Scholars increasingly rely on online platforms to extract latent political concepts from texts. This paper introduces a novel approach—crowdsourced pairwise comparisons—to validate human coding of political texts.
Data & Methods: We test the framework using U.S. Senate campaign ads and State Department human rights reports, employing free software for aggregation.
Key Findings: The method effectively combines coder intuition with computational reliability, addressing concerns about non-expert biases in previous studies.
Why It Matters: Our open-source tool enhances accessibility to text analysis techniques across political research applications.
| A Pairwise Comparison Framework for Fast, Flexible, and Reliable Human Coding of Political Texts was authored by David Carlson and Jacob M. Montgomery. It was published by Cambridge in APSR in 2017. |
