Climate Change Case Study | Moka

Climate Change
Case Study

Explore how communities on Twitter talk about climate change.

*Best viewed in Chrome on Desktop.

Created by
Katya Obyedkova
Karen Hao

“Climate change does not respect border; it does not respect who you are - rich and poor, small and big. Therefore, this is what we call ‘global challenges’, which require global solidarity.”  Ban Ki-moon

The US is the only developed nation that still faces widespread climate denial. An October 2016 Pew Research study showed that overall, only 48% of American adults agree that climate change is human-made. Public literacy around climate change is also highly community dependent. The same study found heavy political divides for nearly all of the facts and policy proposals that they surveyed for about climate change. More recently, a March 2017 New York Times article also showed that opinions on climate change differ by region.

Despite being aware of our nation’s highly divided discourse, we don’t have a nuanced understanding of how different communities talk about climate change. How often do they broach the subject, if at all? Which articles resonate across different communities? Which garner a polarized response?

We decided to explore these questions with Moka, a tool for analyzing community-based discussions on Twitter. We focused on comparing seven communities: Democrats and Republicans, coastal cities and coal-mining states, libertarian think tanks and conservative think tanks, and climate change activists. Using Moka, we analyzed all of the tweets within each community between August 1 and August 9 that shared an article about or related to climate change.

How do we define these communities?
Hover over the circles to learn what parameters we use for streaming tweets for each of the communities.
SomeTextSome text
While some communities are defined by location, some are defined by a network of people.

We chose to examine these seven communities to cover a wide range of political actors and perspectives. Based on our understanding of how opinions about climate change differ by political preference, we wanted to juxtapose the discussions happening among Democrats and Republicans—two communities composed of its respective members of Congress. Similarly, we sought to juxtapose coastal cities and coal-mining states—rough location-based proxies for the Democratic and Republican parties’ constituents.

The decision to examine climate change activists, conservative think tanks, and libertarian think tanks—all communities that generate research and engage policymakers—was to tease out how influential each of them are in shaping the national discourse.

What are these communities saying about climate change?
Click on the circles to see the top 8 articles for each of the communities. Hover over the comment bubble to see what people say.
Some text
Note that if one Twitter account posts the same article multiple times we count it as one post.

Over our data collection period, all seven communities shared climate change articles a total of 2,774 times. Whereas coastal cities and Democrats tweeted more often about the reversal of climate change policies and climate “doom” scenarios, conservative and libertarian think tanks and Republicans focused more on criticisms of advancing the climate agenda. Coal-mining states seemed to fall somewhere in the middle, sharing a mix of criticism against the Trump administration for its climate denial and against environmentalists for their alarmism.

Within the climate change activist and left-leaning communities, a significant portion of the conversation was also tied up on emphasizing that climate change is real and will have disturbing repercussions. Most of these articles, however, did not reach right-leaning audiences. There was also a distinct lack of deeper discussions around policy proposals, technology solutions, or other action plans. This was particularly marked for the two policymaking communities, Democrats and Republicans, which instead focused their tweets on partisan politics.

Interestingly, the draft of the White House climate report and the corresponding New York Times article were shared widely in both coastal and coal-state communities, but not among Democrats or Republicans.

How does the media landscape vary across communities?
Hover over bar segments to see article titles and how many times they were shared. Click through the menu bar below to see different communties.
Some text
Some text
The bar chart shows up to 40 top publications.

By looking at which publications were tweeted the most, we can observe trends about which media brands have established the most trust within each community. (Note we chose not to visualize Republicans, conservative think tanks, or libertarian think tanks because they shared only two, four, and two articles, respectively.)

For climate change activists, coastal cities, and coal-mining states, a select few media brands dominated the discussion, though many more pitched in. Climate change activists, for example, cited 315 unique news sources, the top five of which accounted for 30% of all shares. Coastal cities cited 303 sources with the top five covering 37% of all shares; coal-mining states cited 210 with the top five covering 33%.

Among the long tail of cited publications, climate change activists shared more brands focused specifically on the environment, such as Climate Central, InsideClimate News, Grist, and Yale E360. Coastal cities and coal-mining states, on the other hand, shared more general media brands, such as the Washington Post, CNN, The Hill, and Huffington Post. This suggests that the public typically receives its climate change news from the sources they already consume.

Notably, we also found a stark difference in the volume of climate change articles shared among Democrats versus Republicans. Given that both communities contain roughly the same number of accounts, it becomes abundantly clear that climate change is not top of mind for Republican policymakers.

How often do these communities share the same content?
View the overlap between the top 10 articles.
Unexpectedly, coal-mining states and coastal cities had a larger overlap in the top 10 articles shared than climate change activists and Democrats.

By looking at the overlap between different communities, we can begin to infer what types of content resonates across disparate audiences. Across coastal cities and coal-mining states, for example, three of the six overlapping articles were reporting on the Trump administration; two were related to jobs and economics. In contrast, climate change activists and Democrats overlapped on two articles criticizing the Trump administration and another two articles focused on the harmful impacts of climate change.

What can we learn from our analysis?

From studying seven Twitter communities across nine days, our analysis is by no means complete, but it provides a starting point for developing a more nuanced understanding of the national climate change discourse.

In addition to confirming the wide disparities across how different communities discuss and share news about climate change, we also surfaced fresh insights about the state of our national discourse. Perhaps our most striking find was that by focusing predominantly on whether or not climate change is real, policymakers and climate change communicators are missing an opportunity to engender more sophisticated dialogue. The resulting naive and muddled public comprehension of climate change played no small part in the election of a president and Congress that have demonstrated poor conviction for advancing a national climate agenda.

Despite these challenges, however, there is good news. A May 2017 study on climate change communication conducted by George Mason University and Yale University found that 52% of registered voters think climate change should be a high or very high priority for the administration and 68% think the US should reduce its greenhouse gas emissions regardless of what other countries do.

“Millions of Americans are willing to work together to demand climate action by the government and companies,” George Mason University concluded. “But this potential mass climate movement remains largely unorganized, with many people sitting on the sidelines waiting to be engaged.”

Methodology: We use Twitter's API to stream tweets from the selected communities. Some communities are defined through a list of Twitter accounts and some are defined by location. If a community is defined by accounts, we stream all their tweets and filter for the ones that link to an article and have climate change related content. For location based communities, we stream tweets that originate from specific locations defined by geographical bounding boxes and then filter for links to articles and climate change related content. To identify whether content is climate change related, we check for a set of keywords in a tweet’s text field, hashtags, and extended url.

The top 8 articles in each community: We calculate how many times a link was posted in a particular community, however, the same articles are often posted under different links. To solve this problem, first we extract the canonical url for each link to use for further analysis. If one account tweeted a link multiple times, we still count it as 1 post. We relied on declared html meta tags to extract the canonical url, publication, title, and image. In the rare instances where this information was incomplete or missing, we added it manually.

Media Landscape: For each publication we count how many articles it published and for each article how many times it was shared. Again, if one account tweeted a link multiple times, it is still counted as 1 post. We display up to the 40 most popular publications.