Toward a unified understanding of climate anxiety: Examining measures of climate and eco-anxiety

PLOS One | https://doi.org/10.1371/journal.pone.0339562 December 29, 2025 5 / 19 Expanding the focus beyond personal behavior change is crucial, as system-level transformations are essential for addressing climate change [39,40]. A recent development in this field is the readiness for climate action model [41,42], which combines individual behaviors weighted according to CO2 impact, with these system-level behaviors—that is, acceptance of climate policies and political participation—as indicators. In sum, different measures exist that assess climate change- and eco-related anxiety; the degree of their overlap is not yet clear, and validation efforts require the use of relevant convergent constructs and criteria that are appropriate for general population, non-clinical samples. The present study therefore aims at investigating the validity of climate and eco-anxiety by testing competing measurement models for the CAS and HEAS scales that are frequently discussed in the literature [6,9,12] as well as joint models for both scales. We then test the correlation of the emerging factor(s) with neurot- icism, risk perception of climate change–related health risks, climate change–related fatigue, and self-efficacy to show individual emission-reducing behavior. Moreover, we investigate the emerging factors’ criterion validity as correlations with the indicators of readiness for climate action. Materials and methods Data collection and participants Data for this study was collected as part of a larger project, the Planetary Health Action Survey (PACE; [42,43]). PACE is a

serial cross-sectional online survey, with each wave consisting of a non-probabilistic sample of approximately 1000 partic- ipants. The sample is quota-representative of Germany’s adult general population (18–74 years) for the characteristics of federal state (non-crossed) and age and gender (crossed, meaning quotas set for each age–gender combination). For the sample characteristics, see S1 Table in the Open Science Framework (OSF; https://osf.io/wvbd3/). We used data from the September 2023 PACE wave collected September 19–20 through an online access panel provider (Bilendi). Because the panel provider maintains reliable population quotas only for individuals aged 18–74, participants were drawn from this age range, and only complete responses were included in the final dataset (N = 1003). The study was hosted on the plat- form Tivian/Unipark. Participants needed, on average, 27 minutes to complete the full survey. With the given sample size, one-sided testing, a significance threshold of ɑ = .05, and a power of 1 – β = .95, our sensi- tivity allows for the detection of a true correlation of ρ = .10 [44]. Ethical approval The PACE study received ethical approval from the Institutional Review Board at the University of Erfurt (No. #20220525) and is in line with the general ethical guidelines of the University of Erfurt and the German Research Foundation. Partici- pants were informed about confidentiality, anonymity, and data-sharing practices, and they provided their informed con-

sent prior to enrollment in the online study. In addition, data collection and analysis were conducted anonymously. Measures The measures are presented in the order used in the study. The German and English versions of all measures from the PACE study, as well as the data and code for calculating the variables and reliabilities, can be found in the OSF (https:// osf.io/wvbd3/). Due to copyright, please refer to the original works for the German and English versions of the CAS, HEAS, and neuroticism items (CAS [6,9]; HEAS [7,12]; Neuroticism [45]). Since this study was part of a larger data collec- tion, more variables were assessed than are reported here. More details are reported in the project’s study protocol [42]. Constructs were assessed in blocks, with all items within each block presented in randomized order. All variables were modeled as latent constructs (for details on the models for risk perception of climate change–related health risks, self- efficacy to show individual emission-reducing behavior, and readiness for climate action, see [41]). Reliability was esti- mated as factor saturation McDonald’s ⍵, since it provides a more appropriate and robust measure of reliability than more