Generational differences in climate-related beliefs, risk perceptions and emotions in the UK

The study investigates whether there is a generational gap in engagement with climate change across cognitive (beliefs), risk perceptions, and affective (emotions) dimensions. Prior work often suggests younger people are more concerned than older adults, potentially due to earlier exposure and greater anticipated impacts, with growing attention to climate-related anxiety among youth. However, empirical findings are mixed, with some studies showing minimal or no age differences in climate concern. The authors posit that inconsistent outcome measures (beliefs vs concern vs risk vs emotions) may contribute to this ambiguity. They conceptually distinguish beliefs (e.g., anthropogenic causation, urgency, temporal proximity), risk perceptions (likelihood/seriousness to self and society), and emotions (worry, fear, guilt, outrage), proposing a cognitive-to-affective continuum and a hierarchical relationship where beliefs underpin risk perceptions, which in turn underpin emotional responses. The research question is whether generational differences are present across these distinct components and whether the gaps persist over time (2020–2022).

Prior literature indicates older individuals are more likely to express skeptical views about climate change’s reality, causes, and impacts, while younger people often report higher environmental and climate concern. Yet meta-analytic and cross-national work has found negligible or inconsistent age effects for environmental concern and climate worry. Conceptual clarity is emphasized: beliefs are propositional cognitions; risk perceptions encompass perceived likelihood/seriousness, generalized concern, and personal worry; emotions include worry, fear, guilt, and anxiety, with fear/anxiety potentially more debilitating than concern/worry. The authors present a comprehensive model situating beliefs, risk perceptions, and emotions along a cognitive–affective continuum, with emotions reflecting higher engagement. Generational labels (Post-War/Silent, Boomers I and II, Gen X, Millennials, Gen Z) are widely used in Western contexts and in media/academic discussions of climate engagement, providing a framework for cohort-based comparisons.

Design: Three cross-sectional, nationally representative online surveys in the UK by CAST, administered by DJS Research. Data collection windows: Wave 1 (29 Sep–26 Oct 2020; n=1893, including Scotland booster n=485 and Wales booster n=467), Wave 2 (28 Aug–22 Sep 2021; n=1001), Wave 3 (5 Sep–6 Oct 2022; n=1087). Quotas ensured representativeness for gender, age, region, and socioeconomic status. Ethical approvals: EC.20.08.11.6068 (2020), EC.21.08.10.6385 (2021), EC.22.07.12.6597 (2022). Informed consent obtained. Measures: Climate-related beliefs (3 items): perceived causes (1–5; 0 if “no such thing as climate change”), perceived temporal distance (“already feeling effects” recoded to binary 1 vs all others 0), perceived urgency (1–5). Risk perceptions (3 items): perceived impacts on the UK (1–5 bipolar), perceived threat to self/family (1–5), perceived threat to the UK (1–5). Emotions (4 items): worry (1–5), fear (1–5), guilt (1–5), outrage (1–5). Recoding: “already feeling effects” endorsed by 66.8% (2020), 64.6% (2021), 68.4% (2022). Generational groups: Gen Z (1996–2010), Millennials (1981–1995), Gen X (1965–1980), Boomers II (1955–1964), Boomers I (1946–1954), Post-war/Silent (1928–1945). Due to low n, Post-war combined with Boomers I. Covariates: gender, education (degree vs no degree), political orientation (z-scored 11-point left–right), and home nation. Analytical approach: (1) Separate linear regressions for continuous outcomes and logistic regression for the binary temporal proximity outcome, using Gen Z as reference; covariates included. Analyses run separately by year (2020, 2021, 2022). (2) Multilevel repeated-measures models treating nine of the ten variables (excluding the binary temporal proximity) as Level-1 repeated measures nested within individuals (Level 2). Model 1 included generational group and covariates; Model 2 added measure-type indicators (risk perceptions vs beliefs; emotions vs beliefs) and their interactions with generational groups. Intraclass correlation (ICC) estimated from a null model. Software: R 4.0.2, RStudio 2021.09.0+351; stats and lme4 packages. Code and questionnaires available at https://doi.org/10.17605/OSF.IO/DKRCB.

- Descriptive patterns (Table 1): Across 2020–2022, younger groups (Gen Z, Millennials) generally show higher means for climate-related beliefs, risk perceptions, and emotions; the pattern is strongest in 2020 and less pronounced in 2021–2022. Proportions endorsing “already feeling effects” were high across all years (about two-thirds each year). - 2020 regressions (Table 2): No significant generational differences for perceived causes or urgency. Boomers II were more likely than Gen Z to say effects are already being felt (OR=2.70, 95% CI [1.24, 5.84]). Compared with Gen Z, older groups (Gen X, Boomers II, Boomers I/older) reported lower perceived threat to self/family and to the UK and lower emotional responses (worry, fear, guilt, outrage), indicating a generational gap primarily in risk perceptions and emotions. - 2021 regressions: Fewer significant differences. Boomers II and Boomers I/older showed higher belief in anthropogenic causation than Gen Z in 2021 (difference not persistent in 2022). Older groups (Boomers II; Boomers I/older) and Gen X more often endorsed “already feeling effects”: OR=2.30 (95% CI [1.15, 4.72]) and OR=4.04 (95% CI [2.00, 8.19]) for Boomers II and Boomers I/older, respectively. Emotional differences persisted, especially lower fear/guilt/outrage among Boomers vs Gen Z. - 2022 regressions: Older groups continued to be more likely to report “already feeling effects”: Boomers II OR=4.66 (95% CI [2.40, 9.04]); Boomers I/older OR=3.42 (95% CI [1.76, 6.65]); Gen X OR=2.54 (95% CI [1.35, 4.81]). Major differences in risk perceptions largely disappeared. Emotional differences remained, particularly between Boomers and Gen Z. - Multilevel repeated-measures models (Table 3): ICCs indicated substantial shared variance across measures attributable to individuals (ICC=0.39 in 2020; 0.36 in 2021; 0.33 in 2022), reflecting general climate concern. In 2020, Baby Boomers exhibited lower engagement across repeated measures, driven by lower risk perceptions and emotions; Gen X showed lower emotions vs Gen Z. In 2021, overall generational gaps narrowed for beliefs, with persistent differences in emotions (and some risk perceptions). By 2022, differences in risk perceptions further diminished, while differences in emotional engagement remained significant across all years. - Overall synthesis: Generational differences are most consistent and largest for affective engagement (fear, guilt, outrage), with younger generations reporting stronger emotions. Differences in beliefs (e.g., anthropogenic causation, urgency) are smaller and narrowed from 2020 to 2021/2022. Contrary to expectations, older generations were more likely to state that the UK is already feeling the effects of climate change in 2021 and 2022.

The findings address the research question by showing that generational differences vary across cognitive and affective components: beliefs and perceived impacts show smaller and diminishing differences over time, while emotions (fear, guilt, outrage) consistently differ by generation, with younger groups reporting stronger negative emotions. This clarifies mixed prior evidence by demonstrating that outcomes along the cognitive–affective continuum are not interchangeable and can yield divergent age patterns. The unexpected higher temporal proximity belief among older groups may reflect shifting baselines and declining remarkability of temperature anomalies, making current changes more salient to those with longer historical reference frames. Persistent emotional differences are consequential because emotions strongly predict policy support, pro-environmental action, and activism; thus, stronger emotions among younger generations may partly explain their higher levels of climate engagement and activism. The narrowing of belief gaps from 2020 to 2021/2022 may relate to period effects such as increased media attention (e.g., COP26, UK Net Zero Strategy), protests, and extreme events, which raised salience across age groups. Overall, the study underscores the importance of analyzing beliefs, risk perceptions, and emotions separately when examining generational patterns.

This study jointly examined generational differences in climate-related beliefs, risk perceptions, and emotions using three UK-representative surveys (2020–2022). It shows that the most robust generational gap lies in affective engagement—especially fear, guilt, and outrage—while differences in cognitive beliefs are smaller and have narrowed in recent years. Older generations were paradoxically more likely to state that climate effects are already being felt, possibly due to shifting baselines. The work contributes a comprehensive framework that clarifies why prior research found mixed age effects and highlights emotions as a key dimension of generational climate engagement. Future research should employ longitudinal designs to disentangle cohort, developmental, and period effects; expand to non-Western contexts to test generalizability; investigate consequences of emotional engagement for wellbeing and behavior; and develop communication and intervention strategies to bolster constructive emotional engagement among older generations while supporting younger people’s mental health.

- Cross-sectional design prevents disentangling cohort, developmental (aging), and period effects. - Generational groupings vary in time span (e.g., Baby Boomers cover almost two decades), potentially blurring boundaries; individuals near cutoffs may resemble adjacent cohorts. - Small sample sizes for Gen Z (many not yet 18 during surveys) may reduce power to detect differences versus Millennials. - UK-based sample and Western generation labels may not generalize to other cultural contexts. - The temporal proximity item was dichotomized due to skewed distributions, potentially reducing nuance. - Online panel sampling, while quota-controlled for representativeness, may still have unobserved biases (acknowledged implicitly by design).