We examine the role of judicial policy preferences in influencing whether judges in Nazi Germany sentenced defendants charged with serious political offences to death. We find that judicial policy preferences, measured by the depth of the ideological commitment of the judge to the Nazi Party worldview, were an important determinant of whether judges imposed the death sentence. Judges more committed to the Nazi Party were more likely to impose the death sentence on defendants belonging to organised political opposition groups, those accused of violent resistance and those with characteristics to which Nazism was intolerant.

There is a growing body of research examining the manner in which authoritarian states use courts as instruments of governance (Moustafa, 2014). This literature has found that courts are frequently deployed to exercise state power vis‐à‐vis opposition groups (Moustafa, 2007; Rajah, 2012). Regime‐related theories, developed in political science, suggest that authoritarian regimes use the courts to further their political goals through establishing a dual legal structure in which they leave the ordinary judiciary largely to exercise independence in the routine administration of justice, while transferring politically sensitive cases to courts specifically established to hear such cases (Toharia, 1975; Linz, 2000; Guarnieri, 2010). In such politicised courts, judges are expected to implement the will of the regime. But such expectations are not always fulfilled. Indeed, there is an emerging literature that suggests that some judges in authoritarian courts behave more as agents for change than puppets of the regime. For example, Ip (2012, p. 331) finds that the Supreme People's Court of China has acted ‘with considerable autonomy to influence an ever‐widening range of policy domains’. In another example, Moustafa (2007) asserts that the Egyptian Supreme Constitutional Court has been responsible for an expansion of political rights in that country, contrary to the interest of the authoritarian state. Ip (2011, p. 1) refers to several other such instances as well, maintaining that courts ‘in at least some authoritarian polities are more autonomous and activist than previously assumed, occasionally defying their regimes, sometimes with impunity’. The main point here is that there is mounting evidence that judges in authoritarian regimes do not necessarily always exercise their discretion to advance the interest of the state. If one accepts this point, two questions naturally arise: what determines how judges will behave in political courts in authoritarian regimes and to what extent judges do, in fact, implement the will of the authoritarian state? We answer these questions by examining the factors that determined whether judges in Nazi Germany sentenced to death defendants charged with high treason and treason, the most serious of political offences. Nazi Germany provides a good example of the bifurcation of the criminal justice system in one‐party states. When the Nazis came to power in January 1933, cases of treason and high treason were tried before the Supreme Court (Reichsgericht). Thus, between September and December 1933, it was the Supreme Court which heard charges of high treason against defendants alleged to have burned down the Reichstag (Germany's parliament) on 27 February 1933. This was a politically sensitive case for the Nazis because the fire, which they blamed on Communists, had been the catalyst for sweeping legislation that had dramatically diminished the rights of those opposed to the regime. For the state, the trial did not go as hoped. Four of the five defendants were acquitted for lack of evidence. Hitler later described the verdict as ‘laughable’ and the judges as ‘senile’ (Picker, 1951). In response, in April 1934, the Nazis enacted further legislation establishing the People's Court (Volksgreichtshof) and transferred jurisdiction for cases of treason and high treason to it. In so doing, the regime created a court that was specifically designed to hear serious political offences. Traditionally, authoritarian states, fearful of judicial review, have been regarded as reluctant to cede independence to the courts. More recently, however, there has been increasing recognition that some authoritarian regimes may be willing to provide autonomy to their judiciaries. In Nazi Germany, judges of the People's Court, for example, were given a high level of discretion over sentencing. Sentencing options were elastic with respect to both the type and severity of punishment handed down. Moreover, the state provided no specific guidelines on sentencing, meaning that judges were free to impose sentences for treason and high treason that increased in severity up to capital punishment. In addition, since decisions of the People's Court (and Supreme Court before August 1934) were final, no appeal was possible (Koch, 1989; Marxen, 1994; Zarusky, 2011; Geerling et al., 2016). The literature developed to explain judicial behaviour posits a range of competing theories on how judges exercise discretion. Ongoing debate exists over whether judges are guided more by the law or ideology. At one end of the spectrum, formalists contend that judges simply interpret and apply the law without reference to their own preferences. At the other end, there are those who subscribe to attitudinal models of judicial behaviour, which give a central role to the policy preferences of judges as measured by their political ideology. A midway position is occupied by agency theorists who, while recognising the importance of judicial policy preferences, suggest that judges’ ability to implement their preferences are constrained by the institutional environment as well as oversight by courts higher in the judicial hierarchy (Epstein and Knight, 1998; Segal and Spaeth, 2002; Garoupa et al., 2013). We test the attitudinal model using data on the imposition of the death sentence for treason and high treason in the Supreme Court and People's Court from the time the Nazis came to power on 30 January 1933 until the end of World War II. About 98% of the defendants in our sample were sentenced in the People's Court from 1934–45.1 Our hypothesis is that because the People's Court was established by the Nazi regime to hear offences against the state, by its very nature it was highly politicised and that judicial policy preferences (reflected in the judge's political ideology) should play an important role in how cases were decided. To put it differently, our contention is that the extent to which judges were willing to implement the will of the authoritarian state depended on how closely their own political ideology aligned with that of the regime. The potentially central role of judicial policy preferences was reinforced by the fact that the applicable laws regarding the sentence to be imposed for treason and high treason were vague and that decisions of the People's Court were not subject to review by courts higher in the judicial hierarchy. The judges who heard the cases in our sample were a fairly homogenous group. All but one of the 26 judges in our sample was a member of the Nazi Party or one of its ancillary organisations.2 Merely being a member of the Nazi Party, though, did not necessarily signal strong ideological commitment to Nazi values. It could be motivated by other factors, such as professional advancement. Thus, in order to differentiate between judges in terms of their ideological commitment to the Nazi worldview, we use a binary variable set equal to 1 if the judge joined the Nazi Party before the Nazis came to power in January 1933. Hereafter, these pre‐1933 judges will be referred to by the German term for such early members of the Nazi party, the Alte Kämpfer (old fighters). We assume here that the Alte Kämpfer held personal views that were strongly aligned to the values of the Nazi Party. Indeed, given that they had all joined the party before its seizure of power, their dedication to the movement and its causes was in most cases deeply held. The first step of our analysis examines the role of the policy preferences of the judges in influencing the likelihood that a defendant charged with treason or high treason received the death sentence. All judges of the Supreme Court and People's Court were white, ethnically German males. Hence, there is no variation in the gender or race of the judges in the sample. In all cases, however, we control for other important judicial characteristics such as the judge's age, professional background (had they served on the Bench during the Weimar Republic, prior to the Nazis’ coming to power?) and religion (were they raised as Roman Catholics?), in addition to our main variable of interest, the ideological commitment of the judge to Nazism (did they join the party before 1933?) We also control for the offence type, defendant's prior criminal history, including prior political and non‐political offences, demographic and human capital characteristics and political affiliation. We then interact our proxy for judicial ideology with various characteristics of the defendant, to examine whether judges who were more ideologically committed were more inclined to impose the death penalty on members of the better organised oppositional groups, those engaged in violent resistance against the state and individuals with characteristics against which the Nazis had an ideological prejudice or showed intolerance. This step of the analysis provides a direct test of whether those judges most ideologically aligned to the Nazi regime exercised their policy preferences so as to further the interests of that state. We find that overall Alte Kämpfer were 3.64 percentage points more likely to impose a death sentence than other judges. As a check on our results, instead of employing a dummy variable for Alte Kämpfer status, we use years of Nazi Party membership. We find that for each additional year of Party membership, judges were 1.22 percentage points more likely to sentence a defendant to death. We find that relative to their less ideologically committed brethren, Alte Kämpfer were 17.4 percentage points more likely to impose the death sentence on those engaged in violent resistance. Similarly, Alte Kämpfer were 5.9 and 4.8 percentage points more likely to impose the death penalty on members of the comintern‐aligned Communist and social democratic parties respectively. The targeting of these parties by Alte Kämpfer made political sense. From the regime's perspective, the greatest and most imminent threat for most of its existence lay in the two mass left‐wing parties of the period, the Communists (KPD/KPÖ) and social democrats (SPD/SPÖ). These were parties that the Alte Kämpfer had literally fought on the streets during the years of the Weimar Republic and, in the 1930 and 1940s, they remained the political movements best positioned to challenge Nazi supremacy. It was not until the final stages of the war that most conservative and nationalist political movements broke from the consensus that underpinned Hitler's rule (Weitz, 2009, pp. 81–127; Benz, 2014, pp. 18–22). We also show that while the Alte Kämpfer were more likely to target Communists overall, the Court's treatment of Communists was more nuanced and varied over time. Specifically, we find that relative to the period following Operation Barbarossa in June 1941, Communists were 13.41 percentage points less likely to receive the death penalty prior to World War II, and 8.23 percentage points less likely to receive the death penalty in the period when the non‐aggression pact with the Soviet Union was in force (August 1939 to June 1941). Hence, there is evidence that in the earlier years of the regime and during the period that the Nazis had an agreement in place with the Soviets, they executed fewer Communists, either in an attempt to avoid an escalation of resistance or build bridges to individuals in the Communist movement who might still be convinced to accept National Socialist rule. Among judges with an ideological commitment to Nazism, we also find considerable evidence of bias against defendants with specific characteristics. Compared to judges who did not join the Nazi Party before 1933, the Alte Kämpfer were thus more likely to sentence devout Roman Catholics (24.7 percentage points), defendants with partial Jewish ancestry (34.8 percentage points), juveniles (23.4 percentage points), the unemployed (4.9 percentage points) and foreigners (42.3 percentage points) to death. The hatred of extremist Nazis like the Alte Kämpfer towards Jews and foreigners more generally is, of course, well known. Yet, Nazi ideology was also deeply antagonistic towards all organised religion, especially those varieties, like Roman Catholicism, which sought to maintain their independence from Nazi control and influence. Individuals who attempted to assert Roman Catholic values in place of Nazi values were thus harshly persecuted. Furthermore, core to the Nazi worldview was the belief that those elements of German society which acted to weaken or hold it back from attaining its National Socialist destiny should be quickly and mercilessly expunged. As a consequence, one would expect that, unlike judges of other persuasions who might regard relative youth and economic disadvantage as mitigating factors at times of sentencing, the ideologically driven Alte Kämpfer would have shown markedly less sympathy or tolerance to juveniles or the unemployed who had chosen to challenge the regime's right to rule.3 We also examine whether judges who were exposed to distinct historical periods (the Revolution of 1918–9 and period of hyperinflation – June 1921 to January 1924), which may have shaped these judges views with respect to Nazism, were more likely to impose the death sentence. We find that judges who came of age during the hyperinflation were 8.3 percentage points more likely to sentence a defendant to death and that Alte Kämpfer, whose hometown or suburb lay in a centre of the Revolution of 1918–9 were 5.8 percentage points more likely to sentence a defendant to death. Our findings are important because they provide insights into the practical realities of judicial empowerment by providing rare empirical evidence on how the exercise of judicial discretion in authoritarian states is reflected in sentencing outcomes. We find that judicial policy preferences, measured by the political ideology of the judge, did matter in Nazi Germany. This, however, only tells part of the story. Courts set up by authoritarian states to sentence the political opponents of the regime are often depicted as kangaroo courts. Indeed, initial post‐war assessments of the People's Court portrayed it as little more than a vehicle for the show trials of the regime's most prominent opponents. It was labelled as merely a ‘court in name only’, a ‘terror court’ or ‘blood tribunal’ with largely preordained outcomes (Rüping, 1984, 1985; Jahntz and Kähne, 1992). While few authors would contend that the People's Court judgments were impartial or that its judges were not subservient to the wishes of the regime, we offer here a more nuanced interpretation of its history. Our results show that, rather than having a preordained outcome in which death was the only permissible sentence irrespective of mitigating circumstances, a range of judicial and defendant characteristics as well as the offence committed influenced whether the death sentence was imposed or not. The existing economics literature on sentencing in capital cases has focused mainly on gender and racial disparities in the likelihood of being sentenced to death for homicide in the United States (Iyengar, 2011; Alesina and La Ferrara, 2014; Kastellec, 2015). In a related study, Argys and Mocan (2004) examine the factors that influence whether individuals already on death row for committing homicide in the United States end up being executed. Kubik and Moran (2003) consider the role that election‐year political considerations have in determining the timing of executions in the United States. We extend this literature on the determinants of receiving, or implementing, the death penalty in several directions. First, instead of situating the study in a modern day democracy (the United States) as existing studies do, we extend this literature to examine the imposition of the death penalty in a court of an authoritarian regime of historical significance. While there is a growing interest in the use of capital punishment in modern day authoritarian states, our understanding of what determines whether courts in such states impose the death sentence is scant. A lack of information is an issue here. While Amnesty International does produce limited data on death sentences,4 information on judge and defendant characteristics in specific cases is hard, if not impossible, to come by. By studying a politicised court in an historically important authoritarian state, we shed light on sentencing more generally in authoritarian states. Second, instead of focusing on cases of homicide, our sample consists of individuals convicted of the most serious political offences (treason and high treason). As such, ours is one of the first studies to examine the determinants of sentencing in a political court, established to sentence political opponents of an authoritarian regime. In addition to factors considered in previous studies (demographic characteristics of the defendant, offence type and prior criminal history), we explore for the first time the role of both the political affiliation of the defendant and the ideological orientation of the sentencing judge.5 Third, our research contributes to the literature on the workings of the People's Court itself. As noted above, earlier histories of the court depicted it purely as an instrument of terror. Recent studies based on archival research, however, have challenged this overly simplistic assessment, emphasising instead its continuities with German legal practices (Gruchmann, 1988; Marxen, 1994; Schlüter, 1995). Our results lend further weight to this more nuanced assessment of the People's Court. Our study is related to Geerling et al. (2016) which also examines the determinants of sentencing for treason and high treason in Nazi Germany. We, however, differ from that study in important respects. First, we analyse cases over a longer time frame. We consider all such cases until the end of World War II; Geerling et al. (2016) examines cases only from the pre‐war years. Second, our focus is on the imposition of the death penalty; Geerling et al. (2016) looks at sentencing more generally. Third, and perhaps most importantly, we examine the role of both defendant and judicial characteristics in influencing sentence outcomes; Geerling et al. (2016) only considers defendant characteristics. Given that Geerling et al. (2016) does not consider judicial characteristics, they do not examine the role of judges’ policy preferences in sentencing, which is the key objective of this study. The article proceeds as follows. Section 1 discusses the institutional background. Section 2 outlines the positive political theory of criminal sentencing and our key testable propositions. The data and method are outlined in Section 3. The main empirical results are presented in Section 4 and we offer robustness checks in Section 5. Section 6 concludes.

1 Institutional Background When the Nazis came to power in January 1933, Articles 80–92 of the Reichsstrafgesetzbuch (Federal Penal Code), first promulgated in 1871, defined the acts of high treason and treason in Germany. High treason consists of those acts that internally undermine the power and integrity of the state, while treason concerns itself with acts which harm the state externally. The code defined three categories of high treason: attacks that threatened the head of state; attempts to change or alter the territory of the federal state; and violations of the constitution (Lauf, 1994; Marxen, 1994; Schlüter, 1995). The Reichstag fire and subsequent trial provided the impetus for the restructuring of the German judicial system. In April 1934, the Law Amending Provisions of Criminal Law and Criminal Procedure created the People's Court and transferred jurisdiction of high treason and treason to it from the Supreme Court. Each case of treason or high treason in the People's Court was heard before a single presiding judge. In addition to the presiding judge, four others sat on the case. Three were lay members of the Court, drawn from the Nazi Party and its ancillary organisations to provide ‘political expertise’; the fourth was another judge (Zarusky and Mehringer, 1998; Gruchmann, 1988). While it was hoped that each trial would be conducted on a collegiate basis, with each member required to sign the final decision, in practice the court operated according to the Führer principle, with the presiding judge dominant (Koch, 1989, p. 53). The role of other members of the panel in the trial was purely advisory; they neither voted on guilt nor decided on the punishment imposed. These were the prerogative of the presiding judge, who also provided the formal reasons for his decision. As Freisler explained in 1941: Next to the prosecution it would be the [presiding] judge's task to see to it that everything at issue would be cleared up swiftly and thoroughly. He was to lead and to decide; fellow professional as well as lay judges representing the Volk could advise but not decide. Responsibility lay with the presiding judge alone (cited in Koch, 1989, p. 81). Hence, in all practical respects, decisions of the People's Court were decisions of a single judge.6 The adoption of the Law Amending Provisions of Criminal Law and Criminal Procedure brought all decrees with respect to high treason and treason into line. The death penalty was applicable to all crimes listed in articles 80–83. The People's Court now decided in the first and last instance in all matters which previously came under the jurisdiction of the Supreme Court. No appeal against a decision of the presiding judge was possible. The presiding judge was also given significant flexibility in determining punishment. In all cases, judges were free to hand down sentences that increased in severity: prison, penitentiary with hard labour and the loss of civil rights and capital punishment (Zarusky and Mehringer, 1998).

2 Positive Political Theory of Criminal Sentencing In the literature on positive political theory, judges are modelled as strategic policy maximisers who seek to maximise their preferences, subject to constraints imposed by other factors, such as adherence to legal doctrine (Schanzenbach and Tiller, 2006; Shavell, 2006; Cross et al., 2012). Such an approach has been used to model the sentencing preferences of US judges under Federal Sentencing Guidelines (Schanzenbach and Tiller, 2006; Fischman and Schanzenbach, 2011). Our conceptual framework draws on these studies, although there are important differences, given that these studies model sentencing preferences of judges in a democratic common law country (the United States) sitting in Federal District courts and subject to appeal to higher courts in the judicial hierarchy. In our conceptual set up, we assume that there are two actors: the authoritarian Nazi regime with a preference for remaining in power and judges of the Supreme Court and People's Court who have their own preferences over the sentencing outcomes. Thus, for a given crime and set of defendant characteristics, some judges would prefer to impose the death penalty while others may opt for a lighter sentence. We assume that the preferences of different judges are derived from their ideological beliefs. This approach is consistent with the US literature which asserts that policy preferences on sentencing are reflected in political persuasion, namely, whether the judge in question is a Democrat or Republican. This relationship is typically proxied by looking at whether the appointing President was a Democrat or Republican. In the US context, the conventional wisdom is that Democrat judges prefer more lenient sentences for street crimes and harsher sentences for white collar crimes than their Republican counterparts (Schanzenbach and Tiller, 2006). Rather than the Democrat/Republican dichotomy, in this article, we use the depth of the judge's ideological commitment to the Nazi Party to measure his policy preferences. To the extent that judges have discretion in applying the law, they will do so in a manner consistent with their view of the world, informed by their ideological policy preferences. We assume that the ideology of Alte Kämpfer was closely aligned with the interests of the Party and Nazi state, and, as a result, they were more likely to exercise their discretion to reinforce the position and status of the regime. It is worth noting that our measure of ideological preference is much stronger than that typically used in US studies, because it relies on the judge actually joining the Nazi Party in the period before there was any professional advantage in so doing. We take it as given that the main objective of the authoritarian state is to remain in power and stamp out opposition to its rule. In the courts, this objective manifests itself in several preferred sentencing outcomes, which can be summarised as follows. First, the authoritarian state would generally prefer capital punishment for those opposed to it. The proviso to this statement is that sometimes the authoritarian regime may benefit from the judge being more lenient than his inclination would dictate, for example when dealing with milder offences of people who can be reformed or from environments in which people can be swayed by a display of mercy. It is generally recognised that if the Nazi regime had tried to kill all of its adversaries from the start, it risked stiffer and more violent opposition. Instead it tried for some time, where possible, to convince with leniency individual opponents to reconsider their positions, cease resistance and accept Nazi rule. Second, the authoritarian state would prefer the death sentence for members of the most organised opposition groups – particularly if its strategic attempts to display mercy failed to convince many to stop their resistance – given that they represent the greatest threat to the state. Third, the authoritarian state would prefer the death sentence for those engaged in violent resistance against it. Fourth, the authoritarian state would prefer harsher sentences up to, and including, the death sentence for those against whom the regime had an ideological bias. In Nazi Germany, these prejudices included inter alia those of partial Jewish ancestry, foreigners, juveniles, the long‐term unemployed and Roman Catholics. In positive political theory, judges express their sentencing preferences subject to having to comply with legal doctrine. In common law countries, legal doctrine is usually taken to be adherence to a system of precedent and fear of judicial reversal by a Court higher in the hierarchy. This was not the case in Nazi Germany, however, where there was no judicial review of decisions of the People's Court. Consistent with the Germanic civil law tradition, judges of the Court provided written reasons for their decision but there was no system of precedent and cases were not decided with reference to earlier cases. While legal obedience is less of a constraint in a court not subject to appeal, judges are still required to act within the constraints of the legislation, so it is not totally irrelevant (Cross et al., 2012). In Nazi Germany, adherence to legal doctrine meant in practice that judges had to decide cases with reference to the legislation governing sentencing for treason and high treason. The relevant legislation was the Law Amending Provisions of Criminal Law and Criminal Procedure, enacted in April 1934. Articles 80–83 of the Penal Code made a broad range of treasonous or high treasonous offences punishable through either long terms of incarceration or capital punishment. This degree of flexibility in sentencing meant that the judges had enormous discretion and that adherence to legal doctrine prescribed in the Penal Code imposed only a very weak constraint. Judges possessed significant latitude in determining the type and severity of punishment imposed. Some of the literature on courts in authoritarian regimes has focused on the circumstances in which judges exercise their preferences in defiance of the regime (Ip, 2011, 2012). If the starting point of analysis is that the preferences of judges as a group differ from those of the regime, one could begin with the assumption that the preference of judges is to impose fewer death sentences than preferred by the state. If one takes this approach, it follows that if the Alte Kämpfer wanted to impose fewer death sentences, they could more easily do so than other judges because their loyalty to the regime was beyond question. As a result, their deliberations would not be as likely to come on the radar of the state. However, for non‐Alte Kämpfer judges, most of whom were also members of the Nazi Party, it might be argued that their loyalty to the state could be questioned because they had joined the Party later, potentially making the Party leadership suspicious of their motives and commitment. In a principal (state) agent (judiciary) sense, the non‐Alte Kämpfer may have been subject to a higher level of monitoring; there was no equivalent principal agent problem for the Alte Kämpfer. If one accepts this reasoning, one might then actually expect non‐Alte Kämpfer judges to sentence more people to death than Alte Kämpfer judges in order to demonstrate their loyalty to the state and because they were less able to get away with imposing a lighter sentence consistent with their preferences. While we accept this is one way to view the situation, we do not see judicial behaviour in Nazi Germany as a principal agent problem but rather as an issue of how judges acted at the margin of discretion. All of the judges were personally appointed by Hitler, with advice given by the Justice Minister. The mindset of all the judges on the People's Court was broadly aligned to that of the regime.7 The non‐Alte Kämpfer judges were certainly not rebels. Hence, we do not see it as particularly useful to view sentencing behaviour in Nazi Germany in terms of defiance of the state, as opposed to the exercise of judicial discretion at the margin. It was this discretion that was influenced by ideological commitment. Our hypothesis that the extent to which judges of the Supreme Court and People's Court were willing to implement the will of the Nazi regime depended on their political ideology, or degree of political commitment to the authoritarian state, suggests the following testable propositions: Proposition 1.Alte Kämpfer will be more likely to impose the death penalty for treason and high treason. Proposition 2.Alte Kämpfer will be more likely to choose capital punishment for defendants who were members of the most regime‐threatening oppositional groups. Proposition 3.Alte Kämpfer will be more likely to choose capital punishment for defendants who were engaged in violent resistance against the regime. Proposition 4.Alte Kämpfer will be more likely to apply the death penalty to individuals with characteristics against which the Nazi Party had a strong ideological prejudice. Proposition 5.The Nazi regime will benefit from judges being more lenient to individual defendants than their inclination might otherwise dictate in certain situations if such leniency is conceived as being conducive to avoiding even stiffer opposition to the regime. The psychology literature suggests that individuals’ identify formation are most affected by major events in early adulthood (Grotevant, 1987). Previous research in the economics and finance literature suggest that early life events impact on the management styles of CEOs (Malmendier et al., 2011; Schoar and Zuo, 2017). We examine whether sentencing outcomes were affected by discrete historical events in the judges’ early adulthood that may have had a profound psychological impact and, thus, particularly shaped these judges’ views with respect to Nazism. The two historical events are as follows: Was the judge's home town or suburb located in a centre of the Revolution of 1918–9? Was the judge aged 18–21 during the period of hyperinflation (June 1921 to January 1924)? We expect that judges whose hometown was in a centre of the 1918–9 Revolution or were aged 18–21 during the period of hyperinflation to be more anti‐Communist, anti‐democratic and anti‐Weimar in their outlooks. As each of these characteristics was core to Nazism, possessing them in strength was consistent with a deeper holding of Nazi values. We expect that judges whose hometown was in a centre of the 1918–9 Revolution or were aged 18–21 during the period of hyperinflation to be more anti‐Communist, anti‐democratic and anti‐Weimar in their outlooks. As each of these characteristics was core to Nazism, possessing them in strength was consistent with a deeper holding of Nazi values. Proposition 6.Judges affected by discrete historical events in their early adulthood that are likely to have contributed to a deeper sense of Nazism will be more likely to impose the death penalty for treason and high treason.

3 Data and Method 3.1 Data The data come from the official state records of cases tried before the People's Court and Supreme Court, 1933–45, as collated by the Resistance as High Treason (Widerstand als Hochverrat) project. The Widerstand als Hochverrat microfiche series comprises around 70,000 pages of court files that document the judicial prosecution of more than 6,000 men and women charged with high treason and treason who appeared before the central courts of the Third Reich. This study focuses on cases where resistance took place within the territorial boundaries of Germany following the Anschluss (union) with Austria in March 1938. Altogether, this amounted to 1,334 cases presided over by 35 judges and involving 4,177 defendants. Twenty‐seven of these judges served in the People's Court. To measure a judge's ideological commitment to Nazism, we sought details about their background and career. The Bundesarchiv (Federal Archive) in Berlin contains personal files of judges who served on the Supreme Court and People's Court as well as Nazi Party membership records which survived the war. We were able to locate records for 26 of the 35 judges (21 of the 27 judges from the People's Court) and supplemented this knowledge with leading legal encyclopaedias of the Third Reich (Klee, 2003). These files provided information on the judges’ age, career prior to the Nazis coming to power (whether they had served on the bench during the Weimar Republic), whether they joined the Nazi Party (if so, when) and whether they were brought up as a Roman Catholic. In total, the final sample consisted of 26 judges who presided over 1,157 cases involving 3,579 defendants; that is, 85.68% of all the defendants charged with treason and high treason for activities carried out within Germany.8 For each of the 26 judges in the sample, Table 1 provides information on their period on the Court, the date they joined the Nazi Party (if available), the number of defendants in the sample whom they sentenced and the proportion of defendants in the sample whom they sentenced to death. Of the 26 judges, seven were Alte Kämpfer. Of the six judges who sentenced more than 40% of defendants to death, half were Alte Kämpfer. The two judges who sentenced more than 50% of defendants to death (Freisler and Merten) were both Alte Kämpfer.9 At the other end of the spectrum, the ten judges who did not impose any death sentences were all non‐Alte Kämpfer. In terms of simple averages, the 19 non‐Alte Kämpfer sentenced, on average, 26.3 people to death and the seven Alte Kämpfer sentenced 76.9 people to death. Among the Alte Kämpfer, Thierack, who sentenced just 5% of the defendants who appeared before him to death, appears to be relatively ‘mild’. Table 1. List of Judges Included in the Sample Name of Judge Alte Kämpfer? Number of defendants in sample sentenced Proportion of defendants in sample sentenced to death Date joined Nazi Party+ Years on Court Albrecht No 450 0.44 1 May 1933 1938–45 Bruner No 63 0.00 1934–6 Coninx No 1 0.00 1934–34 Crohne Yes 217 0.36 1 September 1932 1941–4 Diescher No 79 0.49 1 May 1937 1942–4 Diester No 3 0.00 1 April 1933 1938–38 Engert Yes 463 0.23 2 March 1927 1936–42 Freisler Yes 199 0.61 9 July 1925 1941–4 Granzow No 202 0.34 1 May 1933 1940–4 Greulich No 22 0.00 1 May 1933 1936–44 Groβpietsch No 3 1.00 1 May 1938 1944–44 Hartmann No 779 0.21 1 May 1933 1934–44 Illner No 11 0.18 1 November 1938 1942–4 Köhler No 100 0.30 1 May 1933 1935–44 Lämmle No 223 0.09 1 May 1933 1935–45 Linz No 19 0.00 1933–4 Löhmann No 68 0.43 1 May 1933 1934–44 Mengelkoch No 7 0.00 1933–4 Merten Yes 311 0.51 1 December 1932 1938–44 Niethammer No 1 0.00 1933–33 Rheinisch No 45 0.00 1933–4 Schaad No 79 0.00 1 May 1933 1934–6 Springmann No 3 0.00 1 May 1933 1935–6 Stier Yes 142 0.41 1 June 1932 1943–4 Thierack Yes 41 0.05 1 August 1932 1936–42 Zieger Yes 48 0.27 1 December 31 1942–3 Overall Proportion Yes = 0.269 Total defendants sentenced: 3,579 Proportion sentenced to death = 0.306 Thierack is one of the best known Nazi jurists. Following his time as President of the People's Court, he went on to become the Reich Minister of Justice. Previously, he had been leader of the National Socialist jurists’ organisation, the so‐called Rechtswahrerbund and Justice Minister in Saxony. While he is known as a fervent and committed Nazi, he was not a fanatic, such as Freisler. Perhaps, given his ministerial and broader leadership experience, one might conclude he was someone – within the Nazi context – who was more considered and balanced. In late 1943, he cautioned Freisler for his extremism (Wagner, 1974, pp. 26–27). Moreover, it should be remembered that he was in the People's Court (June 1936–August 1942) before its most radical period and sentenced only 41 defendants. These were probably the crucial factors in explaining the relatively low proportion of death sentences he imposed. The reason why some judges’ tenure ended in 1934 is that the People's Court was established that year and when the People's Court was established, many Supreme Court judges were pensioned off. Other judges’ tenure ended in 1936. In 1936, the status of the People's Court was changed to that of an ordinary constitutional court, which stood alongside the Supreme Court. Some of the older judges were also pensioned off at this time. The main reasons so many judges stopped in 1944 was that fewer cases were heard in 1945, due to lost territory and record keeping at the end of the war was far less conscientiously carried out. In many cases, the records from 1945 were so limited for the cases that were heard that we could not include them in our sample for lack of basic information. One might be concerned that if more lenient judges resigned or were forced to resign over time, then there still could be some selection going on. If we only focus on the non‐Alte Kämpfer judges who served relatively longer periods on people's court, however, we do not see any evidence of lenient judges being forced out of the People's Court. Consider the following non‐AK judges and served for comparable periods on people's court. Greulich (1936–44), Lämmle (1935–45) and Hartmann (1934–44) sentenced around 20%, or less, of defendants to death, whereas Albrecht (1938–45), Köhler (1935–44) and Löhmann (1934–44) sentenced 30–40% of defendants to death. Table 2 contains data on the frequency of capital punishment for the sample as a whole as well as whether the judge was an Alte Kämpfer. Of the 3,579 defendants in our sample 1,094, or 30.6%, received the death sentence. Alte Kämpfer sentenced 37.8% of defendants appearing before them to death, while their brethren who had not joined the Nazi Party prior to 1933 sentenced 25.7% of defendants to death. The difference in means of the proportion of defendants sentenced to death between Alte Kämpfer and non‐Alte Kämpfer (12.1%) is significant at 1%. Table 2. Death Sentence Death sentence Alte Kämpfer Yes No Total Yes 15.03% 15.53% 30.56% [N = 538] [N = 556] [N = 1,094] No 24.67% 44.76% 69.43% [N = 883] [N = 1,602] [N = 2,485] Total 39.70% 60.29% 100% [N = 1,421] [N = 2,158] [N = 3,579] Proportion death penalty 0.378 0.257 Difference in means (0.013) (0.009) 0.121*** (0.015) Figure 1 shows the distributions of the percentage of death sentences for Alte Kämpfer and non‐Alte Kämpfer judges pre and post‐1939 for those judges who sentenced at least 50 defendants. For both the pre‐war period and during World War II, visually the kernel density for Alte Kämpfer dominates the kernel density for non‐Alte Kämpfer; however, the Kolmogorov–Smirnov test for equality of distribution indicates no statistically significant difference for either period. Figure 2 plots the percentage of death sentences imposed (on the Y axis) against the number of years that the judge had been in the Nazi Party (on the X axis). It suggests a positive relationship between the length of time that the judge had been a Party member and the proportion of defendants whom he sentenced to death. Figure 1 Open in figure viewerPowerPoint Kernel Density Plot of Distribution of the Percentage of Death Sentences Given by Alte Kämpfer and Non‐Alte Kämpfer Judges Notes. Sample restricted to judges who decided at least 50 cases. Colour figure can be viewed at wileyonlinelibrary.com. Figure 2 Open in figure viewerPowerPoint Percentage of Death Sentences Versus Number of Years in the Nazi Party Note. Colour figure can be viewed at wileyonlinelibrary.com Table 3 contains information on independent variables. A description of each variable is contained in online Appendix A (Table A.2). In just under 40% of cases, the presiding judge had joined the Nazi Party prior to 1933. We also have data on the judge's age (mean age was 54.88 years), whether the judge had been on the bench during the Weimar Republic (49.62% had) and whether the judge was brought up as a Roman Catholic (2.91% had been). Table 3. Descriptive Statistics for Independent Variables Continuous independent variables Variable Observations Mean SD Min Max Age 3,579 37.78 10.50 16 77 Judge age 3,579 54.88 6.77 41 67 Dummy variables Variable Frequency Percent Defendant/case characteristics Education 630 17.60 Male 3,112 86.95 Unemployed 297 8.30 Partial Jewish ancestry 120 3.35 Foreigner 72 2.01 Juvenile 56 1.56 Catholic 275 7.68 Treason only 102 2.85 Minor high treason only 248 6.93 Major high treason only 1,861 52.00 Multiple offences 1,368 38.22 Violence 286 7.99 Pre‐major political 288 8.05 Pre‐minor political 407 11.37 Pre‐non‐political 570 15.93 Comintern Communist 2,374 66.33 Social Democratic 165 4.61 Other left wing 362 10.11 Right wing 492 13.75 Member of Nazi Party 92 2.57 Served in WWII 187 5.22 Judge characteristics Judge Weimar 1,776 49.62 Judge Catholic 104 2.91 Alte Kämpfer 1,421 39.70 Hometown 1918 Revolution 2,681 74.91 Hyperinflation 142 3.97 Time/region effects Sentenced 1933 21 0.59 Sentenced 1934 117 3.27 Sentenced 1935 200 5.59 Sentenced 1936 233 6.51 Sentenced 1937 184 5.14 Sentenced 1938 187 5.22 Sentenced 1939 135 3.77 Sentenced 1940 113 3.16 Sentenced 1941 125 3.49 Sentenced 1942 508 14.19 Sentenced 1943 661 18.47 Sentenced 1944 950 26.54 Sentenced 1945 145 4.05 Bavaria 179 5.00 Berlin 1,032 28.83 Brandenburg 53 1.48 Bremen 39 1.09 Hamburg 252 7.04 Hessen 124 3.46 Lower Saxony 127 3.55 Mecklenburg 24 0.67 North Rhine‐Westphalia 443 12.38 Rhineland 73 2.04 Saarland 93 2.60 Saxony 438 12.24 Saxony‐Anhalt 124 3.46 Schleswig Holstein 71 1.98 Thüringen 56 1.56 East Prussia 21 0.59 Silesia 68 1.90 We have data for a number of demographic, ethnic and human capital characteristics of the defendant. The mean age of defendants was 37.78 years, 17.6% had completed high school or above, 86.95% were male, 8.3% were unemployed, 3.35% were of partial Jewish ancestry, 2.01% were foreigners, 1.56% were juveniles and 7.68% were Roman Catholic. We also know the severity of the offence with which the defendant was charged: treason (2.85%), minor high treason (6.93%), major high treason (52%) or multiple offences (38.22%) and if the offence was committed with violence (7.99%). Information on the defendant's prior convictions (political and non‐political offences), the year in which the defendant was sentenced, as well as the region in which their resistance activities took place is available as well. Finally, we have data on the defendant's oppositional political affiliation, broken into four classifications. ‘Comintern Communist’ denotes membership of either the Communist Party of Germany (KPD) or Communist Party of Austria (KPÖ), whereas ‘Social Democratic’ indicates that the defendant was a member of either the Social Democratic Party of Germany (SPD) or the Social Democratic Party of Austria (SPÖ). The KPD/KPÖ and SPD/SPÖ were the main organised left‐wing opposition groups to the Nazis. ‘Other left‐wing’ here refers to those who were members of one of numerous smaller, less well‐organised left‐wing opposition groups, such as the Revolutionary Socialists,10 Anarchists/Syndicalists,11 Trotsksyists,12 KPDO,13 or other Communist splinter groups and self‐help organisations.14 ‘Right‐wing’ denotes that the defendant belonged to a right‐wing or conservative group opposed to the Nazi regime: namely, the Schwarze Front,15 BJ,16 Catholic group,17 National Conservatives18 as well as those who participated in the 20 July plot,19 the White Rose20 and the Legitimists movement in Austria.21 Table 4 compares the judge characteristics of Alte Kämpfer and non‐Alte Kämpfer. If Alte Kämpfer were more experienced, they might have received more complicated cases leading to a higher likelihood of death sentence. The only characteristic, though, for which there is a significant difference between Alte Kämpfer and non‐Alte Kämpfer is being Roman Catholic. For each of the possible proxies for experience (age, served as a judge in the Weimar Republic, years of experience on the People's Court and total years of experience as a judge on any court) there is no significant difference between Alte Kämpfer and non‐Alte Kämpfer. Hence, it is unlikely that any systematic difference between Alte Kämpfer and non‐Alte Kämpfer is behind the observed differences in death sentencing rates in Table 2 and Figure 1. Table 4. Differences in Characteristics of Alte Kämpfer and Non‐Alte Kämpfer Judges Judge characteristics Mean Difference (3) = (1) − (2) Alte Kämpfer Non‐Alte Kämpfer (1) (2) Age 52.13 55.49 −3.38 (8.26) (7.33) [3.35] Roman Catholic 0 0.32 −0.32* (0.48) [0.18] Judge in Weimar Republic 0.71 0.68 0.03 (0.48) (0.47) [0.21] Years of experience in People's Court 1.98 3.01 −1.03 (1.56) (2.48) [1.03] Total years of experience as a judge on any court† 29.57 24.05 5.52 (7.73) (6.41) [4.59] 3.2 Random Assignment The validity of our findings rest on the assumption that the Court used random assignment of cases to judges, implying that sentencing outcomes can be fairly compared across judges (Ashenfelter et al., 1995; Abrams et al., 2012; Eisenberg et al., 2012; Sorensen et al., 2012; Lim et al., 2016). On first impressions, such an assumption appears questionable. The People's Court was divided into senates, each of which with defined responsibilities. At its inception, the Court had three senates: senates 1 and 2 were responsible for high treason cases and senate 3 for treason cases. As the number of cases steadily increased, a fourth senate – specialising in treason cases – was established from November 1935. On 1 November 1941, a fifth senate was added and in December 1942, a sixth senate, both of which heard cases of treason and high treason (Wagner, 1974, p. 26; Koch, 1989, p. 46). In practice, however, this apparent division of labour between senates proved largely illusory. Over its history, Presidents of the Court redefined the responsibilities of each of the senates on numerous occasions. From the annual business plans of the court, one can identify between 1934 and 1945 at least two major changes of foci for each senate, with the exception of senate 6, the last to be created. Thus, the allocation of duties between senates for 1939 shows senate 1, originally a pure high treason senate, taking on cases of treason for the first time. All high treason and treason cases from Austria and cases of economic sabotage were also added to its jurisdiction at this time. In addition, senate 2 was now assigned cases based on Section 5 of the first decree from 28 February 1933 (arson, use of poison gas, floods and damage to railways) and senate 4 with cases involving damage to military property. From October 1942, senates 3 and 4 became pure treason senates; from 1 January 1943, only senate 3 was confined to cases of treason. A short time later, however, senate 3 had to take on a wider variety of cases, too. Moreover, by the latter stages of the war, senates 5 and 6 had taken over responsibility for most cases from Austria and Bavaria (Wagner, 1974, p. 26; Koch, 1989, p. 46). The jurisdictional confusion created by this shifting foci of the senates was compounded by the regular intervention of Freisler in his role as president of the Court. During his presidency, cases of the highest profile from the regime's perspective were assigned to senate 1 (his senate) or senate 2, irrespective of their nature and circumstances. He also reserved for himself the right to have the last word in all cases in which there were any discrepancies or uncertainty as to jurisdiction. In effect, such interventions served only to override and obfuscate the precise responsibilities of the different senates (Wagner, 1974, p. 26; Wieland, 1989, pp. 80–81). Moreover, Freisler's actions, which eventually induced the Minister of Justice, Thierack, in November 1943 to demand an end to the arbitrary jurisdictional extension of senate 1, further reinforced an already existing tendency for cases to be concentrated in just two increasingly generalist senates: senates 1 and 2 (Wagner, 1974, pp. 26–27). Given that that those charged with treason or subversion almost invariably were charged with high treason as well, most defendants perforce ended up before senates 1 and 2, the only senates empowered to hear such a combination of charges. Indeed, nearly three quarters of all cases in our sample came before senates 1 and 2 and only 1.5% before the two courts dedicated to hear treason cases: senates 3 and 4. Case allocation was thus not based on any strong or abiding notion of senates having clear specialisations based on either case or defendant characteristics. As Wagner (1974, p. 26), a leading authority on the history of the court notes in this respect: ‘the multiplication of senates, the constant growth in the number of cases, the fact that particular cases involved both high treason and treason, as well as the Freisler type of leadership brought with it an ever stronger blurring of jurisdiction within the individual senates’. Koch (1989, p. 229), another historian of the court, concluded similarly: ‘the division of jurisdiction between the various senates was not always clearly defined, and especially under Thierack and Freisler the distinctions were blurred’. But it was not only the jurisdictions of the senates that were in constant flux. Most People's Court judges were not tied to just one senate either, during their tenure presiding over cases in different senates or in senates whose responsibilities had significantly changed over time. An indication of the varied senate experiences of People's Court judges is given in Table 5 for those judges in our sample who heard more than 20 cases. Together, the judges reported in Table 5 account for 90.8% of all cases and 92.5% of defendants in our sample. Table 5. Allocation of Most Active Judges of the People's Court Across Senates Judges Senate varieties 1a 1b 1c 2a 2b 2c 3a 3b 4a 4b 4c 5a 5b 6a Albrecht Y Y Y Y Crohne Y Y Engert Y Y Granzow Y Y Hartmann Y Y Y Y Y Y Lämmle Y Y Y Y Merten Y Y Y Y Y Friesler Y Y Y Köhler Y Y Y Y Y Löhmann Y Y Y Stier Y Thierack Y Y Y Zieger Y Y Y Y Sources. Wagner ( 1974 1989 1989 . Wagner (), Koch () and Wieland (). As Table 5 neatly illustrates, apart from Stier (who heard only 48 cases), all other important judges experienced a number of different jurisdictions and senates: on average about three over their careers as People's Court judges. Others experienced more. Hartmann presided over cases in as many as six different varieties of senates. On closer examination then, the historical record of the People's Court reveals a complex institutional history in which senates and their judges frequently changed jurisdictions. This constant change meant that there was in practice little real or enduring specialisation in the allocation of cases. Judges do not appear as a rule to have specialised in certain types of defendants. Instead, albeit by default rather than design, most People's Court judges during their tenure, including Freisler, typically encountered a wide and varied range of defendant characteristics and cases. It was a pattern of experience that in effect at least was in many ways consistent with what one would expect from random assignment. To test the plausibility of this contention, we begin by examining differences in observed characteristics of cases assigned to Alte Kämpfer and non‐Alte Kämpfer. The results are presented in Table 6. The first two columns show the mean case and defendant characteristics for cases heard by the two groups of judges. The third column is the difference in means without controlling for senate fixed effects and the final column is the difference in means, controlling for senate fixed effects. There are several significant differences in the raw means in column (3); however, consistent with the assignment procedure described above, within senates, there are few significant differences. Controlling for senate fixed effects, in column (4), the only significant differences between Alte Kämpfer and non‐Alte Kämpfer are with respect to Catholic and juvenile defendants. The difference in mean figures in Table 6 suggest that Alte Kämpfer are 1 percentage point more likely to hear cases with juvenile defendants and 2.1 percentage points more likely to hear cases in which the defendant is Catholic. Table 6. Differences in Characteristics of Cases Assigned to Alte Kämpfer and Non‐Alte Kämpfer Judges Variable Mean Difference in means Alte Kämpfer versus Non‐Alte Kämpfer Judge Alte Kämpfer Judge Non‐Alte Kämpfer Judge Without senate FEs With senate FEs Age 38.724 37.162 1.562*** −0.366 (10.850) (10.222) [0.358] [0.392] Education 0.220 0.147 0.073*** 0.001 (0.415) (0.354) [0.013] [0.015] Male 0.858 0.877 −0.019* 0.008 (0.349) (0.328) [0.012] [0.013] Unemployed 0.033 0.116 −0.083*** −0.004 (0.179) (0.320) [0.009] [0.009] Partial Jewish Ancestry 0.037 0.032 0.005 −0.003 (0.188) (0.175) [0.006] [0.007] Foreigner 0.025 0.017 0.009* 0.006 (0.157) (0.128) [0.005] [0.006] Juvenile 0.022 0.012 0.010** 0.010** (0.146) (0.107) [0.004] [0.005] Catholic 0.106 0.058 0.048** 0.021** (0.307) (0.234) [0.009] [0.011] Treason only 0.054 0.012 0.043*** 0.009 (0.226) (0.107) [0.006] [0.006] Major high treason 0.422 0.585 −0.163*** −0.017 (0.494) (0.493) [0.017] [0.016] Multiple offences 0.476 0.320 0.156*** 0.019 (0.500) (0.467) [0.016] [0.017] Violence 0.101 0.066 0.034*** 0.006 (0.301) (0.249) [0.009] [0.010] Pre‐major political 0.091 0.074 0.017* −0.008 (0.287) (0.261) [0.009] [0.011] Pre‐minor political 0.091 0.129 −0.038*** −0.016 (0.287) (0.335) [0.011] [0.013] Pre‐non‐political 0.148 0.166 −0.018 0.019 (0.356) (0.372) [0.013] [0.014] Comintern Communist 0.645 0.676 −0.031* 0.018 (0.479) (0.468) [0.016] [0.018] Social Democratic 0.030 0.057 −0.027*** −0.021 (0.169) (0.232) [0.007] [0.008] Other left wing 0.103 0.100 0.004 0.008 (0.305) (0.300) [0.010] [0.012] Right wing 0.177 0.112 0.065*** 0.002 (0.381) (0.315) [0.012] [0.013] Another way to test whether the assignment was random across Alte Kämpfer and non‐Alte Kämpfer judges is to run a logit regression in which a dummy variable equal to 1 if the judge is Alte Kämpfer is regressed on defendant and case characteristics. The results are presented in Table 7. In the first two columns, when we do not control for senate fixed effects, the two characteristics for which there are significant differences across the groups are offences involving violence and defendant's political affiliation. However, in the third and fourth columns, when we control for senate fixed effects, none of the case or defendant characteristics are significant. Table 7. Alte Kämpfer Status Regressed on Case and Defendant Characteristics Variables No senate fixed effects Senate fixed effects Marginal effects × 100 z‐statistic Marginal effects × 100 z‐statistic Age −0.872 −1.424 −0.840 −0.163 Age‐squared 0.0115 1.547 0.00804 0.162 Education 4.298 1.512 −3.527 −0.162 Male 1.278 0.468 −0.906 −0.145 Unemployed −3.288 −0.758 1.865 0.148 Partial Jewish ancestry 4.734 0.876 13.40 0.163 Foreigner −4.194 −0.657 1.320 0.128 Juvenile 13.05 1.219 15.16 0.162 Catholic 4.051 0.953 3.783 0.161 Treason only −12.53 −1.585 5.369 0.154 Major high treason 1.412 0.324 −5.025 −0.161 Multiple offences 1.359 0.320 −3.460 −0.157 Violence −7.490* −1.824 2.322 0.153 Pre‐major political 4.592 1.377 3.146 0.161 Pre‐minor political −1.723 −0.534 −2.011 −0.157 Pre‐non‐political 2.611 0.935 2.333 0.160 Comintern Communist 13.81*** 3.575 2.043 0.156 Social Democratic −15.06** −2.514 −8.282 −0.162 Other left wing 16.24*** 3.422 4.084 0.161 Right wing 17.79*** 7.509 Observations 3,298 3,298 3,215 3,215 Time fixed effects Yes Yes Location fixed effects Yes Yes Senate fixed effects No Yes Log likelihood −833.992 −511.838 Pseudo R2 0.63 0.768 As a further test we adopt a Monte Carlo simulation methodology similar to Abrams et al. (2012) to assess the degree to which case assignment was random. A key challenge for testing for randomisation is the lack of counterfactual scenarios in the data. We observe only one realisation with all the possible allocations of cases across judges. To address this, we use Monte Carlo simulations to create hypothetical scenarios and then compare the observed realisation (our dataset) with the distribution of all the hypothetical counterfactual scenarios to draw inferences about whether the observed realisation could have been picked by chance among all simulated scenarios. In implementing this testing strategy, we assume that if cases are randomly assigned to judges, observable defendant characteristics should have the same moments across different judges. For instance, in our data set, the average defendant age is around 38 years, meaning that if cases were randomly assigned, we would expect the average age of the defendant appearing before each judge to be around 38 years as well. However, as pointed out in Abrams et al. (2012, p. 360): ‘The difficulty in determining whether a data set results from random assignment is in quantifying exactly what it means for most judges to have defendants with mean age of around [38]’. It is unrealistic to expect that the average defendant age for each judge to be exactly 38, so the question becomes how much variation in the average defendant age can be attributed to sampling variability? The way to answer this question is to create a simulated data set by randomly assigning defendants to different judges who served in the same senate in the same year and measuring the variability in defendants’ ages across various such randomly assigned samples. If the variation in age in the observed sample is within the bounds of variability in age present in the simulated samples, then we can conclude that the allocation is random and any inter‐judge variation is due to sampling variability only. We provide more details on the method in online Appendix A. The results, which we also report in online Appendix A, provide strong evidence that assignment of observable defendant characteristics across judges is indeed random. Taken together, the lack of significant mean differences in observed characteristics of case and defendant characteristics across Alte Kämpfer and non‐Alte Kämpfer judges once senate fixed effects are controlled, the results from the simple selection specification in which Alte Kämpfer status is regressed on case and defendant characteristics, again controlling for senate fixed effects and the Monte Carlo simulations point to random assignment. Hence, any differences in the imposition of the death sentence can be attributed to the characteristics of the judges and their preferences, rather than non‐random allocation of cases across judges. 3.3 Estimation Strategy Having established that the allocation of cases to judges was random, we now examine inter‐judge variation in the imposition of the death sentence. We employ a logit model to examine the probability of receiving the death sentence. The dependent variable is a binary variable set equal to 1 if the defendant was sentenced to death; zero otherwise. In all cases, in addition to our proxy for judicial policy preferences, we include a full set of controls as given in Table 3 and described in online Appendix A. There was considerable variation in sentencing practices over time. For example, the death penalty was imposed with much greater regularity in World War II than before the war. Similarly, there was considerable variation in sentencing practices across regions of activity. Hence, all models also include time‐specific and location fixed effects to account for this variation. We cluster standard errors at the senate level.

4 Results Before presenting the main results from the logit model, we estimate a collapsed specification for each judge, in which, using OLS, we regress the number of death sentences over their time on the Court on judge characteristics, controlling for the number of cases each judge heard. The results are presented in Table 8. The results in panel (a) suggest that Alte Kämpfer sentenced 1.4–2.2 times more defendants to death than non‐Alte Kämpfer judges. The results in panel (b) suggest that Alte Kämpfer sentenced 47.6 to 53.5 more defendants to death than non‐Alte Kämpfer judges. Table 8. Number of Death Sentences Regressed on Judicial Characteristics Variables Panel (a) Panel (b) ln(Number of death sentences) as dependent variable Number of death sentences as dependent variable (1) (2) (3) (1) (2) (3) Alte Kämpfer 2.175** 2.203** 1.419** 47.59* 53.53* 22.61 (0.841) (0.834) (0.678) (25.27) (26.13) (15.95) Median age of judge 0.109* 0.0278 3.499* 0.287 (0.0540) (0.0469) (1.693) (1.104) Roman Catholic −1.332 −0.837 −26.36 −6.872 (0.907) (0.715) (28.43) (16.83) Prior Judicial experience in Weimar Republic −1.840* −0.340 −55.26* 3.905 (0.960) (0.840) (30.10) (19.77) Number of cases decided 0.00673*** 0.265*** (0.00174) (0.0410) Constant 1.570*** −2.792 −0.234 29.26** −118.1 −17.24 (0.437) (2.708) (2.201) (13.11) (84.85) (51.82) Observations 26 26 26 26 26 26 R2 0.218 0.469 0.696 0.129 0.356 0.792 Table 9 presents the logit results for the likelihood of receiving the death sentence. We present four specifications. In column (1), we include only Alte Kämpfer. In column (2) we include all case and defendant characteristics. In column (3) we add judge characteristics. The final column is the full specification. Table 9. Logit Marginal Effects for Death Sentence Variables Specification 1 Specification 2 Specification 3 Specification 4 ME × 100 z‐statistic ME × 100 z‐statistic ME × 100 z‐statistic ME × 100 z‐statistic Age 1.382** 2.311 1.332** 2.184 0.552 1.571 Age‐squared −0.0129** −2.184 −0.0124** −2.058 −0.00540 −1.571 Education 3.147 0.892 4.407 1.295 5.640*** 4.358 Male 15.03*** 3.762 14.74*** 3.763 10.03*** 3.655 Unemployed −46.40*** −5.440 −42.73*** −6.272 −8.715 −1.288 Partial Jewish ancestry 0.191 0.0235 −1.438 −0.174 −3.554 −0.630 Foreigner −1.558 −0.175 −1.946 −0.216 −5.080 −0.807 Juvenile −9.475 −1.536 −11.04* −1.704 −19.65*** −4.634 Catholic 2.412 0.530 2.082 0.472 2.258 0.657 Treason only 83.38*** 5.781 79.65*** 5.681 54.69*** 14.40 Major high treason only 66.33*** 4.215 61.97*** 4.150 35.61*** 4.228 Multiple offences 88.62*** 5.993 83.83*** 6.442 53.41*** 8.205 Violence 14.91*** 4.674 14.88*** 5.432 13.02*** 4.752 Pre‐major political 1.675 0.515 2.144 0.581 5.256* 1.913 Pre‐minor political 5.036* 1.827 4.550* 1.783 3.506** 2.283 Pre‐non‐political −4.859 −1.483 −4.779 −1.449 0.431 0.240 Comintern Communist −0.866 −0.104 0.225 0.0268 1.171 0.222 Social Democratic −35.82*** −3.689 −32.63*** −3.332 −13.70* −1.928 Other left wing −15.00* −1.920 −14.00* −1.701 −7.285 −1.261 Right wing −20.68* −1.660 −21.50* −1.717 −14.05 −1.597 Judge Age 8.158** 2.121 3.618** 2.227 Judge Age‐squared −0.0702* −1.937 −0.0340** −2.245 Judge Weimar −5.430** −2.280 0.0865 0.0404 Judge Catholic −14.76 −1.339 −9.123 −1.483 Alte Kämpfer 11.88** 2.190 4.024** 1.992 9.152** 2.358 3.636** 2.434 Observations 3,579 3,579 3,579 3,579 3,579 3,579 3,579 3,579 Time fixed effects No No No Yes Location fixed effects No No No Yes Log likelihood −2,173.972 −1,690.374 −1,665.838 −1,369.845 Pseudo R2 0.013 0.233 0.244 0.378 Consistent with our first proposition, the sign on Alte Kämpfer is positive and significant in all four columns. If we do not control for other variables, Alte Kämpfer were 11.88 percentage points more likely to impose the death sentence. Controlling for defendant characteristics, this figure is 4.02 percentage points and controlling for judge and defendant characteristics together it is 9.15 percentage points. In the full specification, in which we also control for offence type and fixed effects for the location of the offence and year of sentencing, Alte Kämpfer were 3.64 percentage points more likely to impose the death sentence than other judges on the court. Of the other judicial characteristics, we find that judge age has a non‐linear effect on the likelihood of imposing the death sentence. In the full specification, the likelihood of imposing the death penalty peaks at age 53. In specification (3), judges who were on the bench in the Weimar Republic were 5.43 percentage points less likely to impose the death penalty, but this variable is insignificant in the final specification. Among defendant characteristics, those with at least a high school education and males were 5.64 and 10.03 percentage points, respectively, more likely to receive the death sentence, while juveniles were 19.65% less likely to face capital punishment in the full specification. Older defendants were also more prone to receive the death sentence, although the coefficient is only weakly significant in the full specification. Prior criminal history mattered as far as having previous political convictions. In the full specification, relative to those without any prior convictions, having prior major or minor political convictions increased the likelihood of receiving the death sentence, although having previous non‐political convictions was insignificant. In column (4), being affiliated with an organised political group either had no effect (Comintern Communist) or a negative effect (Social Democratic, other left wing, right wing splinter) on the likelihood of receiving the death sentence, relative to those defendants who were not affiliated with any political group. In the final column, we add the offence type. The offence type mattered. Those charged with more serious political offences – major high treason (35.61 percentage points) or treason (54.69 percentage points) or multiple offences (53.41 percentage points) – were more likely to be sentenced to death than those charged with minor high treason. Similarly, those who committed a violent offence were 13.02 percentage points more likely to receive the death sentence than if the offence did not involve violence. Table 10 reproduces Table 9 but employing years of Nazi Party membership instead of the dummy variable for Alte Kämpfer status. Because we only have information on the date of appointment for 20 of the 25 judges in the sample who were Nazi Party members, the number of observations is less than in Table 9. We enter the controls in the same order as in Table 9. In specification 1, with no controls, for each additional year of Nazi Party membership, the probability that a defendant is sentenced to death is 4.07 percentage points higher. When defendant characteristics are added in specification 2, this figure drops to 2.46 percentage points and in the third specification, in which we control for defendant characteristics and judge characteristics other than years of Party membership, for each additional year of Nazi Party membership, the probability that a defendant is sentenced to death is 3.05 percentage points higher. In the final specification with a full set of controls, for each additional year of membership, the probability of receiving the death sentence is 1.22 percentage points higher. The average duration of Nazi Party membership for judges in the sample is 9.54 years. This suggests that the hypothetical average judge in terms of duration of Nazi Party membership would be 11.64 percentage points more likely to sentence a defendant to death than a judge who was a freshman Party member. Table 10. Marginal Effects for Death Sentence with Years of Nazi Party Membership Variables Specification 1 Specification 2 Specification 3 Specification 4 ME × 100 z‐statistic ME × 100 z‐statistic ME × 100 z‐statistic ME × 100 z‐statistic Years of Nazi Party membership 4.069*** 4.353 2.463*** 5.838 3.053*** 5.588 1.221*** 3.169 Observations 3,506 3,506 3,506 3,506 Time fixed effects No No No Yes Location fixed effects No No No Yes Log likelihood −1,966.799 −1,625.455 −1,602.56 −1,359.047 Pseudo R2 0.096 0.253 0.264 0.376 Table 11 reports the conditional probability that defendants with various characteristics receive the death sentence depending on whether they appear before a judge who is Alte Kämpfer. In the online Appendix B, we present the corresponding logit regressions, which are consistent with the conclusions from Table 11. The results are consistent with Proposition 1. Alte Kämpfer were more likely to impose the death penalty on members of the most organised opposition to the Nazi state (defendants who were members of Comintern Communist or Social Democratic parties) than judges who did not join the Nazi Party prior to 1933. Alte Kämpfer were 5.9 percentage points more likely to sentence a member of a comintern‐aligned Communist party to death than other judges. Similarly, Alte Kämpfer were 4.8 percentage points more likely to hand down the death penalty to members of a Social Democratic party. By contrast, Alte Kämpfer were relatively more lenient with members of the right wing and other left wing groups. They were 2 percentage points more likely to sentence a member of a right wing group to death but 10.5 percentage points less likely to sentence members of smaller left wing groups. Table 11. Conditional Probability of Receiving Death Sentence if Appearing Before an Alte Kämpfer Versus Non‐Alte Kämpfer Judge Defendant characteristic Conditional probability of receiving the death sentence if appearing before: Alte Kämpfer judge Non‐Alte Kämpfer judge Comintern Communist 0.708 0.649 Social Democratic 0.548 0.500 Other left wing 0.395 0.500 Right wing 0.326 0.306 Violence 0.875 0.701 Unemployed 0.209 0.160 Partial Jewish ancestry 0.647 0.299 Foreigner 0.656 0.233 Juvenile 0.279 0.045 Roman Catholic 0.705 0.458 Consistent with Proposition 1, we find that Alte Kämpfer were 17.4 percentage points more likely to impose the death sentence on those engaged in violent resistance than their less ideologically committed brethren. Consistent with Proposition 1, Alte Kämpfer were more likely to sentence to death defendants with particular characteristics than non‐Alte Kämpfer. These characteristics – being unemployed, of partial Jewish ancestry, a foreigner, juvenile or a devout Roman Catholic – were characteristics against which the Nazis were either prejudiced or highly intolerant. Alte Kämpfer were 34.8 percentage points more likely to sentence a defendant who was of partial Jewish ancestry to death; 4.9 percentage points more likely to sentence a defendant who was unemployed to death; 42.3 percentage points more likely to sentence foreigners to death; 23.4 percentage points more likely to sentence juveniles to death and 24.7 percentage points more likely to sentence Roman Catholics to death. We next examine whether the Court was more lenient to individual defendants with certain characteristics and/or at certain points in time. In Table 12, we examine whether the Court was more lenient to Comintern Communists before World War II and during the non‐aggression pact between Nazi Germany and the Soviet Union, from August 1939 to June 1941. Consistent with our fifth proposition, we find that on average Comintern Communists were 12.47 percentage points less likely to be sentenced to death before World War II, relative to during World War II. We find that Comintern Communists were 13.41 percentage points less likely to be sentenced to death before World War II, and 8.23 percentage points less likely to be sentenced to death while the non‐aggression pact between Nazi Germany and the Soviet Union was in force, relative to the period following the invasion of the Soviet Union. Table 12. Treatment of Comintern Communists in the Pre‐war and Non‐aggression Pact Years Variables Marginal effects × 100 z‐statistics Marginal effects × 100 z‐statistics Comintern Communist 4.528*** 4.011 4.415*** 3.654 Pre war period −77.46*** −7.455 −75.36*** −5.830 Non‐aggression pact period −8.232** −2.185 Pre war period × Comintern Communist 64.99*** 7.776 61.95*** 6.079 Pact period × Comintern Communist −2.954 −1.585 Observations 3,579 3,579 All other controls Yes Yes Time fixed effects No No Location fixed effects Yes Yes Log likelihood −1,533.754 −1,453.704 Pseudo R2 0.304 0.34 We also examined whether defendants who were members of the Nazi Party and defendants who served in World War II were treated more leniently. The regime may benefit from leniency being shown towards these defendants, especially if the offences were relatively minor, given that the Wehrmacht were short of personnel as the war progressed. In regressions not reported, we find that Nazi Party members were 6.50 percentage points more likely to be sentenced to death relative to defendants with no political affiliation and World War II veterans were 3.25 percentage points more likely to be sentenced to death than those who did not serve in the War. We also find, though, that those who were charged with minor high treason were 39.57 percentage points less likely to be sentenced to death where the reference is those charged with either of the more serious offences (treason or high treason). The relatively few defendants in the sample who were Party members or served in the War who were charged with minor high treason were also treated leniently. There were 14 Nazi Party members and three individuals who served in World War II who were charged with minor high treason, none of whom were sentenced to death. The proportion of death sentences handed down for treason and high treason escalated during the war. In our sample, 1,077 defendants were sentenced in the period 1933–9; of whom, just 19 (1.76%) received the death penalty. Meanwhile, 2,502 defendants were sentenced during wartime; of whom 1,075 (42.97%) received the death penalty. Figure 3 plots the conditional probability of receiving the death sentence over time. There was not an upward linear trend as the war progressed; rather, the conditional probability of receiving the death sentence is much higher after 1941, peaking in 1942, corresponding with the aftermath of the failure to capture Moscow in January 1942, before falling off somewhat in 1943 and 1944. Figure 3 Open in figure viewerPowerPoint Conditional Probability of Receiving the Death Sentence Over Time When we included a dummy variable for World War II in the final specification in Table 9, instead of time dummies, the likelihood of being sentenced to death in World War II was 34.11 percentage points higher than before the War. With a dummy variable for World War II, in lieu of time dummies, Alte Kämpfer were 3.2 percentage points more likely to sentence a defendant to death. We interacted the dummy variable for Alte Kämpfer with the dummy variable for World War II. The results are in panel (a) of Table 13. Alte Kämpfer were 1.72 percentage points more likely to sentence a defendant to death in World War II. We also interacted Alte Kämpfer with individual war years to test if Alte Kämpfer became more fervent, relative to non‐Alte Kämpfer following specific events towards the end of the war – for example, the failure to capture Moscow (January 1942), defeat at Stalingrad (February 1943) and the Normandy landings (June 1944) influenced sentencing. The results are in panel (b) of Table 13. Between 1939 and 1941, Alte Kämpfer were more likely to sentence defendants to death than non‐Alte Kämpfer; however, between 1942 and 1944, the difference is negligible and in 1945 non‐Alte Kämpfer were 3.05 percentage points more likely to sentence defendants to death than Alte Kämpfer judges. It seems that while ideological differences were important in contributing to differences in sentencing outcomes up to, and including 1941, in the last few years, when it became clear that the tide of the war had turned and defeat appeared inevitable and the proportion of death sentences increased, all judges sentenced more defendants to death and ideology became a less important driver. Table 13. Effects of World War II on Sentencing Outcomes Variables Marginal effects × 100 z‐statistic Panel (a) Alte Kämpfer 23.64*** 4.626 WWII dummy 45.62*** 11.62 Alte Kämpfer × WWII dummy −21.92*** −4.136 Observations 3,579 All other controls Yes Location fixed effects Yes Time fixed effects No Log likelihood −1,539.973 Pseudo R2 0.301 Panel (b) Alte Kämpfer 25.71*** 7.816 Year 1939 16.34*** 2.875 Year 1940 23.43*** 3.372 Year 1941 10.47** 2.122 Year 1942 55.01*** 12.69 Year 1943 43.02*** 9.674 Year 1944 38.92*** 8.726 Year 1945 35.24*** 4.276 Alte Kämpfer × Year 1939 −21.23** −2.468 Alte Kämpfer × Year 1940 −17.66*** −3.080 Alte Kämpfer × Year 1941 −8.913* −1.908 Alte Kämpfer × Year 1942 −25.98*** −8.394 Alte Kämpfer × Year 1943 −23.44*** −6.303 Alte Kämpfer × Year 1944 −24.23*** −4.742 Alte Kämpfer × Year 1945 −28.76*** −6.130 Observations 3,579 3,579 All other controls Yes Location fixed effects Yes Log likelihood −1,361.793 Pseudo R2 0.382 Table 14 examines whether judges who were aged 18–21 during the period of hyperinflation (June 1921 to January 1924) or whose hometown or suburb lay in a centre of the Revolution of 1918–9 were more likely to impose the death sentence. There is mixed support for our sixth proposition. Judges aged 18–21 during the period of hyperinflation were 8.3 percentage points more likely to impose the death sentence, but this is only weakly significant. The coefficient on the dummy variable denoting that the judge's hometown or suburb was in a centre of the 1918–9 Revolution was insignificant. We also tried interacting Alte Kämpfer with both life events to see if Alte Kämpfer judges who experienced these events were more likely to sentence to death. There was not enough variation in the hyperinflation specification so the interaction term was dropped. However, Alte Kämpfer judges whose hometown or suburb was located in a centre of the Revolution of 1918–9 were 5.84 percentage points more likely to impose a death sentence than Alte Kämpfer judges who did not come from such a location.22 Table 14. Effect of Events in Early Adulthood on Sentencing to Death Variables (1) (2) (3) (4) Marginal effects × 100 z‐statistic Marginal effects × 100 z‐statistic Judge aged 18–21 during hyperinflation 8.308* 1.660 Judge's home town or suburb was in a centre of 1918–9 Revolution −2.236 −1.476 Observations 3,579 3,579 3,579 3,579 Time fixed effects Yes Yes Location fixed effects Yes Yes Log likelihood −1,370.767 −1,371.074 Pseudo R2 0.378 0.378

5 Robustness Checks Our main results include defendants sentenced for treason and high treason in the Supreme Court and People's Court. Defendants tried in the Supreme Court were arguably treated more leniently given that it was not established as a political court to hear cases against political opponents of the Nazis. It must be remembered that what Hitler perceived as the lenient treatment of the majority of defendants in the Reichstag fire trial was the catalyst for the establishment of the People's Court in the first place. As a first robustness check on our findings, we redid the analysis excluding the 73 defendants sentenced in the Supreme Court. The results corresponding to those for the full sample in Tables 9 and 11 are reported in online Appendix C. The results, including the magnitude of the marginal effects, are almost identical to Tables 9 and 11. Excluding defendants tried in the Supreme Court has no effect on the results. Freisler is remembered as the most notorious Alte Kämpfer of the Nazi period with a proclivity to impose the death sentence. In our sample, Freisler sentenced 199 defendants; of whom, he sentenced 61.31% to death. This was approximately 10% higher than any other judge in the sample (see Table 1). Could it be that Freisler's sentencing behaviour as the alpha Alte Kämpfer is driving our results? To explore this issue, we redid the analysis excluding the 199 defendants sentenced by Freisler. The results corresponding to those for the full sample in Tables 9 and 11 are reported in online Appendix D. The results are very similar. Hence, excluding Freisler from the sample makes very little difference to the main results.23 As a further robustness check, we split our sample into the periods 1933–9 and 1940–5. The models for the pre‐war period did not converge, using logit because of the lack of variation in several independent variables with respect to receiving the death penalty. The results for defendants sentenced in wartime corresponding to those for the full sample in Tables 9 and 11 are reported in online Appendix E. The coefficient on Alte Kämpfer in the baseline results (corresponding to Table 9) continues to be positive, but ceases to be significant. The results for the conditional probability tables are almost the same as those reported in Table 11. Overall, if we restrict our sample to defendants sentenced during the war, we no longer find support for Proposition 1, but support for Propositions 1 remain unchanged. Of the seven Alte Kämpfer judges in our sample, two (Engert and Freisler) joined the Party before September 1930 when the Nazis became the largest party on the right. In September 1930, no one could have known for sure, or even considered it highly likely, that Hitler would become Chancellor. Some at the time believed that the Nazi Party had, in fact, peaked. Moreover, until 1933, the conservative side of German politics actively worked to keep him out (as, of course, did the left). Hence, those who joined the Nazi Party between September 1930 and January 1933 were still displaying a strong sense of belief in Hitler's party. This said, as a further robustness check, we examine if it makes any difference if we redefine our ideological commitment dummy variable such that it equals 1 for the ‘early’ Alte Kämpfer (Engert and Freisler) and zero otherwise. The results corresponding to Tables 9 and 11 are reported in online Appendix F. Engert and Freisler were 8.3 percentage points more likely to sentence a defendant to death than the other judges, compared with 3.6 percentage points for Alte Kämpfer in Table 9. The conditional probability table corresponding to Table 11 supports Propositions 1. Defining Alte Kämpfer in terms of early Alte Kämpfer makes no difference to the results. Finally, in the main results, we control for regional differences in the location where the offence was committed. One might think that Alte Kämpfer might target regions in which resistance activities were highest. While resistance was scattered all over Germany, the heaviest concentrations was in Berlin, Vienna, Saxony, North Rhine Westphalia and Hamburg. In a further robustness check, we interacted Alte Kämpfer with each of these five regions separately in the one specification. We found that Alte Kämpfer were 4.4 percentage points more likely to sentence a defendant to death if the offence was committed in Berlin, but the other interaction terms were insignificant.

6 Summary and Discussion We examine the factors influencing the likelihood of imposing the death sentence in Nazi Germany for treason and high treason. More than 95% of the cases in our sample were heard in the People's Court, a court established by the Nazis in 1934 to hear cases of serious political offences. Earlier histories of the People's Court depicted it as no more than a ‘blood tribunal’, in which judges conducted show trials with predetermined sentences consistent with the will of the Nazi regime. In recent years, however, there has been a noticeable shift in the scholarship on the role of courts in authoritarian regimes, such that it has now generally come to be recognised that judges in these contexts do not always simply exercise their discretion to facilitate the goals of the state. This change in thinking about the role of courts in authoritarian systems is reflected in recent, more nuanced assessments of the People's Court, too. What this new literature has lacked, though, is direct empirical evidence on the determinants of judicial discretion in such contexts and, in particular, why some judges appear more willing to implement the will of the state than others. We hypothesise that judicial policy preferences, as reflected in the ideological commitment of the Court's judges to Nazi values, should have been important in determining the sentence imposed, especially for defendants who belonged to the major groups opposed to the state, those engaged in violent resistance against the state, and those who possessed specific characteristics to which the Nazis were known to be strongly antagonistic. Our main results support such an attitudinal approach to the exercise of judicial discretion. As predicted, we find that judges with deeper ideological commitment to the values and aims of Nazism were indeed more likely to impose the death penalty than those who did not share the same level of commitment. Similarly, these ideologically committed judges were also more inclined to hand down the death sentence to members of the most organised opposition groups, those involved in violent resistance against the state, and defendants with characteristics repellent to core Nazis beliefs. The results remain fairly robust to alternative ways to define ideological commitment, including using a continuous variables for years of Party membership, and alternative time periods. We find some evidence that judges exposed to historical events in Germany that are likely to have shaped their pro‐Nazi leanings were more likely to impose the death sentence either via a direct effect (hyperinflation) or when interacted with Alte Kämpfer (the Revolution of 1918–9). There were, of course, other factors at play, too. The seriousness of the offence involved and the defendant's prior political convictions also had a clear influence on whether the death sentence was imposed. Another interesting finding of our article is that, like their counterparts hearing criminal cases in western democracies, Nazi judges, notwithstanding the intense political nature of the trials over which they were presiding, also took into account characteristics of the defendant in sentencing that were unrelated to the case, such as age, educational attainments and gender. With regard to these aspects, our results parallel in many ways with those of existing studies for the United States (Mustard, 2001; Sorensen et al., 2010, 2012), although they do differ noticeably with respect to the impact of educational attainment. Studies of the United States Federal Courts, for example, suggest that on the whole the better educated receive lighter sentences (Mustard, 2001). We, however, found the converse in our context, a result that derived in large part from the fact that better educated resisters to the Nazi regime tended to hold more senior roles in oppositional groups. None of our findings, of course, can or should be taken to signify that those who appeared before Nazi courts received fair trials. They manifestly did not. Rather what the evidence presented here tells us is that a more nuanced understanding of courts operating in such contexts is called for. Moreover, our findings suggest that a fuller investigation of sentencing practices and judicial discretion outside the United States, including in authoritarian states, is warranted, not just so that we understand those systems better, important as that might be, but also because in gaining such knowledge we may be better placed to provide a clearer perspective on which aspects of sentencing are generally experienced irrespective of judicial tradition and which are unique to specific contexts.

Notes 1 Our sample consists of 1,157 cases involving 3,579 defendants. Of these, 33 cases (73 defendants) or 2.04% of our sample were tried in the Supreme Court and 1,124 cases (3,506 defendants), or 97.96% in the People's Court.

Our sample consists of 1,157 cases involving 3,579 defendants. Of these, 33 cases (73 defendants) or 2.04% of our sample were tried in the Supreme Court and 1,124 cases (3,506 defendants), or 97.96% in the People's Court. 2 It should not be assumed that the sole judge in the sample who never joined any party organisation, Wilhem Bruner, was significantly independently minded. Indeed, his appointment to the People's Court was possible only because the authorities deemed him to be politically ‘safe’ (Gruchmann, 1988

It should not be assumed that the sole judge in the sample who never joined any party organisation, Wilhem Bruner, was significantly independently minded. Indeed, his appointment to the People's Court was possible only because the authorities deemed him to be politically ‘safe’ (Gruchmann, 3 For a good overview of Nazi ideology, see Kershaw ( 2000

For a good overview of Nazi ideology, see Kershaw ( 4 Amnesty International has statistics on death sentences imposed for many authoritarian states for the period 2007–14 but many of these are approximations and there is no corresponding data on cases in which the death penalty was not imposed or case level characteristics.

Amnesty International has statistics on death sentences imposed for many authoritarian states for the period 2007–14 but many of these are approximations and there is no corresponding data on cases in which the death penalty was not imposed or case level characteristics. 5 Argys and Mocan ( 2004 2003

Argys and Mocan ( 6 Until the last months of the war, every defendant who appeared before the People's Court by law had to be represented by a defence counsel. Under section 3 of article 4, the choice of defence counsel had to be approved by the presiding judge. That permission could be withdrawn at any time (Marxen, 1994 Wahlverteidiger ) or a court‐assigned counsel ( Pflichtverteidiger ). Given that this choice was conditional on the court's approval and politically unreliable lawyers had already been removed from the bar, the principle of free advocacy was an illusion. Hans Richter ( 1934 Deutsche Justiz on the day when the People's Court was first enacted: ‘In cases where the state entrusts the highest court on account of their importance for the general public, care should be taken that only defence counsels who enjoy general trust be involved’. In any case, a raft of court regulations, blocking or severely restricting inter alia the defence counsel's access to the indictment, court files, the defendant, the evidence and the witnesses, ensured that no real defence could be mounted. Furthermore, a lawyer who made an unfavourable impression on the court or took the interest of their client too seriously could expect to be removed from the case and threatened with the withdraw of their licence to practice (Wagner, 1974

Until the last months of the war, every defendant who appeared before the People's Court by law had to be represented by a defence counsel. Under section 3 of article 4, the choice of defence counsel had to be approved by the presiding judge. That permission could be withdrawn at any time (Marxen, ) or a court‐assigned counsel ( ). Given that this choice was conditional on the court's approval and politically lawyers had already been removed from the bar, the principle of free advocacy was an illusion. Hans Richter ( on the day when the People's Court was first enacted: ‘In cases where the state entrusts the highest court on account of their importance for the general public, care should be taken that only defence counsels who enjoy general trust be involved’. In any case, a raft of court regulations, blocking or severely restricting the defence counsel's access to the indictment, court files, the defendant, the evidence and the witnesses, ensured that no real defence could be mounted. Furthermore, a lawyer who made an unfavourable impression on the court or took the interest of their client too seriously could expect to be removed from the case and threatened with the withdraw of their licence to practice (Wagner, 7 See Ott and Buob ( 1993

See Ott and Buob ( 8 The personal files can be found at Bundesarchiv BArch R 3001. The surname of the judge and signature number in parenthesis: Albrecht [50165, 50167], Bruner [52916], Coninx [53671], Crohne [53759], Diescher [54191], Diester [54196], Engert [55261, 55265], Freisler [56247], Granzow [57872], Greulich [57983], Grosspietch [58154], Hartmann [59107, 59108], Illner [61600], Köhler [63914], Lämmle [65592, 65593], Löhmann [66929], Merten [68165, 68166], Schaad [73551], Springmann [76972], Stier [77504], Thierack [78253], Zieger [81308, 81309]. The NSDAP membership cards can be located on microfilm at Bundesarchiv BArch NSDAP Zentralkartei 31 XX and Bundesarchiv BArch NSDAP Ortskartei 3200. The surname of the judge and signature number in parenthesis: Diescher [ZK 31XX E0111], Diester [ZK 31XX E0112], Freisler [OK 3200 E0079], Illner [OK 3200 J0013], Köhler [OK 3200 J0013], Merten [OK 3200 O0067], Schaad [OK 3200 T0003], Zieger [ZK 31XX T9157].

The personal files can be found at Bundesarchiv BArch R 3001. The surname of the judge and signature number in parenthesis: Albrecht [50165, 50167], Bruner [52916], Coninx [53671], Crohne [53759], Diescher [54191], Diester [54196], Engert [55261, 55265], Freisler [56247], Granzow [57872], Greulich [57983], Grosspietch [58154], Hartmann [59107, 59108], Illner [61600], Köhler [63914], Lämmle [65592, 65593], Löhmann [66929], Merten [68165, 68166], Schaad [73551], Springmann [76972], Stier [77504], Thierack [78253], Zieger [81308, 81309]. The NSDAP membership cards can be located on microfilm at Bundesarchiv BArch NSDAP Zentralkartei 31 XX and Bundesarchiv BArch NSDAP Ortskartei 3200. The surname of the judge and signature number in parenthesis: Diescher [ZK 31XX E0111], Diester [ZK 31XX E0112], Freisler [OK 3200 E0079], Illner [OK 3200 J0013], Köhler [OK 3200 J0013], Merten [OK 3200 O0067], Schaad [OK 3200 T0003], Zieger [ZK 31XX T9157]. 9 This excludes Groβpietsch who only sentenced three defendants in the sample.

This excludes Groβpietsch who only sentenced three defendants in the sample. 10 The Revolutionary Socialists were a radical socialist group, which had broken off from the SPD.

The Revolutionary Socialists were a radical socialist group, which had broken off from the SPD. 11 ‘Anarchist/Syndicalist’ refers to membership of an anarchist or syndicalist organisation.

‘Anarchist/Syndicalist’ refers to membership of an anarchist or syndicalist organisation. 12 Trotskyists were Communists belonging to a party adhering to the views of Leon Trotsky.

Trotskyists were Communists belonging to a party adhering to the views of Leon Trotsky. 13 ‘KPDO’ stands for the Communist Party of Germany‐Opposition. As its name suggests, the KPDO was an alternative Communist Party comprised of Communists who had left or been forced out of the KPD because of their opposition to the party leadership's embrace of Stalinism.

‘KPDO’ stands for the Communist Party of Germany‐Opposition. As its name suggests, the KPDO was an alternative Communist Party comprised of Communists who had left or been forced out of the KPD because of their opposition to the party leadership's embrace of Stalinism. 14 Communist splinter groups were other, small Communist parties not affiliated to the Comintern. ‘Communist Help Organization’ refers to organisations set up to provide relief to the families of victims of Nazi oppression.

Communist splinter groups were other, small Communist parties not affiliated to the Comintern. ‘Communist Help Organization’ refers to organisations set up to provide relief to the families of victims of Nazi oppression. 15 The ‘ Schwarze Front ’ (Black Front) was a ‘left‐wing’ Nazi group, which was opposed to Hitler and had split from the Nazi Party in 1930.

The ‘ ’ (Black Front) was a ‘left‐wing’ Nazi group, which was opposed to Hitler and had split from the Nazi Party in 1930. 16 ‘BJ’ stands for Bündische Jugend , a loosely assembled association of German hiking groups which espoused a traditionalist, nationalistic and anti‐democratic worldview.

‘BJ’ stands for , a loosely assembled association of German hiking groups which espoused a traditionalist, nationalistic and anti‐democratic worldview. 17 ‘Catholic group’ refers to all organisations, most importantly the Centre Party ( Zentrum ), that worked to preserve Catholic values in Germany.

‘Catholic group’ refers to all organisations, most importantly the Centre Party ( ), that worked to preserve Catholic values in Germany. 18 National‐conservative’ refers to all organisations that espoused a conservative agenda. These organisations tended to be nationalistic, anti‐Semitic, monarchist and militaristic.

National‐conservative’ refers to all organisations that espoused a conservative agenda. These organisations tended to be nationalistic, anti‐Semitic, monarchist and militaristic. 19 The 20 July plot refers to the conspiracy to assassinate Hitler on 20 July 1944.

The 20 July plot refers to the conspiracy to assassinate Hitler on 20 July 1944. 20 The White Rose was a non‐violent, intellectual resistance group centred on the University of Munich.

The White Rose was a non‐violent, intellectual resistance group centred on the University of Munich. 21 Legitimists rejected the Anschluss with Germany in 1938 and campaigned for the restoration of the Habsburg dynasty in Austria.

Legitimists rejected the with Germany in 1938 and campaigned for the restoration of the Habsburg dynasty in Austria. 22 We also interacted both dummies with whether the defendant belonged to a Comintern Communist or Social Democratic party, but both interaction t