Wrongful Conviction Rates and Death Penalty Support: Acceptability Hypothesis in Japan

This study tested the hypothesis that presenting people with information about wrongful convictions that are considered an unacceptable cost of administering justice lowers support for the death penalty. We recruited 815 participants using an online crowdsourcing service in Japan, randomly assigned them to eight groups with varying combinations of presence or absence of information emphasizing losses caused by wrongful convictions and asked them to answer questions on the death penalty. Results of the preregistration experiment showed no effect of loss emphasis, but the groups provided with information on wrongful conviction rates were less likely to support the death penalty than those with no information; the higher the wrongful conviction rate, the lower was the support. This study’s findings, which replicated a previous study’s findings with non-U.S. participants, support the acceptability hypothesis and confirm the institutional and cultural universality of the effect of wrongful conviction information on public attitude toward capital punishment.


Background
According to statistics from Amnesty International (2021), in 2021, there were 2,052 death sentences and 579 executions worldwide (excluding data for countries such as China and North Korea, whose information is not publicly available).The number of executions has been on the decline, with 144 countries legally or de facto abolishing the death penalty.The 2021 figures are the second lowest since 2010, when Amnesty began keeping records.This international trend toward the abolition of the death penalty can be attributed to factors such as political leadership promoting policies to abolish the death penalty, increased attention to human rights issues, and the rejection of claims regarding the deterrent effect of the death penalty (Johnson, 2019).In the context of such international trends, Japan is the only G7 country other than the United States to retain the death penalty as a statutory punishment.The crimes to which the death penalty is applied in Japan include murder, robbery, arson, and other serious crimes resulting in the loss of human life, with three (two) executions in 2019, zero (three) in 2020, and three (three) in 2021, and roughly the same number in other years (Amnesty International, 2021; The numbers in parentheses indicate the number of death sentences in the same year).The application of the death penalty in Japan is based on nine criteria, including the seriousness of the crime, such as the number of victims, the motive for the crime, and the brutality of the crime.The only method of execution is hanging (Article 11 of the Penal Code).In Japan, public opinion toward the death penalty is rooted in moral intuitions about retribution and atonement (Johnson, 2020); therefore, it is highly supported.According to a survey conducted every 4 years by the Japanese government, 80.8% of the general-public respondents (1,572 people) supported the death penalty in 2019 (Cabinet Office & Government of Japan, 2019), higher than the 55% in the United States (Gallup, 2022) and 68% in China (Liu, 2021).Among those who opposed the death penalty (9.0%, 142 respondents), the risk of false convictions was the largest reason (50.7%) cited.The Declaration by the Japan Federation of Bar Associations (Japan Federation of Bar Associations, 2016) also raised opposition citing the risk of wrongful convictions.However, despite the concerns about the risk of wrongful convictions, most Japanese people continue to support the death penalty; therefore, these concerns have not increased public support toward the abolition of the death penalty.
Outside of Japan, however, the risk of wrongful convictions has been shown to be a factor that suppresses support for the death penalty.In the United States, the ''innocence frame'' (Dardis et al., 2008), in which the mass media emphasize judicial errors, has become popular, and public opinion and policy toward wrongful convictions have changed (i.e., the innocence movement; Norris, 2017;Norris et al., 2020).This trend has led to a growing body of empirical research on the impact of information about wrongful convictions on public attitudes toward the death penalty, particularly regarding the relationship between information and support for the death penalty, on which we focus.In an experiment conducted with U.S. college students, participants who were presented with an essay detailing the frequency with which innocent individuals are sentenced to death were more likely to exhibit decreased support for capital punishment than the control-group participants (Lambert & Clarke, 2001;Lambert et al., 2011).In an experiment, Liang et al. (2019) presented a group of participants with essays revealing that the death penalty is being abolished worldwide or on the deterrent effect of capital punishment.They found that these essays had no effect or, contrary to their predictions, increased support for the death penalty.However, essays on wrongful convictions significantly decreased the support for the death penalty.In another experiment, Dardis et al. (2008) presented newspaper articles reporting on wrongful convictions (Dardis et al., 2008); the group that read the article ''Wrongful convictions as a flaw in the system'' was more likely to mention the content of this article than the group that read a newspaper article that critiqued the pros and cons of capital punishment from a moral perspective.
In sum, presenting people with information about wrongful convictions has been shown to change their attitudes toward the death penalty, although there are exceptions (e.g., Bohm et al., 1991).Notably, information presented in the form of an essay or newspaper article contains ''one-sided information bias'' as a persuasive message and does not comprise objectively evaluable ''facts'' (Wu, 2021).In Wu's (2021) experiment, different participant groups were deliberately presented with different wrongful conviction rates for murder cases (i.e., experimentally manipulated).Participants were assigned to one of four groups: three with wrongful conviction rates of either 0.027, 1, or 4.1%, and one with no information, and asked whether they agreed or disagreed with the death penalty conviction.The support for the death penalty in the 1% group was similar to the group with no information.However, participants in the 0.027 and 4.1% groups were less likely to support the death penalty than those in the no-information group, with the 4.1% group showing the lowest percentage of supporters.While discussing these results, Wu (2021) argued that information about wrongful convictions decreases support for the death penalty when it reaches a level where it is no longer accepted as a cost of administering justice.Although Wu's (2021) experiment is groundbreaking, the current challenge is that it needs to be refined in several ways to generate more evidence in support of the acceptability hypothesis.The first and most unique refinement was conceived by combining two theoretical frameworks: framing effects (Chong & Druckman, 2007) and retributive justification of punishment (Carlsmith & Darley, 2008).The first change to Wu's original experiment involves manipulating the information presented on the loss of the real criminals that escape due to wrongful convictions, in order to alter acceptability levels across groups, as well as replicating Wu's probability manipulation: varying the probability of wrongful conviction rates presented to different groups to examine its effect on acceptance rates.The relationship between probability and acceptability is not perfectly linear, and probabilities are interpreted differently depending on people's beliefs and knowledge (de Finetti, 1974;Sjo¨ber, 1979;Smith et al., 1991).In particular, lower probabilities are subjectively weighted by probability-weighting functions (Tversky & Kahneman, 1992); thus, they can greatly impact decision-making.In an experiment examining the intention to vaccinate 1,000 Americans (Kaplan & Milstein, 2021), the participant groups were presented with varying information on the probability of side effects such as fever and arm pain, and yet no difference was found in the strength of their intention to vaccinate.Even in experiments in which the probability of experiencing serious side effects was manipulated at four levels, from 1 per 100 to 1 per 100,000, the differences in probability did not predict the acceptability of the drug (Wilhelms et al., 2018).In Wu's (2021) experiment, contrary to prediction, the 0.027% conviction rate was more effective in reducing support for the death penalty than the 1% conviction rate.It is possible that the participants misinterpreted 0.027% as 2.7%, as Wu discussed; however, the probabilities may have been manipulated unsuccessfully.Therefore, in this study, we manipulated the loss caused by wrongful convictions.The cognitivepsychology literature has robustly confirmed that humans tend to make loss-averse judgments and are less likely to accept alternatives that emphasize losses (i.e., the framing effect; Chong & Druckman, 2007;Gong et al., 2013;K€ uhberger, 1998;Levin et al., 1985;Tversky & Kahneman, 1981).The public's attitude regarding the punishment of offenders is based on retributive justification, which seeks to offset crime through punishment (Brubacher, 2018;Carlsmith, 2006;Carlsmith & Darley, 2008;Finckenauer, 1988;Keller et al., 2010;Rucker et al., 2004;Watamura et al., 2011).The public must therefore view the ''other side of the coin'' of a wrongful conviction, that is, the real culprit going unpunished, as a loss.
In this study, in addition to varying the four wrongful conviction rates presented to different participant groups, we also manipulated the presence or absence of the sentence ''the real criminal is getting away with it.''We predicted that participants' support for the death penalty would be affected by both the wrongful conviction rate and the emphasis on loss.Although we did not expect a specific interaction between these two factors, we aimed to explore the possibility of an interaction.
The second area of improvement is the addition of a manipulation check for acceptability to confirm that the artificial intervention of the wrongful conviction rate was effective.Wu's (2021) experiment did not measure how acceptable each wrongful conviction rate (0.027, 1, and 4.1%) was to the participants; therefore, it remained unclear whether the acceptability level would be lower if the wrongful conviction rate was higher.High wrongful conviction rates may result in reduced support for capital punishment because of decreased confidence in the judiciary (Norris & Mullinix, 2020).In this study, participants were asked to rate the item ''this level of wrongful conviction rate is compelling for meting out justice'' on a 6-point Likert-type scale.
Third, it is essential to have some evidence of institutional and cultural universality.Although research on the effects of wrongful conviction, rate information has been conducted mainly in the United States, the same results may not be replicable in countries and regions with different justice systems (Stack, 2004).The presence of cultural differences have been confirmed in moral judgments (Barrett & Saxe, 2021;Haidt et al., 1993;Vaisey, 2009) and could influence the acceptability of wrongful convictions as the cost of justice.
This study aimed to refine Wu's (2021) experiment to examine the acceptability hypothesis regarding the effect of information about wrongful convictions in the Japanese context.Participants were assigned to four groups, one group for each murder-conviction rate (0.027, 1, or 4.1%) and one group that was given no conviction-rate information.Up to this point, we followed the same procedure as Wu (2021) did.However, in our study, another set of four groups were prepared.Each additional group was randomly presented with wrongful conviction rates (manipulated at four levels as before) along with written information emphasizing the loss caused by false accusations.In addition, we conducted a manipulation check for measuring the acceptability of the wrongful conviction rate.This study proposes the following hypotheses.
H1. Presenting information about wrongful conviction rates decreases the percentage of participants who support the death penalty.H2.The higher the rate of wrongful convictions, the lower the percentage of participants who support the death penalty.H3.Emphasizing the loss caused by wrongful convictions decreases the percentage of participants who support the death penalty.
We did not propose a specific hypothesis on the interaction between wrongful conviction rate and loss because of the lack of a theoretical basis particular to these variables.

Methods
The study involved an online experiment with a betweenparticipant design and was conducted after completing the unscreened-preregistration process on the Open Science Framework's (OSF) preregistration service.An ethical review was conducted by the first author's university ethics review committee.

Participants
Participants were Japanese men and women aged 18 years or older with the right to vote.In Wu's (2021) experiment, the average number of participants per group was approximately 120, and we sought to analyze the same number of participants.Estimating that 40% of the data would be missing because of the participants' failure to provide informed consent or the exclusion of invalid data, having 200 participants per group (1,600 persons in eight groups) was deemed appropriate.The sample was collected from a panel of an online crowdsourcing company.It has a registered panel of over 4 million Japanese people throughout the country and is generally representative of the Japanese population in terms of various attributes such as gender, age, education, income, occupation, and region of residence.In this study, we randomly sampled from that panel using Excel's random function.Thus, the participants were a microcosm of the Japanese population, except they all had access to the Internet.A total of 1,339 people provided informed consent and participated in the experiment in exchange for a reward of shopping points worth USD 0.3.In online research, some people participate solely for the gratuity; therefore, attention checks are necessary because such respondents tend to answer questions without reading the questionnaire (Alvarez et al., 2019).Thus, this study included two questionnaire items asking respondents to make a specific choice, such as ''Please choose AGREE,'' and the 272 respondents who chose either one of the incorrect answers were excluded.Additionally, in accordance with Wu's (2021) process, the 252 participants who answered ''don't know'' regarding their attitude toward the death penalty were considered as missing.In total, 815 participants (female = 331, male = 484, M age = 50.8years, SD = 14.9) were finally included in the analysis.Raw data for the entire sample before exclusion are available at the OSF.

Design
Participants were randomly assigned to eight groups using the Excel function; the sample size for each group ranged from 84 to 127.The wrongful conviction rate was manipulated at four levels: no information, 0.027, 1, and 4.1%.Both probability (%) and frequency were presented for the wrongful conviction rate.The number of convicted people per 100,000 was also presented because the percentage, if provided on its own, could be misinterpreted (e.g., 0.027% misinterpreted as 2.7%) (Wu, 2021).Loss due to wrongful convictions was manipulated at two levels, with and without emphases, and combined with the wrongful conviction-rate information to form a total of eight groups.For example, participants in the group emphasizing a 0.027% loss were told, ''One study estimates that about 0.027% (27 per 100,000) of those convicted of murder are actually innocent.Thus, this same number of true culprits also escape unpunished.''The group with no information on wrongful convictions was told that ''some true criminals are escaping unpunished due to false accusations.''Information regarding all eight patterns are presented in the Supplemental file.

Procedures
After providing informed consent, participants answered questions presented on a transition screen.After the wrongful conviction rates and loss treatments were applied, participants were asked to respond to the question, ''Do you agree or disagree with the death penalty for those convicted of murder?''using a 4-point scale ranging from 1 = strongly disagree to 4 = strongly agree.Then, six groups that were presented with information on wrongful conviction rates were asked to rate the acceptability-check item, ''Even if the wrongful conviction rate is this high, it is unavoidable for the sake of justice,'' on a scale of 1 = strongly disagree to 6 = strongly agree.The covariates were perceived risk of victimization (self), perceived risk of victimization (family member), fear of crime, perceived crime rate, empathy toward criminals, and empathy toward victims.These items and options were the same as in Wu's (2021) study (see Supplemental file).Demographic variables were also measured.

Analytic Strategy
The analysis was performed using the same strategy outlined in the preregistration stage.First, Pearson's chi-square test of independence and binomial logisticregression analysis were performed for the four groups without emphasizing loss, using the statistical software HAD version 16.0 (Shimizu, 2016).Attitudes toward the death penalty were coded as 1 for strongly agree and agree and 0 for strongly disagree and disagree, and this coded score was used as the dependent variable.Next, based on the regression model, the predicted probability (predicted margin) of support for the death penalty for each group was calculated using SPSS.The analysis strategy up to this point was precisely the same as in Wu's (2021) study.In addition, we used data from all eight groups, including those with loss emphasis, to determine the variance in attitudes toward the death penalty as the dependent variable, and information on wrongful conviction rate and loss emphasis as independent variables.

Results
Descriptive statistics for the data analyzed are presented in Table 1.In this study involving a Japanese sample, 71.4% (582/815) of the total respondents supported the death penalty, higher than the 53.6% of the U.S. sample in Wu's (2021) study.Among the four groups with no loss emphasis, the rate of support for the death penalty was the highest in the no-information group at 89.0% (113/127), the 0.027% group at 68.6% (70/102), the 1.0% group at 75.3% (70/93), and the lowest in the 4.1% group at 57.1% (48/84).A Pearson's chi-square test of independence revealed a significant difference in support for the death penalty among the four groups (x 2 = 28.915,p \ .001),and the residual analysis revealed significantly higher support for the death penalty and lower disapproval rate in the no-information group (ps ł .001).In contrast, the 4.1% group had a lower rate of support and higher rate of disapproval (ps ł .001).Binomial logistic-regression analysis was conducted with the no-information group as the reference group (Table 2).The results showed that compared with the no-information group, the three groups that were presented with the wrongful conviction rate had significantly lower rates of support for the death penalty (ps \ .01).While Wu (2021) found no difference between the no-information and the 1% groups, our experiment showed a significant difference between the groups; thus, Hypothesis 1 was supported.Notably, there were significant differences in the support for the death penalty even among the four loss-emphasis groups (x 2 = 18.051, p \ .001),with the no-information group showing higher support than the other three groups (ps ł .001):85.4% (88/103) for the no-information group, 62.7% (64/102) for the 0.027% group, 64.4% (65/101) for the 1.0% group, 62.1% (64/103) in the 4.1% group.
The predicted probability of support for the death penalty in each group was then calculated based on the logistic-regression analysis (Figure 1).Among the four groups without the loss emphasis, the highest support for the death penalty was found in the no-information group (85.4%), followed by the 0.027% (78.3%), 1% (69.0%), and 4.1% (57.9%) groups.The predicted probabilities in Wu's (2021) study produced an inconsistency wherein the 1% group was more likely to support the death penalty than the 0.027% group.However, in our study, support for the death penalty consistently decreased as the wrongful conviction rate increased.Thus, Hypothesis 2 was supported.
A two-factor analysis of variance was conducted using data from all eight groups, with information on wrongful conviction rates and loss emphasis as independent variables and attitudes toward the death penalty as dependent variables.The results showed that the main effect of loss (F[1,807] = .85,p = .36,partial hp 2 = .001)was non-significant; therefore, Hypothesis 3 was not supported.The interaction was also not significant (F[3,807] = .64,p = .59,partial hp 2 = .002).The main effect of the wrongful conviction rate was significant (F[3,807] = .151,p \ .001,partial hp 2 = .053),and multiple comparisons using the Holm method revealed that the 0.027, 1, and 4.1% groups had significantly lower support for the death penalty than the no-information group (ps \ .001,ds = .433-.629).Acceptability scores were examined, and the mean acceptance ratings for the presented wrongful conviction rates were 2.80 (SD = 1.47), 2.98 (SD = 1.40), and 2.69 (SD = 1.39) for the 0.027, 1, and 4.1% groups, respectively.Analysis of variance results showed no significant difference between the means (F[2,582] = 2.026, p = .133,partial hp 2 = .007).The relationship between higher wrongful conviction rates and lower acceptability was not confirmed.

Discussion
Using a refined version of Wu's (2021) experiment, this study tested the acceptability hypothesis regarding the effect of information about wrongful convictions.The experiment and analysis were conducted in accordance with the preregistered procedures.The results of the chisquare test and the analysis of variance confirmed Hypothesis 1, that is, the presence of information on the wrongful conviction rate decreases support for the death penalty.Although the manipulation check did not lead  to a change in people's acceptability ratings, the predictive probability calculated based on the logistic regression analysis showed that the rate of support for the death penalty consistently decreased as the wrongful conviction rate increased, confirming Hypothesis 2. Although Wu's (2021) predictive probability included an inconsistency in which the 0.027% wrongful conviction rate had a stronger effect on reducing death penalty support than the 1% rate did, the relationship between wrongful conviction rates and death penalty support shown in our study was perfectly consistent across the four levels of wrongful conviction rates.The findings indicate that the relationship between higher wrongful conviction rates and lower support for the death penalty among the participants in Japan mirrors the relationship found in participants in the U.S., suggesting that the effect of the wrongful conviction-rate information may be institutionally and culturally universal.Hypothesis 3, which states that an emphasis on loss decreases the support for the death penalty, was not confirmed.We found no difference in support for the death penalty between the groups that were presented with the information that the number of true culprits who had escaped was equal to the number of wrongful convictions and those that were not presented with such information.No interaction with wrongful conviction rates was found.

Theoretical Implications
By combining two theoretical frameworks, namely, the framing effect (e.g., Gong et al., 2013;K€ uhberger, 1998;Levin et al., 1985;Tversky & Kahneman, 1981) and retributive justification (Brubacher, 2018;Carlsmith, 2006;Keller et al., 2010;Rucker et al., 2004;Watamura et al., 2011), this study examined the effects of experimentally manipulating the loss caused by false accusations on attitudes toward the death penalty.We found no difference in support for the death penalty between groups with or without information that emphasized the unique loss from retribution justification, given that the number of true criminals evading justice is the same as the number of people falsely convicted.However, we replicated Wu's (2021) result that wrongful conviction-rate information decreases support for the death penalty.More importantly, the consistency found in the relationship between higher wrongful conviction rates and progressively lower support for the death penalty advances the knowledge gained from Wu's (2021) study, the results of which contained some inconsistencies.For example, our findings expanded on Wu's (2021) conclusions by suggesting that the acceptability of the cost of justice could influence attitudes toward the death penalty.Wu (2021) also assumed a tipping point in the wrongful conviction rates somewhere between 1 and 4%, wherein the support for the death penalty would decline sharply; however, the present study suggests that this change is rather gradual.Further, by replicating the results in a sample of non-Americans, we confirmed that the acceptability hypothesis has some universality that transcends judicial systems and cultural differences.Evidently, social and cultural differences may play moderating factors in the impact of wrongful conviction rate information on public support for the death penalty.In Japan, access to information about the death penalty is extremely restricted (i.e., secrecy; Holmes, 2020).In contrast, in the U.S., where the ''innocence frame'' (Dardis et al., 2008) has become widespread, statistics on death-row inmates and executions are routinely published, and in China, media coverage of wrongful conviction cases is increasing (Jiang et al., 2010).These differences in daily media coverage and information disclosure could strengthen or weaken the impact of the wrongful conviction rates.The United States is an individualistic and loosely knit society, whereas Japan is a collectivistic and tightly knit society that is characterized by harmony with the inner group and adherence to social norms (Gelfand, 2012).These characteristics could have influenced the tendency toward a high tolerance of failure to treat offenders who violate norms fairly (i.e., wrongful convictions).This could explain why the effect of presenting information on the wrongful conviction rate on death penalty support may be somewhat mitigated in Japan.Nevertheless, as the true wrongful conviction rate of death row inmates has not been publicly available in Japan, if such information were to be presented, it would raise more concerns about the risk of wrongful convictions than in the past, which might affect public support for the death penalty.

Practical Implications
One of the practical implications of this study is that, regardless of the magnitude of the probability, simply appealing to the fact of a wrongful conviction affects the public's attitude toward the death penalty.Based on the predicted probability results, even the lowest level of wrongful conviction rate (at 0.027%) can be expected to reduce support for the death penalty by 6 to 7%.The results also suggest that the larger the probability, the stronger the effect.In addition, this study has some interesting implications regarding the background behind this effect.The opinion of Supreme Court Justice Thurgood Marshall on the death penalty (Furman v. Georgia, 1972) is summarized in the following three hypotheses: (1) public support for capital punishment is due to the lack of knowledge about capital punishment, (2) exposure to information about capital punishment reduces support for capital punishment, and (3) information about capital punishment is less effective for those who support capital punishment on retribution grounds than for those who support it on deterrence grounds.If public ignorance is related to their attitudes toward the death penalty, as Marshall's first hypothesis asserts, then the current high rates of support in both the United States and Japan can be explained by the possibility that they are poorly informed about wrongful convictions.The various essays used in the experiments of previous studies (e.g., Lambert & Clarke, 2001;Lambert et al., 2011) might have contained one-sided information bias as a persuasive message (Wu, 2021); therefore, it is not surprising that such information reduces support for capital punishment.Conversely, the result that objectively assessable wrongful conviction rates reduce support does not warrant such a ''natural'' criticism, and thus may be considered a stronger demonstration of the validity of the Marshall Hypothesis.This revelation may have important implications for international discussions in the future.While the judicial systems and cultures underlying capital punishment vary across countries and regions, people's understanding of the objective facts of the death penalty could be the common starting point for discussion.Another practical implication of this study's findings can be found in the context of jury selection in criminal trials (Haney et al., 2022).Even though some candidates may be too willing to consider the death penalty, exposure to information on wrongful conviction rate may encourage them to make a more prudent decision.

Limitations and Directions for Future Research
The study has some limitations.First, manipulationchecked acceptability was low and almost constant, regardless of the wrongful conviction rate.High wrongful conviction rates decrease people's trust in the judiciary (Norris & Mullinix, 2020), and the relationship between higher wrongful conviction rates and lower support for the death penalty may be influenced by lower trust in the judiciary.Therefore, both acceptability and trust in the judiciary need to be examined while manipulating wrongful conviction rates.Second, participants must be categorized based on their reasons for supporting the death penalty.Attitudes toward capital punishment are an expression of personal values (Vollum et al., 2004(Vollum et al., , 2009)), and some people are unlikely to change their attitudes toward the death penalty.Marshall's third hypothesis predicts that information has a weak effect on those who support capital punishment on retribution grounds.Although some experiments have tested this hypothesis using wrongful conviction rates, their effect may be weak for those who value retributive justice.
Third, this study used the same wrongful conviction rates as those used in Wu's (2021) experiment in the American context.However, as a conceptual follow-up to a previous study (Fabrigar & Wegener, 2016;Schmidt, 2016), the actual wrongful conviction rates in Japan should be used.Fourth, the duration for which information on wrongful conviction rate reduces the support for the death penalty is not clear; in both Wu's (2021) study and ours, the procedure involved a continuous sequence of steps from presenting the information on wrongful conviction rates to collecting the response.It would be worthwhile to investigate whether the effect persists even after a certain period.

Conclusions
This study examined the acceptability hypothesis, which assumes that presenting information about wrongful convictions at a level that is no longer considered an acceptable cost of administering justice, decreases the support for the death penalty.The results showed that the emphasis on the loss due to wrongful convictions did not have effect on the participant's degree of support; however, we confirmed that the groups provided with the wrongful conviction rate were less likely to support the death penalty than the groups that were not provided with such information.The acceptability hypothesis was also supported; the study showed that the higher the wrongful conviction rate, the lower the support for the death penalty.Replicating previous studies with non-American samples confirmed the institutional and cultural universality of the effect of providing information on the wrongful conviction rate.Future research should examine the effect of people's trust in the judiciary, their reasons for supporting the death penalty, actual wrongful conviction rates, and the persistence of the effect of information regarding the wrongful conviction rate over time.

Note.Figure 1 .
Figure 1.Predicted probability of supporting the death penalty for various wrongful conviction rates.

Table 1 .
Participant Characteristics in the Study on Death Penalty Support in Japan (N = 815).
a Chi-square test with coded variables.

Table 2 .
Logistic-Regression Analysis of Wrongful Conviction Rates and the Support for the Death Penalty.