




已阅读5页,还剩13页未读, 继续免费阅读
版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
The Effects of Published Polls on Citizens公开民意调查对市民的影响Modem political opinion polls are accompanied by two constants: the debate about their quality on the one hand (- The Methodological Strengths and Weaknesses of Survey Research), and the debate about their alleged effects in the run-up to elections and voting behavior on the other. The latter includes numerous political efforts (some of which have been implemented) to prohibit publications of opinion polls prior to voting (Foundation for Information, 2003; The Legal Status of Public Opinion Research, in the World). Furthermore, this alleged effect goes hand in hand with the strong assumptions of journalists concerning the impact of publishing poll results (Donsbach & Antoine. 1990; Hardmeier, 2000; Attitudes of Journalists Toward Public Opinion Research) and soon gave rise to analyses about the press-polling connection (Crespi, 1980; Ladd, 1980) or the news medias use of polls ( The News Medias Use of Opinion Polls). The effects of published polls on voters are the focus of the present chapter.现代政治民意调查伴随着两个不变的事物:一个是对调查质量的讨论(调查研究方法论的优势和劣势),另一个是选举预备阶段和投票行为中所谓的效应。后者包括了许多为防止民意调查在选举(信息基础,2003; 民意调查在世界上的法律地位)之前公开的政策(其中一些已经开始实施)。此外,这个所谓的效应与记者对公开民意调查结果影响的有力假设密切相关(Donsbach & Antoine. 1990; Hardmeier, 2000; 记者对公开民意调查的态度),并且很快引发了对压力投票关联的分析,以及媒体对民意调查的利用的分析(媒体对民意调查的利用)。本章将集中讨论民意调查发布对投票人的影响。The remainder of this chapter is organized along three answers to the question why there have been so many assumptions about a strong impact of published polls. The first answer implies a differentiation between the dependent variables (effect on what?) and summarizes the corresponding meta-analytical findings. The second answer summarizes the current media impact research that provides important theoretical tools for evaluating possible effects: the role of predispositions as well as issue and campaign characteristics that may intervene in processes of influence. The meta-analytical finding that the literature on polling impact merges very heterogeneous effects gives rise to the third answer: The discussion about effects of opinion polls needs to differentiate to a much greater extent between different effects and possible processes, which at times even cancel each out. This may again highlight the complexity of these possible effects; however, these statements may also serve as a basis for politicians and researchers to draw some final conclusions.本章的剩余部分将会围绕同一个问题的三个答案展开,这个问题就是为什么对公开民意调查强烈影响的假设会如此之多。第一个答案暗示了因变量(对什么产生影响?)之间的区别,并总结了相应的元分析调查结果。第二个答案则是对当前对媒体影响的调查进行了总结,这些调查为评估可能的结果提供了理论工具。人们的意向与议题、竞选特点一样,会干涉影响过程。元分析调查的发现,民意调查产生的影响方面的研究文献能融合不同种类的效应,这引出了第三个答案:对于民意调查效应的讨论需要在更大程度上将不同的效应和可能的过程区分开,这些有时候甚至会相互抵消。这样就又一次突出了这些可能效应的复杂性;然而,这些陈述也能够为政治家和研究人员得出最终结论提供基础。RESEARCH EVIDENCE研究证据When looking at the history of the debate on the impact of published polls, the persistence of assumptions about a strong impact stands out. This can be explained by two ideal-typical circumstances: either research supports the strong impact assumption or respective research evidence is so modest or contradictory that making such assumptions is particularly easy. In the present case, however, the matter is somewhat more complicated. The body of research is by no means modest. For their meta-analytical overview compiled in 2000, Hardmeier and Roth (2001) identified 34 studies that could be evaluated quantitatively. Moreover, there are more than 30 studies that could be reviewed qualitatively, as well as a comprehensive list of references including numerous recapitulating monographs and various theoretical analyses, most of which follow the traditional line of the Rational Choice Theory.看着对民意调查的影响的讨论历史,我脑中出现了一个有关强烈影响的假设,久久挥之不去。这可以由两种理想型情况来说明:研究支持强烈影响假说的情况,还有各自的研究证据太过保守或太过矛盾,致使这种假设变得格外简单的情况。然而在现在这个例子中,问题就变得格外复杂。研究的主体一点也不温和。在2000年编译的元分析概观中,Hardmeier 和 Roth (2001)定义了34种研究,都可以进行定量评估。另外,还有30多种研究以及包括大量专著摘要和多种理论分析的参考书目可进行定性审查,在这些理论分析中,大多数遵循理性决策理论的传统观点。Thus, research evidence is not at all sparsebut rather confusing as the findings are very disparate. Furthermore, the studies have applied various theoretical approaches and were executed in different disciplines, all of which has led to a lack of mutual reception of the results. All in all, the state of research is characterized by results that do not accumulate very well, so it stands to reason that politicians and regulatory bodies complain about confusing and inconsistent results (Dach, 1997). The present chapter therefore aims to contribute to the accumulation of the state of research, to include literature from non-English speaking areas and to alleviate the prevailing empirical and theoretical confusion. One the one hand, this effort is based on the method of meta-analysis, which allows a quantitative summary of the impact strength measured in published studies. On the other hand, we present an analytical framework to distinguish between possible effects both theoretically as well as in terms of causality.由于调查结果之间联系甚少,因此,问题不是研究证据不足,而是这些证据非常令人困惑。此外,这些研究运用了不同的理论方法,并且调查遵循的准则各有不同,所有这些都成为了导致结果缺乏相互包容性的原因。总而言之,研究状况的特点就是研究结果不能很好的聚集,因此,政治家和监管部门抱怨结果令人困惑和不一致也就不足为奇(Dach, 1997)。本章旨在促进研究状况的积累,包括非英语地区的文献。另外,本章将缓解当前的经验和理论困惑。一方面,这种努力是基于元分析方法上的,它允许了在已发布的研究中测量过的对冲击强度的定量总结。另一方面,我们提出了一个在理论上和因果关系上区分可能的影响的分析框架。DISTINGUISHING THE DEPENDENT VARIABLES: EFFECTS ON WHAT?区分因变量:影响什么?When discussing effects in the context of opinion polls, it is not mainly (Studying Elite vs. Mass Opinion) are concerned with, but the general voters. So we distinguish between two dependent variables or factors that potentially can be influenced: participation in elections on the one hand (turnout) and the voting behavior or the intention to vote for a specific party or candidate on the other (preference). In the first case, we can identify mobilizing and demobilizing effects and the special case of potential effects of exit polls published on Election Day.2 In the latter case, the effect of changing ones voting intention or actual vote towards the party or candidate (in cases of referenda: issue positions) that are leading in the polls is usually referred to under the generic terms of a bandwagon effect3; the effect of changing in the direction of the party, candidate or issue position that is trailing behind in the polls is termed an underdog effect. There are a few other hypotheses, some of which refer to the effects of published polls on so-called tactical voting. For instance, in a multi-party system voters might vote for the party of their second choice in order to facilitate a certain coalition, or to avoid an absolute majority (for an overview of the different effects hypotheses see Donsbach, 2001, p. 22). 当讨论民意调查的影响时,我们主要关心的并不是精英(研究精英和大众观点)而是普通选民。因此我们区分可能被影响的两个单独的变量或因素:一方面,参与选举(“到会者”),另一方面,投票行为或者对某个特定党派或候选人的投票意向(“偏好”)。在第一种情况下,我们可以定义动员或者遣散效应以及在选举日公布的特殊案例出口民意调查的潜在影响。在第二种情况下,改变一个人的投票意向或者对政党或候选人的实际投票走向(在公投中:问题立场)在民调中领先的效果通常被称为 “从众效应”;改变支持党派和候选人在民调中落后的效果被称为“哀兵效应”。有一些其他的假设,其中涉及到所谓的“战略投票”。例如,多党制选民可能为其第二选择党派投票以促进某个联盟,或者避免绝对多数(不同效应假设的概论,见Donsbach, 2001, p. 22)Table 47.1 gives an overview of the main characteristics of the studies included in the meta-analysis. The majority of the studies have investigated opinions towards parties or candidates as the dependent variable.Table 47.1 Overview of the findings included eristics the meta-analysisCharacteristics n %Number of studies 34 100.0Year of publication - Before 1980 12 35.2- After 1980 22 64.7Countries- USA 53 71.6- United Kingdom 10 13.5- Canada 7 9.4- Germany 3 4.1- Mixed 1 1.4Findings(Unit of analysis)Design - Experimental design(field and laboratory)- Survey design(panel and cross-sectional)- Aggregate dataDependent variable- 1. Turnout- 2a.Election: election, preference for and feelings toward a person or party- 2b.Issues: preference for and feelings toward an issueEffects for 2a and 2bBandwagon(n=45); without aggregate data studiesUnderdog(n=17); without aggregate data studies 表47.1对元分析里研究的主要特点做了一个概述。大部分的研究把对政党或候选人的看法作为因变量研究。The results of the meta-analysis confirm that this focus on preferences in the context of elections and referenda is well founded: all in all, the influence of polls or exit polls on voter participation is marginal (Hardmeier & Roth, 2001). Statistically, the zero hypothesis cannot be refuted: with the exception of the two findings in the Jackson (1983) study, whose method was heavily criticized,4 the effect coefficient rtet equals 0.0176.5 In addition, the Fail-Safe- /V-Test6 suggests the results could easily be refuted. Merely another eight findings without the effect would be required to be able to plead for maintaining the zero hypothesis. Furthermore, the minus sign suggests that opinion polls have a demobilizing effect, if any at all.元分析的结果证实,在选举和公投背景下对自己偏爱的关注十分合理,总而言之,民意调查或出口民意调查对选民参与的影响是微不足道的(Hardmeier & Roth, 2001)。在统计学上,零假设不能被反驳:除了两个调查结果之外,Jackson的调查方法受到了严厉的批判,其效应系数为0.0176 。此外,Fail-Safe-N-Test 表明结果很容易被推翻。仅仅只有另外八个不受影响的发现能维持零假设。此外,负号表示民意调查有调动效应,如果算的话。There is also no strong effect of published polls on voting intention or voting for a party, candidate or issue position. The impact variables, calculated in terms of a tetrachoric correlation coefficient, are weak: if findings relying on aggregate data are excluded because of their rather weak caused evidence, and if conventional meta-analytical collections for sample size of the studies (Hunter, Schmidt, & Jackson, 982, p. Fricke & Treinies, 1985, PP. 124 ff.) are applied,7 the r is 0,1102 (for bandwagon effects) and -0.0336 ( or which sums up to a net effect), which sums up to a net effect of 0.0536.民意调查对投票意向、投票党派、候选人以及问题立场的影响并不明显。以四项相关系数累计的影响变量很弱:如果因为异常薄弱的因果关系证据而依靠集合数据的发现被排除在外的;如果对样本量研究的传统元分析修正被采纳(Hunter, Schmidt, & Jackson, 1982, p. 41; Fricke & Treinies, 1985, PP. 124 ff.)那么效应系数则为从众效应0.1102,哀兵效应0.0336,总计净效应为0.0536.If the statistical dependence of some findings originating from the same studies is taken into account, the effect coefficients amount to 0.1205 (for bandwagon effects) and -0.0361 (for underdog effects), which sums up to a net effect of 0.0431 and an absolute effect coefficient value of 0.0787. By conventional standards these are very weak effects, but by no means insignificant when considering the real world importance of treatment effects (Rosenthal & Rubin, 1982, p 157). If, for instance, published poll results would increase the support for party A by 5 points from 48% to 53%, this would correspond to an rtet value of 0.0875 but it would also change the outcome of the election.如果将来源于相同研究的一些发现的相关统计性考虑在内,从众效应系数为0.1205,哀兵效应为0.0361,总计净效应为0.0431,绝对效应系数值为0.0787.按照传统标准,这些都是非常弱的效应,当考虑到真实世界待遇效应的重要性时,这些就没有意义了(Rosenthal & Rubin, 1982, p 157)。如果,例如,公开民意调查的结果能够使A党派的支持率从48%提高5个百分点到53%,刚好对应效应系数0.0875,不过,它将会改变选举的结果。Furthermore, when applying an estimate for significance (z-value) and the Fail-Safe-N- Test along with the meta-analytical methods correction for sample size and test for independence and homogeneity, the weak bandwagon effect will actually turn out to be quite robust. The integrated z-value (15.74) indicates a significant result for the 42 bandwagon findings. Furthermore, more than 55 significant findings implying the opposite effect would be required to fall short of the usual criteria (Fail-Safe-N-Test).此外,当对意义(Z值)和故障安全试验以及样本量元分析校正法还有独立性和均一性测试进行评估时,微弱的从众效应就会变得异常活跃。集成的z值(15.74)反映了42个从众发现的有意义的结果。此外,超过55个有意义的发现表明相反的效应将达不到寻常标准(Fail-Safe-N-Test)。The obvious heterogeneity of the findings suggests that there are important intervening variables such as the methods applied in the study, the year of publication or the way of conveying opinion results to the public. The bandwagon effect is somewhat stronger in (field) experiments (however, there were no replications after 1990), and in studies using trend data rather than snapshot results.发现结果的明显异质性表明确实存在一些重要的干涉变量,例如研究所运用的方法、公布的年份,或者公布民意调查的方式。从众效应在实验或者运用趋势数据的研究中比在快照结果中稍微强烈一些(然而在1990年后就没有反响了)。PREDISPOSED VOTERS AND ISSUES选民倾向和问题The rather weak effects are not surprising considering the lines of theoretical development in the field of media effects research. While the first theorists of mass media communication proceeded on the assumption that the impact of the media was considerable and influenced the masses psychologically, the first empirical studies demonstrated almost the opposite or a very weak impact, and since the early 1970s, research has been drawing attention to the selective effect of the media. Today, media effects are seen in the context of personal prerequisites and the respective prevailing conditions.当考虑到媒体效应研究领域的发展时,这些异常微弱的效应也就不足为奇了。在大众传播媒介的第一批理论家坚持并发展了“媒体的影响非常大,且在精神上影响着受众”这一假设时,第一个经验研究显示的结果却几乎相反:媒体对大众的影响甚微。而且自从20世纪70年代之后,研究的重点就转移到了媒体的选择性影响方面。现如今,媒体影响在个人预备知识和各自现行条件方面都有体现。In this, the concept of predisposition is important. It was initially understood as a possible determining factor in (political) behavior, but was then increasingly regarded as the actual perception hurdle or filter (Schmitt-Beck, 2000). Moreover, along with the actual behavioral part in the persuasion processes, information processing became the focus of attention and the object of detailed empirical investigations. As a result, the effect of the media was gradually understood as g based on attitudinal selectivity. This concept highlights two main intervening factors: the interpretation of an effect at one of the so-called six levels of information processing (McGuire, 1968), and the influence cognitive structures or heuristics on information processing (Tversky&Kahnem,1974).在这里,偏向性的观念非常重要。它最初被理解为对(政治)行为的潜在决定因素,之后又渐渐被当成实际观念的障碍或者过滤器(Schmitt-Beck, 2000)。此外,与劝说过程中的实际行为一样,信息处理成为了关注的焦点和细节理论研究的客体。结果,媒体的影响被逐渐理解成为基于态度选择的基础。这个观念强调了两个主要的干扰因素:对所谓六个级别的信息处理的中断反应(McGuire, 1968),以及认知结构或启发模式对信息处理的影响(Tversky&Kahnem,1974).Predispositions can be defined with regard to social psychology as well as sociology. In the first case, party identification typically acts as a filter; in the second case, this role is assigned to lines of conflict within society. This means that both individuals and issues can be predisposed. Concerning individual predisposition, Ceci and Kain (1982) made an important initiating contribution when they demonstrated that poll-induced effects occur most frequently among the undecided voters. Joslyn (1997) has supported this finding while looking at the intersection of predisposition and public opinion context and differentiating not only between assimilation (for the undecided voters) but also reinforcement and contrast. Furthermore, the concept of predispositions has been shown to be applicable to issues as well. The studies by Hardmeier and Roth (2003), Mutz (1992), and Kaplowitz, Fink, DAlessio and Armstrong (1983) show that published poll results have a stronger impact when people have weak or no predispositions towards the issue at hand.偏向性可以定义为社会心理学或者社会学。首先,政党认同典型地充当过滤器的作用;其次,这个角色被分配进了社会冲突线路。这就意味着两个个体和议题都应该预先处理。关于个人倾向,Ceci 和 Kain (1982)做出了重要的初始贡献,当时他们证明了投票诱导效应最易发生在犹豫不决的选民中。Joslyn(1997)在倾向性和舆论背景交集之时就支持这一调查结果,不仅区分了同化(对于犹豫不决的选民)而且区分了增援和对照。此外,人们证实了倾向性的观念也适用于议题。Hardmeier and Roth (2003), Mutz (1992), and Kaplowitz, Fink, DAlessio and Armstrong (1983)的研究表明,公开民意调查的结果在人们对于手边的问题倾向性很微弱或者没有倾向时,影响最大。On the other side, empirical findings concerning reactions to poll data when people or issues are highly predisposed are less clear-cut, and studies found effects in all directions. Along with zero or reinforcing effects, tendencies away from the poll stimulus have been noticed by Joslyn (1997) or Ceci and Kain (1982) at the individual level, and by West (1991) or Hardmeier and Roth 2003 (see Figure 47.1) at the level of issues. Ceci and Kain interpret this as oppositional reactivity. West ascertains an anti-govemment reflex, and the data reported by Joslyn or Hardmeier and Roth point to a ceiling effect.另一方面,当人们或者是议题具有高度倾向性时,有关民调数据反应的经验结果就没那么清楚了,而且研究发现,影响是多方面的。随着零反应或加强反应作用,远离投票刺激的倾向已经被Joslyn (1997) 以及 Ceci and Kain (1982)在个体层面 , West (1991) ,Hardmeier 以及 Roth 2003 (see Figure 47.1)在议题水平上察觉了。Ceci and Kain (1982)将这个理解为对抗反应性。西方认为这是一个反政府反应,Joslyn和Hardmeier 以及 Roth所报道的数据暗示了天花板效应。This allows establishing another superordinate finding: especially when predisposed subjects are confronted with survey results, their reactions are anything but mechanical. A stimulus does not produce a typical response. The effects thus defy manipulative and deliberate intervention (which is a positive aspect), but also currently resist theorizing. While the theoretical arguments for reactions in the direction of the proposed stimulus, that is, a bandwagon effect, seem rather sophisticated (as above), arguments for the counter-reaction, or even an underdog effect, are more frequently developed ad hoc, and with a less well understood theoretical basis. Thus, the simple impact assumption has received yet another differentiated answer.这样,就确立了另一个更高级的发现:尤其当倾向主体面对调查结果时,他们的反应通常很呆板。一个刺激物并不能造成一个特有的结果。这个效应因此不仅能巧妙反抗或者是有意介入的干扰(这是一个积极地方面),而且能够抵制理论化。然而在提出的刺激反应的理论探讨,也就是说从众效应看起来相当复杂,并且对反作用的探讨异常频繁,甚至哀兵效应的发展都异常迅速的情况下,发展者对理论的理解却缺乏深度。因此,这个简单的影响假设又有了另一个不同的答案。However, the idea that even issues are predisposed has more scope for development. The fact that some issues can trigger resistance or counter-reactions suggests that this is linked to the nature of the messages conveyed to the public (Zaller, 1992). Evidently there are issues where people lack the cueing messages necessary to identify the direction of impact and the political implications, is was obviously the case with the issue of a revision of the tourism tax rate as shown in Figure 47.1. At the time of the study,
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 年月日动态课件
- 年度安全培训考评报告课件
- 年度全员安全培训
- 娱乐主播培训课件教学
- 工业投入产出联系课件
- 兰州事业单位笔试真题2025
- 2024年上饶市机关事业单位招聘考试真题
- 2025年甘肃省公务员考试真题
- 物业公司总经理经营责任合同5篇
- 方孔网交易合同6篇
- 新学期教学工作会议上校长讲话:把功夫下在课堂里把心思放在学生上把质量落到细节中
- 医疗机构基孔肯雅热防控卫生监督检查表
- 2025年全国青少禁毒知识竞赛题库(答案+解析)
- 快消品包装行业可持续性发展报告2025:包装印刷行业绿色转型
- 信鸽裁判证管理办法
- 抑郁症病例分析报告
- 痛风性关节炎鉴别
- 《老年冠心病慢病管理指南(2024版)》解读
- 会计信息系统应用 课件 项目三 总账管理系统
- 2025年河北大学版(2024)小学信息科技三年级(全一册)教学设计(附目录 P179)
- 2025至2030全球及中国工业I和和O模块行业发展趋势分析与未来投资战略咨询研究报告
评论
0/150
提交评论