Predictors of Willingness to Engage in Climate Mitigation Behavior and Eco-Paralysis Among Youth in the United States

Abstract

Addressing climate change will require both individual and large-scale collective action focused on systemic forces, particularly from younger generations. There is evidence that climate distress can predict both action and eco-paralysis, suggesting a need for additional research that examines under what conditions salience and distress are associated with mitigation behavior. Using a diverse survey sample of 1603 teens and young adults (ages 13–22) from across the United States, this study examines the predictors of willingness to engage in climate mitigation behavior for youth and compares them to the predictors of eco-paralysis. Results indicate that climate distress is a significant positive predictor of willingness to engage in individual and collective climate action, but that it also predicts eco-paralysis. Notably, we find that climate contempt, when tested alongside distress, positively predicts eco-paralysis and negatively predicts willingness to engage in individual and collective action. Implications for research and practice on climate mitigation behavior among youth are discussed.

1. Introduction

Climate change is a global threat that necessitates global action [1,2]. While individual-level behavior like recycling and energy conservation is likely to be a necessary factor in addressing the climate challenge [3], it is generally understood that climate mitigation must largely occur at the collective level [4] in order to be effective. For example, current research widely demonstrates that system-level influences, including construction, agriculture, and transportation infrastructure, are a much larger contributor to greenhouse gas emissions—a primary driver of climate change [5]—compared to individual-level behavior [6,7]. However, global systemic action to reduce carbon emissions has been slow and largely ineffective [8]. Given the demonstrated challenges of technological solutions and complexities of broad geopolitical action [9], realizing the necessary changes to system-level sources of greenhouse gas emissions requires that individuals engage in behaviors that support outcomes capable of more effectively influencing system-level contributors (i.e., collective behavior), like voting, boycotts, and supporting organizations dedicated to affecting systemic change. That is, collective behavior of individuals, defined in this study as behavior that is aimed at influencing broader systems and structures that significantly affect climate outcomes, is likely to be an indispensable component of future climate action in addition to individual-level behavior [4,10].

There is a considerable amount of research to date that examines predictors of climate mitigation behavior at both the individual [11] and collective level [12]. Research in this area typically focuses on predictors in one of two categories: internal factors—or factors that relate to one’s beliefs and affect—and external factors—or factors that relate to one’s environment. Curiously, some of the same factors that predict mitigation behavior, especially climate distress or anxiety [13], have also been shown to be predictors of eco-paralysis, i.e., the tendency to err on the side of inaction in the face of one’s deep fear and concern about the environment [14]. Research to date suggests that the disparate effect of climate distress and anxiety on climate action may be due to the co-occurrence of other feelings and beliefs not sufficiently accounted for in current work (i.e., other emotional states). What is less understood, then, is what combination of internal and external factors—alongside climate salience and distress—predict willingness to engage in individual and collective behavior and to what extent those same factors predict eco-paralysis. Understanding how the same set of predictors relate to these key outcomes may lend important insights into what motivates—and demotivates—pro-environmental and climate mitigative behavior, and how those factors can be leveraged to more effectively encourage climate action.

Comparing the predictors of willingness to engage in climate mitigation behavior to the predictors of eco-paralysis specifically among teens and young adults is crucial for several reasons. First and most obviously, climate change is an ongoing threat. As such, young people stand to face greater impacts of climate change than their parents and grandparents and will play an essential role in the future of climate mitigation [15]. At the same time, research shows that establishing pro-environmental behavior among youth makes it more likely that they will engage in that behavior as adults [16], as past mitigation behaviors are often predictive of future behavioral intention [17]. Gaining a better understanding of the factors that predict both individual and system-level behavior among youth, then, will arguably improve our collective capacity to successfully respond to this ongoing threat.

Second, despite the instrumental role youth will arguably play in the future of climate mitigation, there is growing evidence that climate and eco-anxiety are on the rise among youth [18]. Some studies show that climate and eco-anxiety can lead to eco-paralysis [19]. Eco-paralysis is the antithesis of pro-environmental behavior and, as a result, a very real threat to successful climate mitigation. Identifying the combination of variables that predict willingness to engage in pro-environmental behavior—particularly in terms of one’s affect—may shed light on how to mitigate the conceivable rise in eco-paralysis due to recent increases in climate and eco-anxiety.

Third and finally, while a good amount of research exists that examines predictors of climate mitigation behavior, much of that research is focused on adults. Younger populations, especially those under the age of 18, are comparatively understudied compared to adults in the United States when it comes to climate perceptions and pro-environmental behavior [20], despite their increasing involvement in meaningful environmental movements [15]. Furthermore, when young people are the focus of pro-environmental behavior research, most studies to date focus on individual-level behavior [21], and much of the existing research on youth takes place in educational settings [22], often evaluating a specific intervention or educational program [23,24] or informing the design of such programs [25]. Comparatively fewer studies have examined predictors of both individual and collective behavior across the United States.

To improve our understanding of what motivates and demotivates willingness to engage in climate mitigation behavior, this study tests the predictive power of previously identified predictors of climate mitigation behavior among youth, ages 13–22, in the United States. More specifically, it compares the factors that predict willingness to engage in individual and collective behavior to those that predict eco-paralysis, controlling for key covariates, including political ideology. Results show that climate distress is a strong positive predictor of both action (individual and collective) and inaction (eco-paralysis), duplicating prior results of a disparate impact of climate anxiety and distress on mitigation action. Most notably, we find that climate contempt is the strongest positive predictor of eco-paralysis and a key negative predictor of willingness to engage in individual and collective action, suggesting that the combination of distress and contempt may be a crucial factor in determining when climate anxiety and distress is motivating and when it is debilitating.

1.1. Predictors of Mitigation Behavior

Research on the antecedents to climate mitigation (i.e., pro-environmental) behavior has increased significantly in recent years [26]. Existing work spans a multitude of disciplines, including psychology [27], economics [28], sociology [29], sustainability [30], and communication [31]. Numerous theories have been used in prior studies, mostly stemming from the field of psychology [26], with the Theory of Planned Behavior and Value–Beliefs–Norm Theory among the most common [21]. Stemming from these diverse disciplines and theories, a considerable number of predictors of pro-environmental behavior have emerged [21,32]. In an attempt to organize the large array of antecedents of climate mitigation and pro-environmental behavior found in existing literature, scholars often rely on the distinction between individual/internal factors (e.g., attitudes and beliefs) and contextual/external factors (e.g., one’s communication environment) [22,32,33].

Internal predictors of behavior generally refer to factors that have to do with an individual’s attitudes and beliefs surrounding a given behavior. While the existing literature has found numerous internal predictors for climate mitigation behavior, the current study examines two that have been found to be particularly important for youth: climate salience and climate distress. Generally speaking, the majority of research on environmental action supports the idea that people are more likely to act on climate change if it is salient for them [34] and if they feel some level of concern over it [17,35]. Often, climate concern among youth manifests itself through negative emotional affect [36]. Indeed, research demonstrates that feelings of anxiety or distress about climate change are significantly related to engaging in pro-environmental behavior [37], especially among young people [13,38]. In the context of mitigative behavior aimed at addressing climate change, this suggests that the extent to which one feels like climate change is a salient issue and the extent to which they are distressed about it should predict willingness to engage in both individual and collective behavior. As such, we offer the following hypotheses:

H1: 

Willingness to engage in individual mitigation behavior will be positively predicted by (a) climate salience and (b) climate distress.

H2: 

Willingness to engage in collective mitigation behavior will be positively predicted by (a) climate salience and (b) climate distress.

In addition to salience and emotional distress, external factors have been shown to play an important role in predicting mitigative behaviors. External factors primarily include elements of one’s environment, as environmental factors are generally understood to provide modeling and reinforcement that shape efficacy and outcome expectations, and, ultimately, behavior. Research shows that environmental and climate information from interpersonal networks [39,40,41], social media [42], and mass media [43] all predict pro-environmental and climate mitigative behaviors. Understanding the relative predictive power of these different information sources is paramount for encouraging mitigative behavior, as research suggests that all information sources are not created equal in terms of their ability to motivate action [44]. To build on existing work that examines how one’s social environment impacts their behavior, the current study examines the predictive power of exposure to climate-relevant information within one’s social environment in their interpersonal, social media, and mass media networks. To that end, this study tests the following hypothesis:

H3: 

Willingness to engage in individual mitigation behavior will be positively predicted by learning about climate change through (a) interpersonal social networks, (b) social media, and (c) mass media.

H4: 

Willingness to engage in collective mitigative behavior will be positively predicted by learning about climate change through (a) interpersonal social networks, (b) social media, and (c) mass media.

1.2. Predictors of Eco-Paralysis

Generally speaking, climate anxiety has two manifestations: cognitive impairment—where one’s ability to cognitively function is affected—and functional impairment—where one’s anxiety interferes with their day-to-day activities [14]. While prior research shows that climate anxiety and distress are on the rise among youth [18] and that climate anxiety and distress can be a motivating factor in terms of climate mitigation behavior [14,45], some studies show that climate and eco-anxiety can also lead to eco-paralysis [19]. Widely considered to be a maladaptive response to climate anxiety, eco-paralysis is the tendency to err on the side of inaction in the face of one’s deep fear and concern about the environment and climate. While somewhat similar to the concept of efficacy, eco-paralysis is distinct in that it is not necessarily reflective of a belief regarding one’s capabilities, but rather a behavioral freeze response to emotional overwhelm. Given that an issue must, to some extent, be salient in order for one to feel distress over it and elect to avoid taking action, we offer the following hypotheses about the relationship between eco-paralysis and salience, distress, and one’s information environment:

H5: 

Eco-paralysis will be positively predicted by (a) climate salience and (b) climate distress.

H6: 

Eco-paralysis will be positively predicted by learning about climate change through (a) interpersonal social networks, (b) social media, and (c) mass media.

To understand under what conditions climate or eco-anxiety is motivating rather than maladaptive, there are two explanations supported by recent research. First, there is evidence to suggest that the amount of climate anxiety matters; that is, too much or too little may lead to an unwillingness to engage in mitigative behavior, creating an inverted U-shaped relationship between climate anxiety and environmental action [46]. Second, some have argued that climate anxiety must be experienced alongside additional non-affective variables, like self-efficacy or specific value orientations, in order to be sufficiently motivating [45]. Additionally, when faced with negative emotions like guilt or helplessness, research shows that youth often cope by distancing themselves from their emotional responses [36], which may function to decrease their effectiveness in motivating action.

However, what is less understood is what other affective responses, when paired with salience and distress, may help explain the puzzling and sometimes inconsistent relationship between climate distress and eco-paralysis. One underexplored affective factor that may be a relevant predictor of eco-paralysis—particularly when combined with climate distress and anxiety—is contempt. Contempt is generally studied within the context of political discourse and identities [47,48,49]. In this context, it is often defined as a negative appraisal of or resentment toward another person characterized by the belief that the other person is somehow unworthy or inferior [50], and that others should similarly reject them [47]. Importantly, research on contempt has found that it often functions alongside other emotions and can lead to disengagement from the issue(s) or person(s) at hand [51].

The role of contempt has not been sufficiently studied within the context of climate change and climate mitigation behavior. Climate contempt—generally understood to be a morality-based affectual orientation [52]—is characterized by feelings of hostility or resentment toward the climate crisis [53]. Given prior research that establishes the key role of contempt in one’s political beliefs and behaviors, and the highly politicized nature of climate change in the United States, it is reasonable to assume that contempt may be a relevant affectual response to the climate crisis. Given that prior work has also shown that contempt, in particular, can lead to disengagement in the political sphere, it may also, in turn, influence the connection between climate distress and willingness to engage in mitigation behavior. Because climate contempt has not been sufficiently examined in relation to climate mitigation behavior, we offer the following as a research question:

RQ: 

To what extent does climate contempt predict willingness to engage in individual mitigation behavior, collective mitigation behavior, and eco-paralysis?

2. Materials and Methods

2.1. Procedures and Participants

Data were collected via a cross-sectional online anonymous survey through Qualtrics over the course of three weeks in late September and early October of 2024. Participants were residents of the United States aged 13–22 (n = 1603) who previously agreed to participate in survey research in exchange for compensation. Participant recruitment was managed through Qualtrics. Quotas were set ahead of time for gender (~50% female) and parent education (~50% of participants having at least one parent with a college degree). The sample was diverse with regard to gender, socio-economic status indicators, and political ideology. For a complete summary of participant demographics, see Table 1. Participants were provided informed consent prior to their participation and were compensated at a rate previously agreed upon between themselves or their guardians, for those under the age of 18, and Qualtrics. Consent was obtained from their legal guardians for those under the age of 18. All data collection procedures were reviewed and approved by an Institutional Review Board prior to data collection (IRB approval #FY2025-8), and all study procedures were performed in compliance with relevant laws and institutional guidelines.

Table 1. Sample Demographics.

Age			Education Level
	N	%			N	%
13	93	5.8	Current or some middle school		125	7.8
14	119	7.4	Current or some high school		740	46.2
15	147	9.2	High school graduate/GED		212	13.2
16	176	11.0	Current or some college		450	28.1
17	244	15.2	College graduate		33	2.1
18	155	9.7	Current or some post-graduate		39	2.4
19	141	8.8	Post-graduate degree		4	0.2
20	144	9.0
21	192	12.0	Parent Education Level
22	192	12.0	At least one parent has a college degree		791	49.3
Ages 13–17 (N = 779, 48.6%) Ages 18–22 (N = 824, 51.4%)			Neither parent has a college degree		812	50.7
Gender
Male	808	50.4	Race & Ethnicity
Female	782	48.8	White or European	860		53.6
Other	13	0.8	Black or African American		354	22.1
			Multiracial		149	9.3
Political Ideology			Hispanic or Latino/a		98	6.1
Very Liberal	169	10.5	Asian		68	4.2
Liberal	215	13.4	Native Hawaiian or Pacific Islander		29	1.8
Lean Liberal	129	8.0	American Indian or Alaska Native		26	1.6
Neutral/Moderate	576	35.9	Other		11	0.7
Lean Conservative	92	5.7	Middle Eastern or North African		8	0.5
Conservative	147	9.2	Percentages rounded to the nearest tenth. Sample n = 1603
Very Conservative	79	4.9
“I don’t know”	196	12.2

2.2. Measures

Climate Salience. To assess climate salience, participants were asked two questions used in existing work [54]. The first question was “how much had you thought about climate change before today?” and participants answered on a scale of “none at all” (1) to “a great deal” (5). The second question was “how important is the issue of climate change to you personally?” and participants answered on a scale of “not at all important” (1) to “extremely important” (5). These two items were highly positively correlated (r = 0.68, p < 0.001) and were averaged together to represent climate salience (M = 2.95, SD = 1.11).

Climate Distress. Affective anxiety toward climate change was assessed using climate distress, defined as the extent to which one experiences emotional turmoil in response to climate change. Survey items measuring climate distress were based on prior work [55]. Participants were provided with nine statements, including “I feel anxious when I think about the consequences of climate change,” and “My awareness of climate change and its effects takes away from my happiness,” and asked to rate the extent to which they agreed with those statements, from “strongly disagree” (1) to “strongly agree” (5). Taken together, these nine items had high reliability (α = 0.91) and were averaged to create the climate distress variable (M = 3.19, SD = 0.97).

Climate Contempt. Climate contempt was assessed using a preexisting four-item measure [56]. Participants were provided with statements like “It annoys me when people panic about climate change” and “I am annoyed by the constant publicity around climate change” and asked to rate the extent to which they agreed with those statements, from “strongly disagree” (1) to “strongly agree” (5). Taken together, these four items had adequate reliability (α = 0.79) and were averaged to create the climate contempt variable (M = 2.58, SD = 1.01).

Information Sources. For exposure to climate information, participants were asked “How much have you learned about climate change from the following sources of information?” and provided a list of potential information sources, which included interpersonal sources (friends, family, peers/classmates, and professors/teachers), social media, and mass media (newspapers/radio/TV, internet news, and podcasts). Participants responded on a scale from “nothing” (1) to “a lot” (5). For the purposes of this analysis, interpersonal sources were averaged together (M = 2.65, SD = 0.97, α = 0.77), as were mass media sources (M = 2.66, SD = 1.04, α = 0.74). Social media was a single item (M = 3.20, SD = 1.23).

Willingness to Engage in Individual Mitigation Behavior. Willingness to engage in individual climate mitigation behavior was quantified using an existing scale for pro-environmental behavior [57]. Questions were modified from the original source to focus on willingness to engage in the listed behaviors specifically to address climate change (rather than improve the environment generally), and the list of behaviors was reduced to regulate survey length. Participants were asked “How willing would you be to do each of the following things to help reduce the effects of climate change?” and provided seven behaviors to respond to using a willingness scale, from “not at all willing” (1) to “very willing” (5). The behaviors included things like turning off the lights when they are not needed, recycling, and carrying a reusable beverage container. These seven items, taken together, had good reliability (α = 0.87), and were averaged to create the individual action variable (M = 4.08, SD = 0.85).

Willingness to Engage in Collective Mitigation Behavior. To assess willingness to engage in collective climate mitigation behavior, participants were asked “How willing would you be to do each of the following things to help reduce the effects of climate change?” and provided a list of seven system-level actions, including volunteering time at a climate-change non-profit, joining a campaign to influence elected officials to take action on climate change, and signing a petition that supports action on climate change. Participants answered on a scale from “not at all willing” (1) to “very willing” (5). Taken together, the seven items had high reliability (α = 0.91) and were averaged to create the collective action variable (M = 3.39, SD = 1.09).

Eco-paralysis. Eco-paralysis is the tendency to err on the side of inaction in the face of one’s deep fear and concern about the environment and climate. Eco-paralysis was assessed using an existing scale [19]. There were eleven items included in this scale. Participants were provided with statements like “the changes in the global climate are so profound that they can no longer be stopped” and “any attempt on my part to change the global environmental situation would do nothing else than frustrate me,” and were asked to respond with the extent to which they agreed with those statements, from “strongly disagree” (1) to “strongly agree” (5). Taken together, these eleven items had adequate reliability (α = 0.79) (M = 2.94, SD = 0.69).

A correlation matrix for all variables can be found in Table 2. The complete survey protocol can be found in Appendix A.

Table 2. Correlation Matrix.

	Salience	Distress	Contempt	Info–Interpersonal	Info–Mass Media	Info–Social Media	Indiv. Behavior	Coll. Behavior
Distress	0.601 ***
Contempt	−0.322 ***	−0.265 ***
Info–Interpers.	0.461 ***	0.339 ***	0.027 (ns)
Info–Mass Media	0.569 ***	0.385 ***	−0.037 (ns)	0.630 ***
Info–Soc. Media	0.456 ***	0.386 ***	−0.084 ***	0.481 ***	0.576 ***
Indiv. Behavior	0.351 ***	0.397 ***	−0.277 ***	0.206 ***	0.211 ***	0.205 ***
Coll Behavior	0.556 ***	0.621 ***	−0.329 ***	0.370 ***	0.419 ***	0.331 ***	0.592 ***
Eco-Paralysis	0.045 (ns)	0.236 ***	0.334 ***	0.104 ***	0.133 ***	0.119 ***	−0.104 ***	0.005 (ns)

Pearson product correlation values shown; *** p < 0.001; ns = not significant.

3. Results

A single path analysis model tested all relationships simultaneously, using the sem command in Stata v. 15.1 (StataCorp). Internal factors, external factors, and demographic covariates were entered as predictors of the three dependent variables (i.e., individual action, collective action, and eco-paralysis). The dependent variables’ residuals were allowed to covary, accounting for shared variance among the outcomes. Variables are primarily multi-item scales and were treated as approximately continuous; thus, parameters were estimated using maximum likelihood estimation. Standardized coefficients are reported to enable direct comparison of relationship sizes (unstandardized coefficients are available in the supplemental materials).

Analyses include the full sample, as no data were missing. While the original model specification indicated that most relationships were statistically significant, goodness-of-fit statistics indicated a just-identified model (consistent with saturated models), precluding assessment of fit (χ²(0) = 0.00; CFI = 1.00; RMSEA = 0.00). Thus, in a second analysis, non-significant paths were constrained to zero, indicating acceptable model fit (χ²(14) = 22.39; CFI = 0.996; RMSEA = 0.019). The trimming procedure in the second model was intended to allow interpretable model fit statistics and should be considered exploratory rather than confirmatory [58]. Standardized beta coefficients from the parsimonious model are presented in Table 3. Beta weights and significance levels were consistent between the full and parsimonious models. Still, fit statistics should be interpreted cautiously since the model was fitted from observed data.

Table 3. Final, parsimonious path analysis.

Dependent Variable	Independent Variable	Beta	S.E.	95% Confidence Interval
Individual Action
	CC salience	0.17 ***	0.03	0.11, 0.23
	CC distress	0.26 ***	0.03	0.20, 0.31
	CC contempt	−0.15 ***	0.02	−0.19, −0.10
	Parent has college degree	−0.08 ***	0.02	−0.13, −0.04
	Female	0.11 ***	0.02	0.07, 0.14
	Age	−0.10 ***	0.02	−0.14, −0.06
Collective Action
	CC salience	0.17 ***	0.03	0.12, 0.22
	CC distress	0.41 ***	0.02	0.37, 0.46
	CC contempt	−0.15 ***	0.02	−0.19, −0.11
	CC info–Interpersonal	0.05 **	0.02	0.01, 0.10
	CC info–Mass media	0.13 ***	0.02	0.08, 0.17
	Parent has college degree	−0.04 *	0.02	−0.08, −0.003
	Age	−0.10 ***	0.02	−0.14, −0.06
	Conservative political ideology	−0.05 *	0.02	−0.08, −0.02
Eco-paralysis
	CC Salience	−0.06 *	0.03	−0.11, −0.001
	CC distress	0.33 ***	0.03	0.28, 0.38
	CC contempt	0.46 ***	0.02	0.42, 0.50
	Age	0.06 **	0.02	0.02, 0.10
	Conservative political ideology	−0.04	0.02	−0.09, 0.001

N = 1603. *** p < 0.001; ** p < 0.01; * p < 0.05. Values are standardized beta coefficients from a path analysis model. Missing paths were nonsignificant in full model and constrained to 0. Model RMSEA = 0.019, CFI = 0.996. Chi²(14) = 22.39, p = 0.07. Individual Action R² = 0.23; Collective Action R² = 0.48; Eco-paralysis R² = 0.24.

Models were also reanalyzed using robust standard errors and yielded substantively identical results. Due to the skewed distribution of the Individual Action variable, sensitivity analyses using a binned version of the variable with five groups of roughly equivalent size were also conducted. This model also showed findings consistent with the original model. Finally, analyses tested the model within two subsamples divided by age (13–17 years, n = 779; and 18–22 years, n = 824). Patterns of results were substantively unchanged in these analyses as well, suggesting robust findings across age cohorts. The full original model containing standardized beta coefficients and standard errors is displayed in Supplementary Materials (Table S1), along with unstandardized coefficient values (Table S2). Results described below are from the final parsimonious model (Table 3).

Across dependent variables, the internal factors as a set accounted for more variance than external factors or covariates. With regard to individual pro-environmental action, the full model accounted for 23% of the variance. Level of climate distress was the strongest predictor (β = 0.26, p < 0.001), followed by salience (β = 0.17, p < 0.001), supporting H1a and H1b, and climate contempt (β = −0.15, p < 0.001). None of the external factors were significantly predictive of individual action, so H3a, H3b, and H3c were unsupported. Older participants (β = −0.10, p < 0.001) and those whose parents had a bachelor's degree or more in terms of education reported lower intention to perform individual pro-environmental behaviors (β = −0.08, p < 0.001), while females reported greater intention than males (β = 0.11, p < 0.001).

Patterns were similar for willingness to engage in collective climate actions, though the model predicted more variance for this variable (R² = 0.48). Climate distress was an especially strong predictor of intended collective action (β = 0.41, p < 0.001), followed by climate salience (β = 0.17, p < 0.001) and contempt (β = −0.15, p < 0.001), meaning H2a and H2b were supported. Notably, two external factors were also significantly related to intended collective action. Participants who reported greater climate information from interpersonal sources (β = 0.05, p < 0.01) and mass media (β = 0.13, p < 0.001) had greater intentions to engage in collective action, supporting H4a and H4c. Social media (H4b) was not a significant predictor. Older age (β = −0.10, p < 0.001), having a more educated parent (β = −0.04, p < 0.05), and more conservative ideology were all negatively associated with intended collective action (β = −0.05, p < 0.01).

Regarding eco-paralysis, the full set of significant predictors accounted for 24% of variance. Again, none of the external factors were significantly predictive, meaning H6a, H6b, and H6c were unsupported. Greater climate contempt predicted more eco-paralysis (β = 0.46, p < 0.001), and this was the strongest relationship. Climate distress was also a strong positive predictor (β = 0.33, p < 0.001), supporting H5b, while climate salience had a weak but significant negative relationship with eco-paralysis (β = −0.06, p < 0.05), meaning H5a was unsupported. Older participants had greater eco-paralysis (β = 0.06, p < 0.01), while political conservatism was related to lower eco-paralysis (β = −0.04, p < 0.05).

4. Discussion

Teens and young adults are likely to spend the next several decades dealing with and combating the worst effects associated with climate change. Consequently, it is vital to understand the internal and external factors that motivate (or de-motivate) this population to act on this environmental threat. In this study, we hypothesized that climate salience, climate distress, and exposure to climate information from one’s information environment would positively predict willingness to engage in individual and collective climate mitigation behavior for youth in the United States. We found that climate salience and distress were significant positive predictors of willingness to engage in individual and collective climate mitigation behavior, validating prior work [13,17,34,35,38]. While the data utilized in this study cannot establish a causal mechanism between salience and distress and willingness to engage in climate action, these results nonetheless do suggest that salience and distress, taken together, are important statistical predictors of the willingness to engage in individual and collective climate action.

Interestingly, we found mixed results in terms of the predictive of information sources on individual and collective action. Duplicating results of prior research [43], learning about climate change through mass media was significantly and positively predictive of collective action. Counter to some earlier findings [40,42] but in line with others [59], learning about climate change through social media was not a significant predictor of any individual behavior, collective behavior, or eco-paralysis. This finding is particularly interesting, as the majority of youth in the United States spend a large amount of their time on social media sites [60] and rely comparatively less on traditional sources like mass media [61]. Indeed, in the study sample, on average, youth reported learning more about climate change through social media (M = 3.20, SD = 1.23) than they learned through mass media (M = 2.66, SD = 1.04) or interpersonal sources (M = 2.65, SD = 0.67), suggesting that social media in particular should arguably be a significant predictor of behavior.

Despite the importance of social media as an information source about climate change demonstrated in the current sample and widespread use of social media among youth generally, we suspect that learning about climate change through social media was not a significant predictor of individual or collective mitigation behavior for two possible reasons. First, our operationalization of learning about climate change through social media did not account for what kind of information participants were exposed to. For example, it is entirely possible to ‘learn’ about climate change through information presented on social media, without that information being factually correct or particularly motivating of system-level behavior. Indeed, the prevalence of misinformation on social media is well-documented [62], as is the polarization around climate change on social media sites [63]. Second, given the frequency with which teens and young adults use social media [60], it is possible we did not see an effect of learning about climate change through social media because of the vast amount of information teens and young adults are exposed to on a daily basis that both converges with and differs from their own perspectives [64]. This could reduce message consistency, thus reducing the predictive power of learning about climate change through social media on pro-environmental behavior.

Importantly, our measure of information sources asked participants specifically about how much they had learned about climate change through various sources of information. This approach does not capture other important elements of information transmission long known to influence message efficacy and effects, including frequency of information exposure [65], variations in message content and framing [66], and beliefs about or assessments of source credibility [67]. For this reason, it is not possible to know whether the lack of effect found here regarding social media, for example, is actual evidence of null effects or an artifact of the aspects of information transmission not captured by our measurement approach.

This study also sought to examine the role climate contempt might play in helping to explain the differential results to date regarding the role of climate anxiety in motivating climate mitigation behavior. While prior research shows that climate anxiety and distress are on the rise among youth [18] and that climate anxiety and distress can be a motivating factor in terms of climate mitigation behavior [14,45]; some studies show that climate and eco-anxiety can also lead to eco-paralysis [19]. Prior work has identified two explanations for the disparate results regarding climate anxiety and (in)action to date: the level of climate anxiety [46] and the role of other variables like self-efficacy [45]. We found that climate contempt, when combined with distress, powerfully predicts eco-paralysis and negatively predicts individual and collective climate mitigation action. This suggests that in addition to the other variables already identified by prior research, contempt may be a key differentiating factor in determining when anxiety and distress are motivating, and when they are demotivating.

When interpreting the results reported here regarding the potential role of climate contempt in predicting willingness to engage in climate action, there are some important caveats to consider. Most notably, we operationalized climate contempt using one of the very few existing scales available to date. However, this measure may not comprehensively align with the theoretical conceptualization of contempt used in research on political discourse and identities [49] or the conceptualization of climate contempt as a feeling of resentment toward the climate crisis [53]. For example, the measure used here consisted of four items and generally asked about annoyance and surprise regarding climate hysteria. By contrast, the traditional conceptualization of “contempt” found in political literature and, more recently, the concept of climate contempt, generally entails hostility and resentment toward a given object or person. While we do believe that our operationalization captures an important potential component of climate contempt specifically—particularly resentment toward climate hysteria—it does not fully capture contempt as it has been traditionally conceptualized and therefore may lack content validity. To that end, future research should explore the extent to which climate contempt may mirror, and perhaps diverge from, the morality-based concept of contempt found in prior research and explore alternative methods of operationalization.

Interestingly, the three internal variables examined here—salience, distress, and contempt—offer the same level of explanatory power for both willingness to engage in action and eco-paralysis. However, contempt was far more positively predictive of eco-paralysis than it was negatively predictive of individual and collective action. This mirrors prior work that shows negative affect toward climate change may motivate distancing behavior [36], and that contempt in particular may function to encourage disengagement from other people [51] or particular issues. Of course, this does not necessarily mean that the mere absence of contempt will lead to action. Indeed, the relationship between climate distress and mitigation behavior remains complex, and current research suggests that climate distress also must be met with meaning-focused coping strategies [38] to be an effective motivator. While the current study only examined climate contempt and did not account for such coping strategies, future work looking to leverage climate distress as a motivator for climate mitigation behavior should be mindful of these dynamics, especially given that it was also predictive of eco-paralysis. This may be especially relevant given the highly politicized nature of climate change in the United States. It is possible, for example, that the relationship between contempt and willingness to engage in mitigation behavior is unique for political liberals and conservatives. While we elected to control for political ideology in our analysis in an effort to statistically isolate the predictive power of contempt, future research should explore how contempt relates to willingness to engage in mitigation behavior by varying the inclusion method for political ideology (e.g., treating it as a moderator or mediator rather than a control).

These results have several implications for motivating climate action among youth. For example, they suggest that interpersonal and mass media information sources may play an important role in collective action but may be less important for individual action. In terms of disseminating information about climate change through social media channels, despite the fact that social media is the most important information source among youth, our results suggest that learning about climate change through social media may not be an important factor in predicting climate action or inaction. Finally, and perhaps most importantly, these results suggest that engagement efforts should make an effort to avoid creating feelings of contempt about climate change, as contempt may counteract the effect of climate distress in motivating action and may, instead, predict eco-paralysis. Unfortunately, the current study cannot speak to the engagement processes that risk fostering contempt, as the data reported here are not causal and did not test specific engagement strategies. Given that climate contempt is comparatively under-researched, future research should investigate communication strategies that create feelings of contempt. When interpreting our results in the context of climate communication strategies, while our analysis can show the extent to which the variables captured here statistically predict willingness to engage in climate action, it cannot establish the extent to which those variables cause or inhibit willingness to engage in that action or the extent to which they influence actual mitigation behavior. Additionally, the nature of our analysis focused on predictors of willingness to engage in individual and collective behavior, as a path analysis assumes a linear relationship between variables. This approach limits our ability to examine the extent to which the variables studied here may have a reciprocal or cyclical relationship with each other. Future research should address this gap and examine the extent to which, for example, engaging in mitigation action leads to feelings of distress or contempt.

There are important limitations to note for the current study. While our sample size was quite large and demographically diverse, our sample consisted of teens and young adults who previously agreed to complete online surveys in exchange for compensation. As a result, these study results may suffer from selection bias and are not necessarily generalizable to teens and young adults in the entire United States and should not be interpreted as being nationally representative. Our measurement of willingness to engage in climate mitigation behavior at both the individual and collective level was self-report and did not account for actual behavior, whether through self-report or direct behavioral observation. Additionally, willingness to engage in individual-level mitigation behavior was fairly high, on average. While we conducted sensitivity analyses to guard against the threats introduced by a ceiling effect and do not believe it occurred in our analysis, any skewed measure may introduce bias in statistical findings. Self-report methods of assessing behavioral intention are not always as reliable as other measurement methods, like direct observation of behavior [68]. Future work should collect data on actual behavior to see if the predictors of actual (in)action are similar to those found here regarding self-reported willingness to act. Finally, we used a path analysis to test our hypotheses on cross-sectional data, meaning that the current study cannot establish causality or speak to the extent to which feelings of contempt directly cause inaction. Future research in this area should utilize experimental study designs to examine causal mechanisms, particularly regarding the extent to which various engagement processes may risk fostering feelings of climate contempt or other factors that appear to lead to climate inaction.

5. Conclusions

It is generally understood that successful climate mitigation will require action at both the individual and system levels. Because youth will play a key role in future climate action, understanding the factors that predict willingness to engage in both individual and system-level mitigation behavior is vital. Prior research shows that climate distress is on the rise among youth and is paradoxically associated with both willingness to act and with eco-paralysis. What is less understood is what other affective responses, when paired with distress in particular, may help explain the puzzling and sometimes inconsistent relationship between climate distress and mitigation behavior. The results of the current study suggest that climate contempt is a negative predictor of willingness to act and a positive predictor of eco-paralysis. Improving our understanding of how affective responses to climate change, like distress and contempt, operate together may help inform climate engagement strategies that encourage action and reduce the risk of eco-paralysis.