It’s Not What You Know, It’s How You Use It: The Application of Country of Origin Information in Judicial Refugee Status Determination Decisions – A Case Study of Germany

Existing research has emphasized the different forms of expert knowledge available to refugee status determination (RSD) decision makers, as well as the differing conditions under which it is produced. However, little work has been done to address how decision makers interpret, represent, and use such evidence in their written decisions. This study investigates how country of origin information (COI) is used in judicial RSD decisions, taking decisions of Germany’s Higher Administrative Courts on Syrian draft evaders as a case study. The analysis shows that the courts draw different conclusions from the same evidence, utilizing interpretation, framing, and citation styles to amplify or dampen the persuasive force of COI in their reasoning. As such, legal reasoning dominates evidence, meaning that evidence is discursively highly malleable, frequently incidental to legal reasoning, and does not produce legal consensus. These findings raise concerns that decision makers use COI selectively to justify the positions they have adopted, rather than allowing their conclusions to be directed by COI. The article concludes by reflecting on what, if anything, should be done about these seemingly opaque and unaccountable textual and discursive forms of discretionary power.

decision makers use COI selectively to justify the positions they have adopted, rather than allowing their conclusions to be directed by COI. The article concludes by reflecting on what, if anything, should be done about these seemingly opaque and unaccountable textual and discursive forms of discretionary power.

IN TROD UCTION
Probity of evidence is a core challenge in refugee status determination (RSD). Decision makers must assess the credibility of applicants, as well as the situation in their country of origin, in order to weigh up the risk of return. Consequently, country of origin information (COI) is a crucial aid in RSD decision making, albeit one of many pieces of evidence that decision makers must consider. The European Union (EU) Qualification Directive requires that an assessment of an asylum application must take into account 'all relevant facts as they relate to the country of origin at the time of taking a decision on the application' . 1 The ways in which COI is generated, interpreted, and used are therefore of critical importance to the determination of refugee status.
This article adopts a broad conceptualization of COI, defining it as all sources that are used as evidence about a country of origin, including, for example, reports, media sources, and expert written statements. Ideally, such information would be selected and used consistently, according to transparent principles, in order to treat like cases alike. Indeed, existing literature has emphasized the importance of having a unified approach to COI selection, as well as when to rely on it and how to discern which COI is to be relied upon. 2 These considerations are important, but leave open questions about the practices of using COI to inform reasoning in decisions and how these vary. Through a case study of German judicial decisions, the present article fills this gap by asking not only what COI is used, but how it is used.
The inquiry is motivated by the fact that multiple asylum seekers from a particular country may face a similar threat, but different outcomes are reached when their cases are individually considered. 3 This can happen even when decision makers consider the same case scenarios, have the same or similar COI at their disposal, and are in the same national jurisdiction and apply the same legal standards. How is this so?
The article argues that divergent decisions under these circumstances are possible because different decisions employ similar COI in different ways. By analysing a selection of opposing second-instance administrative court decisions in Germany on factually similar cases -Syrian draft evaders in Germany -the article reveals that it is possible to justify 1 Directive 2011/95/EU of the European Parliament and of the Council of 13 December 2011 on standards for the qualification of third-country nationals or stateless persons as beneficiaries of international protection, for a uniform status for refugees or for persons eligible for subsidiary protection, and for the content of the protection granted (recast) [2011] OJ L337/9 (EU Qualification Directive) art 4(3)(a). contradictory positions that reflect different interpretations of the situation in the country of origin in question, using the same key COI, even when most other circumstances of the case are comparable. The focus is on decisions of Germany's Higher Administrative Courts, the final judicial body able to rule on evidence (see part 3). Even though these decisions concern individual cases, their impact is much more far-reaching: they answer questions of general facts that go beyond individual cases and thus represent, in practice, a sort of country guidance decision for lower courts, as well as for the decision making of government officials.
The article does not set out to explain, causally, how COI informs the development of RSD decisions, nor how such decisions are made on the whole. Rather, it provides a more complete picture of how COI is used to substantiate reasoning in judicial written decisions. Part 2 begins by locating the current study within the literature on COI. Parts 3 and 4, respectively, introduce the case study and describe its methodology. The research question is answered by using a form of discourse analysis that borrows methodologically from critical discourse analysis (CDA). 4 The focus of this approach is intertextuality -the way texts are treated in other texts. On this basis, part 5 sets out the study's findings. It shows how COI is interpreted differently in different decisions to support opposed outcomes (section 5.1), and how these varied interpretations influence the application of COI in detail -that is, how COI material is represented and recontextualized when applied in decisions, and how authority is attributed to different sources of COI (section 5.2). The findings are discussed in part 6: in drafting their decisions, judges use certain techniques that make it possible to 'amplify' certain elements of COI, whilst relegating other parts to the background in accordance with adopted positions. In short, the courts employ a large degree of discretion when it comes to applying, assessing, and interpreting COI, as well as how to represent it in their reasoning. Part 7 concludes by asking what can be done to counter such diverging uses of evidence.

PR E VIO US LITER AT UR E ON THE US E OF CO UN TRY OF OR IGIN INFOR M ATION IN R EFUGEE STAT US DETER M IN ATION
Scholars have studied the role of expert knowledge in courts in general, and in the context of RSD, in particular. They have also examined the application of COI in different institutional contexts from a procedural perspective. The body of literature that addresses expert knowledge and COI has focused on cultural evidence and certain kinds of anthropological knowledge. Foblets, for example, refers to the incommensurability of legal and anthropological thinking, as well as how judges assess its usefulness. 5 Research on cultural evidence in court often focuses on how cultural knowledge or ignorance affects the treatment of parties, for example, in matters of intercultural communication. Such cultural expertise is often used to help judges avoid misunderstandings, rather than to 'directly impact legal outcomes' . 6 The focus of this article is quite different. COI for risk assessment in RSD has an impact on the final decision because it serves as a key evidentiary basis for decision making. Lawrance and Ruffer critique the increasing evidentiary burden placed on asylum seekers and the expanding role of different forms of expertise in RSD, 'ranging from country conditions reports, to biomedical and psychiatric evaluations, to the emerging field of forensic linguistic analysis ' . 7 They are concerned about 'increasing dependence on expert testimony' 8 and observe a trend towards a 'standardization of knowledge' 9 which aims at reducing a decision maker's discretion concerning a case's outcome. Rosset and Liodden, however, argue that COI fulfils mainly 'symbolic functions' . 10 COI both legitimizes and substantiates RSD 'by suggesting that decisions that reference COI are well-informed and rational' . 11 In a recent article, Liodden likens COI to a map, because of its 'authority as a seemingly objective depiction of reality' . 12 Continuing the metaphor, she considers specific information from country reports as 'landmarks' on this map, used by decision makers to assess an asylum seeker's credibility by checking if he or she is familiar with these landmarks. Other scholars discuss the actual use of COI as expert knowledge in RSD, detecting 'different epistemic frameworks' 13 between decision makers and country knowledge producers. Some of them have themselves been engaged by courts or tribunals as country experts, and are keen to critically evaluate how decision makers apply their knowledge. 14 A common observation is that the country expert occupies a challenging place in RSD: as decision-making bodies rarely acknowledge the complexity of a situation in a country of origin, but rather try to bend this knowledge until it meets legal requirements, that is, until it provides a sufficient basis to support a binary decision (for example, protection/no protection lawyers and anthropologists; 16 while it is contested for the latter, the former may cut 'directly to the chase: what are the implications of all this "culture" for the credibility of the applicants' stories?' 17 Among scholars who take a legal, procedural perspective, Vogelaar analyses the use of COI in different (inter-)national RSD decision-making bodies, and how this use corresponds to the bodies' own quality standards, as well as to COI guidelines, 18 such as the checklist of judicial criteria for assessing COI by the International Association of Refugee Law Judges, 19 the Common EU Guidelines for Processing COI, 20 the European Asylum Support Office's (EASO) judicial practical guide on COI, 21 and the Austrian Centre for Country of Origin and Asylum Research and Documentation training manual for researching COI (ACCORD). 22 These guidelines mainly address the application of COI by RSD decision makers, offering standards for selecting appropriate sources, assessing their validity, and applying them appropriately. 23  aim is to ensure transparent and uniform use of COI in RSD processes. 24 Yet, Vogelaar concludes that the decision-making bodies she researched lacked transparency and fell short of these guidelines' standards for applying COI. 25 In summary, there is a noteworthy juxtaposition between anthropological analysis of knowledge application and legal perceptions of evidence: the former is concerned with epistemological questions, whilst the latter requires practical applicability. While relevant literature on COI focuses mainly on the anglophone context, it is particularly valuable in highlighting the institutional and political dynamics of the application of knowledge and the conceptual tensions that arise when country knowledge and legal reasoning come together. What is missing is an assessment of how COI is integrated into written RSD decisions with a focus on the gaps and room for manoeuvre left to decision makers, despite the existence of COI and accompanying guidelines. This article shifts the focus to how judges use COI in their decisions, rather than why a particular decision is reached, adopting a discursive lens that reveals broad discretion in the textual use of COI.

C A S E ST UDY: S YR I A N DR A FT E VA DER S IN GER M A N Y
To answer the research question -how is COI used to substantiate legal reasoning in written asylum decisions? -this article examines diverging decisions concerning factually similar cases: Syrian draft evaders (namely men of military age (18-42 years)), who have neither served in the army, nor completed military service, and are now reservists. Between 2016 and 2018, there was considerable debate in Germany about how their asylum claims should be assessed. For context, in Germany, asylum applications are decided at the administrative level by the Federal Office for Migration and Refugees (Bundesamt für Migration und Flüchtlinge, BAMF). These decisions can be appealed, in the first instance, to local Administrative Courts (Verwaltungsgerichte, VG) and, in the second instance, to state-level Higher Administrative Courts (Oberverwaltungsgerichte, OVG). The Higher Administrative Courts are the final factual judicial courts in Germany with the authority to issue guidance to lower courts and governmental agencies.
Between 2011 and 2015, almost all applicants from Syria were granted refugee status, but from early 2016 onwards, the Federal Office for Migration and Refugees predominantly granted subsidiary protection, with only 50 per cent of Syrian applicants granted refugee status between 2016 and 2020. 26 This resulted in an increase in 'upgrade claims' -appeals to Administrative Courts and, eventually, to Higher Administrative Courts for refugee status 24 Albeit uniformity within, rather than between, guidelines as they may vary between themselves; analysing differences and similarities between such guidelines is beyond the scope of this article. rather than subsidiary protection. Even though both statuses prevent removal and guarantee basic rights in Germany, there are some differences in the rights granted. In contrast to refugee status, people with subsidiary protection have only a limited right of family reunification, a time-limited residence permit due for review after one year, and permanent residence is harder to obtain. 27 Between 2016 and 2018, Germany's Higher Administrative Courts revealed considerable divergences in their assessments of whether Syrian draft evaders were persecuted politically and would therefore qualify for refugee protection (see section 4.1).
The Higher Administrative Courts exhibited two opposing views. One view held that the Syrian government considered draft evasion a political act against the regime, and automatically ascribed anti-regime political convictions to draft evaders, rendering them enemies of the Syrian State, hostile, and therefore treatable as the opposition. According to this view, draft-evading returnees would be in danger of political persecution, entitling them to refugee protection in accordance with the 1951 Refugee Convention. 28 The other view suggested that the Syrian government was aware that the majority of the five million forced migrants who left Syria, including draft evaders, did so not on account of political opposition to the regime, but rather out of fear of civil war. If so, military-age returnees would 'merely' face minor, administrative punishment, and might be assigned to military service. According to German law, 'sanctions for evading conscription, even if they emanate from totalitarian states, only constitute significant persecution under refugee law if they not only serve to punish a violation of a general civic duty, but are also intended to affect the person on account of their religion, political conviction or other characteristics relevant to asylum determination' . 29 If courts understand that ill-treatment of draft evaders upon return (torture or inhuman or degrading treatment or punishment) is carried out irrespective of the victim's personal attributes, it is considered 'serious harm', meriting the grant of subsidiary protection status rather than refugee status.
There was no resolution of these diverging views. The only court that could have resolved the inconsistencies, the German Federal Administrative Court (Bundesverwaltungsgericht, BVerwG) did not issue a fundamental, guiding decision because it only considers appeals on points of law of general or universal legal significance beyond the facts of individual cases. The Federal Administrative Court ruled that matters of fact, such as whether conscientious objectors from Syria are subject to political persecution upon return, do not pose legal questions of fundamental importance that require clarification, 30  In Germany's civil law system, courts must marshal the relevant evidence themselves, rather than it being furnished by the parties. 32 Judges in the (Higher) Administrative Courts are both fact collectors and decision makers. 33 However, legal guidelines about how this principle should be implemented are scarce, leaving judges with substantial individual responsibility in fact-finding. COI is difficult to gather, heterogeneous, and changeable due to political and social instability in countries of origin. Courts are required to construct a 'mosaic image' of the country of origin, 34 put together from various sources, to determine if there is a real risk of a person facing persecution or serious harm if returned to that country. This risk assessment requires complex knowledge about the country of origin. Unlike in the United Kingdom (UK), for example, (see part 6), there are no national country guidance cases, and for the reasons set out above, there has not been a Federal Administrative Court decision. Given the different Higher Administrative Courts' decisions, draft evaders might receive refugee protection in one German state and subsidiary protection in another. It is this problem that the article sets out to examine through a discursive analysis of the use of COI in judicial decisions relating to Syrian draft evaders. Not only is Germany an interesting case study in itself, but it also provides conditions under which differences in practices between courts can be observed, while many other elements remain constant because they are in the same national jurisdiction, making this a potentially revealing site for analysis.

Case selection
The research undertaken for this article considered 13 Higher Administrative Courts' decisions, issued between December 2016 and June 2018, seven of which rejected the appeal to upgrade from subsidiary protection to refugee status, and six of which overturned the lower court's decision and thus granted the appeal to upgrade from subsidiary protection to refugee status (see table 1). 35 These were all the first decisions 32 This inquisitorial administrative justice procedure ('Amtsermittlungsgrundsatz', see Code of Administrative Procedure, s 86(1)) is one of the main differences from asylum determination in adversarial systems, where parties provide courts with evidence. Lawyers thus have limited influence over fact collection in German courts. 33 Ida Staffans, Evidence in European Asylum Procedures (Martinus Nijhoff Publishers 2012) 119. The term 'fact collector' is used to convey that German judges do their own research. This differs from 'fact-finding' as understood in the Anglo-American context, which refers to judges deciding what they accept from the expert evidence presented to them. None of the Syrian asylum seekers had their claims refused outright; where an upgrade claim was rejected, the grant of subsidiary protection remained. Whether the strategies courts deploy are also used in jurisprudence related to other countries of origin to which deportations are possible would require further analysis. delivered by Higher Administrative Courts relating to draft evasion by men aged 18-42 years. 36

Analytical approach
The analysis seeks to illuminate how Germany's Higher Administrative Courts use COI in their legal reasoning to substantiate decisions concerning the treatment of Syrian draft evaders. It borrows from Fairclough's CDA. 37 CDA assumes that actual discursive or communicative events (such as texts) must be analysed as embedded in a wider framework of an 'order of discourse' and, even more broadly, a framework of (non-discursive) social practices: text must not be analysed without its context. CDA aims to critically explore 'the links between language use and social practice' , 38 that is, to analyse how power relations are constituted by language. While the present focus is less on broader power relations and social change, and more on the concrete level of text, it nevertheless has its basis in CDA. Decisions are conceptualized as discursive events that can only be fully understood within their particular social practice -the legal system, institutions, and norms, and the legal foundations of subsidiary protection and refugee status. CDA is used to examine how the 'reality' in a country of origin (or at least parts of it) is constructed by courts in order to determine whether, and on what basis, someone receives a grant of refugee status or, instead, a grant of subsidiary protection. That analysis does not examine the 'reality' behind the discourse (what is really going on in the country of origin, and whether courts apply country knowledge correctly) but rather the discourse itself (how courts construct this reality in assessing cases). In this way, the approach is different from a more mainstream legal analysis which usually asks if the law is applied correctly in relation to the facts of the case, or if the decision is based on an appropriate assessment of the situation in the country of origin. CDA facilitates a focus on courts' linguistic toolkits when applying COI. A key feature of CDA concerns the 'external relations' of a text: intertextuality and assumptions. 39 Both refer to the fact that texts are always connected with, and rely on, other texts (here, the decisions rely on COI reports). A core feature of intertextuality is the concept of recontextualization, understood as 'a movement from one context to another, entailing particular transformations consequent upon how the material that is moved, recontextualized, figures within that new context' . 40 Therefore, recontextualization is mainly a matter of framing of information: 'when the voice of another is incorporated into a text, there are always choices about how to "frame" it, how to contextualize it, in terms of other parts of the text -about relations between report and authorial account' . 41 Assumptions, on the other hand, do not refer to explicit external relations of a text. As Fairclough observes, '[w]hat is "said" in a text always rests upon "unsaid" assumptions, so part of the analysis of texts is trying to identify what is assumed' . 42 This section examines how information from COI is recontextualized differently within decisions, depending on different interpretations of key evidence and different factual assumptions (see section 5.1). 43 While Fairclough, a linguist, also focuses on 'internal relations' of texts (eg semantic and grammatical relation within texts), this article considers these only insofar as they concern a text's external relations (here, the use of COI in decisions).
Legal decisions are a particular kind of text and not the type Fairclough examined when he described his tools of analysis (these being policy documents and speeches). One analytically important distinguishing characteristic is that intertextuality is obvious in the decisions. They explicitly refer to extra-legal sources (here, COI), whereas in Fairclough's policy documents, the focus was on 'hidden' references in the text. The analysis here does not aim to 'uncover' external relations to decisions, therefore, but to show how these relations are applied in practice. 44 To ascertain that differences between the decisions were not due to fundamentally different COI, the analysis was developed through an iterative process. First, a quantitative textual analysis was undertaken to count instances of COI citation, and to produce an overview of the COI sources, authors, and types of authors (government sources, expert sources, and media sources) used in the decisions analysed. Table 2 and section 4.3 show that there were no statistically significant differences in authors and types of authors used in decisions that reached different conclusions, 45 indicating that something else must differ between them. Secondly, coding categories were identified through a qualitative analysis of the decisions. A pilot study with two strongly contrasting decisions -one refugee status protection, 46 one subsidiary protection, 47 selected to represent the full spectrum of the discourse of interest -enabled the coding system to be developed. This system of categories (described below) was then applied to all decisions, and refined further where needed. This resulted in an approach that cycled back and forth between the analytical and results levels, with the latter continually informing the former.

Numeric overview of the decisions analysed
The findings show considerable variation overall in the number of instances where COI is referenced in the decisions. 48 However, there is no statistically significant difference in the number of citation instances between decisions granting refugee status and those granting subsidiary protection. 49 This broad numeric similarity across the sample is also evident when considering the use of COI sources from different authors and authoring bodies in the decisions (see table 2). On average, 13.5 different authors or authoring bodies are cited in each decision, but there is no statistically Each reference was recorded to a source of COI as a citation instance, whether occurring directly in the text, in parentheses, or in footnotes. Distribution of number of instances of COI references: minimum = 13, maximum = 232, mean = 84.8, standard deviation = 72.6. 49 Mean_RP = 67.5, mean_SP = 99.6, t = 0.78, p = 0.45 (RP -decision granting refugee status; SP -decision granting subsidiary protection).
The Application of COI in Judicial RSD Decisions -Germany • Page 11 of 27 significant difference between decisions granting refugee status and those granting subsidiary protection. 50 The sources cited were classified into types of authors/authoring bodies, namely, government, (inter-)national non-governmental organizations (NGOs), international inter-governmental organizations, research institutes, media sources, and individual experts. Forty different authors/authoring bodies and 139 individual COI sources were identified across the sample. Government sources were referenced most frequently on average (mean = 8.8), and in all decisions. The popularity of sources then decreased in the order shown in table 2, with individual experts (including expert articles) cited least often (mean = 1.7), and in only seven decisions. Decisions granting refugee status were assessed to see whether they were more or less likely than decisions granting subsidiary protection to cite sources from particular types of authoring bodies. Although the rank order of author types by popularity within the two groups of decisions differed, with NGO sources referenced most often in decisions granting refugee status (compared to government sources referenced most often in decisions granting subsidiary protection), there were generally no statistically significant differences. 51 The fact that there is no statistical basis for distinguishing the use of COI, including in relation to the number or types of COI sources employed, confirms the need for qualitative, textual analysis of the legal reasoning to understand how different conclusions are justified.

R E SULTS
The qualitative discursive results are presented in two sections. The first examines how evidence that is well established across various COI is interpreted differently by courts that reach opposing conclusions. The second examines representational techniques in the decisions to see how evidence is emphasized and de-emphasized.

Interpretation of evidence
This section examines how evidence featuring in COI is interpreted differently and used to support diametrically opposed positions by the courts, based on different assumptions about the following five 'facts': (1) the existence of gaps in detailed information available about the Syrian regime; (2) the Syrian government's actions to ensure its continued survival; (3) the Syrian government's mistreatment of its citizens; (4) the arbitrariness of ill-treatment; and (5) the departure from Syria of many young men to avoid conscription. Even when courts accept these 'facts', this does not lead to uniformity in decision making.
First, all decisions acknowledge significant gaps in evidence concerning the treatment of returning draft evaders in Syria; since there are very few examples of draft evaders returning to Syria, their treatment upon return is not reported in the COI. 50 Mean_RP = 11.5, mean_SP = 15.1, t = 1.38, p = 0.19. 51 The only exception was media sources, which were cited significantly more in decisions granting subsidiary protection than in decisions granting refugee status (mean_RP = 0.7, mean_SP = 3.3, t = 2.11, p = 0.059). Full results are available from the authors on request. However, courts interpret this paucity of information differently and use it to support different viewpoints. Decisions granting subsidiary protection take the position that a lack of information and evidence weakens the case for refugee status. By contrast, decisions granting refugee status regard the lack of evidence as part of the reason why refugee protection should be forthcoming, assuming that the lack of evidence indicates the regime's totalitarian character and the risk of persecution. They suggest that courts must thoroughly acknowledge reports on general ill-treatment and human rights violations and, from this, extrapolate information for the assessment of the risk for returning draft evaders: It is in the nature of things that in regimes that operate largely outside the rule of law and human rights principles and where inhuman persecution is an omnipresent phenomenon, torture and ill-treatment cannot be reliably and comprehensively documented externally, but happen largely unnoticed by the public, or even operate in grey areas. 52 According to these assessments, the very ambiguity and opacity of the Syrian regime are a material consideration. 53 A second difference between the decisions is how they interpret Syria's precarity as a State. While a range of COI material confirms that Syria seeks to consolidate its geopolitical position by increasing its power base and security, 54 this fact is utilized in different ways by the Higher Administrative Courts. Decisions granting subsidiary protection assume that, due to the fight for existence, the regime has only a military interest in returning men of military age. The State needs soldiers and, therefore, instead of imprisoning returnees for a long time or torturing them to the extent that they cannot fight as soldiers afterwards, it is more likely to approach their utility rationally and send them directly into combat, since this is in the State's 'objective interest': 55 Imprisonment in the case of withdrawal from military service is, from the Syrian State's point of view, a rather counterproductive punishment because, as long as it lasts, it prevents the person concerned from using his or her military strength. 56 If the Syrian regime were to subject those who preferred not to take part in combat operations to treatment in violation of human rights, including torture, it would significantly diminish its own potential. However, this would be in contradiction to its primary goal, which has been pursued so far with great severity, of re-establishing the monopoly of power over the entire territory. 57 Decisions granting refugee status, however, assume that a regime fighting for its existence would accuse draft evaders of having weakened the State and thus consider them hostile. Consequently, torture would likely be an act of retaliation for disloyalty, not simply an act of deterrence. The State's struggle for power is then interpreted as increasing, rather than decreasing, the likelihood of torture of draft evaders if they are returned. Here, the State is depicted as irrational, vindictive, and volatile: In conjunction with the character of the Syrian regime that is aiming to achieve its goals without reserve by means that violate human rights, in particular torture, it can be assumed that the Syrian regime has regularly ascribed a disloyal, politically oppositional attitude to people who have evaded military service by fleeing the war abroad. Because, despite the war that threatened the regime's existence, these people did not stand ready for military action and so, from the point of view of those in power, showed behaviour that ran counter to its urgent military needs. 58 Even the withdrawal from military service is perceived as disloyal by the Syrian regime and the conscript is suspected of having a different, oppositional political position. 59 A third 'fact' , well established in the COI, is the long-standing ill-treatment of Syrian citizens by Syrian authorities. Decisions granting subsidiary protection interpret this as indicating that the regime's actions towards citizens are less likely to be politically motivated. The fact that such ill-treatment has been evident since well before 2011 (the year commonly cited as the start of the current conflict) is considered to undermine the view that ill-treatment in the country is motivated by political events: The general use of mistreatment or torture upon return, which cannot be ruled out, ultimately does not, as such, constitute an essential indicator of the political motivation of the persecution; because it should be considered that this behaviour did not arise only due to the current civil war-like situation prevalent since 2011, but that due to the state of emergency that has existed since 1963, the security services in Syria have in practice never been subject to either parliamentary or judicial monitoring mechanisms, and have also always been responsible for arbitrary arrests, torture, and solitary confinement in the past. Decisions granting refugee status, however, consider that the consistency of past ill-treatment indicates the likelihood of future political persecution: It is consistently reported in the utilized evidence that torture, mistreatment, arbitrary arrests and enforced disappearances have been common practices of the Syrian security forces for years and to this day. 61 A fourth issue raised frequently in the COI material is the arbitrariness of ill-treatment in Syria. Again, the courts interpret this fact differently. In decisions granting subsidiary protection, arbitrariness is used to indicate the lack of a coordinated, politically motivated campaign against returned draft evaders and therefore speaks against political persecution. 62 By contrast, in decisions granting refugee status, the arbitrariness of reprisals is not regarded as precluding individual persecution: [A]rbitrariness -as is typically the case with totalitarian or despotic systems of injustice -is not only tolerated by the regime, but also deployed or accepted as a means of maintaining power and of obfuscation. 63 The arbitrariness of violence in the country may also indicate a lack of State power, which could mean that the State is more likely to try to impose its will violently in a desperate attempt to assert authority. In this case, the arbitrariness of violence and ill-treatment could indicate an increased likelihood of political persecution of returned draft evaders.
A fifth 'fact' that the decisions interpret differently concerns the large numbers of draft evaders (evidenced both by COI and the quantity of such protection claims). Decisions granting subsidiary protection interpret the large numbers as driven mainly by applicants' fear of dying in combat, and do not regard it as linked to a Convention ground (such as political conviction). Decisions granting subsidiary protection express this view both as a form of logical reasoning ('fear of war is reasonable') and as part of a belief that refugee numbers are too high to assume that everybody leaving Syria does so for political reasons: The -completely apolitical -fear of war deployment of conscripts is a typical and powerful reason for draft evasion. 64 In view of the cross-cultural phenomenon of fear of a war deployment as a reason to evade military service in times of war, it is obvious to everyone that the flight and asylum requests by Syrian conscripts do not involve political opposition to the Syrian regime, but only have to do with the -understandable -fear of a war deployment. Decisions granting refugee status do not discuss the large numbers of draft evaders leaving Syria as often as decisions granting subsidiary protection. When they do, they tend to view draft evaders as holding political objections. One decision explicitly rejects the subsidiary protection decision line of reasoning: In addition, the alleged plausibility -according to which a postulated 'cross-cultural' phenomenon of fear of a war deployment as grounds to evade military service in times of war makes it supposedly obvious to everyone that flight and asylum requests by Syrian military conscripts commonly have nothing to do with political opposition to the Syrian regime, but solely to do with fear of a war deployment -ignores the fact that the motivations of conscripts in times of (civil) war are commonly not only based on the fear of war deployment, but are also decisively determined by the legitimacy of this military mission and thus also by the legitimacy of the compulsory military service demanded from him … From this perspective, draft evasion -especially if this occurs through an illegal flight abroad -is perceived as an outwardly visible manifestation of disloyalty to the Syrian State in a very clear way, even if there was no such disloyalty in any given case. 66 This section has set out how evidence derived from COI is interpreted differently by German Higher Administrative Courts, leading to different outcomes for people seeking protection in Germany. It has shown how evidence based on COI is unable to produce legal consensus. The results point towards the subordination of evidence to legal reasoning. The next section turns to an examination of the textual techniques used to emphasize and de-emphasize evidence in decisions.

Representation of evidence
Sometimes, evidence is so obviously against a particular line of reasoning that it is impossible to construe it differently. There is, however, a set of techniques available to those writing legal decisions that renders even the most inconvenient factual information flexible and malleable. This section disaggregates and catalogues these techniques. They concern how evidence is represented in text, ranging from explicit practices of negatively evaluating sources in the legal reasoning or framing them in particular ways, to more subtle practices of citation. By using these techniques, decision makers cannot be criticized for omitting key pieces of evidence, but they may effectively reduce or amplify the evidence's persuasive force.

Evaluation of COI
In the decisions studied, it is not uncommon for particular sources or authors of COI to be criticized. This is perhaps the least subtle of the identified techniques, and is most likely to be used when the COI in question is high profile or regularly drawn upon by other decisions, meaning that it would be inadequate to simply ignore it.
Decisions granting subsidiary protection, for example, frequently criticize two authors of COI often used to bolster the reasoning in decisions granting refugee 66 TH (n 34) para 148.
The Application of COI in Judicial RSD Decisions -Germany • Page 17 of 27 status: the Immigration and Refugee Board (IRB) of Canada's 2016 COI 67 and three COI reports issued by the United Nations High Commissioner for Refugees (UNHCR) in 2017. 68 The main points of criticism are that: (1) central statements in the COI do not refer to actual cases; (2) experts quoted by these COI do not sufficiently refer to primary sources, indicating a lack of methodological thoroughness; (3) expert statements in these COI are mostly value statements or subjective assessments; and (4) the COI authoring bodies themselves are politically motivated (especially in relation to the UNHCR reports). 69 One decision on subsidiary protection additionally refers to different COI reports as too old, lacking new evidence, and providing evidence inconsistent with that contained in other COI by the same author. 70 Another approach is to suggest that evidence from COI does not correspond with the 'official' position of the Syrian regime and must therefore be rejected: This assessment, however, … contradicts the official stance of the Syrian regime -which, according to its public declarations, pursues a policy of reconciliationin whose objective interest it can only be to use conscripts for military service. 71 Decisions also employ positive assessments of COI materials that support their reasoning. Whereas some decisions granting subsidiary protection strongly criticize reports by UNHCR, decisions granting refugee status stress that the information provided comes from several sources, and often emphasize their expert status. 72 Such decisions defend the UNHCR reports by pointing to the coherence of the information and observing that UNHCR, by reason of its mandate, has an authoritative voice with respect to RSD. 73 These decisions also employ affirmative, positive language (for example, 'comprehensible on all sides' 74 and 'detailed and comprehensive') in reference to the COI they endorse.

Recontextualization of COI
In addition to explicit evaluation, a range of more subtle techniques are used to recontextualize COI, including the framing of information within decisions, selective inclusion of pieces of information, and minor modifications to quoted material.
Framing of information. The order, presentation, and prominence of information in a decision can affect its persuasiveness. For instance, one decision granting refugee status quotes a COI report referring to the loss of military manpower in the first sentence of the core paragraphs in which the political character of persecution is described. 76 The sentence is directly followed by the assertion that draft evaders are considered 'traitors' .
Another example is information about torture and ill-treatment. In decisions granting subsidiary protection, torture is mainly understood as an arbitrary act, not as political persecution. When it comes to the ill-treatment of draft evaders, such decisions frame it according to the assumption that the Syrian State has a military interest in returning men, by using COI stating that young men are directly returned to combat rather than being imprisoned. 77 Decisions granting refugee status, however, use general information about torture and ill-treatment to underscore the State's violent character and as an indication of political persecution. 78 Key evidence is sometimes relegated to relatively less important parts of the decision text. Decisions granting subsidiary protection only refer to war crimes when discussing whether the future participation of the draft evader in committing war crimes could serve as a basis for protection. 79 Since this is denied, these parts are less important for the overall decision. Some subsidiary protection decisions frame information on war crimes with the statement that these do not indicate political persecution of citizens, and refer to it in the introduction as a sort of preamble, followed by the statement that subsidiary protection status (and not refugee status) is the appropriate outcome considering this context. 80 Decisions granting refugee status, by contrast, usually refer to war crimes against Syrian society at central parts of the reasoning to support their characterization of the State, and use strong wording to amplify their reasoning (for example, 'war of extermination', 81 'particularly repulsive manner', 82 and 'means in violation of human rights'). 83 Different framing of similar evidence is not confined to purpose-written COI. N-tv news reports about a television interview given by Syrian President Assad are quoted in both decisions granting refugee status and those granting subsidiary protection. 84 The core statements relate to how he regards the majority of Syrian refugees as 'good Syrians' but that there is 'of course, an infiltration by terrorists' . The interview is used in four decisions. Three subsidiary protection decisions take Assad's statement as evidence of the Syrian regime's rationality, as Assad distinguishes between 'good Syrians' and those who must be considered terrorists. 85 The single decision granting refugee status that quotes the interview, however, uses the same statement to support the claim that returnees have to expect a loyalty check upon return, and conclude that ill-treatment based on political conviction is likely. 86 Selective inclusion. The decisions can be highly selective in the evidence they include. Some decisions granting subsidiary protection mention COI stating that the official Syrian army is less and less involved in military operations in practice, implying that draft evasion is not considered disloyal behaviour because the army is less relevant for the war. 87 They also quote COI that refers to men living unharmed in Syria, without having been conscripted. 88 In decisions granting refugee status, however, this information is not mentioned. Furthermore, some decisions granting subsidiary protection note the existence of amnesties that show the regime has little interest in political persecution. 89 Only one decision granting refugee status refers to amnesties, 90 asserting that the matter is not relevant to the question of political persecution.
The incorporation of information about the large number of draft evaders fleeing Syria (section 5.1) is another example of the selective use of evidence. While decisions granting subsidiary protection refer to this information to support the assumption that people flee on account of the civil war rather than for political reasons, 91 decisions granting refugee status refer to the large numbers less often. When they do, it is in the context of general information about the war, in order to show its intensity. 92 Sometimes, reports feature in the decisions for the purpose of being dismissed. Decisions granting subsidiary protection cite two UNHCR reports that explicitly refer to draft evaders as 'traitors', 93 only to then reject their validity. 94 At other times, decisions acknowledge only those parts of reports that support their reasoning. Three decisions granting refugee status 95 refer to a decision of the German Federal Administrative Court which states that persecution is likely to be political in character when the draft evader is considered a traitor and therefore faces a risk of being 'punished excessively harshly' , 96 for example, while none of the decisions granting subsidiary protection 85 RLP (n 57) para 50; SL (n 65) para 22; SH (n 57) para 124. 96 quote this part of the federal decision. Connected to the issue of 'traitors', one COI report uses the term 'friend-foe-dichotomy' to describe the logic of the Syrian regime and to suggest that a regime following this logic will persecute draft evaders politically. 97 Even though decisions granting subsidiary protection draw on this report for other information, they ignore this section, 98 unlike decisions granting refugee status, which cite it in central parts of their reasoning. 99 Minor modifications of sources. Certain parts of sources are discussed almost universally in the decisions, serving as touchstone pieces of evidence. However, these parts are framed differently, and even modified, depending on the protection status granted. For example, all the decisions refer to the following paragraph from a Swiss Refugee Council report describing the treatment of deserters and draft evaders: [D]eserters and persons who have evaded military service are imprisoned and sentenced. Torture occurs in detention, and human rights organisations report executions of deserters. Family members are also arrested or put under pressure by the Syrian authorities. Men who are picked up by the security services are usually arrested by the military security service or the air force security service. Some are brought before the military court in Damascus. The Office of the United Nations High Commissioner for Human Rights (OHCHR) has documented cases of torture at both security services. Some of those arrested are sentenced to prison terms by the military court before being drafted, others are warned and sent directly into military service. Many men who are sent directly to combat receive very limited military training and are sometimes sent to the front within a few days. 100 The paragraph could be considered to support the grant of refugee status. Decisions granting subsidiary protection consequently seek to minimize its impact in their reasoning. They mainly paraphrase the penultimate sentence, weakening its meaning by adding the modifying formulations 'even though' and 'only': Even though [s]ome of those arrested are sentenced to prison terms by the military court before being drafted, others are only warned and sent directly into military service. 101 The additional words reduce the rhetorical impact of the source material. In an analysis of how scientific facts become established, Latour and Woolgar note that the use of modalities (statements about statements) becomes less common the closer a proposition comes to attaining factual status. 102 In this quotation from the decision, a reversal of this process can be observed: the italicized additions have the effect of casting the original statement into doubt. The information is also framed by stating that it indicates that the Syrian State wants to send these men directly to its 'suffering army' . 103 By contrast, decisions granting refugee status are much more likely to quote not only this key sentence, but also the following one, clarifying that men are not simply sent into combat, but face severe threat due to limited training before being sent to the front. 104 Another decision granting refugee status adds the phrase, 'this means as cannon fodder', to reinforce the reasoning rhetorically. 105 It uses the term 'cannon fodder' eight times; another decision does so twice. 106 This is the only example in the sample where a terminological addition occurs.
A similar case can be observed with respect to a 2016 report by the German embassy in Beirut, that is used in all the decisions: The Federal Foreign Office has no information that returnees to Syria have suffered assaults/sanctions solely because of their previous stay abroad. However, there are known cases in which returnees to Syria have been questioned, temporarily detained, or have disappeared permanently. This is mainly in connection with activities in proximity of the opposition … or in connection with draft evasion. 107 Decisions granting refugee status refer to this information without modification and in key parts of the reasoning. Decisions granting subsidiary protection treat it differently. They question the source's validity, and relegate it to less important parts of the decisions (separated from the substantive discussion of draft evasion). In addition, one case does not quote the last, seemingly critical, phrase ('or in connection with draft evasion'). 108 In summary, the different techniques of recontextualizing information identified above indicate how a particular part of a COI report can feature in all decisions, but with very different emphases. When COI and evidence do not support the courts' reasoning (and cannot be interpreted in ways that do), courts utilize techniques of framing (strategically locating the evidence within the text in positions that are less prominent), omit certain evidence, and modify sources to reduce their persuasive force. Conversely, where particular COI evidence, or sections within COI reports, provide strong support for a decision, this can be amplified by increasing its prominence within the text, dwelling in detail on certain points, or modifying it to strengthen it.

Quotation of COI
There are also different styles of quoting and citing COI in the decisions. Direct quotations are used to give particular prominence to COI as part of a decision's reasoning. 109 Sources are also referenced in continuous text, 110 which can be effective in closely integrating COI into the reasoning, as well as maintaining its pace, either to build momentum towards an important point, or to note, and then dismiss, information that does not support that reasoning. In the latter case, COI is frequently used in combination with contrast structures. 111 These are used to flag -but reject (or make a counter statement with respect to) -an opposing view. Words and phrases used to achieve this include 'even though', 'although', and 'indeed' at the start of the leading clause, followed by 'but' , 'however' , and so forth, at the start of the next clause. 112 Decision makers use these structures to acknowledge information that does not support their reasoning. The contrast structure allows them to immediately reject elements of COI that challenge their reasoning and move on quickly. This study found the technique was used more commonly in decisions granting subsidiary protection.
Citations given in parentheses at the end of sentences appear in different forms in the decisions. Often, single sources are cited after discrete pieces of information. However, at other times, multiple sources are cited together in parentheses. This can either strengthen the evidence cited, or, conversely, disguise the actual content of the reports. For instance, one decision makes the case that fear of combat, rather than political conviction, is the reason for leaving the country to evade the military draft: Doing military service in Syria is very dangerous because of the combat missions to large parts of the country and the greatly reduced training period … The Syrian army is therefore highly exposed to the problem of conscripts evading military service. 113 The statement is followed by cumulative reference in parentheses to five COI reports. The causal link of the second to the first sentence ('therefore') implies that young men 109 Although they can also simply be used as a way to convey contextual information without having to paraphrase it. 110 That is, via indirect speech. See eg HE (n 53) para 61: ' According to the information from UNHCR in addition to the current country of origin information from February 2017, the government also regards evasion of military service as a political or anti-government activity against which sanctions are imposed' . 111 The term 'contrast structures' comes from Dorothy E Smith, '"K Is Mentally Ill": The Anatomy of a Factual Account ' (1978) 12 Sociology 23. 112 For example: 'Indeed, returning draft evaders face sanctions and compulsory draft to the army. … This is, however, not based on the returnees' political conviction' . See RLP (n 57) paras 138-39 (emphasis added). The structures are used (by both subsidiary protection and refugee status decisions) mainly when referring to COI or to evidence in general. 113 NRW (n 47) paras 66-68.
The Application of COI in Judicial RSD Decisions -Germany • Page 23 of 27 evade conscription because military service is dangerous. Cumulatively quoting COI after this statement implies that these COI reports support this causal statement (and not only the factual statement that many men evade conscription). However, the COI reports cited do not say anything about young men evading the draft because it is dangerous (they do not discuss the subjective reasoning of young men), but simply state that many young men evade military service. By citing the COI reports cumulatively, the statement is arguably given greater authority, and the fact that none of the sources quoted directly covers what the court suggests is disguised. Similarly, Latour observes that scientists sometimes cite academic work not for its precise content, but for its ability to bolster the point the author is making. 114

BR ID GIN G THE G A P: THE COM PLE X IN TER PL AY OF E VIDEN CE A ND A SSU M P TIONS
One challenge of RSD is assessing -based on often-limited evidence -an applicant's risk of being persecuted (or subjected to other serious harm) on return. This is especially true for information about persecutors' motives, which at times can be crucial to the decision but very difficult to discern. While COI can be helpful here, it only partially resolves the (judicial) decision-making challenge. Despite this, COI is an essential building block for legal reasoning.
To answer the study's research question of how COI is used to substantiate legal reasoning in judicial written asylum decisions, this article has applied a social scientific methodology to legal texts. This has some tangible benefits. First, such a methodology can produce knowledge that is unlikely to be found in law reports. Law reports are records of judicial decisions that do not intend to offer critical scrutiny of the textual and discursive techniques employed to support the reasoning giving rise to such decisions. Secondly, by employing quantitative and qualitative analytical techniques in understanding judgments, the social science approach goes beyond the perspective adopted by research on the implementation of guidelines on the use of COI, cited earlier (part 2).
The results, however, are delimited by the scope of the case study offered here. This article does not examine in any detail how the Higher Administrative Courts' decisions were received and interpreted by decision makers in lower courts, or the ways in which these decisions were drafted in anticipation of their reception by different audiences, such as legal representatives and the applicants themselves. Furthermore, there is such a large volume of COI that, when faced with the task of collating and forming a view on so many sources, judges must, almost inevitably, employ a pragmatic degree of editing. Aside from these practical considerations, the findings here are noteworthy because they reveal how malleable 'facts' can be in decisions -even key facts commonly established across different COI sources. This opens a window onto a range of rhetorical and persuasive inter-textual techniques that have hitherto been largely overlooked in the legal and socio-legal literature.
There are various ways that the findings may be interpreted. The results could imply that courts assume a view and then approach evidence with the intention of interpreting 114 Bruno Latour, Science in Action: How to Follow Scientists and Engineers through Society (Open University Press 1987) 40. and representing it so as to support whatever position they hold. This is similar to what Fairclough calls 'bridging assumptions' -that is, 'assumptions which are necessary to create a coherent link or "bridge" between parts of a text, so that a text "makes sense"' . 115 The findings could indicate that the interpretation and representation of evidence in decisions is highly dependent on the courts' implicit and explicit assumptions. Factual assumptions about the logic of the Syrian regime, and the motivations of those who evade the military draft, are then used to create a coherent link with the application of COI.
One concern is that judges use COI instrumentally. They show little hesitation in arranging complex, partial, and contradictory evidence in a way that gives the impression of rather too consistently supporting whichever case is being put forward, and by side-lining elements of the COI that do not support their reasoning. If it is the case that decisions of such consequence approach COI in this manner, then this should be cause for concern. How, for example, are courts' positions determined if evidence is incidental to their adoption?
A potential framework for understanding the role of assumptions in the decision-making process is the theory of confirmation bias in judicial assessment. 116 An analysis of confirmation bias as part of the process of the production of a decision is beyond the scope of this article, but it represents an interesting future line of enquiry. Although the findings here show that bridging assumptions can be traced in the final decision, and in guiding how the 'facts' from COI are used, their role in the decision-making process needs further analysis.
An alternative interpretation of the results is that courts are informed by the COI, but that the process of learning and deliberation that occurs is obscured from view when written decisions are examined. The end product -the decision -that draws on COI to support particular findings, does not necessarily reveal how the decision maker selected, considered, and evaluated the evidence. If this is the case, though, we still have little insight into how courts' positions are reached, which is an issue of transparency in itself. The way decisions are written may deserve more analytical scrutiny: for example, judges may draft decisions in a particular way to avoid committing an error of law. Such questions again go beyond the scope of this article, but may also play a part in the presentation of COI.
Yet another alternative could be that, when it comes to the application of COI, courts consider two interrelated 'levels' that influence each other. The positions that are assumed are informed by evidence contained in the COI; they rely on empirical premises (first level). However, the conclusions drawn by the courts (second level) do not follow logically from these premises. In other words, empirical premises derive from COI, but courts allow presuppositions to influence the way they approach and utilize evidence in the COI to draw their conclusions. For example, the empirical premise that the Syrian regime is under military pressure does not logically lead to a single conclusion on the behaviour of this regime towards returning draft evaders. The courts rely on assumptions to bridge this gap between an empirical premise and the conclusion

CON CLUS ION
Overall, the analysis here suggests that, by the time decisions are written, courts' approaches to evidence have often turned from genuinely inquisitive to instrumental, meaning that COI is applied according to pre-assumed positions, rather than testing these positions against the evidence. In the decisions examined in this article, a wide range of evidence was interpreted in opposite ways, including in relation to the ill-treatment of draft evaders, shortages of military manpower, and human rights violations. There were also many examples where courts referred selectively to a COI report, merely drawing on information useful for their reasoning. Alternatively, if they felt it necessary to cite a piece of evidence that did not support their reasoning, they drew on a selection of evaluative, framing, and citation techniques to reduce its impact. Neither the trend towards a 'standardization of knowledge', 117 nor the attempt to improve decision making by demanding the application of common standards, will solve this problem of evidence being adjusted as required.
What, then, might be done to address the apparent opacity in the use of COI in RSD decisions indicated in these results? One option might be to rely on national (rather than state-level) country guidance cases (as has been done in the UK, for example). There has been a lively discussion in Germany on this matter, although both scholars and legal practitioners stress that it would be difficult to implement. 118 While the decisions of Higher Administrative Courts can be considered as country guidance cases, there can be up to 15 such cases. Moreover, the independence of the courts is guaranteed by the Constitution. 119 As such, decisions of higher courts can never fully determine lower courts' decision making. While Administrative Courts must acknowledge a Higher Administrative Court's ruling, they are not bound by it. 120 A more viable option might then be the creation of a central COI research unit. 121 Such a unit would allow better integration of interdisciplinary and consistent country knowledge in RSD and reduce the likelihood that COI is rejected by courts, based on concerns about an author's partisanship. EASO's COI unit is an example of such a body providing independent COI reports. 122 A broader problem exists, however, in relation to how COI is understood and utilized by judges. Requiring judges to cite in particular ways, to locate information at particular parts in their decisions, or to use particular citation practices could be seen as a form of micro-management that threatens their autonomy and independence. What is more, even if this degree of procedural guidance or training were developed, 123 judges may still find ways around it, as described above. Without more consistency in practice, the textual techniques discussed here afford judicial decision makers a large degree of unregulated discretion. Such discretion is capable of rendering evidence incidental to a decision, which is a disconcerting finding. This can also be seen when it comes to Higher Administrative Courts' reactions to a more recent decision of the Court of Justice of the European Union (CJEU) on draft evaders from Syria, in which the CJEU states that there is a 'strong presumption' that draft evaders will be persecuted politically upon return. 124 The decision, however, has been treated and interpreted very differently by German courts, based on differing assessments of the situation in Syria. 125 Of course, a German case study has specificities that limit the extent to which the findings can be generalized to other countries. For instance, decision makers who consider the merits of asylum appeals in Australia and New Zealand are tribunal members rather than judges, and operate in a different legal context. However, this article exposes a general dilemma. Even though guidelines may not achieve consistency in how COI is used, there may nevertheless be a case for good practice guidelines on citation practices, evaluative reasoning, and writing techniques aimed at those drafting decisions in asylum cases. However, any procedural guidelines may very well not have their desired effect as long as judges and other decision makers have the discretion to focus less on what can be known through COI, and more on how it can be used to support their reasoning.  While courses offered by the body responsible for nationwide training of German judges, the German Judicial Academy (Deutsche Richterakademie), as well as by the European Judicial Training Network, increasingly cover so-called soft skills and the challenges of assessing credibility or burden of proof, there is no obligatory curriculum of continuing training for German judges. Training endeavours might therefore reach only a small number of judges. 124 Case C-238/19 EZ v Bundesrepublik Deutschland [2020] ECLI:EU:C:2020:945, para 57. 125 While, for instance, the OVG North Rhine-Westphalia (NRW), 22 March 2021, 14 A 3439/18.A upheld its rejection of refugee status, the OVG Berlin-Brandenburg (BB), 29 January 2021, 3 B 109.18 changed its jurisprudence in response to the CJEU decision and can now be considered a decision granting refugee status.
The Application of COI in Judicial RSD Decisions -Germany • Page 27 of 27