林喜芬、马勇 | 量刑建议、锚定效应与刑事司法公正

Sentencing Recommendations, Anchoring Effect and Fairness in Criminal Justice

——An Empirical Study Based on a Sample of 520 Sentences in K City

*林喜芬

上海交通大学凯原法学院教授

*马勇

中国人民大学财政金融学院教授

摘 要：锚定效应是决策领域中非常普遍且强劲的一种认知现象。通过520个量刑样本的量化分析，本文发现：无论是以法官判处的自由刑量（freedom penalty）、罚金量（fine penalty）还是缓刑量（term of probation）作为考察对象，检察院的量刑建议对法院量刑裁判均产生了显著影响，即“锚定效应”是存在的。此外，“定罪罪名发生变化”、“庭审情节发生变化”以及“辩护人的存在”会弱化检察院量刑建议的“锚定效应”。作为一种认知偏差，锚定效应在量刑领域的存在，进一步增强了法律现实主义分析在中国刑事司法领域应用的必要性，同时，也在制度层面要求法官需要采取相应机制来保障自由裁量的公正行使。

The anchoring effect is a powerful and widespread cognitive phenomenon in the decision-making field. Our quantitative analysis of a sample of 520 sentences indicates that the sentencing recommendation of the public procuratorate has a marked influence upon the court’s sentencing judgment. That is, whether we are investigating the freedom penalty, the fine penalty or the term of probation imposed by the judge, we find that the anchoring effect does exist. In addition, a change in the court conviction or in the trial sentencing circumstances and the availability of a defense lawyer may weaken the anchoring effect of the procuratorate’s sentencing recommendations. As a form of cognitive bias, the presence of the anchoring effect in the area of sentencing further highlights the necessity of applying analyses based on legal realism to the field of criminal justice in China; and at the institutional level, it demands that judges adopt corresponding arrangements to ensure the impartial exercise of discretion.

I. Introduction: Judicial Discretion and Judicial Fairness

How do judge think and exercise their discretion? This is not a simple question. Mechanical legalism, the outcome of European rationalism of the 18th-19th centuries, thought of the judge as a vending machine into which facts (minor premises) and laws (major premises) were inserted, and which then disgorged judgements (conclusions). Legal realism has questioned legalist theory, arguing that judges’ rulings are  influenced not only by country-specific culture, institutional context and political beliefs, but also by psychological factors (cognitive bias or personal background and feelings). This theory may be readily acceptable abroad (especially in countries with the Anglo-American legal system), but it remains a challenge to the sensitivities of jurisprudential and legal circles in China. China is still, after all, in a critical stage of progress towards the modern rule of law, and judicial thinking centered on formal rationality has yet to be fully absorbed. Moreover, if we start from strict legal positivism, we urgently need an answer to the predetermined question of whether Chinese judges are influenced by factors outside legal provisions. Are their judgments affected by latent cognitive bias (anchoring effects, cognitive fluency, hindsight, and so forth)? Some scattered recent research has shown that latent features such as cognitive fluency and hindsight do play a part when Chinese judges make decisions. Given the problematic of this article—the question of sentencing fairness in criminal justice—we use a sample of 520 sentences from K City courts to see whether criminal judges in China are influenced by the anchoring effect, and more specifically, whether the sentencing recommendations of the public procurator exert an anchoring effect upon the sentence imposed by the judge. 

Since 1979, the central task of China’s rule of law reform of criminal justice has been the establishment of judicial proceedings modeled on due process. In practice, however, more than 80 percent of defendants in criminal cases plead guilty. In the judicial practice of the great majority of criminal cases, at least, the crucial issue for equity in criminal justice is not the legitimacy of the conviction process but the justice of the sentencing procedure. Previous sentencing practice in China has faced two major difficulties, those of “sentencing procedure is not independent enough” and “judges have too much discretion over sentencing.” How, then, do we achieve just and standardized sentencing procedures? What is required is, firstly, the trialing a relatively independent sentencing procedure that would free the defense case from the predicament arising from the inseparability of convicton and sentencing procedures; that is, it is hard for the accused to maintain his innocence while simultaneously asking for a lighter sentence on the grounds that he repents of his crime. Second is the implementation of a system of procuratorial sentencing recommendations, with a view to restricting judicial discretion. The latter measure was trialed in a similar pilot program in base-level procuratorates as early as around 2000; the Supreme People’s Procuratorate formally issued “Opinions on the Implementation of a Pilot Program of Sentencing Recommendations from People’s Procuratorates” in 2005; and in 2010, the Supreme People’s Court, the Supreme People’s Procuratorate, the Ministry of Public Security, the Ministry of State Security and the Ministry of Justice jointly formulated the “Opinions on Several Issues concerning the Regulation of Sentencing Procedures (for Trial Implementation)”, making clear the basic principle of “ensuring the independence of sentencing procedure,” and conferring the power to make sentencing recommendations on the procuratorate and the right to express sentencing opinions on the defense. In 2014, the Standing Committee of the National People’s Congress issued a “Decision” on  launching a pilot program of expedited sentencing and a system of leniency for those who plead guilty and accept punishment. Since then, it has become normal in some pilot sites for the public procurator to make relatively precise sentencing recommendations. The development of this system has seen sentencing recommendations applied increasingly extensively, and  they are now  viewed as an important measure for the standardization of courtroom sentencing. One could say that the degree of rule of law justice in routine criminal cases is determined by the relationship between the “prosecution” (sentencing recommendation) of the public procurator and the “judgement” (sentencing decision) of the judge. 

However, the question is: in the course of limiting judicial discretion and realizing judicial fairness, does the public procurator’s sentencing recommendation imperceptibly influence or even dominate the sentence imposed by the judge? In sentencing, are judges more inclined to rely on cognitive heuristics such as anchoring effects, as other decision-makers do, especially when time limits on trials pressure them to expedite their handling of complex case information? More specifically, does the sentencing recommendation really serve merely as a reference for the judge? What about the problem of “the more serious the charge, the harsher the sentence?” And for the accused, does no mention of a sentencing recommendation (especially one with a specific numerical value) mean that the prosecution has partly lost the case? These questions, which relate to the judicial fairness of the great majority of criminal cases, demand an urgent in-depth empirical answer along the cognitive dimension.

II. Literature Review and Problem Assumptions

As a cognitive phenomenon, the anchoring effect was first proposed and systematically expounded by Amos Tversky and Daniel Kahneman. It refers to the fact that people make quantitative estimates by starting from an initial value, and these initial values act as an “anchor” constraining their estimations. In making a decision, people unconsciously pay too much attention to their initial information. Since this theory was proposed, there have been a number of developments in the field of general cognition. Firstly, researchers have provided unimpugnable experimental data proving that the decision-making process is influenced by the anchoring effect. Whether the reference values are related to the content of decision-making or are totally random, and whether or not they are numerical, the anchoring effect is always strong. Secondly, as the literature indicates, in the course of decision-making, anchoring effects are strongest where there is greater ambiguity in the contents of decision-making; greater reliability in the sources and reference values of the reference information; less familiarity on the part of the decision-makers with decision-making contents; greater attention given by decision-makers to decision-making contents; and their greater involvement in the decision-making process. Finally, some research has shown that there are ways of weakening anchoring effects.

The above research was all carried out within a laboratory environment and the test takers were all college students, so the question was raised as to whether this kind of anchoring effect would exist in professional decision-making fields such as justice, value estimates, purchases, negotiations and self-efficacy evaluation. In particular, would the anchoring effect also apply to judges, with their wealth of experience and strict professional training—judges whose decisions are so important to all the parties in a case and to the public (and may even be a matter of life or death)? A series of subsequent studies continues to demonstrate the existence of extremely strong anchoring effects. In the field of legal decision-making, a good deal of the research has been done on tort compensation, with a series of studies finding that the higher the plaintiff’s compensation claim, the stronger the support received from the court. A similar phenomenon is found in jury trials: the plaintiff’s prior claim for compensation affects not only the amount of compensation but also on the probability of winning the case. Ironically, a legal cap on compensation may instead become an anchor, increasing the amount of compensation awarded by the judge. A cap on punitive damages has a similar anchoring effect.

In the area of sentencing, Fitzmaurice and Pease’s research on England indicates that the sentence given by the court of appeals may be influenced by the sentencing judgment of a lower court. In the Spanish judicial system, one study found that in 81.75 percent of cases of violation of sexual freedom, the sentence was influenced by the anchoring effect; and another found that 63.6 percent of judgments were influenced by the sentencing recommendations of public prosecutors or the sentencing judgments of the court of the first instance. This anchoring effect was especially significant in cases where the accused was convicted. The existence of anchoring effect is further demonstrated by a set of empirical studies of German judges, which showed that the effects were unaffected by the relevance or randomness of the public prosecutor’s sentencing recommendation or the judge’s experience. All of these studies indicate that the judge’s sentencing decision can easily be influenced by anchoring effect because it involves the estimation of numerical values, and this is the case even if the decision-maker is an experienced judge. Other research has found that the sentencing recommendation of a parole officer has an even greater influence upon the judge than does that of the public procurator. As empirical research on judges and prosecutors has found, the public prosecutor’s sentencing recommendation can even affect the sentencing opinion of the opposing defense attorney, and then influence the judge’s sentence. In other words, the prosecutor’s sentencing recommendation may weaken the sentencing defense presented by defense counsel.

It should be noted that to date, Chinese academics have failed to pay sufficient attention to the influence of cognitive bias in the field of criminal justice, nor have they made an empirical analysis of the relationship between sentencing recommendations and sentencing judgments in situations where anchoring effects apply. Therefore, this article, as an empirical study of the sentencing judgments of base-level Chinese judges, may make a contribution in three ways. Firstly, it demonstrates the actual (rather than laboratory) circumstances and cognitive influences on Chinese judges’ sentencing, thereby compensating for the absence of some countries in comparative law. The main reason the comparative law field lacks empirical analysis of actual circumstances or experimental research based on real judges and prosecutors is the difficulty of obtaining samples—it is not easy to gain access to written judgments or to examinations for judges. Secondly, our paper issues a warning, from the cognitive dimension, about a series of measures in the current institutional reform of sentencing recommendations, with the aim of preventing the possible abuse of power by the procuratorate in the area of sentencing recommendations. Thirdly, the paper aims to more truly reflect the practice of sentencing recommendations in other countries. Previous research has mainly examined recommendations and sentencing in relation to the freedom penalty, but we analyze not only the question of the freedom penalty but also the application of the fine penalty and probationary (suspended) sentences. At the same time, the paper hews more closely to the particular practice of sentencing recommendations in China; unlike the practice and research occurring in other countries, the public procurator’s sentencing recommendations in China suggest a range rather than a specific figure. Below, we use our quantitative analysis of a sample of 520 sentences in K City to focus on the following questions. First, do the procuratorate’s sentencing recommendations (including maximum, minimum and average values) have an anchoring effect on the judge’s freedom penalty sentence? Second, do they generate an anchoring effect on the judge’s decision on the fine penalty? Third, do they have an anchoring effect on the term of probation decided by judges? Fourth, when a “change of court conviction” or a “change of trial circumstances” takes place, does the anchoring effect alter accordingly? And fifth, does the availability of defense counsel weaken the anchoring effect of the procuratorate’s sentencing recommendations?

III. Empirical Analysis: The Anchoring Effect of Sentencing Recommendations on a Judge’s Decisions

 1. Data and variables

This paper is a quantitative empirical study based on data collected by the authors from the files of 411 criminal cases tried in the court of first instance in K City in East China in 2015. K City’s courts were selected for this sample for the following reasons: firstly, K City is a typical mid-sized city in East China, with quite a high level of economic development and rule of law construction and quite well qualified judges. To a certain extent, it is representative of those areas where China’s pilot program of sentencing recommendations has been relatively successful; so to a certain extent, this enables one to deduce the prospects for practice in other place where rule of law conditions are less propitious. Secondly, not only do the criminal cases in the K City courts exhibit a broad range of types of crime, but the distinguishing features of the accused are also similar to those of criminal cases in the more highly developed cities and counties in today’s China; that is, there is a large transient population. Except for individual missing items, all files basically include such legal documents such as sentencing recommendations, statements of charges and judgments. Because some cases are joint crimes, we take the accused, rather than the cases, as sample units. After eliminating sample units which lacked one or two types of legal document, we ended up with a sample of 520 sentences. 

Although the sample comprises 520 sentences, it actually consists of more than 1,500 legal documents (including sentencing recommendations, statements of charges and judgments), so the statistical work was very onerous. Once the sentencing sample was set up, we input sample data for the 520 sentences and extracted such variables as basic information on the accused; sentencing recommendations (including the maximum and minimum freedom penalties and a calculation of the average freedom penalty, as well as the amount of fine penalties and the term of probationary sentences); and change of trial sentencing circumstances, change in the court conviction, and presence or absence of defense counsel (see Table 1).

Table 1 Explanation of Data Processing of Variables in the Documents

2. Empirical methodology

In order to analyze whether the public procurator’s sentencing recommendations would affect the sentence imposed by the judge, we set up the following regression model.

Yi=C+βXj+ε     i=1,2,3; j=1,2,3…6. (1)

Here, the explained variable Yi (i=1,2,3) is the sentence delivered by the judge. Specifically, it employs three basic indicators: the term of the freedom penalty given in the sentence(Y1); the amount of the fine penalty given in the sentence (Y2); and the term of probation given in the sentence (Y3). The explanatory variable Xj (j=1,2,3...6.) covers key areas of the public procurator’s recommended freedom penalty, comprising the minimum recommended sentence (SRMinimum, X1); the maximum recommended sentence (SRMaximum, X2); the average of the maximum and minimum recommended sentences (SRAverage, X3); whether a probationary sentence was recommended (RProbation, X4); and whether a fine penalty was recommended (RFine, X5). In addition, given that a change in trial sentencing circumstances may directly influence the sentence delivered by the judge, we also include changes in trial sentencing circumstances (Trialchange, X6) in the regression analysis. C in Model (1) is a constant term, and ε is the residual regression in the model.

It should be noted that Regression Model (1) is a simple regression model with only one explanatory variable. Its purpose is to conduct a separate investigation of the influence of each explanatory variable on the explained variable (Yi). On this basis, we establish the following regression model in order to further observe the simultaneous influence of various factors on the explained variable (Yi):

Yi=C+αXj+β1X4+β2X5+β3X6+ε    i, j=1,2,3  (2)

In this model, Yi (i=1,2,3) is defined as before; Xj (j=1,2,3) is the minimum, maximum and average value of sentencing recommendations from the public procurator respectively, and X4, X5 and X6 are the other three explanatory variables, as follows: whether a probationary sentence was recommended; whether a fine penalty was recommended; and whether there was any change in sentencing circumstance in the trial. Compared to Regression Model (1), Regression Model (2) simultaneously includes a number of factors that may influence the sentencing judgment. However, it should be noted that in reality, the three variables of the public procurator’s minimum, maximum and average freedom penalty recommendations are often highly correlated. To improve the results of our estimations, we have introduced these variables into the model one by one, with a view to avoiding multicollinearity in the regression. In terms of methodology, because all the data used in our paper are cross-section data, we will use the Ordinary Least Squares (OLS) method to estimate the models.

3. Empirical results

(1) Baseline results

In Table 2, we give the regression results obtained by taking the freedom penalties imposed in the judges’ sentences (Y1) as explained variables. Here, the first seven equations correspond to the regression results of Model (1), while the last three equations correspond to those of Model (2). As the results in Table 2 show, the minimum recommended sentence (SRMinimum) is statistically significant at the 1 percent confidence level, and the estimated value of the regression coefficient is 1.024. This means that, other things being equal, the minimum recommended sentence has a marked positive influence on the term of the freedom penalty imposed by the judge. The latter increases by 1.024 months for every month’s increase in the public procurator’s minimum recommended sentence. The regression coefficient of the maximum recommended sentence (SRMaximum) is 0.795  and is statistically significant, indicating that the term of the freedom penalty pronounced by the judge will increase by 0.795 months for every month’s increase in the maximum recommended sentence. Similarly, the regression coefficient of the average recommended sentence (SRAverage) is 0.902 and is statistically significant, indicating that the term of the freedom penalty imposed by the judge will increase by 0.902 months for every month’s increase in the average recommended sentence. Although the variable of the public procurator’s recommendations for probationary sentences (RProbation) is not significant in the simple regression (Regression 4), after the influence of other variables is controlled, this variable has a significant negative effect upon the term of the freedom penalty imposed by the judge (Regression Models 8 and 9), indicating that in cases where the public procurator recommends a probationary sentence, judges are more inclined to give a lower sentence. This is quite understandable. On one hand, offenders eligible for probationary sentences have certain favorable conditions with regard to penalties, i.e., probation is only applicable to offenders who could be sentenced to criminal detention or to three years or less of fixed-term imprisonment; on the other hand, the application of probation also means that the procuratorate holds that the circumstances of the offender’s crime were relatively minor and there was little threat to the public. Therefore, the public procurator’s recommendation of a probationary sentence signals to the judge that the offender’s penalty needs to be assessed more leniently and that there is little likelihood of a threat to the public. The regression coefficient of recommendation of the fine penalty variable (RFine) is significant in some cases but not in others, and the signs are inconsistent. This indicates that for the present, the data in this paper do not allow for a conclusive judgement on whether the public procurator’s recommendation of a fine penalty influences the term of the freedom penalty imposed by the judge. Finally, the regression of changes to trial sentencing circumstances (Trialchange) is positive and is significant when other variables are controlled. This indicates that if the trial sentencing circumstances are aggravated, the term of the freedom penalty delivered by the judge will show a marked increase.

Table 2 Public Procurators’ Sentencing Recommendations and Judges’ Imposition of Freedom Penalty Sentences

Employing the same structure as Table 2, Table 3 reports the regression result when the fine penalty in the judge’s sentence is taken as the explained variable. First, and completely consistent with theoretical predictions, the regression coefficient of whether the public procurator recommends a fine penalty is positive and highly significant at the 1 percent confidence level (see Regression 5 in Table 2). This indicates that the public procurator’s recommendation on the imposition of a fine penalty as supplementary punishment does have an important influence upon the judge’s final judgement on a fine penalty. Second, we found that the public procurator’s recommendation of the freedom penalty also affects the amount of the fine penalty imposed by the judge. Specifically, as shown in regression results (1)—(3) in Table 3, we found that the public procurator’s minimum, maximum and average sentencing recommendations all have a statistically positive effect upon the amount of the fine penalty imposed by the judge. In particular, for every month’s increase in the public procurator’s minimum sentencing recommendation, the fine penalty imposed by the judge increases by 90.323 RMB; for every month’s increase in the public procurator’s maximum sentencing recommendation, the fine penalty imposed by the judge increases by 68.567 RMB; and for every one unit increase in the public procurator’s average sentencing recommendation, the fine penalty imposed by the judge increases by 78.555 RMB. This finding deserves our attention. There is no necessary relationship between the freedom penalty and the fine penalty, but our findings indicate that even a source of information that is unrelated in terms of contents, i.e., the freedom penalty, still has an anchoring effect upon the level of the fine penalty in the judge’s judgement. To some extent, this verifies and expands Kahneman’s findings in the field of general cognition, those of Chapman and Bornstein in the field of civil compensation, and those of Englich and Mussweiler in the field of sentencing in Germany; that is, totally unrelated reference information still affects decisions, even with such richly experienced professional decision-makers as judges. In addition, the regression results in Table 2 show that the regression coefficients of changes to the trial sentencing circumstances are not significant, which indicates that such changes are not directly related to the judge’s imposition of a fine penalty. Finally, what is interesting is that when the influence of other variables is controlled (Regression Models 7-9), we find that where the public procurator recommends a probationary sentence, the judge imposes a higher fine penalty. This is possibly because the judge may, in imposing a sentence, make an implicit trade-off between the freedom penalty and the fine penalty, resulting in probation being accompanied by a relatively higher fine penalty. In other words, judges are reluctant to impose either harsh or lenient sentences across the board, that is, in both components of the punishment at the same time: when the freedom penalty is lenient, the fine penalty may be rather more severe, and conversely, when the freedom penalty is relatively severe (with no chance of probation), the fine penalty may be slightly less.

Table 3 Public Procurators’ Sentencing Recommendations and Judges’ Imposition of Fine Penalty Sentences

In Table 4, we continue our analysis of the influence of the public procurator’s sentencing recommendations upon the term of probation imposed by the judge. Firstly, consistent with the survey data discussed by Pan Shenming, Zhou Wen and Tan Jianxia and with the experience of some previous local judicial pilots, we find (in Regression Models 4 and 7—9) that whether the public procurator recommends probation (RProbation) has a positive and statistically highly significant effect on the term of the probationary sentence (at the 1 percent significance level). This indicates that the public procurator’s recommendation for probation does have an important influence upon the judge’s decision to impose a probationary sentence. In addition, the results in Table 4 also show that the effect of the public procurator’s recommendation of a fine penalty (RFine) on the term of probation set by the judge is negatively significant. Once again this verifies the analytical result in Table 3, i.e., the judge may be making an implicit trade-off between criminal penalties and the fine penalty, so that probationary sentences and fine penalties are in an inverse relationship. Lastly, when the influence of other variables is controlled, neither the minimum, maximum or average sentencing recommendations (SRMinimum, SRMaximum, or SRAverage) of the public procurator nor changes in trial sentencing circumstances (Trialchange) has a significant influence upon the term of probation set by the judge.

Table 4 Public Procurators’ Sentencing Recommendations and Judges’ Imposition of Probationary Sentences

(2) Further discussion

The analysis above indicates that the public procurator’s sentencing recommendation has a significant effect upon the sentence imposed by the judge, which means that the anchoring effect is supported by the data in this paper. In reality, however, the effect of the public procurator’s sentencing recommendations upon the judge’s decision may be constrained by related factors that intensify or weaken the anchoring effect. In this section, we focus on examining the following three influencing factors: 1) Where a change of trial circumstances occurs, is the anchoring effect of the public procurator’s sentencing recommendation affected? 2) Where there is a change of conviction, does the sentencing recommendation affect the judge’s sentence? And 3) Does the presence of defense counsel influence the anchoring effect of the public procurator’s sentencing recommendations?

The above questions can be analyzed by introducing the interaction term of the public procurator’s sentencing recommendations and related variables into the regression equation. Specifically, we can set up the following regression models:

Y1=C+αXi+βXi*Trialchange+ε     i=1,2,3        (3)

Y1=C+αXi+βXi*Courtchange+ε   i=1,2,3        (4)

Y1=C+αXi+βXi*Counsel+ε           i=1,2,3        (5)

Here, Models (3), (4) and (5) are used respectively to analyze the three questions above. The explained variable Y1 in the model is the term of the freedom penalty assigned by the judge and the explanatory variables Xi (i=1,2,3) use three simultaneous indicators: the public procurator’s minimum, maximum and average sentencing recommendations (SRMinimum, SRMaximum and SRAverage respectively). The newly introduced interaction term indicators include the change of trial sentencing circumstances; court change of conviction; and the   presence or absence of defense counsel (Trialchange, Courtchange and Counsel respectively). C is the constant term and ε is the residual in the model regression. Table 5 gives the results of the analysis based on the three regression models above. 

In accordance with the analytical goals of this section, our focus in the three regression models is on the signs and significance of the interaction terms. Generally speaking, if the sign of the interaction term is significantly positive, it means that occurrence of these circumstances will strengthen the anchoring effect of the public procurator’s sentencing recommendations; conversely, if the sign of the interaction term is significantly negative, it indicates that occurrence of these circumstances will weaken the anchoring effect of the public procurator’s sentencing recommendations. Therefore, from regression results (1) to (3) in Table 5, one can see that the interaction term of the public procurator’s sentencing recommendation and a change of trial circumstances (SRMinimum*Trialchange) is highly significant at the 1 percent confidence level and has a positive sign, which indicates that in cases where trial circumstances change, the judge’s final sentence will have an additional positive marginal increment over that recommended by the public procurator, that is, in such cases the judge takes the initiative to adjust the term of the penalty upwards. At the same time, we also observed that the regression coefficients of the public procurator’s minimum, maximum and average sentencing recommendations, namely items (1) to (3) in Table 5, are all clearly lower than the numerical values of the corresponding coefficients of the related variables in Table 2. This indicates that where trial sentencing circumstances change, the relevant sentencing recommendations from the public procurator have less effect on the judge’s final judgement. These results show that in cases where trial circumstances are aggravated, the anchoring effect of the public procurator’s sentencing recommendations will be somewhat diminished. Similarly, if we look at the influence of court changes to conviction, we can see from the regression results in items (7) to (9) in Table 5 that the interaction term of the public procurator’s sentencing recommendations and court changes to conviction (SRMinimum*Courtchange) is significantly positive at least at the 5 percent confidence level, indicating that where the change makes the conviction more serious, the term imposed by the judge’s final sentence will have an additional positive marginal increment over the term recommended by the public procurator. This demonstrates that in cases where the conviction becomes more serious, the anchoring effect of the public procurator’s sentencing recommendations will likewise become weaker. Furthermore, in comparing the numerical values of the regression coefficients of the interaction terms SRMinimum*Trialchange and SRMimum*Courthange, we find that although both reinforce the anchoring effect of the public procurator’s sentencing recommendations, a change in the conviction does so to a greater extent than a  change in trial circumstances. With regard to the influence of the defense counsel, the coefficients of the interaction terms in Regression Models (4) and (6) in Table 5 are all significant and negative, indicating that the presence of counsel significantly weakens the influence of minimum and average procuratorial sentencing recommendations on the sentences imposed by the judge; in other words, the anchoring effect of procuratorial recommendations upon the judge’s sentence is much weaker when defense counsel is present than when this is not the case. And finally, although a change in trial circumstances, a court change to conviction and the presence of defense counsel all influence the anchoring effect, empirical data suggest that the following merit attention. First, although the anchoring effect of the public procurator’s sentencing recommendations in the three situations are to some extent weakened, they still exist. This indicates that such sentencing recommendations, made prior to the trial, have intrinsic advantages. Second, the corresponding coefficient is negative where defense counsel is present, indicating that the presence of counsel not only weakens the anchoring effect of a sentencing recommendation from the procuratorate, but also encourages the court to impose a lighter sentence than would be the case without counsel. This proves that defense counsel not only is engaged in defending, as opposed to prosecuting, the accused, but also has an independent significance of its own.

Table 5 Public Procurators’ Sentencing Recommendations and Judges’ Imposition of Freedom Penalty Sentences: Further Analysis of the Intensifying or Weakening Mechanisms inAnchoring Effects

(3) A summary of the empirical results

The analysis above leads to the conclusion that whether we examine the freedom penalty, the fine penalty or probationary term imposed by the judge, we can see that in all of them, the sentencing recommendations of the public procurator have a significant effect on the sentence the judge imposes; that is, the anchoring effect is present. Further, we find that changes to court conviction and trial circumstances and the presence of a defense lawyer will correspondingly weaken the anchoring effect of the procurator’s sentencing recommendations. The first two situations are important mainly because trial circumstances and the type of crime for which the defendant was convicted are important substantive factors affecting sentencing, while the presence of a defense lawyer is an important procedural factor affecting sentencing. These findings provide some new evidence and new approaches to a deeper understanding of the mechanism of changes to anchoring effects.

IV. Conclusion  

As Herbert A. Simon’s bounded rationality model points out, people’s decisions or judgments are usually time-constrained; therefore, in complex situations, the decisions or judgments they are able to make tend to entail satisficing rather than optimizing or maximizing their interests. Accordingly, such decisions can readily be influenced by the cognitive heuristics discovered by Daniel Kahneman and Amos Tversky.

The findings of our empirical data are as follows. Firstly, in terms of cognitive mechanisms, sentencing recommendations, although professedly “a mere reference,” are not dispensable; on the contrary, they constitute a kind of obdurate “anchor” or constraint. Whether they are directed at freedom penalties or fine penalties; whether they target the criminal punishment itself or the way it is implemented; whether they give a specific numerical value or a proposal that “may be applied”; whether the sentencing recommendations are relevant to the contents of the judgment or not; and whether the decision-maker is an inexperienced novice or an experienced senior judge, the anchoring effect has an objective existence. In this sense, it would seem that the procuratorate, as prosecuting party, should continue to provide sentencing recommendations, as these help to strengthen the power of the prosecution.

Secondly, our research argues at the same time for the other side of criminal justice, the prosecuted party. Neither the accused, who seeks to preserve his own rights and interests, nor the criminal defense lawyers who aim to protect his rights can afford to think lightly of, still less ignore, the significance of the cognitive dimension of defense sentencing opinions. The defense may have a number of misgivings or concerns, such as never having encountered anything similar before, and may be afraid that holding a trick in reserve during the pretrial proceedings could take the public procurator by surprise and arouse antipathy among judges or public procurators, and so on. However, regardless of these concerns, it remains very necessary to present one’s own “anchor,” which can to a certain extent limit the anchoring effect of the public procurator’s sentencing recommendations. Our research also demonstrates that in Chinese criminal justice, the presence of a defense lawyer does bring about a rather better outcome for the defense.

Thirdly, judges who uphold legal principles and norms face a paradox: they have to remain independent of procuratorial sentencing recommendations while at the same time deciding on criminal punishments which objectively reflect the facts of the crime and embody rational legal norms. The original purpose of establishing the procuratorial sentencing recommendations system was to limit the judge’s discretion over sentencing by means of these recommendations, since otherwise, if judicial discretion was given free rein, judges could impose arbitrary sentences. In reality, however, such recommendations have had a strong impact upon the sentences delivered by the judge. We can predict that this impact will be strengthened as the judges’ caseloads become heavier. A possible solution for this problem could be the formulation and application of a set of relatively detailed guidelines for the standardization of sentencing running parallel to the trial of the sentencing recommendations system. Moreover, the important judicial principle of “letting both sides have a say” is not merely empty talk, nor does its function simply remain at the level of procedural law—that is, of equal protection; rather, it involves deeper underlying cognitive and behavioral scientific principles. Therefore, in order to avoid the one-way influence of procuratorial sentencing recommendations, the judge should listen carefully to sentencing opinions of the defense.

原文刊载于《中国社会科学（英文版）》2018年第39卷第3期

（图片来自网络素材）

 中国法与社会研究院 

分享海内外法社会学的前沿景观

学术成果 | 活动报告 | 论坛通知

投稿、荐稿请联系：

[e-mail]