ABSTRACT

In this article we examine PolitiFact’s fact-checking process. We collect a random sample of 858 fact-checks and evaluate them in the light of criteria based on or inspired by fact-checking literature and the International Fact-checking Network’s code of principles. Our analysis reveals the following: in general, PolitiFact fared well. However, from the point of view of the criteria, its practices leave room for improvement. The biggest issue is complex propositions. These are statements containing multiple claims, i.e., more than one proposition. In 279 cases (33% of our sample), PolitiFact checks a complex proposition and assigns one truth rating to it. This is problematic as the reader might misinterpret the truthfulness of an individual claim. PolitiFact also checks claims that we considered uncheckable. These are statements whose truthfulness cannot be defined in practice, e.g., claims about the future and vague claims. In 92 cases (11% of our sample), PolitiFact checked a claim like this. The article ends with a discussion about the limitations of the criteria used here.

摘要

在本文中，我们研究了PolitiFact的事实检查过程。我们收集了858个事实检查的随机样本，并根据基于事实检查文献和国际事实检查网络原则准则或受其启发的标准对它们进行了评估。我们的分析揭示了以下内容：总体而言，PoliitiFact表现良好。但是，从标准的角度来看，其实践留有改进的空间。最大的问题是复杂的主张。这些是包含多个主张的陈述，即，多个主张。在279个案例中（占我们样本的33％），PoliitiFact会检查一个复杂的命题，并为其分配一个真实等级。这是有问题的，因为读者可能会误解单个主张的真实性。PolitiFact还会检查我们认为不可检查的索赔。这些是在实践中无法确定其真实性的陈述，例如，关于未来的主张和模糊的主张。在92个案例中（占我们样本的11％），PoliitiFact检查了此类索赔。本文最后讨论了此处使用的标准的局限性。