A computer analysis of 20 years of data found that corporate funding influenced both the content and specific language used to encourage public skepticism of climate change.
Writing in the Nov. 23 online edition of the Proceedings of the National Academy of Sciences, Yale University sociologist Justin Farrell said the analysis offers a new level of understanding about the role of corporate money in polarizing a public discussion. Farrell is an assistant professor of sociology at the Yale School of Forestry and Environmental Studies.
Farrell’s analysis is based on two data sets. The first is a social network of 4,556 individuals with ties to 164 organizations that have been skeptical of climate change. The second data set is a compendium of every text those organizations produced about climate change from 1993 to 2013. The latter includes 40,785 texts such as policy statements, press releases, website articles, conference transcripts, published papers, and blog articles.
A good deal of research already exists about how polarization affects individual attitudes and behavior relating to climate change, Farrell explained. Yet there is little data about the underlying organizational factors at play.
“We’ve never had this level of data,” Farrell said. “The text analysis is entirely computational, and it shows an ecosystem of influence.”
The study mentions several themes that became more dominant among climate change “contrarians” that received corporate funding. One was the idea that climate change is cyclical in nature; another prevalent theme was the positive benefits of carbon dioxide.
“They were writing things that were different from the contrarian organizations that did not receive corporate funding,” Farrell said. “Over time, it brought them into a more cohesive social movement and aligned their messages.”
The study employed a recently developed approach to topic modeling — a computer-assisted content analysis process —that factors in metadata such as the year a text was written and its link to corporate funding. The process enables reliable content analysis on massive collections of text, with topics emerging from the data as algorithms learn the hidden patterns within a database.
(Image courtesy of Shutterstock)