Exploring Global Perspectives: How Science Communication Shapes Climate Change Attitudes Across 68 Countries – Insights from the TISP Dataset

In this section, we describe how the TISP dataset was collected and prepared before being published. While some details can be found in other TISP publications, this article offers the most thorough explanation of the methods used in gathering the TISP dataset.

Ethical Compliance

We obtained ethical approval for our study from the Institutional Review Board at Harvard University in August 2022, which deemed it exempt from full review. All authors ensured their respective institutions also reviewed the survey to guarantee ethical compliance in the countries where we collected data. Informed consent was secured from every participant before they took the survey.

Pre-registration

To boost credibility and transparency, we pre-registered our study methods with OSF on November 15, 2022, before collecting any data. This pre-registration documented our study design, data collection methods, variables, and sample size, which we determined using power analyses. Minor deviations occurred; for instance, we exceeded our target sample size due to additional funding, but we were unable to achieve targets in some countries because of limited local respondents. Despite challenges, we collected data from 68 countries, adjusting quotas and extending the data collection period to August 2023 as necessary.

Participants

The TISP dataset includes detailed records from 71,922 participants across 68 countries. Initially, we gathered 72,135 responses, but had to remove duplicates. This dataset reflects over a quarter of all countries worldwide, covering diverse income levels according to World Bank classifications. We primarily recruited participants through online panels, ensuring a consistent method across nations.

Procedure

We used a structured survey with balanced quotas based on age and gender. The age distribution included five categories, each accounting for 20%: 18-29, 30-39, 40-49, 50-59, and 60+. The gender distribution was 50% male and 50% female, and participants who preferred not to disclose their gender were not subjected to quotas. Most surveys were done online using Qualtrics, but some data was collected face-to-face in the Democratic Republic of the Congo. Our team provided extensive resources and support for consistent data collection across different countries.

The survey was conducted between November 30, 2022, and August 27, 2023, with an average completion time of around 18 minutes.

Measures

The questionnaire comprised 111 variables, although a few responses from some countries were missing certain items. Full questionnaires are available online for review, and users are encouraged to refer to the core questionnaire for accurate variable labels and codes. Participants provided demographic data and their responses to various questions relating to science and societal issues, ensuring a broad understanding of public perspectives.

Informed Consent

All participants received a consent form detailing the study’s purpose and ensuring data anonymity.

Demographic Data

Participants shared information regarding their gender, age, and education levels.

Attention Checks

To ensure validity, participants completed attention checks during the survey process, with those failing instructed to exit the survey.

Understanding of Science

Participants were given a definition of science to ensure clarity in the questions that followed, which were randomized to avoid bias.

Information Exposure

The survey asked participants about their exposure to scientific information across various media over the past year, including news and social media.

Perceived Benefits of Science

Respondents shared their beliefs on how scientific research benefits people in their country and which regions gain the most from scientific work.

Data Pre-processing

The TISP data underwent meticulous pre-processing to ensure accuracy. We merged data from all research groups, excluded incomplete responses, removed duplicate entries, and filtered out outliers. For income data, we converted local values to U.S. dollars and applied necessary transformations for analysis. Ultimately, we computed post-stratification weights to ensure representative statistical analyses.

Sample Characteristics

The cleaned dataset now presents a complete picture of participants from diverse backgrounds and regions. Detailed overviews of sample characteristics are available, providing insights into the demographics represented in this extensive dataset.