Political beliefs and behaviors are predominantly shaped by the torrent of linguistic communication surrounding us every day, but existing models only capture tiny fractions of this complex process. This work develops new techniques in automated text analysis and causal inference to model the elaborate interplay between speech, belief, and behavior, revealing how language reflects ideology, affects vote intention, and shapes long-term opinion change.
Part 1 develops a new method for measuring individual ideology, showing that in legislatures, speech data can be used to predict roll-call votes, to infer ideology even when the voting data are uninformative, and to detect—in ways the roll-call cannot—how changes in leadership affect the agenda.
Part 2 turns to ideological change, examining political advertisements to reveal the complex ways speech affects belief and behavior. Rather than testing a single theory of persuasion at a time, a new bottom-up approach is developed that estimates the myriad simultaneous effects of hundreds of different presidential campaign ads in 2004. The effective underlying themes and strategies are then inferred using new text analysis methods, which are shown to be able to predict the persuasive effects of ads based only on their text.
Part 3 turns to the vast bulk of linguistic behavior, interpersonal communication. Examining millions of online discussions, Part 3 develops a new model of political psychology that posits an interconnected network of ideas, and hypothesizes that argument consists in exchanging ideas, facts, and topics that are relevant to, but missing from, what one's interlocutor has previously said. A Bayesian topic model is adapted to infer this conceptual network, and can predict which topics discussants will deploy in response to each other. Panel vector autoregression methods are then used to infer how these arguments affect online voting behavior and long-term opinion change.
In toto, this work shows how the complex interplay between speech and belief can be modeled using new techniques in text analysis in ways heretofore impossible. It develops a new model of political psychology and language that is predictively useful, and should be broadly applicable to domains such as legislatures, campaigns, and online.
|Advisor:||Nagler, Jonathan, Laver, Michael|
|Commitee:||Beck, Nathaniel, Dawes, Christopher, Egan, Patrick, Rosenthal, Howard|
|School:||New York University|
|School Location:||United States -- New York|
|Source:||DAI-A 74/04(E), Dissertation Abstracts International|
|Keywords:||Campaigns, Machine learning, Political behavior, Political communication, Public opinion, Text analysis|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be