Yale Program on Climate Change Communication

About Downscaling Climate Opinions

This version of the Yale Climate Opinion Maps is based on survey data from 2008 through fall 2024. Public opinion about global warming is an important influence on decision-making about policies to reduce global warming or prepare for the impacts, but American opinions vary widely depending on where people live. So why would we rely on just one national number to understand public responses to climate change at the state and local levels? Public opinion polling is generally done at the national level because local-level polling is very costly and time-intensive. Our team of scientists, however, has developed a geographic and statistical model to downscale national public opinion results to the state, congressional district, and county levels. We can now estimate public opinion across the country and a rich picture of the diversity of Americans’ beliefs, attitudes, and policy support is revealed.

Our national surveys show that 72% of Americans think global warming is happening. Our new YCOM model estimates, however, show that only 53% of people in Emery County, Utah agree. Meanwhile 72% in neighboring Grand County, Utah believe global warming is happening.

Explore the maps by clicking on your state, congressional district, metro area, or county and compare the results across questions and with other geographic areas. The Congressional District Map reflects the 118th Congress (Jan. 2023 – Jan. 2025). You will find that not all congressional districts have data due to litigation regarding the adoption of new redistricting plans. In 13 states where districts were in litigation at the time of census data collection, we are delaying the release of estimates. In all cases, keep in mind that the margin of error increases as you go to finer geographic scales (i.e., from state to county levels). Thus, it is not possible to know whether a county estimate of 42% is statistically different from an estimate in a nearby county of, say, 44%. You can also click each shade of color on the legend to highlight just the locations matching that value. 

Beneath each map are bar charts displaying the results for every question at whichever geographic scale is currently selected. For the bar charts and the maps alike, multiple responses to each survey question are grouped into positive and negative categories as defined on the Survey Question Wording tab.

See the ‘Methodology’ section for more information about uncertainty estimates. In some cases, numbers that should sum to 100% or differences from the national average that should sum to zero are off by one percentage point; these effects are due to rounding errors.

This research was funded by the Schmidt Family Foundation, the U.S. Energy Foundation, the MacArthur Foundation, the Heising-Simons Foundation, King Philanthropies, and the Grantham Foundation. For further questions about these maps or what they mean, please see our Frequently Asked Questions (FAQ) tab above.

Methodology

This site provides estimates of U.S. climate change beliefs, risk perceptions, and policy preferences at the state and local levels – a new source of high-resolution data on public opinion that can inform national, state, and local decision-making, policy, and education initiatives. The estimates are derived from a statistical model using multilevel regression with post-stratification (MRP) on a large national survey dataset (n>32,000), along with demographic and geographic population characteristics.

The modeling employs data from 31 nationally representative surveys of American adults (18+) conducted from November 2008 through December 2024. The surveys were administered by Ipsos and drawn from the Ipsos KnowledgePanel®, an online panel of members drawn using probability sampling methods. All questionnaires were self-administered by respondents in a web-based environment, and computers were loaned to individuals who were chosen to participate but did not have access. Respondents came from all 50 states and 2,379 of 3,144 counties. Sample weights for all respondents were calculated by Ipsos to be nationally representative post-survey to match U.S. Census Bureau norms. For respondents who have taken the survey multiple times, only their most recent response was kept in the data, and all previous responses were removed. This resulted in 3,108 responses being removed. 

Multilevel modeling with poststratification (MrP) was used to estimate the spatial distribution of climate opinions at state, county, congressional district, and metro area levels (Howe et al. 2015; Mildenberger et al. 2016, 2017). Dependent variables were first recoded into binary format, with positive response values grouped and coded as “1” (e.g., “Somewhat favor” and “Strongly favor”) and non-positive values coded as “0” (“Somewhat oppose”, “Strongly oppose”, “Don’t know”, Refused); see “Survey Question Wording” on the interactive map tool tab for a full list of questions and how each was recoded to binary). MrP modeling then proceeded in two phases. First, the multi-level model is constructed by predicting individual survey responses as a function of both individual-level demographics (gender, race, education, and a three-way interaction term among these variables) and geography-level covariates. Second, “post-stratification”, or spatial weighting, is performed using the fitted model, where population-weighted opinion estimates for each demographic-geographic subtype are aggregated based on the subtype population distribution within each geographic subunit. 

The YCOM model uses four geography-level covariates: percent of people who drive alone to work, percent of same-sex households, percent of CO2 emissions per capita, and percent of people who voted Democrat in the most recent election. In order to ensure accurate and current estimates, two of the data sources used for these covariates in the current YCOM model (version 8) have been updated to reflect more current data. 

The U.S. Census Bureau’s American Community Survey (ACS) variable which was previously the source of same-sex household data was discontinued in 2018. Therefore, we replaced this covariate with the ACS variable “Coupled Households by Type”, which includes an almost identical survey question most recently asked in 2020. CO2 data from the Vulcan Project, originally published in 2010, were replaced with updated 2024 estimates of CO2 emissions and supplemented with new data from Crosswalk Labs, which provides estimates of CO2 emissions at the census tract level. Presidential vote share was acquired from the Redistricting Hub for the 2020 election, as the vote share data for 2024 is not yet available. All new sources of data were compared with their outdated counterparts and found to be highly correlated. 

Validating models is essential for producing accurate results. Our original YCOM model estimates were validated using three different methods. First, cross-validation analyses were conducted within the dataset. The dataset was divided into two sets of respondents, with one part used to run the model and the other kept aside for validation. The model estimates were then compared to the results of the set-aside respondents to directly quantify the percentage of correct answers the model predicted. These cross-validation tests were repeated multiple times using different sample sizes and dividing the data in different ways. Second, the model estimates derived from the full dataset were compared to the results of independent, representative state- and city-level surveys conducted in California, Colorado, Ohio, Texas, San Francisco, and Columbus, Ohio in 2013. The mean absolute difference between model estimates and validation survey results was 2.9 percentage points (SD = 1.5) among the four states (CA, TX, OH, CO) and 3.6 percentage points (SD = 2.9) among the two metropolitan areas (Columbus, OH, and San Francisco, CA), well within the margins of error for the survey results alone (at a 95% confidence level). Estimates have also been validated internally through a series of technical simulations. Third, some model estimates were compared with third-party survey data collected by other researchers in previous years.

Our current model estimates were validated by comparing modeled estimates with weighted survey averages at the national level and for the five most populous states. The mean absolute error between modeled estimates and weighted survey averages across all variables was 0.49 percentage points at the national level and 3.84 percentage points at the state level.

For the 2024 model estimates, uncertainty ranges are based on 95% confidence intervals using 99 bootstrap simulations. These confidence intervals indicate that the model is accurate to approximately ±7 percentage points at the state and congressional district levels, and ±8 percentage points at the metro and county levels. Such error ranges include the error inherent in the original national surveys themselves, which is typically ±3 percentage points.

Trends over time from 2010–2020 are provided for 16 state-level climate opinions using data from Marlon et al. (2022). Estimates for 2021, 2022, 2023, and 2024 are generated using our conventional MRP model described in Howe et al. (2015).

Marlon, J. R., Wang, X., Bergquist, P., Howe, P., Leiserowitz, A., Maibach, E., Mildenberger, M., and Rosenthal, S. “Change in US state-level public opinion about climate change: 2008–2020.” Environmental Research Letters 17, no. 12 (2022). 124046.

Howe, P., Mildenberger, M., Marlon, J., & Leiserowitz, A. (2015) “Geographic variation in opinions on climate change at state and local scales in the USA,” Nature Climate Change 5. DOI: 10.1038/nclimate2583.

The YPCCC is pleased to offer our downscaled climate change opinion estimates to the public. These data are distributed under the following terms of use. This is a legal agreement between you, the end-user (“User”) and Yale University on behalf of the Yale Program on Climate Change Communication (the “YPCCC”).  By downloading the survey data made available on this web site (“Data”) you are agreeing to be bound by the terms and conditions of this agreement.  If you do not agree to be bound by these terms, do not download or use the Data. The YPCCC hereby grants to the User a non-exclusive, revocable, limited, non-transferable license to use the Data solely for (1) research, scholarly or academic purposes, (2) the internal use of your business, or (3) your own personal non-commercial use.  You may not reproduce, sell, rent, lease, loan, distribute or sublicense or otherwise transfer any Data, in whole or in part, to any other party, or use the Data to create any derived product for resale, lease or license.  Notwithstanding the foregoing, you may incorporate limited portions of the Data in scholarly, research or academic publications or for the purposes of news reporting, provided you acknowledge the source of the Data (with express references to the YPCCC, as well as the complete title of the report) and include the following legend: The YPCCC bears no responsibility for the analyses or interpretations of the data presented here.

THE DATA IS PROVIDED “AS IS” WITHOUT ANY WARRANTY OF ANY KIND, EITHER EXPRESS OR IMPLIED, ARISING BY LAW OR OTHERWISE, INCLUDING BUT NOT LIMITED TO WARRANTIES OF COMPLETENESS,  NON-INFRINGEMENT, ACCURACY, MERCHANTABILITY, OR FITNESS FOR A PARTICULAR PURPOSE.  THE USER ASSUMES ALL RISK ASSOCIATED WITH USE OF THE DATA AND AGREES THAT IN NO EVENT SHALL YALE BE LIABLE TO YOU OR ANY THIRD PARTY FOR ANY INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE OR CONSEQUENTIAL DAMAGES INCLUDING, BUT NOT LIMITED TO, DAMAGES FOR THE INABILITY TO USE EQUIPMENT OR ACCESS DATA, LOSS OF BUSINESS, LOSS OF REVENUE OR PROFITS, BUSINESS INTERRUPTIONS, LOSS OF INFORMATION OR DATA, OR OTHER FINANCIAL LOSS, ARISING OUT OF THE USE OF, OR INABILITY TO USE, THE DATA BASED ON ANY THEORY OF LIABILITY INCLUDING, BUT NOT LIMITED TO, BREACH OF CONTRACT, BREACH OF WARRANTY, TORT (INCLUDING NEGLIGENCE), OR OTHERWISE, EVEN IF USER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

The YPCCC has taken measures to ensure that the Data is devoid of information that could be used to identify individuals (e.g., names, telephone numbers, email addresses, social security numbers) who participated in or who were the subject of any research surveys or studies used to collect the Data (“Personally Identifying Information”).  However, in the event that you discover any such Personally Identifying Information in the Data, you shall immediately notify the YPCCC and refrain from using any such Personally Identifying Information. This license will terminate (1) automatically without notice from the YPCCC if you fail to comply with the provisions of this agreement, or (2) upon written notice (by e-mail, U.S. or otherwise) from the YPCCC.  Upon termination of this agreement, you agree to destroy all copies of any Data, in whole or in part and in any and all media, in your custody and control. This agreement shall be governed by, construed and interpreted in accordance with the laws of the State of Connecticut. You further agree to submit to the jurisdiction and venue of the courts of the State of Connecticut for any dispute relating to this Agreement. Please use the following citation in any work that makes use of the data and documentation as follows: Howe, Peter D., Matto Mildenberger, Jennifer R. Marlon, and Anthony Leiserowitz (2015). “Geographic variation in opinions on climate change at state and local scales in the USA.” Nature Climate Change, doi:10.1038/nclimate2583. Direct any questions to the YPCCC at climatechange@yale.edu.

Frequently Asked Questions

What do these maps depict?

The maps depict estimates of the percentage of American adults (age 18 and over) who hold particular beliefs, attitudes, and policy preferences about global warming. The estimates were generated from a statistical model that incorporates actual survey responses, but combines these responses with demographic data from the U.S. Census to estimate opinions for different groups of people based on information such as their gender, race and ethnicity, and educational attainment; they also take into account changes in public opinion over time.

Where do the survey data underlying the estimates come from?

The data underlying the maps come from a large national survey dataset ( >32,000 respondents) collected between 2008 and 2024 as part of the Climate Change in the American Mind project led by the Yale Program on Climate Change Communication and the George Mason University Center for Climate Change Communication. Reports from the individual surveys are available here: CCAM Reports.

How many people did you survey in my state/county? How can you say anything about opinions in my geography if you didn’t survey anyone there?

The number of respondents surveyed varies for each geographic unit but is very large across the U.S. as a whole. Since our surveys are nationally representative, the number of respondents in each state and county is proportional to its share of the national population. In other words, larger states and counties will have more respondents than smaller states and counties. In some smaller counties, very few, if any, respondents may have been surveyed. In those cases, our model uses information about the population of the county, information from other counties in the same state and region, and information about the characteristics of the county (such as how the county voted in previous elections) to make its estimates. While this may sound surprising, it makes sense when we consider the strongest influences on a person’s individual opinions about climate change in the U.S. – political views. Knowing the political leanings of a county’s (or other unit’s) residents, along with the proportion of different demographic groups within that county, allows us to predict resident’s climate opinions very well. Adding additional information (e.g., about rural/urban context, carbon-intensive activities, etc.), further improves predictions. Moreover, for any given geographic unit, information is pooled from nearby units and from higher-level geographies (e.g., state-level information is included for all counties within that state). All of this information together allows us to produce reliable estimates (see next question) for all counties even if the number of individuals sampled within a specific county is small.

How accurate are the estimates?

No model is perfect and there are uncertainties in the model estimates. To validate the model, we conducted independent surveys in four states (CA, TX, OH, CO) and two metropolitan areas (Columbus, OH and San Francisco, CA) and compared the survey results to our model estimates. On average, the model estimates differed from the survey results by 2.9 percentage points among the four states and 3.6 percentage points among the two metropolitan areas, within the survey margins of error. A series of technical simulations estimate that the model has an average margin of error of ±7 percentage points at the state and congressional district levels, ±8 percentage points at the metro and county levels. Such error ranges include the error inherent in the original national surveys themselves, which is typically ±3 percentage points. The model uncertainties are smaller at broad geographic scales (e.g., the state level), and are larger at finer geographic scales (e.g., at the county and city levels). The model estimates also tend to be conservative, so geographic areas with extremely high or low measures are not estimated as well as areas with values closer to the national average for each survey question.

Why are some Congressional Districts gray (missing data)?

Due to redistricting, many states have congressional district boundaries that were not yet settled at the time of census data collection, or that did not yet have census counts available for those districts. Census counts are required for the modeling, so we cannot make estimates of public opinions about climate change for geographic areas that are missing census data. We will update the model and maps as soon as the districts are settled and the census data for those areas are available. Current maps reflect all the census data available through January 2025.

What does the gray color mean on some of the bars beneath the maps?

The gray area reflects people who refused to answer the question or said “don’t know”. We do not provide specific values for the gray areas because we did not develop estimates for these particular responses.

Do the maps account for differences in population density across the country?

No, the maps depict the estimated proportion of people within each geographic area who would answer each question as indicated. We have not adjusted the maps based on population density differences. It is important to keep in mind that some geographic areas may be large, but have few residents (e.g., Wyoming), while other geographic areas may be small, but have many residents (e.g., New Jersey). 

Do these maps reflect changes in opinions due to recent extreme weather events like Hurricane Helene?

Perhaps. The maps may reflect the impacts that specific extreme weather events had on public opinion in a given geographic unit. If public opinion in a particular area has been influenced by local events it is possible that the model would detect such an influence. However, data from specific events or types of events are not explicitly built into the model as predictor variables.

Can I use the data?

Yes. We encourage you to explore the maps and use the results in your own work. The data are available on our Data Download tab at the top of this page so that you can do your own analyses and create your own visualizations. If you publish an academic paper using these data please acknowledge the source by using the following citation:

Trends over time from 2010–2020 are provided for 16 state-level climate opinions using data from Marlon et al. (2022). Estimates for 2021, 2022, 2023, and 2024 are generated using our conventional MRP model described in Howe et al. (2015).

Marlon, J. R., Wang, X., Bergquist, P., Howe, P., Leiserowitz, A., Maibach, E., Mildenberger, M., and Rosenthal, S. “Change in US state-level public opinion about climate change: 2008–2020.” Environmental Research Letters 17, no. 12 (2022). 124046.

Howe, P., Mildenberger, M., Marlon, J., & Leiserowitz, A. (2015) “Geographic variation in opinions on climate change at state and local scales in the USA,” Nature Climate Change 5. DOI: 10.1038/nclimate2583.

If you publish a news article, visualization, blog post, or other publication using these data or maps, please include the link to the Yale Climate Opinion Maps website and attribution to the Yale Program on Climate Change Communication.