Are judges influenced by legally irrelevant circumstances?

,

It is fundamental for the rule of law that judges are not influenced by legally irrelevant circumstances.There could, for example, be a circumstance in the personal history of the plaintiff that inspires †Email: my.bergius@jur.lu.seV C The Author(s) [2020].Published by Oxford University Press.All rights reserved This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
sympathy for the plaintiff as a person, but is irrelevant to the legal merits of the plaintiff's case, or a circumstance about the defendant that provokes disgust for the defendant as a person, but is legally irrelevant.Such circumstances may move the private feelings of the judge, but they must never be allowedto influence the legal decision.The aim of the present study was to investigate if judges are affected by an admissable legally irrelevant negative personal circumstance.Judges generally believe that they are capable of ignoring irrelevant information and that they are objective and impartial in their decision-making (Wistrich et al., 2004;Dhami and Belton, 2017).Research has however shown that a large number of legally irrelevant factors may influence legal decision making.For example, legal decisions may be influenced by personal characteristics of the judge such as age, gender and political orientation (Coontz, 2000;Kulik et al., 2003).Similarly, irrelevant aspects of the defendant, such as their race, gender or age, have been shown to affect judges legal decision-making (Doerner and Demuth, 2010).A large portion of research on how legally irrelevant factors affects legal decision-making has focused on aspects of the involved parties, such as judge or defendant characteristics.In addition to this, a number of studies have examined the effect of other types of legally irrelevant information on legal decision-making.For example, Guthrie et al. (2000) found that anchors can trigger intuitive judicial decision-making and that judges who were presented with a legally irrelevant anchor were affected to award an injured party lower compensation compared to judges who were not presented with the anchor.Similarly, Englich et al. (2006) found that irrelevant anchors affected judges' sentencing decisions, even when judges were informed that the anchor was random.Experience and expertise did not reduce the anchoring effect.
Another type of legally irrelevant information that can be presented in court is irrelevant evidence.In many jurisdictions, irrelevant evidence will be inadmissible.This is the case in for example in North America, where the majority of research outlined in the introduction was conducted.Since the present study was conducted with a sample of Swedish judges, some remarks on the Swedish legal system are in order before moving on to a description of how irrelevant evidence affects legal decision-making.Two important principles in Swedish laws of evidence are the principle of free evaluation of evidence and the principle of free submission of evidence (Ekelo ¨f et al., 2009).This approach to evidence is different from many other judicial systems where the submission of evidence and the boundaries thereof are more rigorously regulated.In comparison to systems with regulated evidence law the Swedish system has a very low degree of regulation.The kind of regulated law of evidence that does exist in for example the Federal Rules of Evidence does not exist in the Swedish Code of Judicial Procedure.There are however some rules in the Swedish Code of Judicial Procedure that limits the possibility for parties to introduce evidence.The possibility to adduce written statements are limited and there is also a regulation that limits the possibility to hear a witness about classified information that he or she possess (Swedish Code of Judicial Procedure; SCOJP, 35 chapter 14 § and 36 chapter 5).Swedish courts also have the possibility to reject evidence if they make the assessment that the evidence a party wants to put forward is unnecessary or unreasonably expensive (SCOJP; 35 chapter 7 §.).There is no regulation of character evidence that limits the ability to invoke criminal charges or other personal circumstances as evidence.According to the preparatory works of the Swedish Code of Judicial Procedure it is not the legislator's intent that the possibility to reject evidence should be used to prevent information about criminal charges or other character evidence from being put forward by a party.It appears from the preparatory work that character evidence normally should be regarded as permitted.A general ban of character evidence had not been consistent with the principle of free evidence and therefor the possibility to reject character evidence should only be used in exceptional cases when the evidence is presented with the sole purpose of harassing the person (NJA II 1943, p. 451).According to Swedish law of evidence parties are therefore allowed to put forward evidence regarding a part or a witness previous criminal record in both civil and criminal cases.This means that evidence concerning criminal record are allowed concerning both the question of a person's credibility and the question of guilt.Therefore, irrelevant circumstances, sometimes in the form of previous convictions, in accordance with the current Swedish evidence law can be presented in Swedish courts.
Even if irrelevant evidence is rarely deemed inadmissible in Swedish legal contexts, the effect of inadmissible evidence on legal decision-making is relevant to the present study as it is an area of research that has examined the effect of irrelevant evidence on legal decision-making.Decades of research into the effect of inadmissible evidence on legal decision-making has found that mock jurors are indeed influenced by inadmissible evidence in their decisions (see e.g., Sue et al., 1973;Carretta and Moreland, 1983 for early works; see Steblay et al., 2006, for a meta-analysis).The majority of research into the effect of inadmissible evidence on legal decision-making have studied how mock jurors' decisions are affected, but a few studies have examined professional judges and found that their decisions are as likely to be affected by inadmissible evidence (e.g., Landsman and Rakos, 1994).
Thus, it seems that even if evidence is legally irrelevant, it can still affect judges' decision making.The very nature of the decision making can also leave room for biases.For example, in deciding whether a person (A) has committed libel and should compensate the injured party (B), judges need to consider whether the condition for liability is fulfilled-namely if the A caused the injury through negligence.In this case, what is legally relevant is for the judge to consider reasons for classifying A's behaviour as negligent or non-negligent.However, judgements such as these may be influenced by hindsight bias.In deciding whether someone's' behaviour was negligent, judges should consider how foreseeable the damage was.Oeberst and Goeckenjan (2016) argued that since legal decisions are always made after the fact (i.e., the damage has already occurred), the damage will be perceived as more foreseeable, and found that professional judges' negligence judgements were influenced by hindsight bias.
Moreover, irrelevant evidence may affect judge's decision making.Irrelevant aspects of the plaintiff such as previous criminal records have been shown to affect legal decision-making (Wistrich et al., 2004).In a trial between A and B, information about A's past behaviuor towards an unrelated person (C) has nothing to do with the present situation and is therefore legally irrelevant.Even if the information about A's past behaviour is legally irrelevant, it could still be relevant outside the law.A heuristic that works well outside the law could therefore turn into bias if it is used by a legal decision maker.Is A's claim fair?Outside the law people may use a rule of thumb that we will call the clean-hands-heuristic.According to this heuristic you are not entitled to complain about a behaviour if you behave in the same way yourself.Parents often use the clean-handsheuristic to handle conflicts between children: 'You can't complain about your sister taking your sweater without permission, since you took her sweater without permission yesterday.'However, using this heuristic in legal decision-making would be a mistake.Since 'clean hands' is an irrelevant circumstance in legal circumstances, the clean-hands-heuristic becomes a bias, when it is imported into the law.

Thepresent study
Research had demonstrated that a number of legally irrelevant factors, such as irrelevant anchors, inadmissible evidence and characteristics of the decision maker or the persons the decisions concern, may affect legal decision-making (e.g., Guthrie et al., 2000;Coontz, 2000;Steblay et al., 2006;Doerner and Demuth, 2010).The aim of the present study was to examine how irrelevant negative information about the plaintiff affected professional judges' decision making.To our knowledge, this has only been investigated in one previous experiment (Wistrich et al., 2004).In our study, the 'irrelevant negative information' consists in the fact that the plaintiff (A) in a libel case against (B) has previously committed libel against an unrelated third party (C).A does not have 'clean hands' with regard to libel, but this is irrelevant to the legal merits of As claim against B.

Participants
The experiment was conducted through a questionnaire that was distributed to all professionally trained judges at municipal courts in Sweden.A total of 629 judges received the questionnaire, and 260 (41%) responded.After screening for outliers, the final sample consisted of 256 judges (119 females, 137 males).Information on age was not collected.The participating judges had varying experience, from so-called fiscal judges who were undergoing judge training, which in Sweden means serving in various courts for a number of years, to experienced judges who had long ago completed their training.No data on years of experience was collected.

Procedure
Data for this study was collected as part of a larger data collection.Chief judges at all Swedish District Courts (N ¼ 48) were contacted and with their permission the test questions were sent to all judges currently employed at a District court.In total, there were ten vignettes, each followed by one question, which were printed on paper and bunched together as a booklet with a cover page and a consent form describing overall aims of the project.These documents were sent to the participants through regular post together with a returned envelope.The vignette and questions used for the present study was placed at the 9 th position in the series of 10 test questions.The remaining test questions were asked in relation to different vignettes and were part of a different research project, and will therefore not be reported here.Participants returned the questionnaires through regular post using the provided return envelope.

Design
The study employed a between subjects design with two conditions.The experimental condition, which will be referred to as the Irrelevancy condition and the control condition.Participants in the Irrelevancy condition were provided with legally irrelevant negative information about the plaintiff (Magnus P), while participants in the control condition were not provided with this information (Magnus P).

Materials
The study was conducted using a vignette describing a case of libel.There were two versions of the vignette, the experimental version (irrelevancy condition) and the control versions.Both versions are presented below.

Control version
In the control version, the information about Magnus P's previous conviction and the post on Facebook was left out and was therefore not known by the judges.In all other aspects, the vignette was identical to the experimental version.

Measures
At the end of the vignette, the judges were asked to specify the amount in Swedish Kronor (SEK, as of July 2018, 1 SEK equals to approximately 0.10 USD) they would consider reasonable compensation for Lisa H to pay Magnus P in damages.The amount specified by the judges was used as the outcome variable.

Results
The data (N ¼ 260) was checked for missing values, outliers, and assumption of normality before conducting the analysis.Inspection of data showed no missing value, and outliers were checked by transforming the participants' scores into z-scores.The participants (n ¼ 4) whose z-scores were higher than 3 were considered outliers and were removed.The assumption of normality was tested through Shapiro Wilk (S-W) test.Results of S-W test for normality (SW ¼ 0.72, df ¼ 256, p ¼ 0.0001), and Skewness ¼ 2.11, and Kurtosis ¼ 4.43 suggest that the data violated the assumption of normality.
Since the assumption of normality was violated, the data (N ¼ 256) was analysed using the Mann-Whitney U-test.The Mann-Whitney U-test is one of the most powerful non-parametric tests (Landers, 1981) and is recommended when the assumptions of a t-test has not been met.
The Mann-Whitney U-test of the difference between the Irrelevancy condition (the group that was provided with irrelevant information) and the Control condition showed that the participants in the Irrelevancy condition suggested a significantly lower amount of damage compensation than the participants in the Control condition; U ¼ 5531, z ¼ 4.6, p < 0.001, r ¼ 0.29.See Table 1 for descriptive statistics from the two conditions.

Discussion
The aim of the present study was to examine if professional judges were affected by a legally irrelevant circumstance in their decision making.Swedish judges read a vignette depicting a case of libel, and half of the participants received legally irrelevant information that the plaintiff had previously been convicted of libel himself.The judges were then asked what damage compensation the plaintiff should be paid by the defendant.The results show that the professionally trained judges were influenced by irrelevant circumstances.The median amount of damage compensation decided by the judges in the control condition was 10 000 SEK, while the judges who received the legally irrelevant information settled for 7500 SEK.Thus, the irrelevant information about the plaintiff's previous conviction influenced the judges to award him 25% less in the present case.
The findings are in line with those reported by Wistrich and colleagues (2004), who found that legally trained judges from the USA who were exposed to information about a plaintiff's previous criminal record awards less in pain and suffering damages.Even though there are similarities between their study and ours there are also differences.First of all in the legal context.As reviewed in the introduction, the view on the admissibility of character evidence are as shown above quite different between the USA and Sweden, with the Swedish legal system relying heavily on principles of free submission of evidence.There are also differences in the scenarios that were tested.While both studies tested the effects of a plaintiff s previous conviction the background, Wistrich and colleagues ( 2004) presented judges with a scenario where the plaintiff was previously convicted of fraud and the case at hand was a product liability suit.The previous conviction had nothing to do with the scenario at hand.The scenario was not based on a clean-hands-heuristic but rather that the plaintiff was unlikable in general.In our scenario, the plaintiff was previously convicted for the same crime, libel, as he now sought compensation for, which could trigger the clean hands heuristic, that he is not entitled to complain about a behaviour that he himself has been found guilty of.Thus, the present study extends the findings from previous research, showing that information about a plaintiff's previous conviction can affect legal decision-making under different circumstances, and in legal system where such evidence is treated very differently.
The results add to a body of research showing that legal decision-making can be influenced by legally irrelevant factors, and that even professional judges are not immune to its influence.Even if Swedish rules of free evaluation and submission of evidence means that irrelevant evidence such as that used in the present study (information that the plaintiff had a previous conviction of libel), would not be inadmissible, the results are in similar to the findings by Landsman and Rakos (1994) and Wistrich et al., (2004) who found that professional judges were affected by inadmissible evidence.
Some of the participants were interviewed after the experiment, and participants in the irrelevancy condition were asked specifically whether they were influenced in their decision by the information about Magnus P's previous libel case.All of them said that they had not been influenced by this circumstance.This unawareness is very interesting, and should be investigated further.This is in line with previous research showing that people are often unaware of factors that influence their judgements (e.g., Nisbett and Wilson, 1977).A greater awareness among judges about their vulnerability to irrelevant circumstances could help them resist influence.
In the introduction, we proposed a 'clean hands heuristic'-that you are not entitled to complain about another's behaviour if you behave the same way yourself.This heuristic may hold up well outside the law, but becomes problematic when used in legal decision-making.The judges who received information that the defendant had been convicted of libel himself, used this piece of information to award him less damage compensation than judges who did not have this information about the plaintiff.Thus, the judges in the irrelevancy condition may have reasoned that the plaintiff did not deserve a high amount of damage compensation, since he had previously behaved the same way himself, a reasoning that is in line with the clean hands heuristic.The plaintiff's previous conviction was legally irrelevant to the case of libel presented to the judges however, and the findings thus illustrate that judges may use heuristic reasoning in their decision making.
The results call for further research on how to counteract the effect of irrelevant information on legal decision-making.For example, research into the effect of inadmissible evidence on juror decision making has found that juror deliberations can temper the effect of inadmissible evidence (London and Nunez, 2000).In real life, professional judges in Swedish district courts evaluate the evidence and then write up a verdict, in which they state the grounds for their decision.Being required to state the grounds for ones' decision may mitigate the effect of irrelevant circumstances, but it is also possible that it would not be helpful since people may not be aware of the circumstances that affected their decision people are often unaware of factors that influence their judgements (Nisbett and Wilson, 1977).Further research could examine whether deliberation or having to state the grounds for the decision can counteract the effect of irrelevant evidence on legal decisionmaking.

Limitations
Before moving on to the conclusions, the present study has some limitations that should be addressed.Research into legal decision making has been criticized for having limited application to actual legal decision making (Dhami and Belton, 2017).It must be remembered that the experiment was conducted by a vignette.The judges were thus presented with condensed information in text format.It does not fully correspond to the environment in which legally trained judges normally decides a case, where there is much more information to be considered and this information is presented orally.It is possible that the judges who participated in the experiment had acted with more prudence in a real case.Also, we did not ask the judges for their years of experience, meaning we were unable to examine if experience meant that judges were more or less likely to be affected by the irrelevant information.Further research into professional judges' decision making could benefit from taking experience into account.Still, the present study used a rigorous sample consisting of nearly half (41%) of professional district court judges in Sweden meaning that we were able to provide important data using an understudied population.

Conclusions
In conclusion, the self-image of professional judges who tend to believe that they are capable of being objective and unaffected by irrelevant circumstances (Dhami and Belton, 2017;Wistrich et al., 2004) does not seem to hold up.Our experiment shows that judges were influenced by negative circumstances in the plaintiff's personal history.Further research is required to identify methods to mitigate the effect of irrelevant evidence on legal decision-making.

Funding
The research is founded by the Swedish Research Council.

Table 1 .
N, median damage compensation chosen by the judges in SEK, mean rank and sum of ranks for the Mann-Whitney U-test of difference between the control and irrelevancy condition Lisa H is studying law at Stockholm University.On 20 September 2012, she writes on her blog that she has been sexually harassed by one of her teachers, university lector Magnus P. Lisa H names Magnus P as the perpetrator and writes that he has been sending dirty text messages to her.Lisa H's story quickly becomes the focus of attention and her blog receives record number of visitors.Magnus P says that Lisa H is lying and sues her for libel.During the trial, Lisa H admits that she has not received any text messages from Magnus P.She had heard a rumour about a teacher who sent dirty text messages to female students and guessed it was Magnus P. Lisa H is convicted of libel.During the trial it is brought up that Magnus P himself has been convicted of libel in the previous year.Magnus P had made a post on Facebook about a colleague at Stockholm University, ¨ Professor Susanne O, and in the post wrongly stated that she made a career by having an affair with her boss.