Environmental Institutional Determinants of Climate Behavior Among Taiwan’s Public Officials

Abstract

This study investigates how climate change literacy (CCL) and institutional contexts shape the climate-related behaviors of Taiwan’s public officials. Drawing on a 2024 national survey of 1940 civil servants, we apply hierarchical and comparative regression analyses to examine the relative influence of knowledge, affective dispositions, and organizational supports. Results show that solution-oriented knowledge exerts greater behavioral influence than factual awareness. At the same time, affective resources—particularly self-efficacy and environmental identity—are the strongest and most consistent drivers of engagement. Institutional factors further condition these relationships: central officials’ behaviors are shaped by departmental mandates and bureaucratic constraints, whereas local officials rely more on supervisor support and prior project involvement. These findings integrate literacy research with institutional perspectives, demonstrating that effective climate governance requires both individual agency and enabling organizational contexts. Policy implications include strengthening leadership training, creating experiential learning opportunities, and streamlining administrative structures across governance levels to accelerate climate action.

1. Introduction

Climate change is among the defining challenges of the twenty-first century, with far-reaching environmental, social, and economic consequences [1,2,3]. Meeting this challenge requires not only technological innovation and robust policy frameworks but also the active engagement of public officials, who translate statutory visions into administrative practice [4,5]. International initiatives such as the European Green Deal (EGD) demonstrate how comprehensive governance frameworks can accelerate renewable energy adoption, mainstream circular economy policies, and enhance regional leadership [6,7]. These experiences highlight administrative capacity as central to achieving carbon neutrality. In Taiwan, the Climate Change Response Act similarly emphasizes the need for institutional readiness and governance capacity to realize long-term climate objectives [8,9].

1.1. Climate Change Literacy and the Knowledge-Behavior Gap

Individual responses to climate change shape both mitigation and adaptation efforts, with broader implications for sustainable development and human well-being [10,11]. Research shows these responses are influenced by socioeconomic conditions, psychological dispositions, cultural orientations, and institutional contexts [12,13,14,15,16,17,18]. Among these factors, climate change literacy (CCL) is widely recognized as a multidimensional construct that offers both the conceptual foundation and practical tools for understanding climate change, enabling informed decision-making and pro-environmental behavior [19,20]. Higher climate literacy has been linked to greater risk perception, more substantial concern, and higher policy support [21,22,23]. Yet, findings are inconsistent: knowledge alone often proves insufficient—and at times counterproductive—for motivating behavior, as information can reinforce existing beliefs rather than prompt action. This “knowledge-action gap” has been documented globally across diverse populations and contexts [24,25], revealing a fundamental paradox: even individuals with strong climate science understanding frequently fail to translate this knowledge into consistent pro-environmental behavior. This suggests that literacy operates indirectly through affective factors such as concern and self-efficacy, or with institutional supports [24]. In Taiwan, while citizens and students report high awareness and concern, their actual participation in climate action remains limited, revealing a persistent “knowledge–behavior gap” [25,26,27]. This underscores the need to investigate how institutional and organizational conditions facilitate or constrain the translation of literacy into action.

Assessing climate change literacy has demonstrated robust generalizability across diverse populations, geographical contexts, and cultural settings. Bibliometric analysis of climate change literacy research between 2001 and 2021 reveals that the knowledge-affection-behavior framework—or variations emphasizing knowledge, attitudes/affect, and behavior—has become the dominant theoretical approach in climate literacy assessment globally [20]. This widespread adoption across continents, from African contexts to international comparative studies spanning 119 countries [19,22], demonstrates the framework’s cross-cultural validity and applicability.

The framework’s generalizability stems from its grounding in established psychological theories that explain human environmental behavior universally. A comprehensive review of empirical and experimental studies on climate policy support identifies knowledge, risk perception (affective component), and personal efficacy as consistent predictors of climate-supportive attitudes and behaviors across diverse national contexts [21]. Similarly, research demonstrates that while cultural worldviews moderate the relationship, the fundamental structure linking climate knowledge, risk perception, and behavioral intentions remains stable across different cultural contexts [23].

Climate literacy in Africa was assessed using similar foundational knowledge questions covering climate science basics, causes, and projected impacts [19]. The measurement approach typically employs multiple-choice questions or Likert-scale agreement items about scientific consensus, greenhouse gas effects, and climate change causes. In the affective domain, risk perception and climate concern are universally measured components. Assessed perceived risks from climate change and personal concern levels [22]. Measured affective responses, including worry, concern, and perceived threat, demonstrating that these emotional and evaluative responses to climate change are cross-culturally relevant despite cultural variation in intensity [23]. Personal efficacy—belief in one’s ability to contribute to climate solutions—emerges as another standard affective measure [21]; The behavioral domain is where the most significant variation occurs, as behaviors must be contextualized to population characteristics.

1.2. Policy Vision to Administrative Practice: The Critical Role of Public Officials

Civil servants occupy a unique and critical position in climate governance that distinguishes them from the general public. Unlike citizens who primarily influence climate action through voting and personal behaviors, civil servants directly shape, implement, and enforce climate policies that affect entire populations. While climate literacy is recognized as foundational, its translation into policy outcomes depends heavily on civil servants, who coordinate cross-agency planning, manage budgets and regulations, and facilitate collaboration with stakeholders [28,29]. Their climate literacy directly determines whether ambitious climate targets remain aspirational or become actionable reality [30,31].

Taiwan’s Climate Change Literacy (CCL) survey initially focused on the general public and students, but was later expanded to officials. Enhancing workplace engagement and embedding climate literacy into routine administrative practice across central and local agencies has become increasingly critical [32,33,34]. This progression—from policy vision capacity building and everyday implementation—positions officials’ CCL as a pivotal mechanism for accelerating mitigation and adaptation, advancing sustainable procurement reforms, and strengthening place-based resilience planning [26].

1.3. Behavior Differences Across Governance Contexts

Recognizing the central role of public officials, it becomes essential to understand how organizational and institutional contexts shape their capacity to act on climate knowledge. Research on organizational behavior consistently shows that leadership support, resource provision, and organizational culture strongly shape employees’ willingness to adopt innovative or sustainability-oriented practices [35,36,37,38]. In the public sector, supportive supervisors and environmentally oriented organizational climates are linked to stronger pro-environmental engagement, suggesting mechanisms through which knowledge and affection translate into action [39,40].

At the same time, institutional arrangements define the opportunities and constraints for implementing climate policy. Scholarship highlights how leadership clarity, mandate design, resource availability, and intergovernmental coordination condition the mainstreaming of climate policy at subnational levels [41,42,43]. Local governments often operate with tighter capacity constraints and more immediate stakeholder pressures than central agencies, producing different behavioral responses even under the same statutory frameworks [44,45,46]. Cross-level comparisons between central and local governments are therefore essential for identifying institutional heterogeneity and understanding how different incentive structures and governance contexts mediate the relationship between knowledge, affection, and behavior.

Central agencies are typically responsible for policy design and inter-ministerial coordination, while local governments focus on implementation, community outreach, and disaster response [41,47]. Understanding how supervisory support, departmental involvement, and cross-level dynamics influence officials’ ability to translate their knowledge and affection into action is crucial for assessing implementation readiness and identifying capacity gaps that may hinder Taiwan’s broader climate governance objectives. These considerations motivate the specific research objectives outlined below.

1.4. Climate Change Literacy in Asian Contexts

While climate literacy research has predominantly focused on Western contexts, emerging studies from East and Southeast Asia offer significant theoretical contributions while revealing critical research gaps. These Asian studies make several important theoretical contributions. First, they challenge the centrality of climate concern in communication strategies, demonstrating that climate literacy was a stronger predictor of policy support than concern itself, and that media coverage exerted significant effects through literacy rather than direct emotional engagement [16]. Second, they reveal formative rather than reflective measurement structures for climate literacy, where different dimensions (causes, consequences, engagement) may show inconsistent relationships with predictors and outcomes [23]. Third, they highlight the critical role of cultural and institutional contexts: high institutional trust in China [16] and political neutrality in Taiwan [48] create different pathways from literacy to action compared to Western polarized contexts.

However, significant gaps remain in the English-language literature. Most peer-reviewed climate literacy studies available in English focus on China and Taiwan [48,49]. While climate literacy research in other East Asian countries (Japan, South Korea) may exist in local languages, limited English-language publications constrain cross-national comparative analysis. In Southeast and South Asian contexts, available research has largely emphasized policy analysis and educational frameworks [50,51,52] rather than large-scale empirical assessments of public climate literacy. While some studies have examined the relationship between climate literacy and pro-environmental behaviors [53], systematic investigation across Asian populations remains limited.

1.5. Research Objectives

Against this backdrop, this study systematically evaluates the climate change literacy (CCL) of Taiwanese public officials, focusing on the interplay between knowledge, affection, and behavior. Beyond providing a baseline assessment of literacy levels, the research highlights the institutional contexts that shape whether climate awareness translates into action. Specifically, it examines how supervisory support and departmental involvement influence the relationship between CCL and behavioral practices. By comparing behaviors across various institutional and organizational settings, this study extends the application of CCL frameworks to the field of public administration. In doing so, it addresses a key gap in the climate governance research. It offers practical insights for designing capacity-building programs, strengthening administrative readiness, and supporting Taiwan’s long-term goals of carbon neutrality and resilience.

2. Materials and Methods

This study examines the climate change literacy (CCL) of Taiwanese public officials, focusing on the knowledge, affective, and behavioral domains, as well as the institutional factors that shape their engagement in climate policy. Understanding how officials perceive, internalize, and respond to climate change information is crucial for implementing effective mitigation and adaptation policies [54,55]. The methodological framework is built upon established national CCL surveys in Taiwan and incorporates organizational perspectives from public administration studies [56].

2.1. Data Sources and Sampling Procedures

Data were collected through a cross-sectional survey of Taiwanese public officials in 2024. The process of questionnaire construction and survey implementation is illustrated in Figure 1. A stratified quota sampling strategy was used to ensure representativeness across three dimensions: (a) government levels (central vs. local), (b) policy domains, and (c) administrative ranks. This design aligns with best practices in governance research, as stratified sampling minimizes selection bias and improves coverage of diverse populations [57].

The questionnaire (full version in Appendix A) was distributed primarily online through official channels, with Fax and mail options for agencies with limited internet access. Telephone follow-ups were conducted to confirm delivery and encourage participation, thereby reducing missing data.

Before launch, the survey underwent a rigorous validation process involving both expert review and cognitive pre-testing. First, content validity was established through expert consultation with four specialists, who evaluated item relevance and contextual appropriateness. Items were refined based on their feedback to ensure suitability for Taiwan’s civil service context.

Subsequently, the instrument was cognitively pre-tested with 56 public officials to assess item clarity, difficulty, and discrimination. Cronbach’s α analysis was conducted to evaluate internal consistency, and items with inadequate item-total correlations or those reducing scale reliability were removed or revised. The final instrument demonstrated satisfactory reliability, with Cronbach’s α values for the affective and behavioral dimensions exceeding 0.70 (see Appendix A for dimension-specific coefficients).

Following data cleaning to remove incomplete or invalid cases, the final analytical sample comprised 1940 valid responses. Participation was voluntary and anonymous, with eligibility restricted to active government employees aged 20 or older.

2.2. Measurement of Climate Change Literacy

The CCL framework builds upon the National Environmental Literacy Survey [26,56,58]. It conceptualizes CCL as a multi-dimensional construct with three domains—knowledge, affect, and behavior—representing understanding, emotional response, and participation in climate issues. Each domain was operationalized through sub-dimensions and measured with items designed to reflect both individual and institutional contexts. Figure 2 illustrates the framework.

• Knowledge Domain. The dimension assessed officials’ understanding of the scientific, contextual, and strategic aspects of climate change, through three sub-domains: (a) content knowledge—fundamental concepts such as the greenhouse effect, anthropogenic impacts, and global emissions trends; (b) issue knowledge—the broader context, including natural variability, the human–climate relationship, and evolving policy frameworks; and (c) strategy knowledge—knowledge of mitigation and adaptation strategies at national and international levels. Items were multiple-choice or true-false, scored dichotomously and aggregated into a composite score of knowledge literacy.
• Affective Domain. This dimension assessed officials’ values, attitudes, and motivation for climate action, with five sub-domains: (a) sensitivity—perceiving climate impacts and their extent; (b) values—recognizing of stakeholder responsibilities and the need for cross-sector cooperation; (c) self-efficacy—believing in one’s own ability to adapt, communicate, and cooperate on climate issues; (d) sense of hope—a positive psychological state involving persistence, support from others, and knowledge of strategies; and (e) environmental identity—seeing that environmental protection/environmental problem-solving is essential to individuals and even part of one’s self-image. Constructs were measured with five-point Likert-scale items (1 = strongly disagree to 5 = strongly agree), and mean scores were calculated for each sub-dimension.
• Behavioral Domain. This dimension assessed how public officials translate knowledge and attitudes into action. Sub-domains included: (a) individual skills, which include the ability to collect, apply, and plan climate change information and activities, and to build partnerships across sectors; (b) individual behavior, which refers to actions to mitigate and adapt to climate change.; and (c) civic engagement, including generating intention and experience in collective climate action. Items were rated on a five-point frequency scale (1 = never to 5 = always) and averaged to create action scores.

To capture the organizational settings in which knowledge and affection are translated into action, the 2024 survey asked about officials’ duties and support. Specifically, respondents reported: (1) prior involvement in climate-related projects, (2) the extent to which current work relates to climate issues, and (3) perceived supervisor support for integrating climate considerations. These factors were used as institutional variables in regression analyses to test how professional engagement and organizational support shape behavioral outcomes.

2.3. Data Processing and Statistical Analysis

Data were analyzed using Stata 15.1. To examine mechanisms linking literacy to behavior, hierarchical regression analyses (HRA) were conducted [16]. Independent variables were entered sequentially: knowledge and demographics, then affective domains, then institutional variables. This stepwise approach tested how institutional contexts contribute to explaining behavior and whether administrative structures influence action [54].

Ordinary least squares (OLS) regression analyses were further used to compare central and local officials [20,26,53]. Both dummy variable and split-sample analyses were used to test whether literacy and institutional factors varied significantly across levels of government. This is how institutional culture and administrative roles influence behavior [59,60]. Five hypotheses guided the study:

H1. 

Higher knowledge literacy predicts stronger behavioral engagement.

H2. 

Higher affective literacy predicts stronger engagement.

H3. 

Prior or current involvement in climate tasks predicts higher engagement.

H4. 

Supervisor support enhances engagement.

H5. 

Central and local officials differ significantly in behavioral engagement, reflecting institutional heterogeneity.

3. Results

3.1. Demographic and Background Assessment

Table 1. Sample distribution by demographic characteristics.

Variables	Description	Freq.	Percent	Cum.
Gender	Male	889	45.82	45.82
	Female	1051	54.18	100.00
Age (years)	29 years and under	270	13.92	13.92
	30–39	626	32.27	46.19
	40–49	621	32.01	78.20
	50–59	343	17.68	95.88
	60–69	80	4.12	100.00
Education level	Junior high school	7	0.36	0.36
	Senior high school	34	1.75	2.11
	Junior college	115	5.93	8.04
	Bachelor’s degree	998	51.44	59.48
	Master’s degree	745	38.40	97.89
	Doctoral degree (PhD)	41	2.11	100.00
Seniority	0–9	867	44.69	44.69
	10–19	646	33.30	77.99
	20–29	263	13.56	91.55
	30–39	155	7.99	99.54
	40 years and over	9	0.46	100.00
Government Level	Central	1106	57.01	57.01
	Local	834	42.99	100.00

presents descriptive statistics for the 1940 valid responses, providing a profile of Taiwan’s administrative workforce. The gender distribution was balanced (54.2% women; 45.8% men). The largest age groups were 30–39 (32.3%) and 40–49 (32.0%), followed by 50–59 (17.7%). Younger officials (<29) accounted for 13.9%, while only 4.1% were 60–69. This pattern indicates that most respondents were mid-career professionals, consistent with the civil service structure.

Educational attainment reflected a highly qualified workforce: 51.4% held a bachelor’s degree, 38.4% a master’s, and 2.1% a doctorate. Only 7.0% reported below-tertiary education, meaning over 93% had tertiary education or higher. This profile positions officials well to address complex governance challenges, such as implementing climate policy.

In terms of tenure, 44.7% had <10 years of service, 33.3% had 10–19 years, 13.6% had 20–29 years, and 8.0% had 30–39 years. Fewer than 1% reported 40 years or more. This suggests a relatively junior workforce, balanced by a notable group of mid- to long-tenured officials contributing institutional knowledge.

In Taiwan’s climate governance framework, central agencies focus on policy formulation and coordination while local governments handle implementation and community engagement [61]. Regarding affiliation, 57.0% of respondents worked in the central government, and 43.0% in the local government. This split enables analysis of institutional differences, providing context for interpreting how literacy relates to behavior across governance levels.

3.2. Knowledge and Behavioral Scores Across Professional Backgrounds

Table 2. Knowledge and Behavioral Scores by Field of Expertise.

Field of Expertise	Freq.	Percent	MK	MB
Earth and Environmental Sciences	81	4.18	12.65	3.49
Life Sciences	62	3.20	12.00	3.45
Biological Resources	37	1.91	11.73	3.47
Humanities (Literature, History, Philosophy)	52	2.68	10.77	3.09
Mathematics, Physics, and Chemistry	25	1.29	10.72	3.12
Engineering	310	15.98	10.52	3.22
Medicine and Health Sciences	54	2.78	10.33	3.45
Foreign Languages	51	2.63	10.31	3.05
Finance and Economics	144	7.42	10.31	3.06
Architecture and Design	57	2.94	10.23	3.37
Information Technology	113	5.82	10.20	3.06
Law, Political Science, and Public Administration	225	11.60	9.99	3.05
Mass Communication	31	1.60	9.90	3.26
Education	46	2.37	9.89	3.09
Social Sciences and Psychology	58	2.99	9.76	3.03
Management	112	5.77	9.73	3.13
Recreation and Sports	27	1.39	9.00	2.85
Arts	40	2.06	8.63	3.30
Other: not specified	415	21.39	9.32	2.90
Total	1940	100.00	10.16	3.12

Note. (a) M_K denotes the mean score for the Knowledge Domain; M_B denotes the mean degree for the Behavioral Domain. (b) Fields of expertise are ranked in descending order by mean knowledge score (M_K).

presents the mean scores of the Knowledge Domain (M_K) and Behavioral Domain (M_B) by field of expertise. The Knowledge Domain was scored based on 22 items, with a maximum possible score of 22 points, while the Behavioral Domain was measured using a five-point Likert scale. For the sub-domain of Individual Skills, respondents indicated their level of agreement with each statement (1 = Strongly Disagree; 5 = Strongly Agree). For Individual Behavior and Civic Engagement, respondents rated the frequency of their actions (1 = Never; 5 = Always).

Results reveal that officials with training in Earth and Environmental Sciences (M_K = 12.65; M_B = 3.49) and Life Sciences (M_K = 12.00; M_B = 3.45) exhibited the highest knowledge levels. In contrast, officials with backgrounds in Arts (M_K = 8.63; M_B = 3.30) and Recreation and Sports (M_K = 9.00; M_B = 2.85) scored substantially lower in knowledge.

Notably, behavioral scores did not fully mirror knowledge differences. For instance, respondents in Medicine and Health Sciences (M_K = 10.33; M_B = 3.45) and Architecture and Design (M_K = 10.23; M_B = 3.37) had relatively moderate knowledge scores, yet their behavioral scores were comparable to higher-knowledge groups. Conversely, Engineering, which constituted the largest professional group (15.98% of respondents), demonstrated moderate performance in both domains (M_K = 10.52; M_B = 3.22), suggesting neither disciplinary advantage nor disadvantage in climate engagement.

This pattern indicates that while professional expertise contributes to knowledge acquisition, it does not guarantee proportionally higher behavioral engagement, further highlighting the persistence of the knowledge–behavior gap identified in our main analysis. Importantly, across all professional backgrounds, knowledge scores remained moderate (ranging from 39% to 57% of the maximum possible score), while behavioral scores clustered around the midpoint of the scale (2.85–3.49 out of 5), suggesting room for improvement in both climate literacy and engagement across the civil service.

3.3. Regression Statistical Results

Prior to conducting the regression analysis, multicollinearity diagnostics were performed to assess the independence of predictor variables.

Table 3. Definitions of Variables and Multicollinearity Diagnostics.

Dependent Variables	Definitions	VIF	1/VIF
Action	Degree of engagement in climate change behaviors	-	-
Knowledge Domain
Content knowledge (CK)	Understanding of basic climate concepts	1.25	0.80
Issue knowledge (IK)	Understanding of climate-related risks	1.20	0.84
Strategy knowledge (SK)	Possessing actionable knowledge	1.47	0.68
Affective Domain
Sensitivity	Degree of awareness of climate change impacts	2.54	0.39
values	Degree of acknowledgment of climate change responsibility	3.31	0.30
Self-Efficacy	Confidence in one’s ability to address climate change	1.92	0.52
Sense of Hope	Confidence in achieving climate goals	2.43	0.41
Identity	Degree of environmental self-identification	2.65	0.38
Institutional variables
Once	Prior involvement in climate-related work	1.38	0.72
Related	Degree of alignment between job responsibilities and climate change	1.97	0.51
Support	Degree of supervisor support for climate change	1.50	0.66
Demographics
Gender	Respondent’s gender	1.11	0.90
Age	Respondent’s age	2.74	0.36
Edu	Years of education	1.11	0.90
Seniority	Years of service	2.74	0.37

presents the operational definitions of all variables along with their VIF and tolerance (1/VIF) values. All VIF values ranged from 1.11 to 3.31, well below the commonly accepted threshold of 5 [62], confirming the absence of problematic multicollinearity in the regression models.

Hierarchical regression (

Table 4. Hierarchical Regression Results Predicting Climate-Related Action.

	(1)	(2)	(3)	(4)	(5)	(6)
Variables	Action
CK	0.021	0.017	0.013	0.020	0.016	0.013
	(0.018)	(0.013)	(0.013)	(0.018)	(0.013)	(0.013)
IK	−0.024	−0.019	−0.018	−0.018	−0.016	−0.015
	(0.021)	(0.015)	(0.014)	(0.021)	(0.015)	(0.015)
SK	0.041 ***	0.0068	−0.0046	0.039 ***	0.0049	−0.0067
	(0.0086)	(0.0061)	(0.0062)	(0.0086)	(0.0061)	(0.0063)
Sensitivity	-	−0.015	−0.0049	-	−0.011	−0.00088
	-	(0.025)	(0.024)	-	(0.025)	(0.025)
values	-	0.040	−0.00088	-	0.037	−0.0015
	-	(0.037)	(0.037)	-	(0.037)	(0.037)
Self-Efficacy	-	0.61 ***	0.56 ***	-	0.61 ***	0.56 ***
	-	(0.020)	(0.021)	-	(0.020)	(0.021)
Sense of Hope	-	0.0093	0.020	-	0.012	0.021
	-	(0.022)	(0.021)	-	(0.022)	(0.021)
Identity	-	0.13 ***	0.14 ***	-	0.13 ***	0.13 ***
	-	(0.023)	(0.023)	-	(0.023)	(0.023)
Once	-	-	0.053 *	-	-	0.046
	-	-	(0.029)	-	-	(0.029)
Related	-	-	0.067 ***	-	-	0.065 ***
	-	-	(0.014)	-	-	(0.014)
Support	-	-	0.026 *	-	-	0.028 **
	-	-	(0.014)	-	-	(0.014)
Gender	0.067 **	0.0030	−0.0066	0.050	0.0029	−0.0035
	(0.034)	(0.024)	(0.024)	(0.034)	(0.024)	(0.024)
Age	0.0024	−0.0033 *	−0.0029	0.00079	−0.0039 **	−0.0035 *
	(0.0027)	(0.0019)	(0.0018)	(0.0027)	(0.0019)	(0.0019)
Edu	0.056 ***	0.026 ***	0.020 ***	0.064 ***	0.032 ***	0.024 ***
	(0.011)	(0.0075)	(0.0074)	(0.011)	(0.0076)	(0.0076)
Seniority	0.0011	0.0035 *	0.0030	0.0022	0.0043 **	0.0037 *
	(0.0028)	(0.0019)	(0.0019)	(0.0028)	(0.0019)	(0.0019)
Constant	1.79 ***	−0.0012	0.13	2.09 ***	0.060	0.14
	(0.19)	(0.15)	(0.15)	(0.24)	(0.18)	(0.18)
City				Yes	Yes	Yes
Observations	1940	1940	1940	1940	1940	1940
R-squared	0.043	0.542	0.556	0.075	0.548	0.560

Note. (a) *** denotes significance at the 1% level, ** denotes significance at the 5% level, and * denotes significance at the 10% level. (b) Standard errors are in parentheses.

) tested the effects of knowledge, affective, and institutional factors on behavior. In the baseline model, strategy knowledge (SK) was positively associated with behavior (β ≈ 0.04, p < 0.001), while content (CK) and issue knowledge (IK) were not significant. These results partially support H1 and align with prior research emphasizing the importance of solution-oriented knowledge [63,64].

Affective variables showed robust effects, strongly supporting H2. Self-efficacy was the most potent predictor (β ≈ 0.56–0.61, p < 0.001), consistent with social cognitive theory and studies linking efficacy beliefs to pro-environmental action [65]. Environmental identity was also positively associated (β ≈ 0.13–0.14, p < 0.001), confirming that self-perception as an environmentally responsible individual strengthens engagement [66].

Institutional factors also played a role, supporting H3 and H4. Departmental involvement (“related”) and supervisor support (“support”) both showed significant positive effects (β ≈ 0.07, p < 0.001; β ≈ 0.03, p < 0.05). These results underscore the importance of organizational relevance and hierarchical support in enabling action [67,68,69,70,71,72]. These suggest that the organizational context is crucial, alongside individual literacy and attitudes.

Control variables were included in all models. Education was positively associated with behavior (β ≈ 0.02–0.06, p < 0.001), while gender and age showed weak or inconsistent associations. Importantly, the extent to which the regression model explains variance in the dependent variable improved substantially: R² rose from 0.04 in the baseline to 0.56 in the complete model, showing that affective and institutional variables added significant explanatory power.

Separate OLS models for central (N = 1106) and local (N = 834) officials (

Table 5. OLS Regression Results by Government Level (Central vs. Local Officials).

	Central		Local
	Coef.	Std.	Coef.	Std.
CK	0.019	(0.017)	0.0053	(0.020)
IK	−0.016	(0.019)	−0.014	(0.023)
SK	−0.016 *	(0.0084)	0.0024	(0.0096)
Sensitivity	−0.0038	(0.033)	0.0052	(0.038)
values	0.013	(0.049)	−0.011	(0.058)
Self-Efficacy	0.55 ***	(0.028)	0.57 ***	(0.034)
Sense of Hope	0.032	(0.028)	0.015	(0.035)
Identity	0.13 ***	(0.029)	0.13 ***	(0.036)
Once	0.016	(0.040)	0.093 **	(0.043)
Related	0.088 ***	(0.019)	0.023	(0.023)
Support	0.017	(0.018)	0.054 **	(0.024)
Constant	−0.16	(0.27)	0.24	(0.26)
Control var.	Yes		Yes
City	Yes		Yes
Observations	1106		834
R-squared	0.561		0.575

Note. (a) *** denotes significance at the 1% level, ** denotes significance at the 5% level, and * denotes significance at the 10% level. (b) Standard errors are in parentheses.

) provided clear support for H5. Both groups relied heavily on self-efficacy and environmental identity, with consistent magnitudes (central: β = 0.55 and 0.13; local: β = 0.57 and 0.13, all p < 0.001).

Institutional variables diverged. For central officials, departmental relevance was a significant predictor (β = 0.09, p < 0.001), whereas supervisor support and prior task involvement were not. For local officials, by contrast, supervisor support (β = 0.05, p < 0.05) and prior project involvement (β = 0.09, p < 0.05) were significant, while departmental relevance was not. This indicates that local engagement depends less on formal mandates and more on managerial encouragement and hands-on experience, consistent with research on resource-constrained local governments [73].

Interestingly, strategy knowledge (SK) had a small but significant adverse effect among local officials (β = −0.016, p < 0.05). This suggests that awareness of strategies may heighten perceptions of bureaucratic or political constraints. This paradox echoes a prior study, showing that knowledge does not automatically lead to implementation without supportive institutions [74].

4. Discussion

This study examined how officials’ knowledge, affective, and behavior interrelate across mitigation, adaptation, and civic participation. Building on this, it explored how supervisory support strengthens the translation of knowledge and affection into actions, particularly when supervisors endorse integrating climate issues into daily tasks. It also assessed the influence of departmental climate experience, testing whether prior involvement fosters more proactive cultures. Finally, it compared central and local officials to evaluate how institutional contexts such as hierarchy and governance style shape behavioral engagement. Together, the findings offer theoretical and practical insights into the knowledge–behavior gap, psychological dispositions, and organizational contexts in climate governance.

Findings partially supported H1, which states that strategy knowledge predicts behavior, while content and issue knowledge do not. This underscores a key point: factual and contextual knowledge, though necessary, are insufficient for behavioral change without actionable, solution-oriented understanding [75]. Strategy knowledge provides feasible tools that help close part of the knowledge–behavior gap [26].

This finding converges with cross-national CCL research demonstrating that cognitive understanding alone weakly predicts behavior [13,25]. Our results replicate this pattern in an East Asian public administration context, demonstrating the robustness of this relationship across cultural and institutional settings.

However, our explicit differentiation of knowledge types (content/issue/strategy) reveals important nuances absent from most CCL surveys, which treat knowledge as unidimensional. This distinction suggests that international comparisons may systematically underestimate the behavioral variance explained by knowledge because they conflate different knowledge forms with distinct behavioral implications. By disaggregating knowledge into conceptual understanding (content/issue) versus solution-oriented knowledge (strategy), we show that not all knowledge types equally influence behavior—a finding with methodological implications for future CCL measurement and cross-national comparison.

H2 was strongly supported. Self-efficacy was the most potent predictor (β ≈ 0.56–0.61, p < 0.001), consistent with social cognitive theory and prior findings [76]. Environmental identity also had positive effects (β ≈ 0.13–0.14, p < 0.001), confirming that viewing oneself as an environmentally responsible actor strengthens behavioral consistency. By contrast, sensitivity, values, and hope were nonsignificant, suggesting that action depends less on awareness or moral stance than on capacity and identity alignment.

Self-efficacy’s dominance strongly converges with meta-analytic evidence across climate and environmental behavior research. Meta-analysis of climate belief determinants identified personal efficacy as among the strongest predictors globally [14,15,21]. Our finding replicates this universal pattern.

H3 and H4 were also supported. Departmental involvement and supervisor support both exerted significant positive effects (β ≈ 0.07, p < 0.001; β ≈ 0.03, p < 0.05), highlighting the importance of organizational climate and leadership in facilitating the translation of literacy and affect into behavior. Institutional theory emphasizes that agency is embedded within norms and structures, and our findings confirm that organizational contexts amplify the role of literacy and affective dispositions [77,78].

These findings align strongly with research on organizational environmental behavior. A seminal study of European companies demonstrated that supervisory support behaviors significantly predict employee eco-initiatives [79], while recent meta-analyses confirm leadership’s catalytic role in green behavior [34,37]. Our results replicate these patterns in public sector climate governance, extending organizational sustainability theory to administrative contexts.

H5 was clearly supported. Both central and local officials relied heavily on self-efficacy and identity, but institutional pathways diverged. For central officials, departmental relevance was the only significant institutional predictor, suggesting a reliance on mandates but also constraints imposed by bureaucratic awareness. Local officials, by contrast, were shaped by supervisor support and prior involvement, showing that managerial encouragement and practical experience drive frontline engagement.

Figure 3 and Figure 4 illustrate the levels of knowledge and action among central and local government officials using two institutional indicators measured on a five-point Likert scale. In Figure 3, the grouped bar chart shows that central officials (left) demonstrate a tight coupling between departmental climate relevance and both knowledge acquisition and behavioral engagement. As departmental relevance increases from “not at all related” (1) to “very strongly related” (5), knowledge scores rise substantially from 2.57 to 3.57, while action degrees increase from 2.90 to 3.57. This pattern suggests that for central officials, formal institutional mandates simultaneously enhance climate literacy and drive behavioral engagement.

By contrast, local officials (right) show relatively stable knowledge levels across departmental relevance categories (ranging only from 2.63 to 2.90), yet their behavioral engagement still increases steadily with relevance (from 2.90 to 3.73). This implies that even without substantial knowledge gains, local officials are more likely to take climate-related actions when their responsibilities are perceived as closely aligned with climate governance.

Figure 4 presents the results for supervisor support. Among central officials, both knowledge and action levels increase modestly as supervisory encouragement strengthens (from 2.64 to 3.50), reflecting the hierarchical reinforcement of institutional goals. Local officials, however, display a steeper rise in action (from 2.61 to 3.64), indicating that managerial endorsement plays a more immediate and motivational role in promoting behavioral engagement.

Together, these findings show that institutional factors—specifically job relevance and supervisor support—serve as critical pathways linking literacy to action, though their influence operates differently across governance levels.

4.1. Implications

Practically, this study offers guidance for strengthening Taiwan’s climate governance under the Climate Change Response Act. For central agencies, reforms should streamline mandates, clarify responsibilities, and reduce fragmentation, so that knowledge translates into capacity rather than being hindered by institutional constraints. For local governments, policies should prioritize supervisory training, experiential learning, and capacity-building, as these approaches are effective in driving frontline engagement. More broadly, the study contributes to the field of behavioral public administration by highlighting the importance of leadership exemplars, training programs, and internal governance mechanisms in fostering a climate-conscious public sector. Internationally, the findings offer comparative lessons for other multi-level systems, where aligning capacities with institutional supports is essential for effective climate action.

These empirical and theoretical contributions point to several actionable policy recommendations. First, capacity-building strategies must be differentiated by governance level. Central agencies require cross-departmental climate task forces to overcome bureaucratic silos and standardized climate integration protocols that can be adapted across ministries. Local governments need peer-learning networks, mentorship programs, and experiential learning opportunities, such as pilot projects and site visits, rather than relying solely on classroom-based training.

Second, leadership development requires targeted attention. Mid-level supervisors should receive training on integrating climate considerations into routine task assignments, recognizing and rewarding climate-positive behaviors, and creating psychologically safe environments where officials feel empowered to innovate. Performance evaluation criteria should explicitly include supervisors’ support for climate initiatives, creating accountability mechanisms that reinforce desired leadership behaviors.

Third, training programs should emphasize strategy-focused content, including actionable tools such as greenhouse gas accounting methods, climate risk assessment frameworks, and green procurement protocols. Role-specific strategy toolkits should be developed for officials to apply immediately, complemented by accessible repositories of best practices and successful case studies from similar agencies.

Fourth, self-efficacy enhancement mechanisms should be institutionalized through quick-win initiatives that allow officials to experience tangible success, recognition systems that celebrate incremental progress, and ongoing technical support services that reduce perceived barriers to action.

Fifth, cultivating an organizational climate requires structural interventions. Climate impact assessments should be mandated for major policy decisions across all departments to normalize climate considerations in bureaucratic processes. Dedicated budget lines for climate initiatives signal institutional commitment, while inter-agency coordination platforms facilitate knowledge sharing and collaborative problem-solving. Key performance indicators that track departmental climate engagement create accountability and enable continuous improvement.

Finally, governance structure reforms should address institutional barriers. For the central government, streamlining approval processes and granting greater operational flexibility can mitigate bureaucratic constraints. For local government, strengthening vertical coordination mechanisms ensures policy coherence while respecting local autonomy. Experimental governance arrangements, such as regulatory sandboxes, allow innovative climate solutions to be tested before widespread implementation.

These recommendations are grounded in our empirical findings and should be adapted to the specific contexts and capacities of individual agencies. By addressing both psychological dispositions and institutional contexts, these strategies can strengthen climate governance capacity across Taiwan’s public sector and contribute to more effective climate action.

4.2. Strengths and Limitations

A significant strength of this study is its large, nationally representative dataset (N = 1940), providing robust evidence on officials’ climate literacy across central and local governments. It also integrates multidimensional measures of CCL with institutional factors, providing a comprehensive framework that is rarely applied in prior research. Methodologically, the central–local comparison offers a nuanced view of both individual and institutional determinants, thereby contributing to the intersection between environmental psychology and public administration.

Several limitations should be noted. First, the cross-sectional design limits causal inference; future work should use longitudinal or experimental approaches to capture the dynamic processes linking literacy, institutions, and behavior. Second, reliance on self-reports may introduce social desirability bias, especially on politically salient topics. Third, while distinguishing central and local officials, the study does not fully capture sectoral variation across policy domains. Finally, the Taiwan focus offers valuable insights but may limit generalizability; comparative research across diverse political and institutional settings would extend validation.

5. Conclusions

This study provides systematic quantitative evidence on how Taiwan’s public officials translate climate literacy into behavioral engagement, addressing a critical gap in climate governance research in democratic East Asian contexts. Our findings reveal that effective climate governance requires the interplay of individual agency and supportive institutional contexts, with important implications for both theory and practice.

At the individual level, our analysis demonstrates that strategy knowledge—actionable, solution-oriented understanding—drives behavioral engagement more effectively than factual or contextual knowledge alone. This challenges the assumption that information provision is sufficient for behavioral change and underscores the necessity of equipping officials with practical tools to bridge the knowledge–behavior gap. Moreover, self-efficacy and environmental identity emerge as the most potent psychological predictors of climate action, suggesting that behavioral change depends less on awareness or values than on perceived capacity and identity alignment.

Critically, institutional factors significantly condition officials’ climate engagement. Departmental involvement and supervisor support enable the translation of knowledge and affect into action, confirming that organizational climates and leadership play essential roles. Cross-level comparisons reveal a striking institutional divergence: central officials’ behaviors are shaped primarily by departmental mandates yet constrained by bureaucratic structures, whereas local officials rely more on supervisory encouragement and experiential involvement. This divergence indicates that uniform capacity-building approaches may be ineffective; different governance levels require differentiated intervention strategies tailored to their distinct institutional realities.

Theoretically, this study bridges environmental psychology with institutional perspectives, demonstrating that effective climate governance cannot be understood through individual-level factors alone. Our integration of social cognitive theory with institutional theory reveals that psychological dispositions and organizational contexts interact to shape behavioral outcomes. This microfoundational perspective advances climate governance research by elucidating the mechanisms through which macro-level policies translate—or fail to translate—into frontline action. Significantly, our findings extend the knowledge–behavior gap literature by demonstrating that this gap is not uniform across institutional contexts. Individual literacy and action are embedded within broader organizational and governance structures, which shape whether and how knowledge translates into behavior.