A Caltech Library Service

New Perspectives in Political Communication


Adams-Cohen, Nicholas Joseph (2019) New Perspectives in Political Communication. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/7TDG-4R42.


This dissertation contains three chapters exploring the nature of political communication and public opinion formation by analyzing social media data. Each chapter uses original sets of Twitter data to examine the public’s response to major shifts in public policy (Chapter Two), the differences between partisan networks (Chapter Three), and how citizens engage with gun policy after mass shootings (Chapter Four).

Chapter Two examines how public opinion towards gay marriage changed before and after the legalization of same-sex marriage as a result of the 2016 Obergefell v. Hodges Supreme Court decision. Exploiting the variation in state law prior to the Court’s decision, I use a difference-in-difference approach to find causal evidence that citizens residing in states where the Supreme Court overturns state laws are more likely to have a negative opinion of the federal decision.

In Chapter Three, I collect an original dataset of Twitter conversations about the American political parties to develop a supervised learning algorithm that classifies users as liberal or conservative, using these labels to then map out separate ideological network structures. Analyzing these networks, I find significant differences in how conservative and liberal citizens form online networks, leading to important consequences for information diffusion and action coordination.

In Chapter Four, I examine how messages from the political and media elite concerning gun control impact citizen engagement with gun policy issues in the wake of high-profile mass shootings. I analyze the impact of elite messaging with a panel data set of sixty thousand partisan Twitter users, data that includes each user’s full Twitter history as well as information on which accounts they follow. By building this Twitter panel, I am able to better determine which elite messages each user receives and whether the recipient chooses to engage with gun policy. I find that elite messages increase the likelihood a user will engage with gun policy issues, but further determine that we must broaden the notion of elite to include users only considered influential on the Twitter platform.

Item Type:Thesis (Dissertation (Ph.D.))
Subject Keywords:Public opinion, parties, text analysis, network analysis, machine learning, causal inference
Degree Grantor:California Institute of Technology
Division:Humanities and Social Sciences
Major Option:Social Science
Thesis Availability:Public (worldwide access)
Research Advisor(s):
  • Alvarez, R. Michael
Thesis Committee:
  • Katz, Jonathan N. (chair)
  • Rosenthal, Jean-Laurent
  • Agranov, Marina
  • Alvarez, R. Michael
Defense Date:25 April 2019
Non-Caltech Author Email:njadamscohen (AT)
Record Number:CaltechTHESIS:06102019-125021538
Persistent URL:
Adams-Cohen, Nicholas Joseph0000-0003-2251-1744
Default Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:11734
Deposited By: Nicholas Adams-Cohen
Deposited On:10 Jun 2019 23:07
Last Modified:04 Jun 2024 21:38

Thesis Files

PDF - Final Version
See Usage Policy.


Repository Staff Only: item control page