Closed divilian closed 3 years ago
Conover- Political Polarization on Twitter create a corpus of political communication- a tweet containing at least one politically relevant hashtag cluster analysis of retweet and mention networks manually annotated users political identity retweet and mention networks homogeneous vs heterogeneous content
Laver- Extracting Policy Positions from Political Texts Using Words as Data reference texts with known positions, generate word scores from reference texts, score new text using the word scores
Lowe- Scaling Policy Preferences from Coded Political Texts propose an alternate approach to estimating policy positions based on the logarithm of odds ratios
Slapin & Proksch -A Scaling Model for Estimating Time-Series Party Positions from Texts scaling algorithm called WORDFISH- estimate policy positions based on word frequencies in text
Monroe- Fightin' Words: Lexical Feature Selection and Evaluation for Identifying the Content of Political Conflict Dirichlet prior Laplace prior p.396-398 political polarization
Lowe- Understanding Wordscores Given R documents or ‘‘reference texts’’ with known positions or scores on a policy dimension, Wordscores attempts to estimate the scores of L out-of-sample documents, the ‘‘virgin texts.’’ To do so the method first estimates scores for each word type occurring in the reference texts and then combines these wordscores into a score for each virgin document. It is important to distinguish the two parts: estimating wordscores and estimating document scores using wordscores because they are, at least in principle, independent parts of the method. There is usually a third and final part of the method that rescales virgin document score estimates, so they can be more easily compared with the reference text scores.
Add to Zotero whatever looks worth adding.