Introduction

Just over a decade ago, David Johnson and Frank Zimring pointed to the fact that about 95% of the people in Asia live in countries that retain the death penalty as a sentence of the court, and 90% of the globe’s executions take place there; they further opine: ‘Asia should be important to students of capital punishment…for the same reason that Hawaii is of interest to volcanologists: because that is where the action is.’ (Johnson and Zimring 2006: 91). In the decade since, there is clear evidence that scholars have taken this exhortation to heart, with a growing number of studies looking to countries with distinct social, cultural, and political institutions. On China, where criminal law currently specifies 46 death-eligible offences, there is a small but growing list of empirical studies, published in English, on Chinese attitudes towards capital punishment (Jiang and Wang 2008; Oberwittler and Qi 2009; Wu et al. 2011; Jiang et al. 2018). Similarly on Japan, there are several studies that have examined public opinion and its determinants (Jiang et al. 2010; Sato 2013; Sato 2018) and indeed empirical research from the Republic of South Korea (Choi et al. 2017). Yet, on Taiwan, a retentionist country, there has been little published empirical research on attitudes to the death penalty in the last two decades, and what little Taiwanese research exists has utilised time limited datasets, restricted telephone-based interviewing, and only available in Chinese (see Chiu 2006; Hsien 2008; Chou 2012; Yang 2016). Although both Taiwan and USA are retentionist countries, they clearly differ significantly in most variables of interest to the comparativist, including economic development and political and cultural background (Greenberg and West 2008). Hewitt et al. (2004) using data from small and uneven college samples (N less than 250) compared opinion in Taipei and Denver. Their research is the only such direct comparison yet published between USA and Taiwan, but the limitations of the study make interpretation of similarity and difference difficult.

Using survey data collected from face-to-face interviews in a national stratified random sample across Taiwan, our study aims to (1) examine public attitudes towards death penalty abolition and (2) specifically analyze the role of ‘value conflict’ in public consideration of both abolition and in the context of other alternatives. A study of this kind has both theoretical and public policy significance.

On policy, findings from our research importantly add to the very limited empirical evidence on Taiwanese attitudes toward the death penalty and consequently advance comparative knowledge on ‘making sense of difference’ (Nelken 2010). Public opinion may influence public policy; in China, public opinion is often cited as a prime justification for retention (see Jiang 2015; but also see Bakken 2013; Wang 2015). There is certainly evidence on the significance of public opinion for policy-making in Taiwan itself (Bae 2011; Jen 2015). Yet, no doubt extended quantitative and qualitative comparative work needs to be undertaken to understand how unpopular norms become adopted elsewhere (Sarat and Boulanger 2005; Greenberg and West 2008). Strong popular opinion in support of the death penalty (‘our people want this’) around East Asia has so far produced differing outcomes in terms of policy reforms that embrace international human rights norms—with Taiwan more actively embracing such reform on the death penalty than, for example, Japan ( Bae 2011; Sato 2018). However, on global policy shifts, we agree with Greenberg and West’s well evidenced cross-sectional study of 193 countries that ‘the circumstances that have favoured elimination may not ever develop in some countries, and where they do, they may not produce the same outcomes as they have in Western Europe.’ (Greenberg and West 2008: 335). The results of our particular study will be of interest to those concerned with the death penalty ‘debate’ generally, and no doubt may form ammunition as part of the ‘battle-of-arms’ among retentionists and abolitionists both in Taiwan and elsewhere.

Theorising value conflict and ambivalence behind public attitudes on such a vital issue in a non-Western context is also important, and we now turn in our next section to a consideration of the contribution of value conflict, ambivalence and moral psychology to a more nuanced understanding of attitudes at the individual and collective levels.

Value Conflict, Moral Judgement and Ambivalence

Legal philosophers have long pointed to the notion of ‘value’ as a moral principle directing that particular actions or attitudes be promoted or avoided, and that values ‘have judgemental force’ (Dworkin 2011: 118). Yet, in any moral community, at any one time, there will be a plurality of conceptions of any particular value; this does not mean that that members of the moral community share a single conception, but arguably, any individual’s conception will have limits to its flexibility, and will have a tipping point, where the person considers that something that she considers important in that value is lost. In addition, as Stuart Hampshire notes ‘our everyday and raw experience is of conflict between contrary moral requirements at every stage of anyone’s life.’ (Hampshire 1983: 151). At the societal level, politicians are most times prepared to justify policies, such as capital punishment, in terms of prioritizing the claims of one value over those of another: security and safety over freedom. And indeed, as Isaiah Berlin argues, a world of value pluralism, in which values are pursued to their maximum level, does not allow combination without loss (Jahanbegloo, 2007). Yet, Berlin further concludes that this does not mean in every case that we must choose only one value above the other, or that in the act of choosing we negate the other. We can be exclusive in a weak sense, because one value may have to be diminished in order to increase the other—on a case by case basis. While debate rages among philosophers about how best to best understand value ‘conflict’ and the extent to which conflict can be resolved by reinterpretation of a value (see Winter 2016), for our purposes, we can note that there is agreement that though the content of moral claims cannot be independently determined by reference to external ‘brute facts’, our deployment of them carries judgemental weight broadly determined by our pre-existing moral intuitions.

Recent advances in moral psychology help theorize such matters, and in particular, the simple observation by social psychologists that various people moralize in different ways has led to the development of a number of theories and measures which could capture the observed differences in moral judgments and moral reasoning (Sinnott-Armstrong, 2008). Currently, debate centres around moral foundations theory (Haidt 2013; Graham et al. 2013). Moral foundations theory was originally designed for cross-cultural and cross-national work, and it turns out that moral communities in the USA were often like separate nations in some ways. Their moral ‘matrices’ built on differing moral foundations. According to Haidt, when confronted with an opportunity for moral judgement, we arrive instantly at most of our judgements by intuition and only afterwards employ moral reasoning in the form of ex-post justifications (Haidt, 2012). The presumed dominance of intuition over reasoning in moral judgement has become the subject of fierce criticism (Greene, 2013). Yet, moral foundations are arguably not only a valuable analytical tool for understanding the underlying mechanisms of the formation of attitudes but, in capturing the plurality of moral judgements, can also become the potential basis for an effective shifting of these attitudes. There are several studies in different policy areas which provide convincing indications that using moral foundations, knowledge to (re)frame issues can reduce conflict and produce attitudinal shift (Feinberg and Willer 2012; Fernbach et al. 2013; Day et al. 2014).

Ambivalence can be seen to differ from simple rational expressed preference in at least two ways. First, preferences are normally expressed in either positive or negative terms; the nature of ambivalence is to hold ‘opposing affective orientations’ (Smelser 1998: 4). Second, preferences are normally regarded as relatively stable, whereas ambivalence may be unstable, expressing itself in different and sometimes contradictory ways. ‘Public opinion’ surveys serve often to delegitimize ambivalence via the process often of depicting the world as though it is divided into people who are for or against something, ‘a clear distortion of the social-psychological reality of public feelings’ (Smelser 1998: 11).

Sato’s (2013, 2018) research on public support for the death penalty in Japan and the results of our present study point to ambivalence at the individual level, yet it is important to recognise the role of ambivalence in all its manifestations (Smelser 1998). In the USA, where research on public opinion and public discourse on the death penalty have been subject to extensive study, the extent of ambivalence is manifest across different parts of the criminal justice system and in wider public discourse (Zimring 2003; Keys and Smith 2010; Turow 2004) Footnote 1. Frank Zimring identifies the contradictions in the American political experience, identifying ambivalence in the death penalty debate in issues such as due process concerns, moratoriums, scrutiny of executions and vigilante values. He concludes that American ambivalence about executions is a product of contradictory impulses about limits on governmental power (Zimring 2003: 12).

Further evidence of the application of ambivalence at the level of public discourse has been found in Kita and Johnson’s (2014) content analytic study of newspaper reporting on death penalty issues in Japan. Two main signs of ambivalence in reporting were identified: first, ambiguity at the aggregate level in how issues were framed, and second, the use by key criminal justice actors of the ‘ineluctability’ of their actions, seeming to deny their own agency—especially prosecutors and judges in pursuing a death sentence. Overall, Kita and Johnson’s results add further empirical evidence, outside of the USA, in helping to understand the role of ambivalence surrounding the death penalty. While there are no direct studies of ambivalence at the level of public discourse in Taiwan, one can only speculate on signs of ambivalence in recent law reform in relation to the death penaltyFootnote 2 (see also Liao 2010; Bae 2011: 47-48).

Key Dimensions of Value Conflict Related to the Death Penalty Debate

In Taiwan, some 50 crimes are death penalty eligible, only 20 of which involve offences relating to loss of life. Table 1 provides a breakdown of Taiwan’s death sentence and execution data between 1992 and 2018 and since 2010, there have been in excess of 30 executions (Bae 2011; Jen 2015). Until the late 1990s, Taiwan conducted more executions per capita than the USA, but following the move from one-party rule in 2001, executions rapidly declined (Johnson and Zimring 2009). Indeed, there was a tacit Ministry of Justice moratorium in the mid 2000s (Liao 2010). In the face of a public regard for retention of capital punishment, governmental policy and legal reform in Taiwan on the death penalty has pursued abolition of mandatory death sentences for certain crimes, more intense judicial review and in application of sentences by the Minister of Justice—in other words, execution. While China is a very different political context, we agree with Michelle Miao’s recent assessment where she notes ‘what matters is not the status of public opinion but the state of the public mood’ (Miao 2013: 244). Public opinion polling and ‘mood’ matters politically in Taiwan—indeed, both local and international NGOs explicitly recognise this fact both in their interpretation of how and when government decides on executions, and indeed in their own attempts to shift opinion via specific miscarriage cases (TAEPD 2015; Amnesty International 2015).

Table 1 Death sentences and executions in Taiwan: 1992–2018
Full size table

The parameters of the abolition/retentionist debate in Taiwan and indeed across the world are now well rehearsed (for Taiwan see Liao 2008, 2010; and elsewhere see Hood and Hoyle 2015). Essentially, there appear to be four recognised dimensions: individual versus collectivist social order, perceptions of insecurity/fear of crime, wrongful conviction/acquittal and the differing philosophical rationales justifying use of the death penalty. These dimensions of scholarly debate on the death penalty are best seen as forms of value judgement (with no absolute right or wrong answer) that scholars and members of the public make.

Collectivists, for example, tend to take the view that the state has the obligation to maintain law and order. Despite the various definitions of collectivism, scholars tend to agree that collectivist society stress ‘we’ consciousness, collective identity, emotional dependence, group solidarity, sharing, duties and obligations, need for stability and group decision (Brewer and Chen, 2007: 134). Therefore, supporters of the death penalty place greater value on protection of law and order than that of individual life and rights when wrongdoers pose serious threats to collective identity, social solidarity and well being. This view is commonly shared by Taiwanese legal scholars (i.e. Chen, 2001; Liao, 2008, 2010).

The fear of crime dimension is constituted through the individual’s direct and indirect experience of victimisation. Taylor et al. (1979) and Lester (1998) found that people strongly support the death penalty when they have greater direct and indirect experience of victimisation. Hou (1997), Wu (1997) and Hsien (2008) all found that in Taiwan, fear of crime is highly correlated with the positive attitudes to the death penalty when other variables are controlled. Scholars acknowledge that the link between fear of crime and punishment, including the death penalty, is complex and has both instrumental and expressive aspects (Tyler and Boeckmann 1997; Hollway and Jefferson 1997; Garland 2010; Lee 2011; Kury and Kuhlmann 2014).

Turning to the ‘wrongful acquittal/conviction’ Footnote 3 dimension, in Taiwan, the Chiang Kuo-ching case in 2011 has made the public more aware of this issue. Chiang, a 21-year-old soldier, was found guilty of a brutal child rape in Taiwan in 1996, was executed in 1997 and only later adjudged to be ‘not guilty’ in 2011. Evidence was adduced at the 2011 appeal that the military police used old style torture methods to elicit his confession. In 2011, another suspect confessed and confirmed Chiang’s innocenceFootnote 4. Relatedly, Huang and Lin, in a recent mock jury trial experimental study in Taiwan, found that prior to participating in mock jury deliberation, subjects were more likely to view wrongful acquittal as more undesirable than wrongful conviction (Huang and Lin, 2014: 370-371). Finally, the varying justifications for use of the punishment are, of course, well recognized: retribution, deterrence, incapacitation, rehabilitation and more recently, restorative ideas. For retributivists, the death penalty may even have an expressive and communicative function and can be justified by the principles of equality and proportionality with the crime (Hsu, 2014, 2015). Deterrence and rehabilitation arguments remain mainstream justifications for punishment (Tonry, 2011: 3).

The ideas surrounding the principle of restorative justice are of course also well understood in Taiwan with ‘remorse, compensation and mediation’ practices underlying much civil and criminal law in Taiwan (Alarid and Wang, 2001; Jou and Hebenton, 2011). The principle of ‘restoration’ is particularly addressed in the sentencing decisions of Taiwan’s Penal Code. However, the formal term ‘restorative justice’, on the other hand, is a rather unfamiliar term for the Taiwanese public and only became a pilot practice for criminal justice in Taiwan in 2009 (Ministry of Justice, 2014).

In the context of the abolitionist/retentionist debate, value judgements of the type reviewed above are arguably considered and weighted by the public, most likely weighing some more than others, and usually inconsistently. Value conflict is a fundamental aspect of human psychology. Social psychologists recognize that the characteristics of attitudes are too complex to be captured by the simple dichotomy between ‘agree’ and ‘disagree’ (Craig, et al. 2005: 6). The research literature demonstrates that attitude strength is generally related to importance (Krosnick and Abelson, 1992), certainty (Gross, Holtz and Miller, 1995), elaboration (Petty and Cacioppo, 1986), intensity (Krosnick and Abelson, 1992) and commitment (Abelson, 1988). Craig et al. argues that often, an individual holds conflicting values which are related but distinct towards an object (2005: 6). Scientists have experimented with value conflict on many controversial issues such as abortion (Alvarez and Brehm, 1995; Craig et al. 2002) and homosexuality (Craig et al. 2005). Our paper illustrates empirically the role of value conflict in public consideration of both death penalty abolition and in the context of other alternatives, taking Taiwan, a developed and democratic country, as an example of a country with apparent strong public support for the death penalty. We use the 2014 national face-to-face large sample survey conducted by Academia Sinica (Taiwan’s premier research institution) and funded by the Taiwan Alliance to End the Death Penalty (TAEDP)Footnote 5.

Methodology

Data and Sample

The first part of the results for this paper is based on a face-to-face interview survey conducted by Academia Sinica across Taiwan between December 2013 and January 2014, with a random sample of 4082 adults (957 were not eligible), aged 20 and older (see Academia Sinica 2015 for the data file); 2039 completed the survey, with a 65% completion rate and 35% rejection rate, and this compares favourably with the normal response range (48–59%) for national ‘face-to-face’ surveys in Taiwan (Academia Sinica 2017: 21). Interviews were conducted with respondents using computer-assisted personal interviewing (CAPI) with an iPAD device. Based on the total sample, the margin of sampling error is at the 95% confidence level. The survey used a stratified three-stage probability proportional to size sampling method. The first stage selected the primary sampling units—county/town/city/area. The secondary sampling units were village and community, and final sampling units were individual. The primary and secondary sampling units were based on the most recent population census by the service population, industry population, aged 15–64 population, aged 65 and older population, higher education population and population density. Fifty-eight primary units and 116 secondary units were selected. Sampling was weighted to match the national demographics of gender, age, education, region and population density. Demographic weighting targets were based on the most recent National Household Registration Database figures for the aged 20 and older population (see Table 2). All reported margins of sampling error include the computed weighting design effects. The survey instrument contained 104 questions. Two methodological limitations should be acknowledged. The first possible limitation is that alongside the general death penalty related questions, the questionnaire had five death penalty-related ‘scenario’ questions, and none of these questions formed part of our analysis. While there is nuanced import in scenario style questions, these are as yet unstandardized, and as such are, in our opinion, poor comparative proxy measures. The other possible limitation is that the questionnaire was designed in two forms: one was to ask about attitudes toward the death penalty at the beginning of the survey (Questionnaire A with a sample size of 1023), and the other was placed at the end of the survey (Questionnaire B with a sample size of 1016). To control the possible survey design effect on the main dependent variables, only respondents from questionnaire B were included in the multivariate analysis. However, two sub-sample sizes, demographic characteristics and death penalty core values of the respondents in forms A and B were not statistically significant.

Table 2 Demographic characteristics of respondents
Full size table

Measurements

Attitudes Toward Death Penalty and its Alternatives

There were four sets of questions regarding attitudes to the death penalty. The first was a direct question ‘do you agree/disagree to abolish the death penalty?’. Overall, 9.67% agreed and 87.68% disagreed with the abolition proposal for questionnaire A and 14.76% agreed and 81.69% disagreed on questionnaire B. The provision of information (including confidence in police, lack of evidence on deterrence, legal system, the legal grounds and facts about sentencing and executions, ‘miscarriage’ cases in Taiwan and international trends) did apparently change public attitudes to some degree. To control for this variable, we chose to analyze questionnaire B (where all the information was fully covered prior to the direct question). In addition to the direct question, the questionnaire had three alternative schemes.

The first alternative was the ‘suspended death penalty’; the question asked was ‘do you agree/disagree to replace the death penalty with suspended death penalty, which is 25 years imprisonment and possible parole afterwards, if the offender shows remorse and with low risk of reconviction?’ (SD) Table 3 indicates that the support for abolition increases from 14.76 to 41.15% for the SD alternative. The second alternative was ‘do you agree/disagree to replace the death penalty with a life sentence without possible parole?’ (LWOP). Results indicate that support for abolition for the LWOP scheme increased up to 47.90%. The third alternative, ‘do you agree/disagree to replace the death penalty with a life sentence without possible parole, with hard labour and with compensation for the victim’s family?’ (LWOP with hard labour and restitution, LWOP_L_R), increased support for abolition to 70.77%. Seemingly, for Taiwanese, support for abolition is contingent on the conditions of the alternative. These headline results echo McCord’s argument that ‘when the public is asked to express an opinion about the desirability of capital punishment in the abstract without being presented with a particular set of facts or possible alternative sentences, the degree of support for capital punishment is about as high as one would expect to see generated on any contentious issue in pluralist democracy.’ (McCord, 1998:7-11) and allegiance to the death penalty dramatically weakens when the public is presented with alternatives such as LWOP or LWOP_L_R. This illustrates the point that alternatives to capital punishment can indeed change public attitudes to abolition.

Table 3 Public attitudes to capital punishment and three alternatives
Full size table

Core Values Related to Death Penalty Attitudes

Based on the existing research literature on death penalty attitudes, we chose 10 questions from the survey to represent respectively individualism, collectivism, feelings of neighbourhood safety, fear of crime, likelihood of wrongful conviction and acquittal, as well as the four classic rationales of punishment (see Butler et al. 2017). Questions were presented in terms of a 4-point scale, strongly agree to strongly disagree. So on individualism, the question highlighted the importance of the dignity of each person’s life. On collectivism, the role of the state as protector of law and order was stressed, as opposed to individual human rights. Feelings of safety, measured by perception of neighbourhood safety and fear in terms of worry about self and family being victims of crime. Likelihood of wrongful conviction was measured in terms of likelihood that an executed person being factually innocent and acquittal in terms of likelihood that a factually guilty person is wrongly acquitted (see our endnote no. 3).

Turning to the four justifications for punishing offenders, the question of rehabilitation was measured in terms of the likelihood that murderers can be rehabilitated. While restorative justice tapped into remorse and due forgiveness by the victim’s family. Deterrence was assessed by agreement with the notion that the death penalty is the most effective punishment to prevent reoffending. Finally, retribution measured agreement with the classic ideal of ‘a life for a life’.

Table 4 presents percentage agreement on core values in the abolition and retention groups. It indicates that among all core values for questionnaire B, feelings of neighbourhood security has the highest mean, and the lower ones would be individualism and concerns for wrongful conviction. There is also a significant mean difference between wrongful conviction (m = 1.93) and wrongful acquittal (m = 2.35). Thus, it is fair to describe Taiwan as a young democracy, with low individualism, lower concerns for wrongful conviction than wrongful acquittal but a safe society. As for support for four rationales for punishment, rehabilitation is the lowest, where restorative justice (forgiveness) and retribution are the highest.

Table 4 Core values among abolition and retentionist groups for questionnaire B
Full size table

Table 4 also shows the differences for the abolitionist and retentionist groups in terms of their four sets of core values. Among abolitionist supporters, strong value conflict appeared on ‘Death penalty is violation of human dignity (70%) and State should prioritize law and order (53%)’, ‘high possibility of wrongful conviction (85%) and acquittal (90%), ‘good sense of neighbourhood safety (97%) and yet reasonable amount of fear of violent crime (53%)’, but more consensus on rationales of capital punishment with ‘rehabilitation (77%), restorative justice (90%), deterrence (48%) and retribution (42%)’. Greater value conflict appeared on values such as ‘high possibility of wrongful conviction (79%) and acquittal (92%)’, ‘good sense of neighbourhood safety (88%) and yet reasonable amount of fear of violent crime (57%)’ and ‘belief in both restorative justice (72%), deterrence (73%) and retribution (72%) and more consensus on the balance of the State role and individual human rights’. Table 4 reveals that both abolitionist and retentionist groups have ambivalent attitudes on the fundamental dimensions of debate on the death penalty.

Value Conflicts

Our measurement of value conflict derives from comparing pairs of contrasting core value sets and calculating their relative effects according to the formula below:

$$ \mathrm{Value}\ \mathrm{conflict}\ {\mathrm{C}}_{\mathrm{i}}=\left({\mathrm{V}}_{i1}+{\mathrm{V}}_{i2}\right)/2-\mid {\mathrm{V}}_{i1}-{\mathrm{V}}_{i2}\mid \dots \dots . f1 $$

where i = value set 1 to 7; 1 and 2 are subsets in each value set

Ci scores between − 0.05 (strongly disagree coded 1 and strongly agree coded 4 for each value in the set) to 4 (strongly agree on both conflict values in each set). In other words, the higher C scores, the more value conflict or ambivalence the respondent has. Each value set can be described as follows:

  • Value conflict set 1: individualism vs collectivism

  • Value conflict set 2: feelings of safety vs fear of crime

  • Value conflict set 3: likelihood of wrongful conviction vs likelihood of wrongful acquittal

Value conflict set 4 contains four rationales for capital punishment and their paired value conflict contrasts. The four subsets are rehabilitation vs deterrence, restorative justice vs deterrence, rehabilitation vs retribution, and restorative justice vs retribution.

Results

Who Would Un/Change their Attitudes to Abolition? The ‘Ambivalent’ and ‘Hard-Core’ Groups

To understand the role of value conflict in public consideration of abolition within the context of the other alternatives, we first constructed four analytic groups (See Table 5):

Table 5 Percentage of attitudes un/change to abolition when alternative is offered
Full size table

In Table 5, groups 1 and 4 can be viewed as ‘ambivalent groups’ because their attitudes to the death penalty are subject to change when the alternative option is offered, whereas groups 2 and 3 can be seen as ‘hard-core’ retentionist, maintaining their firm attitudinal position. Results indicate that about 35 to 625 of the respondents would change their attitudes to abolition when different alternatives were offered. Group 3 supports abolition firmly with or without alternatives. Number of cases in Group 4 was too rare to comment upon. As we can see from Table 5, the shifts in attitude to abolition were most likely to happen for the LWOP_L_R alternative and least likely to change for the SD alternative. Since our aim was to analyze shifts in attitude, we do not further analyze group 3 (the other hard-core group), nor do we include group 4 (the other ambivalent group) because of small sample size. Furthermore, groups 1 and 2 are the most interesting targets to be analyzed.

Multivariate Analytic Models: the Relationship Between Capital Punishment and Value Conflict

We developed a two-step process to analyze what made the death penalty supporters change or not.

Step 1: Dependent variable is the degree of support of abolition and independent variables are sets of value conflict when core values are controlled (see Table 6). Linear multiple regressions were applied to these two models (models 1-1 and 1-2) due to a continuous dependent variable. Models 1-1 and 1-2 test the hypotheses that when controlling for core values, to what extent does value conflict explain variation of attitudes to abolition.

Table 6 Regression models of value conflict and public attitudes to abolition of the death penalty (questionnaire B)
Full size table

Step 2: The level of value conflict within group 1 and 2 when comparing SD , LWOP, LWOP_L_R alternatives and abolition. ANOVA results were presented in models 2-1 to 2-3 (see Table 7). Percentage of attitude change was used as a dependent variable for the three models. Models 2-1 to 2-3 analyse that differences in value conflict between respondents shift their attitudes from opposition to support for abolition with alternatives (ambivalent group, coded 1), compared with those who maintain their opposition to abolition (hard-core group, coded 2).

Table 7 ANOVA results of value conflicts between ambivalent group and hard-core group#
Full size table

Results of our analyses in Table 6 show how seven sets of value conflicts sit in relation to abolition. When core values are controlled, model 1-1 shows that greater ambivalence on wrongful conviction/acquittal leads to slightly stronger support for abolition (b = 0.128). Among all core values, people who value greater individualism (b = 0.618), greater concern for wrongful conviction (b = 0.109) and less belief in deterrence (b = 0.114) tend to have significantly stronger support for abolition. Model 1-2, on the other hand, when controlling core values, more value conflicts around wrongful conviction/acquittal (b = 0.116), restorative justice/retribution (b = 0.122) and less value conflict on rehabilitation/retribution (n = − 0.109) provide greater support for abolition. Similarly, significant core value, such as those in model 1-1 can be found also in model 1-2, namely, greater individualism (b = 0.585), greater concern for wrongful conviction (b = 0.107), rehabilitation-oriented (b = 0.162) and less deterrence-oriented (b = 1.081), provides more support for abolition. In other words, both models reveal that value conflict on wrongful conviction/acquittal is important to slightly increase the likelihood of abolition. And when the value of retribution conflicts with restorative justice also increases the support of abolition. However, when retribution and rehabilitation value conflict exists, the support of abolition would be reduced. About 20% of the abolition attitudes variation is accounted for by the regression model in both models (R2 = .211 and .216 respectively). The low R-squared might indicate noise and high-variability of the data which were falling further from the regression line; however, the interpretation of the significant independent variables are the same for both high or low R-squared models.

Table 7 further contains results of applying ANOVA analyses to examine if there could be a relationship between value conflict and the two groups (ambivalent group versus hard-core group), when three alternatives were offered, namely a replacement of suspended death penalty (SD), a replacement of life sentence without parole to the death penalty (LWOP) and a replacement of life sentence, hard labour and restitution (LWOP_L_R). The results indicate that compared to the hard-core supporters of the death penalty, those who considered the SD acceptable held more value conflicts around individualism/collectivism, rehabilitation/deterrence and rehabilitation/retribution, whereas those who considered the LWOP acceptable held more value conflicts around individualism/collectivism only, and those who preferred LWOP_L_R had conflicts across all values except neighbourhood safety/fear of crime. In other words, the results of our alternative-specific analyses show that the LWOP ambivalent group was least likely to shift their attitudes due to value conflicts they had, but SD and LWOP_L_R ambivalent groups were more likely to shift. The LWOP_L_R ambivalent group appear to have the greatest flexibility to shift attitude.

Compared to Table 6, which demonstrated value conflict predictors of attitudes toward abolition, Table 7 reveals the same predictors on attitude shifts. The ANOVA analyses illustrate that value conflict is much more powerful in explaining how attitudes shift than why the public held certain attitudes. Those who were willing to shift their opposition to abolition and towards support for the suspended death penalty, or life without parole or life without parole plus hard labour and restitution (ambivalent groups) face greater value conflicts than those who maintained firm opposition to abolition in the face of all other alternatives to the death penalty (hard-core group 2). The pattern is consistent across all types of value conflict. Group 1 who was willing to change their attitude favourably to abolition obviously has greater value conflicts of restorative justice vs deterrence, rehabilitation vs deterrence and individualism vs collectivism than Group 2. The same group were even more likely to change their attitude towards abolition with LWOP and LWOP_L_R alternatives available than group 2, when group 1 held value conflicts greater, apart from safety vs fear of crime.

We can conclude from the above findings that if a person holds two opposite core values in relation to capital punishment, they are more likely to be undecided on the issue, especially for the LWOP_L_R ambivalent group. We should also recall the key findings in Table 5, which revealed that more than 60% of respondents supported the replacement of capital punishment with the LWOP_L_R. That is to say, the majority of the retentionist public can shift attitudinal position, if offered appropriate alternatives.

Discussion

Countries of the world vary significantly in their use of the death penalty; some using it, others retaining it but rarely using it and some rejecting the penalty (Johnson and Zimring 2009; Amnesty International 2018). Yet, large-scale cross-sectional studies of the sources of variation in possession and use of the death penalty have found that ‘no single characteristic neatly distinguishes between abolitionist and nonabolitionist states.’ (Greenberg and West 2008: 332). Historically too, Western democracies functioned for long periods without particular opposition to the execution of numbers of criminals (Garland 2010; Hood and Hoyle 2015). There is a complexity to understanding this phenomenon of the death penalty, its retention, use and abolition. This complexity is mirrored also in variation in values; for example, retribution. As we referred to in our earlier section on values and moral cognition, there is a significant body of research on public opinion formation that has identified core values (moral foundations) as a fundamental source of general political attitudes (Jacoby 2006). Core values refer to an individual’s deeply held beliefs about what is right and important in life (Rokeach 1973), and in the realm of attitudes to the death penalty, these beliefs serve as shortcuts for forming opinion about issues that are complex. The only multi-country survey to ask directly about retribution (albeit in the context of imprisonment) shows substantial cross national differences. In Gallup International’s (2000) Voice of the People Millennium (VPM) Survey, in 59 countries, the retributive option ranged from a low of 13% in Denmark to a high of 54% in South Korea, with the USA falling at the average of 30%. These cross-sectional data demonstrate that even in a highly globalized world, the prevalence of retribution as a core belief varies substantially across national populations. In the context of the death penalty, commentators generally highlight the complexity of such core beliefs on opinion formation (Unnever et al. 2005; Butler et al. 2017).

In Taiwan, from our analyses of the 2014 survey, there is clearly complexity both to public opinion and arguably in Taiwan’s public discourse on death penalty issues. Although the results of the 2014 survey reveal that overall level of support for the death penalty in Taiwan is undeniably very high, our research demonstrates that people frequently view abolition/retention through complex multiple lenses. Most people are, to a greater or lesser extent, ‘ambivalent’ in considering abolition of the death penalty.

Public attitudes in Taiwan are simultaneously committed to many underlying values which can be viewed as in conflict: agreeing on individualism and collectivism, concerns for both wrongful conviction and acquittal, feeling safe in their neighbourhood, while generally fearful of crime, and with degrees of belief in rehabilitation, deterrence and retribution. Arguably, the aggregate decision the public make reflects the balance of their conflicting values—but what is needed is a more nuanced understanding of how the existence of ambivalence might help us explain how the public makes their judgments based on core and conflicting values. Indeed, as previously mentioned in our introduction to the paper, Sato’s (2013, 2018) research on public support for the death penalty in Japan also points to ambivalence at the individual level. How to account for strength of particular value conflicts between, for example, Japan and Taiwan would be a worthwhile comparative issue to explore. Executions in Japan, probably more so than in Taiwan, are shrouded in secrecy, as are deliberations by judges and lay judges about life and death sentencing decisions (Johnson 2013: 180). This raises the question of whether citizens receive insufficient information about how the death penalty is deployed. More broadly, while our value conflict perspective undoubtedly has its explanatory limits, drawing upon some of the insights from Savelsberg’s (1999) theoretical work on the institutional organisation of knowledge production in a given country and its impact on political and legal decision-making may be highly pertinent. One key advantage of Savelberg’s thesis, from your perspective, is that it has a micro (cognitive dilemmas) and macro component.

Value conflict may also provide purchase on how better to explain the instability of public attitudes to the death penalty. Our findings indicate that when a scale of ‘appropriate’ alternative options to the death penalty is made options, there is always space for punishing the most severe crimes: somewhere between a death sentence and imprisonment (for example, suspended death penalty, life without parole and with hard labour work and restitution). Hence, the most attractive alternative to the public is life without parole plus hard labour work and encompassing restitution to the victim’s family. For the public of Taiwan, notions of ‘restitution’ and being ‘victim-oriented’ would appear to appeal more strongly than concerns over the plight of offenders. This raises the intriguing issue of the gap in Taiwan’s penal spectrum, and whether comparable alternatives should be designed and implemented in Taiwan’s legal system which could serve a similar retribution and incapacitation function that the Taiwanese public expects the death penalty to perform. Taiwan does not have ‘LWOP’Footnote 6, and any such proposal would entail considerable costs, as well as denial of human agency for the possibility of rehabilitation (see Kleinstuber et al. 2017 for the US experience and the unintended consequences of LWOP). Yet, LWOP and its various combinations can offer the public a persuasively ‘retributive’ alternative and instrumentally reduce support for retention.

Like many other sensitive political policies in contemporary Taiwan, for example, single-sex marriage, now legalized (British Broadcasting Corporation News 2017; Ho 2019) or even ‘Taiwan independence’ from the People’s Republic of China (British Broadcasting Corporation 2019), capital punishment is often presented as an issue of extremes, with an abolitionist group at one end of a spectrum pitted against retributivist penal populism at the opposite end. Differences between these two supposed ‘groups’ seem impossible to resolve. Our study, however, discloses that numbers of the public hold both positive and negative attitudes towards issues related to the death penalty (i.e. the role of the State, Type I and II errorsFootnote 7 made by the legal system, actual safety and fear of crime and aims of punishment). The ambivalent groups who might lack a firm opinion when alternative choices are proposed or attach relatively little importance to capital punishment are by no means incapable of change. Abolitionist reform groups are sometimes uncomfortable with noting and privileging public opinion on the death penalty, which in debate is often portrayed as based upon emotional logic, fuelled by revenge. Yet, 82% of the public supports the death penalty in Taiwan; this surely should not be portrayed as merely faulty logic, based on inaccurate information or lack of public understanding. Recognition of the value conflicts behind public attitudes to the death penalty is a step forward in understanding the relationship that people hold towards punitiveness and forgiveness and in the complexity of their cognitive affect towards both victims and offenders.

Findings of this study have some implications for advocates of both abolition and retention. While for many criminologists there are serious questions about whether anything useful about the deterrent value of the death penalty can ever be learned from any data that are likely to be available (National Research Council 2012; Muramatsu et al. 2018), for abolitionists, our results indicate a continuing need to redouble efforts on distributing information about the known limited deterrent effects of the death penalty. Overall, belief in deterrence among the Taiwanese public (68%) is higher even than Oberwittler and Qi’s (2009) general population study in China (59%). Similarly, abolitionists face what one could term the ‘challenge of rehabilitation’. Among the Taiwanese public, only some 16% agreed with the question that execution is unnecessary because all criminals can be rehabilitated, compared with 45% in the Chinese population (Oberwittler and Qi 2009). Our view is that the high percentage on the Chinese side may best be understood as a legacy of the Mainland Chinese belief in reform or ‘ganhua’ (Dikotter 2002; Bakken 2005). Part of a constant striving to educate the Chinese people in terms of the imperative of virtue, the idea according to which even the most hardened criminal was capable of repentance and self-rehabilitation through a process of moral transformation (ganhua) was at the heart of the penal philosophy of modern China. As much as in the imperial past, and drawing on ideas of Confucius and Mencius on ‘perfectibility of human nature’, Bakken notes, the principle of ganhua, was transformed in twentieth century China ‘from a mere principle of penology to a more general political concept.’ (Bakken 2005: 4). In contrast, in Taiwan, a loosely regulated mass media report regularly on very poor prison conditions and comparatively little spending on prison services; it is little wonder, therefore, that the public assumes the opportunity for any effective rehabilitation of prisoners is vanishingly small (see Jou 2014).

For retentionist enthusiasts in Taiwan, our results reveal the importance in strengthening appeals to societal order (collectivism) over individualist claims. Here, one should understand collectivism not just as part of traditional culture but as a cultural form of organizing society that can define a modern type of society as well (Abbot 1970). Hofstede (1979) found that there exist ‘I’ and ‘We’ type societies of culture; normally, the I type is found in modernised capitalist Gesellschaft-type societies, while the latter dominate Gemeinschaft societies. However, Taiwan, like Singapore, stands out, with a We value set (see Bakken 2000: 106-111).

Turning to priority areas for future research on the death penalty in Taiwan, we would identify two. First, in terms of continuing research on attitudes about the death penalty, it is imperative that funding in Taiwan allows for regular, ideally biennial, national stratified sample face-to-face surveys. Sample size should be large enough to allow meaningful analyzes on demographic and other characteristics of those identified as ambivalent (value conflict/weakly held sentiment). There needs to be a set of questions asking respondents the degree and strength to which they endorse identified core values, based on prior research (Krosnick and Abelson 1992). From our results and those of others outside of Taiwan, one may predict that those who have conflicted core values should be more likely to choose combinations of conditions that allow them to support the death penalty. On the other hand, those without conflicting values would more likely produce an end of spectrum response along the lines of ‘no one’ or ‘all’ should be executed. Further, there is a pressing need for intensified collaborative work with psychologists and others to generate empirical agendas around theoretical developments; perhaps based on a cognitive-dissonance reduction model, of the type by Bohm and his colleagues some time ago, seeking to assess death penalty opinion consistency with particular personality traits and core values (Bohm et al., 1993; Unnever et al. 2005). For example, in relation to rehabilitation, examining the potential value conflict associated with it as a core value, and the recognition of Taiwan’s low pecuniary spend per capita on prison (Jou 2014). We would argue that the synergy generated from such collaboration will be of immense benefit to obtaining a more satisfactory grasp of attitude formation and shift. Core values, while perhaps slow to change, are not immutable. In the USA, many commentators attributed the decline in support for the death penalty during the 1960s to a reframing of the death penalty debate in terms of ‘fairness’. Another prong to such a research agenda can be derived from our earlier discussion of moral cognition. While we broadly agree with Taiwan-based psychologist KK Hwang that ‘an understanding of a particular culture is required for obtaining a comprehensive understanding of moral reasoning in that culture.’ (Hwang, 1998: 211), an empirical approach using a suitably amended moral foundations theory to analyse, intuitions, reasoning and emotions constituting moral judgement on the death penalty in Taiwan may pay dividends, especially since we know that moral foundations theory has successfully devised interventions aimed at changing (political) attitudes.

Second, as we opined at the beginning of our discussion above, public discourse about the death penalty as a social institution in Taiwan is complex. Yet, unlike in the USA, for example, where we know a considerable amount about the significance of central ‘narratives and that supply the institution with its dominant meanings’ (Garland 2010: 61), in Taiwan, we know very little. How people in Taiwan imagine and describe the process of sentencing and execution both reflects and constitutes policy and policy change. Yet, there has been no systematic research on how Taiwanese people, NGOs and government communicate through mass and social media about the death penalty. In this regard, Kita and Johnson’s (2014) landmark study of Japan’s newspaper reporting is something surely to be emulated in Taiwan. Their study, as well as revealing previously poorly recognized tropes, identified the importance of certain ways the death penalty is framed, and how the flexibility of such framing can suggest new possibilities for Japan’s abolitionist movement. Like Japan, Taiwan is a ‘hypermediated’ society; the diversity of Taiwan’s media is reflected in the plethora of outlets and intense competition. Taiwan, with a population of barely 23 million, had more than nine 24-h news channels, around 100 television channels in total (and only one state ‘run’ channel). There are about 25 national daily newspapers, but four of them (Liberty Times, China Times, Apple Daily and United Daily News) account for 85–90% of the market share (Jou, 2014: 298). The nation’s press operates in a media environment that is one of the freest in Asia, and Taiwan was indeed ranked the top Asian nation for the sixth consecutive year (World Press Freedom Index, 2018). Thus, Taiwan’s media environment is distinct, even from Japan, in at least three ways. First, a highly competitive media environment forces much of commercial media toward sensationalism. Second, political party affiliations of most of the commercial media influence the level and quality of coverage around issues and interest groups, and finally, media organizations have developed various perceptions of issues and interest groups regarding their non-partisanship, autonomy and credibility. This is the environment within which the cultural realm of death penalty public discourse occurs, and its complexity urgently needs research.