CHAPTER 3 DATA COLLECTION AND ANALYSIS were sections devoted to migration, national security and armed conflict. For instance, regarding migration-related detention as many as 47% of replies lacked disaggregation by sex and some countries struggled with providing information whether children were unaccompanied or not. Interestingly, while Global Study data for the snapshot day was usually disaggregated by sex, this information was lacking in regards to the annual statistics. This data loss may be due to the organisational and technical divergences between the systems used by certain State agencies (e.g. prison authorities) and the national bodies responsible for compiling yearly statistics (especially in the case of decentralised/federal States). In countries with strict data protection laws, the reason may be also the deliberate deletion of information during the archiving processes. Nevertheless, as high-quality data remains necessary for tracking long-term trends, public authorities should ensure that the legal framework as well as the technical means in place facilitate official statistics by, inter alia, providing data disaggregated by the key characteristics. Data disaggregation is also essential for the monitoring of State compliance with international human rights law. For this reason, various UN bodies request data disaggregated by certain features, most frequently by age16, sex/gender17 or disability18. Notwithstanding, to ensure adequate protection of children as well as implementation of SDG No. 16.2, public authorities should intensify their efforts to collect data on other relevant features. Due in part to rapid scientific progress, data are playing an everincreasing role both in policy-making and decision-making. The UN 2030 Agenda stresses that ‘quality, accessible, timely and reliable disaggregated data will be needed to help with the measurement of progress (SGDs) and to ensure that no one is left behind.’19 This statement should apply to all vulnerable groups, in particular children whose lives will be increasingly co-shaped by data-based solutions such as algorithmic risk assessments used in sentencing.20 Therefore, States should take all appropriate measures to ensure that the data collected on children taken into any kind of detention are disaggregated at least by 16 CRC-Committee (OPAC), Concluding observations: Saudi Arabia, CRC/C/OPAC/SAU/CO/1, 2018, para. 15; CRC-Committee, Concluding observations: United Kingdom (2016), CRC/C/GBR/CO/5, para. 77. 17 CERD-Committee, Concluding observations: Canada, CERD/C/CAN/CO/21-23, 2017, para. 34; CEDAW-Committee, Concluding observations: Italy, CEDAW/C/ITA/CO/7, 2017, para. 50. 18 CRC-Committee, Concluding observations: Georgia, CRC/C/GEO/CO/4, 2017, para. 30; CRPD-Committee, Concluding observations: Argentina, CRPD/C/ARG/CO/1, 2012, para. 30. 19 UN General Assembly, Transforming our world: the 2030 Agenda for Sustainable Development, A/RES/70/1, 25 September 2015, para. 48. 20 Although algorithmic methods have advanced rapidly in recent years, their performance depend primarily on the quality of ‘training’ data. For the discussion on the accuracy and fairness of using risk assessment tools in the juvenile justice see: Richard Berk, ‘Accuracy and Fairness for Juvenile Justice Risk Assessments’, Journal of Empirical Legal Studies, Vol. 16(1), 2019, pp. 175–194. 46