International Public Opinion on Climate Change

Yale Climate Communication, Data for Good at Meta, Rare’s Center for Behavior & the Environment

Anthony Leiserowitz, Marija Verner, Jennifer Carman, Seth Rosenthal (2023)

Cross-national survey of climate beliefs, risk perceptions, and policy support across multiple countries. Provides the comparative baseline most single-country studies lack.

Link to study

What the evidence shows

This is a headline-level snapshot of climate attitudes across a very wide set of countries. It asks Facebook users what they know about climate change, whether they think it is happening and whether people cause it. It also asks how worried they are, how much harm they expect, how important the issue is to them, and how high a priority it should be for their government.

The report is short (about 13 pages of findings) and descriptive. Each section reports which areas score highest and lowest, how many of the 110 areas show a majority, and a regional breakdown. The charts in Appendix II hold the full per-area distributions. It does not test messages, explain why attitudes differ, or compare groups within countries.

The survey is a partnership between Yale, Meta's Data for Good programme and Rare. It follows an earlier 2022 edition (fielded March–April 2022). Companion reports from the same dataset exist, including one on extreme weather and vulnerability in lower-income, lower-emitting countries.

Key Findings

Belief that climate change is happening is near-universal. A majority in all 110 areas say yes, after being shown a short definition. 19 areas are at 90% or higher (El Salvador 94%). The lowest are Haiti (64%), Australia and the Netherlands (70%), and Laos and Austria (71%).
Belief in human causation is much weaker than belief in occurrence. 109 of 110 areas have a majority saying humans cause it at least partly. But only 16 areas have a majority saying it is mostly human-caused. Portugal is highest (61%), then Spain (59%) and Finland (57%). Haiti (18%) and Indonesia (20%) are lowest.
Worry is widespread. A majority are "very" or "somewhat" worried in 108 of 110 areas. South Korea and Puerto Rico are highest (93%). The Netherlands (45%) and Yemen (47%) are lowest.
People see harm to others as more likely than harm to themselves. In 80 of 110 areas a majority say climate change will harm future generations "a great deal". Only 24 areas have a majority saying it will harm them personally "a great deal". Personal importance sits in between: 81 of 110 areas have a majority calling it "extremely" or "very" important.
Most want government to prioritise it. A majority in 103 of 110 areas say it should be a "very high" or "high" government priority. Puerto Rico (92%) and El Salvador (90%) lead. Yemen (38%) and Turkey (45%) are lowest per the executive summary.
———
Curious how this insight applies to your work? Test Your Message ->
———

Stand-out data points

80 vs 24 areas: majorities expecting "a great deal" of harm to future generations vs to themselves personally.
45 of 110 areas: majorities who say they know at least a moderate amount about climate change (Finland 89%, Hungary 85%; no majority in any Sub-Saharan African area).
36%: respondents in Benin and Haiti who say they have "never heard of" climate change.

Examples of how Understories uses this research

Actionability 103 of 110 areas want climate change to be a high government priority. That supports messages asking for policy or political action. The report has no data on what individuals will do, so it cannot support scores on personal-action asks.
Emotional resonance Worry is high almost everywhere (108 of 110 areas), so a message does not need to create concern. It needs to give existing concern somewhere to go. Concern for future generations (80 areas) is a stronger emotional hook than personal threat (24 areas).
Credibility The gap between "it's happening" (all 110 areas) and "mostly human-caused" (16 areas) means credibility claims about cause need more support than claims about occurrence. Flag messages that treat human causation as already accepted.
Non-maleficence The risk is over-reading country rankings: treating a "low-concern" country as a target for fear-based messaging, or treating "high worry" as permission to amplify alarm. The report identifies no manipulative tactics, so Understories should not infer any from it. It should flag the personal-harm gap as a place where exaggerating personal risk would be misleading.
Explainability: Specific enough to cite as baseline context, e.g. "81 of 110 areas have a majority calling climate change very important to them personally (YPCCC, 2023)". It is not specific enough to support a scored recommendation about message wording, framing or tone.

non-exhaustive