15. The overall concern raised with us by our six complainants is that the Home Office has failed to compensate them for the impacts they suffered after being accused of deception in English language tests. In all their cases, tribunals have found they did not cheat in the test. The impacts they experienced are serious, ranging from detention to years spent unable to continue in education or embark on careers. They believe that the decisions the Home Office took in 2014 and 2015 when it cancelled their leave in the UK were maladministrative. We have looked at the specific concerns they have highlighted about what happened and the process the Home Office followed.
UKVI failed to gather or consider relevant evidence before coming to a decision to cancel their leave. They did not consider the circumstances of each case and/or any evidence the applicants might have. UKVI solely relied on information from the Company.
16. Our ‘Principles of Good Administration’ say public bodies must comply with the law and have regard for the rights of those concerned. They should act according to their statutory powers and duties and any other rules governing the service they provide. They should follow their own policy and procedural guidance, whether published or internal.
17. Our principles say public organisations should take account of all relevant considerations, ignore irrelevant ones and balance the evidence appropriately when making decisions. Organisations necessarily assess risk in decision-making and should do so reasonably.
18. The Home Office relied on evidence from the Company to accuse our complainants of deception. They did not consider the particular circumstances of each individual before making the decision that they had cheated and cancelling their leave. In each case, they had evidence from the Company that a ‘proxy’ test taker had taken their English language test instead of the complainants themselves. We have considered if it was reasonable for the Home Office to rely on this evidence to make an allegation of deception.
19. After Panorama exposed widespread cheating in its test centres, the Company used voice recognition software to identify tests which had been taken by proxies – third parties who took multiple tests on behalf of candidates. The Company then checked this evidence using two human listeners. This evidence said that nearly all TOEIC tests taken through the Company between 2011 and 2014 were either invalid or questionable. The Home Office says that this still represents a very small percentage of all student migrants during this period as separate companies were also providing English testing arrangements who were not involved in the cheating scandal. They have also highlighted that the criminal investigation into TOEIC fraud has identified over 110 suspects (college directors, test centre administrators, agents and proxy test takers) since 2014. Nevertheless, the high number of questionable and invalid tests raised questions about how a system could operate with such widespread fraud and how such a high proportion of test-takers could feasibly have been cheating. In this context, in subsequent years, concerns were raised about the validity of the Company’s evidence in 2014 and 2015, in particular the voice-recognition software they used to identify tests taken by proxies. (The software was new and had not been used in this context previously.) Concerns have also been raised about the handling of the data by the Company and the integrity of the process it used to attribute tests to candidates. These were considered in detail in the case of DK/RK, which we look at further below – it did not find these concerns to be substantiated.
20. Our complainants have highlighted a number of reports which raised questions about the validity of the Company’s data. The National Audit Office published a report looking at the Home Office’s response to cheating in English Language tests in May 2019. This said it was difficult to establish how many people had been wrongly accused of cheating. Estimates ranged from 1% (the expert engaged by the Home Office in 2016) to a National Union of Students expert review which said that voice recognition software could have made mistakes in 20% of cases and human listeners in an even greater number.
21. The following month, the All Party Parliamentary Group (APPG) on TOEIC also released a report. This was critical of the Company’s evidence and said it could not be relied on. In particular, the APPG report said there were a large number of anomalies in the data and a lack of proof to link individual recordings to the people who sat each test. The report suggested it was implausible that nearly all tests taken were suspicious. It was also critical of the Home Office’s failure to seek expert evidence in relation to the reliability of the Company’s evidence until 2016.
22. Following the APPG report, the courts considered in detail the reliability of the Company’s evidence in DK & RK (ETS: SSHD evidence; proof) India [2022] UKUT 00112. The Upper Tribunal found that the evidence relied on by the Home Office to establish fraud in individual cases – in other words the evidence from the Company - was amply sufficient to decide it could demonstrate there was sufficient evidence of deception. The court said there was no good reason to conclude the evidence did not accurately identify those who cheated. Each case fell to be determined on its own facts and evidence but the context for this consideration was that there were thousands of fraudsters and the process for identifying them had not been shown to be generally inaccurate.
23. This judgement considered evidence from a number of experts and witnesses. It said that the potential error rate in the voice recognition process (in terms of false positives, where an individual was accused of fraud when they had not cheated) may be between one and three percent but there was no basis for thinking it was any higher than this. The court noted that the existence of possible error is a feature of all, or nearly all, evidence. In terms of ‘chain of custody’ concerns – such as failure by the Company to properly record test data – the court said this was only a theoretical possibility and was not supported by any detailed evidence. The Court highlighted other common-sense considerations – such as that the Company was very unlikely to seek to overstate the number of tests which were fraudulently conducted by its own centres.
24. Our complainants say they were found to be innocent at the end of the process. They say they should have had the opportunity to disprove the allegation of cheating before the Home Office issued decisions to curtail (cancel) their leave to remain and removal notices for them to leave the UK. The complainants believe this would have given the Home Office access to evidence which later led to them being found to be credible by the courts. The complainants say this would have prevented them being issued with a curtailment/removal notice and having to go through the process of proving their innocence. For example, in Mr B’s case the Tribunal – which took place before the judgement in DK/RK - found he was a reliable witness and likely did not cheat on his TOEIC test. They based this on his account of taking the test and having a good level of English after studying in Bangladesh and graduating from University in August 2015. He had also supplied other English language certificates. In his tribunal hearing, the judge found the Home Office had not discharged the burden of proof and raised concerns about the accuracy of the Company’s evidence.
25. Given the conclusions reached by the courts in DK/RK, the accepted legal position is that the evidence from the Company was amply sufficient for the Home Office to decide those with invalid tests results would have their leave to remain cancelled and be issued with removal notices. It follows that these were the ‘relevant considerations’ they needed to take into account in our complainants’ cases.
26. We have considered whether it would have been feasible or necessary for the Home Office, before making this decision, to seek to gather individual evidence from each of the many thousands of individuals involved, particularly given the urgency of taking action in a context of wide-spread criminality. At the end of 2019, the Home Office published new guidance which brought in a ‘minded to refuse’ process which means allegations of false representations are put to applicants in a letter to give them the opportunity to respond directly. This process was not in place at the time the decisions in English language tests were taken. The Home Office told us that in respect of TOEIC fraud there was no mechanism at the time that could have identified individuals who may have been innocent of the allegation of cheating from a cohort of 34,000 people.
27. We have looked at this point in more detail below in terms of the process the Home Office followed. We have taken account of the Home Office view about the need for an urgent response to wide-spread criminality, the evidence provided by the Company, the very large numbers of individuals involved and the mitigations the Home Office put in place to avoid incorrect accusations (such as listening to the test recordings and inviting those with a questionable result to take a further test). Taking all these considerations into account, we do not think that it was maladministrative for the Home Office to make decisions to cancel leave without inviting each applicant to make further representations. The process followed by the Home Office included a right of appeal for those accused of cheating which allowed for individual evidence to be put forward.
28. At the time the Home Office made these decisions, the evidence showed the information from the Company was reliable – although it knew it might contain a small error rate. The Courts have confirmed this through DK/RK – it has now found there was no good reason to conclude the evidence from the Company didn’t accurately identify those who cheated. This meant the Home Office were entitled to make the decisions they did, but these decisions were rebuttable by the individuals involved. Each case should be assessed on its own facts and evidence, but the context of this assessment is that there were thousands of fraudsters and the process of identifying them has not been shown to be generally inaccurate.
29. We have not seen any evidence to suggest concerns with the conclusions reached in DK/RK- the position reached in this case about the Home Office having sufficient evidence to make an accusation of cheating based on the Company’s evidence has been upheld and strengthened in subsequent caselaw. The court in DK/RK said that where there is evidence from the Company that points to a test having been taken by someone else (a proxy test taker) this requires a believable explanation.
30. We find that it was not maladministration for the Home Office to rely on the evidence provided by the Company when making the decision that our complainants had cheated and to cancel their leave. We consider below the complainants’ subsequent concern that there was no fair process to enable them to put forward their own individual evidence to rebut this accusation of cheating.
Despite knowing some of the people accused would be innocent, UKVI did not create a fair process to allow those affected to challenge their decision. At the time of its decision in their cases, UKVI only allowed response or challenge by an out of country appeal, which was later found by the courts not to be an effective remedy. Our complainants’ only alternatives, and only way of remaining in the UK was to request a judicial review or make a claim on human rights grounds.
31. We have reached the view that the Home Office had sufficient evidence to accuse our complainants of fraud based on the information provided by the Company.
32. The process they followed when they did this needed to be in line with our Principles and the relevant legislation. Our Principles say public bodies should act according to their statutory powers and duties and any other rules governing the service they provide. In decision-making they should have regard for the relevant legislation. Our Principles also say that if applying the law, regulations or procedures strictly would lead to an unfair result for an individual, the public body should seek to address the unfairness.
33. The Home Office used section 10 of the Immigration and Asylum Act 1999 to issue removal decisions in cases where they decided deception had been used in an immigration application. Section 10 meant an individual who had been identified as using false representations in order to gain leave to remain by deception could be removed from the UK and limited to an out of country appeal only (i.e. the individual had to leave the UK and return to their country of origin in order to launch an appeal). The effect of this was that an individual was served with two notices at the same time - both a curtailment notice (cancelling their leave to remain) and a removal notice under section 10 (informing them of the Home Office’s intention to remove them back to their country of origin). Appeal rights were out of country in all cases where an applicant was accused of making false representations (not only cases where fraud was alleged in English language tests). The Home Office used this process in cases where they had evidence of fraud in English language tests until 20 October 2014 when the Immigration Act 2014 came into effect. This created a single power of removal and streamlined the removal process. A one stop process was introduced in which the immigration decision to cancel leave and the enforcement decision commencing the removal process were included in the same letter. The legislative context at the time the Immigration Act 2014 came into effect was that appeal rights were removed for all points-based applications in 2014 and 2015.
34. We have considered whether it was appropriate for the Home Office to take a ‘business as usual’ approach to these cases, using Section 10 and subsequent legislation. In particular, we have considered the circumstances of these cases, which involved thousands of people accused of widespread organised fraud, and whether being restricted to an appeal right outside the UK was fair.
35. The Home Office followed the legislation in place at the time to cancel the leave of those accused of cheating in English language tests, including our complainants. It follows that the Home Office’s decision to do this where they had evidence of fraud was lawful.
36. We have considered whether the Home Office should have considered doing something other than taking a ‘business as usual’ approach. Relevant factors included in particular the scale of the numbers accused of deception, the possibility of errors in the data and the potential impact on the courts in the coming months and years due to the volume of cases involved.
37. It is clear the Home Office needed a rapid and robust response given the scale of the fraud that took place. It says, and we accept, the urgency of the situation was a significant and relevant consideration in its decision making. We also know the Home Office implemented a number of measures which show it did consider the possibility of error and unfair impacts. For example, it treated people with ‘questionable’ results differently to those with ‘invalid’ results, despite the Company cancelling the test results of both groups.
38. The Company also told the Home Office it had implemented systems to attempt to minimise any error rate. The Home Office told us it had not simply taken the Company’s word for it about the numbers of invalid and questionable test results. It provided us with information about the actions it took to assure itself that the evidence from the Company was robust. Staff from the Home Office travelled to the Company’s head offices to understand the assurance process all TOEIC tests went through as part of establishing whether individual candidates had cheated by using a proxy test taker. The Home Office said where software identified a proxy, two trained human assessors reviewed the recordings before a decision was taken whether to cancel the result as invalid. The Home Office said it had ensured appropriate steps were taken to understand the Company’s processes and data. It said the data provided by the Company was also reviewed by a Home Office analyst, and it later employed a forensic speech and acoustics expert to independently analyse the Company’s methodology.
39. While questions have been raised repeatedly about the extent of an error rate in the Company’s data, we have not seen that, at the time the Home Office made the decisions to cancel leave (in 2014 and 2015), it had evidence to suggest there was a significant problem with the information provided by the Company. This is supported by the judgement in DK/RK discussed above. While the Home Office could not be sure no errors would occur, the information they had at the time suggested these would be relevant to only a very small percentage of the decisions made.
40. Through subsequent caselaw, Balajigari (which was not decided on English Language tests), a ‘minded to refuse’ process was developed where individuals accused of false representations can have the right to respond to allegations, either in writing or via an interview. At the time, the Home Office was lawfully following primary legislation when it made the decisions to cancel leave. They did not have evidence to suggest the data from the company was widely inaccurate, they had taken steps to reduce the potential for people to be wrongly impacted and there was an appeal process for people to challenge the decision.
41. The complainants have said they were innocent. Therefore, they considered the process followed by the Home Office was not fair because the way to challenge the decisions to cancel their leave to remain and remove them was via an out of country appeal right. They said the out of country appeal was later found by the courts not to be an effective remedy. The court case in question is Ahsan v SSHD which was decided in 2017 – after the Home Office had made decisions in our complainants’ cases. It concerned the fairness of an out of country appeal for several people who had been accused of cheating in an English language test.
42. The Court of Appeal in this case said an out of country appeal would not give the Appellants a fair way to challenge the decisions to remove them from the country. However, the Court of Appeal’s decision did not amount to a blanket finding, but was focused on the specific circumstances of the students’ cases. In particular, relevant considerations included both whether the matter could be fairly decided without hearing their oral evidence and also the fairness of the out-of-country appeal process for people who could not access video-link facilities to give this evidence if the video links were not realistically available in their countries of origin. These circumstances were likely to be shared by a majority of those accused of fraud in English Language tests.
43. As a result, the Home Office changed its approach following the judgement in Ahsan. After this date, it used a workaround to allow those accused of cheating in English Language tests to make an in-country appeal. This effectively meant allowing students to make an appeal from within the UK rather than returning to their country of origin. Commonly, this involved asking or allowing the students to make a leave application on human rights grounds, which the Home Office would then consider. If the Home Office went on to refuse the application, the refusal notice carried the in-country right of appeal to the Tribunal.
44. Our complainants have not said they tried to access their original out of country appeal right and were prevented from fairly doing so by the concerns raised by the appellants in Ahsan (access to oral evidence/video link facilities). They simply assert the requirement to leave the UK in order to appeal is unfair. After Ahsan, the Home Office altered its position to ensure those accused of cheating did have access to a fair mechanism to challenge the accusations against them. This meant our complainants were eventually able to access tribunal hearings where they successfully put forward arguments that they had not cheated in their English Language tests and were subsequently granted leave to remain in the UK.
45. We have not seen evidence to suggest that prior to the judgement in Ahsan the Home Office should have considered that an out of country appeal right was unfair. It was the standard approach, enshrined in legislation, in cases involving allegations of false representations. Previous caselaw, such as Gazi, had found out of country appeal rights to be an adequate remedy (though the issues with giving oral evidence in many of these cases were not raised). In the years following the original decisions in English Language test cases, the caselaw has evolved, and it follows that the Home Office’s approach to these cases has rightly also evolved to take this into account. We have to consider whether the Home Office’s actions were maladministrative at the time they took the decisions based on the information they had then. We have not seen evidence to suggest this. They issued decisions in line with appropriate legislation and took action to reduce the potential for people to be wrongly accused. We have not seen evidence to suggest that they knew the appeal rights these decisions attracted were unfair, and we have seen they changed their approach when the courts highlighted this.
46. We do not uphold the complaint that the Home Office did not follow a fair process to allow those affected to challenge their decision.
UKVI’s compensation decision failed to take account of their tribunal decisions and growing number of reports such as ones by the National Audit Office, the Public Accounts Select Committee and an All-Party Parliamentary Group, all of which suggested UKVI’s decision-making and decision-making process had been flawed and unfair to those who had been wrongly accused.
47. UKVI’s Ex-Gratia Policy is designed to provide redress for maladministration. Payments are made at the discretion of the Home Office and depend on the individual circumstances of each complaint. Deciding whether to make a payment is a matter of judgment and the reasons for decisions must be clearly recorded. The guidance says ex-gratia payments will be considered where maladministration has been identified by the department. Maladministration is generally agreed to be a lack of care, judgement or honesty in the management of something. The guidance says that if a decision on an application is reasonable (given the caselaw and/or Home Office guidance at the time) but subsequently held not to be sustainable, maladministration will not have been found.
48. The UK Central Government Complaint Standards say organisations should give fair and accountable responses. This requires public bodies to give a clear and balanced account of what happened, based on established facts. It gives clear references to any relevant legislation, standards, policies, procedures or guidance, based on objective criteria.
49. Our complainants made a claim to UKVI for ex-gratia compensation on the basis that its decisions in 2014 and 2015 to revoke their leave were maladministrative. They said:
• UKVI had not put forward evidence to the tribunal hearings in support of its assertion they had cheated. The complainants considered that the tribunal decision meant UKVI’s earlier/original decisions to revoke leave had failed to take into account all relevant considerations.
• UKVI relied on data from the Company when making decisions to cancel leave rather than making individual factual assessments of each claim. There was a failure to use a robust and reliable system, with inbuilt safeguards, to make serious accusations of fraud.
• It was not possible to challenge UKVI’s decision on the facts between the date the decisions were made (in 2014 and 2015) and the dates of our complainants’ tribunals (in 2019-2020) because there was only an out of country appeal right which was found by the courts [on Ahsan] not to be fair.
• Mr S also said he was not sent a notice to inform him his leave had been curtailed. This was before the Panorama programme aired. His leave was curtailed because the licence of his sponsor college had been revoked and not because of allegations of cheating.
• Mr Y said that when his appeal was allowed the judge said he should be treated as if he had been residing lawfully in the UK for more than 10 years. Mr Y said this meant the Home Office should reimburse NHS costs that he paid during this period.
50. The Home Office has refused to provide compensation on the basis the courts have consistently stated over the years that it had enough evidence to decide there was evidence of deception (leading to decisions to cancel leave to remain and issue removal notices). It said that the existence of appeal rights in primary legislation is a safeguard to allow for the possibility that the facts of a case about cheating may be differently interpreted and weighed up by different decision makers. It disagreed it was maladministrative to accuse students of cheating who were later found by the court to be innocent.
51. In its responses to our complainants’ ex-gratia claims, the Home Office said it would only offer an ex-gratia payment where it accepts on the facts, established on the balance of probabilities, that its actions were maladministrative and resulted in a financial loss. The Home Office said it did not believe this threshold had been met in the complainants’ cases. It noted the evidence (such as the NAO report) referred to by the complainants, but said the courts (citing caselaw references) had consistently found that UKVI had met the evidential burden to make an allegation of cheating.
52. Mr S’s ex-gratia claim included a complaint about the curtailment of leave prior to 2014. In its response to Mr S, the Home Office said that curtailment action had been taken correctly because his college’s sponsor licence had been revoked. It said the other actions it had taken, including detaining Mr S, were reasonable and necessary in order for them to carry out their duty to maintain effective immigration control. It maintained its position that its actions were not maladministrative.
53. As per the two heads of complaint above, we have not found maladministration in the Home Office’s actions and decisions at the time they made accusations of deception and issued decisions to our complainants with out of country appeal rights. The Home Office’s complaint responses to the six complainants were issued at different times, but its position in all its responses (that it had grounds to believe students with invalid test results had cheated) has been strengthened by the caselaw of DK/RK and subsequent cases. The Home Office’s complaint responses reflected what we have found, took account of evidence provided by the complainants and gave reasons why compensation was refused. For these reasons, we consider the Home Office’s responses to the six complainants were in keeping with its ex-gratia policy and the UK Government Complaint Standards.
54. We appreciate that this decision is not what our complainants were hoping for. A number of years after these events, they were found by tribunals not to have cheated and now have valid leave to remain in the UK. We have considered the evidence available to the Home Office and the processes it was expected to follow at the point the original decisions to curtail leave to remain were made. In some cases, this happened many years prior to the tribunal decision. We have not found that the Home Office acted with maladministration when it made decisions in 2014 and 2015 based on what it knew at that time. In making this finding, we do not seek to diminish the experiences our complainants have described to us or the on-going impact on their lives.