Research and analysis

GPS expansion pilot evaluation

Published 13 January 2025

Applies to England and Wales

Authors

Home Office Analysis and Insight (HOAI)

Acknowledgements

We would like to thank the GPS expansion operational team for preparing and providing information about the pilot and HOAI analysts to support with the scoping and quality assurance of the analysis and report.

Executive summary

The project

This report presents findings of the Evaluation of the GPS expansion pilot, a pilot testing the expansion of the use of electronic monitoring (EM) as a condition of immigration bail. The pilot started in June 2022 and was extended until December 2023. This report presents findings based on data of individuals who joined the pilot between June 2022 and October 2023.

The pilot involved monitoring 2 groups of individuals undergoing immigration bail: 1) tagged individuals experiencing EM and traditional face-to-face reporting; and 2) a control group experiencing traditional reporting only. This evaluation compares information about the behaviours of these 2 groups to understand if EM had an impact on compliance.

Key findings

This report suggests that GPS tagging did not improve compliance behaviours of asylum claimants on bail conditions. Interim findings in October 2022 and August 2023 presented a similar picture. The pilot underwent significant operational challenges interlinked to compliance behaviours, from setting up reporting and tags efficiently to legal challenges to the process.

Compliance rates dropped over time following a similar trend across tagged and control groups; however, individuals in the tagged group tended to go out of contact more quickly than in the control group. After 16 months, there was no statistically significant difference between the proportions of individuals out of contact (66% in both groups).

The use of an EM tag does not appear to be a barrier to absconding. At the end of the pilot, 14% individuals in the control group were marked as absconders against 16% in the tagged group. It took on average 219 days for individuals to be officially declared absconders, this timeframe was similar across both control and tagged groups.

Future behaviours

Due to the non-random nature of the sample in this study, it is not possible to predict future behaviours; however, findings from this pilot show that under these circumstances the use of this particular technology did not appear to have improved compliance.

1. Introduction

1.1 Background

Tackling irregular migration is a key aim of the Home Office (HO). The majority of small boat arrivals claim asylum, in the year ending June 2023, 90% (36,169 of 40,386 arrivals) had an asylum claim recorded.^{[footnote 1]}

Some asylum claimants subject to immigration control who are deemed inadmissible or undergoing assessment are detained in Immigration Removal Centres (IRCs). They can be released into the community if there is no imminent prospect of removal.

Asylum claimants who are released are subject to immigration bail which can be granted either by the Secretary of State (SoS) or an Immigration Judge (IJ) to claimants meeting certain criteria (for example, being held for immigration matters) and willing to follow conditions (for example, being restricted on where they can live, whether they can work, attending appointments).^{[footnote 2]}

As part of their bail conditions, asylum claimants must maintain contact with the Home Office. This entails meeting an Immigration Enforcement Officer and scanning a barcode in one of the 17 Reporting and Offenders Management centres (ROMs) across the United Kingdom. The most common reporting frequency is fortnightly, and sometimes digital or telephone reporting is used instead of in-person reporting. Frequency and mode of reporting are determined by the IJ and SOS as part of the bail conditions.

Many individuals who arrive via dangerous routes might be incentivised to be compliant and report as requested, as non-compliance will affect their asylum claim or result in more stringent bail conditions. However, if in some instances asylum seekers believe their likelihood of being granted asylum in the UK is low, they might be less inclined to be compliant with any conditions of immigration bail and are at higher risk of absconding.

If individuals do not maintain contact with the Home Office, this presents difficulties in progressing their case and will also impact the ability to remove them if they no longer have a right to remain in the UK.

1.2 The pilot

The Home Office established a pilot, which took place from June 2022 to December 2023, to expand the use of electronic monitoring (EM) as a condition of immigration bail in England and Wales.^{[footnote 3]} The aim was to understand the impact of the use of electronic monitoring to inform future roll out and operational decisions. In this pilot, electronic monitoring involved installing a fitted device (a ‘tag’) to the ankle of an asylum claimant on the day of release from detention.

Electronic monitoring was intended to ensure individuals on immigration bail remained traceable, in contact with Immigration Enforcement and were compliant with their bail conditions. The same technology is also used by His Majesty’s Prison and Probation Services (HMPPs) to support community-based sentences or introducing individuals back to the community.

Electronic monitoring under the pilot was used where its application did not breach individual rights based on the European Convention of Human Rights.^{[footnote 4]} It was not used for certain categories of individuals, such as pregnant women, minors, and those with specific vulnerabilities, such as disabilities and medical conditions. Asylum claimants went through a GPS tagging suitability assessment (Annex A) before being enrolled in the pilot. People with vulnerabilities were not tagged and subsequently became part of the control group.^{[footnote 5]}

The pilot involved monitoring 2 cohorts of individuals undergoing immigration bail: 1) a control group undergoing traditional in-person reporting in reporting centres (ROMs) across the country (without tagging); and 2) a group of individuals undergoing traditional reporting but also fitted with a tag. Participants were recruited until October 2023.

The nature of the selection process as well as the prevalence of individuals with certain characteristics (for example, a high number of Albanian nationals) meant that the sample was not randomised and results in this report are not generalisable and representative of all potential cohorts.

After December 2023, all individuals who underwent electronic monitoring went back to traditional reporting as the pilot ended and there was no legal basis to maintain the use of tags. Those who did not receive asylum decisions remained subject to traditional reporting requirements even as the pilot ended. Some individuals might have become non-compliant or absconders at a later stage, but it was outside the scope of this pilot to monitor if the ceasing of the use of tags affected an individual’s compliance, or what happened to people following this, in terms of apprehension and removal outcomes.

1.3 The evaluation

The aim of the evaluation was to understand whether EM was an effective way to improve and maintain regular contact management with asylum claimants on immigration bail. This is the final concluding report from the evaluation. The evaluation specifically examines the impact of EM on compliance. This has been pursued through a mixed-methods design including quantitative quasi-experimental and qualitative approaches.

Home Office Analysis and Insight (HOAI) produced 2 interim reports in October 2022 and August 2023 which noted that GPS tagging had not appeared to improve compliance behaviours of asylum claimants on bail conditions. This final report covering the full 16 months of the pilot corroborates these findings. GPS tagging in this instance does not appear to have an impact on compliance.

For the purpose of the pilot, non-compliance is defined as being ‘out of contact’; however, data also includes information about reporting compliance, tag compliance and whether asylum claimants are officially absconders. The list below sets out different forms of compliance.

List of definitions of compliance used in this pilot

Out of contact (for control and tagged group)
Behaviours indicating non-compliance differ across the control and tagged groups due to the different monitoring modes (electronic versus traditional). For the control group, non-compliance is equivalent to lack of reporting, for the tagged group it is lack of reporting and non-compliance with the tag. The GPS project team determined contact status on a case-by-case basis.

Tag compliance (for tagged group only)
This is determined by the occurrence of tag breaches, which can take place in multiple forms; for the purpose of this evaluation and based on data availability we focused on battery and strap tampers only. A strap tamper breach could be pulling out or cutting a tag with scissors, while a battery breach takes place when the tag has not been recharged.

Reporting compliance (for control and tagged group)
A person is considered non-compliant with reporting when they miss 2 or more consecutive in-person reporting events when reporting fortnightly.

Absconder (for control and tagged group)
Those who are considered out of contact are not necessarily official absconders, but they might become absconders. Further action may be taken to bring them back into compliance. Where that is not effective, the individual will be considered an absconder and the Home Office will stop funding the tag provided by the contractor, that means the order will be ceased. This process might take up to 2 months. Data on absconders was imputed by caseworkers or Immigration Enforcement (IE) officers in Atlas and not directly determined by the GPS project team. See Annex B – Glossary for the department’s definition of absconder.

The objective of this evaluation was to answer the following research questions:

1. Process: how was EM implemented?​ How efficient was the process?
2. Compliance: did EM improve rates of compliant behaviour amongst asylum claimants subject to immigration bail?
3. Compliance over time: how was contact maintained over time?
4. Drivers: which factors were associated with compliance?​
5. Outcomes: was there a relationship between electronic monitoring and asylum/removal outcomes?

To produce this report, HOAI collaborated with the GPS expansion pilot project team to collect relevant information:

• datasets containing numerical and categorical data for all individuals in the control and tagged groups (1,200) who joined the pilot from June 2023 to the beginning of October 2023. Two months of extra compliance information for cases enrolled in September and October 2023 were recorded to fully understand their behaviour and ensure robustness of findings.^{[footnote 6]} Information was sourced from Home Office’s Case Management Systems (Atlas and CID) and provided by the GPS supplier (Electronic Monitoring Services, EMS)

• case-working notes describing asylum claimants’ experiences, legal challenges and outcomes. This information was analysed, restructured and presented as case studies (Person A to E) throughout the report. Additional qualitative evidence was gathered through interactions with operational staff

The data presented in this report has some limitations affecting the robustness of the findings. These limitations should be considered when interpreting the results:

• the sample is not randomised: this is due to the selection process and the characteristics of the small boat arrivals:

  • selection process: detained asylum seekers who received bail were referred to the pilot; after referral they were allocated to the tagged group through an assessment based on various criteria including medical conditions. If they fulfilled all the assessment criteria and they were willing to participate, they became part of the tagged group. If they refused to be tagged or did not meet the assessment criteria, they became part of the control group, together with other individuals on traditional reporting

  • small boat arrivals: there is a high proportion of Albanian nationals in the sample, who at the time featured more prominently in the small boat arrivals (88% in the tagged group and 68% in the control). This high proportion of Albanian nationals reflects the timing of the pilot, which drew data when Albanian arrivals were at their peak. This trend changed from quarter 3 of 2023 where Afghan nationals featured most prominently, see Figure 4 in Irregular migration to the UK, year ending September 2023 - GOV.UK (www.gov.uk). This means results of this report cannot be generalised as we do not have enough information about the behaviours of asylum claimants from other nationalities. There was also a high prevalence of male individuals (99%) and the top age groups, covering two-thirds of the whole sample, were aged 18 to 24 years (34%) and aged 25 to 29 years (30%)

• inconsistent report frequency: 16% of individuals across tagged and control groups appeared to report at non-standard frequencies. This means individuals went through different experience which are not fully comparable. Reporting frequency is decided by SOS or IJ when determining the bail conditions, this can be changed later on by HO officials. Changes on mode and frequency of reporting are based on specific individual circumstances (for example, medical conditions, individuals selected for imminent relocation). For consistency, data of individuals reporting fortnightly were filtered to conduct statistical tests presented in this report

• missing information on asylum and removal outcomes: there was limited information about the migration and asylum outcomes because several individuals did not receive an asylum result by the end of the pilot. Therefore, it was not fully possible to investigate relationships between compliance and migration outcomes

2. Overview of the process

2.1 Key findings

Process: how was EM implemented? How efficient was the process?

• setting up reporting and installing a tag were not always efficient processes. It took 60 days on average to set up an individual on traditional reporting, both in the control and tagged group, and 166 asylum claimants refused to be tagged on the day of release
• the majority of individuals across both groups underwent fortnightly traditional face-to-face reporting (1,011 out of 1,200, 84%). To enable like with like comparisons, only individuals who reported fortnightly were considered when analysing information about compliance
• the tagged and control groups have similar demographic characteristics. There is a high prevalence of young, male and Albanian individuals
• the pilot saw a high number of legal processes and challenges, which are interconnected with compliance and tagging, and therefore had an impact on delivery

2.2 Enrolment

Over the course of the pilot, approximately 18,000 asylum claimants were released on bail from detention^{[footnote 7]}, 990 of these were referred to the pilot. Before release, 390 refused to participate in the pilot or were not suitable based on assessment (Annex A and therefore became part of the control group (600), together with another 210 people undergoing traditional reporting. Six hundred were enrolled and tagged on release from detention (Figure 1).

Figure 1: Referral and enrolment process

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Flowchart showing 7 boxes highlighting various groups of individuals going through the referral and enrolment process.

Figure 2 shows that it was quicker and easier to assign a person to the control group after release than setting up the person as part of the tagged group. The control group reached a total of 600 individuals in July 2023, whereas the tagged group reached this total in October 2023.

Figure 2: Cumulative amount of people in the pilot (tagged versus control group)

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Line chart showing the cumulative number of individuals in the pilot covering the period from March 2022 to November 2023.

The first 3 months (from June to August 2022) and summer 2023 (from June to August 2023) were periods which saw peaks in number of people tagged on a weekly basis as part of the pilot (see Figure 3).^{[footnote 8]}

Figure 3: Number of people tagged weekly March 2022 - November 2023

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Bar chart showing the number of people tagged by week from March 2022 to November 2023. One case was tagged in March 2023 because Immigration Judges (IJ) had power to tag people before the start of the pilot.

Setting up reporting and setting up a tag for asylum claimants were 2 different processes having different timeframes. Setting up a tag in the tagged group was immediate, a tag was set up on the day of release from detention, setting up traditional reporting was slower, it took 60 days on average to set up an individual on traditional reporting over the course of the pilot. This long timeframe implied a high-risk of absconding.

It is, however, important to note that the process of setting up reporting was overall slow because at the start of the pilot, tagged individuals were not required to be on traditional reporting. The project team worked on implementing reporting for tagged individuals and this overall improved contact management for individuals who did not have contact management before.

However, tagging was not always an efficient process, 166 asylum claimants refused to be tagged on the day of release, even if they agreed to the bail conditions. These instances caused additional paperwork and disruption in detention. After tag refusal, 150 became part of the control group, but 16 were tagged and continued to be part of the tagged group. One individual after their tag appointment decided to stay in detention rather than being release on bail with a tag.^{[footnote 9]}

2.3 Reporting frequency

The majority of the individuals across both groups underwent fortnightly traditional face-to-face reporting (1,011 out of 1,200, 84%), with the remainder subject to weekly (5%), monthly (9%) or ad-hoc reporting (Table 1). In addition, 117 (19%) individuals in the control group and 11 (2%) in the tagged group underwent digital reporting instead of in-person reporting.

To enable like with like comparisons, only individuals who have been listed as reporting fortnightly have been considered when analysing information about compliance in this report. This reduces the total number of people in the tagged cohort to 502 and the control cohort to 509.

Table 1: Frequency of reporting in the tagged and control groups

Frequency	Tagged group (%)	Control group (%)	Total (n)	Percentage
Fortnightly	84%	85%	1,011	84%
Weekly	7%	2%	59	5%
Monthly	8%	11%	112	9%
Quarterly	0%	0%	2	0%
Ad-hoc	0%	1%	4	0%
N/A or unknown	1%	1%	12	1%
Total	100%	100%	1,200	**100%

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Percentages in the tables throughout the report do not add up to 100% due to independent rounding.

2.4 Characteristics

The individuals who participated in the pilot, or were part of the control group, were members of 4 different cohorts:

• DAC PRE: individuals in DAC (Detained Asylum Casework) process pre-asylum decision^{[footnote 10]}
• DAC POST: individuals in DAC (Detained Asylum Casework) process post-asylum decision
• TCU: individuals referred to the Third Country Unit, who are eligible for the MEDP (Migration and Economic Development Partnership) scheme with Rwanda or other third countries schemes. These individuals had received a Notice of Intent (NOI) stating that their claim is inadmissible, and they are eligible to be relocated to another third safe country
• NRPC: individuals in the National Returns Preparation Command (NRPC) cohort, who were not accepted into DAC due to capacity constraints or inability to access legal aid. This cohort was added in March 2023, due to the lower utilisation of tags than anticipated. Information about this cohort includes historical data taken from PRAU (Performance Reporting and Analysis Unit) for the control group

Only 52 individuals were part of the DAC POST group (4%), 309 were part of the DAC PRE (26%), 265 were part of the NRPC group (22%) and a majority (48%, 574) were part of TCU. Figure 4 shows the number of individuals across cohorts broken down along the control and tagged groups. This disproportional distributions across tagged and control groups render comparison across cohorts more challenging.

Figure 4: Number of individuals across groups in tagged and control groups

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Bar chart highlighting number of individuals belonging to specific cohorts (DAC, NRPC and TCU) across tagged and control groups.

A higher proportion of individuals were subject to Secretary of State (SoS) bail in the tagged group in comparison to the control group. Of those in the tagged group, nearly a third (206, 34%) were granted bail by an Immigration Judge (IJ bail) and two-thirds (394, 66%) by the Secretary of State (SoS bail). The control group is more balanced, 332 individuals (55%) were granted IJ bail and 268 (45%) SoS bail.

The tagged and control groups have similar demographic characteristics. There is a high prevalence of male individuals (99%). Top age groups, covering two-thirds of the whole sample, were aged 18 to 24 years (34%) and 25 to 29 years (30%). Albanian nationals featured as the most common nationality among the tagged group (88%) and control group (68%). However, this high proportion of Albanian nationals tends to reflect the timing of the pilot, which drew data when Albanian arrivals were at their peak.

The behaviour of Albanian nationals cannot be generalised to other nationalities, the findings in this evaluation do not include evidence showing how other nationalities behaved or would have behaved similarly or differently from Albanian nationals, there is not enough data for proving behavioural patterns (including compliance or non-compliance) in other nationalities.

Across both groups 58% of the individuals were identified as Adults at Risk (AAR) levels 1, 2 or 3 when in detention (see Annex B - Glossary for definitions), with the highest proportion of individuals (143, 24%) at level 2 in the tagged group as well as in the control group (243, 40%):

Table 2: Adults at risk levels across tagged and control groups

AAR Level	Tagged group (%)	Control group (%)	Total (n)	Percentage
Level 1	21%	19%	243	20%
Level 2	24%	40%	386	32%
Level 3	3%	8%	65	5%
N/A	52%	33%	506	42%
Total	100%	100%	1,200	100%

2.5 The process and legal challenges

The asylum claimants on the pilot are individuals navigating legal processes with their legal representatives and this implies travelling through the system through different steps.

Person A – the process

Person A, an asylum claimant who had arrived in 2022, were served a Notice of Intent to be relocated to Rwanda, but then was referred to the National Referral Mechanism (NRM) and received a Positive Reasonable Grounds decision.

They were subsequently granted Immigration Judge (IJ) bail and were released with a tag. Whilst in the community, 11 weeks after the initial referral, they received a Positive Conclusive Grounds decision, and a Discretionary Leave (DL) application to stay in the UK was opened.

Person A forgot to charge their tag device once and missed a reporting appointment because they mixed up dates, but they otherwise went through the asylum application process and were compliant with the tag and reporting throughout the 16 months of the pilot.

Person A’s story shows that asylum claimants on this pilot went through various legal processes, including:

• applications for bail (IJ and SOS)
• NRM (National Referral Mechanism) applications for modern slavery cases (see Annex B – Glossary for definitions)
• asylum applications or other immigration applications to stay in the UK
• removal and return processes, such as voluntary returns, NOI and relocations
• legal challenges against the pilot mainly based on medical grounds

These legal processes had an impact on operational delivery and were interconnected with compliance, as Figure 5 shows. They could result in either negative or positive outcomes for the claimants and/or the Home Office. This means they could either challenge the system and be a barrier to tagging or facilitate and speed up the process of people through the system.

Figure 5: Overview of the process and legal steps

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Flowchart describing the arrival, asylum and contact management process experienced by asylum seekers.

3. Compliance

3.1 Key findings

Compliance: did EM improve rates of compliant behaviour amongst asylum claimants subject to immigration bail?

Compliance over time: how was contact maintained over time?

Notwithstanding the limitations of the sample, EM did not appear to affect compliance of individuals on bail conditions as part of this pilot. Compliance rates dropped following a similar trend across tagged and control groups; however, individuals in the tagged group tended to go out of contact more quickly than in the control group. Use of an EM tag did not appear to be a barrier to absconding.

• out of contact: at the end of the pilot, there was no statistically significant difference between the proportions of individuals in contact or out of contact in each group, 66% of them was out of contact in both the control and tagged groups
• there was a statistically significant association between tag compliance and reporting compliance in the tagged group, where individuals were more likely to be compliant with either both or neither, as opposed to be compliant with only one of the methods. This suggests that tagging was not effective as a method to improve contact with individuals less likely to be compliant with reporting, as the same individuals who are not compliant with one method are likely to not be compliant with the other
• reporting compliance: there was not a statistically significant difference between reporting compliance in the tagged and control groups. Overall, at the end of the pilot only 32% (162 out of 502) of the tagged group were compliant with reporting, and 34% (174 out of 509) of the control group were compliant
• tag compliance: only 146 out of 502 (29%) people were fully compliant with the tag, while the majority (356, 71%) had at least one battery breach or strap tamper recorded. There was a total incidence of 1,142 battery breaches and 265 strap tampers
• absconding: at the end of the pilot, there was no statistically significant difference in absconding between tagged and control groups, with 14% of the control group and 16% of the tagged group marked as absconders

3.2 Contact status

The majority of people in both groups (664 out of 1,011, 66%) at some point went out of contact by missing consecutive in-person reporting requirements and/or (where applicable) having tag breaches rendering the device out of operation for a period of time (for example, by not recharging once the battery was out of power).

For the control group, the ‘out of contact’ measure is the same as reporting compliance, but for the tagged cohort it is a combination of reporting and tag compliance. Individuals out of contact were not immediately considered to be absconders (see Introduction and Annex B – Glossary for full list of definitions).

At the end of the pilot, there was no statistically significant difference between the proportions of individuals in contact or out of contact in each group, 66% of individuals were out of contact in both the control and tagged group (Table 3). Electronic monitoring did not appear to affect compliance of individuals on bail conditions in this pilot.

Table 3: Number of people who are out of contact at the end of the pilot

	Tagged Group (%)	Control Group (%)	Total (n)	Total (%)
Out of Contact	66%	66%	664	66%
In Contact	34%	34%	347	34%
Total	100%	100%	1,011	100%

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

In the tagged group, contact status was determined by elements of 2 methods of compliance: tag compliance and reporting compliance. A tagged person was out of contact if they were not reporting and were non-compliant with the tag. In this group, there is a statistically significant association (Logistic regression: p<0.001***, coefficient = 3.674) between tag compliance and reporting compliance, with individuals more likely to be compliant with either both or neither, as opposed to be compliant with only one of the methods.

Out of the 502 in the tagged cohort and on fortnightly reporting, 121 people were compliant with both methods, and 307 were compliant with neither, at the end of the pilot. This left only 23 individuals compliant with tag but not reporting, and 41 with reporting but not the tag (Table 4).

This suggests that, for this pilot’s sample, tagging was ineffective as a method to improve contact with individuals less likely to be compliant with reporting, as the same individuals who were not compliant with one method were likely not compliant with the other.

Table 4: Relationship between reporting compliance and tag compliance

Reporting compliance	Tag compliance		Total
	No (not compliant with tag)	Yes (compliant with tag)
No (not compliant with reporting)	307 (88%)	23 (16%)	330 (67%)
Yes (compliant with reporting)	41 (12%)	121 (84%)	162 (33%)
Total	348 (100%)	144 (100%)	492 (100%)

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Ten cases who had no information on reporting compliance were not included in this table and the analysis.

3.3 Contact status over time

Individuals in the tagged group tended to go out of contact more quickly than in the control group. After the first week of the pilot, 6% of individuals in the control group were non-compliant, compared to 29% of the tagged group. Individuals went out of contact after 53 days on average in the tagged group, whereas in the control group, individuals went out of contact after 116 days on average (mean numbers).

This difference likely reflects the point at which non-compliance is detected - as this will be immediate for tagged people whereas for untagged people it will only become apparent when they miss 2 or more consecutive reporting events. In any case, it appears that tagged individuals went out of contact more quickly than people reporting face-to-face.

After 100 days, 47% of the tagged group were still in contact, while 67% of the control group were still in contact. These proportions, which are also presented in Figure 6, take into consideration that tagged individuals went out of contact but came back into contact throughout the pilot, such as a tag recharged after battery breach.

This means that even if compliance rates dropped following a similar trend across tagged and control groups, the tagged group was more unstable (individuals were going out of contact and were brought back into contact by EMS or Home Office more frequently) than the control group. This is demonstrated by evidence on breaches (see Tag compliance section).

Over the first 300 days, compliance rates dropped in a relatively stable way for both groups. Over the longer term, at the end of the pilot, compliance rates reached the same level across the control and tagged group, in both groups around a third of the sample were still in contact (34%).

Figure 6: Proportion of individuals maintaining contact over time, by tagged status

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Line chart showing the proportion of individuals maintaining contact over time (number of days). The analysis on compliance over time covered a smaller number of individuals due to missing data on dates.

3.4 Reporting compliance

Individuals in both the control and the tagged cohort were required to report face-to-face, by visiting a ROM at a set frequency. For 84% of the overall sample, this was fortnightly, while 5% were reporting weekly and 9% monthly. To enable comparisons, compliance analysis only assessed the individuals on fortnightly reporting.

There was no statistically significant difference between reporting compliance in the tagged and control groups.^{[footnote 11]} Overall, at the end of the pilot only 32% (162 out of 502) of the tagged group were compliant with reporting, compared with 34% (174 out of 509) of the control group (Table 5).

Table 5: Number of people who were compliant with face-to-face reporting

Reporting compliance	Tagged Group (%)	Control Group (%)	Total (n)	Total (%)
Yes (reporting compliant)	32%	34%	336	33%
No (not reporting compliant	66%	66%	665	66%
Unknown	0%	0	1	0%
N/A	2%	0	9	2%
Total	100%	100%	1,011	100%

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Timeframes around reporting non-compliance appeared to differ across the 2 groups. On average, it took 201 days for a person to become non-compliant with reporting, in the tagged group this process was much quicker than in the control group, only 136 days on average against 269 (mean numbers).

3.5 Tag compliance

The individuals in the tagged group were required to comply with their EM tag, this means to continue to charge and wear it. Only 146 out of 502 (29%) people were fully compliant with the tag, while the majority (356, 71%) had at least one battery breach or strap tamper recorded. Out of the 173 tagged people still in contact at the end of the pilot, 82 committed at least a breach throughout the pilot (47%). This means they were not compliant with their tag but continued to report face-to-face or came back into contact at a later stage.

Battery breaches occurred when the tag ran out of battery and was not charged on time – the guidance recommended charging once every 24 hours. Strap tampering took place when the ankle strap of the EM tag was tampered with, often by pulling or cutting with a pair of scissors. In the instances where the individual successfully broke or removed the tag, EMS would send a team to the recorded address to attempt to retag the individual.

Overall, 52 (10%) individuals had reported both battery and strap breaches, 120 (23%) had reported battery breaches only, 186 (37%) had strap breaches only, and 123 recorded no breaches (24%) (Figure 7). There was a total incidence of 1,142 battery breaches and 265 strap tampers.

Figure 7: Number of people tagged by breach type

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Bar chart showing the number of individuals committing different breaches, for example, strap tamper and battery breaches.

The greatest number of battery breaches recorded for a single individual was 81, this was clearly an outlier since the average (mean) number of breaches for an individual was 6.3. The greatest number of strap tampers committed by one person was 3, and the average (mean) was one.

A greater proportion of people had multiple battery breaches than multiple strap tampers: 224 of the 243 (92%) people with strap tampers had only one incident, compared to 74 out of 181 (40%) people with battery breaches who had only one incident.

It is worth noting that battery breaches were often accidental, the project team explained that participants complained that the battery was not lasting for as long as advised, experienced difficulties with the charging cable not being long enough to sit comfortably during charging and problems scheduling their lives around charging the device and the stigma attached. On the other hand, strap tampers were more likely to be intentional and could not be resolved by the individual, instead requiring the EMS team to conduct maintenance in person.

Additionally, the majority (316 out of 502, 64%) of people reporting fortnightly had their tag ceased throughout the pilot. Some of these tags were ceased due to the individual having absconded – and removing the tag in the process – whereas others were a result of legal or medical challenges (Figure 8).

Figure 8: Number of tags ceased by reasons

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Notes:

1. Bar chart showing the number of individuals ceasing their tag based on reason, for example, voluntary departure.

Information shared by the project team indicated that legal challenges had questioned the proportionality of the use of electronic monitoring for individuals who had demonstrated good behaviour and compliance with reporting, and so could be considered at low risk of absconding. Medical challenges were more varied, as issues arose concerning things such as skin irritation, constricted blood flow, impact on mental health. Person B’s story is an example of a medical challenge.

Person B – mental health challenge

During Person B’s induction in detention, they were offered voluntary return, which they refused. No health issues were reported, but they were reported to have missed their asylum interview due to illness.

They were refused Immigration Judge (IJ) bail as the judge was content that the asylum claim would be decided in a reasonable timeframe. After one month in detention, they were referred to the National Referral Mechanism (NRM) and released on Secretary of State (SOS) bail with electronic monitoring as a condition. They received a Negative Reasonable Ground decision and were also assessed as an Adult at Risk Level 1.

A month after release, Person B was admitted to a mental health facility and their solicitor requested the removal of the EM tag as they had been admitted for a long-term mental health condition which could worsen by the presence of the tag. Person B’s solicitor referred to a letter that recognised that a GP had identified scars consistent with torture on the individual.

Person B’s tag was subsequently removed.

The project team also discovered that a few medical challenges were fraudulent. Some individuals removed the tag themselves but provided falsified medical documents to claim that the tag was removed by a medical professional (see Person C’s story). While EMS would attempt to retag these individuals, it was often considered difficult to do so as the person could refuse entry to their home or not be present at the agreed time.

Person C – fraudulent documents

Person C was encountered during an Immigration Enforcement visit and placed into detention. Three days later they claimed asylum, received SOS bail, and declined the option of voluntary departure. Person C had a young daughter living in the UK, and requested to be released without tag because of their family circumstances. They were released from detention after 10 days wearing a tag.

Two weeks later they attended an asylum interview, but soon after a report of strap tampering was received as they had attempted to remove the tag. Electronic Monitoring Services (EMS) made a visit to attempt to retag Person C, but they received Withdrawal of Consent from the property owner, preventing them from entering the property.

The EMS team were informed by the Reporting and Offender Management team that Person C had provided a screenshot of their mobile phone that showed a letter from a hospital explaining that the tag had been removed as it was too tight. Contact with the hospital revealed the medical letter to be false, and Person C was eventually retagged following a number of unsuccessful attempts.

During the process, Person C remained fully compliant with their reporting conditions.

3.6 Absconders

When individuals went out of contact the operational team attempted to re-establish contact before declaring them ‘absconders’, a process which could take a couple of months (see Annex B – Glossary). It took on average 219 days for individuals to be officially declared absconders, this was similar across both the control and tagged groups (211 days in the tagged group and 227 in the control group).

Out of those on fortnightly reporting, 153 absconded (23% of people who were out of contact). At the end of the pilot, there was no statistically significant difference in absconding between the tagged and control groups, with 14% of the control group and 16% of the tagged group marked as absconders.

Table 6: Number of people recorded as having absconded

	Tagged Group (%)	Control Group (%)	Total (n)	Total (%)
Absconded	16%	14%	153	15%
Not absconded	67%	86%	771	76%
N/A	17%	0%	87	9%
Total	100%	100%	1,011	100%

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Out of the fortnightly subset, 87 people were marked as ‘N/A’ for absconding, all of whom are in the tagged group. It is unclear why, as the individuals at the end of the pilots had a range of characteristics and were at different stages in the asylum process: 47 were still awaiting interview, 14 were refused asylum but were still in the country, 7 withdrawn their applications, 5 voluntarily departed, 17 were marked as being out of contact and 9 had the use of their tag ceased.

Out of the 80 absconders in the tagged group, the use of the tag was ceased because of them absconding in 16 cases. In 7 instances, the absconding date and the date of tag ceasing is the same, so we could not determine in which order this happened. The majority of people from the tagged group were declared official absconders after the use of their tag had ceased (52).

For 27% (out of 52) this happened after the first month after tagging ended, 17% after the second month, and 12% after the third – the remaining 45% became absconders more than 3 months after the tag was ceased. Five people were marked as absconders but had not yet had their tags ceased.^{[footnote 12]}

Use of an EM tag did not appear to be a barrier to absconding. The proportion of individuals who had their tag ceased and absconded (52 out of 326, 16%) was similar to the overall rate of absconding across both groups (15%), and so there is no evidence individuals were more likely to abscond once tagging ceased or sought tagging to be ended with the intention of absconding.

4. Drivers of non-compliance

4.1 Key findings

Drivers: which factors were associated with compliance?

Outcomes: was there a relationship between electronic monitoring and asylum/removal outcomes?

Each person who enters the UK has different experiences as part of their journey and asylum process. As such, everyone will have different resources available and different motivations to comply or abscond, so it is difficult to determine what the main drivers of non-compliance are. Regression analysis presented in this section identified some possible relationships between variables; however, due to the non-random nature of the sample, it is not possible to predict with certainty future behaviours and generalise beyond people included in this pilot:

• nationality: there appeared to be an association between Albanian nationals in this sample and being out of contact, where Albanian nationals in the pilot were 2.34 times as likely to go out of contact than non-Albanian nationals. However, this does not indicate that any or all other nationalities are more likely to be in contact, due to the disproportionately small numbers of other nationalities included in this pilot preventing any further analysis. This finding should be treated with caution
• group: being part of the tagged or control group did not appear to have any impact on contact status or reporting compliance, but the tagged cohort were 2.7 times as likely to abscond than the control
• asylum support: people receiving asylum support were 79% less likely to go out of contact compared to those who were not receiving support
• asylum accommodation: if they had been offered asylum accommodation but were not living there, they were 3.96 times as likely to go out of contact; however, if they were living in asylum accommodation, they were only 39% more likely to go out of contact compared to those who were not given accommodation
• migration outcome: anecdotal evidence provided by Home Office’s staff and case studies signalled that if an asylum claimant believed in the success of its application and was at low risk of removal, they tended to remain compliant and in contact with the Home Office

4.2 Key factors

Various characteristics were recorded and considered as potential drivers of non-compliance.^{[footnote 13]} A logistic regression of these characteristics and their association with contact status identified 6 significant factors behind the likelihood of someone going out of contact (see Model 1 Annex C).

Compared to all other nationalities taken together, Albanian nationals included in this pilot were twice as likely to go out of contact (2.349). Whilst this could indicate an association between being an Albanian national and being out of contact, it does not however indicate that any or all other nationalities are more likely to be in contact than Albanian nationals.

It is important to note that Albanian nationals were the predominant nationality in the sample (88% in the tagged and 68% in the control group), and there was insufficient information about behaviours of other nationalities to fully test assumptions around their compliance behaviours. Table 7 shows a breakdown of contact status and the proportions of Albanian nationals across the control and tagged groups.

Table 7: Proportion of Albanian people out of contact for the tagged and control group

	Albanian (tagged)	All Other (tagged)	Albanian (control)	All other (control)	Total
Out of Contact	301 (66%)	28 (64%)	284 (78%)	51 (34%)	664 (65%)
Still in Contact	157 (34%)	16 (36%)	79 (22%)	95 (66%)	347 (35%)
Total	458 (100%)	44 (100%)	363 (100%)	146 (100%)	1,011 (100%)

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

People receiving asylum support were 79% less likely to go out of contact compared to those who were not receiving support.

If they had been offered asylum accommodation but were not living there, they were almost 4 times as likely to go out of contact, however if they were living in asylum accommodation, they were only 39% more likely to go out of contact compared to those who were not given accommodation.

This regression found that when controlling for all other statistically significant factors, the tagged cohort were 43% less likely to go out of contact. As we know the proportion of people who went out of contact is approximately equal across the 2 cohorts, we can determine that other factors may be more significant drivers of non-compliance than whether an individual is tagged.

Bail route, NRM referral and Adult at Risk level were not found to be statistically significant. Sex could not be analysed due to the sample composition (99% of individuals were male).

Table 8 and 9 shows breakdown of contact status across cohorts and bail routes, there was no significant difference in proportions between the various groups.

Table 8: Proportion of people across different cohorts who were out of contact for the tagged and control group

	DAC (tagged)	NRPC (tagged)	TCU (tagged)	DAC (control)	NRPC (control)	TCU (control)	Total
Out of Contact	140 (67%)	83 (52%)	106 (78%)	90 (84%)	44 (51%)	201 (63%)	664 (64%)
Still in Contact	67 (33%)	75 (48%)	31 (23%)	16 (16%)	41 (49%)	117 (37%)	347 (36%)
Total	207 (100%)	159 (100%)	136 (100%)	106 (100%)	85 (100%)	318 (100%)	1,011 (100%)

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Table 9: Proportion of people out of contact across bail routes for the tagged and control group

	SOS Bail (tagged)	IJ Bail (tagged)	SOS Bail (control)	IJ Bail (control)	Total
Out of Contact	228 (66%)	101 (63%)	159 (70%)	176 (61%)	664 (64%)
Still in Contact	113 (44%)	60 (37%)	65 (30%)	109 (39%)	347 (35%)
Total	341 (100%)	161 (100%)	224 (100%)	285 (100%)	1,011 (100%)

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

A similar model was run except with reporting compliance as the outcome instead of out of contact (see Model 2 in Annex C). The same variables were statistically significant, and the direction of affect (increase or decrease in compliance) was consistent. The difference in magnitude is relatively minor.

The model was also run with absconding as the outcome (see Model 3 in Annex C), though only asylum support, given accommodation, and being tagged were found to be statistically significant factors. People were 67% less likely to abscond if they were receiving asylum support but were twice as likely to abscond if they had been offered accommodation but were not living in it. This model found that, controlling for all other factors, the tagged cohort were 2.5 times more likely to abscond than the control.

The same model was then re-run with data on the control and tagged groups separately. This enables identifying whether the impact of set characteristics on compliance is similar across both groups.

In the control subset, all factors remained statistically significant apart from whether an individual was living in provided accommodation. Each factor was also in the same direction as the overall model (whether it was associated with being more likely or less likely to be out of contact), but to different magnitudes.

In contrast, for the tagged subset only whether they had been given asylum accommodation was a statistically significant indicator of contact. This could suggest that individual’s responses to being tagged are less driven by their characteristics and more by their personal experience, making it more difficult to predict if specific cohorts will be compliant with the conditions.

Each factor was also analysed separately for the control and tagged groups. Additional findings (Annex D) contain further results of regression analysis on various factors influencing compliance.

4.3 Migration outcomes

One of the key questions of the evaluation was to try to understand whether there is a relationship between asylum/removal outcomes and compliance across both groups. However, information provided do not give us a clear picture of the relationships between these variables, for 2 main reasons:

• more than half of the asylum claimants (652, 54%) did not receive a result of their asylum claim at the end of the pilot, they were waiting for their interview appointment or interview result (Table 10). It took up to 512 days for some individuals to have an interview and they received 108 days of legal representation on average

• there was a high number of potential outcomes for the asylum claimants. This means that statistical tests to measure relationships could not be performed on these categorical variables, or when performed did not show any relationships

Table 10: Asylum/removal outcomes across tagged and control group

Outcomes	Tagged group (%)	Control group (%)	Total (n)	Percentage
Awaiting decision	12%	11%	133	11%
Awaiting interview	47%	40%	519	43%
Granted	0%	3%	19	2%
Refused (still in the country)	15%	13%	171	14%
Removed	2%	0%	15	3%
Voluntary departure	5%	4%	52	5%
Withdrawn	12%	18%	180	15%
Other (N/A)	7%	12%	111	9%
Total	100%	100%	1,200	100%

Source: Dataset produced by GPS Expansion Pilot Project Team (Home Office 2023).

Once an asylum claimant was removed or received leave to remain, the person stopped reporting and if the person was wearing a tag, their tag was ceased.

Even if numerical evidence does not show associations between variables, anecdotal evidence provided by Home Office staff signalled that if an asylum claimant believed in the success of its application and was at low risk of removal, they would remain compliant and in contact with the Home Office, either with traditional reporting, the tag or both. An example of this is Person D.

Person D – compliant refugee

Person D arrived in the UK seeking asylum and submitted a modern slavery claim. They received an NOI letter and were offered voluntary departure, which they refused.

In detention, Person D attempted self-harm, was treated by staff and an Assessment Care in Detention and Teamwork (ACDT) plan was opened, but closed after 10 days as the person did not have further intentions to self-harm. At this stage, they also became part of the MEDP cohort and they were accepted for relocation to Rwanda.

They then submitted a Human Rights claim, 10 days later they were granted Immigration Judge (IJ) bail and then released with a tag after 5 days. Legal representatives asked for removal of the tag because of post-traumatic stress disorder (PTSD) symptoms due to trauma from torture they had previously experienced in their home country. The Home Office made the decision to continue using EM tagging, due to a high risk of absconding, as the individual lacked compliance with traditional reporting.

Sixteen months after claiming asylum Person D’s claim was granted. Their tag was ceased. They had been fully compliant with the tag and registered no breaches.

Some of the individuals on the pilot underwent completely different experiences than Person D, their outcomes were negative, and they were not compliant. Person E is an example of an asylum claimant who did not receive asylum and subsequently absconded.

Person E – absconder irregular migrant

Person E arrived irregularly in the UK seeking asylum. The initial asylum interview did not find any known mitigating circumstances or family ties which would aid their case.

They were granted SOS (Secretary of State) bail a few weeks later as the asylum decision was still outstanding. At this point, Person E refused EM (electronic monitoring) as they were planning to live in a property with small children and felt that the use of the tag would elevate stress. The decision to use EM tagging was maintained because of the lack of medical evidence supporting this claim.

Two months later, Person E’s substantive asylum interview took place. A month after this, asylum was refused on the basis that fear of return was unfounded. Ten days later, data from the GPS tag began to be processed to locate Person E and Immigration Compliance Enforcement (ICE) were tasked to pick them up for removal.

Person E submitted a claim to the National Referral Mechanism. The Home Office sent correspondence to say that the individual had missed all reporting events and reminded them that they were still subject to EM conditions. A Negative Reasonable Grounds Decision was completed 2 weeks after Person E raised the claim.

The next week, a strap tamper was registered but the Home Office could not conduct a visit as Person E’s location was unknown. New bail reporting instructions were sent but later the Home Office discovered that person E had vacated their accommodation, leaving behind the cut-off EM tag, SIM card and phone, and the staff had no means to locate or contact them.

Person E was recorded as an absconder. They were monitored for just short of 6 months and did not attend any of their reporting events.

Person D and E are 2 examples, showing how compliance align with migration outcomes. However, due to low numbers, it is not possible to generalise and identify specific trends.

5. Annexes

Annex A: GPS pilot suitability assessment

Name:

Date of birth:

Nationality:

Home Office Reference:

Consideration of individual circumstances and Convention Rights

In this section you must set out the facts of the case and the claimant’s individual circumstances. You must give consideration as to whether the application of EM would infringe the claimant’s Convention Rights. Factors which you must take into account include:

• strong independent medical evidence to suggest that an EM condition would cause serious harm to the person’s mental or physical health
• a claim of torture been accepted by the Home Office or a Court
• a positive conclusive grounds decision in respect of a claim to be a victim of modern slavery
• whether the person’s mental capacity is deemed to be a bar to understanding the EM conditions and therefore their ability to comply for example, a person suffering with dementia
• the individual is suffering with phlebitis or similar conditions which cause swelling of the lower legs
• disability, injury, frailty or age-related conditions which may impact on the person’s ability to wear and/or maintain the device
• any evidence of suicide risk

Initial triage

• question 1 - method of entry
• question 2 - age (as at today’s date)
• question 3 - is the claimant’s age subject to age dispute processes (including awaiting a Merton compliant age assessment)? Yes/No
• question 4 - as the claimant ever been detained under sections 37 or 41 of the Mental Health Act 1983? Yes/No
• question 5 - if question 4 is yes, does the claimant remain under a supervision order? Yes/No
• question 6 - will the claimant reside at an address in Scotland or Northern Ireland? Yes/No

Where the individual initially arrived in the UK via a legal route, is aged 17 and under or the answer to any of questions 3 to 6 is ‘yes’, EM is not appropriate, and no further action should be taken.

This is not an exhaustive list. Where any of those factors exist, it is unlikely to be appropriate to continue with EM without a very strong rationale for doing so. Where you consider that EM remains appropriate, you need to set out the rationale for your decision. Assistant Director authority will be required in such circumstances.

Where the claimant has made representations, you must also specifically address each ground upon which the claimant relies.

Bail conditions
You must liaise with DAC/NRC colleagues to determine appropriate bail conditions. The checklist below will be used by the EM Hub to set those conditions as part of the release process.

Residence conditions
Most individuals are likely to require accommodation through the asylum support system although some may make their own accommodation arrangements (such as living with family members). Please include details of the address here.

Reporting conditions
There may some individuals who, as well as being subject to EM, may also have a condition to report to a ROM or police station. Where this is the case, please include details of the reporting location and the time and date of reporting, including frequency for recurring events. For example, reporting to Lunar House ROM between 10:00hrs and 13:00hrs on Wednesday XXdateXX and each Wednesday thereafter.

Supplementary conditions
Those subject to EM can be required to meet some supplementary conditions. There are 3 main types of supplementary conditions:

• curfews
• inclusion zones
• exclusion zones

The use of supplementary conditions within the pilot will be restricted to exceptional circumstances, generally involving criminality or national security considerations. You should liaise with relevant colleagues in the Home Office should such circumstances arise, and it may be that alternative provisions (such as more frequent reporting, or EM in accordance with the Statutory Duty) are more appropriate.

In any cases in which a supplementary condition is considered appropriate, you must be able to justify the use of the supplementary condition in relation to the risk of harm, reoffending and/or absconding and/or any previous offending pattern. You should clearly set out those justifications (including responding to any representations received from the claimant) here and seek approval from a Deputy Director before any supplementary condition can be assigned.

Annex B: Glossary

Legal definition of absconder

A person can only be an absconder if they have both:

• previously been served with one of the following forms

• BAIL 201- notification of immigration bail to a person detained or liable to be detained

• IS96 - notification of temporary admission to a person who is liable to be detained

• IS82 series - notice of refusal of leave to enter

• IS99 - recognizance of applicant (bail cases)

• ICD0343 - imposition of reporting restrictions (restriction orders)

• IS248 - notice of restriction to a person who has made an in-country, in-time claim for asylum

• subsequently failed to comply with one or more of the conditions of immigration bail set by the Secretary of State (SoS) or the First-tier Tribunal (FTT), exercising their powers under Schedule 10 to the Immigration Act 2016

Or they either:

• proceed through the immigration control without obtaining leave to enter

• leave the border control area or Border Force accommodation without permission

• escape from detention and their whereabouts are unknown and all mandatory procedures to re-establish contact with them have failed

Source: Home Office, Non-compliance and absconder process v8 - GOV.UK (www.gov.uk).

Adult at risk: Level 1, 2 and 3

An individual will be regarded as being an adult at risk if:

• they declare that they are suffering from a condition, or have experienced a traumatic event (such as trafficking, torture or sexual violence), that would be likely to render them particularly vulnerable to harm if they are placed in detention or remain in detention

• those considering or reviewing detention are aware of medical or other professional evidence which indicates that an individual is suffering from a condition, or has experienced a traumatic event (such as trafficking, torture or sexual violence), that would be likely to render them particularly vulnerable to harm if they are placed in detention or remain in detention, whether or not the individual has highlighted this themselves

• observations from members of staff lead to a belief that the individual is at risk, in the absence of a self-declaration or other evidence

Levels consist of the following:

• Level 1: A self-declaration (or a declaration made on behalf of an individual by a legal representative) of being an adult at risk should be afforded limited weight, even if the issues raised cannot be readily confirmed

• Level 2: Professional evidence (for example from a social worker, medical practitioner or nongovernment organisation (NGO)), or official documentary evidence, which indicates that the individual is (or may be) an adult at risk should be afforded greater weight. Such evidence should normally be accepted and consideration given as to how this may be impacted by detention. Representations from the individual’s legal representative acting on their behalf in their immigration matter would not be regarded as professional evidence in this context

• Level 3: Professional evidence (for example from a social worker, medical practitioner or NGO) stating that the individual is at risk and that a period of detention would be likely to cause harm, for example, increase the severity of the symptoms or condition that have led to the individual being regarded as an adult at risk, should be afforded significant weight. Such evidence should normally be accepted and any detention reviewed in light of the accepted evidence. Representations from the individual’s legal representative acting on their behalf in their immigration matter would not be regarded as professional evidence in this context

When considering the likely risk of harm for the period identified, decision-makers are entitled not to place decisive weight on assertions that are unsupported by medical evidence.

Given the difficulty involved in validating cases in which the only evidence available is the self-declaration of the individual concerned, the distinction between such cases and cases of those who are not considered to be at risk may not be great. However, the expectation, where the weight of the evidence is at level 1, is that this will act as a flag to all those involved in managing the case, and that particular attention will be paid to pursuing voluntary return options and progressing the case.

The flag should also act as an alarm should additional risk issues emerge as the case progresses, particularly if the person is already detained or, if not, following their detention. However, caseworkers should not usually disagree with medical evidence unless there are very strong reasons for doing so - for example, a finding by an independent tribunal that rejects the same evidence or credibility concerns arising from other sources (such as an asylum casework decision). Such matters may be taken into account in deciding the weight that should be afforded to evidence and could result in a reconsideration of the weight of the evidence.

Source: Home Office, Adults at risk in immigration detention - GOV.UK (www.gov.uk).

National Referral Mechanism (NRM)

The National Referral Mechanism (NRM) is a framework for identifying and referring potential victims of modern slavery and ensuring they receive the appropriate support. Modern slavery is a complex crime and may involve multiple forms of exploitation. It encompasses:

• human trafficking
• slavery, servitude, and forced or compulsory labour

An individual could have been a victim of human trafficking and/or slavery, servitude and forced or compulsory labour.

Victims may not be aware that they are being trafficked or exploited, and may have consented to elements of their exploitation, or accepted their situation. If you think that modern slavery has taken place, the case should be referred to the NRM so that the relevant competent authority can fully consider the case. You do not need to be certain that someone is a victim.

This guidance explains how first responders should complete the referral form before it is considered by the relevant competent authority within the Home Office.

Source: Home Office, National referral mechanism guidance: adult (England and Wales) - GOV.UK (www.gov.uk).

Annex C: Results from multiple logistic regressions

Logistic regressions results (out of contact, reporting compliance, absconder):

• Model 1 (M1): Out of Contact
• Model 2 (M2): Reporting Compliance
• Model 3 (M3): Absconder

	β -Coefficient (M1)	Odds Ratio (M1)	Confidence Interval (M1)	β -Coefficient (M2)	Odds Ratio (M2)	Confidence Interval (M2)	β -Coefficient (M3)	Odds Ratio (M3)	Confiednce Interval (M3)
Tagged	-0.400**	0.670	-0.689, -0.113	0.381**	1.463	0.094, -0.67	0.946***	2.546	0.621, 1.256
Albanian	0.854***	2.349	0.437, 1.269	-0.786***	0.455	-1.203, -0.365	0.429	1.535	-0.095, 0.994
Asylum support	-1.556***	0.211	-2.285, -0.871	1.680***	5.367	0.972, 2.446	-1.087*	0.337	-2.042, -0.219
Given accommodation	1.377***	3.962	0.692, 2.146	-1.784***	0.174	-2.594, -1.005	0.661*	1.936	0.065, 1.237
Living at accommodation	-1.018*	0.361	-1.831, -0.224	1.313**	3.716	0.496, 2.164	-0.653	0.520	-1.732, 0.361
Constant	0.255		-0.142, 0.656	-0.296		-0.037, 0.762	-2.042***		-2.609, -1.521

Logistic regression results (out of contact), control and tagged group compared:

• Model 4 (M4): Out of contact (control)
• Model 5 (M5): Out of contact (tagged)

	β – Coefficient (M4)	Odds Ratio (M4)	Confidence Interval (M4)	β – Coefficient (M5)	Odds Ratio (M5)	Confidence Interval (M5)
Albanian	1.294***	3.646	0.748, 1.842	-0.123*	0.884	-0.854, 0.564
Asylum support	-1.712***	0.180	-2.717, -0.809	-0.814	0.442	-1.991, 0.341
Given accommodation	1.294*	3.645	0.333, 2.404	1.470**	4.347	0.481, 2.684
Living at accommodation	-0.740	0.477	-1.792, 0.305	-1.095	0.334	-2.569, 0.341
Constant	-0.029		-0.531, 0.471	0.690		0.024, 1.402

Annex D: Regression results for individual factors

When examining factors individually, some were found to have an effect on the different forms of compliance. Where factors did have a statistically significant effect, this was consistent across both reporting compliance and out of contact, the latter being the inverse. In some cases, the control group had a statistically significant effect whereas the tagged group did not. This further suggests that people respond differently to the tag conditions and their response cannot be predicted through the factors that have been monitored in this pilot. Analysis of these factors was purely explorative and conducted for informing decisions for building models presented in Chapter 4.

Bail Route
The control group had lower odds of being reporting compliant if they had received SOS bail, which was the same for the tagged group but to a smaller extent. Bail route had the same effect on the odds of going out of contact: individuals who received SOS Bail had higher odds of going out of contact than those who received IJ Bail. Bail route was not significantly associated with tag compliance. Overall, the effect of bail route was comparable between the tagged and control groups.

Albanian Nationals
Albanians in the control group had lower odds of being reporting compliant and higher odds of going out of contact compared to all other nationalities. There was no relationship between Albanian nationality and compliance in the tagged group. As Albanians in the control group were less compliant than other nationalities but indistinguishable from others in the tagged group, it is possible that for some Albanians the tag was therefore effective at decreasing the likelihood of being non-compliant, but this was not the case for all.

Asylum support
The control group had higher odds of being reporting compliant if they were receiving asylum support, and were less likely to go out of contact. Asylum support was not a statistically significant factor for the tagged group.

Given accommodation
The control group were twice as likely to be compliant if they had been offered asylum accommodation, but the tagged group were less likely to be reporting compliant and twice as likely to be out of contact. Living at Accommodation: In the control group, those still living in asylum accommodation had higher odds of being reporting compliant and were less likely to go out of contact. Living at asylum accommodation was not a statistically significant indicator of compliance for the tagged group, despite whether they’d been offered accommodation having been associated with lower compliance.

NRM
People in the control group had lower odds of being compliant if they had received either a positive reasonable grounds or negative conclusive grounds decision (bearing in mind people who receive a positive reasonable grounds decision will then be assessed for conclusive grounds). People in the control group with negative conclusive grounds or positive reasonable grounds decisions had lower odds of being reporting compliant. The other NRM decisions were not significantly associated, and none were significant in the tagged group.

Cohort
People in the NPRC or TCU cohorts had higher odds of being compliant compared to the DAC cohort. This was contrasted in the tagged group, as TCU were less likely to be compliant compared to DAC cohort, but there was no statistically significant difference in compliance between NRPC and DAC cohorts.

1. Irregular migration to the UK, year ending September 2023 - GOV.UK (www.gov.uk), this data covers the time period September 2022 to September 2023. Data covering other timeframes (Irregular migration to the UK, year ending December 2022 - GOV.UK (www.gov.uk) and Irregular migration to the UK, year ending March 2024 - GOV.UK (www.gov.uk)) shows similar percentages (90%, 99%). ↩

2. SOS bail is decided by the Home Office, whereas IJ bail is independent of the Home Office. This means that while the Home Office can oppose bail and/or propose conditions of immigration bail, the IJ is not obliged to follow the Home Office’s wishes. Asylum claimants apply for bail if they are held in an immigration removal centre, a detention centre or a prison on immigration matters, see here criteria and conditions: Immigration detention bail: Before you apply - GOV.UK (www.gov.uk). ↩

3. Immigration bail conditions - Electronic monitoring (EM) expansion pilot - GOV.UK (www.gov.uk)]. ↩

4. European Convention on Human Rights - The European Convention on Human Rights (coe.int). ↩

5. EM is also used for monitoring Foreign National Offenders (FNOs); however, FNOs were not in the scope of this pilot. FNOs also undergo electronic monitoring through non-fitted devices, a technology that has never been tested on asylum claimants. ↩

6. Nineteen cases who were tagged in October 2023 did not have 2 full months of compliance data recorded. ↩

7. Eighteen thousand is an estimate based on quarterly counts, Home Office Immigration detention detailed datasets, year ending September 2023 - GOV.UK (www.gov.uk). ↩

8. This because the initial cases were part of the MEDP (Migration and Economic Development Partnership) pre-IMA (Illegal Migration Act) cohort, but they then stopped being tagged because of legal challenges to the MEDP scheme, then the cases continued to increase due to widening the groups who were tagged. ↩

9. This is based on anecdotal evidence provided by Home Office operational staff. ↩

10. Home Office - Detained Asylum Casework (DAC) - GOV.UK (www.gov.uk). ↩

11. Neither chi-squared test nor logistic regression (p = 0.67) found compliance and group membership to be associated. ↩

12. Discrepancies might be due to lack of up-to-date information on absconding on internal systems. ↩

13. Sample size: n = 1,011. Independent variables were selected for analysis based on the data available and requests from the project team. The out of contact model with multiple independent variables was created by adding all possible variables and removing the variables which were not statistically significant one at a time. The same variables were then used with the dependent variables reporting compliance and absconder for models 2 and 3. Categorical variables which had multiple outcomes were not included as the limited sample size would have hindered meaningful analysis, especially of complex categorical variables. ↩