Recently, academic attention has been directed toward the collegial feature of judicial decision-making.1 The literature on the influence of colleagues on judicial-decision-making, however, predominantly focuses on the role of judge-colleagues. Other co-workers of judges —namely, the judicial assistants that assist judges in preparing and processing cases—have received more limited attention. This is especially true for judicial assistants at lower courts. This oversight is surprising, since in many jurisdictions both the number of judicial assistants and the duties assigned to them have increased.2

The rising prevalence and prominence of ‘judicial assistants’, or simply ‘assistants,’ has increased the urgency around better understanding their role in the judicial decision-making process. An obvious point of attention, in studies on judicial assistants, is the level of influence that they wield.3 Explaining assistants’ influence on adjudication is an important follow-up focus point, which is explored in this article.

Most of the research that goes beyond a mere description of the role of judicial assistants originates from the US, with a strong emphasis on law clerks at the US Supreme Court.4 The majority of these explanatory studies use a version of principal-agent theory that is adapted to the specific setting of the US Supreme Court (see more on principal-agent theory in the next section). In this setting, assistants are appointed to assist one justice, with whom they work closely. The clerkship is a temporary position, usually lasting one year. Decision-making at the court is, furthermore, highly politicised. The studies using principal-agent theory to explain assistants’ influence, in this specific setting, focus particularly on justices’ selection of assistants as a tool to control adverse selection and on the congruence of the political attitudes between assistants and justices. These studies’ findings cannot simply be generalised to other levels of court and other judiciaries—precisely because of the exceptional conditions of assistant-justice collaboration in the US Supreme Court. Moreover, a small body of mostly qualitative studies suggests that principal-agent theory is altogether insufficient to explain the role of assistants in judicial decision-making. These studies indicate that contextual factors—such as whether a case is decided upon by a single judge or a panel of judges, the complexity of court cases, and time pressure—also explain assistants’ influence on adjudication.

For both reasons, our goal is to provide an alternative explanatory model to understand judicial assistants’ influence on adjudication. To achieve this goal, we develop and test a version of principal-agent theory that can be applied more universally than the one that has been used thus far. Additionally, we test the explanatory power of several contextual factors. The data used for this research were collected from a survey of administrative law judges at Dutch district courts. Apart from the theoretical aim to contextualize and enrich principal-agent theory for the explanation of judicial assistants’ influence, our study also aims to provide insights that are relevant for legal practice. There is a general awareness that the involvement of judicial assistants in adjudication can have advantages such as reducing caseloads and backlogs, relieving judges of routine, administrative tasks enabling them to focus their attention on the more complex aspects of adjudicating and providing a soundboard for judges. At the same time there are also concerns that employing assistants may have disadvantages such as a diminishing judge’s sense of personal responsibility for judgments and an abdication of judicial responsibilities. This study increases the insight into the conditions under which assistants influence judges’ decisions and what strategies judges use to control the risks of involving assistants in adjudication in addition to selecting assistants.

In the next section, we provide our hypotheses. Subsequently, we explain the method of our study. We then present our findings and discuss them. In the last section we draw our conclusion.

Hypotheses regarding the influence of judicial assistants

Principal-agent theory, operationalised in terms of attitudinal differences

While several more descriptive and qualitative studies exist on the role and duties of judicial assistants in a wide range of jurisdictions,5 the quantitative studies that attempt to explain influence are predominantly US centred, with a strong bias towards the US Supreme Court. The dominant theory used in these studies is principal-agent theory.6

Principal-agent theory applies in circumstances where one actor (the ‘agent’), is able to make decisions on behalf of, or that impact, another actor (the ‘principal’).7 These circumstances create a dilemma when the two actors may have different interests and asymmetric information (the agent having more information), such that the principal cannot directly ensure that the agent is always acting in the principal’s best interest. This is particularly true when activities that are useful to the principal are costly to the agent, and where elements of what the agent does are costly for the principal to observe.8 Various mechanisms may be used to align the interests of the agent with those of the principal. A distinction is made between two types of control mechanisms.9 The first type is related to adverse selection, that is when the agent has private information about the costs of exerting effort or the valuation of a good before an employee contract is concluded. An example of this type of control mechanism is the selection of new recruits who are likely to share the principal’s interests. The second type of control mechanisms pertains to moral hazards, wherein the agent becomes privately informed after an employee contract is concluded. Examples of this type of control mechanisms are monitoring and incentivizing performances.

Judges, as principles, employ clerks—their agents—to help them cope with their workloads by performing particular duties, such as writing first drafts of opinions. However, the delegation of duties also enables clerks to ‘develop private information that would allow them to steer decisions in a particular direction’.10 The principal must therefore ‘be concerned about potential avoidance of duty or shirking on the part of the agent, which can arise when there is a divergence between the preferences of the two’.11 This means that judges may want to reap the benefits of involving assistants in judicial decision-making, while at the same time controlling the risks of assistant defection or underperformance. As is typical of principals, judges basically face two risks: adverse selection—which could result in recruiting untrustworthy assistants—and moral hazard, which could result in collaborating with assistants that do not share the same goals as the judge.

The most important tools that contemporary Supreme Court justices possess to reduce the possibility of assistant defection are on-the-job monitoring of the assistants’ performance and selection. It is the latter tool that has been extensively studied. Ditslear and Baum have established that justices endeavour to select clerks that share their ideological preference and that this trend has become stronger during the last few decades.12 Subsequently, Peppers and Zorn13 have additionally established that justices’ inclination to select like-minded clerks influences justices’ decisions regarding their merits. The probability of a liberal vote by justices in the US Supreme Court increases, as the cadre of judicial clerks is composed of more Democrats than Republicans. Apart from replicating the latter finding, Kromphardt has established, more specifically, that justices are less likely to cast a liberal vote when their chamber is composed homogenously of conservative clerks.14 Furthermore, two studies have shown that ideological differences between clerks and judges play a role when it comes to screening certiorari petitions for Supreme Court hearings.15 Clerks prepare memoranda regarding these petitions in a pool; the memos are distributed amongst the chambers of all justices in the pool. The ideological distance between the pool clerks and pool justices impacts judges’ votes. Both studies show that receiving a clerks’ recommendation from an ideological opponent, versus from a trusted ally, makes a judge less likely to vote consistently with the clerk’s recommendation. However, in the study of Black and Boyd,16 this is found to be true only for those petitions that are of middle-high quality (not for petitions of low or high quality).

In these American studies, the control issues that justices face as principals are thus consistently narrowed down to political-ideological divisions, which justices attempt to solve via their selection strategies. This specific operationalisation of principal-agent theory is particularly suitable for the setting of the US Supreme Court. The focus on selection mechanisms is logical, given that the justices typically select their personal clerks annually and nearly always personally interview the final candidates.17 Justices’ hiring choices can therefore massively decrease the risk of defection by personal clerks. Focusing on ideological differences and partisanship also makes perfectly good sense in the context of the US Supreme Court. The US Supreme Court is highly politicised, given the political nature of the court cases it decides. The process of appointing the justices themselves (after nomination by the US president) also considers the justices’ political attitudes and partisanship as important selection criteria. Researchers suggest that the court has become increasingly politicised during the last decades.18 The setting of the US Supreme Court is quite unique. This implies that the dominant operationalisation of principal-agent theory, therein, is not necessarily similarly suitable for other settings. In fact, there are reasons to assume that such an operationalisation is unsuitable for most other court settings.

First, the focus on clerk selection is not obvious for most other court settings. In many settings, judges are not as involved in selecting the assistants. Also, several courts have different set-ups, in which judicial assistants are not appointed to individual judges. For example, many US courts also employ so-called ‘staff-attorneys’. These are professional, centralised research staff members who routinely serve for multiple years. An inventory among member states of the Council of Europe also reveals that, in the majority of European member states, a judicial assistant position can be a permanent career.19 Additionally, a substantial number of member states organise judicial assistance so that assistants are assigned to panels of judges, or work in a pool for multiple judges.20 This implies that, in comparison to the setting of the U.S. Supreme Court, the relationship between judges and assistants is less personal and the personal selection criteria of judges (for instance, related to their political preferences) play a much smaller role in the selection of assistants.

Second, in most court settings apart from the US Supreme Court, the influence of partisanship and political attitudes upon judges is likely to be limited. ‘This is because judges have an incentive to protect the legitimacy of their institutions, and judicial legitimacy is thought to suffer when partisanship is seen as a basis for the court’s decisions’.21 Indeed, research has shown that ideology plays a smaller role in hiring judicial clerks at the lower court levels (circuit court and district court).22 Two surveys—among federal district court judges and federal court of appeal judges, respectively—showed that only 15.1% of district court judges and 15.8% of appellate court judges mentioned political ideology to be a factor in hiring law clerks. Consecutively, 0.3% of district court judges and 1.7% of appellate court judges considered political ideology to be the most or the second most important selection criterion, respectively. This was a much lower priority level than they gave to judicial assistants’ personalities and law school class ranking.23 In line with the previous observations, legal specialisation—rather than political orientation—was found to be the most important selection criterion for selecting judicial assistants at the German Constitutional Court.24 Thus, there is ample research indicating that judicial clerks’ political attitudes do not have a significant impact on judicial decision-making in settings apart from the U.S. Supreme Court.

This does not mean, though, that principal-agent theory cannot provide a valuable explanation of judicial assistants’ influence on judicial decision-making. On the contrary, a more universally operationalised version of the theory is still expected to offer a powerful explanation. First, the distinction between principal and agent seems suitable to characterise the relationship between the judge and the judicial assistant. After all, as judges are the ones appointed by law to adjudicate, they are ultimately responsible for the opinions and judgments. Judicial assistants are subordinates and all their duties derive from the judges’ responsibility to adjudicate. Second, the basic assumption that underlies principal-agent theory—that judges seek to benefit from the involvement of assistants in judicial decision-making, while at the same time controlling the risk of assistant defection—appears realistic. However, whereas clerks’ incompatible political attitudes are deemed to pose a significant risk for justices in the US Supreme Court, they are unlikely to be perceived as such by judges in most other court settings. The latter are likely to perceive other vulnerabilities in collaborating with judicial assistants. We therefore proceed by identifying five other factors that determine the magnitude of the moral hazard judges face when involving assistants in judicial decision-making, which also account for courts apart from the US Supreme Court.

Factors determining the magnitude of the moral hazard of employing judicial assistants

The first factor that may guide judges’ decision-making behaviour and how they deal with employing judicial assistants consists of their role orientations. Whereas the attitudinal model of the principal-agent theory focuses on controlling the content of judicial decision-making (Democratic/liberal vs. Republican/conservative views on what is considered a proper outcome of adjudication), role orientations are concerned with the process of decision-making. Role orientations concern a judge’s conception of what is considered to be proper decision-making behaviour by a judge.25 Focussing on the process of decision-making, Gibson notes that judges can have ‘narrow’ or ‘broad’ role orientations. ‘Narrow’ orientations reject the legitimacy of allowing anything but strictly legal criteria to influence judicial decision-making. ‘Broad’ orientations are more open to allowing non-legal criteria in the process of decision-making.

A new development among Western judiciaries has been observed, which may change judges’ role orientation. A paradigm of managerialism has been introduced into court organisations, which alters the thinking around adjudication.26 This development also offers judges a new set of values toward which to orientate their judicial behaviour. In the managerial paradigm, private-sector principles and techniques are applied to the court organisation. Non-legal criteria, such as effectiveness and efficiency and accountability, are thus also considered to legitimise judicial decision-making. Differentiation of duties and the creation of internal stratification are important ways in which administering justice can be organised more efficiently. Judicial assistants can play a key role in achieving greater efficiency and cost-effectiveness. This perspective thus marks the desirability of involving subordinate staff members, such as judicial assistants, in judicial decision-making. As such, the managerial perspective results in broader ideas around judicial decision-making than were present when courts were led by the traditional rule of law paradigm.

Within a rule of law paradigm, judges are—to a certain degree—viewed as being the personification of the judiciary, one of the three independent branches of the government. To carry out their special duty of administering justice, judges are surrounded by several unique safeguards that aim to uphold their independence: impartiality, integrity and competence. These provisions justify that judges are entrusted by society with the responsibility to adjudicate. Holvast argues that this rule of law paradigm means that involvement in adjudicational duties by non-judges, such as judicial assistants (who are not surrounded with comparable safeguards), is likely to be perceived as inappropriate or at least problematic.27 A particular concern of delegating duties to judicial assistants, from a rule of law role orientation, is that judges will sacrifice at least part of their autonomy to make a fully independent and impartial decision.28

Judges may have internalised either of these two prevailing normative stances toward the employment of judicial assistants. A managerial role orientation is less restrictive than a rule of law role orientation, in terms of allowing other than strictly legal criteria to influence judicial decision-making. Therefore, it is expected that judges who have internalised the former role orientation allow assistants more influence than do those who have internalised the latter role orientation.

It is also conceivable that judges’ perceptions of the risks and benefits of employing assistants have an effect on assistants’ influence on adjudication—one that is independent of judges’ role orientation. Judges may, on the one hand, perceive risks around involving assistants in judicial decision-making, such as abdicating responsibilities for adjudication, generating extra tasks related to supervising assistants and corroding the authority of the judge. On the other hand, judges may perceive benefits. For example, the involvement of assistants can enable judges to delegate simple, time consuming tasks so they can focus on the more demanding, complex aspects of adjudication; can provide judges with a sounding board; and can help prepare hearings, opinions and verdicts. In relation to judges’ role orientations and their weighing of the risks and benefits of involving assistants in judicial decision-making, we formulate our first three hypotheses:

  • A judge’s managerial role orientation is associated with the judicial assistant having more influence on the judge’s decisions (hypothesis 1);
  • A judge’s rule of law role orientation is associated with the judicial assistant having less influence on the judge’s decisions (hypothesis 2);
  • The more a judge is convinced that the benefits of involving assistants in the decision-making process outweigh the risks of doing so, the more influence judicial assistants have on judges’ decisions (hypothesis 3).

Related to the weighing of risks and benefits is the amount of trust—or lack thereof—an assistant deserves, in the eyes of the judge. Trust can be defined as the confidence that another party to an exchange will not exploit one’s vulnerability.29 In this context trust means that judges have confidence that assistants will not execute the duties that have been delegated to them in such a way that it harms the position or reputation of the judge. The level of trust of the judge in the assistant is another factor that can be expected to define the influence of judicial assistants. Generally, three aspects of trust are distinguished: competence, benevolence and integrity.30 An ethnographic study in Dutch district courts showed that judges particularly referred to the former aspect of trust, in relation to their collaboration with judicial assistants.31 Several judges mentioned relying more on certain judicial assistants than on others, because they generally ‘do a better job’ or are ‘better assistants’. Judges attributed these qualifications mainly to the educational background of judicial assistants and to their experience and expertise. A particular competency that was also regularly mentioned as an important factor, in trusting judicial assistants’ work, was their accuracy. Thus, in some situations, a judge’s expectation that an assistant is deserving of trust seems to play a role in how much influence assistants have on judge’s decisions.

This study also demonstrated that the extent to which judges appeared to rely on the competency of assistants, depended not just on the competency of the assistant but also on that of the judge. Especially when a very experienced and knowledgeable assistant was working with a judge who was new to a specific field of law, this could result in judicial assistants dominating the decision-making.32 When a judge had little experience and the assistant had a great deal of experience, this resulted in the judicial assistant being particularly highly involved in adjudication. This suggests that the experience level of the judicial assistant, relative to that of the judge, appears to affect their influence.33 A US study also suggests that judges’ relative experience affects the influence of judicial assistants on judicial decision-making.34 That research shows that newcomers to the court are systematically more likely to follow an assistant’s pool memo recommendation, than are their more senior colleagues.

Two hypotheses can be formulated based on the role of trust. The first of these focuses on the trustworthiness of the assistant and the second on the assistant’s relative competence:

  • The more trust a judge places in a judicial assistant, the more influence this assistant has on the judge’s decisions (hypothesis 4);
  • The more experience a judicial assistant has, relative to the judge, the more influence this judicial assistant has on the judge’s decisions (hypothesis 5).

Contextual explanations

So far, we have only discussed hypotheses that are derived from principal-agent theory. Yet a modest body of, primarily, qualitative studies suggests that contextual factors also play a role in assistants’ influence on adjudication. In our study we will assess how three contextual factors affect the influence of judicial assistants, but without formulating concrete hypotheses about their impact. The reason why we abstain from formulating hypotheses about these contextual factors is that the empirical evidence on how these factors are related to judicial decision-making is either insufficiently solid or inconclusive.35

The first factor concerns the composition of the adjudicating corps: are the cases adjudicated by a single judge or a panel of judges? The suggestion that this factor may play a role in assistants’ influence is based on a study by Holvast. The majority of judicial assistants she interviewed mentioned that their involvement is usually greater during single-judge decision-making than in cases that are decided upon by a panel of judges.36 In more complex, single-judge cases, the judicial assistant regularly functions as an important discussion partner for the judge. In panel decision-making, the judges are frequently so engaged with each other that the opinion of the assistant is less relevant. This suggests that assistants have more influence in cases that are decided by a single judge than those that are decided by a panel of judges.

A second contextual factor that may determine the influence of assistants is the complexity of a court case. The findings with regard to this factor point in opposing directions. On the one hand, Holvast found that, in most instances, simple cases were decided primarily by following precedent.37 The outcome of these cases was rather clear-cut. Hence, neither the judge nor the judicial assistant really made a difference regarding the outcome of the judgment. In legally complex cases, there was more room for variable outcomes. This also resulted in more room for assistants to influence the judicial decision. This was especially true when the assistant had a great deal of expertise and the judge valued his or her opinion.

On the other hand, Holvast also found that more complex cases tended to receive greater attention from the involved judge(s), which seemed to diminish the assistants’ influence.38 Yet, when Black and Boyd tested this hypothesis in their study on the influence of law clerks, they did not find support for it.39 There are other studies, though, which also question the idea that judges grant assistants more discretion in complex cases than they do in simple cases. For instance, Swanson and Wasby exposed that the higher the percentage of routine cases on the court’s docket (as perceived by the surveyed judges), the higher the percentage of cases in which a clerk changes the judge’s opinion.40 The researchers’ interpretation of this finding is that—in cases with clear-cut legal outcomes, where clerks provide highly constraining laws or facts that cannot be evaded—judges have no option but to change any preconceived opinions to conform to the legally dictated result.

Thus, although several studies suggest that the complexity of court cases is associated with the influence of assistants, the empirical evidence is inconclusive about the direction of this association. The greater prevalence of precedents in simple cases may either reduce or enhance the room for assistants to influence judges’ decisions, while the increased attention that judges pay to complex cases may reduce assistants’ influence.

Time pressure is a third contextual factor that may affect the influence wielded by judicial assistants.41 Again, the available data on how this factor affects the influence of assistants points in different directions. On the one hand, Holvast found that judges had less time to discuss cases when they experienced time pressure, which could result in their deciding cases alone instead of discussing them with their judicial colleagues or assistants.42 On the one hand, she also found that, when judges were under pressure to produce output, they were occasionally more receptive to the input of judicial assistants and less critical regarding the products (such as draft-judgements) they created.43 This amplified receptivity increases judicial assistants’ potential to influence judicial decision-making. The effect appeared to be strengthened when judges were dealing with substantial workloads and were operating under severe time pressure. The suggestion that time pressure provides assistants more influence was supported by a study from the US.44 This research established that—in state high courts without an intermediate court below them—law clerks were more influential in changing their judges’ opinions, regarding proper case outcomes, than in state high courts with an intermediate court below them. The researchers suggest that this reflects the typically larger workload placed on judges in former high courts, which could lead them to defer more often to clerk recommendations.

Thus, time pressure may reduce the assistants’ influence as it leaves judges less time to discuss the merits of cases, but it may also increase assistants’ influence because it makes judges more receptive for assistants’ suggestions on how to adjudicate.

We will compare the anecdotal or inconclusive findings of the existing research about the association between the composition of the adjudicating corps, case complexity and time-pressure and assistants’ influence on adjudication with the findings of this study.



We tested our hypotheses in a research project conducted among Dutch district court judges working in Administrative Law. Dutch district courts handle all court cases in the first instance. This results in wide variations among cases (from very simple to highly complex), which are adjudicated under greatly differing circumstances. We limited our study to administrative law judges for practical reasons. The second author had the best contacts among judges within this legal domain, for instance, and administrative law judges are approached less often to participate in research than are criminal judges. Since judges have the ultimate responsibility for judgments, we focused on judges and did not include judicial assistants in our research.

In order to contextualize our explanatory model of assistants’ influence, we briefly discuss the role of judicial assistants in the Dutch court system in general and in administrative law cases in particular.45 In the Netherlands, judicial assistants are not assigned to specific judges. Rather, judicial assistance work in a pool for multiple judges. Since judicial assistants are not assigned to an individual judge, a large part of the selection and recruitment of assistants is completed by court managers rather than judges. Although the minimum educational requirement for new judicial assistants is a degree from an institute of higher professional education, most applicants possess a university law degree. Employment as a judicial assistant is not a temporary position; rather, it can be a lifelong career. Judicial assistants assist judges in cases from the beginning of the process up to the writing of the judgment. Administrative law assistants in particular perform all four duties through which they may influence judges’ decisions.46 These duties consist of screening, for example by deciding whether a court case will be send to a single judge or a panel of judges, deliberating cases with the judge, preparing bench memos and drafting judgments.

Our research data were collected by way of a survey, distributed among all Administrative Law judges working at Dutch district courts. Approval for the distribution of the survey was granted first by the Council of the Judiciary (Raad voor de rechtspraak) and subsequently by the National Board of Administrative Judges (Landelijk Overleg Vakinhoud Bestuursrecht). The survey consisted of five sections: 1) the perceived extent and desirability of a clerk’s influence on judicial decision-making, 2) the collaboration of the respondent with a clerk in a concrete court case, 3) the role orientation of the respondent, 4) background characteristics of the respondent and 5) a vignette of a fictitious but realistic court case. The first four sections are used for this paper, while the fifth part is used for another paper.47

In total 90 respondents filled-in the questionnaire. Ten questionnaires were deleted because they were only partly completed. At the start of our data collection, in total 272 FTE were working as an administrative law judge, while on average judges work 0.925 FTE.48 This means the response rate was approximately 27% (80/(272 x 1.081)). This rate is lower than the average response rate of 35.7 percent that Baruch and Holtom found for studies utilizing data from organizations, but it is within the range of one standard deviation of 18.8 that they established in their study.49 A potential reason for our somewhat lower than average response percentage has been the process we had to follow in collecting the data. Data were collected in two rounds. Since we were not allowed to send individual judges an email-invitation to complete the online questionnaire for privacy reasons, we had to rely on team leaders to distribute this email among their team members.50 After two weeks, a second email was distributed to remind judges to complete the questionnaire. The first round yielded 31 questionnaires that were filled in completely. In order to boost the response, we were granted permission by the National Board of Administrative Law Judges to conduct a second round of data collection in five of the eleven administrative law divisions of district courts in the Netherlands – Rotterdam, Amsterdam, Haarlem, Utrecht and Arnhem. During this second round we personally visited these courts to briefly present our research to the judges at team meetings, to answer possible questions about the questionnaire and to invite them to complete our questionnaire.51 Additionally, the team leader of administrative law judges in Zwolle offered to personally bring the questionnaire to the attention of his team members and collect and post the completed questionnaires to us. This second round of data collection increased the response substantially. However, as mentioned, only six out of eleven courts were included in the second round. Moreover, the second round brought to light that not all judges had received or read the email that was sent in the first round of data collection. Both factors may have negatively influenced the ultimate response rate.

It is challenging to assess how our selection procedure may have affected the generalizability of our findings. It was not possible for us to determine which judges received the invitation email in the first round, which results in us not being able to assess how this affected the representativeness of our sample. What we can say is that in comparison to the total population of judges in the Dutch courts, the average age of the sample is similar (50.5 (population) vs 50.1 (sample) years), while males are overrepresented in the sample (36.9% (population) vs 51.4% (sample)).52 Judges working at the courts that we visited during the second round of data collection are also overrepresented in our sample. We are unable to determine whether the (lack of) representativeness of these different categories in our sample has had an impact on the test results of our hypotheses.


The dependent variable was based on the influence that judges attribute to assistants in a concrete court case. Judges were asked to recall the latest court case, in which they collaborated with a judicial assistant, and to report the extent to which the assistant had contributed to the outcome of the court case. The answer categories were the following: 1) ‘none,’ 2) ‘very inconsiderable,’ 3) ‘inconsiderable,’ 4) ‘considerable,’ and 5) ‘very considerable.’ The court cases varied in content. They involved, for example, reuniting family members who have stayed in the country or origin with a family member who has migrated to a host country, real estate appraisal, the denial of social welfare payments and the licensing of electric vehicle charging posts.

Of the five variables that were used to test the principal-agent theory, a managerial role orientation emphasises—among other aspects—the values of efficiency and effectiveness in judicial decision-making. Respondents were asked to indicate whether they disagreed or agreed with five Likert items, with answers ranging from ‘disagree strongly’ to ‘agree strongly’, as well as a category of ‘don’t know/no opinion.’ Table 1 shows that the factor loadings of the five items range between 0.32 and 0.79, which together do not constitute a reliable scale (Cronbach’s α = 0.51). As all items loaded sufficiently high on the underlying factor and no better alternative to measure the managerial role orientation was available, we have decided to use this scale nonetheless. This means that we test hypothesis 1 with an unreliable scale, an issue we return to in the discussion.

Table 1

Principal Components Analysis of five Likert items indicating a managerial role orientation (N = 63).

Items Factor loadings

(1) The quality of judicial decisions suffers severely from court budgeting53 0.70
(2) Judges insufficiently take into account whether the costs of their efforts outweigh the benefits 0.65
(3) Judges ought to give better account of the time spent on handling cases 0.79
(4) It is undesirable that the budget of courts is dependent on the number of cases handled54 0.32
(5) Team leaders ought to conduct themselves more as managers 0.38
Eigenvalue 1.8
Variance explained (%) 35.5
Cronbach’s alpha 0.51

Scale scores were calculated as the mean score for every respondent having a valid score on at least three of the five items used to measure the managerial role orientation. Higher scores on this scale indicate that judges have internalised a managerial role orientation.

A rule of law role orientation centres, among other aspects, on the constitutional independence and impartiality of the judge. This orientation was measured with eight Likert-items, emphasising the value of the judge as autonomous decision-maker and the legal provisions that justify judges’ autonomy of decision-making. Table 2 shows that the factor loadings of the eight items range between 0.30 and 0.75 and together constitute a scale of relatively low but acceptable reliability (Cronbach’s α = 0.64).55

Table 2

Principal Components Analysis of eight Likert items indicating a rule of law role orientation (N = 49).

Items Factor loadings

(1) The selection procedure for judges needs to be stricter 0.46
(2) Judgments contain too much jargon56 0.40
(3) Politicians interfere too much with individual court cases 0.70
(4) Citizens lack the required knowledge to assess the correct value of judicial decisions 0.62
(5) The council of the judiciary poses a threat to the independence of the judge 0.41
(6) Adjudication by laymen is undesirable, because it undermines the quality of the judiciary 0.58
(7) The use of algorithms in judicial decision-making jeopardises the quality of the judiciary 0.68
(8) The approval of having ancillary positions in addition to being a judge ought to be restricted further 0.52
Eigenvalue 2.4
Variance explained (%) 29.7
Cronbach’s alpha 0.64

Scale scores were calculated as the mean score for every respondent having a valid score on at least four of the eight items used to measure the rule of law role orientation. Higher scores on this scale indicate that judges have internalised a rule of law role orientation.

Trust was measured based on the extent to which the judge believed that the particular judicial assistant they had collaborated with, in the concrete court case, possessed the qualities of experience, assertiveness, precision, speed and orientation toward service. The answering categories ranged from: 1) To a very small extent, 2) To a small extent, 3) To some extent, 4) To a large extent, 5) To a very large extent. Index scores were calculated by adding the scores for the five qualities and subsequently dividing the sum by five. A high score indicates that a judge puts a high level of trust in this assistant.

Assistants’ relative experience was measured by subtracting the standardised score for the respondents’ own experience as an administrative law judge, in years, from the standardised score for the experience they attributed to the judicial assistant they had collaborated with in the concrete case. The answer options were 1) very little, 2) little, 3) some, 4) much, 5) very much.

The risk-benefit perception of involving judicial assistants in the decision-making process was measured with ten Likert items. The answer categories ranged from ‘disagree strongly’ to ‘agree strongly’ and ‘don’t know/no opinion’. Table 3 shows that the factor-loadings of the ten items range from 0.34 to 0.74 and together constitute a reliable scale (Cronbach’s α = 0.70).

Table 3

Principal Components Analysis of ten Likert items indicating the risk-benefit perception of involving judicial assistants in judicial decision-making (N = 75).

Items Factor loadings

The contribution of judicial assistants to judicial decision-making
(1) …provides an invaluable assistance to the judge 0.68
(2) …harms the reputation of the judiciary57 0.65
(3) …enables the judge to focus on his core tasks 0.40
(4) …dilutes the ultimate responsibility of the judge for decision-making58 0.47
(5) …relieves the judge 0.34
(6) …influences the outcome of the case in improper ways59 0.61
(7) …diminishes the authority of the judge60 0.74
(8) …keeps the judge on edge 0.43
(9) …causes the judge to pay less attention to the specific circumstances of a particular court case61 0.35
(10) …burdens the judge with extra work in terms of controlling and managing judicial assistants62 0.56
Eigenvalue 2.9
Variance explained (%) 29.3
Cronbach’s alpha 0.70

Scale scores were calculated as the mean score for every respondent having a valid score on at least nine of the ten items used to measure the risk-benefit perception. Higher scores on this scale indicate that respondents are more convinced that the benefits of involving assistants in judicial decision-making outweigh the risks.

The first of the three contextual factors consisted of whether the concrete court case in which the respondent had collaborated with a judicial assistant was decided by a single judge or by a panel of judges. The reported complexity of the case ranged from (with percentages given in brackets): 1. Very simple, 2. Simple, 3. Neither simple nor complex, 4. Complex, 5. Very complex. Time pressure was measured by asking respondents how much time they had at their disposal to solve the concrete case. The answer categories (with percentages given in brackets) were: 1) Much too little, 2) Too little, 3) Precisely enough, 4) Too much, 5) Way too much.


Table 4 shows to what extent principal-agent theory and contextual factors explain judicial assistants’ influence, in a concrete court case in which the judge collaborated with a particular judicial assistant.

Table 4

Judicial assistants’ influence on a concrete court case explained by principal-agent theory and contextual factors (N = 76, correlations [2-tailed significance], Betas).

r Beta (model 1)# Beta (model 2)# Beta (model 3)#,~

Principal-agent theory
Managerial role orientation (hyp. 1) 0.22*   0.19*   –     0.23*  
Rule of law role orientation (hyp. 2) –0.01 0.03     –     0.01    
Perceived risk-benefit (hyp. 3) 0.45** 0.23*   –     0.21*  
Trust (hyp. 4) 0.62**   0.45** –     0.43**
Relative experience of assistant (hyp. 5) 0.27*   0.12     –     0.15    
Contextual explanation
Panel judgment (vs. single-judge) 0.02     –     0.07     0.06    
Complexity case –0.09     –     –0.13     –0.12    
Time pressure 0.00     –     0.09     0.12    
R2% 44.3** 1.3     45.8**

# In model 1 all principal-agent variables are regressed on the assistants’ influence, in model 2 all contextual factors are regressed on the assistants’ influence, in model 3 both the principal-agent variables and the contextual factors are regressed on the assistants’ influence.

~ The variance inflation factor (VIF) for the variables in model 3 ranges between 1.19 (rule of law role orientation) and 1.70 (complexity court case), all below the rule of thumb that if VIF>10 then multicollinearity is high63 (a cutoff of 5 is also commonly used).64

* p < 0.05, ** p < 0.01.

The five hypotheses are derived from principal-agent theory. In support of hypothesis 1, a managerial role orientation correlates positively with judicial assistants’ influence (r = 0.22, p < 0.05). This correlation remains significant when controlling for the other principal-agent factors (ß = 0.19, p < 0.05) and when also controlling for the contextual factors (ß = 0.23, p < 0.05). The more judges emphasise the values related to managerialism in their role orientation, the more influence they allow assistants in concrete court cases. Contrary to hypothesis 2, there is no significant correlation between a rule of law role orientation and assistants’ influence. A role orientation that emphasises values related to the rule of law does not prevent judges from allowing assistants’ influence (r = –0.01, p = n.s.; ß = 0.03, p = n.s. when controlling for other principal-agent factors and ß = 0.01, p = n.s. when controlling also for contextual factors). In support of hypothesis 3, there is a strong positive correlation between judges’ perception that the benefits of involving assistants in judicial decision-making outweigh the risks, and assistants’ degree of influence (r = 0.45, p < 0.01). This correlation becomes substantially weaker when controlling for the other principal-agent factors, particularly trustworthiness. Yet it remains significant (ß = 0.23, p < 0.05), even when also controlling for the contextual factors (ß = 0.21, p < 0.05). In support of hypothesis 4, there is a strong positive correlation between judges’ trust in the particular assistant they are collaborating with in concrete court cases and said assistant’s influence (r = 0.62, p < 0.01). This correlation decreases, but remains significant, when controlling for the other principal-agent factors (ß = 0.45, p < 0.01) and when also controlling for the contextual factors (ß = 0.43, p < 0.01). In support of hypothesis 5, there is a positive correlation between the experience of the judicial assistant, relative to the experience of the judge (r = 0.27, p < 0.05), and the assistant’s level of influence. Yet, this correlation is no longer significant when controlling for the other principal-agent factors (ß = 0.12, p = n.s.) and when controlling also for contextual factors (ß = 0.15, p = n.s.). Thus, it cannot be concluded that judges allow assistants more influence when assistants have more relative experience.

Table 4 also presents the correlations between three contextual factors and assistants’ influence. None of these correlates are significant. Moreover, this lack of significance does not alter after controlling for other contextual factors and factors related to principal-agent theory. The literature review already showed that multiple interpretations of contextual factors are possible and that relatively little previous research has been conducted, upon which to build solid hypotheses. It is possible that our measures were insufficiently specific to capture the complexities and nuances involved in these contextual factors.

This means that three out of the five hypotheses derived from principal-agent theory are supported by our findings, while none of the contextual factors are associated with assistants’ influence on adjudication. In total, 45.8% of the total variance is explained by the two sets of explanations.


In this section, we first provide interpretations for not finding support for two principal-agent related hypotheses. Subsequently, we discuss the limitations of our study.

A reason for not finding that a rule of law role orientation is negatively correlated with judicial assistants’ influence on judgments may be that the assumption on which this hypothesis is based is wrong. Contrary to this assumption, an orientation among judges that views their role as that of core adjudicators entrusted with the responsibility by law—and accompanied by unique safeguards to promote independence, impartiality, integrity and competence—may not be incompatible with allowing assistants influence on judgments. It could be argued that—if assistants’ input is limited to administrative, supportive and consultative tasks and if judges remain ultimately responsible for judgments—then the legitimacy of adjudication is not compromised by the involvement of assistants in judicial decision-making. This argument becomes even stronger when judicial assistants also enjoy several safeguards similar to those of judges, such as adherence to a professional code of conduct and minimal requirements of training and expertise. It could thus be understood that a rule of law role orientation is not negatively correlated with assistants’ influence. The tenability of this interpretation could be explored, by probing deeper into how the category of judges that has firmly internalised a rule of law role orientation think about the (un)desirability of delegating certain tasks and responsibilities.

Furthermore—in line with our expectation regarding the hypothesis on relative experience—we find that the assistant’s experience, relative to the judge, is positively correlated with the assistant’s influence. However, contrary to our expectations, this correlation is no longer significant after controlling for the other factors that are linked to principal-agent theory (and contextual factors). This suggests that relative experience captures yet another aspect of the level of trust a judge places in an assistant. This would explain why relative experience ceases to correlate significantly with assistants’ influence when controlling for trust. This interpretation is also congruent with principal-agent theory, in that this theory predicts that assistants’ influence is positively correlated with trust—i.e., how slight the principal (the judge) perceives the risk of defection by the agent (the assistant) to be. Black, Boyd and Bryan also fail to find support for the relative experience hypothesis: Justices are not less likely to follow a pool clerk’s recommendations when they have served more time on the court.65

Our study has several methodological limitations. First, the number of valid observations, at 80, is rather small, just as is the response rate of around 27%. A higher response rate would have reduced the risk of potential biases. Second, we assessed the dependent variable using a single indicator (namely, the influence of judicial assistants in a concrete court case as perceived by the judge). This perceived influence may not accurately reflect assistants’ actual influence. Ideally, assistants’ actual influence should be measured by way of an experimental study, as we have done in another paper.66 However, it is challenging to devise an appropriate experimental design for this particular study. This is because such an experimental study would require the comparison of judge’s decision with and without collaboration with a judicial assistant on the same court case, which is not feasible. What is conceivable, is to increase the number of indicators measuring assistants’ perceived influence. For instance, respondents could be asked to recall more than one court case in which they collaborated with a particular assistant and to attribute the amount of influence to assistants in each of these cases. Another option would be to supplement the influence a judge attributes to an assistant in a particular court case with the influence the assistant attributes to herself in the same case. Third, we encountered some challenges in measuring the judges’ role orientations. The reliability of the managerial role orientation scale was not sufficient and the number of missing values, for some items measuring both role orientations, was considerable. This means that the validity of the measurement of both role orientations could be improved. These methodological limitations to our study imply that our conclusion needs to be interpreted with some caution.


The aim of this study was to test the extent to which principal-agent theory and contextual factors explain judicial assistants’ influence on adjudication. Principal-agent theory has thus far been operationalised primarily in terms of congruence between the political attitudes of judges and assistants, to explain assistants’ influence on justices’ decisions in the US Supreme Court. We have derived five hypotheses from this theory, which can offer a more universal operationalisation that is better suited for explaining judicial assistants’ influence in a wide range of court settings. As expected, we find that three of the principal-agent hypotheses—judges’ managerial role orientation, judges’ trust in judicial assistants and judges’ perception that the benefits of involving assistants in adjudication outweigh the risks—positively correlate with assistants’ influence in concrete court cases. Contrary to our expectations, we did not find the other two principal-agent hypotheses—judges’ rule of law role orientation and the experience of assistants relative to the experience of the judge—to be correlated with assistants’ influence. Likewise, none of the contextual factors—panel judgments (vs. single-judge judgments), complexity of court cases and time pressure—are correlated with assistants’ influence. Thus, while several of the hypotheses relating to principal-agent theory explain the influence that judges allow assistants, contextual factors do not. Our study shows that principal-agent theory can be fruitfully applied to settings other than the politicised US Supreme Court. It also shows that apart from the recruitment of suitable assistants, trust building is another strategy that judges use to control the risks of involving assistants in adjudication. By collaborating with specific assistants in daily practice, judges learn by experience which duties they can or cannot delegate safely. This insight provides an important nuance to more general discussions about the potential risks of involving assistants in adjudication. To determine whether contextual factors (as well as the principal-agent related hypotheses that were not found to be significant) can explain judicial assistants’ influence, further research is required. Replicating an expanded version of this study, in other legal domains and jurisdictions, is therefore warranted.

Additional File

The additional file for this article can be found as follows:

Influence judicial clerks_V2.sav

SPSS data file DOI: