I. Study Overview
The aim of the current study was to measure the variation in attitudes of judges towards various aspects of judicial decision-making, based on their prior professional experience. The main hypothesis of this study was that a different amount of professional experience might make judges have different opinions towards their own judicial activity. This hypothesis was inspired by psychological studies focused on the importance of experience in decision-making, which are explored in more detail in section II. The idea that experts might think differently than novices is probably intuitively known for quite some time (often expressed anecdotally by both advocates and judges themselves). Yet no empirical study of judicial decision-making has yet measured the dynamic view that judicial attitudes can change over time with levels of experience. There are almost no longitudinal studies done on judges, and certainly none in Europe. A few cross-sectional studies included the age or experience as one of their variables, but these variables were counted as one among the many generic background factors and never given central importance in the findings discussions.
To explore the impact of professional experience, the current study operationalised the concept by splitting it into 3 different variables: judicial experience (number of years on the bench/tenure), criminal legal experience (number of years spent dealing with criminal cases), and pre-appointment legal experience (number of years gained in law practice before becoming a judge). While potentially overlapping, these three variables also reflect different aspects of professional expertise of judges, and allow the study to explore in more detail what it is exactly that makes judges change their attitudes towards their sentencing practice.
While ‘experience’ could presumably be operationalised in several ways, the current study aimed to measure the level of experience of judges not by self-reported measures, but by using objective variables such as the amount of years spent on the bench, or the amount of years in legal practice. This methodological choice was directly inspired by psychological studies on expertise and decision-making, most notably dual processing accounts of decision-making, Gary Klein’s extensive work on the decision-making of experts such as firefighters, airplane pilots and military officers. Most notably, although Kahneman and Klein disagreed on many points, they agreed that professional experience is built under three necessary conditions: prolonged practice, routine and feedback. In other words, the [prolonged] exposure to certain kind of practice (be it legal, judicial or criminal) of judges seems to be a solid predictor of their professional experience.
The study was ambitious in aiming to include judges from all levels of courts, and not just focus on judges from appellate or supreme courts. Due to this fact, it had to tame its other ambitions in three other ways. First, it was designed as a sociological study of attitudes (expressed by the judges themselves), and not a psychological experiment testing their internal mental processes. The European judges have yet to be as comfortable as American judges with undergoing psychological testing. That said, judicial experience is not seen as a potentially contentious factor affecting the judicial image, so it was expected that judges would not hide their opinions or falsify their answers in any way (at least consciously).
Second, while experience might be expected to influence attitudes towards many aspects of the judicial activity, for the sake of scientific clarity the study only focused on one type of judicial decision-making: sentencing. Sentencing was considered one of the most challenging and important duties a judge has; on one hand, due to the importance of the consequences for the parties involved, and therefore the pressure to get the “right sentence” in a case; on the other hand, due to the multitude of aggravating and mitigating factors a judge has to take into account (or disregard) when deciding the right sentence. For this reason, the questions focused on the respondents’ attitudes towards their sentencing activity.
Third, while empirical studies on judicial decision-making are perhaps often conducted with the unspoken desire to describe characteristics that are universal, applicable cross-jurisdictionally, each researcher is bound to focus on one jurisdiction and hope that the study shall have the chance to be replicated in other jurisdictions in the future. This study makes no exception. While it was designed to be conducted in any world jurisdiction, the study was eventually conducted in one European jurisdiction, namely Romania.
Perhaps surprisingly, Romania has a key advantage in constituting a good case study for this type of study: it has a hybrid judicial appointment and training system, combining the typical characteristics of continental jurisdictions (e.g. the recruitment of young judges, with no prior legal experience) with a common law jurisdiction (e.g. focus on judicial recruitment of legal professionals with significant legal experience). Having access to two different samples of judges, with different levels of experience (as well as exposure to different types of judicial training) but with the same historical and cultural background proved fortunate, because it was able to measure the impact of expertise on various judicial attitudes, without having to worry about a range of confounding variables that cannot typically be controlled for in cross-jurisdictional studies, such as different socio-political conditions, different judicial systems or different historical and cultural backgrounds. Judges from both types of recruitment and training routes were included in the study, along with a wide range of judges that have already served for years on the bench. The details of how the sampling was made and how the interventions took place are detailed in section III.
There are two key questions (and corresponding findings) that shall be presented in this article:
- Sentencing tools – How do judges in Romania say they use different tools in sentencing, and how does their prior experience relate to the way they use sentencing tools;
- Sentencing factors – How do judges say they approach the weighing of sentencing factors (and whether their approach changes with experience).
Section II briefly overviews the main theories and studies that have influenced the current research. As this is an interdisciplinary study, it drew influences from judicial studies (traditionally developed within political science and psychology), sociological studies on sentencing, as well as developmental theories of experiential learning. Due to space constraints, section II only focuses on the direct influences.
II. Theoretical Framework
While in the United States empirical legal studies judicial behaviour have been conducted for almost a century, judicial studies are still in their infancy in UK and Europe. The scholarship so far developed in the United States has proven a valuable lesson for European researchers when designing empirical studies that assess judicial decision-making. The initial attempts in the States, most probably inspired by the legal realist movement, focused on attempting to prove that the legal model (whereby a judge is merely applying the law in a case and none of her background or personal beliefs play any role) is not depicting how judicial decision-making actually works. For this reason, the earlier empirical studies measured one variable – the individual judge’s political orientation – as the key variable that shows there is much more to judicial decision-making than mere rule-applying. In time, the models derived became more sophisticated, either by incorporating a strategic dimension explaining the dynamics between judges, or the importance of economic factors in the decision-making process, or by incorporating the idea that judges are preoccupied by how they are perceived by various audiences, be it the general public, their colleagues, the mass media or colleagues from other branches of the government. Perhaps the most complex models were the ones to acknowledge the multi-faceted nature of the judicial decision-making activity, and to argue that it is simply wrong to only focus on just one type of factors that might influence judges.
Apart from the research activity produced in political science departments across the US in the past 70 years, psychologists have also brought incredibly important contributions to our understanding of the decision-making process. Most often, in these studies, judges were considered first and foremost as human beings, subjected to the same range of biases and unconscious processes as other people. Given this fundamental assumption, a flurry of studies made on judge or non-judge participants who were given judicial decision-making tasks revealed important problems and biases. Another strand of psychological studies took a step further and looked at how professional decision-makers differ from average human beings. These studies revealed that expert decision-makers differ from novice decision-makers in the way they make decisions – be it the factors they take into account, the weight they give to different factors, the speed of decision-making, or the fact that they have “hunches”.
In addition to research conducted on judges more generally, some sociological and criminological research has specifically focused on judicial sentencing. The empirical research on sentencing is split into two major strands of research: behavioural research on sentencing, on one hand, and research on attitudes towards sentencing, on the other. The common premise of both strands is that there is a whole range of factors that impact on the decisions of judges when they sentence. What distinguished the two strands was a methodological assumption on how to best find out what those factors are. The behavioural research adopted indirect measures of judicial attitudes (for instance, votes in panels, variation in sentencing decisions etc.), while the other strand of studies adopted direct measures of judicial attitudes (surveys, interviews, judicial notes). Almost none of these studies looked at judicial experience as an independent variable, apart from two major exceptions: Hogarth’s 1971 study on Canadian magistrates, and Jacobson and Hough’s 2007 study on English judges’ attitudes towards mitigation factors in sentencing.
Hogarth’s study was a sophisticated, mixed-method sociological study, the first to measure (1) the magistrates’ age, legal background, years on the bench, and to verify if these explained variations in their attitudes; (2) their attitudes towards legal instruments (such as guideline judgments from higher courts); and he was highly innovative in (3) looking at how “magistrates search for and use information in the process of coming to decisions” – which meant he asked magistrates to rank sentencing factors, as well as the sources of information used in deciding a sentence. One of Hogarth’s major findings was that
About 50 per cent of the variation in sentencing behaviour could be accounted for by knowing nothing about the cases and relying solely on three pieces of information about the magistrate [n.b. attitudinal variables].
This finding is very strong supporting evidence for the idea that studies that interview or survey judges are important. Unlike what critics of direct measures of attitudes might claim, the characteristics of judges and their beliefs about sentencing are much more powerfully predictive than just examining details of a case.
Finally, the current study is inspired by one of the major educational theories of the 20th century – the theory of experiential learning – through David Kolb’s theory of the Learning Cycle and the Learning Styles Inventory (LSI). Kolb’s theory is useful in explaining why it is that, during the formal training process, judges might have different needs at different points of the learning process, depending on their prior exposure to training and their prior legal experience. More importantly for this paper, experiential theory is also useful in explaining differences between judges in using different informal learning tools have in their day to day job, depending on the stage of their career development. Interestingly, although Kolb’s Learning Cycle has been replicated with learners from many professions, across more than 1,000 studies, it has never so far been validated on judges.
III. Methodology and Sampling
A. Research Questions and Variables Measured
This part of the study tested the hypothesis that judges of different levels of expertise will have different informal learning preferences and will approach sentencing differently. This hypothesis was based on the assumption (derived from the theoretical framework) that judges do not just learn during judicial training, but also from a variety of situations and sources outside of the training context. To explore this hypothesis, the research asked judges to share their thoughts on three main research questions:
- How judges in Romania say they use different tools in sentencing;
- How does their prior experience relate to the way they use sentencing tools;
- How do judges say they approach the weighing of sentencing factors (and whether their approach changes with experience);
Sentencing Tools. The first two questions refer to the so-called ‘sentencing tools.’ Sentencing tools represent the wide variety of legal materials and sources of information that the judge can use throughout his court practice in attempting to identify what the best sentence is in a particular case. In this sense, sentencing tools are complementary to formal training. In Romanian sentencing practice, the legally binding tools are:
- Sentence ranges (stipulated in the Criminal Code);
- Sentencing guidelines (stipulated in the Criminal Procedure Code);
- Landmark cases (“Appeals in the Interest of the Law” judgments);
While the non-binding sentencing tools are:
- Sentencing remarks from own court or other courts;
- Pre-sentence reports;
- Prosecutorial recommendations of appropriate sentence.
While the Romanian justice system has a hierarchical and unified structure that is typical of other European continental countries, whereby rights of appeal are automatically accepted and appeal judges have the power to review the sentence given by lower courts, its sentencing framework is much more hybrid, combining elements from both civil and common law jurisdictions.
Just like in many European countries, Romanian law defines and classifies criminal offences based on severity, offers a diversity of available punishments (custodial, community, fines, suspension of rights, security measures), and indicates a mandatory sentencing range for each offence. Sentencing ranges are comparable to most European countries, although it allows a wider judicial discretion on theft and rape, and it is slightly less lenient on burglary. Romanian criminal law also specifies instructions on how to compute a sentence for single, repeated and multiple offences, how to discount sentences for guilty pleas, and what sentencing factors judges ought to take into account in sentencing. Together, these sentencing guidelines are binding, but still allow judicial discretion. The timing of the sentencing process is also very similar to many other European jurisdictions – it takes place at the end of the litigation, after the defendant has been found or has pleaded guilty.
Similar to other continental justice systems, Romania does not officially acknowledge the doctrine of precedent, and so judges are not bound by judgments formulated in courts. Yet since 2005 a new legal instrument, called the “Appeal in the Interest of the Law” (RIL), allows the High Court of Cassation and Justice to issue guideline judgments. These judgments give guidance on interpretations of statutes, including on sentencing practices, and they are binding.
Apart from sentence ranges, sentencing guidelines and RILs, which are binding, three other non-binding tools constitute a source of information for judges in their sentencing practice:
- Sentencing information systems – sentences from previous cases – although there is no doctrine of precedent and Romanian judges are not bound by previous judgments of their court or by judgments from other courts, these judgments were still mentioned as useful sources of information on best judicial practices in sentencing. All Romanian judges have access to ECRIS, an internal database containing judgments from all courts. This system is similar to other sentencing information systems implemented in Canada, Australia, Scotland or Israel.
- Pre-sentence reports – in a case, the Romanian judge can ask the probation officer to write a pre-sentence report on the background and characteristics of the defendant. By law, the probation officer is instructed to “assist the court in the sentencing process”.  These reports are expected to constitute an important source of learning for the judge, as it provides insights into the defendant’s reoffending patterns, social background and circumstances, character and likelihood of reoffending. Pre-sentence report are a widespread judicial practice in countries such as England and Wales, United States, Canada, and New Zealand. In Europe, the Committee of Ministers has recommended all COE member states to allow probation services to formulate pre-sentence reports “on individual alleged offenders in order to assist, where applicable, the judicial authorities in deciding whether to prosecute or what would be the appropriate sanctions or measures”.
- Prosecutor’s sentence recommendation – in Romania, when charging a defendant, the prosecutor also customarily suggests the best offence allocation and, at the end, can suggest the appropriate sentence. The judge is not bound to follow the prosecutor’s recommendation, but it is expected that the prosecutor’s suggestion would be a good source of insight for the judge in learning what the customary sentence is in a specific type of case. This practice is also encountered in other European countries such as the Netherlands, or Germany.
Sentencing Factors. Most jurisdictions around the world stipulate lists of sentencing factors, in both common law and civil law traditions.  The Romanian sentencing guidelines stipulate that the most important two general criteria a judge ought to take into account are the general harm caused (considered an objective element of the offence) corroborated with the degree of dangerousness of the defendant (considered a subjective element of the offence). The guidelines then detail the underlying aggravating and mitigating factors that could account for the two criteria. On one hand, the harm caused is understood in two ways – first, towards society at large (the sentencing guidelines mention “socially-protected values”) – and secondly, towards the victim. On the other hand, the dangerousness of the defendant can be assessed by taking into account the offender’s previous convictions, his socio-economic and personal background, or his behaviour throughout the court proceedings.
Although the guidelines strive to be specific in detailing 7 main criteria judges must take into account, they do not also provide a hierarchy of importance to guide the judge in assessing which factor is more important than another, or which one to prioritise against another. For this reason, the survey included a ranking question that forced judges to assess the importance they assign to various sentencing factors.
In addition to the research questions pertaining to formal training and informal learning practices, a range of demographic and attitudinal input variables were collected for each respondent:
- The level of professional experience (split into 3 separate variables: years in the legal profession, years on the bench, years in hearing criminal cases, where applicable);
- The type of prior legal experience, if applicable;
- The amount of prior exposure to initial, induction and continuous training;
- The judicial specialty, if applicable (civil, criminal, etc).
Experience for the purposes of this study is a multi-layered concept that focuses not on general life experience but on professional experience more specifically. In addition, it distinguishes between the professional legal experience gained before judicial appointment, which is assumed to be qualitatively different from judicial experience, the experience of being a judge in general, as well as asking more specifically about the experience gained in hearing criminal cases, which is assumed to have an even tighter connection with attitudes towards sentencing. In order to explore the effect of these experience-related variables, the sample was stratified in such a way so as to contain respondents that have varying degrees of legal, judicial and criminal judicial experience. In addition, the data analysis phase included multivariate analysis to identify possible collinearity between experience-related variables and to identify genuine impact of each type of variable.
The main sample consisted of 226 judges with judicial experience ranging from 0 to 35 years on the bench (mean 11 years). They were randomly selected from all levels of the court and from all around the country. Half of the respondents had no judicial experience at the time of the survey, while the other half had already been hearing cases in court. In addition, out of those without judicial experience, the stratified sample contained both new appointees without any legal experience as well as appointees with at least 5 years of legal experience, to isolate legal experience from the potentially confounding effect of judicial experience.
Thirty-nine percent of the sample (N 88) consisted of young newly recruited judges through the continental-style judicial appointment route: Romanian citizens who have an undergraduate law degree are allowed to participate in a highly competitive state exam organised by the National Institute of Magistracy (NIM) – for this reason they shall be called in short “NIM route judicial appointees”. The number of allocated places for new judges is around 100 every year. The newly recruited judges from the sample are representative of the typically young and inexperienced law graduates specific to this type of judicial recruitment in continental jurisdictions. First, 85% of them had graduated from the law school during the past 4 years (median graduation year 2012), which suggests indeed they are young. Second, only 24% of them had some legal experience, and that experience was in the majority of cases not longer than 2 years. In contrast, as all other law graduates around the world, the Romanian law graduates had some legal internship experience before being appointed, the most frequent being a court internship (60% of respondents).
Twelve percent of the sample (N 28) consisted of the other category of newly appointed judges, more specific to common law systems, where candidates are required to have at least a few years of legal experience in addition to an undergraduate degree in law. For the purposes of clarity this category of respondents will be defined as “direct route judicial appointees”. The reason why the sample was relatively small was that in 2014 only 73 judges were recruited in total through this route, so the respondents represent 38% from the 2014 cohort. These respondents are both similar and different to the first category of respondents: similar in that they had just begun their judicial career and therefore have no experience on the bench; different in that they have nevertheless at least 5 years of practical legal experience. On average direct route appointees had 7 years of legal experience, with 95% of them having between 5 and 9.5 years of legal experience (SD 3.65). The typical legal background direct route appointees seem to come from is either advocacy (43%) or the police forces (57%).
Finally, forty-nine percent of the sample represents judges already appointed with varying degrees of judicial and sentencing experience (N 110). The judicial experience of this category of respondents, which for the sake of clarity will be called “experienced judges”, ranged from 8 months to 35 years on the bench, with 12 years of experience on average. 95% of respondents have between 4 and 20 years of experience. The experienced judges were also asked about their experience in criminal law more specifically: 95% had prior experience in criminal law; out of those, 70% gained that experience solely as judges (so, one can assume, while hearing criminal law cases and sentencing), while others also indicated additional roles in which they gained criminal law experience.
A secondary sample consisted of prosecutors (N 236) ranging from 0 to 29 years of prosecutorial experience (average experience 11 years). Although this is a study on judicial decision-making, therefore focusing mainly on judges, prosecutors were also included as a secondary sample in order to offer an external viewpoint from legal professionals who undergo the same training and have the same legal background, but have a different role in court. While the questions addressed to judges asked them about their own sentencing, the prosecutors were asked about how they think judges conduct their sentencing. This brought about an original and innovative vantage point on judicial decision-making. Although the focus of this paper is to present the findings on judicial attitudes towards decision-making, specific mentions will be made when prosecutors had statistically significant differences in their perceptions of judicial activity.
All the participants were recruited with the kind help of the Romanian National Institute of Magistracy (NIM), in reference to the stratified sampling requirements of the study.
The study consisted of two phases: the qualitative exploratory phase and the main quantitative phase. The qualitative exploratory phase was meant to help test and refine the questions, which were then included in the survey in the quantitative phase. It consisted of observations and semi-structured interviews with judicial appointees from all routes and with judicial trainers in charge with conducting sentencing training for both newly appointed judges (“direct route” and “NIM route”) and for experienced judges (continuous training).
The main phase consisted in a 20-question pen and paper survey. The survey was designed to not take longer than 10 minutes to fill in, and for that purpose it mainly contained closed single choice, multiple choice, rating and ranking questions. In addition, each type of respondent as described above received a survey only pertaining to their particular profile, which meant that no question skipping was necessary and no unnecessary was asked. That said, most questions were formulated similarly/identically so as to allow cross-sample comparisons.
The survey was applied during the centralised training sessions for all the types of respondents described: for the “NIM route” initial appointees, at the end of their criminal law module (including sentencing) during their first year of judicial training; for the “direct route” appointees, at the end of their criminal law induction training, taking place within the first month of appointment; for the experienced judges and prosecutors, during several continuous training sessions on criminal law and sentencing. The research project was presented at the beginning of the session, endorsed by the Institute, and the survey also contained a short description of the research project aims, as well as the assurances that the participation is entirely voluntary and anonymous. The researcher was in the room during the application of the survey and was ready to answer any queries.
All these techniques ensured a high response rate (70%). In total, 510 surveys were received. During the data clean-up and analysis phase, 470 surveys were retained (226 judges and 236 prosecutors).
The findings will be presented in the next section in turn.
A. Factors in Sentencing
The following figure summarises the aggregate responses from 216 judges, split by the three samples based on their past judicial and legal experience. The respondents had to rank all the 7 sentencing factors on a scale in order of their importance. The figure shows the percentage of respondents who chose one factor among their top 3 choices.
Figure 1: Factors that judges think ought to influence them in sentencing (N 226)
While the 7 sentencing factors were ordered depending on their overall importance for all the 216 respondents, the figure also shows the visual differences between subsets of the respondents. The first thing worthy of notice is that the ordering of the 7 sentencing factors is roughly the same across all 3 samples.
Perhaps not surprisingly, the general harm (done to society and the legal order) was considered among the most important sentencing factors. In many criminal justice systems, civil and common legal alike, harm to society or to the social values is seen as a definitional element of the criminal offence.
None of the category distributions were normally distributed (Shapiro-Wilk sig. <.05 for all factors for all subsets) therefore all the tests used subsequently on this question were non-parametric tests.
The first question was if there were any statistically significant differences in the three subsets of judges in their appreciation for different sentencing factors, as the figure above might be visually misleading. The second question was if the judicial experience in general (and experience in criminal cases in particular) could account for the differences among respondents.
Table 1: Variation of attitudes towards sentencing factors with experience (N 226)
|As experience (in each category) increases|
|Experience category||Greater appreciation of||Lesser appreciation of|
|Legal practical experience||General harm|
|Judicial experience||Defendant’s future prospects||Harm to the victim|
|Experience in criminal cases||Previous convictions||Harm to the victim|
Since the sentencing factor scores constitute an ordinal variable, a Kruskal-Wallis test was run for each of them. As the figure above also suggests, newly appointed judges with prior legal experience think the general harm caused ought to be significantly less important in sentencing than the other two categories of judges (H(2)=7.604 p= .022 mean rank 130 vs 105 and 105). It was not clear just from this test whether this might be due to their legal practice background or another factor.
In addition, newly appointed judges seem to give significantly more importance to the specific harm caused to the victim than experienced judges (H(2)=8.891 p=.012, mean rank 93 vs 118). This was further explored as it signalled that the experience on the bench might render judges less sensitive to the victim’s harm in time. A binary variable exp_prof_yes was computed across all judges’ subsets, variable which indicates if the respondent has any experience in hearing cases or not. An ordinal regression model confirms that judges with experience on the bench are less likely to consider harm to the victim an important factor in sentencing (Est. -.316 SE .245 p .035). A potential explanation for this series of findings is that when hearing cases, judges are also exposed to the defendant – their claims, their profile, their future prospects – and that seems to render them more sensitive towards the defendants as well not just towards the victim. This is further strengthened by an additional regression model confirming that judges with experience seem to take more into consideration the defendant’s future prospects in their sentencing (Est. .661 SE .245 p .007).
Even if Figure 1 above might visually suggest that judicial experience also has a significant impact on what judges think about the defendant’s previous convictions or the effect of the sentence on the defendant, the regression models do not seem to confirm this intuition (“previous convictions” p .118; “effect of sentence on defendant” p .331). An alternative hypothesis, which cannot be tested in this study, is that age (or life experience) is actually a predictor, as the 3 subsets of judges also differ in their age. Age was not a focus of this study but it could constitute the object of future studies.
While judicial experience does not correlate with a higher concern for previous convictions, this correlation was discovered for the secondary sample consisting of prosecutors. As mentioned in the introduction, prosecutors’ views on judicial sentencing are not a focus of this article, but this was a particular finding worth mentioning as it could throw light on the judges’ attitudes as well. The more experienced prosecutors are, the more they seem to believe judges do indeed consider previous convictions (Est -.046 SE .015 Wald 9.757 p .002) more important for their sentencing. In fact, an aggregate figure representing the prosecutors’ responses would show that they think previous convictions are more important than the harm caused to the victim in judges’ minds. This could mean that 1) prosecutors might indeed see in court that judges of different ages take previous convictions into account differently, while, at the same time, 2) being slightly biased due to the nature of their efforts into thinking that judges care about previous convictions more than they actually think.
B. Useful sentencing tools
Because experience was a central variable in the study, a natural question to ask was what do judges do in order to gain the experience they need to do their job? This question asked the respondents to reflect on what tools they find more useful in their sentencing activity. Each of the eight tools suggested needed to be rated on a 4-point Likert scale (very useful, relatively useful, doesn’t influence, counterproductive). The eight tools are of very different kinds and they emerged as the main sources of inspiration in the exploratory phase, but they naturally clustered in the survey findings. The figure below summarises the responses:
Figure 2: Tools that judges consider useful in their sentencing practice (N 221)
In this case the figure represents all judicial respondents, without indicating the respondents of the 3 subsets of judges, because the figure would be too complicated. This is compensated for by indicating in the following few pages where significant differences between the respondents were found, and on what basis. The responses were ordered based on the percentage of respondents considering a particular sentencing tool very useful, because there seemed to be a significant difference in which the respondents distinguished between “very useful” and “relatively useful”, which warranted a clustering of categories.
There are several aspects which need to be explained regarding the options given to respondents – as they might be particular to the Romanian justice system or they might have an ambiguous meaning. They will be explained in the subsections below. As the figure might already suggest, there are 3 main clusters of sentencing tools judges had significantly different perceptions of. They will be presented in turn.
The main finding of this study regarding the use of various tools in sentencing is that the more experienced judges are, the more likely they are to diversify the range of tools they use in their activity. Different types of experience seem to account for an increase in appreciation for different types of tools: for instance, judges with more experience on the bench seem to appreciate landmark cases more, while those with more sentencing experience also appreciate pre-sentence reports more, whilst taking prosecutor’s recommendations into account less. These findings are summarised in the table below and will be explored in more detail afterwards:
Table 2: Variation of attitudes towards sentencing tools with experience (N 226)
|As experience (in each category) increases|
|Experience category||Greater appreciation of||Lesser appreciation of|
|Legal practical experience||Sentencing guidelines
Sentencing remarks (own court)
|Judicial experience||Landmark cases|
|Experience in criminal cases||Landmark cases
Sentencing remarks (own court)
C. Binding sentencing tools
As previously explained, there are three types of sentencing tools that Romanian judges can use in their activity and are well-known in other world jurisdictions as well: the sentence ranges for every type of offence, the sentencing guidelines and landmark cases. As Figure 2 shows, between 73% and 80% Romanian judges considered each of these tools very useful to their activity, and almost all judges (98-99%) considered them either relatively useful or very useful.
It is important to note that by sentencing guidelines Romanian judges understand the legal indications stipulated by the criminal procedure code on how to compute a sentence by taking into account various aggravating and mitigating factors. While the Romanian judiciary does not have separate sentencing guidelines formulated and updated by a dedicated sentencing council, the criminal procedure code computation methods are very similar in giving precise indications on the steps and variables judges have to take or have to disregard while sentencing.
In addition, since it is a civil law jurisdiction, Romanian law does not follow the doctrine of the precedent. For this reason, it is not the case that every decision given by a higher (or a judge’s own court) is binding for future judgments. This is why, in this question addressed to the Romanian judges, a clear demarcation between “similar sentences given at one’s own court” and “landmark cases” was made. The Romanian judiciary has nonetheless a growing corpus of decisions that are binding – decided by the Romanian Supreme Court in view of ensuring consistency in judgment. The landmark cases resulted, including cases specifying specific sentences for a specific offence, or sentencing computations, are binding for Romanian judges.
While for sentence ranges there was no significant difference between the 3 types of judges surveyed, they had different views on the importance of sentencing guidelines and landmark cases.
Sentencing guidelines seemed to be considered significantly more useful by newly-appointed judges who had legal experience (mean rank 132) than newly appointed judges with no legal experience (mean rank 104) (H(2) = 6.758 p .034). An ordinal regression model testing that the amount of legal experience might significantly account for a change in attitude towards sentencing guidelines confirmed that, indeed, judges with more previous legal experience are more likely to consider sentencing guidelines useful (Est. .467 SE .235 Wald 3.953 p .047).
Landmark cases seemed to be considered more useful by experienced judges (mean rank 116) that by newly-appointed judges with no legal experience (mean rank 99) (H(2) = 8.047 p .018). An ordinal regression model testing that the existence of judicial experience might account for a change in attitude towards landmark cases confirmed that judges with no experience on the bench are less likely to consider landmark cases useful than experienced judges (Est. -.733 SE .340 Wald 4.656 p .031). The difference between judges with no sentencing experience and those with sentencing experience is even starker (Est. -.810 SE .341 Wald 5.644 p .018). In addition, prior legal experience before becoming a judge seems again to increase appreciation, this time for landmark cases (Est. .343 SE .160 Wald 4.576 p .032).
D. Non-binding sentencing tools
Apart from landmark cases issued by the Supreme Court, Romanian judges are not bound by sentences given in similar cases in their own court or other courts. That said, most judges (78-84%) do consider those useful or very useful sources of inspiration for their sentencing decisions, with the majority of them (64-67%) seeing them useful but perhaps not essential for their activity.
Yet again, the amount of prior legal experience before becoming judge seem to play a significant role in increasing appreciation for similar sentences as sources of inspiration in sentencing. Ordinal regressions confirm that, the more legal experience judges had before being appointed, the more likely they are to consider similar sentences from their own court (Est. .197 SE .098 Wald 4.073 p .044) and from other courts (Est. .210 SE .097 Wald 4.712 p .030) as useful sources of inspiration in their own sentencing practice. In addition, sentencing experience itself seems to have an impact: judges who never sentenced before are much less likely to appreciate sentences from one’s own court useful than judges who have at least some sentencing experience (Est. -.663 SE .306 Wald 4.680 p .031).
A second category of non-binding sentencing tools are made of the suggestions that other court actors make to judges regarding the appropriate sentence. In this survey, judges were asked how useful they consider the pre-sentence reports formulated by probation officers and the recommendations prosecutors make with regards to the appropriate sentence during the trial. While most judges considered these suggestions useful overall, only a very small percentage of them (6-16%) considered these recommendations essential for their sentencing decision. In fact, 43% of respondents considered that the prosecutors’ recommendations play no role whatsoever in their decision.
Newly appointed judges with prior legal experience were significantly less likely to think pre-sentence reports are useful in sentencing than judges with no legal experience (Est. 1.428 SE .474 Wald 9.068 p .003). Yet the amount of sentencing experience seems to have an opposite effect – that of making judges more reliable on pre-sentence reports than less experienced colleagues (Est. .104 SE .035 Wald 8.669 p .003).
Finally, the judges’ reliance on prosecutor’s recommendations seems to decrease as they gain sentencing experience (Est. .622 SE .288 Wald 4.663 p .031), which could be part of the explanation for why it is one of the least preferred sources of inspiration in sentencing.
V. Discussion and Conclusions
This research has sought to show that judicial decision-making is a dynamic process, whose ingredients might change in time and with experience. Judicial studies have so far sought to draw an accurate picture of what influences judges in their activity, but far too little attention was paid to the idea that judges’ behaviour and mental processes might change with time, and therefore an empirical study that does not measure time-related variables offers just a momentary “snapshot” of judicial decision-making. The chief finding of this study is that, indeed, judges of various levels of experience pay attention to different aspects of their activity, and can thus arrive at different results.
This article explored two more specific research questions:
- How do judges say they approach the weighing of sentencing factors (and whether their approach changes with experience);
- How do judges in Romania say they use different tools in sentencing, and how does their prior experience relate to the way they use sentencing tools.
With regards to the factors that judges think influence them in establishing a sentence, it was good to confirm that general harm remains the most important factor taken into account. This somewhat replicates Hogarth’s and Tombs’ studies on sentencing factors. But, perhaps more interestingly, this study found that experience on the bench increases the judge’s concern for the defendant and his future prospects, while judges without experience are more likely to empathise with the victim. The study found that there is a clear tension between the two.
The judges do not sentence in a vacuum, nor do they think they do. This study’s results on sentencing tools shows that judges use a variety of sources of inspiration when they need to calculate a sentence. Not surprisingly, they give a lot of attention to the instruments they are bound to take into account (sentence ranges, sentencing guidelines, landmark cases), which suggests they conform to the requirements of the rule of law. But at the same time, they also draw inspiration from a range of non-binding tools. Once again, experience seems to play a significant role in which instruments are seen more useful than others, and the findings suggest that the more experienced judges are, the more they diversify the range of tools they use in sentencing. Almost all sentencing tools display some level of variation with experience. Although a few studies explored the way in which magistrates search and process sentencing-related information, none of the previous research asked judges to rank sentencing tools by their relative usefulness to sentencing practice, which makes this study unique.
The author is aware that the focus on Romania might, at first glance, seem limited, but the actual findings are intuitive and interesting enough to warrant an extrapolation to other jurisdictions. Any initiatives in applying the same methodology for studies in other jurisdictions are much welcome. For instance, future research could attempt to combine an attitudinal study with behavioural studies of sentencing (e.g. including sentencing exercises or analysis of sentencing remarks), while maintaining an interest for expertise-related variables. Even more desirable would be a longitudinal study of judicial sentencing, which would entail willingness of judges to participate throughout their career.
 PhD Candidate in Judicial Studies at University College London, Faculty of Law.
 Gary Klein, Sources of Power: How People Make Decisions (MIT Press 1998); Amy L Baylor, ‘A U-Shaped Model for the Development of Intuition by Level of Expertise’ (2001) 19 New Ideas in Psychology 237; Barbara A Spellman, ‘On the Supposed Expertise of Judges in Evaluating Evidence’ <http://papers.ssrn.com/abstract=977854> accessed 24 December 2014; Daniel Kahneman and Gary Klein, ‘Conditions for Intuitive Expertise: A Failure to Disagree’ (2009) 64 American Psychologist 515.
 With the notable exception of Lee Epstein and others, ‘Do Political Preferences Change? A Longitudinal Study of U.S. Supreme Court Justices’ (1998) 60 The Journal of Politics 801. More recently, Niv and Lachman have discovered that more experienced Israeli judges are more likely to consider other judges’ sentencing more lenient than novice judges, although the samples were rather small and it is not very clear from the research design if the study is indeed longitudinal (if it has the same respondents over time). Moshe Bar Niv and Ran Lachman, ‘Judges’ Perspective on the Level of Punishment | EJLS – European Journal of Legal Studies’  European Journal of Legal Studies s IV.2 <http://www.ejls.eu/23/243UK.htm> accessed 9 September 2017.
 Sheldon Goldman, ‘Voting Behavior on the United States Courts of Appeals, 1961-1964’  The American Political Science Review 374; S Sidney Ulmer, ‘Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases: 1947-1956 Terms’ (1973) 17 American Journal of Political Science 622; Orley Ashenfelter, Theodore Eisenberg and Stewart J Schwab, ‘Politics and the Judiciary: The Influence of Judicial Background on Case Outcomes’ (1995) 24 Journal of Legal Studies 257; Kenneth L Manning, Bruce A Carroll and Robert A Carp, ‘Does Age Matter? Judicial Decision Making in Age Discrimination Cases*’ (2004) 85 Social Science Quarterly 1; Lee Epstein, Andrew D Martin and Andrew W Martin, ‘Does Age (Really) Matter? A Response to Manning, Carroll, and Carp’ (2004) 85 Social Science Quarterly 19.
 Daniel Kahneman, Thinking, Fast and Slow (1st edn, Farrar, Straus and Giroux 2011); Jonathan Evans, ‘Dual-Processing Accounts of Reasoning, Judgment, and Social Cognition’ (2008) 59 Annual Review of Psychology 255.
 Gary Klein, ‘Jumping to Conclusions Can Make for Good Decisions’ (2010) 177 Science News 32; Klein, Sources of Power (n 2).
 Kahneman and Klein (n 2) 241.
 Oliver Wendell Holmes, ‘1897. The Path of the Law’ (1896) 10 Harvard Law Review 457; Karl N Llewellyn, The Bramble Bush: On Our Law and Its Study (New York: Oceana Publications 1930).
 S Sidney Ulmer, ‘The Political Party Variable in the Michigan Supreme Court’ (1962) 11 J. Pub. L. 352; Glendon Schubert, ‘Judicial Attitudes and Voting Behavior: The 1961 Term of the United States Supreme Court’  Law and Contemporary Problems 100; Goldman (n 4).
 Jeffrey A Segal and Harold J Spaeth, The Supreme Court and the Attitudinal Model (Cambridge University Press 1994); Lee Epstein and Jack Knight, The Choices Justices Make (SAGE 1997).
 Richard A Posner, How Judges Think (Harvard University Press 2008).
 Lawrence Baum, Judges and Their Audiences: A Perspective on Judicial Behavior (Princeton University Press 2009).
 Paul H Robinson and Barbara A Spellman, ‘Sentencing Decisions: Matching the Decisionmaker to the Decision Nature’  Columbia Law Review 1124; Jennifer K Robbennolt, Robert MacCoun and John M Darley, ‘Multiple Constraint Satisfaction in Judging’ in David E Klein and Gregory Mitchell (eds), The Psychology of Judicial Decision Making (Oxford University Press 2010).
 Chris Guthrie, Jeffrey J Rachlinski and Andrew J Wistrich, ‘Inside the Judicial Mind’ (2000) 86 Cornell Law Review 777; Chris Guthrie, Jeffrey J Rachlinski and Andrew J Wistrich, ‘Blinking on the Bench: How Judges Decide Cases’ (2007) 93 Cornell L. Rev. 1; Keith J Holyoak, ‘Analogy’ in Robert G Morrison (ed), The Cambridge Handbook of Thinking and Reasoning (Cambridge University Press 2005).
 Gary Klein, Intuition at Work: Why Developing Your Gut Instincts Will Make You Better at What You Do (Currency/Doubleday 2003); Kahneman and Klein (n 2); Livingston Armytage, Educating Judges: Towards a New Model of Continuing Judicial Learning (Martinus Nijhoff Publishers 1996); Marta Sinclair, ‘Misconceptions About Intuition’ (2010) 21 Psychological Inquiry 378.
 George Everson, ‘The Human Element in Justice’ (1919) 10 Journal of the American Institute of Criminal Law and Criminology 90; Edward Green, Judicial Attitudes in Sentencing; a Study of the Factors Underlying the Sentencing Practice of the Criminal Court of Philadelphia. (Macmillan; St Martin’s Press 1961); Franz Exner, Studien über die Strafzumessungspraxis der deutschen Gerichte (Wiegandt 1931); Roger Hood, Sentencing in Magistrates’ Courts: A Study in Variations of Policy (Stevens 1962); William Edler von Eyben, Strafudmaling (GEC Gads Forlag 1950); Jorge Sobral Fernandez, ‘Judges and Their Decisions’ in Santiago Redondo, Advances in Psychology and Law: International Contributions (Walter de Gruyter 1997).
 John Hogarth, Sentencing as a Human Process (University of Toronto Press in association with the Centre of Criminology, University of Toronto 1971).
 Jessica Jacobson and Mike Hough, ‘Mitigation: The Role of Personal Factors in Sentencing’ (Prison Reform Trust 2007).
 Hogarth (n 17) 229.
 ibid 351.
 David A Kolb, Experiential Learning: Experience as the Source of Learning and Development (Prentice-Hall 1984).
 Alice Y Kolb and David A Kolb, ‘The Kolb Learning Style Inventory — Version 3.1 2005 Technical Specifications’ (2005).
 Carlo Guarnieri, Patrizia Pederzoli, and Cheryl Thomas (English Editor) The Power of Judges: A Comparative Study of Courts and Democracy (Oxford University Press 2002).
 Which means that virtually all judges hearing criminal cases have sentencing powers in Romania.
 Noul Cod Penal (v 1 Feb 2014) 2014 ss 188–445 Partea Speciala.
 ibid 53–70, 107–12, 115–25.
 Romania has the same sentence ranges as most European countries for offences such as intentional homicide (10+ years), robbery (2-5 years), assault (up to 2 years), sexual assault (2-7 years), and drug offences (2-5 years).
 For theft, Romania has a very wide range (6 months – 3 years) allowing judges a wider sentencing discretion than most other European counterparts. For rape, Romania offers again a wider discretion, by allowing judges to award custody anywhere between 3 and 10 years – other European jurisdictions seem to be split, with some countries awarding 2-5 years, while others award 5-10 years of custody.
 For burglary, Romania resembles only England and Wales (2-7 years), as most European countries are more lenient with this category of offence (less than 2 years custody).
 Noul Cod Penal (v 1 Feb 2014) (n 26) ss 35–35.
 Noul Cod de Procedura Penala 2014 s 374.
 Noul Cod Penal (v 1 Feb 2014) (n 26) s 77.
 Noul Cod de Procedura Penala (n 32) s 474.
 Michael Tonry, ‘Punishment Policies and Patterns in Western Countries’ in Richard S Frase and Michael Tonry (eds), Sentencing and Sanctions in Western Countries (Oxford University Press 2001) 21–4; Uri J Schild, ‘Statistical Information Systems for Sentencing: The Israeli Approach’ (2000) 14 International Review of Law, Computers & Technology 317.
 Parlamentul Romaniei, Legea 252/2013 privind organizarea si functionarea sistemului de probatiune 2013 s 32.
 Cyrus Tata and others, ‘Assisting and Advising The Sentencing Decision Process The Pursuit of “Quality” in Pre-Sentence Reports’ (2008) 48 British Journal of Criminology 835.
 Green (n 16) 22; Robert M Carter and Leslie T Wilkins, ‘Some Factors in Sentencing Policy’ (1967) 58 Journal of Criminal Law, Criminology and Police Science 503; Vladimir Konecni and Ebbe B Ebbesen, ‘The Mythology of Legal Decision Making’ (1984) 7 International Journal of Law and Psychiatry 5.
 Hogarth (n 17) 240–61.
 Christina Rush and Jeremy Robertson, ‘Presentence Reports: The Utility of Information to the Sentencing Decision’ (1987) 11 Law and Human Behavior 147.
 Committee of Ministers, Recommendation CM/Rec(2010)1 on the Council of Europe Probation Rules 2010 s 42.
 Noul Cod Penal (v 1 Feb 2014) (n 26) s 390.
 Sentencing Commission for Scotland, ‘Sentencing Guidelines Around the World’ (2015) 37 <https://www.scottishsentencingcouncil.org.uk/media/1109/paper-31a-sentencing-guidelines-around-the-world.pdf> accessed 28 November 2016.
 ibid 42.
 “Almost all common law jurisdictions have placed certain mitigating and aggravating factors on a statutory footing.” (Julian Roberts, ‘Punishing, More or Less’ in Julian V Roberts (ed), Mitigation and Aggravation at Sentencing (Cambridge University Press 2011) 6.)
 Andrew Ashworth, ‘Re-Evaluating the Justifications for Aggravation and Mitigation at Sentencing’ in Julian V Roberts (ed), Mitigation and Aggravation at Sentencing (Cambridge University Press 2011) 21.
 In order to isolate the effect of judicial experience on the attitudes of respondents towards sentencing tools and sentencing factors from any other potentially-confounding variables, the study asked both non-experienced respondents (experience=0 years) and experienced respondents (experience >0 years) the same questions. In addition, asking both novice “NIM route” and novice “direct route” appointees the same questions, the study was also able to isolate the effect of the legal experience variable.
 The selection was made by an automated program operated by the Romanian National Institute of Magistracy for selection of judges for centralized judicial training sessions. Each court in the country has a number of allocated places, the judges have to opt in themselves but the algorithm does not allow them to participate in training more than once in a couple of years, to ensure all judges get the same access to judicial training overall.
 In October 2013 3,733 candidates applied for the 200 places available (19 candidates per place). Out of the 200 selected trainees, 100 will become judges and 100 will become prosecutors. 88 out of the 100 judges of the 2013 generation were included in the study Institutul National al Magistraturii, ‘Tabel Rezultate Proba Eliminatorie de Verif. Cunostinte Jur. (30.08.13)’ (2013) <http://www.inm-lex.ro/fisiere/d_198/Tabel%20rezultate%20proba%20eliminatorie%20de%20verif.cunostinte%20jur.(30.08.13).pdf> accessed 20 March 2015.
 74% of the respondents had at least one legal internship during their studies, median 3 weeks long.
 Apart from being rather small, this sample is also biased geographically because 37 of the judges were located remotely from the survey (based on the location of their courts) and for technical reasons they could not be included in the study. While the small sample prevents generalisations, it does nevertheless offer interesting insight into how judges with some legal experience differ in their attitudes from the first category of respondents, but also from the judges who already gained judicial experience.
 14% as lawyer/legal advisor, 6% as prosecutor, 2% as police officer and 7% as other (clerk, legal academic).
 The survey also contained questions regarding attitudes towards the judicial training offered on sentencing, but the corresponding findings are not central for the thesis of this article and are therefore not included here.
 This explains why the difference was marked by an asterisk in Figure 1, being the only significant difference in perception between judges and prosecutors on this matter.
 This is in tension with what the prosecutors themselves think about how judges consider their recommendations – the more experienced are, the more likely they are to think judges consider their recommendations useful (Est. .042 SE .016 Wald 6.670 p .010).
 Although comparisons need to be treated with caution as these previous studies used slightly different categorisations of sentencing factors (Hogarth (n 17) 281 Table 86; Jacqueline Tombs, A Unique Punishment: Sentencing and the Prison Population in Scotland (Scottish Consortium on Crime and Criminal Justice 2004) 49 Table 4.1.).
 Hogarth (n 17) 239–44.
 Along the lines of Jacobson and Hough’s, Hogarth’s or Arce’s mixed-method studies (Jacobson and Hough (n 19); Hogarth (n 17); Ramon Arce and others, ‘Judges’ Decision-Making from within’ in Raymond R Corrado, Rebecca Dempster and Ronald Roesch (eds), Psychology in the Courts (Routledge 2001).)