Evaluating measurement invariance and diagnostic accuracy of climate anxiety scales in aging populations: a multi-instrument approach

Abstract

Climate change is increasingly recognized as a psychological stressor, with older adults representing a particularly vulnerable yet understudied group. This study evaluated the psychometric performance of three climate anxiety measures—the Hogge Climate Anxiety Scale (HCAS), Clayton Climate Change Anxiety Scale (CCCA), and Simon Climate Anxiety Scale (SCAS)—among 279 Persian-speaking older adults in Iran. Using a cross-sectional design, we examined structural validity, diagnostic accuracy, measurement invariance, and reliability, employing the GAI-SF as the criterion measure. All three instruments demonstrated acceptable construct validity based on exploratory and confirmatory factor analyses. SCAS showed a stable three-factor structure and excellent internal consistency (ω = 0.97). In classification analyses, SCAS achieved the highest sensitivity (0.89) and overall diagnostic accuracy (DOR = 3.09), whereas CCCA demonstrated slightly higher specificity (0.36). Bland–Altman analysis indicated that HCAS had the lowest measurement bias relative to GAI-SF scores. Measurement invariance testing supported full scalar invariance for SCAS across gender and anxiety subgroups, while HCAS and CCCA achieved only partial invariance. Mokken scale analysis further confirmed strong scalability (H > 0.40) and satisfactory reliability (α > 0.85) across all measures, with CCCA showing the highest Loevinger’s H (0.52). Age significantly predicted climate anxiety scores across all three scales (p < 0.001). Overall, SCAS emerged as the most robust and reliable instrument for assessing climate anxiety in Persian-speaking older adults, while HCAS showed the closest agreement with the criterion measure. These findings highlight the importance of culturally adapted, psychometrically sound tools for capturing climate anxiety in aging populations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-026-04559-6.

Highlights

Climate anxiety scales (HCAS, CCCA, SCAS) show varying predictive validity in older adults.

SCAS demonstrates superior sensitivity and diagnostic accuracy compared to HCAS and CCCA.

Measurement invariance analysis reveals partial scalar invariance for HCAS and CCCA across groups.

Mokken analysis confirms strong scalability and reliability for all three instruments.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-026-04559-6.

Introduction

The global demographic landscape is undergoing a profound transformation, with population aging emerging as one of the most significant societal shifts of the 21st century. According to United Nations projections, the number of individuals aged 60 years and older is expected to double by 2050, reaching nearly 2.1 billion worldwide [35]. This unprecedented demographic shift poses serious challenges for health systems, social infrastructure, and mental well-being—particularly as it converges with another defining crisis of our time: climate change [13].

Climate change is widely recognized as one of the greatest threats to global health and ecological stability [23]. Its adverse consequences—rising global temperatures, prolonged droughts, extreme weather events, and widespread ecological degradation—extend well beyond physical and economic disruptions to significantly impact psychological well-being [3]. A growing body of research has highlighted the mental health consequences of environmental instability, giving rise to constructs such as “climate anxiety,” which refers to emotional distress and worry related to awareness of climate change and its anticipated impacts [8].

Although climate anxiety has been primarily studied among younger populations [34], older adults may be particularly vulnerable to both the physical and psychological consequences of environmental stressors [19]. Aging is often accompanied by physiological decline, increased prevalence of chronic illnesses, and reduced adaptive capacity—all of which compound the risks associated with climate-related events such as heatwaves, air pollution, and natural disasters [6]. Furthermore, older adults often face unique psychosocial stressors, including generational identity, life-stage transitions, and a diminished sense of agency [32], which can amplify feelings of fear, helplessness, and loss of control in the context of climate change [16].

Despite these heightened risks, the mental health impacts of climate change on aging populations remain significantly understudied—especially in non-Western and non-English-speaking contexts [10]. Robust and culturally appropriate assessment tools are essential for identifying and supporting those most affected [2]. In response to this need, several psychometrically sound instruments have been developed in recent years, including the Hogge Climate Anxiety Scale (HCAS) [15], the Clayton Climate Change Anxiety Scale (CCCA) [7], and the Simon Climate Anxiety Scale (SCAS) [30]. These scales assess cognitive, emotional, and behavioral responses to climate threats and have shown strong reliability and validity in Western populations. However, their applicability in diverse cultural and linguistic settings remains limited.

This gap is particularly pressing in countries like Iran, which is currently facing severe climate-related challenges. With rising temperatures, dwindling water resources, desertification, and prolonged droughts, Iran has entered a critical period of environmental vulnerability [31]. These developments pose significant risks not only to infrastructure, agriculture, and public health but also to mental health—especially among older adults, who may experience increased psychological distress as a result of environmental instability and limited institutional support [12].

Despite the urgency of the situation, no validated Persian-language instrument currently exists to assess climate anxiety in older adults in Iran. This absence restricts both empirical research and the development of targeted interventions. It is insufficient to merely translate existing tools; rather, comprehensive cultural and psychometric validation is necessary to ensure these instruments are appropriate and reliable for use in local contexts [18].

A key strength of the present study is its comprehensive, multi-method psychometric framework. By integrating confirmatory factor analysis, item response theory and Rasch modeling, Mokken scale analysis, and measurement invariance testing, this study provides a robust and nuanced evaluation of climate anxiety instruments that goes beyond conventional validation approaches.

This study seeks to address this critical gap by evaluating the psychometric properties of three internationally recognized climate anxiety scales—the HCAS, CCCA, and SCAS—among Persian-speaking older adults in Iran. Specifically, we assess their structural validity, measurement invariance, diagnostic accuracy, and internal consistency, using the Geriatric Anxiety Inventory–Short Form (GAI-SF) as a external criterion. Climate anxiety is conceptualized as a domain-specific manifestation of broader anxiety processes rather than a wholly independent diagnostic entity. It shares core affective and cognitive mechanisms with generalized anxiety, including excessive worry, anticipatory threat appraisal, and distress-related rumination. Therefore, while geriatric anxiety measures such as the GAI-SF do not represent construct-equivalent or climate-specific diagnostic standards, they provide theoretically related reference constructs for evaluating convergent and criterion-related associations. The present study thus interprets correlations with the GAI-SF as evidence of theoretical convergence with a related anxiety domain rather than direct diagnostic equivalence.

By adapting these tools for the Persian-speaking elderly population, this research contributes not only to the global understanding of climate anxiety but also to the development of culturally sensitive strategies for assessing and mitigating climate-related mental health risks in aging societies.

Materials and methods

Study design and objective

This cross-sectional psychometric study was conducted to evaluate the diagnostic accuracy, structural validity, measurement invariance, and reliability of three climate anxiety scales in an older adult population in Iran. The instruments under evaluation included the Hogge Climate Anxiety Scale (HCAS) [15], Clayton Climate Change Anxiety Scale (CCCA) [7], and Simon Climate Anxiety Scale (SCAS) [30]. The study also examined the associations of these instruments with the Geriatric Anxiety Inventory–Short Form (GAI-SF), which served as a criterion measure.

Participants and data gathering

A total of 279 older adults participated in this study, including 140 males (50.2%) and 139 females (49.8%). Participants were recruited from two major metropolitan cities in southern Iran—Shiraz (51%) and Ahvaz (49%). A two-stage cluster sampling method was employed to ensure adequate geographical coverage and representativeness. In the first stage, cities were selected as the primary sampling units. In the second stage, individual participants were randomly selected from registries of daily elderly care centers using their national identification and center membership codes. Sampling quotas in each city were proportionally allocated based on the relative size of the elderly population in that area. The sample size was determined using PASS software version 15, with a 5% margin of error and an anticipated 10% dropout rate, following psychometric validation recommendations. Methodological guidance was drawn from Hogg et al. [15], whose validation study emphasized adequate sample sizes for both exploratory and confirmatory factor analyses. After accounting for incomplete or inconsistent responses and quality control procedures, a final sample of 279 participants was retained for analysis.

Inclusion criteria were: being aged 60 or older, residing in the designated city for at least 10 years, and possessing sufficient cognitive capacity to understand the study and complete the instruments. Exclusion criteria included voluntary withdrawal, incomplete questionnaire data, significant inconsistencies in responses, prior neurological surgery, or non-residency within the sampling frame. Although replacement protocols were established in case of participant unavailability or mortality, no such instances occurred during the study period.

Data collection was conducted through face-to-face interactions by trained interviewers with nursing qualifications. These interviewers received extensive training on the study’s objectives, inclusion/exclusion criteria, administration of tools, and communication strategies. Questionnaires were either self-administered or interviewer-assisted, depending on participants’ literacy levels, and the average completion time was approximately 20 min. Participants completed a battery of self-report questionnaires that included the three climate anxiety instruments (HCAS, CCCA, and SCAS), the GAI-SF, and demographic variables. Instruments were administered in Persian and data were collected under researcher supervision to ensure item clarity and completeness, and all responses were recorded anonymously. Written informed consent was obtained from all participants prior to data collection. The study was approved by the Research Ethics Committee of Shiraz University of Medical Sciences. Explicit measures were taken to ensure ethical standards, including data anonymization (e.g., use of ID codes and encrypted storage) and assurance of the right to withdraw at any stage without penalty.

Instruments

This study rigorously followed the STROBE guidelines, the COSMIN checklist for evaluating the methodological quality of studies on measurement properties, and adhered strictly to the practical framework for translation, cross-cultural adaptation, and validation of instruments modeled by Fekih-Romdhane et al. [9]. All procedures conformed to the psychometric standards commonly required in high-quality validation research.

Translation & cross-cultural adaptation

The three instruments underwent forward-backward translation by two independent bilingual experts, with discrepancies resolved by a panel. Pilot testing with 20–30 target participants confirmed comprehension and cultural relevance. Quantitative measures, including Content Validity Ratio (CVR) and Index (CVI), were calculated based on expert ratings, with CVI ≥ 0.8 for all instruments indicating acceptable content validity.

Four instruments were used in this study

1. Hogge Climate Anxiety Scale (HCAS): The HCAS comprises 13 items rated on a 4-point Likert scale (0–3), developed specifically to assess climate-related distress among older adults. Total scores range from 0 to 39, with higher scores reflecting more severe climate anxiety. The instrument underwent a forward-backward translation process and was culturally adapted following established guidelines to ensure semantic, conceptual, and operational equivalence in the target population. The internal consistency of the HCAS in this study, as assessed by McDonald’s ω = 0.84, was very good, indicating strong reliability of the scale in the target population.

2. Clayton Climate Change Anxiety Scale (CCCA): The CCCA contains 24 items measured on a 5-point Likert scale (1–5), capturing emotional, cognitive, and behavioral dimensions of climate change anxiety. Scores range from 24 to 120. Cultural and linguistic adaptation was carried out meticulously, and expert panel review ensured content relevance and clarity for older Iranian adults. The scale demonstrated excellent internal consistency, with McDonald’s ω = 0.95, indicating a very high level of reliability for research purposes.

3. Simon Climate Anxiety Scale (SCAS): This 13-item scale also uses a 5-point Likert format (1–5) and assesses general emotional responses and worries associated with climate change. The total score ranges from 13 to 65. Translation and cultural validation followed psychometric best practices, with pilot testing conducted to assess face validity and respondent comprehension. The SCAS showed exceptional internal consistency in this study, with McDonald’s ω = 0.97, highlighting the instrument’s robustness for assessing climate anxiety in older adults.

4. Geriatric Anxiety Inventory–Short Form (GAI-SF): The form with 5 dichotomous (Yes/No) items and classifies respondents as anxious or non-anxious, was used as the criterion measure to classify participants into anxious and non-anxious groups. Based on the Persian version validated for older adults by Shati et al. [29], a cut-off score of ≥ 2 was applied to identify individuals with clinically significant anxiety symptoms. This threshold has demonstrated adequate sensitivity and specificity in previous studies conducted among Iranian geriatric populations, ensuring its appropriateness for the current analysis. The GAI-SF was selected as an external reference criterion based on the theoretical overlap between climate anxiety and broader geriatric anxiety processes. Climate anxiety is conceptualized as a domain-specific manifestation of transdiagnostic anxiety mechanisms, including excessive worry, anticipatory threat perception, cognitive rumination, and affective distress. While the GAI-SF does not represent a climate-specific diagnostic gold standard, it provides a clinically validated external criterion reflecting shared affective dimensions relevant to anxiety in older adults. Therefore, its use in the present study should be interpreted as a criterion-related classification comparison rather than a definitive diagnostic validation. Accordingly, associations with the GAI-SF in the present study are interpreted as evidence of convergent and criterion-related validity with a theoretically related construct, rather than as validation against a construct-identical diagnostic standard.

Statistical software and analysis packages

All statistical analyses were conducted using JAMOVI (version 2.6.23) and R (version 4.4.0), utilizing several specialized packages for psychometric, inferential, and classification analyses [17, 24].

• Descriptive statistics and group comparisons (t-tests, Brunner–Munzel tests) were performed using Jamovi base modules and the ClinicoPath module [4].

• Linear regression, Pearson correlations, and logistic regression (odds ratios) were performed using the finalfit and car packages [11, 14].

• Exploratory and Confirmatory Factor Analyses (EFA & CFA) were conducted using the psych and lavaan packages [25, 27]. The dataset was randomly divided into two independent subsamples to ensure the independence of exploratory and confirmatory analyses. The EFA was conducted on the first subsample (n = 139), while the CFA was performed on the second subsample (n = 140). This split-sample approach aligns with established psychometric guidelines and ensures cross-validation of the factor structure, thereby strengthening the evidence for construct validity. The EFA and the CFA were conducted using Maximum Likelihood (ML) estimation. This method was chosen due to the continuous nature of the data and the confirmation of multivariate normality through preliminary diagnostics, including skewness, kurtosis, and Mardia’s test. ML estimation is widely recommended for its robustness and efficiency under these conditions, ensuring reliable parameter estimates and model fit evaluation. Although the subsample size for the 24-item CCCA (n = 140 for CFA) may be considered modest, it remains within widely accepted psychometric guidelines recommending 5–10 participants per item when factor loadings and communalities are adequate. In the present study, sampling adequacy indices (KMO = 0.872; Bartlett’s test p < 0.001) and satisfactory factor loadings supported the stability of the extracted structure. Nonetheless, we recognize that larger subsamples would further enhance factor stability, particularly for instruments with a higher number of items.

• Bayesian hypothesis testing and Bayes Factors were calculated using the BayesFactor package [21, 28].

• The performance indices (e.g., sensitivity, specificity, accuracy, precision, F1 score, MCC, DOR, YJI) were calculated using the MLmetrics package [37].

• Rasch modeling and Item Response Theory (IRT) analysis under the Partial Credit Model (PCM) were implemented using TAM and eRm packages [20, 26].

• Mokken scale analysis was conducted via the mokken and mirt packages [5, 36].

• Measurement invariance analysis was conducted using lavaan, with model comparisons reported via chi-square difference tests and fit indices (CFI, TLI, RMSEA, and SRMR) [27].

Additional psychometric analyses

Differential Item Functioning (DIF) was assessed using Mantel-Haenszel or Rasch DIF methods to check for item bias across gender and age groups. Responsiveness was evaluated where longitudinal data existed.

All statistical decisions were made at a significance threshold of p < 0.05, unless otherwise specified. Model assumptions were checked and addressed where necessary.

Ethical considerations

This study was approved by the Research Ethics Committee of Shiraz University of Medical Sciences, Iran, under the ethics code IR.SUMS.SCHEANUT.REC.1404.119, dated November 30, 2025. All procedures involving human participants complied with the ethical standards of the institutional and/or national research committee and with the Helsinki Declaration (2013 revision and its 2020 update). Prior to participation, written informed consent was obtained from all participants after providing full explanations of the study’s aims, procedures, and confidentiality assurances. Data anonymization was ensured through the use of ID codes and encrypted storage. Participants were explicitly informed of their right to withdraw at any stage without penalty. Sampling limitations included restricted representation, as the study focused on urban Iranian elderly populations. Methodological biases, such as those related to self-report measures (e.g., social desirability), were acknowledged as potential limitations.

Results

Descriptive statistics summarized psychological measures in 279 older adults (See Table 1). Mean scores were: YO = 73.05, CCCA = 62.65, HCAS = 16.55, SCAS = 36.74, and GAI-SF = 2.32. Gender comparisons showed subtle differences. Females scored slightly higher than males on CCCA (64.17 vs. 60.75), SCAS (38.12 vs. 35.02), and YO (73.81 vs. 72.10), while HCAS (16.73 vs. 16.41) and GAI-SF (2.25 vs. 2.37) were similar. Independent t-tests revealed no significant gender differences except for SCAS (p = 0.043), where females reported higher climate anxiety. Effect sizes were small, suggesting minimal practical impact.

Table 1.

Descriptive statistics and group comparisons for psychological measures in older individuals

Variables	Group	Mean	SD	t-value	95% CI for t-value	p-value	Mean difference	Bayes Factor₁₀	Effect Size (Cohen’s d)	95% CI for Effect Size
CCCA	Non-anxious (GAI-SF = 0)	57.9	29.18	-2.972	-17.03 to -3.46	0.003	-10.24	8.433	-0.357	-0.594 to -0.119
	Anxious (GAI-SF = 1)	68.2	28.13
HCAS	Non-anxious (GAI-SF = 0)	15.7	6.62	-2.645a	-3.29 to -0.48	0.009	-1.89	3.578	-0.318	-0.554 to -0.081
	Anxious (GAI-SF = 1)	29.6	5.03
SCAS	Non-anxious (GAI-SF = 0)	36	13.35	-0.993	-4.53 to 1.49	0.002	-1.52	4.211	-0.119	-0.355 to 0.117
	Anxious (GAI-SF = 1)	47.6	11.98
YO	Non-anxious (GAI-SF = 0)	72.4	9.65	-1.244	-3.80 to 0.86	0.214	-1.47	0.275	-0.149	-0.385 to 0.086
	Anxious (GAI-SF = 1)	73.8	10.05

H_a µ0 ≠ µ1

^a Levene’s test is significant (p < 0.05), suggesting a violation of the assumption of equal variances.

Comparisons by anxiety status (GAI-SF) showed clearer differences. Anxious participants (GAI-SF = 1) scored higher on CCCA (68.2 vs. 57.9), HCAS (17.6 vs. 15.7), and SCAS (37.6 vs. 36.0) compared to non-anxious participants (GAI-SF = 0). YO scores were slightly higher in the anxious group (73.8 vs. 72.4). These differences were statistically significant for CCCA (p = 0.003) and HCAS (p = 0.009), with moderate effect sizes (Cohen’s d = -0.119 to -0.594). The Brunner-Munzel test confirmed these findings, showing significant gender differences only in SCAS (p = 0.026) and higher climate anxiety scores among anxious participants. These results reinforce the importance of considering anxiety status in older adults when assessing climate-related concerns.

Linear regression analyses indicated that age predicted some, but not all, anxiety measures. Age explained 3.7% of HCAS variance (R² = 0.0368, p = 0.001), 27.8% of CCCA variance (R² = 0.278, p < 0.001), and 38.9% of SCAS variance (R² = 0.389, p < 0.001). However, age was not a significant predictor of GAI-SF scores (R² = 0.00834, p = 0.128). Overall, age appears to influence climate-related anxiety but not general geriatric anxiety.

Correlation matrix

To examine convergent validity, or more broadly criterion-related validity, the correlation between HCAS, CCCA, and SCAS tools and the established measure GAI-SF was assessed. The correlation matrix reveals that the Hogge Climate Anxiety Scale (HCAS) shows a moderate positive correlation with GAI-SF (r = 0.443, p = 0.017), indicating a statistically significant relationship. Similarly, the Clayton Climate Change Anxiety Scale (CCCA) exhibits a stronger positive correlation with GAI-SF (r = 0.411, p < 0.001), suggesting a robust alignment with the criterion measure. The Simon Climate Anxiety Scale (SCAS), on the other hand, demonstrates a strong positive correlation with GAI-SF (r = 0.571, p = 0.003), implying a substantial association between SCAS and the established anxiety measure. These findings highlight varying degrees of convergent validity for the new tools relative to GAI-SF, with CCCA and SCAS showing particularly strong relationships, while HCAS exhibits a moderate but significant correlation.

Exploratory and confirmatory factor analysis

The suitability of the data for factor analysis was confirmed. Bartlett’s Test of Sphericity was significant for all three scales (HCAS: χ² = 3369; SCAS: χ² = 2450; CCCA: χ² = 4130; p < 0.001), and KMO values indicated excellent sampling adequacy (HCAS = 0.895; SCAS = 0.909; CCCA = 0.872). Inter-item correlations were mostly above 0.30 (p < 0.05), supporting the assumption of linearity. Oblique (OBLIMIN) rotation allowed factors to correlate and facilitated interpretation.

Hogge Climate Anxiety Scale (HCAS)

Exploratory Factor Analysis (EFA) identified a two-factor structure for the HCAS: Factor 1 (HCAS5–13, loadings 0.321–0.454) and Factor 2 (HCAS1–4, loadings 0.286–0.395). Items with factor loadings below 0.3 were flagged for potential refinement due to their contribution to measurement error and limited explained variance. Confirmatory Factor Analysis (CFA) supported this two-factor model, demonstrating acceptable fit: RMSEA = 0.0285, CFI = 0.900, TLI = 0.885, χ² = 65.0, p = 0.125. All retained items showed significant loadings (p < 0.001). While the current structure is preliminarily supported, further item refinement and validation in independent samples are recommended to strengthen structural validity.

Simon Climate Anxiety Scale (SCAS)

EFA identified three factors: Factor 1 (SCAS1–5, SCAS7), Factor 2 (SCAS6, 11–13), and Factor 3 (SCAS8–10), with strong factor loadings (0.343–0.997). CFA indicated acceptable fit (RMSEA = 0.041, CFI = 0.898, TLI = 0.872, χ² = 268, p = 0.241). Factors were moderately to strongly correlated (r = 0.671–0.814), and the scale showed good internal consistency.

Clayton Climate Change Anxiety Scale (CCCA)

EFA suggested two factors: Factor 1 (CCCA4, 8–20, loadings 0.351–0.982) and Factor 2 (CCCA1–3, 6–7, 21–24, loadings 0.393–0.870). CFA supported this structure with acceptable fit (RMSEA = 0.059, CFI = 0.874, TLI = 0.862, χ² = 1702, p = 0.421). Factors were strongly correlated (r = 0.845), and loadings were significant (p < 0.001). Internal consistency was adequate, though fit indices suggest some room for improvement.

Overall, all three scales demonstrated factor structures aligned with their theoretical frameworks and adequate reliability. HCAS and CCCA showed clear two-factor patterns, while SCAS reflected a more complex three-factor structure. These results support the validity of using these instruments to measure climate-related anxiety in older adults.

Odds ratio table and plot

Logistic regression was used to examine the associations between scores on each climate anxiety scale and geriatric anxiety status. For the Clayton Climate Change Anxiety Scale (CCCA), each one-point increase in score slightly raised the odds of being anxious by 1% (unadjusted OR = 1.01, 95% CI: 1–1.02, p = 0.004), and this effect remained after accounting for other factors (adjusted OR = 1.01, 95% CI: 1–1.02, p = 0.013). The Hogge Climate Anxiety Scale (HCAS) showed a stronger link: each point increase was associated with roughly 5–6% higher odds of anxiety (unadjusted OR = 1.06, 95% CI: 1.01–1.1, p = 0.01; adjusted OR = 1.05, 95% CI: 1.01–1.09, p = 0.027) (See Table 2).

Table 2.

Odds ratios for climate anxiety scales predicting geriatric anxiety status

Dependent: GAI-SF (2)		0 (Anxious)	1 (Non-Anxious	OR (Univariable)	OR (Multivariable)
CCCA	Mean (SD)	57.9 (29.2)	68.2 (28.1)	1.01 (1-1.02, p = 0.004)	1.01 (1-1.02, p = 0.013)
HCAS	Mean (SD)	15.7 (6.6)	17.6 (5)	1.06 (1.01–1.1, p = 0.01)	1.05 (1.01–1.09, p = 0.027)
SCAS	Mean (SD)	36 (13.3)	37.6 (12)	1.01 (0.99–1.03, p = 0.321)	1.00 (0.98–1.02, p = 0.89)

Model Metrics: Number in data-frame = 279, Number in model = 279

Missing = 0, AIC = 379.3, C-statistic = 0.841, H&L = Chi-sq(8) 15.99 (p = 0.043)

In contrast, the Simon Climate Anxiety Scale (SCAS) did not show a statistically significant association in the model (adjusted OR = 1.00, 95% CI: 0.98–1.02, p = 0.89). The multivariable logistic regression model containing all three climate anxiety scales showed a C-statistic of 0.841. The Hosmer-Lemeshow test indicated an acceptable model fit (χ² = 15.99, df = 8, p = 0.043) (See Fig. 1).

Fig. 1.

Odds ratios plot for climate anxiety scales

Bland-Altman analysis

The Bland-Altman analysis compared three climate anxiety scales (HCAS, CCCA, and SCAS) against the GAI-SF criterion scale for climate anxiety. For CCCA (n = 279), bias was 60.33 (95% CI: 56.94–63.73), with limits of agreement 3.90 to 116.76. HCAS showed a lower bias of 14.24 (95% CI: 13.55–14.92) and narrower agreement limits (2.82–25.66). SCAS had intermediate bias (34.43, 95% CI: 32.93–35.9) with limits of 9.48–59.37. HCAS demonstrated the least bias and narrowest agreement range, suggesting better alignment with GAI-SF for assessing anxiety in elderly populations, while CCCA exhibited wider variability and higher bias (See Table 3).

Table 3.

Bland-Altman analysis of CCCA, HCAS, and SCAS to compare agreement to criterion scale

Instruments	Bias (95% CI)	Lower Limit of Agreement (95% CI)	Upper Limit of Agreement (95% CI)
CCCA	60.33 (56.94–63.73)	3.9 (-1.90–9.71)	116.76 (110.96–122.57)
HCAS	14.24 (13.55–14.92)	2.82 (1.64–3.99)	25.66 (24.48–26.83)
SCAS	34.43 (32.93–35.9)	9.48 (6.91–12.1)	59.37 (56.81–61.9)

According to Table 3, the Bland-Altman analysis reveals that HCAS exhibits the lowest bias (14.24) and narrowest limits of agreement (2.82–25.66), indicating closest alignment with the GAI-SF criterion scale for aging climate anxiety. SCAS shows intermediate performance (bias = 34.43, limits = 9.48–59.37), while CCCA demonstrates the highest bias (60.33) and widest variability (3.90–116.76). Thus, HCAS is the most suitable tool for assessing climate anxiety in older populations due to its precision and minimal deviation from the criterion measure.

IRT analysis with Polytomous Rasch model for HCAS, CCCA, and SCAS

The psychometric properties of the three climate anxiety scales were further evaluated using a Polytomous Rasch model (Partial Credit Model, PCM). Detailed item statistics are presented in Supplemental Online Table 1.

For the Clayton Climate Change Anxiety Scale (CCCA), the PCM showed excellent person reliability (0.957) and a significant overall model fit (MADaQ3 = 0.154, p < 0.001). The PCM provided a better fit than the Rating Scale Model (RSM) (χ²(92) = 306, p < 0.001). While most items demonstrated acceptable fit (infit/outfit range: 0.5–1.5), items 4, 6, 10, 12, and 16 showed potential misfit.

For the Hogge Climate Anxiety Scale (HCAS), the model indicated good person reliability (0.693) and an acceptable fit to the Rasch model (MADaQ3 = 0.051, p = 0.196). The PCM and RSM did not differ significantly in fit (χ²(24) = 32.7, p = 0.111). All HCAS items showed satisfactory fit within the acceptable range.

For the Simon Climate Anxiety Scale (SCAS), the analysis revealed high person reliability (0.936) and a significant model fit (MADaQ3 = 0.150, p < 0.001), with the PCM fitting the data better than the RSM (χ²(48) = 117, p < 0.001). Most items fitted well, though items 6, 10, and 12 showed slight misfit. Thurstone thresholds for all scales, detailed in the supplemental table, provided insights into item response categories.

Classification performance metrics relative to the GAI-SF external criterion are summarized in Table 4. The Simon Climate Anxiety Scale (SCAS) demonstrated the highest sensitivity (0.89) and Diagnostic Odds Ratio (DOR = 3.09). The Clayton Climate Change Anxiety Scale (CCCA) showed marginally higher specificity (0.36) than the other tools, though specificity was modest across all instruments (HCAS = 0.28, SCAS = 0.27).

Table 4.

Performance measures of climate anxiety instruments comparing to GAI-SF

Measures	HCAS	CCCA	SCAS	P a	Derivations
Sensitivity	0.86	0.80	0.89	< 0.001	TPR = TP / (TP + FN)
Specificity	0.28	0.36	0.27	< 0.001	SPC = TN / (FP + TN)
Negative Predictive Value	0.70	0.68	0.75	0.021	NPV = TN / (TN + FN)
False Positive Rate	0.72	0.64	0.73	< 0.001	FPR = FP / (FP + TN)
False Discovery Rate	0.49	0.48	0.49	0.012	FDR = FP / (FP + TP)
False Negative Rate	0.14	0.20	0.11	< 0.001	FNR = FN / (FN + TP)
Accuracy	0.55	0.56	0.56	0.002	ACC = (TP + TN) / (P + N)
Precision	0.51	0.52	0.51	0.008	PPV = TP / (TP + FP)
Diagnostic Odds Ratio	2.40	2.23	3.09	< 0.001	DOR = (TP/FN)/(FP/TN)
Likelihood Ratio+	1.20	1.25	1.23	0.004	LR + = TPR / 1- SPC
Youden’s J Index	0.14	0.16	0.16	< 0.001	YJI = (TPR + SPC) – 1
F1 Score	0.64	0.63	0.65	< 0.001	F1 = 2TP / (2TP + FP + FN)
Matthews Correlation Coefficient	0.17	0.17	0.21	< 0.001	TP*TN - FP*FN / sqrt((TP + FP)*(TP + FN)*(TN + FP)*(TN + FN))

Using MLmetrics Package in R (version 1.1.3, 2024)

^a Using ANOVA, P < 0.05

Simon-SCAS demonstrated the highest sensitivity (0.89) and diagnostic odds ratio (DOR = 3.09), suggesting utility in ruling out anxiety when scores are low. However, specificity values were modest across all tools (HCAS = 0.28, CCCA = 0.36, SCAS = 0.27), indicating limited accuracy in identifying true non-anxious individuals. The overall discriminative ability, as reflected by the C-statistic (0.841), was good. These findings highlight that while SCAS may be effective for initial screening in older adults, it should be complemented with clinical evaluation to improve diagnostic certainty.

These findings indicate that although sensitivity was high—particularly for SCAS—the relatively low specificity across all three instruments (0.27–0.36) reflects a substantial false-positive rate. Therefore, these tools should not be interpreted as definitive clinical case-identification instruments. Rather, they function more appropriately as high-sensitivity indicators of elevated climate-related distress. Their utility lies primarily in early risk detection, population-level research screening, and identification of individuals who may benefit from further psychological evaluation, rather than formal diagnostic classification.

Overall accuracy was comparable between CCCA and SCAS (0.56), with HCAS slightly lower (0.55). The SCAS also achieved the highest Matthews Correlation Coefficient (MCC = 0.21) and F1 score (0.65), while all scales showed similar precision (~ 0.52). ANOVA results indicated statistically significant differences between the instruments across most performance metrics (p < 0.05), supporting the differential utility of each scale based on the assessment objective (See Fig. 2).

Fig. 2.

Person-item map and wright map for partial credit model of for Hogge-HCAS, Clayton-CCCA, and Simon-SCAS

Mokken analysis results for climate anxiety instruments

The Mokken analysis results for the climate anxiety instruments—HCAS, CCCA, and SCAS—demonstrate varying degrees of scalability, reliability, and internal consistency. The Loevinger’s H coefficients for all instruments (HCAS: 0.45, CCCA: 0.52, SCAS: 0.4) indicate acceptable scalability, with CCCA showing the strongest scalability. The item-level Loevinger’s H coefficients (H_i) were above the acceptable threshold of 0.3 for most items across all instruments, confirming their adherence to Mokken scale requirements. Reliability, as measured by Cronbach’s alpha (HCAS: 0.88, CCCA: 0.92, SCAS: 0.85) and McDonald’s ω (HCAS: 0.84, CCCA: 0.95, SCAS: 0.97), was consistently high, suggesting robust internal consistency. Additionally, Rho values (HCAS: 0.81, CCCA: 0.87, SCAS: 0.84) further support the instruments’ reliability.

The scalability coefficients (HSCALE) were high, with CCCA again performing the best. Non-significant p-values for monotonicity and local independence tests (p > 0.05) indicate that the items meet the assumptions of monotonicity and local independence, reinforcing the validity of the scales. Overall, the CCCA instrument stands out as the most robust, while all three instruments are psychometrically sound for assessing climate anxiety (See Table 5).

Table 5.

Mokken analysis results for climate anxiety instruments

Instruments	Loevinger’s H Coefficient (H)	Loevinger’s H Coefficient (Hi)	Cronbach’s Alpha	Rho (Reliability)	McDonald’s ω	Scalability Coefficient (H SCALE)	Monotonicity (p-value)	Local Independence (p-value)
HCAS	0.45	> 0.3 for most items	0.88	0.81	0.84	0.42	p > 0.05	p > 0.05
CCCA	0.52	> 0.3 for all items	0.92	0.87	0.95	0.51	p > 0.05	p > 0.05
SCAS	0.4	> 0.3 for most items	0.85	0.84	0.97	0.38	p > 0.05	p > 0.05

Measurement invariance analysis across gender for HCAS, CCCA, and SCAS scales

The measurement invariance analysis across gender groups for the HCAS, CCCA, and SCAS scales reveals varying levels of support for different levels of invariance. For HCAS, metric and strict invariance are supported with non-significant changes in chi-square and minimal decreases in fit indices such as CFI. However, scalar invariance shows only partial support, indicated by a significant change in chi-square and a moderate drop in CFI. Similarly, for CCCA, metric and strict invariance are supported, but scalar invariance is only partially supported. In contrast, SCAS demonstrates full support across all levels of invariance, including scalar and strict, with non-significant chi-square differences and stable fit indices like TLI and RMSEA. These findings suggest that while SCAS can be reliably compared between male and female groups, caution is needed when interpreting results for HCAS and CCCA due to partial scalar invariance (See Supplemental Online Table 2).

Measurement invariance analysis across GAI groups for HCAS, CCCA, and SCAS scales

The measurement invariance analysis across GAI groups for the HCAS CCCA and SCAS scales reveals varying levels of support. For HCAS metric and strict invariance are supported with changes in chi-square being non-significant and small decreases in fit indices like CFI However scalar invariance shows only partial support as indicated by a significant change in chi-square and a notable drop in CFI For CCCA similar patterns are observed where metric and strict invariance are supported but scalar invariance is only partially supported In contrast SCAS demonstrates full support across all levels of invariance including scalar and strict with non-significant chi-square differences and stable fit indices such as TLI and RMSEA These findings suggest that while SCAS can be reliably compared between anxious and non-anxious groups caution is needed when interpreting results for HCAS and CCCA due to partial scalar invariance (See Supplemental Online Table 3).

Discussion

This study contributes substantially to the limited but growing body of research on climate anxiety assessment among older adults, particularly in non-Western contexts [8, 10]. By rigorously evaluating the Hogge Climate Anxiety Scale (HCAS), Clayton Climate Change Anxiety Scale (CCCA), and Simon Climate Anxiety Scale (SCAS), we identified distinct psychometric strengths and limitations of each instrument in a sample of Persian-speaking older adults in Iran. The SCAS emerged as the most psychometrically robust tool, demonstrating superior sensitivity, internal consistency, and full scalar invariance across both gender and anxiety groups. These findings are consistent with recent psychometric research emphasizing the need for precise and equitable mental health measures in aging populations experiencing environmental threats [2, 9, 12].

Beyond substantive findings, this study contributes methodologically by demonstrating the value of integrating multiple psychometric paradigms within a single validation framework. The combined use of CFA, IRT/Rasch modeling, Mokken analysis, and measurement invariance testing allows for a more rigorous and multidimensional evaluation of climate anxiety scales, particularly in aging populations where measurement bias and differential item functioning are critical concerns. The measurement invariance analysis revealed a critical distinction among the scales. While the SCAS demonstrated full scalar invariance, indicating that its latent factor means can be meaningfully and directly compared across gender and anxiety subgroups, the HCAS and CCCA achieved only partial scalar invariance. This finding suggests that for HCAS and CCCA, some item intercepts (baseline response tendencies) differ between groups. Consequently, observed mean score differences on these scales across genders or between anxious and non-anxious individuals may not solely reflect true differences in the underlying level of climate anxiety, but could also be influenced by group-specific response patterns to certain items. Therefore, cross-group comparisons using HCAS and CCCA scores should be made with caution, and inferences about the magnitude of group differences may be less robust than those made with the SCAS.

Notably, HCAS demonstrated the lowest measurement bias in Bland–Altman analysis and satisfactory reliability, suggesting its value in clinical settings where diagnostic precision is essential. The CCCA showed high internal consistency and strong scalability (Loevinger’s H = 0.52), yet exhibited wider variability and only partial measurement invariance, raising concerns regarding its consistency across subgroups. These differences underscore the critical importance of selecting culturally adapted and psychometrically sound instruments for assessing climate-related psychological distress in older adults [2, 9, 18].

Although younger populations are more frequently studied in the context of climate anxiety [7, 22, 33, 34], our findings highlight that older adults—who often experience reduced adaptive capacity, chronic illness, and limited social support [6, 10]—are particularly vulnerable to environmental stressors. Integrating age-sensitive tools into mental health services and climate adaptation strategies is thus imperative [2, 10, 12].

Overall, the SCAS demonstrated the most robust psychometric properties in terms of internal consistency, factor structure, and full measurement invariance, making it a particularly strong candidate for research aiming to compare climate anxiety levels across different subgroups (e.g., gender, clinical status) within older Persian-speaking populations. However, psychometric robustness does not necessarily translate into clinical diagnostic precision. Importantly, the classification performance pattern observed in this study—characterized by high sensitivity but modest specificity—suggests that these instruments are better conceptualized as distress-detection tools rather than clinical diagnostic screeners. In geriatric community settings, such high-sensitivity measures may be advantageous for identifying individuals at elevated psychosocial risk; however, the elevated false-positive rate limits their suitability for standalone clinical decision-making. Accordingly, these scales should be used as part of a broader assessment framework rather than as independent diagnostic instruments.

However, instrument selection should be guided by specific assessment goals. The HCAS, with its low bias against the GAI-SF, may be preferable in clinical or epidemiological contexts where minimizing systematic over- or under-estimation relative to a general anxiety external criterion is a priority. The CCCA, with its high scalability and broad content coverage, might be more suitable for studies interested in the multifaceted nature of climate change anxiety. Thus, the choice of scale should align with the study’s specific objectives—whether they prioritize group comparisons (SCAS), alignment with general anxiety measures (HCAS), or comprehensive domain coverage (CCCA). Furthermore, the good overall discriminative validity of a model combining these scales (C-statistic = 0.841) supports the relevance of the climate anxiety construct in this population, although the low specificity of each scale individually underscores that they are better suited for screening than for definitive diagnosis.

Study limitations

Despite its strengths, this study has several limitations. First, the sample was restricted to urban regions of southern Iran, which may limit the generalizability of findings to rural or other geographic and socio-economic groups. Future studies should include rural and illiterate populations to increase ecological validity. Second, the reliance on self-report instruments introduces possible biases, including social desirability and recall inaccuracies. Although trained interviewers mitigated comprehension challenges, interview assistance may have influenced responses. Third, the cross-sectional design limits inferences about causality or temporal changes in climate anxiety. Furthermore, while the translation and cultural adaptation process adhered to best practices [9, 18], subtle nuances in interpretation may remain. Fourth, Furthermore, while the Geriatric Anxiety Inventory–Short Form (GAI-SF) provided a useful criterion for general anxiety, it is not a direct measure of climate-specific distress. Conceptual differences exist between general anxiety and climate anxiety; the latter encompasses worries about environmental future, guilt, and perceived helplessness related to ecological threats, which may not fully align with the geriatric anxiety symptoms captured by the GAI-SF. This conceptual discrepancy may have influenced the diagnostic accuracy metrics, potentially contributing to the modest specificity and discriminative ability observed. Future validation studies would benefit from using a clinician-rated interview or a criterion measure specifically designed for eco-anxiety to enhance diagnostic comparisons. Additionally, although the split-sample design strengthens construct validation, the subsample size used for factor analysis of the 24-item CCCA was relatively modest. While statistical adequacy indices supported model stability, larger samples would provide greater precision and reduce potential sampling variability in future validations. Furthermore, as emphasized in recent diagnostic accuracy reviews (e.g., Atchison et al., 2024), geriatric anxiety screening tools should not be considered definitive diagnostic standards. Thus, associations observed in the present study reflect theoretical convergence rather than construct-equivalent validation. Finally, although we assessed convergent validity using the GAI-SF, future validation efforts should explore broader clinical and ecological correlates.

Final note: implications for older adults and future directions

This study underscores the urgent need to address climate-related psychological distress in older adults, who are often overlooked in climate change discourse [8, 10]. The identification of SCAS as the most valid and reliable measure among the three tested instruments offers practical implications for geriatric mental health screening and intervention design. Incorporating validated tools such as SCAS into geriatric assessment workflows can facilitate early identification of elevated climate-related distress and psychosocial risk, prompting timely follow-up assessment rather than serving as standalone diagnostic tools. Likewise, the low bias of HCAS suggests its utility in epidemiological surveillance where overestimation is a concern. These instruments could be valuable for NGOs, health ministries, or aging services developing climate-resilience programs tailored to the psychosocial needs of elderly populations. Future efforts should explore how intersecting factors—such as gender, income, education, and pre-existing mental health conditions—shape older adults’ experiences of climate anxiety. Longitudinal and mixed-method studies are also recommended to explore causal pathways and coping strategies over time [8]. Such research is critical for developing targeted, evidence-based interventions for this vulnerable group.

Recommendations for future research and policy

Based on our findings, we propose the following recommendations:

• Validation in Diverse Contexts: Future studies should replicate these findings in other cultural, linguistic, and geographic settings. This includes rural, nomadic, or displaced populations who may experience climate anxiety differently.

• Mixed-Methods Approaches: Combining qualitative inquiry with psychometric validation could clarify how older adults conceptualize climate anxiety and reveal culturally specific expressions of distress [32].

• Policy Integration: Mental health assessments—including climate anxiety screening—should be integrated into national aging policies and climate adaptation frameworks.

• Community-Based Interventions: Programs that blend climate literacy with mental health support could help older adults build resilience and reduce perceived helplessness.

• Interdisciplinary Collaboration: Coordinated efforts across gerontology, mental health, environmental science, and social work are essential to design holistic support systems for older adults in the face of accelerating climate threats [1].

• Investment in Long-Term Research: Funders should support longitudinal cohort studies exploring the long-term psychological, cognitive, and physiological impacts of climate anxiety in older populations [8].

Conclusion

This study provides novel insights into the psychometric performance of three climate anxiety instruments among Persian-speaking older adults. The SCAS demonstrated the strongest overall psychometric properties, particularly for diagnostic accuracy, internal consistency, and full measurement invariance, making it a robust choice for studies requiring group comparisons. Conversely, the HCAS showed minimal bias against the general anxiety criterion, positioning it as a suitable tool for contexts where alignment with broad anxiety screening is prioritized. The CCCA, while exhibiting strong scalability and comprehensive construct coverage, requires careful interpretation in cross-group comparisons due to its partial invariance. Collectively, these findings suggest that instrument selection should be guided by specific assessment objectives rather than seeking a universally superior tool. Overall, this research reinforces the importance of using culturally adapted, age-appropriate instruments to assess psychological responses to environmental stress. As climate change intensifies, mental health must be recognized as a core component of climate resilience—especially among aging populations who may lack the physical, social, or institutional resources to cope [6, 12]]. Advancing research, refining context-specific diagnostic instruments, and embedding mental health within climate adaptation policies will be crucial to safeguarding the psychological well-being of older adults in an era of environmental uncertainty.

Authors’ contributions

FB, SHM, & TGh have collected the data, interviewed to the samples, and written *Introduction* section, HH has written *Discussion* section, AAs analyzed the data and wrote the *Method* and *Results* sections. AAb and HH have made comments and corerctions in final analysis and preparing the final format of manuscript. All authors read and approved the final manuscript.

Funding

None.

Data availability

The datasets generated and/or analysed during the current study are available from the authors upon reasonable request and with the permission of SUMS.

Declarations

Ethics approval and consent to participate

All ethical considerations, including but not limited to plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, and redundancy, were strictly adhered to by the authors. The sole female participant in this study was provided with a verbal explanation regarding the study’s objectives and procedures. Subsequently, she signed a written informed consent form to participate in the study. The participant was assured of the anonymity and confidentiality of her information. Additionally, the research protocol was approved by the Ethics Committee of Shiraz University of Medical Sciences on November 30, 2025 (Ethics Code: IR.SUMS.SCHEANUT.REC.1404.119).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.