Climate change skepticism of European farmers and implications for effective policy actions

Our study used two datasets, the European Social Survey (ESS) Round 8 and Round 10 and measures attribution skepticism (i.e., skepticism about the human causes of climate change, 2016–2017 and 2020–2022), and the Special Eurobarometer surveys on climate change (Supplementary Table 1), collected every two years over the period 2011–2021, measuring impact skepticism across distinct samples. Both the ESS and the Eurobarometer are long-standing, widely recognized social science surveys with comparatively low biases, such as those related to representation and response rates. Country-level data were gathered from various sources and compiled into separate datasets. Data processing was conducted using Stata and Excel, and all steps were thoroughly documented. The commented code is available upon request (Lea Kröner; [email protected]).

The ESS datasets were obtained from the openly accessible Datafile Builder ( which offers harmonized data waves. The Eurobarometer datasets were downloaded from the GESIS Leibniz Institute for the Social Sciences open-access data catalog ( and appended across six waves. Each of the country-level datasets was merged with the corresponding survey datasets using a many-to-one (m:1) merge command, with countryname as the key variable. This variable has identical coding across both the survey datasets and the country-level datasets, ensuring consistency during merging.

Individual level data ESS—attribution skepticism

The eighth and tenth rounds of the ESS were conducted through face-to-face interviews between 2016 and 2017 and 2020 and 2022, respectively. These waves were chosen because they feature the key rotating module on climate and energy⁴⁵. Generally, the ESS questionnaires aim to measure public attitudes, beliefs, and behaviors. We included Austria, Belgium, Bulgaria, Croatia, Czech Republic, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Lithuania, Latvia, the Netherlands, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom in our analysis. This resulted after listwise deletion of missing values on the variables of interest in 75,419 observations (1789 Farmers) across 25 countries, each with a minimum representative sample size of 1500 respondents (or 800 if the population is below 2 million).

Attribution skepticism⁴⁵, the dependent variable in this study, was measured with the question: “Do you think that climate change is caused by natural processes, human activity, or both?”. The response options ranged from 1 “Entirely by natural processes” to 5 “Entirely by human activity”. These were reverse-coded to represent skepticism. The independent variable, whether someone is a farmer or not, was based on the occupation variable. We coded those who belong to the category “Skilled Agricultural, Forestry & Fishery” as farmers and the people belonging to the other categories as non-farmers. Other categories include Armed Forces, Managers, Professionals, Technicians and associate professionals, Services and Sales Workers, Skilled agricultural, Forestry & Fishery, Craft and related trades Workers, Plant and Machine Operators, and Assemblers, Elementary Occupations, and Not applicable.

The control variables included gender with 1 “male” and 0 “female”, age, trust in politicians with 0 “No trust at all” until 10 “Complete trust”, and educational level (scale 1–7). These variables were included due to their established role as determinants of climate change perceptions and their differences between farmers and non-farmers. The questionnaires are available on the ESS website at: https://www.europeansocialsurvey.org/methodology/ess-methodology/source-questionnaire.

Individual level data Eurobarometer special issues—impact skepticism

The Eurobarometer has been regularly surveying the public on behalf of the European Commission and other EU entities since 1973. The Eurobarometer questionnaires focus on topics related to the EU and perspectives on contemporary political and social issues. Face-to-face interviews are performed in the spring and fall and are always based on new samples (“repeated cross-section” design). Except for small countries like Luxembourg and Malta, the standard sample size (in terms of completed interviews) in the standard Eurobarometer surveys is 1000 respondents per country. Additionally, the Eurobarometer has special issue waves. This study considered answers from 6 special issue waves on climate change that took place between 2011 and 2021. After appending the waves, merging the contextual data, and listwise deletion of missing values on the variables of interest, 89,867 observations (1378 farmers) within 28 countries remained. These countries are Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Great Britain, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, and Sweden.

The dependent variable impact skepticism was measured with the reverse-coded question “how serious a problem do you think climate change is at this moment?” Please use a scale from 1 to 10, with ‘1’ meaning it is “not at all a serious problem” and ‘10’ meaning it is “an extremely serious problem”. Moreover, we included a time variable to control for contextual differences between the two waves.

To operationalize the independent variable of whether the respondent is a farmer, the following question was used: “What is your current occupation?”, with 18 response options. These options are Responsible for ordinary shopping, etc./Student/Unemployed, temporarily not working/Retired, unable to work/Farmer/Fisherman/Professional (lawyer, etc.)/Owner of a shop, craftsmen, etc./Business proprietors, etc./Employed professional (employed doctor, …)/General management, etc./Middle management, etc./Employed position, at desk/Employed position, traveling/Employed position, service job/Supervisor/Skilled manual worker/Unskilled manual worker, etc. Furthermore, on the individual level, we controlled for gender, age, and education of the respondents. Moreover, we included a time variable to control for a trend toward less skepticism, where we recoded the waves to a variable ranging from 1 to 6, which were treated as continuous (2011 = 1) (2013 = 2) (2015 = 3) (2017 = 4) (2019 = 5) (2021 = 6). Whilst political orientation is commonly recognized as a determinant of climate change perceptions, we did not include it in the main analyses of both our studies due to a high number of missing values on these variables. However, when we conducted the analyses on a subsample where missing values were handled through listwise deletion and political orientation, the results remained stable. For more details on the questionnaires, see the GESIS Leibniz Institute for the Social Sciences open-access data catalog at https://search.gesis.org/.

Contextual country-level data

We included three independent variables at the country-level and their interactions with being a farmer in the analysis. To measure the economic well-being of the country, we used the average annual indexed GDP per capita in purchasing power parities (PPPs). It is defined as the value of all goods and services produced less the value of any goods or services used in their creation. The volume index of GDP per capita in purchasing power standards (PPS) is expressed in relation to the European Union average set to equal 100. If the index of a country is higher than 100, this country’s level of GDP per head is higher than the EU average and vice versa. Basic figures are expressed in PPS, i.e., a common currency that eliminates the differences in price levels between countries, allowing meaningful volume comparisons of GDP between countries. Please note that the index, calculated from PPS figures and expressed with respect to EU27_2020 = 100, is intended for cross-country comparisons rather than for temporal comparisons. Moreover, the General Innovation Index (GII) per country between 2014 and 2021 is used based on Eurostat data ( To measure what area of the country is occupied by agriculture, we used the average annual agricultural land share per country between 2014 and 2020 based on data from the World Bank ( Agricultural land refers to the share of land area that is arable, under permanent crops, or under permanent pastures. Arable land includes land defined by the FAO as land under temporary crops (double-cropped areas are counted once), temporary meadows for mowing or for pasture, land under market or kitchen gardens, and land temporarily fallow. Land abandoned as a result of shifting cultivation is excluded. Land under permanent crops is land cultivated with crops that occupy the land for long periods and need not be replanted after each harvest, such as cocoa, coffee, and rubber. This category includes land under flowering shrubs, fruit trees, nut trees, and vines, but excludes land under trees grown for wood or timber. Permanent pasture is land used for five or more years for forage, including natural and cultivated crops. Land area is a country’s total area, excluding area under inland water bodies, national claims to the continental shelf, and exclusive economic zones. In most cases, the definition of inland water bodies includes major rivers and lakes. As a measure of the climate risk for agriculture of a country, we defined an index combining the change in the length of the growing season and of number of dry days. Originally, the score runs from positive (climate benefits) to negative (climate costs), but we reverse-coded it, so that countries with a higher risk have a higher value on the scale (see Supplementary Table 6).

Models and statistical techniques

The data were analyzed using multilevel statistical methods. The rationale is that the surveyed individuals in both studies are nested in countries, and relationships between variables on different levels (individual and country) are to be investigated. Multilevel models are designed to adequately include the dependencies of individuals nested in the same country⁶⁰. We included cross-level interaction terms of the country-level factors (i.e., indexed GDP/capita, GII, agricultural share of land (%) and climatic risk index for agriculture) with the independent variable “farmer” on the individual level, to investigate whether potential differences in the effect of being a farmer across countries can be explained by differences in the macro factors across countries. In the Supplementary Tables 4 (Eurobarometer) and 5 (ESS), these form Models 1 to 5. Models 6 to 9 in both tables report the robustness analyses. Models 6a and 6b replace agricultural employment with the share of agricultural land used in the main analyses (Models 5a and 5b). The results do not change. Models 7a and 7b include political orientation in the Eurobarometer and replace political trust with political orientation in the ESS. The results are robust. To assess the representation of farmers, we compared the percentage of farmers in our samples to the share of agricultural employment per country (Supplementary Table 7). For many countries, the survey representation aligns closely with the actual share of agricultural employment. In addition, to address potential bias, we re-ran the analyses excluding countries where farmers were notably underrepresented, marked in red in the table. The results do not change, as shown in Models 8a and 8b. Finally, in Models 9a and 9b, we have run additional regression analyses where we have deleted the 20% countries with the lowest share of farmers. Again, the results are robust.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.