1 of 12

Tweeting (or Tweaking)

Climate Change:

A Research Blueprint

September 2016

Tommy Xie

2 of 12

Background

Earlier research

Follow me at

@tommiexie

www.tommyxie.com

3 of 12

Social Construction of Climate Change

Focus

Climate change media coverage
Framing theory
Prestige newspapers and blogs
U.S. and Chinese media
Longitudinal design

Full journal article at

goo.gl/1qMqcq

Key findings

Newspaper’s major shift in reporting in late 2000s
Bloggers much more skeptical than newspapers
Almost zero skepticism in Chinese media
Who’s responsible: U.S. Government

4 of 12

But things have changed

Traditional & Grassroot Social Media Symbiotic Relationship

5 of 12

New Focus

Twitter

500M/DAY (200+GB)

Key platform for

news distribution
political campaign
social movement

Moments

Social media rely on moments

Spontaneous
Respond to internal/external moments
Change in sampling paradigm

Climate Summit 2014 COP21

6 of 12

Why these moments

Climate Summit 2014 (NYC, Sep/2014) and COP21 (Paris, Nov-Dec/2015) are events in a sequence held by the UN.
Climate Summit 2014, which ignited heated discourse on social media, was a preamble to COP21.
COP21 aims to, for the first time in over 20 years of UN negotiations, a binding and universal agreement on climate, from all the nations of the world.

7 of 12

Research Questions

Network structure

Who are the voices (identity)?
How are they interconnected (identity clusters)?
How are they geographically mapped over time?

Narrative

What are the major frames (topic salience) in the discourse?
How are the frames paired (e.g. immigration & national security) to mold public perception
How is the content sourced (news media/NGO/pundits...)

8 of 12

Research Questions: Two Analytical Dimensions

Moments

CS14
COP21

Narrative

What are the major frames (topic salience) in the discourse?
How are the frames paired (e.g. immigration & national security) to mold public perception
How is the content sourced (news media/NGO/pundits...)

Sentiment groups

Pro
Against
Neutral

9 of 12

Method

Data Capture

TwitterR and StreamR packages for R

Twitter R for querying past tweets (CS14)
Stream R for capturing ongoing tweets (COP21)

Three weeks: estimate 15, 000, 000+ tweets

Analysis

WordStat for semantic analysis for frames
sna package in R for network analysis
Machine-learning (naive Bayes classifier in LightSIDE) algorithm for sentiment analysis
CartoDB for geo-spatial mapping