When Issue Salience Affects Adjudication: Evidence from Swiss Asylum Appeal Decisions

: Immigration is a top concern among citizens across the globe. Research shows that the salience of immigration shapes voters’ political behavior, but little is known about whether it influences judicial behavior. This article theorizes that variation in issue salience influences judges’ behavior when there is a clear connection between the legal and a generally salient, politicized issue. I test this argument drawing on all Swiss asylum appeal decisions reached between 2007 and 2015. I find that higher asylum salience leads judges to decide otherwise similar asylum appeals less favorably. This effect is not restricted to judges affiliated with anti-immigrant parties, unlikely to be driven by accountability pressures, and strongest for those topics known to drive anti-immigrant sentiment in the general public. Together, these findings raise concerns that issue salience threatens the consistency of judicial decisions. The materials required to verify the computational reproducibility of the results, procedures, and in this American Journal of Political Science Dataverse Harvard Dataverse

Although particularly salient in 2015, asylum and refugee issues have featured prominently in the media and political discourses across Europe for decades (see, e.g., Dancygier and Donnelly 2014;Sniderman, Hagendoorn, and Prior 2004). Research shows that the salience of an issue-how much public attention is devoted to an issue-matters for voters' attitudes toward that issue (Weaver 1991) and their electoral behavior (Lavine et al. 1996). This connection appears to be particularly strong for the issue of immigration (Sniderman, Hagendoorn, and Prior 2004). Studies document a positive relationship between immigration salience and anti-immigrant attitudes (Benesch et al. 2019;Hopkins 2010), as well as the weight of immigration attitudes in voting decisions (Dennison and Geddes 2019;Mader and Schoen 2019). We know that, at least indirectly due to its recognized importance for voters, parties and policy makers also re-spond to issue salience (see, e.g., Page and Shapiro 1983). Is the same true for judicial decision makers?
Work that researches effects of issue salience on judges' decisions is rare and includes only few studies that directly analyze whether the salience of the broader issue of judges' cases affects cases' outcomes. 1 Whereas Epstein et al. (2005) find no effect on the presence of war on Supreme Court decisions related to war, Philippe and Ouss (2018) establish that whether crime is covered in the main French evening newscast influences jurors' criminal case decisions, but not judges' . Do these null results extend to decision makers dealing with such a highly politicized and salient issue like immigration? This open question about the influence of issue salience on judicial decisions is particularly pressing with regard to asylum appeals, both for theoretical and policy reasons. First, because recent research finds that immigration (salience) is important for the behavior of voters and parties (see, e.g., Dennison and Geddes 2019), it is crucial to know whether it is similarly relevant for asylum decision makers. Second, although they are individually relatively low-salience judgments about administrative decisions, asylum appeal decisions make up large shares of the overall number of judgments in many European countries (see, e.g., Thomas 2011). They also frequently determine whether an asylum seeker will be allowed to stay in the country of asylum or forced to leave. Thus, if issue salience affects appeal decisions, the resulting overtime inconsistency in decision making could have dramatic consequences, especially if leading to a wrongful rejection.
This study argues that though experts on the issue of their cases, judges are likely to be influenced by shortterm variation in issue salience under the following scope conditions: if the issue is politicized, and there is a clear connection between the salient political and the legal issue. In the case of asylum appeals, this effect is theorized to be negative (i.e., reduce the share of decisions in favor of asylum seekers), due to the problem centeredness of asylum and refugee media coverage (Eberl et al. 2018), the frequent appearance of asylum issues among the top concerns in public opinion polls (Dennison and Geddes 2019), and the documented negative effect of immigration (salience) on voters' immigration attitudes (Benesch et al. 2019).
I test implications of this argument by exploring the effect of asylum salience on the universe of asylum appeal decisions at the Swiss Federal Administrative Court (FAC) between 2007 and 2015. Several features combine to make the Swiss FAC a useful case to study questions of issue salience, asylum attitudes, and judicial behavior.
First, rare data and procedural features permit the creation of high-frequency measures of both issue salience and judicial behavior. In Switzerland, asylum issues have been salient at least since the 1990s, when both the number of asylum requests and support for the anti-immigrant right began to rise (Ruedin and D'Amato 2015). 2 Despite the generally high level of asylum salience for decades, there is considerable variation in the volume of media coverage devoted to asylum and refugees, providing for a high-frequency measure of salience. The volume of asylum appeal decisions is considerable (≈3,000 per year), and appeals are centrally decided and spread out over time. In addition, the structure of the Swiss asy-lum appeal process removes one common obstacle in the way of a causal interpretation of issue salience effects: self-selection. At the FAC, neither judges nor asylum seekers can choose cases with high or low merit or time windows of high or low salience. Judges are exogenously assigned to judicial panels and positions within panels, and, there are relatively strict time limits for appeal filing, rendering it effectively impossible for appellants to choose a filing date as a function of issue salience.
Second, the unique combination of partisan judges with legislative judicial elections and retention elections in a multiparty system provides an institutional background against which I can explore the importance of judges' ideology and independence as potential moderators. Essentially all federal judges, including those handling asylum appeals, are affiliated with a political party and have to be reelected by the national parliament every 6 years. Compared to other contexts, the Swiss FAC therefore offers more fine-grained and directly observable variation in judges' ideology and independence. 3 Third, Switzerland is similar to other European countries when it comes to voters' concerns about immigration (Dennison and Geddes 2019), the extent and problem centeredness of media reporting thereon (Eberl et al. 2018), and the politicization of the judicial selection system. Even though other countries do not informally require judges to be affiliated with political parties, judicial selection systems across Western countries are frequently politicized in the sense that voters, legislators, or ministers are involved in the process (Geyh 2019;Seibert-Fohr 2012). Judges' ideology plays a stronger role in politicized judicial selection systems (Bonica and Sen 2017)-this is, at the time of writing, particularly visible in Central and Eastern Europe, where populist governments have succeeded in staffing courts "with loyalists and curtailing their independence" (Bugarič and Ginsburg 2016, p. 70).
Drawing on the case of the FAC, I gather robust evidence that higher asylum salience-measured as the number of articles on asylum issues in Swiss newspapers-leads asylum appeal judges to decide asylum appeals less favorably. The findings suggest that the success probability of an appeal decided in times of average asylum salience (≈14 circulation-weighted articles per day) is 3 to 6 percentage points higher than of a similar appeal decided in times of a one standard deviation higher asylum salience (+9, ≈23 circulation-weighted articles per day). There is no evidence to suggest that this effect is due to reverse causality (cases lodged during times of higher salience do not appear to have lower merits), or a spurious correlation (salience after the decision does not affect it, controlling for judges' caseload does not substantially reduce it). The shift toward more restrictive decisions during times of high salience occurs across the board, although it is a bit less pronounced for judges affiliated with the Social Democratic Party.
Additional analyses indicate that these effects are unlikely to be driven by accountability pressures and imply the presence of a subconscious/nonstrategic channel: There is no evidence that judges are responding to incentives created by the judicial retention system or are strategically choosing decision dates in response to issue salience. An automated text analysis of the articles on asylum issues in Swiss newspapers (using the structural topic model [STM]) reveals that the salience effects are larger when newspapers report more on topics that are in line with concerns documented to drive public antiimmigrant sentiment (cultural and security concerns).
These findings speak to the importance of investigating how contextual factors can shape judicial decision making. There is a vast literature focusing on how judges' identity, their ideological beliefs, and strategic considerations affect judicial decision making (Kastellec 2013;Sunstein et al. 2007;Voeten 2008), underscoring that "judges are people too" (Clark, Engst, and Staton 2018, p. 349). One relevant aspect of this insight is that if contextual factors matter for the (political) behavior of ordinary citizens, they might also be consequential for judges. Research on the effect of issue salience on judges' decisions, however, is relatively scant and has only recently received scholarly attention (see Blauberger et al. 2018;Canes-Wrone, Clark, and Semet 2018;Epstein et al. 2005;Philippe and Ouss 2018;Shayo and Zussman 2011). This article provides empirical evidence that, at least under certain conditions, judges' behavior can be influenced by the public attention devoted to the issue of their cases.
This study also adds to the literature on inconsistency in judicial decision making. A large body of work studies disparities between individual judges that lead to across-judge inconsistency in decisions including asylum decisions (see, e.g., Kim and Semet 2019;Miller, Keith, and Holmes 2014;Ramji-Nogales, Schoenholtz, and Schrag 2007). This article draws attention to overtime inconsistency in asylum appeal decisions induced by the short-term variation in asylum salience. Furthermore, it points toward the power of the media, political elites, and the public-the agenda setters (Weaver 1991)-in shaping judicial outcomes, because it suggests that whoever contributes to issue salience exerts some influence on asylum appeal decisions.

Issue Salience, Attitudes, and Judicial Behavior
Research has long shown that issue salience influences voters' political attitudes (Weaver 1991) and electoral behavior (Lavine et al. 1996). These effects appear to be magnified when it comes to immigration. Though not a new phenomenon, the "flash potential of anti-immigrant politics" (Sniderman, Hagendoorn, and Prior 2004, p. 35) has been particularly visible in recent years. A number of studies illustrate that the local presence of asylum seekers and the salience of asylum and refugee issues affect citizens' attitudes about asylum and immigration (Hangartner et al. 2019;Benesch et al. 2019) and their voting behavior (Dustmann, Vasiljeva, and Piil Damm 2019;Dennison 2019). Although these studies largely focus on voters, other papers indicate that issue salience effects extend to political elites (see, e.g., Epstein and Segal 2000). But what about judicial decision makers? Is their behavior influenced by the salience of the issue of their cases? Judges are usually required by law to decide cases "independently," "bound only by the law" (Article 2 of the Swiss Administrative Court Act), and treat everyone "equal[ly] before the law" (Article 8(1) of the Federal Constitution of the Swiss Confederation). Specifically with regard to the influence of public debates, the FAC Code of Judicial Conduct states that judges "shall not allow their judgments to be influenced by pressures exerted by the general public [. . .]." 4 Indeed, work on attitude change among experts and elites argues that individuals who are more sophisticated and have been exposed to more information and arguments tend to have more stable attitudes (Jennings 1992) and are less likely to be influenced by news content (Coppock, Ekins, and Kirby 2018;Zaller et al. 1992). This is in line with a recent study on the effect of criminal justice salience on crime case decisions in France (Philippe and Ouss 2018) that finds no effect of the criminal justice system being mentioned in the main French evening news on sentence lengths pronounced by judges, as opposed to jurors. Therefore, we would expect that judges are less likely to be influenced by the salience of the issue of their cases than jurors, or voters when it comes to their political behavior.
Work on judicial behavior, however, has documented that there are contexts in which judges do not act as mechanical interpreters of the law. Despite their expertise and the legal requirements mentioned above, they are influenced by political attitudes and contextual factors. Several studies have found that judges' identities influence case decisions (Boyd, Epstein, and Martin 2010;Kastellec 2013; and, with regard to asylum adjudication, Hangartner, Lauderdale, and Spirig 2020;Ramji-Nogales, Schoenholtz, and Schrag 2007). This relationship is stronger when there is a connection between an aspect of a judge's identity (e.g., race, gender) and the legal issue of the case (e.g., race or sex discrimination). Similarly, research shows that judges' ideology matters particularly when it comes to politicized issues, that is, ideologically contested areas of the law (Sunstein et al. 2007).
Past work also highlights that judges do not operate in isolation from public life. Shayo and Zussman (2011) show that higher local terrorism intensity in Israel leads to an increase in judicial in-group bias, and a handful of studies indicate that in the long run, judicial decisions reflect public opinion trends (see, e.g., Casillas, Enns, and Wohlfarth 2011;Epstein and Martin 2010), if the issue of decisions is politicized (Blauberger et al. 2018;Canes-Wrone, Clark, and Semet 2018). Taken together, there is ample evidence that judicial decision makers might not be outside the realm of public debate, in particular when the topic is politicized and salient.
Consequentially, if salience effects extend to judicial decision makers, we should at least expect them when an issue is politicized and salient, and there is a clear correspondence between the salient political and the legal issue. In the case of asylum, higher salience will push decisions toward the more restrictive, more anti-immigrant position, for two reasons. First, because asylum immigration is perceived as a problem: Media coverage on asylum and refugee issues is largely problem centered (see, e.g., Eberl et al. 2018) and citizens view asylum immigration as one of the most important problems (Dennison and Geddes 2019). Second, because recent studies show that natives become more antirefugee when the number of refugees (and thus the salience) increases (Dustmann, Vasiljeva, and Piil Damm 2019;Hangartner et al. 2019;Sniderman, Hagendoorn, and Prior 2004).
Even if asylum judges' topical expertise and experience shielded them from direct, subconscious asylum salience influence, there are several aspects of the judicial system that could create incentives for judges to choose to at least appear as if they were acting in line with the general perception of the issue (expressed in the media). As a political economy strand of the judicial politics literature emphasizes, judges respond strategically to incentives created by the institutions they are embedded in (see, e.g., Clark, Engst, and Staton 2018;Epstein, Landes, and Posner 2013). With regard to issue salience, accountability-ensuring mechanisms are of particular relevance, because they incentivize judges to consider the preferences of those they are accountable to. One such mechanism is the retention system, incentivizing judges to signal responsiveness to the preferences of those determining their retention. To increase reelection chances in popular elections, nonpartisan judges have been found to decide criminal cases more punitively as media coverage increases and citizens are more likely to be informed about their decisions (Lim, Snyder, and Strömberg 2015) and as the reelection date approaches (Huber and Gordon 2004). Similar mechanisms appear to be at play in U.S. systems with gubernatorial or legislative reappointment (Gray 2017;Shepherd 2009), where judges from minority parties need the support of the majority party. In a multiparty system with legislative retention elections of partisan judges, all judges will require support from MPs of other parties (if no party has a majority). Accordingly, we would expect that higher issue salience incentivizes in particular judges not affiliated with issue-owning parties (or those sharing similar policy positions) to respond to issue salience in the direction of public concern.
The importance that judges attribute to their court's legitimacy (Casillas, Enns, and Wohlfarth 2011) constitutes another, more implicit, accountability mechanism. Although the extent to which the court's institutional legitimacy incentivizes judges to act strategically is still under debate (see, e.g., Giles, Blackstone, and Vining 2008), it would follow from this argument that in times of higher asylum salience, judges would want to avoid getting called out by the media, politicians, or voters for being too lenient and going against the public's will. In sum, although there are reasons to expect that issue salience effects are smaller for judicial decision makers than voters, jurors, or lay judges, this article puts forth several channels through which asylum salience could shape judicial decision making on asylum appeals.

Asylum Appeal Decisions at the Swiss FAC
The FAC handles all appeals lodged against asylum decisions. The appeals deal with a variety of legal issues (e.g., initial asylum decision, removal), but are exclusively about asylum matters and lodged against the Swiss government's migration agency (SEM) that decides asylum requests or previous FAC rulings. 5 The approximately 40 FAC judges are elected into judicial office by the United Federal Assembly, the joint assembly of both houses of the Swiss parliament. Almost all judges are members (or at least supporters) of one of the larger parties represented in the parliament. The informal requirement to be affiliated with a political party is connected to the idea that courts should be representative of Swiss society in terms of its sociopolitical views. The parliament elects judges for terms of 6 years and there are no term limits until mandatory retirement at 68 years. The legislative reelections are essentially uncompetitive retention elections-no federal judge has ever not been reelected, and there are no challengers. 6 After the appeal is lodged and received by the FAC, the court's central chancellery assigns it to one of the two asylum divisions. Conditional on criteria that are exogenous to the merit of the appeal, software that was specifically designed for the court randomly chooses judges for positions on a three-judge panel: chair, second, and third judge. The chair judge's role is similar to the role of an opinion writer, as she is the one who receives the complete case file first and effectively decides which of three sequential, file-based decision-making procedures is used. Given the sequentiality of the procedure, it might come as no surprise that the best simple aggregation rule to explain how judges at the FAC combine their individual preferences into a joint panel decision is that the chair judge's preference is decisive (Hangartner, Lauderdale, and Spirig 2020).

Data and Measurement
Appeals and Judges Data. The asylum appeals data set obtained from the FAC covers the universe of 41,040 Swiss asylum appeal decisions reached between January 1, 2007, and December 31, 2015, including unpublished decisions. The data contain a unique case id, appeal filing and decision dates, the language in which the decision was written, the legal issue of the case, the appellant's country of origin, and the names of the involved judges. These data were merged with biographical information about the judges (such as the judge's date of legislative election to the court, gender, and party affiliation, if any) and additional information about the subset of published cases (whether the appellant had a legal representative). 7 The main analysis focuses on 31,937 decisions, of which half took 30 days or less to decide (the minimum duration is 1 day, the maximum duration is 2,276 days) and on average 12% were granted. 8 The appeal verdicts are coded as "granted" if the substantive (content) part of the appeal decision favors the asylum seeker.
Asylum Salience. In this article, issue salience refers to how much public attention is devoted to an issue, thereby following closely the definition laid out in Wlezien (2005) and Collins and Cooper (2012), where salience refers to the importance that voters attribute to a particular issue. To measure salience, I use the amount of newspaper coverage of asylum issues on the national level. 9 In the case of asylum, a highly politicized issue in Switzerland and many other countries, I argue that this captures both how important the issue is to voters and how concerned citizens are about the issue. Figure 1 shows that evidence from public opinion polls supports this interpretation: The correlation between the share of people who count asylum and refugee issues among one of the top five problems, and the yearly number of newspaper articles on asylum issues is 0.73.
The asylum coverage data set is based on a collection of all articles on asylum and refugee issues in Switzerland published in Swiss newspapers that were available on the online data repository Factiva for the entire study period. 10 Among these 19 newspapers are 7 of the 10 most widely circulated daily newspapers, the 3 most widely circulated Sunday newspapers, and 2 influential weekly news magazines. To account for the relative influence of newspapers and approximate national-level salience, I merge the about 36,900 remaining articles with 2009 circulation data (WEMF AG 2010). Figure 1 provides an overview of the temporal distribution of the articles, illustrating that there is sizable variation in asylum salience over time. It also shows that the coverage of asylum issues is not driven by judges' decisions: The subset of articles that mention the FAC is very small. I employ these data to compute several appealspecific measures of issue salience. The two main measures are the average daily number of circulationweighted articles in the asylum coverage data set during the entire duration of the appeal procedure (Salience (total period)) and in the month before the decision is finalized (Salience (pre-decision month)). I use both measures, because we cannot know when exactly during the decision-making process a decision is reached. Salience (total period) has the benefit of averaging over the entire decision-making period, thereby accounting for the fact that it is unclear at what point in time between appeal filing and decision date the fate of the case was decided. A downside of this measure, however, is that only cases that are decided within a very short period can reach extreme levels of salience (see Table 1) and that the distribution of case-specific asylum salience in the data is skewed right.
To address these concerns, I also use a second measure, Salience (pre-decision month). Estimates based on Salience (pre-decision month) provide a lower bound because they capture asylum salience effects only for those appeals where case success was determined at some point in the month before the decision date. (The fate of cases that lasted for more than a month (≈50% of appeals) was possibly decided before the last month. Arguably, these cases were largely unaffected by issue salience in the month prior to the decision date. This introduces measurement error that tends to bias the corresponding estimates toward 0 and therefore provides a lower bound.) As outlined in more detail in SI Appendix B (pp. 5-7), I compute additional salience measures for supplementary analyses. First, I further vary the time window within which salience is measured. Second, I vary the content (the topics) for which salience is measured. More specifically, I compute topic cluster-specific salience measures to gauge whether it is the salience of particular asylum-related concerns, for example, crime and security concerns, that drive the effect of asylum salience on asylum appeal decisions. To do so, I apply an STM (see, e.g., Roberts, Stewart, and Airoldi 2016) to the German-language subset (83%) of the asylum coverage data set. This approach allows me to generate estimates of article-specific topic shares. There are several topic modeling frameworks, but the STM is well suited for the purpose of this study, because it supports the inclusion of covariates according to which article-specific topic shares and topic-specific words can vary. For a corpus that contains articles from national and regional newspapers, such as the asylum coverage data set, accounting for newspapers' regional foci and articles' publishing date facilitates the identification of meaningful themes. I estimate a model with 80 topics and subsequently group them into 12 topic clusters, thereby relying on common themes identified in studies of media coverage of asylum and refugee issues (see Eberl et al. 2018). The share of articles focusing on refugee accommodation ("Allocation/NIMBY" 11 ), for example, is used as a measure of the salience of local refugee allocation-related concerns.

Empirical Strategy
To study whether asylum salience affects asylum appeal decisions, I start by regressing the probability of an appeal to be granted on measures of asylum salience (Salience): logit (π i jctq ) = αSalience i + ι j Judge j + γ c Country c where Salience i captures asylum salience when each case i is being decided and Country c , Year t and Quarter q are fixed effects for each origin country c, decision year t, and quarter q. I include these fixed effects to adjust for general changes in the grant rate over time, as well as seasonal and country-specific idiosyncrasies. Chair judge fixed effects (Judge) account for the variation in judges' individual preferences. In a second step, taking into account that judges' identity might interact in relevant ways with issue salience, I interact Salience with some of the chair judge's characteristics, including party affiliation. To gauge lower and upper bounds for the effect of asylum salience on grant rates, I perform the main analyses for two measures of asylum salience (Salience (total period) and Salience (pre-decision month)), focusing on different time windows. In addition, as SI Appendices C (pp. 11-14) and 11 "not in my backyard." E (pp. 19-20) discuss in more detail, I conduct a variety of robustness tests, including running models with other measures of case-specific issue salience, fewer and additional fixed effects (e.g., for file language, legal representation, legal issue), controls for potential macrolevel drivers of asylum salience (unemployment, asylum requests), and falsification tests, to support the validity of the main results.
Under what assumptions are the issue salience effects causally identified? One of the identifying assumptions is that the effect does not only emerge after the inclusion of a suppressor variable (Lenz and Sahn 2020). Model 1 in Table 2 includes only chair judge fixed effects to show that the effect is not driven by suppression. A second assumption is that there is no omitted variable that has an independent effect on judges' asylum appeal decisions and is correlated with issue salience. If there were such a variable, I would misattribute its effect to issue salience. One variable that is potentially correlated with asylum salience and could have a similar negative effect on asylum appeal decisions is caseload. One could imagine that when more people seek refuge in a country, both asylum salience and judges' caseload increase. Although I find that caseload is negatively associated with a case's probability to be granted, controlling for caseload does not materially change the estimated asylum salience effect (see SI Table C4, p. 14, for more details).
A third identifying assumption is that issue salience does not influence the assignment of cases to judges. This would be a problem if, for example, some judges could choose cases with lower merit during high issue salience times, which would make them appear more restrictive. Court-internal software, however, ensures that judges cannot choose their own cases (see SI Appendix A.2, pp. 2-3, and Hangartner, Lauderdale, and Spirig (2020) for more details on the exogenous assignment procedure and balance tests).
A fourth identifying assumption is that issue salience does not drive case merit. If it were the case that in times of higher issue salience, appeals with on average lower case merits are lodged, an observed negative effect of issue salience would be due to selection, rather than judges' behavior. Even though this is a minor concern given the rather strict time limits to appeal first instance asylum decisions, there might be other reasons why case merit and issue salience are not orthogonal. SI Table C1 (p. 11) examines whether an appeal's predicted success, based on a set of case-specific indicators, varies by issue salience. The results show that there is no negative relationship between asylum salience and predicted case successan indication that the estimates are not plagued by reverse causation. Note: Coefficients are average marginal effects from logistic regressions. In the main models (models 2 and 4), a one standard deviation increase (≈9 circulation-weighted articles per day) at average salience (≈14 circulation-weighted articles per day) leads to a 6.1 (3.2) percentage point decrease in the grant rate, respectively. Standard errors (in parentheses) are clustered at the chair judge level in all models.

Main Effect of Asylum Salience
I estimate the overall effect of asylum salience on asylum judges' grant rates with two different measures of asylum salience: Salience (total period)-the average daily number of circulation-weighted articles in the asylum coverage data set during the entire decision-making periodand Salience (pre-decision month)-the average daily number of circulation-weighted articles in the month before the decision was finalized. As Table 2 shows, I find a significant negative effect of asylum salience on an asylum appeal's probability to be granted for both measures. Models 2 and 4 report results from the main specification as outlined in Equation (1), models 1, 3, and 5 include different combinations of fixed effects and illustrate that the results are robust across a variety of specifications. All estimates are average marginal effects: An estimate of −0.004 (−0.002) implies that on average, the probability of an appeal to be granted decreases by 0.4 (0.2) percentage points with a one-unit increase in Salience (total period) (Salience (pre-decision month)).
At first glance, these effects might appear to be small. In the context of an average 12% grant rate and substantial variation in issue salience over the study period, however, they constitute substantial negative effects. The predicted success probability of an appeal handled in times of average asylum salience (≈14 circulation-weighted articles per day) is 3.2 to 6.1 percentage points higher than that of a comparable appeal handled in times of one standard deviation above average asylum salience (≈23 circulation-weighted articles per day), depending on the asylum salience measure used (corresponding to models 4 and 2 in Table 2). For the case of a Sri Lankan asylum seeker, handled by the court's median judge, decided in the first quarter of the year 2014, this would imply a 13 to 24% reduction in the success probability (from ≈25% to 19-22%). Substantively, this effect (of a one standard deviation increase in asylum salience) is comparable to the reduction in an appeal's probability to be granted if it were decided by a judge affiliated with the anti-immigrant Swiss People's Party (SVP) instead of the center-right Free Democratic Party (FDP). 12 Further robustness tests and falsification checks are presented in SI Appendix C (pp. 11-14). First, I replicate the main specification from Equation (1) with linear probability and random effect models. The results remain unchanged. Second, I use different measures of salience (during 1, 2, and 3 months before the appeal decision) and perform a placebo check by including controls for issue salience over the same number of months after the appeal decision. The results from these regressions indicate that the effect of issue salience is not a spurious correlation: It does not go away if a measure of salience after the decision is included and it is not affected by asylum salience after a decision is made.

Effect of Asylum Salience by Judge Characteristics
The analyses presented in this section take into account judges' party affiliation, gender, and experience. They provide evidence of an effect that is not restricted to judges affiliated with particular parties or little experience. I start by considering chair judges' party affiliation. As the estimated average marginal effects displayed in Figure 2 show, issue salience decreases an appeal's probability to be granted regardless of the chair judge's party affiliation. 13 The predicted probabilities by partisanship depicted in Figure 3 illustrate in more detail that there is 13 All coefficients of the JudgeParty×Salience interactionsregardless of the definition used for Salience-are substantial, neg-variation-a one standard deviation increase of Salience (pre-decision month) at its mean is associated with a decrease in the grant rate of between 2.3 (Social Democcrats (SP)) and 5.3 (Christian Democrats (CVP)) percentage points-but the effect applies to judges from all parties, from the left (Green Party (GPS), SP) all the way to the right (SVP). 14 Additional results in Figure 2 show that similar to party affiliation, seniority and gender do not appear to be crucial moderators of the effect of issue salience on judicial decisions. The average marginal effects of issue salience are a little smaller for senior (and female) judges. 15 ative, and reach statistical significance at least at the 10% level (see SI Table D1, p. 15).

FIGURE 3 Appeal Grant Rate by Party and Salience
Note: Graphs show asylum appeals' success probability by asylum salience and the chair judge's party affiliation. Quarter is set to the first quarter of the year, Year to 2014, and Country to Sri Lanka. The Green Liberal Party (GLP) is excluded because they only have one asylum judge who joined the FAC in 2013. The black lines above the x-axis indicate the distribution of appeals across asylum salience, measured as Salience (pre-decision month). The shaded areas are 95% confidence bands, based on standard errors that are clustered at the chair judge level.

Mechanisms
How can we make sense of a sweeping, across-the-board negative asylum salience effect? The literature on judicial behavior stresses two main parallels between regular citizens' and judicial decision makers' behavior that are potentially relevant: first, the importance of strategic considerations in the work environment and second of political attitudes in decision making. The first parallel highlights the possibility of a conscious, strategic underpinning of judicial responsiveness to issue salience. The second parallel instead assumes a nonstrategic, potentially subconscious, response to issue salience, similar to the one experienced by voters. Along this line of thought, judges would become more restrictive in times of higher asylum salience, because of short-term shifts in their political attitudes driven by "the same social forces that influence the public" (Casillas, Enns, and Wohlfarth 2011, p. 75).
Although investigating subconscious processes with behavioral data is notoriously difficult, I can derive testable implications from the conscious/strategic channel argument. Existing research provides suggestive evidence that judges (strategically) decide nonsalient cases in line with public opinion because they value the court's institutional legitimacy (Casillas, Enns, and Wohlfarth 2011). Judges' more tangible accountability, however, is to the body that reelects or reappoints them (if there is any). Research suggests that, if it increases their retention chances, judges are more likely to decide cases in a way that appeals to the body in charge of retention as the reelection/reappointment date approaches (Gray 2017;Huber and Gordon 2004). Accordingly, if asylum appeal judges responded to issue salience because of reelection incentives, we would expect to see that the negative asylum salience effect on appeal outcomes is stronger in the year before reelection than after reelection. In the case of a multiparty legislature as reelection body, where judges are in need of support from MPs outside of their own party, we would additionally expect that the issue salience effect is stronger for judges who are not affiliated with issue owning parties (or parties with similar issue positions) in the year before reelection. In the Swiss context, this suggests that we would expect larger effects for nonpartisan and center-left judges.
As results in SI Appendix D (pp. 17-18) show, I find no evidence that the issue salience effect is due to incentives created by the retention system. The estimated effect is not smaller when I drop decisions made in the year of (legislative) reelection from the analy-sis (models 1-2 in SI Table D4, p. 18), and it is not stronger in the year before the 2011 legislative retention elections compared to the year after-not even for independent or center-left judges (see SI Figure D2, p. 17). These results suggest that judges' response to issue salience is unlikely to be conditioned by a system of uncompetitive legislative retention elections. If judges' reactions to issue salience are strategic and meant to signal accountability, they are more likely to be signals to the general public than legislators deciding on their fate.
Another question is which of the topics (or frames) that constitute asylum issues drive the observed effect. Past research on issue salience, immigration attitudes and voting behavior has put forth three main explanations for citizens' hardening immigration attitudes in response to issue salience and the presence of immigrants: cultural concerns, security concerns, and economic concerns (see, e.g., Sniderman, Hagendoorn, and Prior 2004). Cultural concerns speak to people's fear that local values and norms might change due to an increase in the number of immigrants. Security concerns refer to the fear that arriving immigrants might contribute to an increase in crime and terrorism. Economic concerns relate, especially with regard to asylum migration, to the economic burden immigrants are perceived to put on locals and the host country. Although the debate about which of these drivers is most important is far from settled, all of them are closely mirrored in the media coverage of asylum and refugee issues as several studies document (for an overview, see Eberl et al. 2018). If judges responded to issue salience particularly when the issues that have been found to drive voters' anti-immigrant sentiment are covered, that would be a piece of evidence at least consistent with a subconscious, nonstrategic mechanism.
I use topic cluster-specific salience measures to examine whether there are differences in the effect of asylum salience depending on the topics covered in newspaper articles on asylum and refugees. As discussed more extensively in SI Appendix B (pp. 5-7), these measures are based on the estimated proportion of each topic in each article. For the salience of a set of theoretically interesting topic clusters, Figure 4 shows average marginal effects on appeals' probability to be granted.
I take the debates about (potential) asylum center locations ("Allocation/NIMBY") as most closely related to citizens' cultural concerns. One of the central themes of the coverage related to asylum seeker allocation is the intense and outspoken local opposition to Note: Graph displays average marginal effects of the overall asylum salience (All Coverage), salience measured based on the German-language subset of articles on asylum issues (German-Language Coverage) and the salience of a particular topic cluster on the probability of an asylum appeal to be granted. The topic cluster-specific estimates stems from a single logistic regression (based on the specification outlined in Equation (1)) that includes all five topic cluster-specific salience measures. The lines are 95% confidence intervals. asylum seeker reception. 16 The literature on the drivers of opposition to receiving asylum seekers suggests that although economic and security concerns play a role, the preservation of homogeneous communities and antimulticulturalism are frequently crucial motives in local anti-asylum seeker movements (Haselbacher and Rosenberger 2018). The topic cluster labeled "Crime & Threat" connects to security concerns (terrorism, Ebola, and crime), "Cost" to debates about the economic cost of asylum seekers, and "Politics" contains the political dimension of the issue (it being featured in debates about party politics, elections, campaigning, and more). The topic cluster that has the largest negative effect on appeals' grant rate is "Allocation/NIMBY," followed by "Crime & Threat" frames. Debates about economic costs, politics, and humanitarian issues related to asylum and refugees do not, or only to a smaller extent, have a negative effect on their own. If newspaper articles dealing with the local accommodation and reception of asylum seekers increase by one circulation-weighted article per day, an appeal's probability to be granted decreases by about 2.3 percentage points (see SI Table D3, p. 16, for point estimates).
Accordingly, the topic clusters that are closely related to prominent explanations of anti-immigration attitudes in the literature are stronger drivers of the effect of issue salience than less closely related topics, such as "Politics" and "Humanitarianism." The finding that there is a stronger response to topics such as refugee accommodation is consistent with two different explanations: the subconscious hardening of judges' asylum attitudes and a conscious choice to signal accountability to the public. How can we differentiate between the two? One way to shed light on this question is to think about the "costs" associated with deviating from preferred grant rates. Several studies assume that judges incur a cost if they (consciously) vote against their preferred outcome (see, e.g., Fischman 2011). 17 Accordingly, for judges who are consciously responding to accountability-ensuring mechanisms, it would make sense to produce decisions that appear in line with the public's preference for more restrictive decisions, without actually being so. One way to appear as if one produced more restrictive decisions in times of high issue salience would be to strategically influence the duration of the decision-making period depending on whether one intends to grant or reject the appeal. 18 This would create the appearance that more restrictive decisions are reached in times of high salience. Consequentially, we should observe that issue salience increases the duration of cases that will eventually be granted and decreases the duration of those that will be rejected.
Because the outcome variable is duration of the appeal, I employ a case-specific Cox proportional hazard (results from ordinary least squares [OLS] regression models are similar) to test this implication. Its results provide no evidence for strategic decision timing: Regardless of whether a case is eventually granted or rejected, higher issue salience is associated with a shorter duration (see SI Table D4, p. 18, columns 3-5). Together with the previous analyses, these results imply that the observed effect of asylum salience on judicial behavior is less consistent with an account that stresses the expert status and a purely strategic, conscious reaction of judicial decision makers than with an account that takes into consideration a potentially subconscious, short-term hardening of judges' asylum attitudes (similar to the "fleeting effects" caused by priming and framing described in Coppock, Ekins, and Kirby 2018, p. 83). Whether the increase in judges' restrictiveness is exclusively due to subconscious or partly also due to conscious processes cannot be determined conclusively here and is a question left open for future research. I would like to note, however, that it appears we would have to resort to rather unconventional arguments to square the empirical evidence with the existence of a conscious/strategic channel. For it to be the case, judges would, for example, have to respond to issue salience to maintain the court's legitimacy either subconsciously so or nonstrategically (e.g., if judges preferred a costly deviation from their preferred grant rate in low salience times to the strategic choice of decision dates).

Conclusion
In recent decades, asylum and refugee issues have been highly salient among citizens and political elites in many countries around the globe. Research has shown that political elites respond to issue salience (see Epstein and Segal 2000) and that citizens become more anti-immigrant in times of increasing numbers of immigrants when issue salience is high (Hopkins 2010). Building on these insights, this study addresses the question whether judicial decision makers, such as asylum judges at the Swiss FAC, also react to issue salience.
I collect robust evidence that when issue salience increases and newspapers report more on asylum and refugee issues, asylum appeal judges make more restrictive decisions. Asylum appeals are more likely to be rejected when they are decided in times of higher asylum salience. This effect is robust across many model specifications, falsification tests, and measures of salience. The effect is not restricted to judges affiliated with left-wing pro-immigrant (GPS, SP) or right-wing anti-immigrant parties (SVP) and not explained by higher caseload in times of high salience. Furthermore, it does not appear to be driven by accountability mechanisms incentivizing judges to consider the preferences of the body that reelects them or the public. Rather, the findings suggest that at least a part of the effect is due to a potentially subconscious, short-term hardening of judges' asylum attitudes.
These findings have important implications for our theoretical understanding of judicial decision making and the consequences of issue salience. So far, much of the literature has focused on between-judge inconsistency and questions (normative and empirical) about the convolution of long-term public opinion trends and judicial decisions, but has not as extensively discussed (short-term) consequences of other contextual factors. This article shows that one such factor, issue salience, can influence judges' behavior and, thereby, induce overtime inconsistency in judicial decisions. Beyond these theoretical implications, these findings have policy ramifications. If "[j]udges shall not allow their judgments to be influenced by pressures exerted by the general public," 19 as stated in the FAC's Code of Conduct, these are particularly pressing in times when both the number of arriving asylum seekers and the level of asylum salience are high.
Although we cannot directly infer from these findings that issue salience generally affects judicial behavior, this study shows that at least in the case of asylum appeals in Switzerland, where judges are usually affiliated with political parties and reelected by a multiparty legislature, asylum seekers' appeals suffer from asylum salience. How generalizable are these findings? Does issue salience affect judicial behavior in other areas of the law and in other countries? Although both questions about the external validity of my results are challenging to assess, the first appears to be a bit more straightforward. There are several other issues that are politicized, gener-ally salient, and exhibit a clear connection to a legal issue: At the time of writing, this includes, for example, sex and race discrimination, at least in some Western countries. 20 One relevant characteristic of asylum appeals, however, might be that written proof is particularly rare and questions relating to the credibility of asylum seekers' claims frequently lie at the heart of asylum appeal decisions (Thomas 2006). If this makes for a relatively larger inherent uncertainty about asylum case merit, and, therefore, a larger realm of justifiable interpretations, we might expect smaller effects of issue salience in other, similarly salient and controversial, areas of the law. Given that the results are more compatible with a subconscious, nonstrategic mechanism, a last dimension to consider is perhaps how routine, repetitive, and quick (as opposed to rare and with major policy implications) the decisionmaking process of a legal issue is. Research shows that intuitions play a larger role in routine decisions (for which fewer resources are available), rendering them more vulnerable to implicit, subconscious influences (Applegarth 2018;Guthrie, Rachlinski, and Wistrich 2007). Like asylum appeals, many sex and race discrimination cases will be decided in routine procedures at lower courts, for which we would expect larger issue salience effects than for cases at higher courts.
Another possibility is that the Swiss judicial system and media landscape provide an institutional background against which the formation of an asylum salience effect is much more likely compared to other national contexts. Although Switzerland shares many relevant similarities with other Western countries (considerable number of asylum appeals, high political and public asylum salience, extensive and problem-centered media coverage of the issue, right-wing anti-immigrant issue-owning parties), there are two distinguishing features of the Swiss asylum appeal procedure: that judges are affiliated with political parties and that they need to be reelected by a multiparty legislature every 6 years. How relevant are these factors for the effect of issue salience on judicial decisions?
Had the results in this article suggested that there are stark differences in how judges respond to issue salience based on their party affiliation or a decision's proximity to the legislative reelection date, it would have implied that the documented effects could be product of the 20 According to 2018 Eurobarometer data, other issues that citizens across Europe are concerned about are crime, terrorism, social security, unemployment benefits, see https://ec. europa.eu/commfrontoffice/publicopinion/index.cfm/ResultDoc/ download/DocumentKy/83548. Dennison (2019) finds, however, that although the Western European populist right benefitted from immigration salience, it did not from the salience of crime, terrorism, or unemployment issues. unique Swiss conditions. None of the results, however, point in this direction. Rather, they suggest that we may expect similar effects in other countries that have politicized judges (which is not an artifact of the Swiss system, but rather of a politicized judicial selection system, see, e.g., Bonica and Sen 2017) decide on asylum appeals. 21 One aspect that the Swiss FAC shares with some (e.g., the U.S. Board of Immigration Appeals), but not other courts (e.g., German regional administrative courts, partly First-Tier Asylum Tribunals in the United Kingdom) is the de facto absence of face-to-face contact between judges and asylum seekers. 22 Research on the "person-positivity bias" (Sears 1983) and on effects of oral hearings on asylum appeal decision suggests that negative asylum salience effects could be smaller where face-to-face hearings exist. 23 In sum, although the context considered here might provide an upper bound for the size of issue salience effects, the relationship between issue salience and judicial decisions is likely of importance in other Western countries as well. If an issue is politicized, salient, and there is a clear connection to a legal issue, we should expect that short-term changes in issue salience affect politicized judges' decisions in cases on that issue. By studying other courts in other countries, future research should examine questions about the relevance of institutional features and further aspects relating to external validity, such as whether the effect of issue salience travels to other politically salient areas of the law. I hope that the research design employed here can serve as a blueprint for future studies.