contrary to the reported decision of a Presidential panel of this Tribunal; but I need to deal with it before going any further. While Keyu involved a great deal of very learned discussion of the standard of review, it is quite clear that the Supreme Court declined the appellants’ invitation to lay down a new one. Perhaps this was most clearly set out by Lord Kerr at paragraph 278:
… the present appeal … is not the occasion to review the authorities. Final conclusions on a number of interesting issues that arise in this area must await a case where they can be more fully explored. These include whether irrationality and proportionality are forms of review which are bluntly opposed to each other and mutually exclusive; whether intensity of review operates on a sliding scale, dependent on the nature of the decision under challenge and that, in consequence, the debate about a ‘choice’ between proportionality and rationality is no longer relevant; whether there is any place in modern administrative law for a ‘pure’ irrationality ground of review ie one which poses the question, ‘could any reasonable decision-maker, acting reasonably, have reached this conclusion’; and whether proportionality provides a more structured and transparent means of review.
I shall proceed on the basis that the question posed by Lord Kerr, referring to ‘any reasonable decision-maker’, is still the one to be answered in judicial review proceedings of this kind. Referring to a previous Court of Appeal decision, in which he had taken part himself, Lord Mance at paragraph 53 of Kennedy v. Information Commissioner [2014] UKSC 20 alluded to the spectrum of cases between those where “… a low intensity of review is applied to cases involving issues ‘depending essentially on political judgment’”, and “… decisions infringing fundamental rights, where unreasonableness is not equated with ‘absurdity’ or ‘perversity’ and a ‘lower’ threshold of unreasonableness is used”.
This is the spectrum referred to by McCloskey P in Mohibullah: perhaps he considered the concept too well known to require further elaboration. While I certainly agree with Miss Masood that the decision now under challenge had nothing to do with political judgment, on the other hand it did not infringe the applicant’s fundamental rights, which are not engaged by Rules regulating a person’s ability to pursue their studies outside their own country. Nevertheless, it was a decision with potentially serious consequences for her, in terms of the time bar (Footnote: 1) on future visa applications, and I shall bear that in mind.
The other authority Miss Masood referred to on this point was Khawaja [1983] UKHL 8. At paragraph 39 Lord Wilberforce set out the functions of the courts, as will be seen, but first making it clear that the case in hand involved someone not only subject to removal as an illegal entrant who had practised deceit, but detained for that to take place. This applicant, on the other hand, has not been detained, and would have every opportunity to leave this country voluntarily to avoid that happening.
This is the relevant passage from Khawaja:
The court's investigation of the facts is of a supervisory character and not by way of appeal (it should not be forgotten that a right of appeal as to the facts exists under section 16 of the Act of 1971 even though Parliament has thought fit to impose conditions upon its exercise). It should appraise the quality of the evidence and decide whether that justifies the conclusion reached - e.g. whether it justifies a conclusion that the applicant obtained permission to entry by fraud or deceit. An allegation that he has done so being of a serious character and involving issues of personal liberty, requires a corresponding degree of satisfaction as to the evidence. If the court is not satisfied with any part of the evidence it may remit the matter for reconsideration or itself receive further evidence. It should quash the detention order where the evidence was not such as the authorities should have relied on or where the evidence received does not justify the decision reached or, of course, for any serious procedural irregularity.
With the reservation about this case set out at 19, that is the approach I shall follow.
Assessment of evidence Besides suggesting that use of the Look-up Tool was not enough to produce a reliable result on its own, Miss Masood referred to the facts she said should have been taken into account by the decision-maker. The Home Office computerized case-notes [GCID] showed she had been doing a course for the Graduate Certificate in Hospitality Management, with Foundation English, at the University of West London, before taking the ETS test in 2013.
As for the ETS test, the only information available to the decision-maker had been from ETS itself, and from the Home Office witnesses Peter Millington and Rebecca Collings (the so-called generic evidence, since it applied to all ETS cases). In Qadir, this evidence was subjected to the following main criticisms:
Neither witness has any qualifications or expertise, vocational or otherwise, in the scientific subject matter of these appeals, namely voice recognition technology and techniques.
In making its decisions in individual cases, the Home Office was entirely dependent on the information provided by ETS. At a later stage viz from around June 2014 this dependency extended to what was reported by its delegation which went to the United States.
ETS was the sole arbiter of the information disclosed and assertions made to the delegation. For its part, the delegation – unsurprisingly, given its lack of expertise – and indeed, the entirety of the Secretary of State’s officials and decision makers accepted uncritically everything reported by ETS.
On the other hand, the Tribunal in Qadir (a statutory appeal) found (at paragraph 68) that, on the basis of this evidence, the Secretary of State had discharged the evidential burden of raising a case of deception which called for an answer from the appellants; but, considering their own evidence and that of their expert witness Dr Harrison, went on to find that the legal burden of proving deception had not been discharged.
In Gazi (ETS – judicial review) (IJR) [2015] UKUT 327 (IAC) Dr Harrison’s evidence was summarized as follows at paragraph 20:
In principle, the ETS methodology constitutes “a reasonable approach”. However, the specifics of its implementation are insufficiently particularised in the Respondent’s evidence, which suffers from “a lack of technical information and detail”.
The Secretary of State’s evidence fails to acknowledge that the human verification mechanism “is almost certain to have resulted in false positive results”.
The fact of an unknown number of false positive results results in “an unknown number of test takers who have been incorrectly identified as having fraudulently taken the TOEIC test”.
The accuracy and reliability of the ETS results overall cannot be gauged in the absence of sufficient technical knowledge of the process.
This inadequacy could be rectified to some extent by disclosure of the audio material from individual tests, which would facilitate independent auditing through the auditory-acoustic phonetic testing methodology.
The complaint at (v) about failure to disclose voice-recordings is of course in striking contrast to the position now reached in the present case. However, in Gazi at paragraph 34 the Tribunal (McCloskey P) declined to find that the respondent had acted with an ‘improper purpose’ by making a decision to remove the applicant on insufficient evidence, with only an out-of-country right of appeal, on the basis that Dr Harrison’s evidence had come after the decision under challenge, which is not the case here.
As will be remembered, the test this applicant had taken, allegedly by proxy, in 2013, was the TOEIC; but Miss Masood pointed out that ETS’s testing tool had originally been developed for the TOEFL [Test of English as a Foreign Language]. See Mr Millington’s statement paragraph 26:
… the intention had been to further establish the technology within the TOEFL system before applying it to the TOEIC system. However, following the revelations in the Panorama programme [February 2014], ETS decided to deploy it as an analytical tool to retrospectively identify levels of abuse across EPN (Footnote: 2)
Although Mr Millington noted (paragraph 31) that pilot testing of the technology had shown an error rate of less than 2%, Miss Masood’s point was that his evidence did not show whether or not the TOEFL and TOEIC populations were comparable. As for the human verification process which followed the use of the tool, there was nothing to show the error rates involved in that.
Procedural unfairness This formed the other main part of Miss Masood’s case, as to whether the applicant had been given enough opportunity to answer the charge against her. While the gist of it had been given in the decision letter, Miss Masood suggested that the context (see Doody [1993] UKHL 8) required something more. The relevant context was that this was a decision with no right of appeal, and only a limited right of administrative review. At the date of the decision under challenge the Rules did not allow submission of fresh evidence on applying for administrative review of a paragraph 322 (2) decision such as this, though that has since changed. Miss Masood’s submission was that the respondent had had a duty to allow the applicant to make representations addressing the gist of the decision.
This point had been referred to in Mohibullah at paragraph 80:
We recognise that in theory the Applicant could have made representations to the Secretary of State after communication of the impugned decision. However, it is far from clear that the Secretary of State would have been either willing or, more fundamentally, legally empowered to revoke the decision. Furthermore, there is the inescapable and prosaic reality highlighted by [previous decisions of the courts] that the exercise of post-decision representations, in the real world, may well be an arid one.
Miss Masood went on to refer to one of those decisions, SP [2004] EWCA Civ 1750: I have cited the relevant part at some length below, because I agree with Mr Dunlop that it is a case where the context is particularly important. Hooper LJ, giving what was in effect the judgment of the court, said this:
Unless constrained by authority to find otherwise, I have no doubt that contemporary standards of fairness and the principles in PSO 4950 required the appellant to have given SP an opportunity to make representations before the order was made to remove her to the segregation unit for reasons of good order and discipline.
As I have shown, there is a substantial difference between the regime for a 17 year old on her unit and in the segregation regime. Although not formally a punishment measure, removal to the segregation unit is in fact the most severe punishment that can be awarded following disciplinary charges for an inmate of a YOI. Although the purpose of sending the respondent into segregation was to protect others from the risk of harm, nonetheless there must be in the eyes of the detainees and others a substantial punitive element.
The appellant’s own PSO requires all staff must deal with the young women in their care openly and fairly, and to be mindful of their vulnerability (paragraph 46 above). A requirement to give an opportunity to an inmate such as SP to comment on the proposed reasons for removing her to the segregation unit would fulfill [sic] the objective of dealing with the inmate “fairly and openly.”
Respondent’s additional evidence Miss Masood now turned to deal with the respondent’s case, as set out in the evidence which I had allowed Mr Dunlop to put in. She first referred to the well-known principles set out in Fordham’s Administrative Law: public bodies are not to act lightly, and refusal of relief, on the basis that proper consideration of an application would have made no difference, is to be exercised sparingly. Though Miss Masood did not refer to this authority, the principles to be applied by the Tribunal, on a claim such as this, filed before 8 August 2016 (Footnote: 3), are set out in Ermakov [1995] EWCA Civ 42.
Dealing with the evidence itself, Jacqueline John-Rose had used the Look-up Tool to summarize the results from the applicant’s New London College test centre on the day in question, 27 February 2013. The outcome is at exhibit JJR4 to her statement: it shows 88% invalid, and 13% questionable (Footnote: 4) for the day as a whole, and comparable figures of 89% and 11% for the morning only. Adam Sewell, analysing the results for New London College for TOEIC tests taken between 20 March 2012 and 15 May 2013 had given them as 74% invalid, and none genuine, leaving 26% questionable.
However, according to Miss Masood, all this amounted to no more than circumstantial evidence of limited value, and was ‘inherently implausible’, though she did not explain why. She suggested that, if the applicant had been given the chance to do so, then she might have wanted to get in touch with the college about her own results. I inquired whether it was still in being: Mr Dunlop told me that it was the subject of a pending judicial review case of its own; so perhaps its continued existence as a seat of learning may be questionable.
Miss Masood went on to deal with the evidence of Professor French, which he helpfully summarized at p 12 of his report. Some explanation will be required by anyone who has not read the body of it: ‘ASR’ means Automatic Speaker Recognition; ‘trained listener pair confirmation’ is a more detailed description of the ‘human verification process’ already referred to, while ‘false positives’ represent wrong classification of results as invalid, with ‘false rejections’ being the reverse. ‘MFCCs’ are ‘mel frequency cepstrum coefficients’; but, as Professor French provides no explanation for this term, so neither can I.
This is what Professor French says there: it seems best to set it out in full, so Miss Masood’s arguments can be followed.
The conditions used for trained listener pair confirmation, in conjunction with the (albeit unspecified) conservative threshold set for ASR match identification (witness statement of Peter Millington, para 31) would, in my view, have resulted in substantially more false rejections than false positives.
Even though there is still material missing from the body of information called for by Dr Harrison, I am not convinced that the provision of such information could be used to establish a closely specified percentage of false positives.
If the 2% error rate established for the TOEFL pilot recordings were to apply to the TOEIC recordings, then I would estimate the rate of false positives to be very substantially less than 1% after the process of assessment by trained listeners had been applied. This is because:
there were stringent criteria for verification by the trained listeners;
the trained listeners had potentially more speech available from the tests than that processed by the ASR;
the trained listeners had available a much wider range of speech features on which to base their decisions than just vocal tract resonances as reflected in an MFCC analysis performed by the ASR.
Even if the TOEIC recordings were on average somewhat shorter and poorer in quality than the TOEFL pilot test recordings, on the basis of the information that has been provided, I would still estimate the number of false positives emanating from the overall process of ASR analysis followed by assessment by two trained listeners to be very small.
Miss Masood noted that Professor French had said at paragraph 3.3.2 of his report that “It remains unclear exactly how much training the listeners underwent …”. However I note that he goes on to summarize the position as follows:
… it would appear from the information given that the trained listeners at ETS would be in a significantly less advantageous position than forensic phoneticians/speech scientists (minimum of master’s degree in the subject area and 1 – 2 years working under supervision), but in a distinctly better position than untrained lay-listeners attempting the same task.
Again, Miss Masood referred to Professor French’s remarks (at 3.3.6) about Dr Harrison’s comments on the lack of testing for new listeners: to summarize them, Professor French notes Dr Harrison’s description of the induction given as “peer [review] by an experienced analyst until a level of confidence was reached that they were capable of carrying out the work on their own”. He says that, while this was not ideal, new listeners, so far as practical work is concerned, are in the same position as new speech scientists, who are also subject to in-house mentoring, rather than any formal testing.
Finally, Miss Masood referred to the rejection letters from two universities (West of Scotland [UWS], 25 November 2014; and Reading, 4 December that year) refusing her a place as she had previously sat a TOEIC test. While Miss Masood relied on these to show a direct link between the decision-making process and the applicant’s lack of further academic progress, it seems to me even more noteworthy that these rejections had given the applicant reason at least to suspect that her 2013 result would be regarded by the Home Office as at best unreliable, over six months before the decision under challenge on 11 June 2015. If she had been in any kind of doubt as to why, a very little inquiry would have pointed to what was disclosed in the ‘Panorama’ programme as the reason.
RESPONDENT’S CASE
Reliability of the evidence Mr Dunlop pointed out that the challenge to the use of the Look-up Tool had been rejected by the Tribunal at paragraph 35 in Gazi, and at 63 in Mohibullah. As for the further evidence since filed, the Tribunal in Saha (McCloskey P and Judge Rintoul) had regarded it as allaying concerns previously expressed in Qadir . In Abbas [2017] EWHC 78 (Admin) Wm. Davis J had considered the evidence of Dr Harrison, Professor French and Mr Sewell (see 32) at paragraphs 12 - 13, and at 16 concluded that Mr Sewell’s evidence provided significant circumstantial support for the proposition that anyone who supposedly took the TOEIC test at the centre in question was party to a fraud. There was no witness statement filed by the applicant to gainsay this body of evidence.
Mr Dunlop went on to refer to the acknowledgement by the Tribunal at paragraph 57 in MA (ETS – TOEIC testing)[2016] UKUT 450 (IAC) of the range of reasons why people with perfectly good English might engage in TOEIC fraud. I suggested to the parties that, besides these personal reasons, another might have to be added: where, as here, the great majority of a college’s results were classified as invalid, and the rest questionable, the college, rather than the individual students, might be assumed to have taken the initiative. If that is right, and no reason was put forward against it, then the students’ collusion in the process was to an extent understandable; but still something for which individuals would have to take the responsibility. Finally Mr Dunlop referred to the offer to make voice-recordings available in the present case, already noted at 25.
Procedural unfairness Mr Dunlop referred to the point already dealt with at 38 about the warning given the applicant by the rejection letters she had received, before the decision under challenge: a similar situation had been dealt with in Mohibullah at paragraph 82. He referred to the Divisional Court decision (later upheld on this point in Berriew & others [2013] EWCA Civ 199 at paragraph 75) for the proposition that, where consultation would have made no difference, a decision should not be set aside for lack of it. Finally Mr Dunlop sought to refer to Habib (JR/1260/2016); but, since this was an unreported decision of the Upper Tribunal, and he had not obtained the necessary permission under the Practice Directions, I need not deal with that.
APPLICANT’S REPLY
Miss Masood correctly pointed out that the further evidence put forward had only been produced after the hearing in Saha: however, it does not seem to me that this could adversely affect its reliability; nor could the view taken on that by Wm. Davis J in Abbas be adversely affected by his having dealt with the question of deception as a matter of precedent fact (since the previous grant of leave was a condition precedent to the validity of the application for naturalization). She accepted that the ‘Panorama’ programme had been broadcast in February 2014, so putting the allegations of generalized fraud in ETS tests into the public domain; but she said the rejection letters in November and December that year were too vague to give the applicant notice of the problem she was about to face in renewing her leave.
CONCLUSIONS
By the time this applicant made her application for leave to remain on 23 December 2014, she would have been well aware what her recent rejection letters, which meant she was unable to provide the necessary CAS [Confirmation of Acceptance for Studies], were based on. The more explicit of the two is the UWS one, which referred to the Home Office having ‘serious concerns’ on the authenticity of the TOEIC test she had taken. It does not seem to me, at least in the particular circumstances of this case, that there can be anything in Miss Masood’s argument, however carefully put together, that the applicant was entitled to advance warning that the respondent had it in mind to refuse her application on the very same ground.
Turning to the reliability of the evidence on which the decision was reached, as already made clear at 16 – 20, I approach it from the stand-point of whether any reasonable decision-maker could reasonably have reached it, bearing in mind its consequences in terms of the spectrum referred to in the authorities: not so serious as the immediate segregation of the young person in SP, or the imminent removal of the détenu in Khawaja; but serious enough for her to be barred from applying for a visa for a whole year if she left voluntarily at her own expense, and a good deal longer if she did not.
As for Miss Masood’s argument that the validity of the decision could only be judged on the material available at the time, I have to reject it. Putting the situation at its simplest, the raw material was provided to the decision-maker by the Look-up Tool; and it was the validity or otherwise of that raw material which was discussed in turn by Dr Harrison, Professor French and the other witnesses.
It was the whole body of evidence to which I have just referred which has been exhaustively considered by various panels of this Tribunal, many of them Presidential ones, in the authorities to which I have been referred, as well as by Wm Davis J in Abbas. I do not see any grounds for distinguishing the basis of assessment used in those cases. It was a perfectly straightforward question for all the decision-makers, and the fact that Wm Davis J had to approach it as one of fact for himself makes his decision all the more persuasive.
While the state of the evidence in Qadir made it possible for the appellants to satisfy the Tribunal, on their own evidence and those of their expert, that the respondent had not satisfied the legal burden of proving deceit, the evidence which she has put forward since has invariably satisfied both courts and this Tribunal that evidence obtained through the ETS Look-up Tool entitled a reasonable decision-maker to refuse an application made in these circumstances. It is certainly a pity that this evidence was not assembled in the first place; but its effect is very clear.
Finally, there is the position of this applicant to be considered. Having received the administrative review refusal on 26 June 2015, she applied in good time for judicial review on 23 July, and was not obliged to take any further steps in the proceedings till 10 February 2017, when the time given by Judge Blum, following the final decision in Saha, expired. However she did not take any steps to investigate the factual basis for the decision under challenge in her own case, or to put forward any evidence to counter it.
After 10 February, the applicant let the case remain asleep till after 26 April, when Mr Hayes, late or not, did his best to find out what was going on on her side. Even if, as she says, the offer to provide the voice recording used for her test was only communicated to her on the date of the hearing, it was there to take up if her real concern was to test the reliability of the evidential basis for the decision under challenge.
Even without those considerations, however, I regard the decision under challenge as neither procedurally unfair, nor evidentially unreliable, for the reasons already given.
Application
(a judge of the Upper Tribunal)