People as sensors: mass media and local temperature influence climate change discussion on Twitter

Abstract:

We examined whether people living under significant temperature anomalies connect their sensory experiences to climate change and the role that media plays in this process. We used Twitter messages containing words “climate change” and “global warming” as the indicator of attention that public pays to the issue. Specifically, the goals were: (1) to investigate whether people immediately notice significant local weather anomalies and connect them to climate change and (2) to examine the role of mass media in this process. Over 2 million tweets were collected for a two-year period (2012 – 2013) and were assigned to 157 urban areas in the continental USA (Figure 1). Geographical locations of the tweets were identified with a geolocation resolving algorithm based the profile of the users. Daily number of tweets (tweeting rate) was computed for 157 conterminous USA urban areas and adjusted for data acquisition errors. The USHCN daily minimum and maximum temperatures were obtained for the station locations closest to the centers of the urban areas and the 1981-2010 30-year temperature mean and standard deviation were used as the climate normals. For the analysis, we computed the following indices for each day of 2012 – 2013 period: standardized temperature anomaly, absolute standardized temperature anomaly, and extreme cold and hot temperature anomalies for each urban zone. The extreme cold and hot temperature anomalies were then transformed into country-level values that represent the number of people living in extreme temperature conditions.

The rate of tweeting on climate change was regressed on the time variables, number of climate change publications in the mass media, and temperature. In the majority of regression models, the mass media and temperature variables were significant at the p<0.001 level. Additionally, we did not find convincing evidence that the media acts as a mediator in the relationship between local weather and climate change discourse intensity. Our analysis of Twitter data confirmed that the public is able to recognize extreme temperature anomalies and connects these anomalies to climate change. Finally, we demonstrated the utility of social network data for research on public climate change perception.