Mixed Judicial Selection and Constitutional Review: Evidence from Spain

Abstract

Spanish Constitutional Court – Judicial behaviour – Mixed judicial selection – Empirical testing – Decisions of the Spanish Constitutional Court, 1980-2018 – Judicial background – Government – Senate – Congress – Spanish Judicial Council – Invalidation of statutes –Dissent opinions – Shaping politicisation.

Introduction

The attitudinal model has been very influential in the empirical literature on judicial behaviour both in the United States and abroad. ¹ The model suggests that ideology is paramount in explaining how judges vote in collegial courts. However, ideology comes into play in a complex set of various interactions between politics, social experiences, public opinion, judicial philosophies, social norms, modes of behaviour in the judiciary, and many other considerations that vary across jurisdictions. The strategic model has also deserved extensive consideration, since courts seem to react to varying political and institutional realities. ²

Among this complex set of interactions, professional backgrounds or loyalty to the appointing institutions are likely to be relevant. Mixed selection systems are used to appoint many constitutional court judges around the world; that is, a given number of judges are selected by different institutions, some more political (such as executive and legislative branches of government) and some less political (such as judicial councils or judges filling regular courts).

These varying institutional features and interests may materialise in distinct ways when constitutional judges reveal their preferences in their behaviour and opinions in the court. There is, obviously, a normative discussion about diversity in backgrounds and in selecting mechanisms producing better results – more representation versus additional fragmentation, for example. In her important article, Lydia Tiede points out that mixed selection mechanisms can improve the representation of different interests, thus shaping deliberation and decisions. ³ At the same time, she recognises that certain appointing institutions can enhance their position outside of the conventional legislative bargaining by choosing constitutional judges.

Still, the normative discussion is less convincing without supporting empirical evidence that diversity in backgrounds and selecting mechanisms induce meaningful variation in judicial behaviour and output. If there are no, or minimal, behavioural implications, then the normative discussion about mixed selection systems is largely without relevant impact. As an illustration, Lydia Tiede finds differences in politicisation across judges selected by different institutions in Chile and Colombia. ⁴ However, similar results have not been documented previously for Spain ⁵ or Portugal. ⁶ Moreover, supporting evidence for Belgium ⁷ , Italy ⁸ , and Turkey ⁹ is somewhat weak.

Overall, our article aims to provide empirical evidence about the possible importance of professional backgrounds and selection mechanisms in generating different behavioural patterns in constitutional judges, notwithstanding the prevalence of attitudinal explanations and political dynamics.

We investigate two measures of revealed judicial preference – invalidation of statutes and dissenting votes. There are many reasons to suppose that professional backgrounds, on the one hand, and appointees from different selecting institutions, on the other hand, could differ across these dimensions. One immediate example is writing inclinations. It is unlikely that such disposition is the same across all appointees. Writing disposition could depend on the prior legal background (in a regular court, in academia, in the bar, in public office). Recognition of the importance of writing inclinations is not necessarily uniform across selecting institutions. ¹⁰

The Spanish Constitutional Court is a noteworthy case study. It has now been a widely studied institution from an empirical perspective. ¹¹ Concerning professional backgrounds, the Spanish Constitutional Court combines career judges with law professors and – less often – other legal professionals. At the same time, four different institutions select constitutional judges (executive, both houses of parliament, and the governing body of the general judiciary – the judicial council). Furthermore, career judges, for example, have been selected by all four different institutions, allowing us to study the interaction of professional backgrounds and selecting institutions.

We find that both professional background and selecting institutions seem to influence judicial behaviour to a certain degree, both in terms of invalidation of statutes and dissenting rates. However, we should emphasise that we do not detect the influence of attitudinal preferences to be deeply affected by these considerations. Empirical findings of political dynamics in the Court are consistent with previous literature and document a reasonable degree of politicisation.

The article goes as follows: the Spanish Court’s institutional context and the literature review are presented in the next section. In the following section, we summarise prevailing theories of judicial behaviour. Later, we introduce the dataset and discuss the findings from regression analyses. We conclude the article with possible implications.

Institutional context

The Spanish Constitutional Court: mixed judicial selection

There is a vast literature in English about the 1978 Spanish Constitution, the institutional dynamics of the Spanish Constitutional Court, and the controversial quasi-federal structure of Spain. ¹² We concentrate, however, on the specific arrangements described in the introduction above.

The Constitutional Court is composed of 12 judges (magistrados), who, in turn, elect a President from among themselves. ¹³ There are various nominating bodies: each of the parliamentary houses (Congress and Senate), with a 3/5 majority, nominates four judges. The executive branch, typically dominated by a prime minister or cabinet head – in Spanish, Presidente del Gobierno – who legally and politically towers over the cabinet, nominates two. The remaining two are nominated by the Judicial Council (Consejo General del Poder Judicial, referred to in the following text as ‘the Judicial Council’), the governing body of the judiciary. Constitutional judges serve a nine-year non-renewable term.

The most politically charged power of the Spanish Constitutional Court is the ex post constitutional review of legislation approved by the national or regional parliaments. In this article, consistent with previous literature, we focus on constitutional review following a challenge by a political body (recursos de inconstitucionalidad). ¹⁴ The Constitution itself names the political actors who may initiate the petition: (i) the prime minister; (ii) a number (50) of congresspersons or senators; (iii) regional governments, ¹⁵ or a majority of a regional parliament ¹⁶ (only against national laws); or (iv) the state ombudsperson (Defensor del Pueblo). Following the petition, if the Court finds a discrepancy between the challenged law and the Constitution, it declares the relevant portions of the law null and void.

The majority of the Court’s docket in matters of abstract review has concentrated on the Constitution’s provisions on the central government’s powers versus regional powers ¹⁷ (regarding the Court’s entire workload, individual claims against specific violations of constitutional rights and liberties by public bodies and courts vastly outnumber other sources). Some regional governments have initiated most constitutional review cases against national laws allegedly invading the constitutionally reserved competencies of the regions. Other cases have been initiated by the central government (formally, the prime minister) against regional laws allegedly exceeding the powers granted constitutionally to the regions.

Appointing institutions

Table 1 summarises the dataset from the viewpoint of constitutional judges. There have been 63 judges, mostly male and non-career magistrates. ¹⁸ We have 33 judges characterised as Spanish Socialist Workers’ Party (Partido Socialista Obrero Español or PSOE)/progressive and, consequently, 30 judges characterised as Democratic Centre Union (Unión de Centro Democrático or UCD)/People’s Party (Partido Popular or PP)/conservative. This classification reflects party (for executive and parliamentary appointees) or judicial association (for Judicial Council appointees) at appointment. Following the classification by Jorge Bercholc, we also control for moderate ‘centrists’. ¹⁹ They are judges from both UCD/PP or PSOE who are perceived to be less polarised. We label them centrista. ²⁰ Finally, due to Spanish political dynamics, we also consider judges born in the three historical ‘Autonomous Communities’: Catalonia, Basque Country, and Galicia. The label we use in Table 1 for this subgroup of judges is Regions. ²¹

Table 1. Characterisation of Dataset (Constitutional Judges)

In terms of the judicial profiles, given the multiple nominating actors, there are some interesting differences in institutional closeness between appointee and appointer: there is a direct link in executive appointments; slightly less so in parliamentary ones, since party majority and inter-party agreement are required; and a more indirect link at the Judicial Council, where political influence works through the connection between its members, judicial associations, and internal politics inside the career judiciary.

Executive appointees

Constitutional judges appointed by the executive are most obviously related to the personal choices of the prime minister. Executive appointments account for 11 constitutional judges in the period 1980–2018, of which only two (less than 20%) were career judges. Of the remainder, eight were law professors, and one was a public prosecutor (Fiscal). The two career judges were nominated by centre-right cabinets: one by UCD, a Supreme Court justice, and one by PP, a judge in the National High Court (Audiencia Nacional). The executive, proportionately, appoints fewer career judges than any other nominating body.

Parliament appointees

Parliament’s appointments are also directly negotiated by the political parties under specific leaderships. Usually, they require inter-party agreement (to comply with the required super-majority of 3/5 in each chamber for confirming appointments to the Court). Of the 39 parliamentary nominations to the Constitutional Court from 1980–2018, only 10 were career judges; ²² all others were law professors. Here, as in the executive, one observes a difference between the preferred pool of appointees for PSOE and PP. PSOE does not seem to back in parliament (or nominate through the executive, as observed before) candidates who are career judges. PP, in turn, although also often picking law professors, slightly favours career judges. Moreover, the (relative) bias towards non-law professors among PP-backed appointments seems to have increased over time. This trend is evident in the Rajoy period (after 2004).

Judicial Council appointees

A selection more distant from party dynamics is the one exercised at the Judicial Council. From 1980–2018, it has appointed 13 judges, all of them career judges.

The Judicial Council comprises 20 members; the two parliamentary houses have appointed all members since 1985. Therefore, political majorities in parliament translate into political majorities in the Judicial Council. Inevitably, then, the Spanish Judicial Council has seen periods of conservative majority and periods of progressive majority. In other words, depending on the general political cycle in Spain, the Judicial Council is dominated by conservative members (when right-wing parties control parliament) or by progressive members (when the reverse is true in parliament). This variance shows in the Judicial Council’s appointments: seven out of 13 career judges appointed by the Judicial Council were conservative, and six were progressive. ²³

The Judicial Council may be exposed to political cycles, but one feature remains unaltered. It is always (by constitutional design, in fact) dominated by career judges. ²⁴ Career judges (even under a 3/5 majority rule to make the Constitutional Court appointment) comfortably and permanently control the nomination process for the Constitutional Court in the Judicial Council, and it comes as no surprise that all appointees have been career judges. ²⁵

Previous empirical literature

An early pioneering study showed how the judges who sided in favour of or against certain legislation regarding two hotly contested political issues at the time (abortion and expropriation of a controversial business tycoon) formed two distinct clusters, progressive and conservative. ²⁶ These labels, since then used by the media, represent the tendency of a judge to concur significantly more often with judges of the same cluster than with those of the other from 1980–1985. A later empirical study questioned the actual prevalence of party interests and offered a more nuanced empirical characterisation of judicial behaviour in the Spanish Constitutional Court. ²⁷ Yet another study, using ideal point estimation, found consistent alignment divided into groups (conservative and progressive) for non-unanimous decisions from 2000–2009. ²⁸

Looking at 297 decisions (recursos de inconstitucionalidad) from 1980–2006, a more detailed study found that 192 were unanimous, of which 64% were decisions affirming constitutionality and 36% invalidated the challenged statute. ²⁹ The patterns did not seem to vary across time and political cycles. Regression analysis detected some forms of party alliance, but other considerations seemed to influence voting (for example, the involvement of Catalan and Basque nationalist parties). These results suggested that party alignment and ideology play an important role, but in no way do they fully characterise or capture judicial behaviour in Spanish higher courts. ³⁰

More recently, an empirical study focused on cases addressing conflicts of power between national and regional governments from 1981–2014. ³¹ The authors showed that there is no general statistical predisposition to favour either side. In this study, 365 decisions on conflicts of competence between national and regional governments from 1981–2014 were considered, including two critical periods: the 1980s and after 2003. While the 1990s seemed to have low constitutional litigation, the earlier period was dominated by the Basque country (although Catalonia, Galicia, and other autonomous communities also played a significant role). The later period was marked by conflicts about constitutional powers with Catalonia. The authors suggested that a multifaceted arrangement of ideological, legal, and strategic motivations describes judicial behaviour in the Spanish Constitutional Court.

Finally, a new study by us showed that there is evidence consistent with party leadership loyalty from 1980–2018. Using the same dataset we explore in this article, the study documented that the impact of judicial review asked by the judge’s party under the appointing leader is stronger than under a different leader. There are some variations across parties (PSOE and PP) and leadership periods. Still, the broad conclusion is that loyalty to the appointing party leader is likely more relevant than to the appointing party. ³²

These recent empirical studies did not address backgrounds or appointing institutions directly, since the articles were more concerned with political determinants of constitutional adjudication. However, through their regression analyses, they provided evidence that backgrounds and appointing institutions do not seem to play a systematic and statistically significant role.

Theories of judicial behaviour

The standard approach in previous literature about the Spanish Constitutional Court has explicitly focused on the attitudinal model and variations. ³³ For example, an earlier article suggests that ‘both preferences (sincere ideological voting) and incentives (strategic ideological voting) support our hypothesis that the behaviour of Spanish constitutional judges is explained by ideology’. ³⁴ The authors justified this hypothesis based on the appointment mechanism, limited tenure (nine years), significant differences between PP and PSOE concerning Spain’s federal nature (PP more centralist, PSOE less so), the abstract review political actors raise, and limited docket control.

In this article, by following Lydia Tiede, ³⁵ we focus on mixed judicial selection and possible differences in constitutional review. We are not downplaying the ideological aspects. Thus, we report our empirical findings controlling for political variables in ways consistent with previous empirical work. However, this article aims to explore the particularities of the institutional design in place and professional background.

Background

Out of 63 constitutional judges who served from 1980–2008, 25 were career judges. Inevitably the first line of inquiry is to assess the extent to which they exhibited a different behaviour than the remaining 38 individuals, mostly law professors and a few judges belonging to other legal professions (such as practising lawyers or prosecutors). ³⁶ Still, more interestingly, not all these 25 career judges were selected by the Judicial Council: only 13 were so appointed. The other 12 career judges were chosen by other selecting institutions (legislature or executive). Table 2 summarises the specific numbers for all four appointing institutions (executive, judicial council, and both parliamentary houses).

Table 2. Spanish Constitutional Court Judges (1980-2018)

Concerning background, a distinction between career judiciary and other legal professions might matter for different reasons. First, in a civil law system, career judges and other legal professions have significantly different training that could justify varying ways to interpret the law and assess the consequences of invalidating statutes. Second, in a court with limited terms, career judges have a safer and less politically dependent outside option (simply go back to the bench). At the same time, law professors may practise law ³⁷ or accept additional political sinecures. Third, in terms of personal economics, unless career judges plan to practise law after retirement, ³⁸ they experience less volatility in salaries and perks (a lifetime of safe employment within a public bureaucracy) than law professors who can practise law to earn additional money on top of their academic income. Finally, due to the nature of the profession, writing disposition, styles, and even skills, career judges are less likely to be prone to advocacy and more inclined to adjudication. ³⁹

Appointing institutions

The second line of inquiry investigates the possibility that constitutional judges selected by different institutions reveal distinct legal preferences. The reality that all four selecting institutions appointed career judges is significantly helpful in that regard.

As to appointing institutions, there may be different degrees of politicisation, with executive choices being more ideologically oriented and all others requiring more compromise because of how both parliamentary houses and Judicial Council operate in general, and more specifically in the nomination of judges to the Constitutional Court, requiring clear super-majorities. However, as previously explained, party choices (either PSOE or PP) are perceived as paramount in all institutions, even inside the Spanish Judicial Council. Still, different appointing institutions might have unalike agendas that constrain their appointees into revealing dissimilar preferences, even when politicisation is pervasive.

Hypotheses

We investigate two measures of revealed judicial preference – invalidation of statutes and dissent. There are many reasons to suppose that career judges, on the one hand, and appointees from different selecting institutions, on the other hand, could differ across these dimensions. The focus on career judges is primarily practical since the Judicial Council selected no law professors or other legal professionals, as explained in the previous section.

Concerning invalidation of statutes, one can hypothesise that career judges are more deferential to the legislator than law professors and others due to the traditional civil law model. Therefore, we should find career judges more likely to vote against invalidating statutes than other legal professionals (H1). As to dissent opinions, due to writing inclinations and the weight of the civil law tradition, we can posit that career judges should be less likely to file separate opinions than law professors (H2).

As to appointing institutions, we suggest that those who are appointed by the parliamentary houses and the executive should be more deferential to the legislator (loyalty to the appointing institution) than judicial council appointees, given the separation between legislative and judicial branches of government (H3). As to dissent opinions, the opposite reasoning applies: judicial council appointees should be more deferential to the court majority (loyalty to the civil law tradition of a unified court) than appointees by other government branches (H4).

We can easily see that H1 and H3 are potentially in contradiction and pave the way to debate which effect, on average, prevails for the career judiciary appointed by four different institutions. As career judiciary, H1 tells us that they should be more deferential to the legislator in the context of a strong civil law tradition. However, within H3, Judicial Council appointees (who are career judges) should be less deferential to the legislator since institutional loyalty matters. Therefore, for career judges selected by the Judicial Council, the overall effect is unclear. However, for career judges appointed by other institutions, H1 and H3 go in the same direction.

As to H2 and H4, a different pattern applies. Under the civil law tradition, career judges should be less likely to write separate opinions. As to loyalty to a unified court, Judicial Council appointees are also less likely to break the court’s line with a separate opinion. Therefore, for career judges selected by the Judicial Council, H2 and H4 go in the same direction. Nevertheless, for career judges appointed by other institutions, H1 and H3 are in contradiction, and the overall effect is ambiguous.

The interaction of H1 and H3, on the one hand, and H2 and H4, on the other hand, require empirical analysis since, when they move in opposing directions, there is no strong theoretical argument to suggest which should prevail.

Data and regression analysis

Data

We consider all constitutional review decisions in cases initiated by explicit political actors (recursos de constitucionalidad) from 1980–2018 (removal of Rajoy as prime minister after an opposition majority in the Congress backed a no-confidence motion): a total of 8,675 individual votes by constitutional judges in 773 decisions taken by the Spanish Constitutional Court. Of these 773 decisions, unanimous decisions formed a significant number (502).

In Table 3, we summarise additional information. The 773 decisions we study can be divided into five groups: 19 from UCD governments (1981–1982), 163 decisions from the first PSOE government (1982–1996), 84 decisions from the first PP government (1996–2004), 99 decisions from the second PSOE government (2004–2012) and 408 from the second PP government (2012–2018).

Table 3. Characterisation of Dataset (Decisions)

Out of 8,675 individual votes, there are 4,010 (46%) individual votes against invalidating statutes and 4,665 (54%) individual votes favourable to invalidating statutes. When only non-unanimous decisions are considered, the number is 1,396 (45%) and 1,678 (55%), respectively, for a total of 3,074 individual votes.

All logit estimations are clustered by decision (to account for the lack of independence of individual votes within the same decision). The usual interpretation of coefficients in odds ratio (a coefficient below one is a negative impact, while a coefficient above one is a positive impact; the sign of the coefficient is the relevant aspect, not the magnitude of the odds ratio) and statistical significance applies to all tables. ⁴⁰ Table 4 summarises the list of variables included in all following regressions, including descriptive statistics.

Table 4. Variable Descriptive Statistics

Voting against invalidating statutes

Taking the vote for constitutionality as the dependent variable allows us to test our approach directly by identifying which institutional variables explain the decision to invalidate a particular statute. Our econometric models include other multiple variables, like the political and demographic variables, consistent with previous literature about the Spanish Constitutional Court.

We start by showing the standard logit regressions exploring voting against the invalidation of a statute, including specifications for each appointing institution (executive, judicial council, and both parliamentary houses) in Table 5. The findings are consistent with previous literature: in the first column, institutional selection (executive and parliamentary versus judicial selection) and career judiciary (versus law professors and others) play no statistically significant role. At the same time, in the second column, a variable controlling for political cycle (PSOE in power) is positive and statistically significant. Finally, in the third column, multiple variables that capture political dynamics, such as judicial review asked for by the judge’s party and party in power, do impact voting against statute invalidation. The substance of the law also matters (for example, if the statute is about powers of the central government, social policy, or state and church, compared to others, there is a lower likelihood of voting against invalidation). The interaction terms reveal specific dynamics. Judges appointed by the executive seem more sensitive to judicial review asked by the judge’s party. Judges born in Catalonia, Basque Country, and Galicia are affected by judicial review asked by regional parties (such as Convergència i Unió from Catalonia and Partido Nacionalista Vasco from Basque Country).

Table 5. Regression Analysis (Logit; Odds Ratio), Voting against Invalidation of Statute, Clustered by Cases, All Observations.

*** 1% significance, **5% significance, *10% significance

There are two useful insights. According to the third specification in Table 5, executive appointees seem to be willing to vote more often for constitutionality than Judicial Council appointees. Meanwhile, according to the fourth specification in the same table, judges appointed by the executive are less likely to vote against invalidating statutes when the judge’s party petitions for judicial review.

Overall, there are a few indications that background and appointing institutions are relevant in a few specific instances. However, the statistical results indicate that different backgrounds and selecting institutions do not generate important variations in explaining the invalidation of statutes. For example, judicial review asked for by a judge’s party has a negative impact, albeit it is statistically significant in all specifications.

While in Table 5 we analyse all observations, in Table 6 we report the same logit approach, but only for non-unanimous decisions (rulings with at least one dissent). The results are somehow different. Career judiciary seems to reduce the likelihood of invalidating statutes as compared to law professors and others. Appointees by executive and parliamentary houses vis-à-vis judicial council also reduce the likelihood of invalidating statutes, but this is not statistically significant in all specifications.

Table 6. Regression Analysis (Logit; Odds Ratio), Voting against Invalidation of Statute, Clustered by Cases, Non-unanimous decisions only.

*** 1% significance, **5% significance, *10% significance

The results seem to point out that career judges appointed by the Judicial Council could have two opposing effects (career judiciary is positive compared to law professors and others while selection by the Judicial Council is negative compared to executive and parliamentary choices). Therefore, career judges selected by either the executive or the parliamentary houses are more likely to oppose the invalidation of statutes, though the coefficient is only statistically significant for Senate appointees (as we can observe in the last four columns of Table 6).

All other findings in Table 6 are broadly consistent with Table 5 and previous empirical literature on the Spanish Constitutional Court. Although somehow statistically weaker (since statistical significance is not valid in all four specifications), we observe identical dynamics across all selecting institutions in terms of judicial review asked for by the judge’s party and party in power.

Summing up, the findings tend to support both H1 and H3. Career judges are more likely to vote against invalidating statutes than law professors in Table 6 (consistent with H1), while Judicial Council appointees are less likely to vote against invalidating statutes than parliamentary appointees, also in Table 6 (consistent with H3). When in contradiction, H1 seems to dominate over H3, specifically, a statistically significant effect for Senate appointees. The positive effect on invalidating statutes when the judge’s party requests petitions do not seem to vary across appointing institutions (both in Tables 5 and 6).

Dissent opinions

We analyse dissents for all observations, including unanimous decisions (Table 7) and decisions with dissents only (Table 8). Our estimations document some important differences across professional backgrounds and appointing institutions concerning the inclination to write a separate opinion.

Table 7. Regression Analysis (Logit; Odds Ratio), Dissent Vote, Clustered by Cases, All Observations.

*** 1% significance, **5% significance, *10% significance

Table 8. Regression Analysis (Logit; Odds Ratio), Dissent Vote, Clustered by Cases, Non-unanimous decisions only.

*** 1% significance, **5% significance, *10% significance

In Table 7, considering all observations, we detect that career judges are more likely to dissent than law professors and others. Such behaviour is detected overall, but also specifically for judges appointed by both parliamentary houses. At the same time, appointees selected by the Judicial Council seem to dissent less often than those selected by both parliamentary houses. Concerning executive appointees, the results are somehow contradictory between first and later specifications.

In Table 8, considering decisions with at least one dissent only, we have similar findings. Judges are more likely to dissent than law professors and others, specifically for judges appointed by both parliamentary houses. Again, appointees selected by the Judicial Council seem to dissent less often than judges appointed by both parliamentary houses, while the results are contradictory for executive appointees.

We also find that leftist judges are more likely to dissent (both in Tables 7 and 8), thus confirming that attitudinal preferences matter as documented by previous literature on the subject. As to interaction terms, constitutional judges born in the three historical ‘Autonomous Communities’ are more likely to dissent when faced with judicial review petitioned by regional parties.

In conclusion, our results indicate that career judges seem to dissent more often, which contradicts H2: the civil law tradition against open dissent in a court of law does not seem to play the role we anticipated. However, the effect is enhanced for parliamentary appointees over judicial appointees, a finding consistent with H4.

Final remarks

In this article, we have tested the idea that professional backgrounds of judges and the nature of the appointing institution matter. In the context of the Spanish Constitutional Court, we study career judges (as opposed to law professors and other legal professionals). Career judges have been selected by all four appointing institutions (executive, judicial council, and both houses of parliament) from 1980–2018.

Our empirical findings extend and, to some extent, confirm previous empirical literature about the Spanish Constitutional Court. There is a degree of alignment between the way constitutional judges vote and ideological variables. Such alignment can be explained by a combination of attitudinal and strategic reasons. Judges’ backgrounds and appointing institutions present empirical regularities regarding voting against statutory invalidation and dissenting. However, these empirical regularities do not deny politicisation; they merely make it more complicated in terms of court dynamics.

Our approach confirms the claim by Lydia Tiede ⁴¹ that mixed selection mechanisms are likely to reveal different judicial behaviours. Notwithstanding, we show that the patterns are complex. Career judges seem to be not identical to law professors in their behaviour as constitutional adjudicators. Career judges oppose statutory invalidation somehow more often (hence, we can conclude they are more deferential to political branches, which is hardly surprising in a civil law system), while they dissent more often. Appointees by the parliamentary houses seem to oppose the invalidation of statutes somehow more often and dissent more often. The combination of selecting institutions and professional backgrounds creates varying realities.

All our regressions confirm once again that political or attitudinal variables matter. In that respect, they are consistent with past findings. Still, they show that judicial behaviour in a constitutional court is complex and hardly explained by one particular dimension of revealed preference. In that respect, we believe that the factors underlying our results are generalisable to other constitutional courts belonging to the Kelsenian type (centralised constitutional review) embedded in similar institutional environments.

The case of the Spanish Constitutional Court is particularly illustrative because it has a variety of appointing institutions, compared to constitutional courts in Germany, Austria, Belgium, and Portugal, which have less variance in appointing institutions. Countries with similar variety, such as France and Italy, however, do not allow separate opinions. In that respect, the Spanish case provides a better institutional setting to test the importance of mixed judicial selection. Notwithstanding, our findings can be of general interest. For example, mixed judicial selection does not seem to reduce politicisation, if that is a concern. At the same time, the career judiciary seems to exhibit characteristics that are less influenced by the appointing institution. The importance of this aspect depends on the extent to which there is a concern about having a constitutional court with more or fewer career judges.