Local temperature anomalies increase climate policy interest and support: Analysis of internet searches and US congressional vote shares

Highlights

• Temperature anomalies are associated with more internet searches about climate change.

• Temperature anomalies are associated with fewer votes for Republican congressional candidates.

• Effects of temperature anomalies were strongest in Republican leaning districts.

Abstract

Two studies examine the effects of temperature anomalies (relative to ten-year averages) on interest in and support for climate policies. Study 1 analyzes the impacts of local temperature anomalies on information acquisition, namely Google searches, about climate change and climate policies. We find strong evidence that temperature anomalies are associated with increased climate change information acquisition. Our results show that deviations from seasonal norms in both directions (i.e., temperatures higher or lower than expected) predict increased interest. Study 2 analyzes voting for Republican candidates, who in the timeframe of our data were not likely to support climate policies. Analysis of voting records from ten US midterm elections from 2002 to 2020 shows that greater local temperature anomalies are significantly associated with lower vote shares for Republican candidates.

Introduction

Understanding how anomalous weather events affect climate policy support can help communicators to strategically leverage weather events to promote citizens’ attention to and support for climate change policies. Notable weather events may help open policy windows that can be anticipated. Knowledge of how weather experiences affect climate attitudes and behaviors also aids us in generating long-term predictions about how climate policy support will evolve over time in response to changes in weather patterns brought about by climate change (Moore et al., 2019, Ricke and Caldeira, 2014).

Climate policy support is an important pro-environmental behavior because enacting climate policies is likely necessary to mitigate climate change (Pachauri et al., 2014). To date, much research has examined the effects of local weather events (WE) on climate attitudes, but fewer studies have examined their effects on climate behaviors or on climate policy support (Sisco, 2021). In the current study, we address this gap by presenting two studies that analyze how WE impact climate policy interest and support in the public.

Climate attitudes. Past research on the effects of WE on climate attitudes has delivered mixed results (see Howe, 2021, Howe et al., 2019, Sisco, 2021 for reviews). A meta-analysis (of 171 academic studies and 25 polls) that compares predictors of belief in climate change finds that self-reported experience with local changes in weather is one of the strongest predictors of belief in climate change (Hornsey et al., 2016). The effect is similar in size to effects of variables such as trust in scientists and perceived scientific consensus. Citizens have some metacognitive insight about this influence on their climate beliefs. In a representative survey of US citizens conducted in 2018, participants were asked how much different factors affected their views on climate science (EPIC/AP-NORC, 2018). Out of seven potential influences including sources of information like news stories and views of political leaders, participants most commonly reported recent extreme weather events and personal observations of weather in their local areas as their top influences.

Beyond meta-analyses and polls, several studies have sought to examine the effects of WE more directly by examining the associations between objectively measured weather and climate attitudes. When people experience temperatures that are warmer than usual, and perceive them as such, they are more likely to report concern about global warming (Li et al., 2011), a result known as the local warming effect (Zaval et al., 2014). Numerous other studies find that local temperature anomalies increase people’s concerns about climate change and attention to it, at least temporarily (Egan and Mullin, 2012, Hamilton and Stampone, 2013, Joireman et al., 2010, Kirilenko et al., 2015, Lang, 2014, Myers et al., 2013, Sisco et al., 2017). A re-analysis of about 400,000 responses across 170 polls finds that self-reported climate attitudes are modestly responsive to changes in state-level annual temperatures (Bergquist and Warshaw, 2019). Approaching this question with a laboratory experiment, Joireman et al. (2010) demonstrated that experimentally priming participants with heat-related cognitions increased beliefs in climate change.

Climate behaviors. A small but growing literature examines the effects of WE by observing behavioral outcomes instead of relying of self-reports of behaviors or behavioral intentions which can be prone to bias. Zaval et al. (2014) show that abnormal weather affected real monetary donations to climate charities in an online experiment. These results are suggestive, but bear the limitation that these behaviors were observed within a controlled experimental setting with generalizability to real-world settings somewhat unclear.

Some researchers have reported effects of WE on social media messages about climate change (Kirilenko et al., 2015, Mumenthaler et al., 2021, Sisco et al., 2017). Citizens’ online posts about climate change are observable behavioral outcomes, however, these behaviors may have minimal impacts on climate action. This is in part due to the presence of echo chambers in social media networks which can limit the audience of positive messages about climate change to those who already endorse them (Pearce et al., 2014).

Other papers have analyzed searches about climate change on the internet as observed behaviors (Choi et al., 2020, Lang, 2014, Sisco et al., 2021). Herrnstadt and Muehlegger report that climate-related searches increase with extreme temperatures (Herrnstadt and Muehlegger, 2014). Lang (2014) finds that monthly temperature fluctuations predicted US searches about “climate change” and “global warming” from 2004 through 2013. Lang observed that monthly search activity increased with extreme summer heat, but also with colder winter and spring temperatures. Choi et al. (2020) examined monthly search activity in 72 international cities. Choi et al. report that searches about “global warming” increased when local temperatures were abnormally high. In the current paper we attempt to replicate these results. We also extend them by examining whether WE influence searches beyond the keywords “climate change” and “global warming” to searches that are more specific to climate change policies. We use a high temporal resolution of weekly averages. In Table 1 we present a comparison of our internet search study and past studies on the effects of temperature on internet searches.

Others studies have not detected evidence of WE’s impacts on climate change attitudes and behaviors (for example, Brulle et al., 2012, Shum, 2012). Scholars have reasoned that the mixed results may be due to the wide heterogeneity of operational definitions of WE, heterogenous outcome variables, and varied methodological approaches that have been employed as well as varying amounts of media and public attention to the different weather phenomena studied (Howe, 2021, Howe et al., 2019, Sisco, 2021).

Climate policy support. While there has been much past research on the effects of WE on climate attitudes and to some extent behaviors, only a small subset of work has examined WE’s effects on climate policy support. Rudman et al. (2013) examined the impacts of Hurricane Irene and Hurricane Sandy on attitudes toward a (real) green politician. They report that, before experiencing the hurricanes, New Jersey residents had negative implicit attitudes toward the green politician. After experiencing the hurricanes, residents showed more positive implicit preferences for the candidate. Bergquist et al. (2019) find that participants living in Florida reported more willingness to pay higher taxes to protect the environment after Hurricane Irma compared to directly before. Similarly, Demski et al. (2017) report that flood experiences increased reported support for climate mitigation policies in a sample of UK residents.

One highly consequential behavior that is a direct measurement of climate policy support is voting behavior. Hoffmann et al. (2022) show a significant effect of temperature anomalies on votes for Green parties in Europe. Hazlett and Mildenberger (2020) examine the impacts of experiences with wildfires on votes for climate policy measures. They find that proximity to wildfires bolstered support by five to six percent for those living in Democratic-voting areas, with the effect diminished to near zero in areas farther than 15 km away from the fires. Gagliarducci et al. (2019) report positive impacts of local hurricane disasters from 1989 through 2014 on US congress members’ support for environmental policies. Similarly, Herrnstadt and Muehlegger (2014) find that congressional members are more likely to cast pro-environmental votes after their home states experienced abnormal weather. Relatedly, Liao and Junco (2021) find that financial contributions to the Democratic Party increase in response to higher weekly temperatures.

Theoretical mechanisms. Some past work has postulated possible mechanisms by which WE can affect attitudes and behaviors (Giordono, Gard-Murray and Boudet, 2021). Sisco (2021) used a literature review to identify three theoretical mechanisms: affect activation, issue salience, and psychological distance. That is, WE may affect climate attitudes/behaviors by evoking strong emotional responses, by increasing the salience of climate change as a threat, and by decreasing the felt psychological distance from the risks of climate change (Zanocco et al., 2019). Brügger et al. (2021) introduced a conceptual framework organizing relevant psychological theories into three clusters of processes by which WE can influence climate attitudes/behaviors: noticing and remembering a WE, mental representations of the WE and its connection to climate change, and effects of a WE on risk perceptions and decision-making (Brügger et al., 2021).

Overview of the current work. We examine the effects of WE, specifically temperature anomalies, on behavioral measures of interest and support for climate policies. Climate policy support is arguably-one of the most important climate mitigation behaviors in part because it allows one individual to affect the behaviors of others. For example, if a slim majority votes to enact a mandatory recycling policy, then all citizens are required to recycle. A challenge with many personal green behaviors, such as recycling or using public transportation, is that they need to be performed repeatedly over time. In contrast, supporting a climate policy or a candidate committed to climate policies can be done with a highly consequential one-time action of casting a vote.

There are many different manifestations of WE such as temperature anomalies, droughts, hurricanes, flooding, and so on. In the current paper we focus on anomalous temperatures for several reasons. As described above, temperature anomalies have been shown to affect climate attitudes and behaviors in a variety of past studies. Moreover, changes in temperature are some of the most direct consequences of climate change. Given that climate change is sometimes referred to as global warming, increases in temperature are likely still strongly associated with climate change in the public mind (Schuldt et al., 2011).

In the current work, we investigate whether local anomalous temperatures lead to increased climate policy interest and support. We report on two studies that approach this research question with different datasets, levels of analysis, and outcome variables. Study 1 analyzes information acquisition, showing that abnormal temperatures predict increased Google searches about climate change as well as search terms more directly related to climate policies (e.g., “renewable energy”). Study 1 lays the foundation for Study 2 by showing that temperature anomalies are associated with increased attention to climate change and climate policies which is arguably prerequisite for WE to affect voting behavior (Brügger et al., 2021, Sisco, 2021). Study 2 examines the more consequential behavior of voting for political candidates who are likely to support climate policies. We find a significant negative association between local anomalous temperatures and votes cast for Republican US congressional candidates.