The mass public has become affectively polarized—ordinary Americans increasingly dislike and distrust those from the other party, with negative consequences for politics. Drawing on work in political and social psychology, this paper tests two mechanisms for reducing this discord, both of which have been shown to reduce similar biases in other settings: heightening partisan ambivalence, and using self-affirmation techniques. A population-based survey experiment shows that neither strategy reduces affective polarization in the aggregate. But this null finding masks an important heterogeneity: Heightening partisan ambivalence reduces affective polarization for ideological moderates, but increases such discord for those with more extreme ideological identities. Efforts to depolarize the electorate can make it more deeply divided, with important implications for our understanding of contemporary politics and the durability of affective polarization. Few public opinion topics have attracted more attention in recent years than political polarization. While scholars disagree on whether ordinary voters’ issue positions have become more divided (Fiorina, Abrams, and Pope 2005), a consensus exists that affective polarization has increased—voters today dislike and distrust those from the other party more than they did just a few decades ago (Iyengar, Sood, and Lelkes 2012; Abramowitz and Webster 2016). Such affective polarization lowers political trust and makes governance more difficult (Hetherington and Rudolph 2015), hampers interpersonal relations Matthew S. Levendusky is a professor in the Department of Political Science (and, by courtesy, in the Annenberg School for Communication) at the University of Pennsylvania, Philadelphia, PA, USA. The author thanks the following individuals and groups for helpful comments on this paper: the editors and anonymous referees, as well as Jamie Druckman, Don Green, Bryan Gervais, the Penn American Politics Working Group, seminar participants at Columbia University, and seminar participants at the 2017 Annual Meeting of the Midwest Political Science Association. Data were collected by Time-Sharing Experiments for the Social Sciences [NSF Grant Number 1227179 to Jeremy Freese and James Druckman, Principal Investigators]. *Address correspondence to Matthew S. Levendusky, Department of Political Science, 133 South 36th Street, Philadelphia, PA 19104, USA; email: mleven@sas.upenn.edu. Public Opinion Quarterly doi:10.1093/poq/nfy036 Downloaded from https://academic.oup.com/poq/advance-article-abstract/doi/10.1093/poq/nfy036/5088191 by University of Pennsylvania Libraries user on 31 August 2018 (Huber and Malhotra 2017), and hinders economic exchange (McConnell et al. 2018). Is it possible to reduce this affective polarization? As it stems primarily from the power of partisanship to divide the world into ingroups and outgroups, weakening this group-centric thinking should reduce affective polarization. This paper outlines and tests two mechanisms for achieving this goal: increasing partisan ambivalence (Lavine, Johnston, and Steenbergen 2012) and using self-affirmation techniques (Steele 1988). Using a population-based survey experiment, I find no evidence that the self-affirmation treatment has any effect on affective polarization. Heightening partisan ambivalence similarly has no aggregate effect, but this masks an important heterogeneity: The treatment decreases affective polarization for ideological moderates, but increases such polarization for others. Moderates enter the study much less polarized than other respondents, so the treatment works best for those who need it least, and does little to actually reduce discord in the electorate. These results highlight that reducing affective discord will likely be quite difficult, precisely because of its intimate connection to our most fundamental political identity: partisanship. Affective Polarization, Social Identity, and Efforts to Mitigate It Affective polarization refers to a dislike of members of the opposing party, and a tendency to impute negative characteristics to them (Iyengar, Sood, and Lelkes 2012, p. 407).1 For example, the average feeling-thermometer ratings given to the opposing party have dropped considerably since the late 1980s (Iyengar, Sood, and Lelkes 2012), with 2016 representing the apex of such animus toward the other party (Pew Research Center 2016). For some individuals, partisan prejudice now outstrips racial prejudice (Iyengar and Westwood 2015). Such affective polarization is not simply cheap talk—individuals will pay a real economic cost to signal their partisan allegiance (McConnell et al. 2018), and this has effects not only on governance (Hetherington and Rudolph 2015), but also on everyday interactions (Huber and Malhotra 2017). Affective polarization stems directly from identification with a political party. Identifying with a party divides the world into an ingroup (one’s own party) and an outgroup (the opposing party), which leads to ingroup favoritism, the tendency to like one’s own (partisan) group and dislike the outgroup (Tajfel and Turner 1979). Affective polarization is an outgrowth of this classification: Because the other party is an outgroup, they take on a negative connotation and individuals dislike and distrust them (Iyengar, Sood, and 1. Note that affective and ideological (issue-based) polarization are distinct concepts that are only weakly related empirically (Mason 2015)—one can increase while the other decreases (Levendusky and Malhotra 2016). Levendusky Page 2 of 10 Downloaded from https://academic.oup.com/poq/advance-article-abstract/doi/10.1093/poq/nfy036/5088191 by University of Pennsylvania Libraries user on 31 August 2018 Lelkes 2012; Iyengar and Westwood 2015). The contemporary political environment—with polarized elites starkly divided along partisan lines—further accentuates this tendency toward partisan animus and affective polarization. This definition of affective polarization also suggests a mechanism for mitigating it: weakening partisan group-centrism and accompanying biases. Here, I draw on two strategies that have ameliorated similar biases in other contexts and test their ability to lessen affective polarization. If strategies that have worked elsewhere fail here, it tells us something important about how ingrained affective polarization is, and how difficult it likely will be to overcome. The first strategy involves heightening partisan ambivalence, a conflict between individuals’ partisan identities and their short-term evaluations of or feelings toward the parties (Lavine, Johnston, and Steenbergen 2012, pp. 3–5). When individuals are univalent (nonambivalent) partisans, partisanship strongly colors their attitudes, vote choice, economic assessments, and evaluation of new information, precisely because partisanship is so central to political reasoning. But when ambivalence is induced (by, say, reflecting on what one dislikes about one’s own party and likes about the other party), partisanship is displaced from its normally central role in political decision-making, and individuals lessen their reliance on it as a heuristic (Lavine, Johnston, and Steenbergen 2012; Bolsen, Druckman, and Cook 2014). Ambivalence, by suggesting where individuals diverge from their party, also weakens sameparty identification, and the ability of partisanship to divide up the world into ingroups and outgroups (Brewer 1991). So, by ameliorating partisan groupcentrism, ambivalence dampens affective polarization and discord. Hypothesis 1: Priming partisan ambivalence will reduce affective polarization. Second, self-affirmation techniques can be used to the same ends. Selfaffirmation techniques build from the recognition that affective polarization and related phenomena reflect a desire to maintain a positive self-image (Oakes and Turner 1980). Individuals want to think highly of themselves, and favoring their partisan ingroup (and disparaging the outgroup) is a mechanism for achieving that goal. But of course there are also many apolitical mechanisms for maintaining a good self-image. If individuals reflect on, for instance, their musical talent, business acumen, or any other positive apolitical trait, they feel good about themselves and have less need to defend their partisan group, thereby reducing partisan bias and affective polarization (Steele 1988). For example, Binning et al. (2010) show that, around the 2008 election, self-affirmation reduced partisan biases in candidate evaluations. Similarly, numerous studies show that self-affirmation makes individuals more open to new information, and less defensive of their side’s positions (e.g., Cohen et al. 2007). This same logic applies to affective polarization: Affirming some apolitical aspect of an individual’s personality lessens their need to engage in identity-protecting group favoritism, so they will have less need to disparage the other party. Hypothesis 2: Self-affirmation techniques will reduce affective polarization. When Efforts to Depolarize the Electorate Fail Page 3 of 10 Downloaded from https://academic.oup.com/poq/advance-article-abstract/doi/10.1093/poq/nfy036/5088191 by University of Pennsylvania Libraries user on 31 August 2018 However, there should be an important heterogeneity in the effectiveness of these treatments. Respondents with stronger partisan or ideological identities have more efficacious and durable identities that will resist the impact of these types of primes (Mason 2015), and hence the treatment effects will be smaller for these individuals. Hypothesis 3: Both self-affirmation and partisan ambivalence will be less effective at reducing affective polarization for those with stronger identities. Experimental Strategy and Results I investigate whether either strategy—increasing partisan ambivalence or using self-affirmation techniques—can reduce affective polarization using a population-based survey expe

中文翻译：

---

当去极化选民的努力失败时

大众在情感上变得两极分化——普通美国人越来越不喜欢和不信任对方的人，这对政治产生了负面影响。本文借鉴了政治和社会心理学方面的工作，测试了两种减少这种不和谐的机制，这两种机制都已被证明可以减少其他环境中的类似偏见：加剧党派矛盾和使用自我肯定技巧。一项基于人群的调查实验表明，这两种策略都不会总体上减少情感两极分化。但这个无效的发现掩盖了一个重要的异质性：加剧党派矛盾会减少意识形态温和派的情感两极分化，但会增加那些具有更极端意识形态身份的人的这种不和谐。去极化选民的努力会使他们更加分裂，对我们理解当代政治和情感两极分化的持久性具有重要意义。近年来，很少有舆论话题比政治两极分化更受关注。虽然学者们对普通选民的问题立场是否变得更加分歧存在分歧（Fiorina、Abrams 和 Pope 2005），但存在一种共识，即情感两极分化加剧——今天的选民比少数人更不喜欢和不信任对方的人几十年前（Iyengar、Sood 和 Lelkes 2012；Abramowitz 和 Webster 2016）。这种情感两极分化降低了政治信任并使治理更加困难（Hetherington 和 Rudolph 2015），阻碍了人际关系 Matthew S. Levendusky 是政治学系教授（并且，出于礼貌，在美国宾夕法尼亚州费城宾夕法尼亚大学安纳伯格传播学院获得博士学位。作者感谢以下个人和团体对本文的有益评论：编辑和匿名审稿人，以及 Jamie Druckman、Don Green、Bryan Gervais、宾夕法尼亚大学美国政治工作组、哥伦比亚大学研讨会参与者和研讨会参与者中西部政治学协会2017年年会。数据由社会科学的分时实验收集 [NSF 资助号 1227179，授予 Jeremy Freese 和 James Druckman，首席研究员]。*地址与 Matthew S. Levendusky, Department of Political Science, 133 South 36th Street, Philadelphia, PA 19104, USA 的通信地址；电子邮件：mleven@sas.upenn.edu。民意季刊 doi:10。1093/poq/nfy036 宾夕法尼亚大学图书馆用户于 2018 年 8 月 31 日从 https://academic.oup.com/poq/advance-article-abstract/doi/10.1093/poq/nfy036/5088191 下载（Huber 和 Malhotra 2017） ，并阻碍经济交流（McConnell 等人，2018 年）。是否有可能减少这种情感两极分化？由于它主要源于将世界分为内群体和外群体的党派力量，削弱这种以群体为中心的思维应该会减少情感两极分化。本文概述并测试了实现这一目标的两种机制：增加党派矛盾（Lavine、Johnston 和 Steenbergen 2012）和使用自我肯定技术（Steele 1988）。使用基于人群的调查实验，我发现没有证据表明自我肯定治疗对情感极化有任何影响。加剧党派矛盾同样没有聚合效应，但这掩盖了一个重要的异质性：治疗减少了意识形态温和派的情感两极分化，但增加了其他人的这种两极分化。与其他受访者相比，温和派进入研究的两极分化要小得多，因此这种治疗对那些最不需要它的人最有效，并且几乎没有真正减少选民的分歧。这些结果突出表明，减少情感不和可能会非常困难，正是因为它与我们最基本的政治身份：党派关系密切相关。情感极化、社会认同和缓解它的努力 情感极化是指不喜欢对方成员，并倾向于将负面特征归咎于他们（Iyengar、Sood 和 Lelkes 2012，第 407 页）。1 例如，自 1980 年代后期以来，给予对方的平均感觉温度计评分已大幅下降（Iyengar、Sood 和 Lelkes 2012），而 2016 年代表对另一方这种敌意的顶点（Pew Research Center 2016） . 对于某些人来说，党派偏见现在超过了种族偏见（Iyengar 和 Westwood 2015）。这种情感两极分化不仅仅是廉价的谈话——个人将付出真正的经济成本来表明他们的党派效忠（McConnell 等人，2018 年），这不仅会影响治理（Hetherington 和 Rudolph 2015），还会影响日常互动（ Huber 和 Malhotra 2017）。情感两极分化直接源于对政党的认同。认同一方将世界分为内群（自己的一方）和外群（对方），这会导致内群体偏袒，即喜欢自己（党派）群体而讨厌外群体的倾向（Tajfel 和 Turner 1979）。情感两极分化是这种分类的产物：因为对方是外群体，所以他们具有负面含义，个人不喜欢和不信任他们（艾扬格、苏德和 1. 请注意，情感和意识形态（基于问题的）两极分化是不同的经验上仅弱相关的概念（Mason 2015）——一个可以增加而另一个减少（Levendusky and Malhotra 2016）。Levendusky 第 2 页，共 10 页，下载自 https://academic.oup.com/poq/advance-article-abstract /doi/10.1093/poq/nfy036/5088191 由宾夕法尼亚大学图书馆用户于 2018 年 8 月 31 日 Lelkes 2012 年；Iyengar 和 Westwood 2015 年）。当代政治环境——两极分化的精英按照党派界限明显分化——进一步加剧了这种党派敌意和情感两极分化的趋势。情感极化的这种定义也提出了一种缓解它的机制：削弱党派群体中心主义和伴随的偏见。在这里，我利用了两种在其他情况下改善了类似偏见的策略，并测试了它们减少情感两极分化的能力。如果在其他地方行之有效的策略在这里失败，它会告诉我们一些重要的事情，即情感两极分化是多么根深蒂固，以及克服它的难度可能有多大。第一种策略涉及加剧党派矛盾，即个人的党派身份与他们对党派的短期评估或感受之间的冲突（拉文、约翰斯顿、和 Steenbergen 2012，第 3-5 页）。当个人是单一（非矛盾的）党派人士时，党派关系会强烈影响他们的态度、投票选择、经济评估和对新信息的评估，正是因为党派关系对政治推理如此重要。但是，当产生矛盾心理时（比如，反思自己不喜欢自己的政党和喜欢对方的什么），党派之争就不再是其通常在政治决策中的核心作用，个人就会减少对它的依赖。启发式（Lavine、Johnston 和 Steenbergen 2012；Bolsen、Druckman 和 Cook 2014）。矛盾心理，通过暗示个人与政党的分歧，也削弱了同党认同，以及党派之争将世界划分为内群体和外群体的能力（Brewer 1991）。所以，通过改善党派团体中心主义，矛盾心理可以抑制情感两极分化和不和谐。假设 1：引发党派矛盾会减少情感两极分化。其次，自我肯定技术可以用于相同的目的。自我肯定技术建立在承认情感极化和相关现象反映了保持积极自我形象的愿望的基础上（奥克斯和特纳，1980）。个人想要高度评价自己，支持他们的党派内群体（并贬低外群体）是实现这一目标的一种机制。但是当然也有许多非政治性的机制来保持良好的自我形象。例如，如果个人反思他们的音乐才能、商业头脑或任何其他积极的非政治特质，他们自我感觉良好，不需要为自己的党派团体辩护，从而减少党派偏见和情感两极分化（Steele 1988）。例如，宾宁等人。(2010) 表明，在 2008 年大选前后，自我肯定减少了候选人评估中的党派偏见。同样，大量研究表明，自我肯定使个人对新信息更加开放，并且对自己立场的防御性降低（例如，Cohen 等人，2007 年）。同样的逻辑也适用于情感两极分化：肯定个人性格的某些非政治方面会减少他们参与保护身份的群体偏袒的需要，因此他们将不需要贬低另一方。假设 2：自我肯定技术会减少情感极化。当去极化选民的努力失败时 第 3 页，共 10 页 宾夕法尼亚大学图书馆用户于 2018 年 8 月 31 日从 https://academic.oup.com/poq/advance-article-abstract/doi/10.1093/poq/nfy036/5088191 下载然而，这些治疗的有效性应该存在重要的异质性。具有更强党派或意识形态身份的受访者具有更有效和持久的身份，可以抵抗这些类型的素数的影响（Mason 2015），因此对这些人的治疗效果会更小。假设 3：自我肯定和党派矛盾在减少具有较强身份认同的人的情感两极分化方面效果较差。