2.1. Participants
Participants were drawn from the GfK KnowledgePanel. GfK (now Ipsos) uses Address Based Sampling (ABS) to randomly recruit panelists using probability-based sampling methods: the panel is designed to be representative of the United States. ABS uses the Delivery Sequence File (DSF) of the USPS, which improves population coverage relative to traditional random-digit-dialing methods and enables recruitment of harder-to-reach individuals such as younger people or minority groups. Households without an Internet connection are provided a web-enabled device and free Internet services. Once household members are recruited for the panel and assigned to a study sample, they are notified electronically of the opportunity. They can then take the survey through their email link or by visiting their online member page.
Data are from a larger, longitudinal study of responses to hurricanes on the Gulf Coast. The first wave of data was collected between 6 pm 9/8/2017 and 6 am 9/11/2017; all 5,940 eligible KnowledgePanel panelists living in Florida or Texas were invited to participate; 2,774 completed the survey for a response rate of 46.7% during the 60 hours of data collection. The data presented herein are from the fifth wave of data collection, which occurred between 12/2022-1/11/2022. Of 1,766 eligible panelists recruited to participate in Wave 5 (i.e., those who had completed prior waves of data and had agreed to be contacted for future surveys), 1,479 completed the survey for an 83.7% response rate.
2.2. Measures
2.2.1. Climate-change mitigation behaviors
Individual-level climate change mitigation behaviors. Participants were asked to report which of the following behaviors they had engaged in during the past week: 1) “Used public transportation, biked, or walked to work instead of driving”; 2) “Used energy-efficient lightbulbs such as CFLs or LEDs”; 3) “Recycled”; 4) “Taken shorter showers”; 5) “Driven a hybrid or electric vehicle”; 6) “Reduced red meat consumption”; 7) “Ate a more plant-based diet”; 8) “Reduced food waste”; 9) “Composted waste”; 10) “Checked the air in your tires to ensure fuel efficiency”; 11) “Used a smart thermostat”; and 12) “Installed or used low-flow shower heads or faucets”. Items were derived from prior research (Mascatelli et al., 2021). Responses were summed.
Collective-level climate change mitigation behaviors. Participants were asked to indicate which of the following behaviors they had performed in the past year: 1) “Worked with community members to help people prepare for hurricanes or other natural disasters”; 2) “Worked with community members to create green spaces (e.g., plant trees, restore habitat) in my community”; 3) “Signed a petition in support of action to help the environment”; 4) “Signed a petition in support of action on climate change”; 5) “Made a donation in support of action on climate change”. Items were based on prior research (Roser-Renouf et al., 2014). Responses were summed.
2.2.2. Climate change-action emotions
Climate action-related emotions. Respondents were asked, “When you reflect on your ability to take action to address climate change, do you feel”: 1) Hopeful, 2) Confident, 3) Optimistic, 4) Helpless, 5) Powerless, 6) Lacking control, 7) Indifferent, 8) On edge, 9) Uneasy, and 10) Nervous. Respondents reported on each of the 10 emotions. Response options were on a 4-point scale from 1 (definitely do not feel this) to 4 (definitely feel this). Items were also grouped into two composites of positive (i.e., hopefully, confident, optimistic) and negative (i.e., helpless, powerless, lacking control, on edge, uneasy, nervous). Items were derived from prior work (Geiger et al., 2021).
2.2.3. Worry
Worry regarding climate-related hazards. Respondents were asked “How much do you worry about the following personally affecting you or someone in your family in the future?” and “How often in the past week have you had fears about the possibility of the following affecting the community where you live?” for the following hazards: major flooding, nuisance flooding, hurricanes, heat waves, tornadoes, and sea level rise. Participants responded to each question (12 items total) on a Likert-type scale from 1 (never) to 5 (all the time). Reliability was excellent α = .90. Of note, consistent with prior research, we combined these items to measure worry as perseverative cognition (e.g., ruminative or repeated thoughts about the future) rather than worry and fear as distinct states (Williams et al. 2022). Items were derived from prior work (Holman et al. 2008; Sweeting et al. 2020; Williams et al. 2022).
Worry regarding climate change. Respondents were asked how much they worried about climate change “personally affecting you or someone in your family in the future?” and how often in the past week they had fears about climate change “affecting the community where you live?” Participants responded to each of the 2 questions on a Likert-type scale from 1 (never) to 5 (all the time). Reliability was excellent α = .90. Items were derived from prior research (Holman et al. 2008; Sweeting et al. 2020; Williams et al. 2022).
2.2.4. Efficacy
Individual-level climate behavior efficacy. Efficacy regarding individual climate actions was assessed by asking: “Of the actions above that you do, how much will they help reduce the impacts of climate change?” Response options were 1 (not at all) to 5 (completely). Given the low number of respondents (n = 12) in the highest group, groups 4 and 5 were combined.
Collective-level climate behavior efficacy. Respondents were asked, “Of the actions above that you do, how much will they help reduce the impacts of climate change?” Response options were 1 (not at all) to 5 (completely). Given the low number of respondents (n = 10) in the highest group, groups 4 and 5 were combined.
All study specific measures are included in Supplemental Materials I.
2.3. Analytic Strategy.
First, descriptive statistics were calculated for all key study variables and a correlation matrix was constructed. Second, two multiple Poisson regression analyses (appropriate for count data) examined demographic variables (race/ethnicity, gender, income, age, education [Bachelor’s degree or higher = 1, less than a bachelor’s degree = 0] and political party identification [a 7-item scale ranging from 1 = strong Republican to 7 = strong Democrat]) as predictors of 1) individual-level climate change mitigation behaviors and 2) collective-level pro-climate change mitigation behaviors. Third, for each dependent variable (individual- and collective-level climate change mitigation behaviors), a series of Poisson regressions examined each’s association with specific emotions (hopeful, confident, optimistic, helpless, powerless, lacking control, indifferent, on edge, uneasy, and nervous) related to performing climate change mitigation behaviors. Models were built using a hierarchical variable entry strategy as follows: Model 1 included demographics and each specific emotion, Model 2 added worry about climate-related hazards, Model 3 added worry about climate change, and Model 4 added efficacy of actions to reduce the impacts of climate change. Fourth, using an identical, four-model approach, a series of Poisson regressions examined the association between 1) individual- and 2) collective-level climate change mitigation behaviors and composite climate-related positive and negative emotions. Interaction terms between positive emotion and self-efficacy and negative emotions and worry were calculated and examined in post-hoc exploratory analyses. Analyses were preregistered on the Open Science Framework (doi:10.17605/OSF.IO/UDG9A). Procedures were approved by [REMOVED FOR REVIEW].
All descriptive and inferential statistics were weighted using study-specific post-stratification weights. These weights were calculated to adjust the final study sample to the demographic compositions of the states of Florida and Texas for adults 18 and older. Weighting benchmarks were based on the U.S. Census Bureau’s Current Population Survey (March 2021 update), and were calculated using the following demographic cells: gender (male, female), by age (18–29, 30–44, 45–59, 60+), race/ethnicity (White/Non-Hispanic, Black/Non-Hispanic, Other/Non-Hispanic, Hispanic, 2 + Races/Non-Hispanic); household income (Under $25,000, $25,000-$49,999, $50,000-$74,999, $75,000-$99,999, $100,000-$149,999, $150,000 and over); metro/non-metro areas, and education (less than high school/high school, some college, Bachelor’s or higher).