Neutral Citation:  UKUT 00112 IAC
DK & RK (ETS: SSHD evidence; proof) India
(Immigration and Asylum Chamber)
At Field House
THE IMMIGRATION ACTS
Heard on 1 and 2 March, 18, 19 and 25 November 2021
Promulgated on 25 March 2022
THE HON. MR JUSTICE LANE, PRESIDENT
MR C M G OCKELTON, VICE PRESIDENT
THE SECRETARY OF STATE FOR THE HOME DEPARTMENT
For the Appellant DK: Mr P Turner and Mr J Gajjar, of Imperium Chambers
For the Appellant RK: Mr R Ahmed and Mr Z Raza, instructed by Charles Simmons Solicitors
For the Respondent: Ms L Giovanetti QC and Mr C Thomann, instructed by Government Legal Department.
For the Intervener: Mr M. Biggs, instructed by Moorehouse Solicitors
1. The evidence currently being tendered on behalf of the Secretary of State in ETS cases is amply sufficient to discharge the burden of proof and so requires a response from any appellant whose test entry is attributed to a proxy.
2. The burden of proving the fraud or dishonesty is on the Secretary of State and the standard of proof is the balance of probabilities.
3. The burdens of proof do not switch between parties but are those assigned by law.
DECISION AND REASONS
1. This decision forms the next episode in the saga of cases arising from the Test of English for International Communication (“TOEIC”) certificates obtained from test centres in the United Kingdom administering tests set by the Educational Testing Service (“ETS”).
2. The BBC Panorama programme broadcast on 10 February 2014 exposed widespread cheating in tests of English needed for immigration purposes. Subsequent investigation revealed that many thousands of results were obtained by fraud, many of them by impersonation of the candidate at the speaking test. The gross results have been confirmed by criminal convictions of a number of those involved in running test centres.
3. The Tribunal, and the High Court, have been and are concerned with challenges made by those identified as having cheated. In these cases, the result attributed to them has been identified by the testing body, ETS, as provided by a proxy or impersonator. The candidates, however, have claimed that they took the test honestly.
4. In this decision we examine the evidence on which the Secretary of State relies to establish the frauds in individual cases. We conclude that despite the general challenges made, both in judicial proceedings and elsewhere, there is no good reason to conclude that the evidence does not accurately identify those who cheated. It is amply sufficient to prove the matter on the balance of probabilities, which is the correct legal standard. Although each case falls to be determined on its own individual facts and evidence, the context for any such determination is that there were thousands of fraudsters and that the appellant has been identified as one of them by a process not shown to have been generally inaccurate.
5. Because of the evidence of widespread fraud in the test process, the Secretary of State reached the view in a large number of cases that the certificates obtained by individuals, and used in order to obtain extensions of leave, had been obtained dishonestly. The consequence has been decisions adverse to the individuals in question. In some cases existing leave has been cancelled. In other cases a subsequent application for leave to remain has been refused on the ground that previous leave was obtained by deception, that is to say by use of a TOEIC certificate obtained fraudulently.
6. The Secretary of State’s evidence has developed over the course of the litigation relating to these cases, from a stumbling and insecure beginning (see Ahsan v SSHD  EWCA Civ 2009;  Imm AR 531 at  and ). These appeals have been remitted to this Tribunal by the Court of Appeal. They appeared to give an opportunity for an up-to- date evaluation of the state of the evidence produced by the Secretary of State in ETS/TOEIC cases. In this decision we consider the impact and effect of that evidence as a whole. We consider whether it is sufficient to call for a response by the present appellants and others in a similar position. We then determine the appeals before us on the basis of all the evidence adduced in these individual cases. We attempt to give some guidance on the approach to TOEIC/ETS appeals in general.
7. The basic background facts are not in dispute. The account which follows is derived from the witness statements of Rebecca Collings and Peter Millington on behalf of the Secretary of State and the analysis of them and other evidence in some of the earlier cases, particularly R (Gazi) v SSHD  UKUT 00327,  Imm AR 1127; SM and Qadir v SSHD  UKUT 229, R (Mehmood and Ali) v SSHD  EWCA Civ 744,  Imm AR 25 and R (Sood) v SSHD  EWCA Civ 831,  Imm AR 61. Because of the extensive setting-out of the principal evidence in those cases, particularly Gazi, we do not repeat the material here.
8. In 2008 the Secretary of State decided that it would be appropriate to introduce a test of facility in the English language as part of the process for determining whether leave to enter or remain in the United Kingdom for certain purposes should be granted. Between 2008 and 2010 the requirement to show evidence of ability to speak and understand English adequately was introduced into the requirements of the rules for entry clearance for leave to remain or work in the United Kingdom under Tier 1, Tier 2 (General, Ministers of Religion and Sports Persons) and Tier 4 of the points based system, as well as for applications by partners and parents of persons settled in the United Kingdom, and applications for settlement and for British nationality. If an applicant’s country of origin was not a majority English-speaking country, and the applicant did not have, and was not studying for, a higher education qualification that was taught in English, there was a requirement to obtain a satisfactory test result from one of a number of independent testing services. One of them was ETS.
9. In 2010 the Secretary of State decided that that arrangement should be replaced by one in which a small number of testing services would be licensed by the Home Office, and only tests taken with those providers would meet the requirements of the rules. Six providers were approved to work under licence, of which again ETS was one. The licences began on 6 April 2011. The licensee had the responsibility for ensuring the integrity of the test procedure. The tests were TOEIC and the test of English as a Foreign Language (TOEFL), both of which are recognised internationally. The level of competence required varied according to the immigration category under which an application was made. It may be as well to observe at this point that TOEIC is a test and ETS is a testing (or examining) service. Neither is concerned with teaching, or with any other method by which test candidates might acquire facility in English. In some cases, however, a college where English is taught may also be a test centre.
10. During the currency of the licences, the Home Office investigated a number of concerns, including what were regarded as suspicious levels of certificates with top scores issued at centres run under the auspices of ETS. The Home Office received, and apparently accepted, a report from ETS that the results had been investigated and found genuine. Various other measures to reduce fraud were also proposed.
11. On 6 January 2014, five weeks before the Panorama programme was broadcast, the BBC wrote to the Home Office summarising the results of an investigation into the integrity of testing at two ETS centres.
(i) Registered candidates standing aside from the secure computer terminals, allowing other people (“fake sitters”) with superior English language skills to take the oral and written parts of the exam on their behalf. The fake sitters were organised by the very staff who were supposed to ensure the proper conduct of the exam.
(ii) Verification trips, intended to act as proof that the registered candidate sat the exams themselves, being falsified by staff of those centres in order to facilitate this fraud.
(iii) Exam “invigilators” at one centre dictating the correct answers to the registered candidates in the multiple choice part of the exam.
(iv) At the other centre multiple choice exam answer papers were filled out and submitted without the registered entrant even being present.
12. The two centres that the BBC had investigated were Universal Training Centre (“UTC”) and Eden college. The programme was broadcast on 10 February 2014.
13. By then ETS’s licence had already been suspended by the Secretary of State. There followed a very lengthy process in which ETS was required to examine and verify the results obtained at all colleges for which it was responsible. By the end of March 2014 ETS had identified numerous cases of impersonation and proxy testing, using voice recognition software. By the time ETS had analysed 10,000 results, it was clear that a majority of the results had been obtained by fraud. These results were declared “invalid”, on the evidence that the voice identified as having provided the answers to oral tests was a voice that had provided answers to another person’s oral tests. ETS regarded the evidence of proxy test-taking or impersonation in those cases as “certain”. The verification techniques developed during the course of the process, and included both preliminary matching by voice recognition software and then human verification. At one stage the software had identified 33,000 fraudulent results: over 80% of them were then verified as positive by individual human analysis.
14. As well as in this way identifying a large number of results as “invalid”, ETS also classified certain results as “questionable”. A “questionable” result is one in which there is evidence of test administration irregularity, including the fact that the test was taken at a UK testing centre where numerous other results had been invalidated. ETS cancelled (that is to say, withdrew their certification of) the results in both “invalid” and “questionable” cases. The consequence is that the candidates in question have no current certification of their competence in English. Only, however, in cases where the test result was identified as “invalid” did the Secretary of State consider taking action against the individual on the basis of fraud.
15. DK is a national of India. His last grant of leave as a student was the result of an application supported by a certificate obtained from a TOEIC test at UTC on 17 July 2013. That certificate was subsequently cancelled as invalid by ETS and DK’s leave, previously due to expire on 5 June 2015, was curtailed so as to expire on 12 May 2014. DK made subsequent applications for leave as a student, which were refused. On 14 March 2017 he applied for leave to remain on the basis of his family life with his partner. That application was refused, partly on the following grounds:
“In fraudulently obtaining a TOEIC certificate … , you willingly participated in what was clearly an organised and serious attempt, given the complicity of the test centre itself, to defraud the SSHD and others. In doing so, you displayed a flagrant disregard for the public interest, according to which migrants are required to have a certain level of English language ability in order to facilitate social integration and cohesion, as well as to reduce the likelihood of them being a burden on the tax payer.
Accordingly, … your presence in the UK is not conducive to the public good because your conduct makes it undesirable to allow you to remain in the UK. Your application is therefore refused under paragraph S-LTR.1.6. of the Immigration Rules.”
16. DK appealed against that decision. In the First-tier Tribunal, Judge Buckwell dismissed his appeal. Permission to appeal against that decision was refused by the First-tier Tribunal, but granted by the Upper Tribunal, on the ground that it was arguable that Judge Buckwell had misapplied the burden of proof. In the Upper Tribunal, however, Deputy Upper Tribunal Judge Eshun decided that the state of the evidence before the judge was such that that error was not material. DK’s application for permission to appeal to the Court of Appeal was refused by the Upper Tribunal, but granted by Singh LJ, who remarked as follows:
“The appellant seeks permission to appeal on the basis of fresh evidence, namely the APPG report [see below] into the TOEIC/ETS issue. In my view, this raises an important point of principle or practice; and has real prospects of success. Further, and in any event, it provides a compelling reason for this court to consider this appeal.”
17. He went on to indicate that this appeal was to be listed in the Court of Appeal with RK, “which raises the same point about the APPG report”.
18. The Secretary of State conceded DK’s appeal to the Court of Appeal on the ground of error in the application of the burden of proof. The order of the Court therefore allows DK’s appeal by consent, and remits it to this Tribunal.
19. RK is also a national of India. She arrived in the United Kingdom on 23 January 2010 as a student and, on application, was granted further leave as a General Migrant and subsequently as a spouse. On 20 April 2016 she applied for further leave as a spouse. In considering her application the Secretary of State noted that in a previous application, dated 1 June 2012, she had submitted a TOEIC certificate from New London College (“NLC”) reflecting scores from a test taken there on 19 May 2012. That certificate had subsequently been cancelled as invalid by ETS. The Secretary of State expressed herself as “satisfied the course certificate was fraudulently obtained and that you used deception in your application of 1 June 2012”. As a result, the application for further leave was refused in terms similar to those used in the case of DK, set out above.
20. RK’s appeal against that decision to the First-tier Tribunal was heard by Judge Trevaskis and allowed, in a decision made on 2 June 2017. Permission to appeal was refused by the First-tier Tribunal, but granted by this Tribunal. The appeal was then remitted to the First-tier Tribunal. Judge Grant considered the matter afresh. She noted an uncertainty about the date of the test, as follows:
“26. There is an obvious error in the refusal letter which has mistyped the date of the first test as 19th May. The appellant did not sit any test on 19th May she sat the test marked as questionable by ETS on 16th May 2012. She sat the rest of the test the second test [sic] after making her application to the respondent, and this was taken on 19th June 2012 and has been marked invalid by ETS. She has confirmed in evidence that she supplied that second test result to the respondent once she had received it from New London College.”
21. Looking at the evidence as a whole, Judge Grant was satisfied that neither of the tests allegedly taken by DK was in fact taken by her. She therefore dismissed the appeal. Permission to appeal was refused by the First-tier Tribunal, but granted by the Upper Tribunal on the ground that it was arguable that Judge Grant had not clarified or resolved whether the appellant had taken two tests or one test, and which of those tests was invalidated. In this Tribunal, Judge Hanson investigated that ground at length and concluded that the judge’s decision was justified by the evidence before her, including contradictory answers by the appellant. The grounds of appeal against his decision raised the July 2019 report of the All-Party Parliamentary group on TOEIC (“the APPG report”) but, as Judge Hanson pointed out, that was not before him or before Judge Grant. The other grounds, in his view, merely amounted to disagreement with the decision. Permission was, however, granted by Singh LJ, in terms identical to those in DK.
22. The Secretary of State conceded RK’s appeal to the Court of Appeal on the ground that Judge Hanson’s reasons were “insufficient to preclude a concern” that the evidence in relation to the tests and their number had been misunderstood. The Court of Appeal’s order remits RK’s appeal for redetermination by this Tribunal.
PROCEDURE IN THE UPPER TRIBUNAL
23. The appellants’ indication of intention to rely on the APPG report made it clear that they intended to argue that the evidence being adduced by the Secretary of State in these cases was flawed and could not properly establish the use of deception by individual appellants. A case management hearing was arranged and following it, on 14 September 2020, and reflecting a measure of agreement between the parties, the Vice President directed that:
“These appeals will be listed for hearing to determine in the first instance whether in either case the appellant has a case to answer in respect of the Secretary of State’s assertion of dishonest conduct.”
24. Shortly before the case management hearing, the Tribunal had been made aware that there was a proposal that a body called Migrant Voice be allowed to intervene, because of its particular interest in the APPG report. By the time of that hearing, however, no application to intervene had been made by any party.
25. Migrant Voice did in due course apply to intervene, and there was a hearing before us at which the admissibility of the APPG report was considered. Our decision, reported as DK and RK (Parliamentary Privilege; Evidence)  UKUT 61 (IAC) was that the admission of the APPG report into evidence would violate Parliamentary Privilege. The only part of the report which we allowed to be adduced as both admissible and relevant was the verified record of what was actually said to and by three witnesses, Professors Sommer and French and Dr Harrison, at a meeting of the group on 11 June 2019. We also gave permission for Migrant Voice to make written submissions on legal and procedural issues, but not to express its opinions about the Secretary of State’s actions in TOEIC/ETS cases.
26. The hearing before us on 1 and 2 March 2021 was thus limited in scope. The Secretary of State, represented by Ms Giovannetti QC and Mr Thomann, adduced in written and oral form the evidence upon which she relies in general in TOEIC/ETS cases, and in the present cases in particular, to establish that the appellants obtained their TOEIC certificates by deception. Migrant Voice adduced the transcript of the evidence before the APPG, accompanied by written submissions by Mr Biggs. The role of the appellant’s representatives at this stage was to challenge the Secretary of State’s assertion that the evidence on which she relies was capable of discharging the burden of proof upon her. That role was performed in the case of DK by Mr Turner and Mr Gajjar and in the case of RK by Mr Ahmed and Mr Raza. Following the taking of evidence, including cross-examination on behalf of the appellants, and re-examination, the parties made further submissions in accordance with agreed directions. The Secretary of State submitted documentation relating to the prosecution of those involved in the management of ETS/TOEIC testing centres, and all parties made closing submissions in writing. A provisional date had been identified for further submissions to be made orally, but the parties agreed that no further hearing was necessary at this stage.
27. Following that hearing, we considered in detail the evidence so far adduced. We concluded, for reasons which appear in this judgment, that the evidence was ample to discharge the Secretary of State’s burden. We therefore arranged for the hearing to continue, which it did on 18, 19 and 25 November.
28. Before this part of the hearing DK indicated a wish to raise an entirely new ground of appeal, based on his family life and the birth of a child. This would have been a ‘new matter’ within the meaning of s 85(6) of the 2002 Act. By virtue of s 85(5), we could consider it only if the Secretary of State consented to our doing so (see now Hydar v SSHD  UKUT 176 (IAC);  Imm AR 1478). The Secretary of State did not give her consent. This further ground is therefore not before us and we say no more about it.
29. The appellants called oral evidence. Professor Sommer was called on behalf of DK. DK also gave oral evidence, as did his wife. RK gave evidence on her own behalf. There were then closing submissions on behalf of the Secretary of State and each of the appellants. We have taken all the evidence, oral and written, and all the submissions at each stage, into account in making this decision.
30. The Secretary of State’s responses to the classification of an individual’s test as invalid were varied, depending on the individual’s immigration status and applications that the individual made. Substantively, in each case the assertion was that the applicant had engaged in, or was engaging in, some form of dishonesty. A number of the early decisions were removal decisions under s. 10(1)(b) of the Immigration and Asylum Act 1999 on the ground that the person “uses deception in seeking (whether successfully or not) leave to remain”. The authoritative analysis of those cases is in the judgment of Underhill LJ (with whom the other members of the Court agreed) in Ahsan v SSHD  EWCA Civ 2009. The relevant statutory provisions were, however, with many others, repealed or replaced by or under the Immigration Act 2014.
31. In some other cases, leave has been refused under paragraph 322 of the Statement of Changes in Immigration Rules, HC 395 (as amended), which provides, so far as relevant, as follows:
“Refusal of leave to remain, variation of leave to enter or remain or curtailment of leave
322. In addition to the grounds for refusal of extension of stay set out in parts 228 of these rules, the following provisions apply in relation to the refusal of an application for leave to remain, variation of leave to enter or remain or, where appropriate, the curtailment of leave….
Grounds on which leave to remain and variation of leave to enter or remain in the United Kingdom are to be refused
(1A) Where false representations have been made or false documents or information have been submitted (whether or not material to the application, and whether or not to the applicant’s knowledge), or material facts have not been disclosed, in relation to the application or in order to obtain documents from the Secretary of State or a third party required in support of the application;
Grounds on which leave to remain and variation of leave to enter or remain in the United Kingdom should normally be refused
(2) The making of false representations or the failure to disclose any material fact for the purpose of obtaining leave to enter or a previous variation of leave or in order to obtain documents from the Secretary of State or a third party required in support of the application for leave to enter or a previous variation of leave;
32. The decision of the Court of Appeal in Adedoyin v SSHD  EWCA Civ 773;  Imm AR 704 provides the starting point for the analysis of such a case. That decision makes it clear that the reference to “false” in paragraph 322 means dishonestly false.
33. By virtue of paragraph A320 of the Immigration Rules, however, paragraph 322 does not apply to an application for entry clearance, leave to enter or leave to remain as a family member under Appendix FM, and it was applications under Appendix FM which were made by each of the appellants before us. In such cases, the refusal was normally on the ground of “suitability”. The relevant provisions in Appendix FM are as follows:
“Section S–LTR: suitability – leave to remain
S–LTR.1.1. The applicant will be refused limited leave to remain on grounds of suitability if any of paragraphs S – LTR.1.2.2, 1.8 apply
S–LTR.1.6. The presence of the applicant in the UK is not conducive to the public good because their conduct (including convictions which do not fall in paragraphs S-LTR.22.214.171.124.5.), character, associations or other reasons, make it undesirable to allow them to remain in the UK.”
There are similar provisions in relation to indefinite leave to remain in paragraphs S-ILR.1.1. and S-ILR.1.8.
34. The suitability requirements are phrased in more general terms than paragraph 322. They do not refer directly to false representation, and they do not specifically require evidence or proof of dishonesty. The relationship to the suitability requirements and the authorities on dishonesty was examined by this Tribunal in Muhandiramge v SSHD  UKUT 675 (IAC), to which we refer in more detail below.
35. In Shen v SSHD  UKUT 236 (IAC);  Imm AR 971, this Tribunal (Green J and UTJ Goldstein) decided that where an application is refused on grounds of dishonesty the Secretary of State bears the burden of proof of establishing the dishonesty. The burden is discharged in two stages. The documentation may show prima facie evidence of dishonesty, but prima facie evidence will not of itself establish the Secretary of State’s case if there is exculpatory evidence from the appellant. In such circumstances all the evidence must be taken into account in order to determine whether the Secretary of State has proved her case.
36. This is set out at paragraph 25 of Shen (where the false statement related to previous convictions) as follows:
“25. On analysis we believe that the way in which the burden of proof operates is as follows. We accept that if an application form is false in a material way, that this may be relied upon as some prima facie evidence which assists in establishing dishonesty. The inference of deliberate deception can be strengthened by reference to other facts, for example if the conviction is shortly prior in time to the completion of the application form this will furnish circumstantial supporting evidence that the conviction must have been high in the applicant’s mind and any explanation based upon oversight would carry little weight. However, this is not dispositive of dishonesty and it is open to an Appellant to proffer an innocent explanation. If an innocent explanation is advanced (by which we mean one that meets a basic, minimum level of plausibility) then the burden switches back to the SSHD to answer that evidence. At the end of the day the SSHD bears the burden of proof. This is a proposition which is uncontroversial and has been confirmed on many occasions: e.g. JC (Part 9 HC395 – burden of proof) China  UKAIT 00027 para 10; MZ (Pakistan) v Secretary of State for the Home Department  EWCA Civ 919 para 25; Mumu (paragraph 320; Article 8; scope)  UKUT 00143 (IAC).”
37. Muhandiramge, a decision of this Tribunal (McCloskey J, President and UTJ Bruce) presents difficulties at a number of levels in the context of this appeal. It has some similarities, in that the decision under appeal was a refusal of an application for further leave to remain as a family member under Appendix FM, and the ground for refusal was that of suitability. However, the date of decision was 12 September 2014, and the appellant had available to him the rights of appeal as they were under the Nationality, Immigration and Asylum Act 2002 before the changes made by the Immigration Act 2014. The primary concern was whether the decision was one that was made in accordance with the Immigration Rules.
38. The suitability requirement in question was also different from that in the present case. The relevant provision was S-LTR.1.7. - “failure without reasonable excuse to comply with a requirement to provide information”. The appellant had failed to disclose criminal convictions in response to a direct question about them on the application form. The primary question for consideration by the Tribunal was whether the burden of proof was on the appellant to show that he had a reasonable excuse for his failure, or whether it was on the respondent to establish the absence of reasonable excuse, or perhaps the absence of reasonableness in any excuse for his failure.
39. As in the present case, there is no specific reference to falseness or dishonesty in the rule under consideration. Also as in the present case, paragraph 322 was not in issue, because of the effect of paragraph A320 (not the version we have set out above, but its predecessor). Nevertheless, the Tribunal examined a number of other issues, beginning with issues of dishonesty and the authorities on paragraph 322. At paragraph , the Tribunal said this:
“Burdens and standard of proof have progressively, and almost with stealth, become an established feature of decision making in the field of immigration and asylum, law. Their emergence may properly be described as organic. They have featured particularly in cases where it is alleged by the Secretary of State that the applicant has engaged in deception or dishonesty with the result that the application in question should be refused. This discrete line of authority is not recent, being traceable to the decision of the Immigration Appeal Tribunal in Olufosoye  Imm AR 141. In Tribunal jurisprudence, the origins of this particular lineage can be traced to the decision of the House of Lords in R v Secretary of State for the Home Department, ex parte Khawaja  AC 74, which concerned the inter-related issues of procuring entry into the United Kingdom by deception and precedent fact in the Secretary of State’s ensuing decision making process. It is well established that in such cases the burden of proof rests on the Secretary of State and the standard of proof belongs to the higher end of the balance of probability spectrum.”
40. The first two sentences of that paragraph ought perhaps to be read in the light of rule 31 of the Immigration Appeals (Procedure) Rules 1972 and its successors, which place the burden of proof on the appellant in proceedings before an adjudicator or the Tribunal. The last sentence no doubt needs to be read in the light of the speech of Lady Hale in the House of Lords in Re B (Children)  UKHL 35, to which we refer below. Paragraph  of the decision in Muhandiramge sets out the Tribunal’s views on a shifting burden of proof in cases of dishonesty. This paragraph forms the target of detailed submissions by Mr Biggs, which we address below.
41. There follows a discussion on general matters relating to the burden and standard of proof in a variety of different types of case. The conclusion, at the end of paragraph , is the perhaps unsurprising one that “in pithy terms, he who asserts must prove”. At paragraph  the Tribunal continues as follows:
“24. Having regard to the terms of the grant of permission to appeal, three conclusions are, in principle, open to this Tribunal. The first is that the burden of proving that the appellant had a reasonable excuse for non-disclosure of his criminal conviction rested on him. The second possible conclusion is that the Secretary of State had the burden of proving that the appellant did not have a reasonable excuse for the non-disclosure. The third partakes of the “boomerang” discussed in  above.
25. If the second of these conclusions is to prevail, it would operate as an exception to the general rule of evidence discussed immediately above. We are unable to identify any basis of principle, logic, or otherwise favouring this conclusion. It finds no support in any authority brought to our attention. Furthermore, it would be in conflict with the principle that where the truth of a party’s allegation lies peculiarly within the knowledge of his opponent, the latter bears the burden of disproving it [citing authorities].
26. In contrast, adoption of the first of the two [sic] possible conclusions rooted above is indicated by the general rule, duly bolstered by considerations of fairness, logic, and common sense, all long-standing traits of the common law. This conclusion would require the appellant to bear the burden of proving a reasonable excuse for the omission in question and to do so to the civil standard, viz on the balance of probabilities. We readily espouse this conclusion.
27. Furthermore, there is no principled reason favouring the third possible conclusion, namely that the appellant had a (mere) evidential burden, which, if discharged, subjected the Secretary of State to the legal burden of disproving the reasonable excuse canvassed. Given that the appellant is the party in possession of the material information, belonging solely to his knowledge there is no basis in principle or otherwise for this third conclusion.”
42. For those reasons, the Tribunal held that the appellant had the burden of proving that he had a reasonable excuse.
43. As Muhandiramge was not a case of dishonesty, the Tribunal’s remarks on the proof of dishonesty are obiter. It is not clear whether the Tribunal appreciated that paragraph 322 did not apply to the case it was considering: although it cites the authorities on that paragraph, it does not mention paragraph A320. In the end, however, it clearly parts from the conclusions reached in the dishonesty cases, and takes the opposite view as to the incidence of burden of proof in a case where a matter depends on the existence or otherwise of a reasonable excuse for a failure. It is, we have to say, not entirely clear why the existence or otherwise of a reasonable excuse is to be regarded as “belonging solely to [the appellant’s] knowledge”, apparently in distinction to the existence or otherwise of dishonesty by the appellant. Be that as it as it may, Muhandiramge stands as the authority on the construction of paragraph S-LTR.1.7.
44. So far as concerns paragraph S-LTR.1.6., however, the position taken by the parties before us was based, as we understand it, on the following propositions. First, the appellant’s non-suitability arises from the demonstration that he obtained TOEIC result from ETS dishonestly. Secondly, it is for the Secretary of State to prove the dishonesty. Thirdly, if the Secretary of State fails to prove the dishonesty, the appellant’s appeal ought to be allowed on human rights grounds, because the non-suitability as identified in the decision under appeal was the sole reason for the interference with the appellant’s family life by the refusal of the application.
LEGAL AND EVIDENTIAL BURDENS
45. Paragraph 10 of Muhandiramge is as follows:
“One of the more recent reported decisions belonging to this stable is that of Shen. This decision is illustrative of the moderately complex exercise required of tribunals from time to time. Here the Upper Tribunal held, in harmony with established principle, that in certain contexts the evidential pendulum swings three times and in three different directions:
(a) First, where the Secretary of State alleges that an applicant has practised dishonesty or deception in an application for leave to remain, there is an evidential burden on the Secretary of State. This requires that sufficient evidence be adduced to raise an issue as to the existence or non-existence of a fact in issue: for example, by producing the completed application which is prima facie deceitful in some material fashion.
(b) The spotlight thereby switches to the applicant. If he discharges the burden – again, an evidential one – of raising an innocent explanation, namely an account which satisfies the minimum level of plausibility, a further transfer of the burden of proof occurs.
(c) Where (b) is satisfied, the burden rests on the Secretary of State to establish, on the balance of probabilities, that the appellant’s prima facie innocent explanation is to be rejected.
A veritable burden of proof boomerang!”
46. We are far from confident that the relevant passage in Shen, set out above, fully justifies this excursion into the varied metaphors of pendulum, spotlight, and boomerang.
47. The description in paragraph 10 in Muhandiramge may be a fair indication of the decision-making process of some judges, tossing an idea backwards and forwards and testing it against the evidence on each side. We do not venture to say that such a process is wrong. As a statement of either the procedure at an appeal or the law, however, it is wrong. There is no sense in which procedurally a case passes backwards and forwards between the parties, giving either of them new chances or even tactical obligations to meet the evidence so far adduced by their opponent: on the contrary, each side has one opportunity only to produce all the evidence it considers relevant to the case. Further, the burden of proof does not shift from one side to the other during the course of a trial. The burden of proof is fixed by law according to the issue under examination. If it were not so, parties would not know in advance what evidence would or might be necessary to establish their cases.
48. If we may venture to introduce yet another metaphor (but only in order to dismiss it), it is misleading to think of the trial as a tennis match and the evidential burden as a ball:
“There are no points to be gained merely by sending the evidential burden back across the net, and what is more, no one is keeping score…. The shifting of the evidential burden during the trial is therefore … only part of a process of reasoning, sometimes convenient and sometimes dangerous, whereby the judge assesses the probabilities by dividing up the evidence into those facts which tend to support one party and those which his opponent has proved in reply.”
That is how Lord Hoffmann put it, writing extrajudicially in The South African Law of Evidence (second edition, 1970).
49. We must consider the burdens in the present case. In an immigration appeal, the burden of proof is placed by law on the appellant, save in respect of a small number of issues where it is placed on the Secretary of State. Dishonesty by the appellant is one such issue. It is not for the appellant to disprove it but for the Secretary of State to prove it.
50. Difficulties arise because the phrase “the evidential burden” appears to be used in two different senses. Where it is used of a burden on a party who does not have the (legal) burden of proof, it means that a matter that might otherwise come into consideration in discharging that burden does not fall for consideration at all unless the party with the evidential burden adduces sufficient evidence to raise the matter. To take an example from the criminal law (it is not easy to identify examples in the field with which this Tribunal usually deals) provocation as a defence to a charge of murder has to be disproved by the Crown in order to secure a conviction: but no disproof of it is necessary unless there is sufficient evidence to raise the defence as an issue: Mancini v DPP  AC 1.
51. When, however, an evidential burden is said to lie on the party that has the (legal) burden of proof on an issue, it cannot be a matter of making the matter an issue: ex hypothesei it already is an issue.
52. What is identified here is a test of whether the party with the burden of proof has adduced sufficient evidence to enable a finding of fact in that party’s favour. To put that another way, might the trier of fact find the matter proved on the basis of the evidence if that evidence were uncontroverted? Looking again at a criminal trial, this is the test applied at “half time”, when the prosecution case is closed. If at that point the evidence adduced is insufficient to found a conviction, the defendant is entitled to a verdict of not guilty and, more relevantly for the purpose of the present proceedings, is not put to the trouble of having to put a case to counter the accusation.
53. It is, as the Privy Council pointed out in Jayasena v R  AC 618 624, better not to call the evidential burden (in either sense) a burden of proof, because an evidential burden can be discharged by evidence falling far short of proof. Evidential burdens have, however, this in common with the burden of proof: they do not really shift, because their placing can be ascertained in advance of the proceedings. The existence of an evidential burden in the first sense is a matter of law; an evidential burden in the second sense exists as it were in the background of every legal burden.
54. To this extent we agree largely with the submissions made by Mr Biggs and supported by copious authority. The points are relatively basic, but a rehearsal of them was no doubt necessary given this Tribunal’s decision in Muhandiramge.
THE STANDARD OF PROOF
55. We do not, however, accept Mr Biggs’ further submission, for which he cited no authority, that in order to discharge her burden of proof, the Secretary of State would need to offer “cogent” evidence.
56. The standard of proof in relation to an assertion of dishonesty in proceedings such as these is the balance of probabilities. If authority for that proposition is needed, it is to be found in the judgment of Beatson LJ in SSHD v Shehzad  EWCA Civ 615 at  where, it is to be noted, the learned Lord Justice puts inverted commas around the word “shifts” every time he uses it. He goes on to cite the words of Baroness Hale in Re B (Children)  UKHL 35 at :
“Neither the seriousness of the allegation nor the seriousness of the consequences should make any difference to the standard of proof to be applied in determining the facts. The inherent probabilities are simply something to be taken into account, where relevant, in deciding where truth lies.”
57. There is no principled basis upon which it can be prescribed that evidence of any particular nature or quality is required in order to enable the trier of fact lawfully to reach a decision that the burden of proof on the balance of probabilities has been discharged. The trier of fact will take whatever precautions are appropriate in the circumstances, to avoid making a decision that is not justified on the evidence as the final juridical disposal of the dispute on the facts. Finding facts is an individual mental process, not a mechanical one, and it may be that psychologically the trier of fact will respond to an imperative to take a more probing approach to evidence tending to prove a matter where the consequences are likely to be more severe. But the test fixed by the law is the same in every case. That test is whether, when taking the evidence as a whole, the trier of fact is satisfied on the balance of probabilities that the proposition of fact advanced by the party with the burden of proof is made out.
58. The evidence as a whole may consist of elements tending to support the proposition with the greatest of certainty, elements tending to support the proposition with much less certainty, elements that are neutral, and elements tending to undermine the proposition. It is their affect as a whole that counts. There is no requirement that any single element have any particular quality other than admissibility, in order to be allowed to contribute to the whole; and there is no requirement that the evidence as a whole have any degree of cogency except such as actually causes the trier of fact to be satisfied as set out above. If the evidence as a whole has that effect the fact is found; if it does not have this effect it is not found; and in neither case is anything contributed to the discourse by separate evaluation of the persuasive force of each item of evidence.
59. If the matter is examined at the time where only the party with the burden of proof has adduced evidence, so that the whole of the potential evidence is not yet available for consideration, the principle is no different, although the question is. The procedure in a criminal trial, which may be stopped after the prosecution case as indicated at  above, is designed to provide safeguards against wrongful conviction. Even if similar issues are before a civil court, the rules of criminal procedure do not apply. Following the directions given, however, we did test the Secretary of State’s evidence in a manner similar to that applicable at the close of the prosecution case in a criminal trial, in order to determine whether her evidence, if not contradicted, would justify a finding that the appellants (or others in similar circumstances) were not suitable for a grant of leave. Thus, our decision in this case is intended to determine whether it is right to require the appellants, and others similarly circumstanced, to put their case, or whether they are entitled to a finding in their favour simply on the ground that the Secretary of State’s evidence is insufficient to prove their dishonesty, before going on to decide the appeals before us on the evidence.
60. We therefore ask first whether the Secretary of State’s evidence would enable a properly-instructed trier of fact to determine that the burden of proof had been discharged on the balance of probabilities. If the evidence at this point would not support a finding that the matter was proved on the balance of probabilities, the appellants would be entitled to succeed in their appeals. If, however, it would support such a finding, the evidence as a whole falls for consideration in order to decide whether the appeals succeed or fail. With that in mind, we turn to the evidence before us.
THE GENERAL EVIDENCE: WIDESPREAD FRAUD
61. It is the Secretary of State’s case that the evidence shows that there was widespread fraud and cheating in ETS centres. We declined Ms Giovanetti’s invitation to watch the original Panorama programme, outside the ambit of the hearing, preferring instead to encounter evidence wholly in the open arena of the courtroom. The general evidence includes the following.
62. First is the material exposed by that programme on 10 February 2014. A summary from the BBC website was in the bundle of evidence. UTC in Watford was one of the centres featured in the programme. Undercover investigators described the ways in which the Centre’s staff arranged for the students’ tests to be taken by proxy or for the invigilator to read out the correct answers for students to enter in their submissions. Following criminal trials, to which we shall shortly refer, the BBC website described some of the material as follows:
“Undercover footage in which entire rooms of registered candidates stepped aside so that exams could be taken by people who spoke better English was shown at the trials. The plotters were equally brazen when it came to a multiple-choice paper – the secret filming showed an invigilator simply reading out the answers for candidates to copy.
The court heard that “proxy sitters” were paid £50 for each exam they took. Clients were promised a “guaranteed pass” for £500, about three times the official exam fee. The footage from Universal Training Centre showed candidates waiting in a separate room while their tests were faked.”
63. To the BBC’s own material may be added the evidence given at the criminal trials of some involved in the ETS frauds.
64. R v Mohammed and others (T2014 7524) was the trial at Southwark Crown Court of three of those involved in the fraudulent activities at UTC. All three were convicted of sample charges of facilitating, and together with others conspiring to facilitate, breaches of immigration law. In passing sentence on 17 May 2016, the trial judge made reference to a number of features of the evidence that the jury must have accepted as proving the alleged facts beyond reasonable doubt. Those remarks are in evidence before us. They include the following:
“The second undercover, Mr Malik, made it clear that his primary interest was to work and he too was offered and supplied with a false backup letter, a CAS and forged bank statements showing that he had over £10,000 in his account. He was brazenly told by Mr Mohammad and Ms Noreen that someone would take the English exam on his behalf and he attended UTC with an employee of Bright Consultants Services (BCS), Mr Shaikh, who did just that. The footage of that event made for compelling viewing. Mr Malik told the court that every student at the test centre had a pilot sitter and the footage showed Mr Hafeez speaking to the students in a waiting area and instructing them to return to their computer terminals in the event of a raid by the awarding body. Thereafter the students were left in the waiting area whilst the test was completed on their behalf. The footage from that day showed what appeared to be a very well-practiced operation in which everyone at the centre, students, pilots and staff seemed to know what was going on and what was expected of them.
The evidence of fraud from the undercover filming at BCS and UTC was overwhelming but it was not limited to the two undercover students nor could that ever sensibly be said to be the case. It was evident from the undercovers’ interactions with both BCS and UTC that they were well-practised in this fraud and had experience of providing forged documents and arranging fake tests.
[A] spreadsheet recovered from the computer at UTC provided yet further evidence of the extent and scale of this fraud. It showed a number of entries where the student taking the exam had another name shown in brackets. Those bracketed names reoccurred on a frequent basis throughout the document and there was evidence of payment of those named persons. In short, there was good evidence that they were the pilot sitters, referred to by Mr Malik and Mr Shaikh in their evidence. The spreadsheet was further peppered with references to “pilot”, “by their pilot” and “pilot fees”. The spreadsheet in my judgment represented the clearest evidence that UTC, under Mr Hafeez’s direction, was routinely running fake tests.
I reject the submission made on behalf of Mr Hafeez that this was a relatively small-scale conspiracy confined within the walls of UTC. There was, in my judgment, the clearest of evidence from which it could be inferred that he was conspiring with others including staff at BCS. Mr Shaikh had been sent there on three occasions by Mr Mohammad and he described it as “a safe centre for us” where there was unlikely to be a raid. He knew Mr Hafeez as the man in charge and it would appear from the undercover footage that he was known also to Mr Hafeez’s staff at the centre, who quite frankly had to have known what was going on and been in on it.
In short, this was a well organised, well-practised and highly lucrative fraud of the immigration visa system committed over a sustained period of time.
The court heard evidence that of the 795 TOEIC tests taken at UTC during the period of the indictment, 145 had a separate bracketed name (highly suggestive of a pilot sitter) and that those test results were used by the students to make an application to the Home Office to extend their visa. In other words, there is good evidence that this fraud was being followed through to the very end in order to extend student visas. For all of those reasons I access the level of harm caused by this offending as high.”
65. It is clear also that judge accepted the defendants’ arguments that the evidence going to specific fraud was limited to the tests, and that there was no fraud committed against the students themselves:
“I accept that it is not possible to correlate the fraudulent activities with specific visa applications (save for the data concerning the TOEIC tests). Lastly I accept that there was no exploitation of the students using the services at BCS and UTC as they must plainly have been participating in the fraud themselves, given that they were unable to legitimately satisfy the visa requirements.”
66. R v Kologatla and others (Isleworth T0157639; 7640; 7664) was the trial at Isleworth Crown Court of some of those involved in the fraudulent activities at NLC. Ten individuals were charged with conspiracy together and with others to facilitate breaches of immigration law, the particulars of offence specifically alleging that the facilitation was by arranging for “proxy” candidates to take the TOEIC tests in place of the persons ostensibly sitting the examination. There were some pleas of guilty and the indictment was severed, but on 30 September 2016 HHJ Edmunds QC sentenced six of those on the indictment, including three who had been convicted after a trial. He too made reference in his sentencing remarks to features of the evidence that the jury must have accepted as proving the alleged facts beyond reasonable doubt. Again those remarks are in evidence before us, and include the following:
“6.6. Although in the beginning, March and April 2012 Mr Kologatla for NLC made a number of representations to ETS that the college was under pressure to accept low quality identification from candidates or to allow substitutes to take tests on behalf of the real candidates it is clear that he, and those working under him at the college, rapidly decided to subvert the whole testing process by arranging and allowing substitutes to take the tests in collaboration with agents who presented candidates.
7.1. It was apparent that a schedule found on a hard disk in the possession of Kologatla was a comprehensive list of candidates for the TOEIC test. It matched with the candidate sign in sheets for individual tests and other records. The schedule shows recurrent names of “Reps” including on many occasions individual reps for each candidate. They were called “Reps” short for replacement because that is exactly what they were – substitutes to take the tests for the real candidates. As Kologatla was forced to accept in evidence the schedule tracked the progress of the candidates and I am quite satisfied that it also tracked the provision of substitutes and therefore shows the scale of the cheating.
8.1.1. There was also “snapshot” evidence of misconduct on particular days. In particular, there was the evidence of a candidate Bernita Basnet (from Nepal) who attended to take the test on 14 May 2013 having paid her solicitor a substantial fee. Ironically she spoke good English and did not require help. She described how, at the commencement of a test for some 16 candidates which was, according to the records invigilated by Basham [one of the defendants], there was a general announcement that if persons from UKBA came “you will hear a bell and must swap seats with the lady taking the test and pretend to do something”. Ms Basnet then found that a lady sat next to her and, despite her protests, insisted on doing the test for her. This was institutional organised cheating.
9.1. On 15 May 2013 there was an unannounced inspection from ETS which revealed much the same picture in the test room being invigilated by Basham.
10.1. On 31 May 2013 ETS wrote to Kologatla at New London College terminating the contract on the basis of what had been found on 15 May. The letter made clear the allegation that substitutes had been found taking tests. Kologatla replied on 10 June and … clearly acknowledged that there had been widespread cheating and it is implicit that Kologatla is aware of it.
10.2. The thrust of the letter is to complain that cheating was widespread as other test centres and Kologatla’s attempts to explain that letter away in evidence were clearly rejected by the jury.”
THE IMPORT OF THE GENERAL EVIDENCE
67. The evidence showing fraudulent activity in a number of ETS centres (including UTC and NLC) is overwhelming. It is clear beyond a doubt that these were institutions for the manufacture of fraudulent qualifications. This conclusion does not show that any individual certificate was obtained fraudulently. But it has an important part in the evaluation of the evidence as a whole, in that it provides the context.
68. Without that context, an allegation that certificates were fraudulently obtained with the assistance of the managers of colleges franchised by a respected international group, inspected and authorised by the Secretary of State for the Home Department would be met with scepticism. That is because, without that context, any decision-maker would approach the allegation from a common sense starting-point based on those facts. Such a person would, as the saying is, need “a lot of persuading” that the allegation of fraud was made out. When, however, such a fact-finder is shown how, contrary to any preconceptions, the truth is that fraud was frequent and widespread, the individual allegation becomes more plausible.
69. An individual allegation always has to be assessed in the context of the whole of the background evidence. The more plausible the alleged fact is, the more the allegation is likely to be accepted on the basis of such individual evidence as is available. That individual evidence will have different effect according to the background against which it is assessed.
70. That is not to say that the need for individual evidence is reduced. The individual case can never be proved by evidence of generality, unless the general is universal. But the general evidence changes the starting point. The possible response to the assertion of fact moves from disbelief that such a thing could ever – or could regularly - happen, to a specific enquiry about whether one of the events that certainly did happen was associated with the individual under investigation. This feature of the interaction of general and individual evidence is a matter of common experience and is not unrelated to the discussion of cogency.
71. We have rejected Mr Biggs’ submission that, in a case such as this, the Secretary of State’s evidence would have to have a particular quality of “cogency”. The position is, however, that the more unlikely an assertion of fact is, the more persuasive the evidence of it will need to be in order to perform its function. But the evidence in question is the evidence taken as a whole. Where the function of the general evidence is to reduce the unlikelihood of the asserted fact, the individual evidence has a less difficult task to perform.
72. On the other hand, where the general evidence sets up a particular set of facts as likely, individual evidence will have a more difficult task in showing even the possibility of the opposite. As Lord Hoffmann said in re B at :
“If a child alleges sexual abuse by a parent, it is common sense to start with the assumption that most parents do not abuse their children. But this assumption may be swiftly dispelled by other compelling evidence of the relationship between parent and child or parent and other children. It would be absurd to suggest that the tribunal must in all cases assume that serious conduct is unlikely to have occurred. In many cases, the other evidence will show that it was all too likely. If, for example, it is clear that a child was assaulted by one or other of two people, it would make no sense to start one's reasoning by saying that assaulting children is a serious matter and therefore neither of them is likely to have done so. The fact is that one of them did and the question for the tribunal is simply whether it is more probable that one rather than the other was the perpetrator.”
73. In other words, proof on the balance of probabilities does not assess either evidence or probability in a vacuum, but takes account of the given context. That principle applies to the evaluation of the credibility (or general reliability) of evidence, as well as to the overall task of determining facts. As Lord Pearce said dissenting in Onassis and Calogeropoulos v Vergottis  2 Lloyd's Rep 403, HL:
“”Credibility” involves wider problems than mere “demeanour” which is mostly concerned with whether the witness appears to be telling the truth as he now believes it to be. Credibility covers the following problems. First, is the witness a truthful or untruthful person? Secondly, is he, though a truthful person, telling something less than the truth on this issue, or, though an untruthful person, telling the truth on this issue? … And lastly, although the honest witness believes he heard or saw this or that, it is so improbable that it is on balance more likely that he was mistaken? On this point it is essential that the balance of probability is put correctly into the scales in weighing the credibility of a witness. And motive is one aspect of probability. All these problems compendiously are entailed when a Judge assesses the credibility of a witness; they are all part of one judicial process. And in the process contemporary documents and admitted or incontrovertible facts and probabilities must play their proper part.”
74. In appropriate circumstances it may be right to reach a judgment on the facts simply on the basis of the inherent likelihood or unlikelihood of what is asserted. An example of this occurs in the judgment of Stewart J in Kimathi v The Foreign and Commonwealth Office  EWHC 2066 at :
“Were the severed heads incident a core allegation, it would be subject to all the problems well rehearsed earlier in this judgment whereby, after the great lapse of time, it is impossible for the Defendant to begin to investigate. Even without such evidence, I am troubled as to the reliability of TC34's account. It is not the inconsistency in detail that particularly concerns me, but rather the sheer unlikelihood, in the absence of any proper corroboration from any source whatsoever, that this sort of incident ever occurred.”
75. Thus any assessment of whether the burden of proof is discharged in an individual case falls to be determined against the background of the fact that there were many thousands of results obtained fraudulently. The assertion that an individual appellant cheated is not an unqualified assertion to be viewed against the background of general unlikelihood, but is a particular assertion that the appellant was one of the large number who certainly did cheat.
INDIVIDUAL EVIDENCE: LINKING TESTS TO CANDIDATES
76. The individual evidence in each case in which it is said that a certificate was obtained by fraud comprises two elements. The first is the identification of the test result as one in relation to which there is evidence of fraud by the test entry being by a person who had been represented as being the candidate in other tests. This is a matter of voice recognition. The second is the identification of a particular individual as the person whose test entry the result reflects. This is a matter of the integrity of ETS’s process for attributing entries to candidates.
77. We use the word “element” rather than any other in order to avoid raising any suggestion that a conclusion on the evidence as a whole depends on the absolute reliability of any feature of the evidence taken in isolation. This is not a chain, only as strong as the weakest link. The evidence, we repeat, needs to be considered as a whole. Clearly if one element is shown to be unreliable, the whole argument may collapse. But that is different from assessing a case on the basis of the level of positive evidence supporting a proposition. Merely suggesting, or even showing, that one part or another of the evidence is not demonstrably unassailable does not begin to show that an individual conclusion on all the evidence taken as a whole would be unsafe or unmerited.
78. The analysis was undertaken within ETS: the organisation already had in place a number of techniques designed to detect fraud, including fraud perpetrated by substitution of test sitters, but (as has been pointed out by Professor Sommer) not particularly focussed on the possibility of deception by, or encouraged by, the testing centre.
79. On voice recognition, what the evidence amounts to is as follows. The recordings on which the test certificates were based were subject to voice recognition analysis. The analysis was undertaken in the first instance by an automated comparison, in preparation for which the original recordings were recoded from .ogg to .wav, and then from .wav to .spx. The purpose was to identify instances of the same voice appearing in more than one test entry. Where a match was identified, the data was subject to further analysis by human listeners. A conservative approach to declarations of invalidity was adopted, in that it was only if all the analysis pointed to the appearance of a voice attributed to more than one candidate that the score was marked “invalid”.
80. Nevertheless, there is a possibility of erroneously reaching that conclusion. The opinion of Dr Harrison, an expert in audio and speech analysis, was that the initial triage by the automatic process, and the subsequent use of human checkers, was the only reasonable way in which the process could be carried out, given the quantity of recordings involved. He looked carefully at the process and estimated that there might be a “false positive” rate of 1%. That is, in that proportion of cases the techniques used might provide an incorrect basis for a marking of ‘invalid’: this would be the error rate, not of the entire process, but of this final part of it. He thought that there would be substantially more false rejections. The figure of 1% was produced by comparison with a pilot exercise undertaken by ETS as part of a training and development process: even if the results of that exercise were not comparable, he would still estimate the number of false positives emanating from the ETS assessment process as “very small”. Dr French thought the error rate of positives might be higher, perhaps 3%, but (as we understand his evidence) that figure was based on the absence of information he would have liked to see rather than strictly on the information that was available.
81. We heard oral evidence from Mr Sewell, an Intelligence Analyst in the Immigration Intelligence Directorate of the Home Office, whose written evidence, with associated data, supplements the Secretary of State’s case as presented in some previous appeals. He was cross-examined by Mr Turner and Mr Ahmed, but did not add very much to his confirmation of the processes and figures given in the written evidence. He confirmed that the primary investigation was of tests, not of candidates, so that the number of results retrospectively invalidated or questionable did not equate to the number of candidates (because each candidate might have taken a number of tests, under different conditions). He also confirmed, however, that 100% of the tests taken at NLC, and overall 64% of tests taken at ETS-licenced centres had had their result cancelled. Mr Turner concentrated on discovering the extent to which Mr Sewell had access to the data transferred by the test centres to ETS. The response was that the material Mr Sewell dealt with had to be that derived from ETS. There was, he said, no reason to indicate that ETS were complicit in any fraud, although they might have been negligent in supervision. Nor did the data with which he was working show any signs of corruption.
82. The one point on which we found Mr Sewell’s evidence less than persuasive was that relating to the correlation of ETS TOEIC success rates with the success rates of other candidates undertaking other tests at other test centres: this was an issue to which Mr Ahmed’s cross-examination was particularly directed. As we said above, a factor causing the initial enquiries into the ETS centres was what was regarded as a suspiciously large number of high marks. That view was supported by Mr Sewell by comparison with the results of candidates not taking the TOEIC test but the Pearson test with other providers. Despite Mr Sewell’s defence of this process, however, it seems to us that it is not shown to be reliable. First, the correlation is not exact, and in particular the two marking schemes have different gradations at the extremes. Secondly, there may have been a difference in the capabilities of candidates undertaking the different tests, or registering at the different centres, which would mean that the populations were not comparable anyway. This point, however, is of no importance. The Secretary of State’s case is not now based on the distribution of scores. That was a reason for investigation, but once the investigation has taken place it is the results of the investigation, not the reasons for it, that count.
83. Perhaps more to the point, the comparison of the tests would appear to fall outside Mr Sewell’s specific expertise as an analyst, although his processing of the data thereby generated, on which we detect no fault, does not. Similarly, Mr Sewell’s views on language learning, although not shown to be wrong, were based, as he put it, on general knowledge rather than on his expertise. On the relevant matters within his expertise we regard Mr Sewell’s evidence as wholly reliable. He had a clear wide-ranging view of the way the relevant data were collected and processed, of the way in which the test entries were automatically given identification numbers, and of the interpretation of both the numbers and the assessment of the results. He also had the expertise to be able to detect if, amongst the very large number of test results and assessments produced by ETS in the investigation, there was evidence of difficulties or errors in association or transmission. When he did not know the answer to a question, as was the case with questions about ETS’s internal procedure and their employment of an agency (YMB) to assist with its assessment, he said so.
84. The process of checking the linking of results to candidates has been described as using a spreadsheet – the ‘lookup tool’. That system was devised by Mr Green, who gave oral evidence. He did not claim to know very much about the contents of the data, and accepted that he had not traced any material through the system from test entry to spreadsheet entry. His clear evidence, however, was that the lookup tool is reliable, and that no erroneous entry has been identified. Mr Turner cross-examined, and did not make any progress against that assertion. Mr Sewell explained in his evidence how the unique number used throughout the investigation process is generated by the candidate logging in for the test.
85. The important thing is not the way the information is tabulated, but its source. We have before us samples of data entries and evidence of how the identification is generated. First, a candidate’s name and other details are registered with ETS by the test centre. Then, when the candidate attends for the test and enters the personal information demanded, a unique registration number is generated by the system automatically: this number identifies not merely the candidate but also the session and location and the part of the test being undertaken. The accuracy of the number can be checked by seeing that it falls within the range of consecutive numbers allocated automatically to the tests in the session in question. As Ms Giovanetti put it in written submissions, a candidate generates his or her own unique ID/test number, which should appear on the audio files, the test certificates, and all other documentation relating to the test. She also observes that there is no evidence of any error in this regard in relation to either of the appellants. Professor Sommer has made extensive suggestions about what else could have been done to identify test entries with candidates, but those suggestions have to be understood in this context.
86. Professor Sommer’s oral evidence explored and emphasised, but did not add very much to what he had said previously in writing. He again went through the possibilities of error in the “chain of custody”, but accepted that these were merely his suggestions about what could conceivably go wrong, rather than evidence that anything had gone wrong. He did make two points which we regard as of some significance. The first is that he was quite clear that candidates would not remain in ignorance that tests were being taken on their behalf: if they were not displaced from their seat, they would in any event see the cursor on the screen moving other than in accordance with their own entries. Secondly, Professor Sommer’s considered conclusion was that it was very unlikely that there were accidental errors in the production or transmission of results. If there were errors, they were very probably deliberate. He hypothesised that a test centre might want to improve its own record of results by substituting false entries for those actually put in by the candidate.
THE APPG MATERIAL
87. The appellants make it clear that it is their submission that the APPG material changes the position from previous cases where the Secretary of State’s evidence was assessed judicially as sufficient to prove the Secretary of State’s case. As we indicated in our earlier judgment,  UKUT 61 at , what was said to the APPG by those who have given evidence in those appeals (and other similar proceedings) is admissible before us and may be of relevance.
88. There are, however, considerable difficulties in considering it alongside the evidence formally given in these proceedings under the constraints that apply to all judicial proceedings. Understandably, since it was not operating judicially, there is no attempt by the APPG to ensure that what was said by those who spoke to it bore the safeguards applicable to evidence before a court or tribunal, including being amenable to challenge by way of cross-examination. That must be relevant to any judicial analysis of the conversation in the context of an appeal.
89. The difference between the caution employed by Professor Sommer, in particular, in expert opinions for court use and in what he said at the APPG session is striking. In the former circumstances he is cautious about reaching conclusions on the basis of the evidence available to him, as can be seen in the material before us including his description of the material available in another case. In the APPG proceedings, however, he is content to begin his narrative by saying that “it was unsafe for anybody to be relying upon computer files generated by ETS and used by the Home Office as a sole means of making a decision”. On the other hand, he was not able, he admits even in that context, to “identify specific points at which things had gone wrong”, but that “the administrative arrangements and the audit trails were simply not present”.
90. Another major difficulty with the APPG transcript is that it shows that those involved were not entirely well-informed on the materials already available. Nobody seemed to know about the contractual or licensing arrangements between ETS and the Home Office, and the transcript records some statements wholly unsupported by evidence. There is a hint that it was unjustifiable that “over 30,000 people lost their visas because of the allegations that were made that they had cheated”, although in reality there does not seem to be any room for doubt that many more than that number did in fact obtain test results by cheating. There is also the no doubt understandable misapprehension that the question is about whether people could pass a test now, rather than whether they cheated in the test they took.
91. It is perhaps because of this lack of full information that the conversation often seems to lose its structure and mission. Words are put into the mouths of experts; one of the experts appears to interrupt another’s introduction of himself and start questioning him; and following a general invitation one of the experts starts to explain the law of evidence as he understands it because “I’ve read some of their papers”.
92. Even without all those considerations, however, we cannot find anything in the way of facts in the transcript substantially to undermine the existing evidence adduced by the Secretary of State. The conversation really only expands on the possibility that the evidence could have been different. Professor French and Dr Harrison adhere to their previous assessments. Professor Sommer strengthens his opposition to the Home Office, but without adducing any factual or evidential basis justifying what appears to be a change of opinion about the general reliability of the evidence: and even if it is not a change of opinion, it would be clearly wrong for us to regard what he said there as in any way contradicting or superseding his evidence before us.
THE APPELLANTS’ EVIDENCE
93. DK adopted his three witness statements. DK took tests on 17 and 19 July 2013 at UTC. The first witness statement is undated, but was evidently before the First-tier Tribunal. So far as the test is concerned, DK states that he was recommended by his college, the London School of Advanced Studies, to take the TOEIC ETS exam. He says that he did take it, on 17 July 2013. He asserts that he has a good command of the English language, but gives no other details of the test itself.
94. The second witness statement is dated 5 June 2019. This refers to the evidence given before the First-tier Tribunal. DK now confirms that he took the test “on two days”. Referring to the evidence he gave before the First-tier Tribunal, he indicates that he did not remember where he took his test, but it was “not that far from where I was”, in Abbey Wood. He wanted, he says, a test, which was “reasonable for me to get to”. He took one hour in London traffic to reach the centre. Ten to twelve people were present at the speaking test: he had seen no cheating. He did not remember what he was asked about, he thought there were some questions on “sports and other personal involvement”. He received his results through his college.
95. The third witness statement is date 23 August 2020. It is largely a consolidation of the previous two, and appears to add nothing new. He was taken through his witness statements by Mr Turner. He also gave evidence that he had worked as an HGV driver and that he had taken driving tests in English, as well as working in an English environment as a driver. Cross-examined, he said again that he had selected UTC through his college. He thought that Abbey Wood was “maybe in East London”. The college had made no alternative recommendation of a test centre. He had no evidence corroborating his assertion that the choice of UTC was on the college’s advice. Although there were several emails on the subject, he was unable to produce them. He had paid the test sum by cash to the college (not UTC). He had no receipt. He had travelled to UTC from Abbey Wood with a friend, with whom he was no longer in contact: he did agree with Mr Thomann, however, that the friend had shown some dedication in taking him from Abbey Wood to Watford on two occasions and waiting for him to complete the test before driving him back. Mr Thomann put to him that the number of other people being tested was not “ten to twelve” but was in fact twenty-seven. He said that the figure he had given was “just a number”; he was not counting. Mr Thomann pointed out to him that on the day he took the first test, of the twenty seven people there, twenty people took the test by means of a proxy. He said he knew nothing about that.
96. Alluding to the quality of his spoken English at the hearing before us, Mr Thomann drew DK’s attention to the test he took in October 2017 from which DK passed at level B1, a level below that which he had required in 2013 and which his ETS result suggested he had achieved. Mr Thomann put it to DK that the truth of the matter was that he could not afford to fail, and had been guaranteed success by going to UTC which was too good an offer to miss. DK denied that that had been the case. Finally in relation to the location of the test centre, DK said that he had asked for somewhere near Abbey Wood and his college had told him the place to go was Watford.
97. In her evidence, DK’s wife adopted her three witness statements. She did not know him in 2013. The witness statements indicate that it was she who applied to ETS for the voice recordings attributed to DK. In her oral evidence she said that DK was trustworthy and that he had never lied to her. Mr Thomann asked her whether she knew, when they met, that DK’s immigration status was dubious. She said that she had never dealt with immigration and did not know anything about it. Pressed by Mr Thomann, she accepted that she did know that he did not have immigration status but said that at the time she did not know why.
98. RK’s test results derived from tests taken at NLC on 16 May and 19 June 2012. The result obtained from the test on 16 May 2012 was not sufficient to achieve a speaking score of B2. The test taken on 19 June 2012 achieved that result, but the voice recording shows that a proxy test-taker was used. Home Office records show that DK’s application for further leave, lodged on 1 June 2012, was later supplemented with the test result from 19 June.
99. In her oral evidence, RK adopted her two witness statements. Those statements indicate that she was recommended to do the TOEIC test by a friend. She did her own research into centres at which the test could be taken, and discovered NLC was relatively close to where she was living. The journey took her three buses. She did tests on 4 May and 16 May 2012. She made an application for leave to remain on 1 June 2012, which was in due course granted. She had not taken part in any deceit, and had not taken any test at all on 19 June 2012.
100. Cross-examined by Mr Thomann, RK said that she had not been able to provide a statement from the friend who she said had recommended the TOEIC exam, because she had lost contact with her when she had lost her phone. She had not made any other attempt to contact her. In response to questions about her recollection of attending the test, she was sure about the buses, but appeared to be mistaken about the time of the test and, more clearly, about the number of people there: she said about thirty, whereas the records show that there were about eighty at that session. Mr Thomann pressed her very hard on what may be described as her testing history. The score for speaking after the test on 16 May was 150, which was not high enough. She nevertheless insisted that she had taken only one test, and submitted its results to the Home Office, and had been granted leave as a result. She was unable to explain how it was that, despite her claimed ignorance of the fact that her 16 May test had not produced a good enough result, somebody else had known that, and had supplemented her result by a better result in a later test of which she knew nothing. She was asked about the answers she had given on the same topic before the First-tier Tribunal in 2019. A note of that evidence was available to the Tribunal and to the parties. Although she had said that she had submitted a certificate “later on”, she was able to show, with the assistance of questions in re-examination, that, then too, she had denied taking any test in June.
101. There is one further element in the evidence relating to the individual appellants. Each has since obtained from ETS the relevant voice recordings on which their test results were based: DK from July 2013 and RK from June 2012. Neither appellant says that the voice on the recording is his or her own.
102. The voices heard on the recordings are said by ETS to be the voices of proxy test takers where voices are heard on other test entries from the relevant centres. Neither appellant disputes that, and indeed there is no basis for doing so.
103. The voice recognition results show that many test entries are in a voice recognisable as having provided another entry for a different candidate. In other words, at a gross level, the voice recognition results amply confirm the evidence obtained from eyewitnesses of the testing process. This is obviously no surprise. Looked at the other way round, the voice recognition results obtain some confirmation from the eyewitness evidence. In fact, the truth of the matter is that although those who have their own process for voice recognition examination (such as Professor French) can suggest other ways in which this examination of the data could have been made, there is no reason to suppose that the voice recognition process was substantially defective. There may be a false positive rate of one per cent, or even possibly three per cent, but there is no proper basis for saying that the false positive rate was or would be any higher than that. (There would also be a substantial false negative rate, but that does not fall for consideration here: we are not concerned with people who should have been caught as cheating, but were not.) ETS would have no known motive for exaggerating the level of the fraud on their system, and a reputational motive for confining the declared fraud to that clearly demonstrated by the data. We conclude that the voice recognition process is clearly and overwhelmingly reliable in pointing to an individual test entry as the product of a repeated voice. By “overwhelmingly reliable” we do not mean conclusive, but in general there is no good reason to doubt the result of the analysis.
104. The continuity of records between the test candidate and the test entry are the subject of detailed criticism by, in particular, Professor Sommer. As we have pointed out above, he has made numerous suggestions about how the tests could have been operated in such a way as to reduce fraud, and he has indicated places where the evidence linking candidates to entries might not be entirely watertight. In particular, the sound recordings eventually used for analysis do not carry metadata associating them with the recordings received by ETS, because they have been converted by several stages onto a form suitable for use for voice recognition analysis. That does not mean that there have been errors: it simply means that he cannot rule out whatever errors he thinks this hypothetical material might have excluded.
105. Clearly, if there were no general reason to link particular candidates’ input with particular test recordings, that would be a powerful criticism. The circumstances were not, however, that there was any prospect of carelessness or randomness being associated with the continuity of records, either at the point where they were labelled by the test centre or after their transmission to ETS for marking. At the first stage, it is clear from the other evidence that certain test centres were providing a fraudulent service to fraudulent candidates who paid them for it. There is no reason at all to suppose that they would be other than extremely careful to ensure that the fraudulent entries were indeed credited to the fraudulent candidates. The suggestion of any general mix-up at this stage runs counter to the ordinary experience of the provision of a service.
106. The second stage is while the test entries are in the control of ETS. A suggestion of dissociation of entry from candidate at that point strikes at the heart of ETS’s analytical process. If there had been mix-ups at that point it would mean that as an examining authority ETS was unable to be sure that it was, in general, able to attribute the appropriate test results to candidates. Whatever may be said about the level of supervision (or lack of it) by ETS in this scheme, it does not appear ever to have been said that ETS’s examining process suffers from this defect. Indeed, any suspicion of it would destroy ETS’s reputation globally. According to Peter Millington’s statement, not challenged by the appellants on this point, ETS is the largest private not-for-profit educational testing and assessment organisation in the world, administering 50 million tests annually in 25,000 test centres in 192 countries. It is responsible in the USA for the SAT, a college admissions test, taken by 3 million students a year. It also administers the TOEFL (Test of English as a Foreign Language) test, the most widely respected English-language test in the world, recognised by thousands of colleges, universities and agencies in numerous countries, including the UK, the USA, Australia and Canada. It is clear from its international role and continued viability and dominance that (outside these cases) nobody seems to be suggesting that it cannot be relied upon to attribute test entries to candidates correctly.
107. Again, we would not say that the evidence has to be regarded as determinative. There may be room for error (although none of the experts involved has detected any error, as distinct from showing that there is room for error). What is clear here is that there is every reason to suppose that the evidence is likely to be accurate.
108. As Professor Sommer said to the APPG, one of the features of evidence that one would look for is corroboration. He said “it might have been different if there was corroboration, but very often in circumstances there wasn’t”. We are unable to comment on “very often”, but there are two sources of possible corroboration that may well be present when individual cases are examined: the individual’s own account of the test and the evidence (if any) of fraud in the session at which that individual’s test was taken. A further possible source of corroboration may be incompetence in English (i.e. English at a lower level than that required for the test); but it must not be thought that the converse applies: as the then President pointed out in SSHD v MA  UKUT 450 (IAC) at , there are numerous reasons why a person who could pass a test might nevertheless decide to cheat. This is a point that seems to have escaped Professor Sommer in his comments to the APPG.
109. It has been pointed out by the representatives of the appellants and Migrant Voice that there is no evidence from ETS. That is correct, although ETS have had some direct input into some of the witness statements. It is, we believe, quite common for fraud to be proved without direct evidence in support of the proposition being provided by the person whose negligence or lack of oversight allowed the fraud to be committed; and there is nothing in the material before us in the present appeals to suggest that the lack of evidence directly from ETS ought to raise any suspicion that evidence that might exonerate the appellants is being concealed. But be that as it may, we are concerned with evaluating the evidence as it is, not hypothesising about what might be the effect of other evidence: this is not in any sense a case where the lack of particular evidence that might have supported a proposition ought to carry any implication of support for the contrary proposition.
110. The same applies to the various speculations by Mr Turner about ETS’s precise process for analysing the evidence, whether by using a service based in South Korea or not. The witness statement of Ahmad Bidour, who worked for ETS and was involved in inspecting testing centres in the UK and elsewhere, does not help the appellants either. It may show that both ETS and the Home Office were or ought to have been aware that fraud was taking place. But it is very far from suggesting that there was any error in the process of transmission of entries, and of course it is not relevant to the question whether frauds were actually committed by particular individuals, except that it contains personal observations of frauds of the sort alleged in the other evidence.
111. The position is that although Mr Turner wants on behalf of DK to show that other institutions or people were at fault in one way or another, we are concerned only with whether the individual appellants were dishonest. Any faults of others in facilitating dishonesty, or failing to prevent or detect it, are not to the point at all.
112. Nor, we should add, and despite what is suggested by Mr Turner, is there any breach of any duty of candour owed by the Secretary of State in these cases. That duty is a feature of judicial review proceedings: in an appeal the equivalent duty is to ensure that the Secretary of State does not ‘knowingly mislead’ (R v SSHD ex parte Kerrouche (No 1)  Imm AR 610). But even if there were a duty of candour here, there is no basis for saying that it has been breached. There is nothing apparently available to the Secretary of State that ought to have been disclosed to the court and has not been.
113. For the same reasons we are not concerned with the suggestions by various witnesses about different processes of analysis that might have been undertaken, nor with what different techniques for the identification of candidates and test entries might have been used. The possibility that there could have been even more accurate evidence does not impinge at all on the assessment of the accuracy of the evidence that is available.
114. That takes us to a crucial observation about the appellants’ arguments in these proceedings. The appellants’ arguments have been largely directed to demonstrating the possibility of error in the evidence – or error in determining the conclusion to which the evidence points. In particular, attention is drawn to the possibility of a false positive in voice recognition, or a failure in maintaining proper labelling of test data. As we have indicated, the former is assessed to be likely but low; the latter, the “chain of custody” argument, remains only a theoretical possibility not supported by any detailed evidence, and rendered less likely by some of the general evidence. But it is important to appreciate that although these possibilities prevent the data conclusively proving fraud in a scientific sense, they do not substantially remove the impact of the evidence as capable of establishing facts in issue so that a human trier of fact is satisfied of the matter on the balance of probabilities.
115. The existence of possible error is a feature of all, or nearly all, evidence. A defendant in a tort action will not succeed merely by showing that the scientific evidence, of a typically reliable nature, pointing directly to responsibility, may be erroneous in the instant case. If faced with scientific evidence apparently pointing to his liability, he will need to provide some good reason why it should not be trusted in the instant case. Even in criminal cases, where the standard of proof is higher, a case can be proved despite the evidence being shown to be potentially or even statistically fallible.
116. Indeed, even where the evidence in question is of a kind notoriously prone to error it can found a decision that the case is proved beyond reasonable doubt. The most obvious example is identification evidence. A case is withdrawn from the jury (as unable in law to support a conviction) only if (i) it depends on the identification evidence, and (ii) the identification evidence is of poor quality, and (iii) the identification receives no support from any other evidence in the case: R v Turnbull  QB 224. That is despite the numerous attested instances of wrong identification, both in proceedings and in ordinary human experience.
117. Showing that the case is not watertight is not sufficient in civil proceedings to show that it need not be answered, or that it is insufficient to prove a fact in issue. The evidence the respondent relies on in these cases is not shown to be unreliable in any general sense. On the contrary, the very limited concerns that have been raised tend to show that as a class the evidence is highly reliable, although not necessarily wholly free from error. All that the appellants’ and intervenor’s arguments show in reality is that the evidence upon which the respondent relies has a similar feature to almost all evidence in almost all cases: it is not infallible.
118. One thing is clear beyond any conjecture: that there were numerous cases of test results obtained fraudulently by the use of proxies. That is proved at a standard well beyond what is required in these proceedings by the criminal convictions and the evidence that led to them. It was striking that some of the material we were asked to consider in these appeals included assumptions that nobody had obtained results, or leave to remain, by fraud. That is a view that is simply not open to anybody with any regard for the rule of law or the value of truth.
119. In the context of the test centres as ‘fraud factories’, a description that is in our judgment properly applied to both NLC and UTC, it is overwhelmingly likely that those to whom the proxy results are now attributed are those who took their tests by that method. There are two parallel strands to this conclusion, and neither contradicts the other. Each could stand alone, but their combined effect is wholly compelling.
120. The first strand is that tending to show that there is no good reason to think of any error in linking the entries examined and classified by ETS with the entries actually submitted on behalf of the candidates to whom they are attributed. The academic evidence was that the “chain of custody” was not absolutely secure and could have been better. That is a world away from saying that it was not in fact wholly reliable. The assertions are made in wholly general terms and any general assertion of that sort is an assertion that ETS has not been reliable in attributing test entries to candidates. But ETS is not concerned only with centres administering TOEIC to those who need a particular result for UK immigration purposes. ETS is, as we have noted above, an organisation with an international portfolio of important and well-regarded tests. We have not heard of any substantiated claim that in respect of any test they have failed in the most basic duty of an examining board, to ensure that the entry examined and attributed to a candidate is the entry made by or on the authority of that candidate. It is in our view inconceivable that ETS would retain the work that they do, if it was thought that their administration was in this respect unreliable.
121. The speculation on what might have been done in terms of a “chain of custody” in the present cases simply adds nothing. We are not concerned with the standard of proof on evidence that might be needed in a criminal case. We are concerned instead with what is done by a responsible body with a clear interest in securing the integrity of results.
122. The second strand is that tending to show that there is no reason to consider that anybody other than the candidates and the test centres in collusion would have wanted to falsify results in this way. We may test this strand by seeing what opportunities there might be for interference with a genuine entry to turn it (as it were) into a fraudulent one. We use that phrase because it is important to be clear that it would be necessary both to add the fraudulent (proxy) entry to the material examined by ETS and to remove the genuine entry. The genuine entry could not be left with its original identity, because that would prevent the addition of another entry having the same co-ordinates; and it could not be exchanged with another genuine entry because that would result in unexpectedly low marks for the victim of that exchange, with consequent appeals, comparisons of the candidate’s voice and that on the test, and clear evidence of the sort of failure of administration that we have just indicated in our discussion of the first strand as completely absent.
123. What would be needed therefore is a process after a candidate’s genuine entry that could substitute an entry consisting of answers given by a proxy tester. As the evidence before us showed, unsurprisingly, this would require considerable technical ability in breaching the security of the test system. There is no evidence that that could even be achieved. If it could be achieved, it is virtually inconceivable that it would be undertaken without any reward for those taking part. Why should anybody go through the test entries, and take great trouble manufacturing better entries and substituting them, if not at the instance of the candidate? In any event there is not the slightest evidence that anybody did or would act in this way.
124. Professor Sommer speculated that a test centre might act in this way in order to manipulate results and so achieve a favourable reputation. That will not do at all. A college might possibly want to make sure that its students did well. If they did, the college could advertise its success in bringing students to the necessary standard and so get more students. But a test centre could not advertise itself in that way: if the tests are administered properly there is no good reason for examinees at one test centre to do better than at another. Even without advertising, the results, if they became known, would risk exciting suspicion (indeed the level of success at certain test centres was one of the features leading to the analysis in these cases). Anyway, a test centre has no particular interest in getting candidates through a test: if they pass it never sees them again, but if they fail, they have to pay to take the test again.
125. There is no perceptible way in which the proxy test entries could have been inserted in the system after the candidates had taken honest tests; and there is no perceptible reason for anybody to insert or substitute them, except at the instance of the candidate. We are left, therefore, with the time of the taking of the test. The material that achieved notoriety in the Panorama investigation and which was used in the criminal trials as well as in earlier episodes of the ETS litigation in these Tribunals shows what happened there. Two observations need to be made. The first is that it is highly unlikely that any candidate present on one of the occasions when proxies were being used was not fully aware of what was going on. The second is that it is if anything even more unlikely that such a system would then attribute proxy entries to anybody who had not taken part in the dishonest scheme, making whatever payment or other arrangement was in place.
126. The two strands, therefore, amount respectively to the virtual exclusion of suspicion of relevant error by ETS, and the virtual exclusion of motive or opportunity for anybody to arrange for proxy entries to be submitted except the test centres and the candidates working in collusion.
127. Where the evidence derived from ETS points to a particular test result having been obtained by the input of a person who had undertaken other tests, and if that evidence is uncontradicted by credible evidence, unexplained, and not the subject of any material undermining its effect in the individual case, it is in our judgment amply sufficient to prove that fact on the balance of probabilities.
128. In using the phrase “amply sufficient” we differ from the conclusion of this Tribunal on different evidence, explored in a less detailed way, in SM and Qadir v SSHD. We do not consider that the evidential burden on the respondent in these cases was discharged by only a narrow margin. It is clear beyond a peradventure that the appellants had a case to answer.
129. In these circumstances the real position is that mere assertions of ignorance or honesty by those whose results are identified as obtained by a proxy are very unlikely to prevent the Secretary of State from showing that, on the balance of probabilities, the story shown by the documents is the true one. It will be and remain not merely the probable fact, but the highly probable fact. Any determination of an appeal of this sort must take that into account in assessing whether the respondent has proved the dishonesty on the balance of probabilities.
ASSESSMENTS AND DECISIONS
130. The appellants have obtained the test entries attributed to them. It seems to us that this is an important further safeguard, as it rules out the possibility that the invalidation of the result was an inappropriate response to the entry. Appellants may be expected to show an error of that sort if they say there has been one. Where an appellant does not say that the voice on the recording is his or her own, there is no “voice recognition” issue: the only question can be the “chain of custody” issue.
131. The appellants’ cases are that there must have been a “chain of custody” error. They rely on their own assertions about the tests. If credible, and sufficiently comprehensive, such assertions might perhaps, in an individual case, suffice to prevent the Secretary of State establishing dishonesty on the balance of probabilities. In the present cases, however, there are good reasons to disbelieve the appellants’ evidence.
132. DK’s persistent attempt to sideline the undoubted fact that the tests he took were in Watford suggests that he is not prepared to be frank about the difficulty in travelling there. He is not unacquainted with the geography of the United Kingdom; indeed he has worked as a HGV driver. He was not remotely accurate in his recollection of the number of people who took the tests at the same time as he did, and we do not believe his claim that he observed nothing wrong at those sessions, at which a large number of tests by proxy were taken.
133. His wife was not being frank about her knowledge of his immigration status when they met and subsequently. We conclude that the reason for that was that she has always known that it would not bear examination.
134. RK’s position is wholly incredible. Her test result was insufficient for the application she made for further leave. She claims that she did not know that, and made the application, right at the end of the time available to her on the strength of having taken the test. Subsequently a further test result was sent into the Home Office that was sufficient to remedy the insufficiency in the earlier test. She says that she did not take a further test and that she knows nothing of the supplementary submission to the Home Office. If that is right it means that there was a person who not only knew more about her results than she did but also knew what they were needed for and what use had been made of them, and had the facility and documentation to make a further test entry in her name, secure a new satisfactory result and forward it to the Home Office as if from her, all without her knowing anything about it.
135. We do not believe her. In particular we do not believe that she was unaware of the standard required, or that she had failed to meet it in the first test. She must have been aware that she needed a supplementary result and her denials are wholly unpersuasive. In our view they are obviously intended to cover up the fact that she fraudulently obtained a satisfactory test result and forwarded it to the Home Office because without it, at that very time, her leave would expire with no possibility of renewal.
136. RK’s evidence does not, in our judgment, reach even the minimum level of plausibility referred to in paragraph  of Shen. The evidence adduced by and on behalf of DK falls well short of making any impact against the evidence on the other side.
137. The Secretary of State has discharged the burden of proof in both cases. Looking at the evidence as a whole we find that she has amply established that the appellants each employed dishonesty in achieving the test results on which they relied in order to seek further leave to remain in the United Kingdom. Although we have been looking in detail at the evidence relating to the tests DK and RK took, these appeals are on human rights grounds. Because of the conclusions we have reached, the grounds based on the tests give no reason to consider that the decisions were unlawful as being contrary to s 6 of the Human Rights Act 1998. Neither DK nor RK has argued in the course of these appeals that if it were found that they cheated in the tests the decision to refuse them leave would breach section 6. We have dealt above with DK’s “new matter”. The appeals of DK and RK are accordingly dismissed.
C. M. G. OCKELTON
VICE PRESIDENT OF THE UPPER TRIBUNAL
IMMIGRATION AND ASYLUM CHAMBER
Date: 24 March 2022