It depends: Partisan evaluation of conditional probability importance
It depends: Partisan evaluation of conditional probability importance
复制标题
DOI：
10.1016/j.cognition.2019.01.020
复制
发表时间：
2019-07
期刊：
影响因子：
3.4
通讯作者：
Leaf Van Boven;Jairo Ramos;Ronit Montal-Rosenberg;Tehila Kogut;D. Sherman;P. Slovic
中科院分区：
文献类型：
--
作者：
Leaf Van Boven;Jairo Ramos;Ronit Montal-Rosenberg;Tehila Kogut;D. Sherman;P. Slovic
压制恐怖主义等罕见事件的政策往往会限制穆斯林移民等同时发生的类别。评估限制性政策需要清楚地思考条件概率。例如，恐怖主义极为罕见。因此，即使大多数恐怖分子移民是穆斯林——“命中率”很高——穆斯林移民成为恐怖分子的逆条件概率也极低。然而，逆条件概率更适合评估限制性政策，例如限制穆斯林移民时的恐怖主义威胁。我们建议人们对条件概率进行党派评估，认为当他们支持政治规定的限制性政策时，命中率更为重要。在两项研究中，驱逐以色列特拉维夫寻求庇护者、禁止穆斯林移民和前往美国旅行以及禁止攻击性武器的支持者认为“命中率”概率（例如，恐怖分子是穆斯林）比政策反对者更重要，政策反对者认为相反的条件概率（例如，穆斯林是恐怖分子）更重要。这些党派分歧涵盖了右翼和共和党（驱逐寻求庇护者和禁止穆斯林旅行）以及民主党（禁止攻击性武器）所青睐的限制性政策。邀请党派人士采用公正的专家观点在一定程度上减少了这些党派分歧。在研究 2（但不是研究 1）中，计算能力更强的党派之间的党派差异更大，这表明计算能力支持动机推理。这些发现对两极分化、政治判断和政策评估具有影响。即使党派就统计事实达成一致，但他们对这些统计事实的相关性却存在明显分歧。
Policies to suppress rare events such as terrorism often restrict co-occurring categories such as Muslim immigration. Evaluating restrictive policies requires clear thinking about conditional probabilities. For example, terrorism is extremely rare. So even if most terrorist immigrants are Muslim—a high “hit rate”—the inverse conditional probability of Muslim immigrants being terrorists is extremely low. Yet the inverse conditional probability is more relevant to evaluating restrictive policies such as the threat of terrorism if Muslim immigration were restricted. We suggest that people engage in partisan evaluation of conditional probabilities, judging hit rates as more important when they support politically prescribed restrictive policies. In two studies, supporters of expelling asylum seekers from Tel Aviv, Israel, of banning Muslim immigration and travel to the United States, and of banning assault weapons judged “hit rate” probabilities (e.g., that terrorists are Muslims) as more important than did policy opponents, who judged the inverse conditional probabilities (e.g., that Muslims are terrorists) as more important. These partisan differences spanned restrictive policies favored by Rightists and Republicans (expelling asylum seekers and banning Muslim travel) and by Democrats (banning assault weapons). Inviting partisans to adopt an unbiased expert’s perspective partially reduced these partisan differences. In Study 2 (but not Study 1), partisan differences were larger among more numerate partisans, suggesting that numeracy supported motivated reasoning. These findings have implications for polarization, political judgment, and policy evaluation. Even when partisans agree about what the statistical facts are, they markedly disagree about the relevance of those statistical facts.