The UK's ambitious net-zero emissions target by 2050 necessitates significant behavioral changes, which are intrinsically linked to public perceptions of climate change risk. Existing research presents a mixed picture of UK public attitudes, with some showing high levels of concern while others demonstrate skepticism or denial. This study aims to provide a more nuanced understanding of these attitudes by focusing specifically on how people perceive the risks of climate change, rather than broader environmental concerns. It addresses three key research questions: (1) What are the distinct groups of attitudes toward climate change risk (ACCRs) in the UK? (2) How do socio-demographic variables relate to ACCR group membership? (3) How have ACCR group structure and membership changed over time? The study's findings are valuable for developing targeted communication strategies to enhance public engagement with climate change.

The literature on public perceptions of climate change risk reveals diverse attitudes. While many recognize climate change as a serious issue, significant skepticism and denial persist. Previous studies have used various methods like Latent Class Analysis (LCA) and cluster analysis to segment the population based on environmental attitudes and climate change beliefs. Studies like the "Six Americas" classification (Maibach et al., 2011) and other national-level segmentations (Crawley et al., 2020; Rhead et al., 2018) highlight the heterogeneity of public opinion. However, most studies have employed cross-sectional data, limiting the understanding of how these attitudes evolve over time. This study builds on previous work by specifically focusing on the clustering of ACCRs and tracking changes in cluster membership over time using longitudinal data. The review also examines the mixed evidence on the relationship between socio-demographic factors (sex, age, income, education, political affiliation) and climate change attitudes.

The study utilizes data from the UKHLS, a nationally representative longitudinal survey, focusing on waves 4 (2012-2014) and 10 (2018-2020). After listwise deletion of cases with missing values, the analysis includes 38,037 and 31,498 individuals, respectively. The analysis uses five statements related to climate change risk perception, measured using 5-point Likert scales or dichotomous responses (see Table 1). K-means cluster analysis is employed to identify homogenous clusters of ACCRs. The optimal number of clusters (k=3) is determined using Makles' (2012) criteria, considering within-cluster sums-of-squares (WSS), the η² coefficient, and the proportional reduction of error (PRE). The robustness of the k=3 solution is assessed through sensitivity analysis. Multinomial logistic regression is then used to examine the relationship between cluster membership and socio-demographic variables (age, sex, education, income, political affiliation). Finally, transition matrices are constructed to analyze changes in cluster membership between waves 4 and 10.

K-means cluster analysis revealed three distinct ACCR clusters in both waves 4 and 10: "Sceptical," "Concerned," and "Paradoxical." The Sceptical cluster strongly disagreed with statements indicating the seriousness and urgency of climate change. The Concerned cluster showed high levels of concern and supported mitigation efforts. The Paradoxical cluster acknowledged climate change impacts but lacked support for mitigation actions. Multinomial logistic regression revealed that males were more likely to belong to the Sceptical and Paradoxical clusters compared to females. Younger respondents were more likely to be in the Concerned cluster, while older respondents were more associated with the Paradoxical (wave 4) and Sceptical (wave 10) clusters. Higher income was associated with the Concerned cluster. Education and political affiliation were the strongest predictors of cluster membership, with higher education and left-leaning political affiliations strongly associated with the Concerned cluster. Transition matrices indicated a general shift from the Sceptical and Paradoxical clusters towards the Concerned cluster between waves 4 and 10 (Table 4). While the overall cluster structure remained consistent, significant individual-level transitions occurred.

The findings provide a dynamic picture of UK public attitudes towards climate change risk. The persistence of the three-cluster structure over time, despite substantial individual-level transitions, suggests considerable fluidity in opinion. The dominance of the Paradoxical cluster highlights the need for effective communication strategies targeting individuals who acknowledge the risks but are hesitant to support mitigation efforts. The study's findings confirm the influence of socio-demographic factors, but education and political affiliation emerge as particularly strong predictors of ACCR cluster membership. These findings support previous research on identity-protective cognition and the role of worldviews in shaping climate change attitudes. The significant shift towards the Concerned cluster between 2012 and 2020 could be attributed to various factors, including increased media coverage, climate-related events, and evolving public discourse.

This study provides valuable insights into the dynamics of UK public attitudes towards climate change risk, offering a typology of ACCRs and tracking their evolution over time. The identification of three distinct clusters, along with their associated socio-demographic profiles and transitions, offers important implications for the design of targeted communication strategies. Future research should investigate the mechanisms driving individual transitions between clusters and explore the interaction of values, worldviews, and personal experiences with climate change attitudes. Comparative studies across different countries and cultural contexts could also broaden the understanding of these complex attitudes.

The study's reliance on secondary data limits the depth of information available, potentially missing nuanced details. The use of two items not explicitly mentioning climate change introduces some uncertainty in interpretation. The focus solely on risk attitudes might oversimplify the multifaceted nature of climate change perceptions; future studies should include measures of knowledge, policy preferences, and behavior. The subjective nature of cluster analysis and potential response biases due to the interview format should also be considered. Finally, the findings are specific to the UK context and might not generalize to other countries.