UNCOMFORTABLE DATA ENDING DEPRIVATION OF LIBERTY OF MIGRANT CHILDREN THROUGH IMPROVED DATA 2021 2 Lead Author: Łukasz SZOSZKIEWICZ (l.szoszkiewicz@amu.edu.pl) Main Contributors: Hugo BALNAVES, Carolina CANETTIERI, Milica STAJIĆ Study Advisor: Georges YOUNES Design and Illustrations: Ewelina PIEŃKOWSKA How to cite this report: L. Szoszkiewicz, H. Balnaves, C. Canettieri and M. Stajic, Uncomfortable Data: Ending Deprivation of Liberty of Migrant Children through Improved Data, Poznan 2021. Cover Page: Ewelina PIEŃKOWSKA Published: July 2021 This publication has been produced with the financial assistance of the Global Campus of Human Rights under the bilateral agreement n. GC_RLF/POZ/2020 with Adam Mickiewicz University in Poznan (Poland). The contents of this document are the sole responsibility of the authors and can under no circumstances be regarded as reflecting the position of the GC. At least 330,000 children are deprived of their liberty for migration related reasons every year worldwide. Because of the limited data available publicly, little is known about the characteristics of these children. This study aims to verify the existence, accessibility, and quality of data on immigration detention of children. In order to verify what categories of data are recorded by the public authorities, the authors approached 37 states requesting statistics under Freedom of Information regulations. Migration authorities were asked for historical and up-to-date statistics disaggregated by basic demographic features such as gen- der, nationality and accompanied status. Although the findings confirm the scarcity of data, several data-related promising practices were identified. Their implementation can significan- tly improve data capabilities of migration authorities and contribute to better policy making, particularly in measuring the progress toward ending immigration detention of children. This report builds on the experience of the UN Global Study on Children Deprived of Liberty (2019) which was the first scientific attempt to assess the magnitude of the situation of children depri- ved of liberty, including migrant children. SUMMARY 3 This report is the outcome of the research project led by Łukasz Szoszkiewicz (Adam Mickiewicz University). Georges Younes (University of Applied Arts Vienna) provided invalu- able support and advice throughout the whole lifecycle of the project. His exceptional manage- rial and interpersonal skills allowed us to build an excellent research team and create an open and positive working environment, along with a sense of common purpose. The Global Campus of Human Rights alumni Hugo Balnaves, Carolina Canettieri, and Milica Stajić co-authored and co-edited this report, in particular case studies on Mexico and Bosnia and Herzegovina. Thanks to their remarkable commitment and extraordinary teamwork skills, this project was comple- ted smoothly and efficiently. Ewelina Pieńkowska (University of the Arts Poznan) brought incredible creativity in data visualisation and layout that allowed us to translate our scientific findings into a memorable graphic form. We wish to express sincere gratitude for the support from institutions and people who made this study possible. We would like to acknowledge the valuable insights shared by Dzidek Kedzia, Manfred Nowak, Manu Krishan, and Reina Marie-Loader (Global Campus of Human Rights) during the conceptu- alisation of the research process, particularly their consulting for the methodology of this study. Valuable feedback and comments to earlier drafts of the report were provided by Günter Schumacher, Jan Löschner (Joint Research Centre), and Elisa Klein-Diaz (Global Campus of Hu- man Rights). The Global Campus of Human Rights alumni Nona Mohammadi Imir, Noémie Ninnin, and Ne- daa Yousef provided excellent research assistance and their national expertise, when contacting national migration authorities. We are also grateful to the organisations and experts who were extensively consulted for the case studies: Jasminka Džumhur (Human Rights Ombudsperson of Bosnia and Herzegovina), Centre for Migration Studies of Mexico (CEM), and Luis Xavier Carranca Álvarez from the Le- gal Clinic Alaíde Foppa. This research would not be possible without the generous funding of the Global Campus of Human Rights and the Right Livelihood Foundation. The Adam Mickiewicz University provided significant administrative support that allowed the project to be completed in a timely manner despite challenges caused by the COVID-19 pandemic. ACKNOWLEDGMENTS 4 1. BACKGROUND 6 1.1. HOW CHILDREN CAN BENEFIT FROM THIS STUDY 7 1.2. DEVELOPMENTS FROM THE UN GLOBAL STUDY 9 2. RESEARCH PROCESS 12 2.1. COMPOSITION OF DATA MATURITY INDEX 12 2.2. PROCESS OF DATA COLLECTION 15 2.2.1. Scope of the study 15 2.2.2. Information extraction 17 2.2.3. Freedom of Information requests 17 2.2.4. Scarcity of data 20 3. MAJOR FINDINGS 22 3.1. LEGAL FRAMEWORK 22 3.1.1. Definitional vagueness 22 3.1.2. Datafication of immigration laws 23 3.2. DATA ACCESSIBILITY 28 3.2.1. Impact of Freedom of Information regulations 29 3.2.2. Benefits of Open Data portals 31 3.3. TIME TRENDS 32 3.4. LEVEL OF DISAGGREGATION 33 4. SELECTED CASE STUDY: BOSNIA AND HERZEGOVINA 37 4.1. LEGAL FRAMEWORK 38 4.1.1. Places of deprivation of liberty 38 4.1.2. Prohibition of detention of children 39 4.1.3. Scope of personal data collected 40 4.1.4. Data rights of individuals 40 4.1.5. Freedom of Information regulations 41 TABLE OF CONTENTS TABLE OF CONTENTS 5 4.2. AVAILABILITY AND ACCESSIBILITY OF DATA 41 4.3. TIME TRENDS 44 4.4. LEVEL OF DISAGGREGATION 44 4.5. CONCLUSIONS 45 4.5.1. Key takeaways 45 4.5.2. Room for improvement 46 5. SELECTED CASE STUDY: MEXICO 47 5.1. LEGAL FRAMEWORK 48 5.1.1. Places of deprivation of liberty 49 5.1.2. Prohibition of detention of children 51 5.1.3. Scope of personal data collected 51 5.1.4. Data rights of individuals 52 5.1.5. Freedom of Information regulations 53 5.2. AVAILABILITY AND ACCESSIBILITY OF DATA 53 5.3. TIME TRENDS 55 5.4. LEVEL OF DISAGGREGATION 55 5.5. CONCLUSIONS 56 5.5.1. Key takeaways 56 5.5.2. Room for improvement 56 6. CONCLUSIONS AND RECOMMENDATIONS 57 APPENDICES 60 BIBLIOGRAPHY 64 TABLE OF CONTENTS 6 Although immigration detention of children1 is rarely or never practiced in some states, there is a con- cerning number of states who have taken the more securitised approach to detention of migrant children.2 Despite states having legitimate reason for regulating the terms of entry and residency into their territories, it is still important that international human rights obligations are considered in poli- cymaking. International and regional human rights bodies have repeatedly concluded that detention of migrant children is never in the best interests of the child.3 The emerging international consensus is also that detention of children can significantly damage children’s physical, developmental, emotional and psychological health, ‘depriving them of their fundamental rights and their childhood.’4 For this re- ason, UN Global Study on Children Deprived of Liberty (‘UNGS’) highlighted that detention of children is always unnecessary and disproportionate, and states should eliminate this practice. Progress toward ending immigration detention of children is a drawn-out process that requires a sys- temic change. Informed policymaking should be rooted in evidence, meaning that for better protec- tion of migrant children, a robust data collection framework is needed.5 A frequently raised argu- ment is the lack of data on immigration detention of children. However, such a statement is true only to a certain extent. Although data on immigration detention is much more limited than data on migratory flows or international protection, the UNGS has demonstrated that sufficiently reliable data is availa- ble for at least 80 countries,6 and therefore provided a global estimate on the scale of this phenomenon – at least 330,000 children are deprived of liberty for migration related reasons every year. Some categories of data are more available than others such as information on the prevalence of unac- companied children being relatively frequently recorded while records on the percentage of detained migrant children with disabilities are nearly non-existent. The activity of international mechanisms and NGOs is gradually but irregularly increasing the availability of data. Data on the immigration detention of children is regularly asked for by the UN Committee on the Rights of the Child,7 as well as the European Council on Refugees and Exiles (‘ECRE’).8 The International Data Alliance for Child- ren on the Move, a cross-sectoral global coalition of states, international and regional organisations, NGOs, academia, and others was launched in 2020 and aims to develop and improve data collection.9 In this study, we propose that in a significant number of countries, statistics on immigration detention of children can be extracted through Freedom of Information (‘FOI’) requests. 1 This report follows the terminology used in the UNGS as well as by the UN treaty bodies. Therefore, the term ‘immigration detention’ will be used in reference to any facility where children are deprived of liberty for migration related reasons. See: Manfred Nowak, ‘UN Global Study on Children Deprived of Liberty’, Geneva 2019, p. 434. 2 Ibid., p. 454-466. 3 UNGA, ‘Report of the Special Rapporteur on the Human Rights of Migrants: Ending immigration detention of children andproviding adequ- ate care and reception for them’ (2020), 20 July 2020, A/75/183, para. 13. 4 Ibid. See also: Nowak (2019), p. 433. 5 Günter Schumacher, Jan Löschner and Francesco Sermi , ‘Data on Children in Migration’, Ispra 2019, p. 5 and 30. 6 Nowak (2019), p. 455. 7 Information on children in administrative detention for migration related reasons is frequently requested by the CRC Committee under the List of Issues. See for instance: Germany in 2021 (CRC/C/DEU/Q/5-6, para. 13b), Canada in 2020 (CRC/C/CAN/Q/5-6, para. 18d), Zambia in 2020 (CRC/C/ZMB/QPR/5-7, para. 47c), Greece in 2020 (CRC/C/GRC/Q/4-6, para. 17i). 8 AIDA/ECRE, Asylum Information Database. 9 UNICEF, ‘International Data Alliance for Children on the Move’, March 2020. 1. BACKGROUND https://asylumineurope.org/ https://data.unicef.org/resources/international-data-alliance-for-children-on-the-move/ 7 Our aim is to assess the quality of the data, which is determined primarily by its timeliness and level of disaggregation. For the same reason, we decided not to publish actual figures indicating the scale immigration detention of children in respective states, as the overarching goal of this study is to assist and support policymakers in improving data collection framework as well as highlighting promising practices implemented worldwide and to encourage other states to adopt them. 1.1. HOW CHILDREN CAN BENEFIT FROM THIS STUDY The UNGS concluded that detaining children for migration reasons violates international human rights obligations – namely Articles 3, 6 and 37(b) of the Convention on the Rights of the Child of 1989 (‘CRC’).10 Because of the significant harm that detention can do to children, along with the availability of non-custodial alternatives, immigration detention of children can therefore never be in the best interests of the child or considered lawful grounds for detention under international human rights law.11 Both this report, and the UNGS, advocate for a child-centred approach to data in which data serves as a tool for improving children’s rights through: Identification of vulnerable individuals and providing them with adequate care Migrant children on the move are obliged to disclose vast amounts of personal data at various stages of migration procedures, including at border crossings, identity screenings, admission to detention centres, and others. The scope of data collected from them is primarily dictated by the needs of national security and includes, among others, fingerprints, iris scans or verification of identity documents. This perspective pushes other categories of information on the margins, such as data on disability, languages spoken, or health needs. These are either not collected or recorded incomprehensively, making public datasets of questionable value in determining the needs of children on the move and providing them with adequate care, in particular adequate al- ternatives to detention. For this reason, this study advocates for a child-centred data collection system that allows for the disaggregation of data by vulnerable status. 10 The detention of children for migration reasons is inconsistent with the best interests of the child (Article 3), a child’s right to develop (Article 6) and the right of the child to not be deprived of liberty unlawfully or arbitrarily. This position is further supported by the Report of the Special Rapporteur on the Human Rights of Migrants (2020). See: UNGA (2020), para. 13. 11 Nowak (2019), p. 451. BACKGROUND 8 Disclosure and proving the systemic and structural character of immigration detention of children Accessing statistical information is one of the primary means of controlling the exercise of government power and can often contribute to uncover human rights abuses. In the United States, the data released from an FOI request led to the discovery of the 'no-release policy' adopted by the Immigration and Customs Enforcement (‘ICE’).12 Statistical data plays an impor- tant role in front of international mechanisms. For instance, based on the administrative data, the European Committee of Social Rights has recently assessed the scale of immigration deten- tion of children in Greece and determined that there have been violations of the right of children to social protection along with the right to adequate shelter.13 Monitoring and measuring progress in eliminating migration detention of children The widespread nature of immigration detention of children around the world indicates that the practice is systematic. At the same time, a variety of alternatives to detention are being applied in various regions. An appropriate data collection framework is therefore essential to assess the efficiency of legislative and policy measures, document promising practices and support their implementation in other countries, as well as to monitor global progress in ending immigration detention of children.14 The implementation of a child-centred approach to data requires development of an adequate legal and technical environment to ensure that data is available (i.e. collected) and accessible (i.e. rele- ased publicly), the quality of this data is maintained to as high a standard as possible, and also allows to identify vulnerable individuals through functional disaggregation. This report intends to assess the extent to which 37 selected UN member states have effectively implemented this approach which was operationalised in the following recommendation of the UN Global Study: States should collect and make publicly available anonymised data, disaggregated to the gre- atest extent possible while ensuring confidentiality. At a minimum this should be done by age, gender (ideally reflecting, in addition to only ‘female’ and ‘male’, numbers for those whose gen- der identity does not match the sex assigned at birth or on identity documents), unaccompa- nied/accompanied status, nationality and migration status, disability, length of stay (including cumulative length of detention for individuals released and immediately re-detained), and pla- ce of detention. It should also reflect the numbers of children deprived of their liberty on the 12 NYCLU, Jose L. Velesaca v. Chad Wolf et al, 1:20-cv-01803. 13 European Committee of Social Rights, International Commission of Jurists (ICJ) and European Council for Refugees and Exiles (ECRE) v. Greece, complaint no. 173/2018. 14 UNGA (2020), para. 86(k). BACKGROUND https://www.nyclu.org/en/cases/jose-l-velesaca-v-chad-wolf-et-al 9 basis of their own or their parents’ migration status. These data should be made available at least annually.15 Despite the practice of immigration detention of children being inconsistent with international human rights obligations, there is still a long way to go to ensure the practice is eliminated globally, meaning that effective data collection to assess the changes in the area is crucial. Effective data collection poli- cies for migrants will assist in achieving Target 10.7 of the 2030 Agenda for Sustainable Development that aims to facilitate orderly, safe, regular and responsible migration and mobility of people, through the implementation of planned and well-managed migration policies. This report aims to push for pro- gress and follow on from the findings and recommendations of the UNGS focussing specifically on the data on immigration detention of children collected by states. 1.2. DEVELOPMENTS FROM THE UN GLOBAL STUDY Following the findings of the UNGS in 2019, there have been some key developments in the interna- tional effort to eliminate the deprivation of liberty of children and their families. In parallel with the UNGS, the Special Rapporteur on the Human Rights of Migrants, Felipe González Morales, develo- ped a report on ‘Ending immigration detention of children and providing adequate care and reception for them.’ 16 The recommendations of the report are coherent with those in the UNGS, recommending, among other things, the implementation of domestic legislation prohibiting immigration detention of children, releasing detained migrant children and their families to the appropriate alternative care or non-custodial reception facilities, and favouring solutions like case management.17 There were a signi- ficant number of countries that made submissions to the UNGS, with the Morales report also receiving additional submissions from Indonesia, Hungary, Cyprus and twelve other states. The report highli- ghted promising practices adopted by many of the Central and South American states and African sta- tes.18 The report also referred to developments in Thailand in early 2019 where measures were put in place for children to have access to social services and non-custodial alternative care arrangements.19 In 2020, the Committee on the Protection of the Rights of All Migrant Workers and Members of their Families (CMW Committee) made a call for comments for their Draft General Comment No. 5 on migrants’ rights to liberty and freedom from arbitrary detention.20 It sets out a principle of non- -detention of migrant children underlining that the detention of children due to their migratory sta- tus is never in the best interests of the child and constitutes cruel, inhuman, or degrading treatment. In addition, the CMW Committee stressed that places intended for providing care and protection to unaccompanied children ‘do not in reality result in material deprivation of liberty, even if they are not formally detention centres’.21 15 Nowak (2019) p. 490. 16 UNGA (2020). 17 Ibid para. 83. 18 Ibid para. 36-38. 19 Ibid para. 39. 20 CMW-Committee, General comment No. 5 (2020) on migrants’ rights to liberty and freedom from arbitrary detention (draft), 14 August 2020, CMW/C/32/R.2, para. 46-51. 21 Ibid. BACKGROUND https://www.ohchr.org/EN/HRBodies/CMW/Pages/CFI-GC5-2020.aspx 10 In 2020, the Parliamentary Assembly of the Council of Europe adopted a resolution ensuring ‘recep- tion conditions and care for child migrants and refugees fulfil basic rights and needs, taking into acco- unt the special requirements for the protection of children; on no account should children be placed in detention.’22 The European Commission adopted the EU Strategy of Rights of the Child stating, in reference to the findings of the UNGS, that it supports ‘Member States in the development of effective and viable alternatives to the detention of children in migration procedures.’23 The CJEU has delivered a land- mark judgment at the beginning of 2021 concerning the interpretation of provisions of the EU Return Directive and the best interests of the child.24 In the case, the Court ruled that the deportation of a 15-year-old child was inconsistent with the child’s best interests and fundamental rights of the child with predicted potential implications for detained migrant children as well.25 The European Union Agency for Fundamental Rights (‘FRA’) released a report in 2020 on children in migration providing updated data for France, Greece, Italy, Malta, Netherlands, Poland and North Macedonia, showing some increasing trends in immigration detention of unaccompanied children.26 The international discussion on immigration detention is ongoing. International human rights bodies have expressed concern over the detention of unaccompanied migrant children in Greece, with the #FreeTheKids campaign gaining attention worldwide – particularly the effect of sanitary conditions in these detention facilities during COVID-19.27 In Central and South America, immigration detention of children remains considerably less present than in other regions.28 In January 2021, Mexico imple- mented legislation to require migrant children be held in child-welfare shelters rather than immigra- tion detention centres, to better protect children.29 The COVID19 pandemic had significant effects on global migration trends. The near-complete halt of global travel and the closure of most international borders in 2020 has had a significant impact on detention numbers in some countries. In Canada, the number of children detained decreased 92%30 while in the United Kingdom 59%31 (between the end of 2019 and 2020). Therefore, when analysing the data on children in immigration detention, we restrained from making statistical comparisons with the data collected in 2018-2019 under the Global Study questionnaire.32 22 Council of Europe/PACE, Resolution 2324: Missing Refugee and Migrant Children in Europe, 30 January 2020. 23 European Commission, ‘EU Strategy on the Rights of the Child’, 2021. 24 Court of Justice of the European Union, TQ v Staatssecretaris van Justitie en Veiligheid (C-441/19), 14 January 2021. 25 R. O’Donnell, ‘Spotlight on the Best Interests of the Child in Returns of Unaccompained Children & Reflections for the New Pact on Migration and Asylum – EU Immigration and Asylum Law and Policy’, EU Immigration and Asylum Law and Policy, 13 April 2021. 26 European Union Agency for Fundamental Rights, ‘Children in Migration in 2019’, Vienna 2020, pp. 18-19. 27 Hetal Doshi and Ritika Goyal , ‘The Plight of Unaccompanied Migrant Minors in Greek Detention System: A National and International Per- spective’, European Law Blog, 8 July 2020. 28 Nowak (2019), p. 462. 29 Amnesty International, ‘USA and Mexico deporting thousands. Of unaccompanied migrant children into harm’s way’, 11 June 2021. 30 Canada Border Services Agency, ‘Quarterly detention statistics: First and second quarter (Q1-Q2), fiscal year 2020 to 2021’. 31 UK Home Office, ‘National Statistics: How many people are detained or returned?’, Table Det_02b (Number of occurrences of children entering detention). 32 For methodology of the Global Study questionnaire see: Nowak (2019), pp. 33-56. On the other hand, in some countries detention numbers did not decrease, e.g. Australia or Bosnia and Herzegovina. BACKGROUND http://assembly.coe.int/nw/xml/XRef/Xref-XML2HTML-en.asp?fileid=28595&lang=en https://ec.europa.eu/info/files/eu-strategy-rights-child-graphics_en https://eumigrationlawblog.eu/spotlight-on-the-best-interests-of-the-child-in-returns-of-unaccompained-children-reflections-for-the-new-pact-on-migration-and-asylum/ https://europeanlawblog.eu/2020/07/08/the-plight-of-unaccompanied-migrant-minors-in-greek-detention-system-a-national-and-international-perspective/ https://www.amnesty.org/en/latest/press-release/2021/06/estados-unidos-mexico-deportan-miles-ninos-migrantes-situaciones-peligro/ https://www.cbsa-asfc.gc.ca/security-securite/detent/qstat-2020-2021-eng.html https://www.gov.uk/government/statistics/immigration-statistics-year-ending-march-2021/how-many-people-are-detained-or-returned 11 At the same time, due to the novelty and unexpectedness of the crisis, states have responded differen- tly to preventing the spread of infections in detention centres.33 The CMW Committee has expressed concern for the often unsanitary and overcrowded conditions in immigration detention facilities and have called upon states to release detainees and use non-custodial alternatives.34 Although in some countries vulnerable groups, including children, were released from detention,35 others introduced suspension of administrative hearings, restrictions on freedom of movement, and temporary closu- res of previously open facilities (see case studies on Mexico as well as Bosnia and Herzegovina).36 The latter has likely influenced the number of children de facto deprived of liberty, although statistically these people are not classified as detained. 33 United Nations Network on Migration, ‘COVID-19 & Immigration Detention: What Can Governments and Other Stakeholders Do?’, 29 April 2020, p. 4. 34 CMW-Committee, ‘General Comment No. 5 on Migrants’ Rights to Liberty and Freedom from Arbitrary Detention’, 2020, CMW/C/32/R.2, para. 7. 35 For instance, in Mexico unaccompanied children were released due to the order of the federal court. Migrants were also released in Norway, Spain, an Zambia. See: UN Working Group on Alternatives to Immigration Detention, ‘COVID-19 & Immigration Detention: What Can Governments and Other Stakeholders Do?’, p. 9. 36 UNICEF/The Alliance for Child Protection in Humanitarian Action, ‘Technical Note: COVID-19 and Children Deprived of their Liberty’, 9 September 2020. BACKGROUND https://www.unicef.org/press-releases/covid-19-immigration-detention-what-can-governments-and-other-stakeholders-do https://migrationnetwork.un.org/sites/default/files/docs/un_network_on_migration_wg_atd_policy_brief_covid-19_and_immigration_detention_0.pdf https://alliancecpha.org/en/system/tdf/library/attachments/covid-19_and_children_deprived_of_their_liberty_v1_lowres_0.pdf?file=1&type=node&id=37576 12 This report is primarily based on research conducted primarily through the Freedom of Informa- tion requests. Taking this approach allowed to extract primary data on children deprived of liberty for migration related reasons, frequently more detailed than statistics released publicly or reported to the UN treaty bodies. At the same time, this approach allowed us to obtain comparable data (states were requested to share the same scope of information for the same period) and assess which states record cases on immigration detention of children according to the UNGS recommendations. The collected data led us to develop the Data Maturity Index which integrates the most impor- tant recommendations of the UNGS, in particular, migration- and data-related sections. The Data Maturity Index served as an analytical tool to measure the robustness of domestic data col- lection frameworks, including their legal framework, public accessibility, timeliness and level of disag- gregation.37 The index scored the countries on the scale of 0-20 and therefore provided a basis for the selection of case studies. The selection of the abovementioned four components was guided by four main core research questions (as indicated on the infographic on p.14). Their selection was based on the findings from the UNGS,38 taking into account also recent developments in the field ofA data governance39 and the impact of data-based solutions on the enjoyment of human rights by migrant children.40 2.1. COMPOSITION OF DATA MATURITY INDEX The legal framework integrates in particular definitional clarity (i.e. which settings amount to depriva- tion of liberty; exclusion of certain groups from immigration detention) as well as data rights (e.g. the legal catalogue of data collected by the authorities or right to access one’s data). Legal information was extracted mainly from the legislation regulating the status of foreigners, refugees and asylum-seekers as well as immigration enforcement.41 The assessment of the level of protection and enforcement of data rights was extracted from the database ‘Data Protection Laws of the World’ compiled by DLA Pi- per.42 This was due to the fact that migration related regulations usually either referred to regular data protection laws or did not mention privacy laws at all which are, nevertheless, applicable. Therefore, 37 The codebook and coding instructions are included in the Annex I. 38 Nowak (2019), pp. 32-55. 39 Releasing public data is considered to be one of the vehicles of development in the digital age. See: Łukasz Szoszkiewicz, ‘Open Data: Toward Achieving and Measuring the Sustainable Development Goals’. In: Walter Leal Filho et al. (eds) Industry, Innovation and Infrastructure. Encyc- lopedia of the UN Sustainable Development Goals, Springer 2021. See also: IEAG, ‘A World That Counts: Mobilising The Data Revolution for Sustainable Development’, 2014. For this reason, the UN Statistical Commission created a subgroup on open data in 2018. See: UN Statistical Commission, decision 49/105, E/2018/24-E/CN.3/2018/37. The potential conflicts of open data with the right to privacy were analysed by the UN Special Rapporteur on the right to privacy Joseph A. Cannataci in 2017. See: UNGA, ‘Report of the Special Rapporteur on the right to privacy’, 19 October 2017, A/72/43103, para. 77-87. 40 Petra Molnar and Lex Gill, ‘Bots at the Gate: A Human Rights Analysis of Automated Decision Making in Canada’s Immigration and Refugee System’, 2018. 41 In Anglophone countries relevant acts typically include term ‘immigration’ in the title while in other jurisdictions words ‘aliens’ and ‘fore- igners’ are more prevalent. At the same time, in some of the former Soviet countries, relevant laws indicate in their titles that regulations are applicable also to stateless persons. 42 DLA Piper, ‘Data Protection Laws of the World’. 2. RESEARCH PROCESS https://www.dlapiperdataprotection.com/ https://link.springer.com/referenceworkentry/10.1007%2F978-3-319-71059-4_129-1 13 limiting analysis to the fragmented provisions included in the regulations on foreigners could lead to erroneous conclusions on the low level of protection of data rights. The public accessibility refers to the accessibility of data to the general public (e.g. whether data is published by default, released on request, or never disclosed). To assess this perspective, the methodo- logy of data collection included submission of individualised FOI requests to the relevant authorities. The implementation of FOI-based methodology indicates that the accessibility of public data is rela- ted, to certain extent, to the existing FOI regulations. For this reason, the Data Maturity Index inte- grates the selection of FOI-related features extracted from the largest existing database Global Right to Information Rating (RTI) developed by Access Info and Centre for Law and Democracy.43 In some countries allowed statistics on immigration detention of children have been, recently subject to either FOI requests or parliamentary inquiries and responses were subsequently released to the public (e.g. Austria, Czech Republic). In these cases, no new FOI requests were submitted.44 The longitudinal characteristics (time trends) allows, us to track fluctuations throughout the time and measure the implementation of migration policies. In the context of deprivation of liberty, regular and coherent statistics allows us to document progress towards ending immigration detention of children. Data diversity is measured by the level of disaggregation, particularly for the following features: age, gender, nationality, disability and accompanied status. Whenever possible, additional information was extracted, in particular on the length of detention as well as the cost of detention, however this took place in individual cases only. The level of disaggregation allowed us to reflect on certain characteri- stics in migration policies, such us the necessity to provide certain services (e.g. interpretation in the language of certain national group or addressing needs of persons with disabilities) while, at the same time, to hold authorities accountable to human rights standards (e.g. monitoring whether children are not detained). 43 Access Info and Centre for Law and Democracy, ‘Global Right to Information Rating’ (accessed 31 July 2021). The Data Maturity Index includes one ‘FOI regulations’ variable that consists of the following indicators: I2 (The legal framework creates a specific presumption in favour of access to all information held by public authorities, subject only to limited exceptions); I4 (Everyone, including non-citizens and legal entities, has the right to file requests for information); I6 (Requesters have a right to access both information and records/documents); I7 (The right of access applies to the executive branch with no bodies or classes of information excluded); and I29 (The exceptions to the right of access are consistent with international standards). 44 The rationale behind this decision is that previous requests were submitted by the citizens of the respective countries who exercised their right to information (which is frequently limited to the citizens only). Therefore, the analysis of the potential impact of FOI regime on the accessibility of detention related statistics could be best assessed when the applicant is a citizen of a given country. RESEARCH PROCESS https://www.rti-rating.org/ 14 COMPOSITION OF THE DATA MATURITY INDEX RESEARCH PROCESS For the codebook with detailed scoring instructions see Annex I (p. 60) DATA MATURITY INDEX LEVEL OF DISAGGREGATION Is data properly disaggregated? Whether released data is duly anonymised, in particular whether small made public Whether data is disaggregated by nationality Accompanied status Do the statistics indicate whether a child was detained unaccompanied or with a family/legal guardian Disability Whether information on disability is recorded Gender Whether data is disaggregated by gender Age Whether age group 0-18 is distinguished in the statistics TIME TRENDS Is data collected on a regular basis? Regularity Whether data is compiled on a regular basis, for instance annually, quarterly or daily, or compiled ad hoc, for instance when a FOI request is submitted Longitudinal data Whether longitudinal data is available – either for 5 or 10 last years Whether data is compiled on annual basis Last update Whether data from 2019 or 2020 is available – either publicly or extracted through FOI request Whether regulations indicate which settings amount to deprivation of liberty or use euphemisms, in particular restrictions of freedom of movement Scope of personal data collected Whether regulations indicate which data, including personal and biometric data, are allowed to be collected and processed by migration authorities Whether regulations prohibit detention of children or only certain groups such as unaccompanied children and children below certain age threshold Data rights of individuals Whether regular data protection regime is applicable to data of foreigners, e.g. right to access one’s personal data or maximum data retention periods LEGAL FRAMEWORK Is data collection, protection and access regulated? Regulation and enforcement of data protection regulations – variable based on the database of Data Protection Laws of the World by DLA Piper Data shared on FOI request or previous inquiries Whether data was shared on FOI requests submitted within the project or in response to the previous FOI requests and/or parliamentary inquires Whether statistics on the immigration detention of children are released publicly by default, e.g. in the annual reports compiled by the migration authority or published on Open Data portals Data shared with the UN Whether data on children in immigration detention has been shared either within the UNGS; under the last periodic review before CRC or CMW Committees; or with the UN agencies, in particular UNHCR within the last 5 years To what extent legal framework facilitates access to data collected by the authorities – variable based on the Global Right to Information Rating PUBLIC ACCESSIBILITY Is data collected and publicly released? How much time did it take for an authority to complete the submitted FOI request 15 2.2. PROCESS OF DATA COLLECTION 2.2.1. Scope of the study The scope of the study included 37 states and territories from all regions of the world.45 The selection of the countries was guided with the four criteria that included: 1) the practice of immigration deten- tion of children; 2) significant migrant and/or refugee population;46 3) geographical location; and 4) the robustness of freedom of information regulations. The states covered under the research are shown on a map. The prevalence of European countries is due to several reasons. First, European countries, in particular EU Member States, are continuously applying detention measures towards children47 (contrary to, for instance, Latin American states). In 2015-2016 immigration detention of children was allowed in 19 EU Member States and the numbers were increasing in some of them, particularly in France, Greece, and Malta.48 Secondly, Europe remains one of the main destinations for migrants which, in the context of harsh immigration policies in many countries, results in significant numbers of migrants placed in various settings euphemistically named ‘temporary accommodation’ that sometimes amount to deprivation of liberty. The selection of Europe- an countries was additionally guided by the migratory trends and included Balkan states (e.g. Bosnia and Herzegovina, Serbia, and North Macedonia), Nordic states (e.g. Denmark, Finland, Norway) as well as Eastern Europe (Poland, Russia, Ukraine). At the same time, research included two countries that did not practice immigration detention of chil- dren at the time of the UN Global Study (2018-2019), namely Brazil and Ireland. In both cases, our assessment was aiming to verify whether there are no records on children in detention. In both cases, no data on the cases of detention of children were observed.49 45 Mayotte was treated as a separate territory for the purpose of data collection (as it was the case under the UNGS). 46 World Bank, ‘International migrant stock - total (ID: SM.POP.TOTL)’ and ‘Refugee population by country or territory of asylum (SM.POP.REFG)’. 47 Nowak (2019), p. 463. 48 European Union Agency for Fundamental Rights, ‘Children in Migration in 2019’, Vienna 2020, pp. 18-19. 49 In Brazil detention of migrant children is allowed in exceptional cases of deportation and extradition, however, the UNGS found that no chil- dren are detained in practice. The FOI requests submitted under this project through the ‘Fala.BR - Integrated Ombudsman and Information Access Platform’ were promptly answered indicating that the requested data are not collected. RESEARCH PROCESS 16 SCOPE OF THE STUDY Selection of the countries was guided by the following criteria: RESEARCH PROCESS Source: National Geographic, 2015 (for migratory routes) Source: M. Nowak, ‘UN Global Study on Children Deprived of Liberty’ Geneva 2019, p. 463. COUNTRIES THAT DO NOT DETAIN: COUNTRIES WHERE THE FOI REQUEST WAS SENT: Australia Austria Brazil Bulgaria Canada Croatia Czech Republic Bosnia and Herzegovina Finland France Georgia Ireland Angola Belgium Denmark Iraq Jordan Kenya Lebanon Mozambique Spain Turkey Ukraine United States Kazakhstan Malta Sweden Malaysia Mayotte (France) Mexico New Zeland North Macedonia Norway Poland Romania Russia Serbia United Kingdom REPLIED Focus on the countries where children was observed within the UN Global Study (approx. 2/3 of the sample) and countries where data could not be extracted at that time (approx. 1/3 of the sample). Priority given to countries with migrant since 2015. A regional balance while taking into account major migratory routes like the Mexican-American route, the Eastern Mediterranean route, and the Balkan route. Balance between the countries with both robust and also weak (seven countries from the top 30 and bottom 30 according to the Global Right to Information Rating). MIGRATION RELATED DETENTION COUNTRIES WHERE THE FOI REQUEST WAS SENT DO NOT DETAIN UNKNOWN MAIN MIGRATION ROUTES REJECTEDNO REPLY Responses to historical requests were accessed instead of submitting a new request. Anguilla (United Kingdom), Argentina, Benin, Brazil, Chile, Congo, Colombia, Costa Rica, Cyprus, Germany, Ecuador, Gambia, Honduras, Ireland, Iran, Israel, Italy, Laos, Liberia, Luxembourg, Madagascar, Mauritius, New Zeland, Nicaragua, Romania, Panama, Peru, Qatar, ElSalvador, SãoToméandPríncipe,Taiwan,South Africa 17 2.2.2. Information extraction The composition of the Data Maturity Index was developed within an iterative process of consulta- tion, feedback and gradual adjustment of the dataset. Several variables that were identified at the be- ginning as potentially informative were excluded due to the scarcity of data, in particular the average length of detention50 and the cost of detention.51 A separate variable was dedicated to the content of house rules for detention settings, however subsequently abandoned due to its low accessibility.52 At the same time, information on disability was kept as a separate variable even though only three coun- tries record his data.53 This was due to the fact that the collection of data on disability within migrants was explicitly recommended under the UNGS due to the invisibility of persons with disabilities in the statistics.54 The data collection process was implemented by the team of seven researchers with either sociolo- gical or legal background that, in addition, completed a dedicated MA programme in human rights. The selection criteria for coding positions included also linguistic skills to secure access to official sources in French- and Portuguese-speaking regions of Africa, Spanish- and Portuguese-speaking regions of Americas as well as Arabic-speaking regions of Middle East. Therefore, all data collected within the project was extracted from primary sources, in particular Official Gazettes (for legal regu- lations) or statistical reports of immigration authorities (for immigration statistics). In addition, FOI requests were submitted to the relevant authorities to extract comparable information, in particular disaggregation by the most common features. All researchers were provided with a proper training and coding instructions (codebook and guide- lines) and were consulted on a regular basis by the Project Lead to resolve any data related issues. The project ensured the diversity of researchers (particularly in terms of gender and nationality) to integrate perspectives that could have been overlooked at the initial stages of the research process. 2.2.3. Freedom of Information requests To maximise the response rate, most of the individualised FOI requests were drafted in one of the offi- cial languages of the country.55 Researchers were provided with a draft FOI request as well as Letter of Authorisation highlighting the scientific purpose of accessing the data. Requests were submitted accor- 50 Only five countries shared data on the length of detention, namely Australia (disaggregated data with periods varying from 1 to 730 days between 2019 and 2020) Canada (average between 11.5 and 16.3 days in 2020), France (20 hours in 2019), New Zealand (disaggregated data with periods varying from 3 to 6 days between 2011 and 2018), Norway (between 2019-2020, there were 34 stays with a duration of less than one day; 8 stays with a duration 1-3 days; 4 days with a duration 3-7 days), and the United Kingdom (between 2-3 months in Q1 of 2021). 51 This data is publicly released only by the United Kingdom (97.54 GBP in 2020). 52 In some countries house rules for detention settings are adopted in the ministerial decrees (e.g. Belgium, Poland) or in the form of internal docu- ments issued by the central authorities (e.g. United Kingdom) while in others this is left to the discretion of the directors of respective facilities – either with some minimum conditions indicated in the legal act (e.g. Kazakhstan) or without any guidance (e.g. Iraq, Lebanon, Mozambique). 53 Bosnia and Herzegovina, Finland, and Serbia. 54 Nowak (2019), p. 490. Due to a lack of reliable global data, disaggregated by disability, it is difficult to estimate the prevalence of children with disabilities among children in immigration detention. See: Ibid., pp. 191-192. 55 Languages spoken by the research team included: Arabic, Danish, English, Farsi, French, German, Macedonian, Polish, Portuguese, Russian, Serbian, Spanish, and Ukrainian. RESEARCH PROCESS 18 ding to the domestic FOI regulations, internal guidelines56 and, if available, through electronic forms57 or Open Data portals.58 The initial FOI request included the following questions: Annual data on children accommodated in the facilities that have been previously identified and classified as places of deprivation of liberty (in 2019 or 2020). When drafting FOI requests, terms such as ‘detention’ or ‘deprivation of liberty’ were avoided and replaced with the official names of the respective settings (extracted from the domestic legislation). If various types of facilities were in operation (e.g. open, semi-open and closed institutions), FOI request focused on these settings that were classified as having the highest degree of confinement (i.e. closed facilities). The degree of confinement was assessed during the data collection process by researchers (scale 0-5). Disaggregation by age (if publicly accessible statistics did not distinguish between adults and chil- dren), gender, nationality, disability, accompanied status and average length of detention. In total, FOI requests were submitted in 33 countries and replied in 20 of them (response rate: 61%), although in three cases there were refusals (Kazakhstan, Malta, and Sweden). If the authority replied substantively to the initial request, a follow-up request was submitted (replied in 10 cases). In the lat- ter, authorities were asked to provide longitudinal data (2011-2018) disaggregated by gender, natio- nality, disability, and accompanied status. Some authorities followed domestic FOI regulations strictly and, therefore, refused to disclose data due to the applicant not being a citizen.59 At the same time, the approach of some other authorities was more liberal. In the latter cases, the data was released based on the exception for research purposes, even if the applicant was a foreigner.60 In one case, replies to our requests were subsequently released publicly on the website of the respective authority, however, one without anonymising the applicant.61 The detailed log of FOI requests is included in the Annex II (p.62). 56 For instance, Norway requires very detailed information on the purpose of requested data, brief description of the project etc. See: Norwe- gian Directorate of Immigration, ‘Innsyn og tilgang til data’ [Access and access to data]. 57 FOI requests were submitted through the electronic form for Bulgaria (replied), Georgia (replied), Jordan (not replied), and Turkey (not replied). 58 FOI requests were submitted through the Open Data portals for Canada (replied) and Mexico (replied). 59 Canada (disclosed data on the initial request but declined the follow-up request) and Malta. At the same time in Jordan and Spain it was not possible to submit FOI request via electronic form due to the lack of national ID (required field). The FOI requests sent on the general e-mail of the respective ministries were not replied. 60 E.g. New Zealand. 61 North Macedonia (Ministry of Interior), Response to FOI request no 1612-199/1 from 10 February 2021 (for data from 2019) ; Response to FOI request no 1612-804/1 from 16 June 2021. 1 2 RESEARCH PROCESS https://www.udi.no/en/statistics-and-analysis/informasjon-til-forskere-og-studenter1/informasjon-til-forskere-og-studenter/ https://www.mvr.gov.mk/Upload/Editor_Upload/Odgovor_za_Milica_Stajik_11_02_2021.pdf https://mvr.gov.mk/Upload/Editor_Upload/odgovori-prasanja-od-javen-karakter/%D0%93%D0%BE%D0%B4%D0%B8%D1%88%D0%BD%D0%B0%20%D1%81%D1%82%D0%B0%D1%82%D0%B8%D1%81%D1%82%D0%B8%D0%BA%D0%B0%20%D0%B7%D0%B0%20%D1%81%D1%82%D1%80%D0%B0%D0%BD%D1%81%D0%BA%D0%B8%20%D0%BC%D0%B0%D0%BB%D0%BE%D0%BB%D0%B5%D1%82%D0%BD%D0%B8%D1%86%D0%B8%202011%20%D0%B4%D0%BE%202018%20%D0%BA%D0%BE%D0%B8%20%D0%BF%D1%80%D0%B5%D1%81%D1%82%D0%BE_%D1%83%D0%B2%D0%B0%D0%BB%D0%B5%20%D0%B2%D0%BE%20%D0%9F%D0%A6%20%D0%B7%D0%B0%20%D1%81%D1%82%D1%80%D0%B0%D0%BD%D1%86%D0%B8%20%D0%A1%D0%BA%D0%BE%D0%BF_%D0%B5(%D0%BE%D0%B4%D0%B3%D0%BE%D0%B2%D0%BE%D1%80%20%D0%B7%D0%B0%208%20%D0%B4%D0%B5%D0%BD%D0%B0).pdf 19 RESEARCH PROCESS RESEARCH PROCESS age gender nationality disability average length of detention accompanied status Freedom of Information FOI m of ation Composition of Research Team Legal research Research on existing databases Initial FOI request Follow-up FOI request NO REPLY REPLIED Finished report 7 researchers Hello Languages spoken: Arabic, Danish, English, Farsi, French, German, Macedonian, Polish, Portuguese, Russian, Serbian, Spanish, Ukrainian on places of deprivation of liberty, authorities responsible for immigration detention and others Data extraction from the UN sources (UNGS, UNHCR, Open Data Portals, responses to historical FOI requests and reports of migration authorities Annual data on children in immigration detention in 2019 or 2020 Annual data on children in immigration detention in 2011 - 2018 Disaggregation Disaggregation Hallo ¡Hola HejBonjour Oi DATA MATURITY INDEX 20 2.2.4. Scarcity of data Due to the scarcity of data, it was impossible to calculate the Data Maturity Index for some countries. These were cases where there were neither statistics on immigration detention of children publicly available, nor the authority replied to the FOI request. In Sweden, the FOI request was answered, ho- wever, indicating that ‘the requested information is classified as confidential and cannot be disclosed to private individuals’.62 In these cases, only information on legal framework could be extracted, frequ- ently with significant difficulties. For this reason, seven countries were excluded from further analy- sis (Angola, Jordan, Kazakhstan, Kenya, Mozambique, Sweden, and Turkey). In addition, Romania was excluded due to conflicting data released in response to FOI request (declaring no children detained) and reported by ECRE earlier in 2020 (between 5 and 43 children were placed in detention according to various sources).63 At the same time Malta, although not releasing official statistics and refusing to share data in the reply to FOI request, was included in the analysis due to the availability of data in other UN sources, in par- ticular from the UNHCR. The latest annual data on immigration detention of children in Malta comes from 2017 and was shared by the Immigration Police.64 The FOI request submitted to this institution was rejected because the applicant lacked Maltese citizenship, despite the Letter of Authorisation cle- arly indicating scientific purpose. 62 Sweden (Swedish Migration Agency), Response to FOI request from 8 July 2021. 63 Felicia Nica, Country Report: Romania, ECRE 2020, pp. 130-131. 64 In 2017 there were 213 children in immigration detention in Malta. See: UNHCR, ‘Global Strategy Beyond Detention 2014-2019: Progress Report 2018’, 2019. RESEARCH PROCESS 21 DATA MATURITY INDEX – INDIVIDUAL SCORES (SCALE 0-20) RESEARCH PROCESS COUNTRY DATA MATURITY INDEX Australia 3,5 4,1 3,5 16,15,0 Norway 4,0 3,2 3,5 14,74,0 Poland 3,0 4,0 3,0 144,0 Bosnia and Herzegovina 2,5 4,9 3,5 13,93,0 Canada 2,0 4,7 3,5 13,23,0 Georgia 2,0 4,5 3,0 12,53,0 Austria 3,0 4,0 14,53,54,0 Croatia 3,5 4,0 15,54,04,0 Mexico 3,5 5,0 16,54,04,0 United Kingdom 3,5 4,8 15,33,04,0 Bulgaria 4,0 3,3 12,83,52,0 Czech Republic 3,5 3,8 12,83,52,0 Belgium 4,0 2,3 10,32,0 2,0 Denmark 4,0 2,8 9,82,0 1,0 New Zeland 3,8 3,8 10,62,0 1,0 France 4,5 2,9 12,94,0 1,5 Ukraine 3,0 2,3 3,5 10,31,5 Finland 4,3 3,5 15,85,03,0 Spain 3,5 3,0 10,01,52,0 United States 3,3 11,84,0 2,51,0 France (Mayotte) 4,5 8,71,51,7 1,0 Iraq 0,5 5,53,01,0 1,0 Russia 3,0 6,81,01,8 1,0 North Macedonia 3,9 11,44,02,0 1,5 Serbia 4,8 11,34,01,0 1,5 Lebanon 1,0 3,81,01,8 0,0 Malta 3,0 5,71,01,7 0,0 LEGAL FRAMEWORK PUBLIC ACCESSIBILITY LEVEL OF DISAGGREGATIONTIME TRENDS 5 , 3 3, 5 2, , 3 6 2, 1 , 3 1 5 5 3 8 2 3 3 3 4 4, 4 2, ,3 , 3 1 0 4 0 0 3 3 3 0 1 1 1 1 0 2 3 3 3 22 3. MAJOR FINDINGS The methodology of Data Maturity Index guided qualitative research into legislation and data collection practices on immigration detention of children. For this reason, major findings are discussed according to the framework presented in the previous chapter. 3.1. LEGAL FRAMEWORK Immigration laws are becoming increasingly complex and intertwined with regulations on national se- curity, administrative procedure, data protection laws and others. This often involves the difficulty of reconstructing legal norms, including such basic norms as the prohibition of child detention. For this reason, a significant challenge during the research was the definitional vagueness. Another challenge was to determine the obligation to maintain the database as well as the scope of stored data. Intere- stingly, data-related provisions became a significant part of immigration laws (in some cases leading to the introduction of entire chapters devoted to personal data processing, e.g. Croatia, Czech Republic, France, New Zealand). This phenomenon of ‘datafication’ of immigration laws is likely to further de- velop which poses both challenges and opportunities for the protection of human rights. If regulated appropriately, through ensuring data rights of individuals, this can lead to the increasing data availabi- lity and accessibility. 3.1.1. Definitional vagueness The comparative review of legislative framework indicates that most of the countries avoid using the term ‘detention’ (its equivalents in official languages) and use terms such as ‘accommodation’ (e.g. An- gola), ‘temporary accommodation’ (e.g. Bulgaria), ‘restriction of freedom of movement’ (e.g. Croatia), ‘custody’ (e.g. Norway), ‘stopping’ (e.g. Poland, Ukraine, Russia), ‘internment’ (e.g. Spain), or ‘presenta- tion’ (e.g. Mexico). There is also a wide variety of terms used for the settings where detainees are placed, e.g. ‘immigration centres’ (e.g. Bosnia and Herzegovina), ‘guarded centres for foreigners’ (e.g. Poland), ‘reception centres for foreigners’ (e.g. North Macedonia), or ‘temporary detention centres’ (e.g. Rus- sia).65 At the same time, certain terminological trends could be observed, e.g. the term ‘detention’ is prevalent in Anglophone countries (Australia, Canada, New Zealand, United Kingdom but also, for in- stance in Finland or Mozambique) while in the Eastern European countries the most common term is ‘stopping’.66 These terms typically have criminal law connotations (criminal procedure) which have a variety of implications. On the one hand, terminology is not only a technical aspect of drafting legisla- tion, but also affects the formation of attitudes in society.67 For this reason, the use of criminal law 65 During the Global Study at least 40 various terms used to describe detention facilities have been identified. See: Nowak (2019), pp. 435-436. 66 In the official languages: zatrzymanie (Polish), затримання (Ukrainian), задержание (Russian). 67 Jelmer Brouwer, Maartje van der Woude and Joanne van der Leun, ‘Framing migration and the process of crimmigration: A systematic analysis of the media representation of unauthorized immigrants in the Netherlands’, European Journal of Criminology, vol. 14(1), 2017, pp. 100-119. See also: Harald Bauder, ‘Why We Should Use the Term ‘Illegalized’ Refugee or Immigrant: A Commentary’, International Journal of Refugee Law, vol. 26(3), 2014, pp. 327–332. 23 terms to describe the situation of migrants (so called ‘crimmigration’ process) may contribute to the perpetuation of prejudices and the image of the migrant as a person committing a criminal act. On the other hand, unambiguous terminology can hinder application of protective measures, in particular alternatives to detention.68 The terminological diversity poses a challenge for comparative research and requires a case-by-case assessment that takes into account circumstances under which an individual could be placed in a certa- in settings (for instance removal centres are typically closed settings) as well as the general conditions (usually established in law). Although inaccessible in most cases, house rules are the most appropriate documents to determine whether certain place amounts to the deprivation of liberty as they practi- cally determine whether a person is allowed to leave facility at his or her will (alternatively, what is the procedure of leaving such place), whether facility is guarded by armed personnel, whether it is wire- -fenced etc. Often, it is only by evaluating the house rules that it is possible to determine whether per- sons placed in certain facility are subject to the restriction of movement or deprived of their liberty.69 One of the well-established principles of constitutional law is that the limitations on rights shall be established in law, which is intended to limit the arbitrariness of decisions concerning, among other things, deprivation of liberty. According to this principle, the greater the interference with a given liberty, the more precisely it should be regulated in law, particularly at the statutory level. For this reason, it should be expected that the house rules of detention facilities will be established in law, and only certain technicalities will be delegated to subordinate acts (e.g. decrees issued by the compe- tent minister). In fact, legal regulations of some countries clearly oblige competent ministers to adopt house rules (e.g. Poland, Georgia, Spain, Bulgaria, Serbia, North Macedonia) while in others directors of facilities are deemed responsible (e.g. Czech Republic). In France, adoption of house rules was subject to a joint action by the Minister of Immigration and the Minister of Interior.70 The duty of cooperation between two state bodies could be one of the mechanisms to reduce the arbitrariness of the public authorities. 3.1.2. Datafication of immigration laws The comparative assessment of immigration laws unveiled their increasing datafication, particularly in the following areas: 1) processing of personal data, including biometric data; and 2) maintenance of centralised databases. All these areas affect the data accessibility both for an individual as well as the general public as they, for instance, specify the scope of data that could be processed by public autho- rities. 68 Interview with Luis Xavier Carranca Álvarez - representative of the Legal Clinic Alaíde Foppa in Mexico, 15 June 2021 (conducted by Carolina Canettieri). 69 According to the European Court of Human Rights, “[t]he difference between deprivation of and restriction upon liberty is nonetheless merely one of degree or intensity, and not one of nature or substance.” See: European Court of Human Rights, Guzzardi v. Italy, Application no. 7367/76, 6 November 1980. 70 Article 26 of the of the Code of Entry and Residence of Aliens and the Right to Asylum (CESEDA) of 2004, amended by the Ordinance no 2016-1457 of 28 October 2016. MA JOR FINDINGS 24 Processing of personal data (incl. biometric data) Migration laws usually affirm the applicability of the general data protection regulations or do not inc- lude any reference to privacy laws at all. In the latter situation, the general data protection regulations usually remain applicable.71 At the same time, migration laws frequently modify or restrain the appli- cability of certain data rights, in particular the right to access one’s personal data, the right to object to the processing of personal data, and the right to erase personal data without undue delay. In some countries legislation specifically provides for the permanent retention of data on foreigners,72 which remains in conflict with the international standards in this matter. In 2020, European Court of Human Rights ruled out that the indefinite retention of biometric data violates right to privacy (Article 8 of the Convention).73 Excessive data retention periods have also recently raised concerns of Human Rights Committee.74 Most of the countries establish in law the scope of personal data to be collected by immigration officers. The catalogues can include more than 20 categories of data, including some basic demographic data (e.g. citizenship, gender, or age) but also highly detailed information on residence, employment, and education. In some countries, legislation provides a close catalogue of personal data while in others, legal framework leaves it open to interpretation (e.g. Australia, Mexico). Although the catalogues of personal data collected are excessive, aggregated statistics cannot be effectively extracted through FOI requests (see infographic below). 71 One of the most careful legislative approach was taken by New Zealand. Section 31(2) of the Immigration Act of 2009 specifies that “Biome- tric information must be dealt with in accordance with the Privacy Act 2020” while the following subsection (3) specifies that “Subsection (2) is for the avoidance of doubt”. At the same time, still only 128 out of 194 countries introduced legislation on the protection of personal data. See: UNACD, ‘Data Protection and Privacy Legislation Worldwide’. 72 For instance in Bosnia and Herzegovina biometric data of foreigners under surveillance are kept permanently. 73 European Court of Human Rights, Gaughran v. United Kingdom, Application no. 45245/15, 13 June 2020. 74 Human Rights Committee, ‘Concluding observations from 2019 to the report of the Netherlands’, CCPR/C/NLD/CO/5, para. 54; ‘Concluding observations from 2017 to the report of Pakistan’, CCPR/C/PAK/CO/1, para. 36; ‘Concluding observations from 2017 to the report of Switzerland’, CCPR/C/CHE/CO/4, para. 47. MA JOR FINDINGS https://unctad.org/page/data-protection-and-privacy-legislation-worldwide 25 PERSONAL DATA COLLECTED FROM CHILD MIGRANTS Categories of data collected from migrant children PERSONAL DATA BIOMETRIC DATA ADDITIONAL DATA other categories of data which authorities deemed important name gender age residential address citizenship marital status family ties languages spoken employment education type and number of identity document accompanied status details of border crossing legal grounds for detention entry and exit to detention facility outcome of an age determination proceedings signature photograph handprints iris scan audio or video recording of the person Among the countries that 9 in 10 a child established in the CRC (below 18 years) 7 in 10 distinguish unaccompanied children 6 in 10 disaggregates data by gender 5 in 10 disaggregates data by nationality 2 in 10 provides information on the average length of detention 1 in 10 records information on disability of children placed in immigration detention 1 in 10 applies data anonymisation before releasing data publicly MA JOR FINDINGS 26 Significant number of countries provide specific regulations on biometric data in immigration laws. In most cases the scope of biometric data that could be collected by the public authorities is defined in law and limited to fingerprints and photograph only (e.g. Canada, Croatia, France, Poland, Norway, United Kingdom, Bulgaria, North Macedonia). Nevertheless, in some countries additional data could be collected such as iris scans (Australia, New Zealand) or audio or video recordings of a person (Au- stralia). Sometimes the meaning of ‘biometric data’ is either left to the scrutiny of law enforcement75 or includes only exemplary categories of data falling into this concept.76 The two latter situations could lead to the collection of excessive information on foreigners and unnecessary risks related, for instan- ce, to future data breaches. Some legislations provide specifically that biometric data cannot be collected for certain groups, e.g. for a child below 14 years of age (Bosnia and Herzegovina,77 Canada78) or 12 years of age (Croatia79). Another approach was taken by France where the collection of biometric data is allowed, however, for a period strictly necessary for providing ‘care and guidance’, and taking into account the individual situation of a child.80 These limitations are supported by the recent findings that most of data-based technologies such as iris recognition are primarily designed to work with adults and, therefore, under- perform when applied to children.81 Recent studies indicate that the accuracy of facial recognition sys- tems decreases significantly for children below 13 years of age.82 Therefore, processing of biometric data on children should be subject to stricter proportionality test compared to adults.83 In some countries automated decision-making systems (e.g. visa processing, customs clearance) have been recently deployed which required adjustment of immigration acts.84 In France and New Zealand, separate chapters on automated data processing were introduced85 while in Canada and Australia the introduction of an automated system was left to the authorisation by the competent ministers.86 In the latter cases, details on the performance of the system, including rights of individuals subject to data processing were subdelegated to the executive orders which provides more space for arbitrary government intervention. 75 For instance, in Ukraine legislation on immigration does not specify which categories of data fall into the concept of 'biometric data'. See: Ukraine, Law on the legal status of foreigners and stateless persons of 2012, O.J. no. 19-20, p. 179. 76 For instance, in Australia the legislation specifies the scope of biometric data collected (i.e. fingerprints, handprints, photograph, audio or a video recording of a person, iris scan, signature) but also allows to collect any other personal data. See: Australia, Migration Act of 1958, section 5A. In other countries legislation specifies exact categories, e.g. Canada (photograph and fingerprints), France (photograph and fingerprints), New Zealand (photograph, fingerprints, iris scan), Poland (photograph and fingerprints). 77 In Bosnia and Herzegovina, fingerprints and signature cannot be taken from a child below 14 years of age. See: Bosnia and Herzegovina, Law on Aliens of 2015, Article 122(4). 78 Canada, Immigration and Refugee Protection Regulations, SOR/2002-227, Section 12.2(1)(a). 79 Croatia, Immigration Act of 2021, Article 22(4). 80 France, Code for the entry and stay of foreigners and the right to asylum (CESEDA) of 2004 (amended by the Ordinance no 2020-1733 of 16 December 2020), Article L142-3. For New Zealand see: Immigration Act of 2009, Sections 28-32. 81 UNICEF, ‘Faces, Fingerprints & Feet’, July 2019. 82 Joint Research Centre of the European Commission and Institute for the Protection and Security of the Citizen, ‘Fingerprint Recognition for Children’, Luxemburg 2013. 83 European Union Agency for Fundamental Right, ‘Facial recognition technology: fundamental rights considerations in the context of law enforcement’, Vienna 2020, p. 26. 84 Grażyna Baranowska and Łukasz Szoszkiewicz, ‘New International Migration Management Technologies and Their Impact on Sustainability’, Walter Leal Filho et al. (eds) Reduced Inequalities. Encyclopedia of the UN Sustainable Development Goals, Springer 2021. 85 For France see: Code for the entry and stay of foreigners and the right to asylum (CESEDA) of 2004 (amended by the Ordinance no 2020- 1733 of 16 December 2020), Articles L142-1 to L142-5. For New Zealand see: Immigration Act of 2009, Sections 28-32. 86 For Australia see: the Migration Act 1958, Section 5(1). For Canada see: Immigration and Refugee Protection Act of 2001, Section 186(5). MA JOR FINDINGS https://link.springer.com/referenceworkentry/10.1007%2F978-3-319-71060-0_120-1 27 Maintenance of centralised databases Although several domestic legislations include provisions on the purposes and scope of collected data, regulations related to the maintenance of databases and their operation remain limited. At least two countries stand out for their more precise regulations. In Spain, the law obliges public authorities to maintain a dedicated registry of unaccompanied children, including the scope of data included. At the same time, services responsible for the protection of unaccompanied children are obliged to share updated information on the status of these children. In Norway legislation provides additional safegu- ards through the imposition of limitations on search algorithms by listing features which can be used for a search function.87 In some jurisdictions legal frameworks specify an obligation to compile certain statistical information for policy-making purposes, including features by which such data shall be disaggregated (e.g. in Mexi- co regulations require collection and release of statistics on the protection of migrants, disaggregated, among others, by undertaken measures)88. Subjecting personal data of foreigners to the legal regime of public statistics can be a reasonable compromise between data confidentiality and accessibility. This is because statistical service is typically independent from the executive agencies and its officers shall fulfil the statistical secrecy obligations laid down in domestic regulations. As raw data collected for statistical purposes may not be re-used for other purposes, information on individuals that is shared with statistical service and subsequently deleted from databases of immigration authorities is likely to be appropriately protected from third party access while aggregated statistics are released for the general public.89 Subjecting data collection to the regime of public statistics can be particularly beneficial for obta- ining country-level data in states with decentralised administration (e.g. federal states). Regionali- sation of data collection can lead to discrepancies in methodologies as well as limited transparen- cy (as in Germany and Switzerland)90 while centralised statistical programmes aim to harmonise information inflow. In Spain, immigration regulations obligation to compile country-level statistics on migration and international protection and share data for policy-making purposes although no information on detention of children is released.91 In the European Union, comparable data on asy- lum applications and migration from outside the EU are aggregated by Eurostat, although with no particular focus on children.92 Harmonisation with EU standards is likely to play an important factor in potential candidates. It was one of the reasons for the inclusion of immigration statistics (inclu- ding on asylum seekers and irregular migration) in the national statistical programme 2021-2024 for Bosnia and Herzegovina.93 Information on irregular migration and measures undertaken toward migrants were also included in the statistical programme of North Macedonia 2018-2022.94 Never- 87 Norway, Immigration Act of 2008, Section 102e. 88 Mexico, Federal Law on Transparency and Access to Public Information of 2021, Article 69(IV)c. 89 In the European Union, the General Data Protection Regulation allows for re-use of (pseudonymized) personal data for statistical purposes. See: European Union, General Data Protection Regulation, 2016/679, Article 89. 90 Global Detention Project and Access Info Europe, ‘The Uncounted: Detention of Migrants and Asylum Seekers in Europe’, 2015, p. 20. 91 Spain, Organic Law 4/2000 on the rights and freedoms of foreigners in Spain and their social integration of 2000, fifth additional provision. 92 Schumacher, Löschner and Sermi (2019), p. 8. 93 One of the reasons for this planned statistical development is to improve the harmonisation with Eurostat standards. See: Bosnia and Herzego- vina, ‘Statistički program Bosne i Herzegovine 2021-2024’ [Statistical programme of Bosnia and Herzegovina 2021-2024], pp. 7 and 13. 94 North Macedonia, ‘ПРОГРАМА ЗА СТАТИСТИЧКИ ИСТРАЖУВАЊА ЗА ПЕРИОДОТ 2018-2022 ГОДИНА’ [Statistical programme for 2018- 2022], Skopje 2018, p. 20. MA JOR FINDINGS https://bhas.gov.ba/data/Dokumenti/Planovi/STATISTICKI_PROGRAM_BIH_2021_2024.pdf https://www.stat.gov.mk/pdf/Programa20182022.pdf 28 theless, there is no obligation within these statistical programmes to collect data on immigration detention disaggregated by age. UTILISING PUBLIC STATISTICS FOR IMPROVING DATA QUALITY In the United Kingdom, immigration statistics are developed in partnership between the Home Of- fice responsible for internal affairs and the Office for National Statistics.95 The statistics on immigra- tion detention are released quarterly and undergo a compliance check against the Code of Practice for Statistics to ensure trustworthiness, high quality and public value of published data.96 The dataset is accompanied by the detailed methodological user guide indicating the scope of data collected (e.g. list of detention facilities) and data quality assessment. The document informs also on the limitations of statistics, e.g. exclusion of persons detained for less than 24 hours in short-term holding rooms at airports and ports as well as persons detained in police cells. In addition, the document describes chan- ges to the data affecting the statistics, for instance, the reason behind a significant decrease in the number of children detained following 2010 (change of legal framework) or fluctuations in the share of children among detainees (revisions due to the age determination procedure). Overall, the appro- ach taken by the UK authorities facilitates both the quality as well as interpretation of produced data. 3.2. DATA ACCESSIBILITY Accessibility of data on immigration detention is slowly improving yet remains far from satisfying. This is mainly due to the following factors: 1) adoption and/or strengthening FOI regulations; 2) bene- fits of Open Data portals. At the same time, there are several factors that could undermine accessibi- lity of data, namely: national security regulations (or exemptions in these); and opaque interpretation of data protection or FOI regulations. The slow but gradual improvement is supported by the finding that we managed to extract primary data (i.e. directly from public authorities) on the number of children in immigration detention for three ad- ditional countries97 (in the UN Global Study secondary sources were used instead). In some cases, we managed to extract more detailed data than within the UNGS, particularly information on nationality of detainees98 or length of detention.99 In some cases, regular data collection and/or release has only started recently (Poland in 2014, Georgia in 2019). There was no single case, in which the state that shared data under the UNGS questionnaire, claimed not to collect the same data under FOI request. 95 UK Home Office, ‘Home Office Statistical Work Programme 2019/20’, October 2019. 96 UK Home Office, ‘User Guide to: Immigration Statistics’ (updated 27 May 2021.) See also: Office for Statistics Regulation, ‘Code of Prac- tice for Statistics’, February 2018. 97 New Zealand, North Macedonia, and Serbia replied to FOI requests and released requested data. 98 The following countries released data disaggregated by nationality: Australia, Austria, Bulgaria, Bosnia and Herzegovina, Czech Republic, Georgia, Mexico, North Macedonia, Norway, Poland, Serbia, and United Kingdom. 99 See: footnote n 50. MA JOR FINDINGS https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/835676/home-office-statistical-work-programme-201920.pdf https://www.gov.uk/government/publications/user-guide-to-home-office-immigration-statistics--9/user-guide-to-immigration-statistics#immigration-detention https://code.statisticsauthority.gov.uk/wp-content/uploads/2018/02/Code-of-Practice-for-Statistics.pdf 29 At the same time, France (Ministry of Interior) indicated that the data on children detained in Mayot- te is not available, although such data was extracted by NGO in 2017.100 3.2.1. Impact of Freedom of Information regulations To assess the relevance of FOI regulations, our sample included seven states from the top and bottom 30 states of the Global Right to Information Rating. From the countries with the highest score, five replied to both initial and follow-up requests and shared detailed statistics (Brazil, Croatia, Mexico, North Macedonia, and Serbia). On the other side of spectrum were Kenya and Ukraine which neither publish data on immigration detention, nor replied to our FOI requests. At the same time, from the countries with the lowest scores, only one replied and only to the initial request (France) and another one released data to a previous inquiry (Austria). The remaining countries did not reply to FOI reque- sts (Belgium, Denmark, Jordan, Lebanon, and Mozambique). This indicates that there is room for im- provement for FOI regulations in these states to ensure data accessibility in the area of immigration detention. One should note that this goes against the usually applied criteria of income level or data- -related indicators (applied for instance, by the Open Data Watch).101 FOI requests are a powerful yet neglected research tool.102 The advantage of FOI-based framework for data collection is a legal obligation of public authority to answer and, if no exemptions are appli- cable, disclose the public information.103 This obligation is typically implemented to ensure effective government accountability. Previous studies that involved FOI requests concluded that due to the sys- tem of internal reviews and appeals, public authorities are careful when arbitrarily refusing to share data.104 Recent studies demonstrated the value of FOI requests as a tool for collecting data also in relation to terrorist research.105 Nevertheless, some FOI regimes do distinguish between the categories of ‘document’ (written or electronic file), ‘information’ (processed data) and ‘data’ (raw data). In the latter cases, an applicant could be refused to access certain data due to two reasons. Firstly, raw data could be perceived as an ‘internal information’ necessary for the operation of certain authorities but not amounting to public information as defined in FOI regulations. Secondly, legislation could limit the right of access only to the existing documents and/or administrative records without the necessity to engage any additional organisational resources (e.g. to extract and merge information from two or more datasets). Neverthe- 100 Assfam-groupe SOS solidarités, Forum refugiés-Cosi, France terre d’asile, La Cimade, Ordre de Malte France & Solidarité ‘Mayotte, Centres et locaux de rétention administrative: 2017 rapport’, 2018, p. 16. 101 Open Data Watch, Report 2020. 102 Kevin Walby and Mike Larsen, ‘Access to Information and Freedom of Information requests: Neglected Means of Data Production in the Social Sciences’, Qualitative Inquiry, vol. 18(1), 2012, pp.31-42. 103 Nevertheless, in some countries, public authorities are not obliged to respond to FOI requests, e.g. France. See: Yseult Marique and Emma- nuel Slautsky, ‘Freedom of Information in France: Law and Practice’. In: D. Dragos, P. Kovač, A. Marseille (eds) The Laws of Transparency in Action. Governance and Public Management, Palgrave Macmillan 2018, pp. 73–118. 104 This conclusion was formulated in the studies conducted in the United Kingdom which has moderate FOI regulations (According to the Global Right to Information Rating, the UK is classified as 42nd out of 128 countries). See: Kevin J. Brown, ‘COUNTERBLAST: Freedom of Information as a Research Tool: Realising its Potential, The Howard Journal of Criminal Justice, vol. 48(1), 2009, pp. 88–91. See also: Joanna Clifton-Sprigg, Jonathan James, Sunčica Vujić, ‘Freedom of Information (FOI) as a data collection tool for social scientists’, PLOS ONE, vol. 15(2), 2020. This methodology was also found valuable in Canada and the US. See: Alex Luscombe, Kevin Walby and Randy K. Lippert, ‘Bro- kering Access Beyond the Border and in the Wild: Comparing Freedom of Information Law and Policy in Canada and the United States’, Law & Policy, vol. 39, 2017, pp. 259-279. 105 Colin Atkinson, Donna Yates and Nick Brooke, ‘Researching a Risky Business? The Use of Freedom of Information to Explore Counterterro- rism Security at Museums in the United Kingdom’, Studies in Conflict & Terrorism, pp. 1–19. MA JOR FINDINGS https://odin.opendatawatch.com/Report/annualReport2020#sec3-3 30 less, our research indicates that the practice is complex and is arguably evolving toward increasing openness. Out of six countries which limit access only to certain categories of sources, three released the requested data.106 In Norway, we followed a dedicated procedure for accessing data for scientific purpose which required submission of relatively detailed information on the project, however, resul- ted in granting access to very detailed statistics.107 The same procedure was foreseen in Denmark, but turned out unsuccessful. The response rate to our FOI requests reached 61% (20 out of 33 states replied) that indicates the usefulness of this method in the extraction of data on immigration detention. This is arguably becau- se aggregated statistics are not sensitive either for national security or privacy and, at the same time, are compiled on the central level for policy-related and enforcement purposes. For this reason, there is low probability of state’s refusal due to, for instance, the necessity to involve excessive organisa- tional or technical resources, confidentiality, or secrecy. In some states, processing of the voluminous requests could be subject to additional fees, however, the scope of data collected within this project remained within the reasonable expectations for public authorities.108 Previous studies distinguished two common approaches of authorities when handling FOI requests. The first one could be described as ‘all-or-nothing’ while another one as ‘informal negotiations’.109 In the case of accessing statistics on immigration detention, the latter approach was prevalent and most of the authorities shared only selected data. At the same time, the reasons for not releasing the rema- ining information were either not specified or limited to the statement that ‘the data is not collected/ not available’. The most discouraging barrier in applying FOI requests in comparative research is restricting right to access public information to citizens only. There are between 19 and 73 countries worldwide that expressly recognise the right to information of everyone, including non-nationals. The exact number is hard to determine due to the ambiguous language which is differently classified by the Global Inve- stigative Journalism Network (19 countries confirmed)110 and Global Right to Information Rating (73 countries confirmed).111 In the context of the EU-wide research, holding the EU citizenship can further improve the enjoyment of the right to information. Nevertheless, even in some countries with the ri- ght to information restricted to citizens only, we managed to obtain the requested data without the necessity to prove nationality.112 Overall, the right to information appears to be particularly efficient method for data collection in Europe (both EU and non-EU countries) as well as Latin America. The 106 Australia and Norway released a very detailed information that required involving some resources in data retrieval. The data was also released by France, however, only to certain extent (public authorities replied substantially to initial FOI request only). Three countries did not replied to FOI request (Denmark, Jordan, and Ukraine). 107 We were required to provide the following information: 1) title of the project; 2) purpose of the project; 3) expected data of publication of research findings; 4) affiliation; 5) personal details of the applicant; 6) description of requested data; 7) statement whether the requested data refer to children; 8) statement whether project involves personal data; 9) statement whether any requested data amounts to 'sensitive data'; 10) how the data will be utilized; 11) how data will be anonymized; 12) who will get access to data. 108 In two cases, Australia and New Zealand, we were informed on the possibility of additional fees which were, however, not charged. 109 Luscombe, Walby and Lippert (2017), pp. 259-279. 110 We submitted FOI requests to ten such countries and received requested data from seven of them. Sweden was the only one that refused data due to confidentiality while Ukraine and the United States did not reply to our requests. Information on the scope of the right to information was extracted from the Global Investigative Journalism Network. See: Global Investigative Journalism Network, GIJN's Guide to National Rules on Where Foreigners Can Make FOI Requests (updated as of September 2019). 111 Access Info and Centre for Law and Democracy, ‘Indicator 4 of the Global Right to Information Rating’. 112 For instance Canada and New Zealand. At the same time, two countries refused releasing data: Malta (specifically indicating on the lack of citizenship) and Spain. MA JOR FINDINGS https://docs.google.com/spreadsheets/d/1EzPtSAUv6FMGul98nkYIN8YOTmKLQIW0/edit#gid=2025142577 https://www.rti-rating.org/country-data/by-indicator/4/ 31 Inter-American system is recognised as a global leader among regional organisations promoting the right to information.113 Another obstacle in successful implementation of the FOI-based approach to data collection could be the ‘data granularity threshold’. It is likely that if more detailed data would be requested, the response rate would decrease significantly (e.g. information on the cost of detention, daily occupancy rates of immigration facilities). As most of the migration laws do not specify the content of the registries ma- intained by the public authorities, it is challenging to determine the limits of FOI approach to data collection. HARNESSING FOI REGULATIONS FOR COLLECTING DATA ON HEALTH The systematic literature review conducted under the UN Global Study identified only 31 scienti- fic publications that reliably addressed the health of children in immigration detention (published between 1980 and 2018).114 Therefore, WHO warns that significantly more evidence is needed to shed light on this topic and advocate for ending immigration detention. Together with health data, public authorities should also collect basic demographic data, in particular gender, age, and disa- bility which are essential for the identification of health needs.115 So far, one of the most compre- hensive primary data on mental health of detained asylum seekers, including children, was made publicly available by the Australian Human Rights Commission. The dataset was released under FOI regulations on the request submitted by the Guardian newspaper and includes data collected on 191 children.116 3.2.2. Benefits of Open Data portals The increasing popularity of Open Data is gradually shifting the mindset of public sector bodies from ‘disclosure-on-request’ to ‘disclosure-by-default’. The idea to allow everyone to access publicly held data and re-use it freely has various motivations, in particular economic growth, social inclusion (e.g. informing social policies and services) as well as strengthening governmental accountability.117 To en- sure findability, Open Data portals typically serve as a one-stop shop for data collected by various pu- blic authorities, including ministries, law enforcement agencies, national statistical offices and others. Although still in its early stages, the Open Data movement is already impacting the accessibility of information on immigration detention too. Firstly, there is no need to identify the relevant public authorities (nor to navigate through their websites) as all datasets are stored in an Open Data portal. Secondly, the datasets are typically described in an intuitive and comprehensive manner, in particu- 113 Bill Orme, Supporting Access to Information: A Practical Guide for EU Delegations, Brussels 2018, p. 38. In 2010, the OAS approved a Model Inter-American Law on Access to Public Information that influenced legislation in the region and beyond, in particular in Africa. See: Organization of American States, Model Law and Guide for Implementation. 114 Nowak (2019), p. 128. 115 WHO, ‘Policy Brief: Health Concerns among Children Deprived of Liberty’, 2021, p. 12. 116 Sarah Mares, ‘The Mental Health of Children and Parents Detained on Christmas Island: Secondary Analysis of an Australian Human Rights Commission Data Set’, Health Hum Rights, vol. 18(2), 2016, pp. 219-232. 117 Szoszkiewicz (2021). MA JOR FINDINGS https://apps.who.int/iris/bitstream/handle/10665/341885/WHO-EURO-2021-2713-42469-58984-eng.pdf https://media4democracy.eu/wp-content/uploads/2019/04/M4D-Ati-Practical-Guide-web-complet.pdf http://www.oas.org/en/iachr/expression/topics/acceso_otros_ley_modelo.asp 32 lar with metadata, brief descriptions of their content (e.g. the level of disaggregation) and sometimes even tags. For this reason, identifying statistics on children in immigration detention in the countries with the most advanced Open Data portals was particularly easy.118 In some cases we managed to extract historical data which could be archived or removed if published on the websites of the relevant public authorities.119 In addition, Open Data portals of Canada and Mexico are integrated with the database of historical FOI requests that facilitates access to the data previously released. In Bulgaria, on the other hand, the Platform for Free Access to Public Information facilitates the whole FOI request process (from submitting the application and forwarding it to the relevant authority to providing the requested information).120 The advantages of Open Data portals do not limit to searchability only. The lack of awareness on the scope of data collected by the public bodies is frequently indicated as one of the most significant ob- stacles in utilising administrative records for social sciences. Open Data portals provide the overview of available data, typically organised into categories and/or tagged (e.g. migration, administration of justice). Some of the Open Data portals have dedicated Application Programming Interfaces (APIs) that allow applications for accessing data in a machine-readable way.121 This, in turn, make Open Data portals particularly useful for social scientists. 3.3. TIME TRENDS The UNGS revealed numerous discrepancies in data collection methodologies across countries, one of which was the diversity of used timeframes. For this reason, one of the recommendations formula- ted within the Study was to aggregate data at least annually. Our research indicates that in majority of cases extraction of annual data is feasible through FOI requests (65% of countries record annual data), even if statistics published by default use other timeframes (other common formats are: mon- thly, quarterly, and snapshot date at the end of calendar year). Annual data is particularly important for migration research and policy-making as the number of migrants placed in detention can increase in certain seasons (e.g. warm months in Europe) and, therefore, snapshot numbers or ad hoc statistics can be misleading. This refers specifically to countries where detention is applied systematically, ho- wever, for relatively short periods of time.122 Another challenge for comparative research was me- thodological particularities in certain countries – for instance in Denmark statistics are compiled by average weekly occupancy of detention facilities123 while in Canada and the United States they are compiled in financial years instead of calendar years. 118 Canada, France, Mexico, United Kingdom. 119 For instance, Open Data portal of France includes dataset on placements in administrative detention centres between 2008-2013 (disag- gregated by age and unaccompanied/accompanied status). See: Dataset no 15065 (Placement in administrative detention centers). For Canada, information on the content of previous FOI requests allowed to extract data on children in detention between 2014 and 2016 (disaggregated by gender and age groups). See: Data package no. A-2018-11696 (released on the Request file number AI-2020-13283 on 3 September 2020). 120 See: Platform for Free Access to Information of Bulgaria. 121 See, for instance, terms of use for the APIs of the Open Data portal of UK: https://www.api.gov.uk/#getting-in-touch (accessed 12 June 2021). For France see: https://doc.data.gouv.fr/api/reference/ (accessed 12 June 2021). 122 Examples: France, Australia. To control for this practice, researchers can ask for the average length of detention. 123 This methodology is particularly misleading as the capacity of immigration facilities fluctuate throughout in time, in particular due to the increasing inflow of migrants. MA JOR FINDINGS https://pitay.government.bg/PDoiExt/indexExt.jsf 33 In addition to the most up-to date-statistics, FOI requests aimed at the extraction of statistics from the recent decade to assess the availability of longitudinal data. Only six states managed to provi- de statistics for at least 10 years (Austria, Bosnia and Herzegovina, Croatia, France,124 Mexico, and the United Kingdom). On average, states record information on immigration detention starting from 2014.125 3.4. LEVEL OF DISAGGREGATION In all countries covered in this study, immigration laws follow the definition of a child of CRC (individu- als under 18 years of age) and do not create additional age-based categories (such as ‘young offenders ’in the administration of justice which usually includes individuals older than 18 years). This coherent approach significantly facilitates comparative research across countries, however, the factual num- ber of children in immigration detention can be also influenced by the outcomes of age assessment procedures. For this reason, ad hoc data collection can exclude individuals that are to be classified as children afterward. For this reason, the statistical portal of the United Kingdom includes a disclaimer that data on the number of children entering detention is subject to change.126 The level of disaggregation differs significantly across countries. The most accessible is information on the accompanied status (74% of countries that share data) followed by gender (56% of countries) and nationality (48%). These results can be significantly influenced by the EU statistical framework which requires disaggregation of various indicators for unaccompanied children. In addition, the stan- dards developed by Eurostat are becoming a reference for the EU’s (potential) candidate countries what explains disaggregation by these features in Bosnia and Herzegovina, North Macedonia, Serbia, and Ukraine (see subsection 3.1.2 on datafication of immigration laws). The most detailed level of disaggregation was observed in Australia, Canada, and the United King- dom. In all these countries, information disaggregated by gender, nationality, accompanied status, as well as the average length of detention was extracted. Finland, although not providing information on the length of detention, was the only country that released data on disability and additionally keeps records on language and family ties. At the same time, Australia, Finland, and the United States were the only three countries that did not disclose small numbers due to privacy protection concerns (<5 numbers were released instead). The release of small numbers can lead to the re-identification of in- dividuals, particularly when information on their stories is available either in media outlets or social media. The cautious attitude of public authorities in Australia may be the attributed to 2014 massive data breach. In Finland, privacy concerns are not specific to migration statistics, but to all demographic data (due to the low population density). 124 In France annual data on immigration detention of children is available at least since 2008, however not for every years. In response to the follow-up FOI request, annual statistics from 2019 was released. 125 Median: 2014; mean: 2013,5. 126 UK Home Office, ‘How many people are detained or returned?’, 25 February 2021. MA JOR FINDINGS https://www.gov.uk/government/statistics/immigration-statistics-year-ending-december-2020/how-many-people-are-detained-or-returned 34 MIGRATION DATA BREACH IN AUSTRALIA In 2014, the Department of Immigration and Border Protection (now the Department of Home Affairs) published a report on its website in error that contained embedded personal information of all 9,258 people who were being held in immigration detention at the time. The information di- sclosed in the report contained the names, gender, citizenship, date of birth, period of immigration detention, location, boat arrival details, and the reasons which lead them to becoming an unlawful non-citizen under the Migration Act 1958.127 Asylum seekers feared that the disclosure of their information would subject th