Volunteering in Environmental Organizations and Subjective Well-Being: Evidence from a Nationally Representative, Longitudinal Dataset in the US

Abstract

This study uses a nationally representative longitudinal dataset in the US to examine the long-term association of volunteering for environmental, recycling, and conservation groups with a person’s (a) willingness to continue to volunteer later in life and (b) several measures of their mental and physical well-being including perceived social status, optimism, psychological stress, suicidal thoughts and attempts, depressive symptoms and general self-reported physical health. By using Add Health data, we match responses to an environmental volunteerism question in Wave III (2002) with subjective well-being responses in Wave V (2016–2018) to examine the long-term association between these variables. After excluding missing responses, the analysis sample consists of 9800 individuals. After using linear survey regression analyses and several techniques based on propensity scores (stratification, weighting, matching) two key results emerged: first, being involved in environmental groups and organizations early in life showed a significant positive association with more hours spent on volunteering or community service work later in life; and second, people who volunteer in early adulthood are more optimistic, more sociable, have a higher perceived social status, display less stress and depressive symptoms.

1. Introduction

When asked, most people in the United States state that they care very much about the environment; for instance, according to a recent survey 69% favored the U.S. taking steps to become carbon neutral by 2050 [1]. Fewer people, however, translate this stated concern into real commitment through voluntary actions, i.e., volunteering to recycle, clean-up trash, improve conservation areas (see [2,3,4]). According to [4], this gap is not rare—they report that 72% of their respondents admitted that they observe a discrepancy between their actions and their intentions. Volunteering is not an automatic response to stated environmental concerns. So, what can explain the gap between such reported intentions and actions? Are those individuals who volunteer simply doing so as it contributes to a greater cause, or do their actions also improve their own well-being and increase their long-term commitment to the environment? While some evidence exists on how environmental concerns affect subjective well-being (e.g., [5,6,7]), little is known about how volunteering in environmental organizations, a distinct and well-defined type of pro-environmental behavior, impacts physical/mental well-being and translates into a long-term commitment to environment protection.

Herein we address this open question empirically and explore the link between environmental volunteering and measures of well-being. Our paper explores whether past volunteering in environmental organizations leads to greater mental and physical well-being as well as future volunteering. Using unique features provided by the National Longitudinal Study of Adolescent to Adult Health (Add Health) dataset, we measure a person’s level of care about the environment in terms of their self-reported volunteering of community service to environmental organizations. We evaluate the association of such volunteering in early adulthood (at ages 18–26) with several measures of physical and mental well-being later in life (at ages 32–42).

In more detail, this paper aims to examine the long-term adjusted association of pro-environmental behavior in terms of membership in environmental and conservation clubs earlier in life with measures of mental and physical well-being (perceived social status, optimism, neuroticism, and sociability, perceived stress scale, depressive symptoms, suicidal thoughts, physical health) and continuing volunteerism later in life. We use environmental volunteerism to measure pro-environmental behavior. This covariate comes from Add Health Wave III (2002) and is based on the question about volunteering in conservation, recycling, or environmental groups. We match responses to this environmental volunteerism question with subjective well-being responses in Wave V (2016–2018) to examine the long-term association between these variables. In other words, we analyze the long-term (~16 years) association of environmental volunteerism with individual well-being for US residents.

After excluding missing responses, we conducted our analysis with a sample size of 9800 individuals and our methodology is designed to investigate whether environmental volunteerism is related to individual well-being in the long run. Using regression and matching techniques based on propensity scores, we study the impact of volunteering to environmental organizations on an average person’s physical, mental, and emotional well-being in the long run by controlling for observable demographic and socio-economic confounders.

Our study improves on previous attempts to establish a link between environmental volunteerism and individual well-being in four specific ways. First, our measure of environmental concern is based on a very specific question about environmental volunteerism and this question simply asks whether the respondent has volunteered to environmental organizations. This is important because the quality of self-reported measures may differ [8,9] and, as discussed in [10], answers to frequently used questions in the form of “How often do you recycle paper?” can be imprecise and comparisons of responses across people may be problematic. While volunteerism for environmental groups variable that we use was self-reported by respondents, it is a relevant measure of active involvement in conservation and environmental organizations, as opposed to questions about general levels of care about the environment [5,6,11]. Second, we look at several established standardized, psychological measures of well-being such as the CES-D scale (Center for Epidemiologic Studies Depression Scale) and Life Orientation Test. Third, our data comes from a large nationally representative, longitudinal dataset. Fourth, unlike most relevant studies, we evaluate the association of volunteering with individual well-being in the long-term—approximately 16 years later. This long-term perspective matters because some health outcomes are likely to manifest over time.

Examining ongoing volunteer participation in environmental organizations is crucial for environmental policies aimed at changing people’s behaviors and habits [12]. Our analysis can guide efforts to increase volunteerism by emphasizing the otherwise discounted individual benefits of such actions.

2. Literature Review and Background

Past literature has focused on how individuals’ pro-environmental behavior is related to subjective well-being, and whether pro-environmental behavior promotes individuals’ subjective well-being and happiness [5,13,14,15,16,17,18,19,20]. For example, [5,14,16,18,19] examined subjective well-being in the context of climate change, natural hazards, environmental degradation, and disasters, whereas [7,15,17] examined subjective well-being in the context of self-perceived health analyses. Subjective well-being generally refers to how people experience and evaluate their lives and activities in their lives [21]. We define subjective well-being as the degree to which individuals experience positive or negative emotions, and how they evaluate their own lives [22].

Pro-environmental actions are sometimes perceived as costly, effortful, or otherwise unpleasant, potentially leading to a negative impact on subjective well-being (e.g., [23,24]). This implies a negative relationship between pro-environmental behavior and subjective well-being. However, an expanding body of research suggests the opposite—that engaging in pro-environmental actions may, in fact, enhance individuals’ subjective well-being [25].

It has been suggested that the link between pro-environmental behaviors and subjective well-being may be positive because engaging in such behaviors is perceived as meaningful [26,27]. The sense of meaningfulness associated with an action reflects the extent to which it is regarded as important, significant, and morally appropriate [28]. Engaging in meaningful activities fosters a positive self-perception, ultimately contributing to enhanced subjective well-being [27,29].

Our study focuses on a specific form of volunteerism: volunteering in environmental organizations. This focus is meaningful for several reasons [24,30]. First, environmental volunteerism represents a unique intersection of civic engagement and environmental concern, where individuals act not only out of altruism but also out of a commitment to ecological values and sustainability. This dual motivation may lead to distinct psychological benefits compared to other forms of volunteering. Second, environmental volunteering often involves physical, outdoor, and group-based activities—such as habitat restoration or community clean-ups—that are themselves linked to improved well-being through mechanisms like nature exposure, physical activity, and social connection. Finally, prior research on the relationship between volunteerism and subjective well-being has largely treated volunteerism as a general category, without accounting for the cause-specific nature of volunteering [5,6,7]. By examining environmental volunteerism specifically, our study contributes to a more differentiated understanding of how the context of volunteering influences well-being outcomes.

Several studies have shown that general volunteer activities are positively associated with subjective well-being, which is usually measured as self-rated happiness and life satisfaction (e.g., [7,31,32,33,34,35,36]). Among these, [31] specifically focused on public policies to promote volunteerism and consequently subjective well-being, [33,34] focused on voluntary activities and daily happiness, whereas [37] analyzed health benefits of volunteerism. It has been argued that volunteer activities can enhance happiness by providing psychological benefits, such as a sense of usefulness and finding meaning in life [37], and also by increasing levels of social integration through extended social participation and engagement (e.g., [38,39]).

Finally, in the economics literature, the economic benefits of volunteerism have been broadly identified in models examining public goods, private consumption, and investment [40,41,42,43,44]. The public goods model [40,41] focuses on altruistic benefit where the general motivation is to increase supply of public goods. In the private consumption model [42,43], self-value benefits are analyzed, and the general motivation includes joy from the act of volunteering, self-integration, social status, or social ethics. The investment model [44] discusses “exchange benefits” of volunteering such as gaining labor market experience, skills, and contacts and even signaling one’s ability to prospective employers. Microeconomic models along these lines are used to explain volunteer labor supply. Meier and Stutzer [45] provided broader theoretical considerations on the motivation to volunteer.

Our paper contributes to the literature by examining benefits of volunteer work in environmental organizations to the individuals themselves, in terms of their own mental and physical well-being as well as future volunteering in environmental organizations. We provide key insights into the strength, direction and robustness of the relation between environmental community work and subjective well-being.

3. Data and Methods

We use data from Add Health, a U.S. nationally representative school-based survey, which started with a sample of approximately 90,000 school-age adolescents enrolled in grades 7–12 in 1994–1995 in 132 schools nationwide [46]. A nationally representative subset of the respondents from the initial sample was re-interviewed at home in 1994–1995 (Wave I, n = 20,745, ages 12–19), in 1995–1996 (Wave II, n = 14,738, ages 13–20), in 2001–2002 (Wave III, n = 15,170, ages 18–26), in 2008–2009 (Wave IV, n = 15,701, ages 24–32) and in 2016–2018 (Wave V, n = 12,300, ages 32–42).

Our sample includes respondents interviewed in Wave III (2002) and Wave V (2016–2018). Among the 15,170 respondents who took part in Wave III interviews when respondents ranged in age from 18 to 26 years old (7140 males and 8030 females), 12,300 respondents participated in Wave V interviews conducted approximately 16 years later when respondents ranged from 32 to 42 years old (6187 males and 6113 females). In Wave V, these respondents answered questions about general (not only environmental) volunteering, physical and mental health, personality, social support, feelings, and experiences. We draw information on control variables from Wave I at-home interviews (1994–1995).

3.1. Outcome Variables

We investigate if environmental volunteerism is associated with subjective well-being that included measures of mental and physical well-being and continuing volunteering in Wave V. In our analyses, we did not exclude any outcome a priori, but used all available outcomes related to individual well-being. The outcome measures of mental health include perceived social status, measures of optimism and neuroticism (life orientation test), Perceived Stress Scale [47], suicidal ideation, and CES-D (Center for Epidemiologic Studies Depression Scale, [48]). Furthermore, we include the outcome of self-reported good physical health and a measure of continuing volunteerism.

Perceived social status ranges between 1 and 10 and is based on the question “Think of this ladder as representing where people stand in the United States. At the top of the ladder (step 10) are the people who have the most money and education, and the most respected jobs. At the bottom of the ladder (step 1) are the people who have the least money and education, and the least respected jobs or no job. Where would you place yourself on this ladder? Pick the number for the step that shows where you think you stand at this time in your life, relative to other people in the United States” [40].

To measure optimism and neuroticism, we created three binary indicator variables based on three items from the Life Orientation Test asked in Wave V which measure dispositional levels of optimism and neuroticism [49]: “How much do you agree or disagree with each statement about you as you generally are now, not as you wish to be in the future? (i) I am always optimistic about my future, (ii) I hardly ever expect things to go my way, (iii) Overall, I expect more good things to happen to me than bad”.

Wave V collected information on all four items of the Perceived Stress Scale such as “In the past 30 days, how often have you felt that you were unable to control the important things in your life” Each item was measured on a 0–4 scale (0 = Never, 1 = Almost never, 2 = Sometimes, 3 = Fairly often, 4 = Very Often). We summed responses, with summed scores ranging between 0 and 16 where higher scores indicate greater perceived stress (Cronbach’s alpha = 0.78).

Next, we created a variable indicating whether the respondent experienced suicidal thoughts or attempts in the past 12 months by combining affirmative responses to the questions “During the past 12 months, have you ever seriously thought about committing suicide?” and “During the past 12 months, how many times have you actually attempted suicide?” [40].

Add Health also collected responses on five items used in the CES-D scale (Center for Epidemiologic Studies Depression Scale) which measures depressive symptoms such as “During the past 7 days, I felt that I could not shake off the blues” [42]. The possible responses on each item ranged from 1 = never to 4 = always. We created a single depressive scale by summing responses, with the resulting scale ranging between 5 and 25 (Cronbach’s alpha = 0.82).

For physical health, we created a binary indicator of self-reported good physical health based on the question “How is your general physical health” where the indicator variable was coded 1 for the responses “excellent”, “very good” and “good,” or zero otherwise. Research has established that self-rated health has good predictive validity with respect to mortality [50,51].

Finally, we created a binary variable indicating continuing volunteerism in Wave V based on the question “In the past 12 months, about how many hours did you spend on volunteer or community service work?”.

Based on findings in prior literature on relationships between concerns about the environment and one’s well-being, we expected environmental volunteering to be positively related to measures of mental, and physical well-being and likelihood of continuing volunteering.

3.2. Independent Variables

Our main explanatory covariate was a binary indicator variable measuring environmental volunteerism in early adulthood. The covariate comes from Wave III, based on the question “Which of the following types of organizations have you been involved with in your volunteer or community service work in the last 12 months? Indicate all that apply: conservation, recycling, or environmental groups, such as the Sierra Club or the Nature Conservancy”. It is important to note that this environmental volunteerism question was only asked in Wave III and was not asked again in subsequent waves. This is the reason why we used Wave III for this question.

The other explanatory variables, which we used as statistical controls in multivariate analyses, come from Wave I and include basic demographic and socio-economic characteristics of the respondent and their family available in Wave I of Add Health: age, indicators for gender, race (white, Black), Hispanic ethnicity, Picture Vocabulary Test (PVT) as a proxy for cognitive abilities, log of yearly family pretax income, whether either parent had a college degree, and whether the adolescent lived with both biological parents. It is important to account for basic demographics such as race/ethnicity and cognitive abilities, which may determine socio-economic status, in multivariate analysis due to the known and persistent disparities in health [46].

Past literature has identified age, education, income, gender and other socio-economic factors as important factors that may impact levels of volunteerism and well-being and should be controlled in empirical analyses [5,7,52,53,54,55]. After restricting the sample to observations with valid Wave V sampling weights (n = 12,300), non-missing observations on the volunteering measure in Wave III (deleted 2082 observations), and non-missing observations on dependent variables (deleted additional 418 observations), our final analysis sample had n = 9800.

Add Health followed a strict privacy protocol where respondents answered questions on computers read through headphones, reducing social desirability bias when responding to the pro-environmental question compared to other surveys.

3.3. Methods

First, we used linear survey regression (linear probability model) to estimate covariate-adjusted associations of volunteering with conservation organizations in Wave III with four categories of outcome measures: mental, and physical well-being and likelihood of continuing volunteering.

Second, we used propensity score analysis, an alternative method of accounting for observed confounders by creating a matched or weighted sample in which distributions of measured covariates are similar between treated and untreated respondents. In our case, treated individuals were those who reported environmental volunteering and untreated were those who did not. Before conducting propensity score analyses, we checked for covariate balance by comparing standardized differences in means of the included covariates in the treatment model using the user-written Stata 14.2 command ‘pbalchk’ [56], and did not detect significant (>0.5 standard deviation) differences between treated and untreated respondents

In the first step of the propensity score analysis we used the logistic regression model and the independent variables to estimate propensity scores (predicted probabilities for each respondent of participation in conservation volunteering in Wave III, henceforth referred to as treated respondents). The use of logistic regression is a common choice for predicting propensity scores [57]. In the second step we conducted three types of analyses using estimated propensity scores in order to obtain average treatment effect (ATE) estimates of the effect of volunteering: stratification, weighting, matching. In addition, we used a combination of weighting and stratification.

During stratification, we stratified respondents into 10 quantiles (strata) based on the predicted propensity scores. The number of treated respondents ranged between 8 and 50 and the number of untreated respondents ranged between 930 and 975 across 10 strata, and standardized differences in means of observables (balance check) were not significant in any stratum. The estimates represent differences between treated and untreated respondents across 10 strata.

Next, during weighting we used estimated propensity scores to create sampling weights for linear regression adjustment. The sample used in these estimations was restricted to common support, i.e., to the range of propensity scores in which both treated and untreated respondents were observed. First, we present estimates using linear regression with sample weights equal to the inverse probability of treatment (IPT). Second, we present the same estimates with standardized mortality/morbidity ratio (SMR) weights. IPT weights change the distribution of confounders in both treated and untreated groups to be equal to the distribution in the entire sample. Estimates that use IPT weights compare expected outcomes when everyone received treatment to expected outcomes when no-one received treatment. In contrast, SMR weights only change the distribution of confounders for untreated respondents to match it to the distribution among treated respondents. The SMR weighted analysis compares outcomes for treated respondents with what their outcomes would be if they had remained untreated. The IPT and SMR weighted analyses produce the same estimates if treatment has the same effect on everyone; they produce different estimates if respondents likely affected by the treatment are more likely to receive the treatment [57].

In the third type of propensity score analysis, we use matching based on propensity scores. In “greedy matching”, every treated respondent is compared to every untreated respondent to find the closest match. This procedure continues until there are no more possible matches (Stata command ‘gmatch’ by [50]). In Coarsened Exact Matching (CEM) data are temporarily coarsened and exact matches are produced on these coarsened data, and then the analysis is computed on the un-coarsened, matched data [58]. Coarsened exact matching requires fewer assumptions and possesses more attractive statistical properties than other matching methods [52].

Finally, marginal mean weighting through stratification (MMWS) combines the above two methods—stratification and weighting based on the two different propensity-based weights—IPT and SMR [59,60].

All analyses were weighted using Wave V Add Health sampling weights, rendering estimates applicable to the underlying cohort.

4. Results

Table 1. Descriptive statistics.

(A) Descriptive Statistics for Everyone, n = 9800.
	Variable Label	Mean	SD	Min	Max
Outcome measures of subjective well-being
Social status in W5, 1–10	socstatus	5.494	1.874	1	10
I am always optimistic about my future, W5	optimism1	0.786	0.410	0	1
I hardly ever expect things to go my way, W5	optimism2	0.176	0.381	0	1
Overall, I expect more good things to happen to me than bad, W5	optimism3	0.769	0.421	0	1
Outcome measures of mental health
Perceived Stress Scale 4 (PSS-4) score: 0–16, W5	pss4	5.023	2.995	0	16
Suicidal thoughts or attempts in past 12 months, W5	suicideal	0.0720	0.258	0	1
Sum of 5 depression items, W5	cesditems	7.346	2.522	5	20
Outcome measure of physical health
Good or better health, W5	ghealth	0.868	0.339	0	1
Outcome measure of continuing volunteering
Volunteered in the past 12 months, W5	volunteer	0.396	0.489	0	1
Measure of environmental volunteering
Conservation volunteer past 12 months (0 or 1), W3	conserv	0.0270	0.163	0	1
Other covariates
Age, W1		15.01	1.731	11	19
Male		0.422	0.494	0	1
Black		0.202	0.402	0	1
Hispanic		0.143	0.350	0	1
Picture Vocabulary Test score, W1		101.0	14.34	10	141
Log annual family income, W1		3.616	0.851	−2.303	6.907
Both biological parents present, W1		0.556	0.497	0	1
Mom or dad has college degree, W1		0.371	0.483	0	1
(B) Descriptive statistics by conservation volunteering status in Wave III, n = 9800.
	Mean, Volunteered (n = 267)	Mean, Not Volunteered (n = 9533)	Difference		Cohen’s d
Outcome measures of subjective well-being
Social status in W5, 1–10	6.124	5.477	0.647 **		0.345
I am always optimistic about my future, W5	0.843	0.785	0.058 *		0.141
I hardly ever expect things to go my way, W5	0.112	0.178	−0.065 **		−0.173
Overall, I expect more good things to happen to me than bad, W5	0.854	0.767	0.087 **		0.207
Outcome measures of mental health
Perceived Stress Scale 4 (PSS-4) score: 0–16, W5	4.517	5.037	−0.520 **		−0.174
Suicidal thoughts or attempts in past 12 months, W5	0.045	0.072	−0.028 *		−0.105
Sum of 5 depression items below, W5	7.030	7.355	−0.325 *		−0.129
Outcome measure of physical health
Good or better health, W5	0.940	0.866	0.075 **		0.218
Outcome measure of continuing volunteering
Volunteered in the past 12 months, W5	0.584	0.390	0.194 **		0.397
Measure of environmental volunteering
Conservation volunteer past 12 months (0 or 1), W3	1.000	0.000	1.000		6.135
Other covariates
Age, W1	14.816	15.016	−0.200		−0.116
Male	0.487	0.420	0.066 *		0.136
Black	0.124	0.204	−0.081 **		−0.199
Hispanic	0.101	0.144	−0.043 *		−0.123
Picture Vocabulary Test score, W1	106.494	100.813	5.681 **		0.396
Log annual family income, W1	3.814	3.610	0.204 **		0.240
Both biological parents present, W1	0.607	0.555	0.052		0.105
Mom or dad has college degree, W1	0.547	0.366	0.181 **		0.375

Statistical significance for difference in means: * p < 0.05, ** p < 0.01.

A presents descriptive statistics for all variables used in analyses. In Wave III, 267 respondents (2.7% of the sample) volunteered for environmental and conservation organizations (of the original 280 responses, 13 responses were deleted due to missing data on other variables; naturally, the environmental volunteering group is much smaller than the non-volunteering group, but we still had a large enough sample to estimate statistically significant effects and the advantage of having thousands of untreated individuals is that the matching techniques can find close matches for the 267 treated individuals.) This Wave III question asks about a specific type of volunteering—time given to environmental or conservation organizations. However, in Wave V respondents were asked a general volunteering question. We had 39.6% of the respondents performing general volunteering work in Wave V.

When the sample was stratified by environmental volunteering status in Wave III (Table 1B), respondents volunteering in environmental organizations scored significantly better (based on significance testing and Cohen’s d [61] on the well-being measures in Wave V: they had higher perceived social status, volunteered more than others, were more optimistic, felt less stressed, showed fewer depressive symptoms, were in better health and experienced less suicidal ideation and attempts.

Table 2. Estimates from linear survey regression of the effect of conservation volunteering in Wave III on well-being and volunteering outcomes in Wave V, n = 9800.

	Socstatus	Optimism1	Optimism2	Optimism3	Pss4	Suicideal	Cesditems	Ghealth	Volunteer
Conservation volunteer, W3	0.49 **	0.09 **	−0.05 *	0.11 **	−0.57 **	−0.05 **	−0.58 **	0.09 **	0.20 **
	(0.14)	(0.03)	(0.02)	(0.04)	(0.20)	(0.01)	(0.16)	(0.01)	(0.04)
Age, W1	0.03	−0.00	0.00	0.01 *	−0.02	−0.01 *	−0.00	−0.00	0.01 **
	(0.02)	(0.00)	(0.00)	(0.00)	(0.03)	(0.00)	(0.03)	(0.00)	(0.00)
Male	0.02	0.02	0.02	−0.02 *	−0.37 **	0.01	−0.12	−0.02	−0.08 **
	(0.06)	(0.01)	(0.01)	(0.01)	(0.09)	(0.01)	(0.07)	(0.01)	(0.01)
Black	−0.17	0.08 **	−0.00	0.08 **	0.29 *	−0.02	0.03	−0.03	0.02
	(0.09)	(0.02)	(0.01)	(0.02)	(0.13)	(0.01)	(0.13)	(0.02)	(0.02)
Hispanic	0.09	0.03	0.03	0.02	−0.03	−0.01	−0.06	−0.01	−0.00
	(0.09)	(0.02)	(0.02)	(0.02)	(0.13)	(0.01)	(0.15)	(0.02)	(0.02)
Picture Vocabulary Test score, W1	0.01 **	−0.00 **	−0.00 **	0.00	−0.01	0.00	−0.00	0.00 **	0.00 **
	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)	(0.00)
Log annual family income, W1	0.19 **	0.01	−0.02 *	0.02	−0.06	−0.01 *	−0.10	0.01	0.00
	(0.05)	(0.01)	(0.01)	(0.01)	(0.07)	(0.01)	(0.07)	(0.01)	(0.01)
Both biological parents present, W1	0.20 **	0.01	−0.03 *	0.01	−0.19	−0.02 **	−0.24 **	0.02	0.03 *
	(0.06)	(0.01)	(0.01)	(0.01)	(0.10)	(0.01)	(0.07)	(0.01)	(0.01)
Mom or dad has college degree, W1	0.57 **	0.03 *	−0.06 **	0.05 **	−0.42 **	0.00	−0.22 *	0.06 **	0.10 **
	(0.07)	(0.01)	(0.01)	(0.02)	(0.11)	(0.01)	(0.10)	(0.01)	(0.02)
R squared	0.07	0.01	0.03	0.01	0.02	0.01	0.01	0.02	0.05

Note: Regression coefficients and SE in parentheses. Statistical significance for difference in means: * p < 0.05, ** p < 0.01. See Table 1A for the definition of the outcome variables shown in column headings.

presents the results from linear survey regression (linear probability model).

Table 3. Estimates of average treatment effects (ATE) of volunteering in Wave III on measures of well-being and volunteering in Wave V using propensity scores, n = 9800.

	Socstatus	Optimism1	Optimism2	Optimism3	Pss4	Suicideal	Cesditems	Ghealth	Volunteer
Panel A: Stratifying by propensity score in 10 strata
Conservation volunteer, W3	0.50 **	0.09 **	−0.05 *	0.11 **	−0.58 **	−0.05 **	−0.59 **	0.09 **	0.20 **
	(0.15)	(0.03)	(0.03)	(0.03)	(0.20)	(0.01)	(0.16)	(0.01)	(0.04)
Panel B: Using propensity scores for inverse-probability-weighted (IPT) regression adjustment, observations restricted to common support
Conservation volunteer, W3	0.41 **	0.09 **	−0.05	0.11 **	−0.56 **	−0.05 **	−0.55 **	0.09 **	0.19 **
	(0.15)	(0.03)	(0.03)	(0.03)	(0.21)	(0.01)	(0.16)	(0.01)	(0.04)
Panel C: Using propensity scores for weighted (SMR) regression adjustment, observations restricted to common support
Conservation volunteer, W3	0.35 **	0.06 **	−0.03	0.08 **	−0.33	−0.03 *	−0.21	0.05 **	0.15 **
	(0.11)	(0.02)	(0.02)	(0.02)	(0.17)	(0.01)	(0.14)	(0.02)	(0.03)
Panel D: Greedy matching using propensity scores with caliper = 0.0001
Conservation volunteer, W3	0.40	0.07	−0.11 *	0.17 **	−0.39	−0.05	−0.49 *	0.05 *	0.27 **
	(0.23)	(0.04)	(0.04)	(0.05)	(0.32)	(0.03)	(0.24)	(0.02)	(0.06)
Panel E: Coarsened Exact Matching using propensity scores
Conservation volunteer, W3	0.43 *	0.09 *	−0.04	0.08 *	−0.21	−0.03	−0.33	0.08 **	0.20 **
	(0.17)	(0.04)	(0.03)	(0.04)	(0.25)	(0.02)	(0.20)	(0.02)	(0.05)
Panel F: Marginal mean weighting (MMWS weights) through stratification using propensity scores
Conservation volunteer, W3	0.46	0.11 **	−0.08 **	0.06	−0.53 *	−0.04 *	−0.55 **	0.11 **	0.24 **
	(0.24)	(0.03)	(0.03)	(0.06)	(0.22)	(0.02)	(0.18)	(0.01)	(0.05)
Panel G: Marginal mean weighting (IPTW weights) through stratification using propensity scores
Conservation volunteer, W3	0.52 **	0.11 **	−0.08 **	0.07	−0.60 **	−0.04	−0.58 **	0.11 **	0.24 **
	(0.20)	(0.03)	(0.03)	(0.05)	(0.22)	(0.02)	(0.17)	(0.01)	(0.05)

Note: Regression coefficients and SE in parentheses. Statistical significance for difference in means: * p < 0.05, ** p < 0.01. See Table 1A for the definition of the outcome variables shown in column headings. All models include independent variables from Table 2.

presents the results from propensity score analyses. In each table, estimates are presented for mental and physical well-being and continuing volunteering.

In Table 3, Panel 1 presents estimates of ATE after stratification, and Panels 2 and 3 present estimates of ATE after using estimated propensity scores to create sampling weights for linear regression adjustment. Panel 2 presents estimates using linear regression with sample weights equal to the inverse probability of treatment (IPT); Panel 3 presents the same estimates but with standardized mortality/morbidity ratio (SMR) weights. Panels 4 and 5 implement matching based on propensity scores. Panel 4 implements “greedy matching”; Panel 5 implements Coarsened Exact Matching (CEM). Panels 6 and 7 present estimates using marginal mean weighting through stratification (MMWS) where Panel 6 uses MMWS weights and Panel 7 uses inverse probability of treatment weights.

5. Discussion of the Results

In this section, we interpret the results across three categories: (1) mental well-being, (2) physical health and future volunteering, and (3) comparisons with the previous literature.

5.1. Results for Measures of Mental Well-Being

The results from linear regression in Table 2 suggest that compared to respondents who did not volunteer in Wave III, those volunteering with environmental organizations earlier in life had significantly higher perceived social status and were more optimistic on the three measures of optimism/neuroticism. Propensity score results in Table 3 are qualitatively and quantitatively similar to results from survey linear regression in Table 2, although in some cases the effect of volunteering in Wave III lost statistical significance for some outcomes in some models.

Linear regression estimates in Table 2 also suggest that environmental volunteering in Wave III reduced perceived stress, reduced suicidal thoughts or attempts and reduced depressive symptoms. Estimates using propensity score analyses in Table 3 echo findings using linear regression in terms of the signs and magnitudes of estimates; however, most coefficients on volunteering lost their statistical significance in case of SMR weighted regressions (Panel 3), greedy matching (Panel 4) and coarsened exact matching (Panel 5). The smaller coefficients when using SMR weights in Panel 3 suggest that respondents whose mental well-being was more likely to improve due to conservation volunteering were less likely to participate in this volunteering.

It is important to note that there is no evidence in the literature that happier people would volunteer more and no empirical evidence of reverse causality between life satisfaction and voluntary activities [31,33,39]. Therefore, our results suggest that environmental volunteerism is likely to improve individual mental, and emotional well-being in the long run supporting prior research [20,21,22,26]: People who volunteer in early adolescence are likely to become more optimistic, more sociable, have higher perceived social status, display less stress and depressive symptoms.

5.2. Results for the Measures of Physical Health and Future Volunteering

According to linear survey regression in Table 2, environmental volunteering in Wave III was associated with greater likelihood of reporting being in good or better physical health and with a greater likelihood of volunteering later in life, in Wave V. The same result was found in all models using propensity scores in Table 3 with the coefficient on volunteering in Wave III being quantitatively similar across models. These results suggest that being involved in and volunteering time for environmental organizations earlier in life translates to better physical health and more hours spent on volunteering or community service work later in life.

We note that R-squared values in all of the models in Table 2 are low (<0.1), which indicates that the included basic demographic and socio-economic controls as well as the environmental volunteering together explain very little variation in the measures of mental and physical well-being and in future volunteering across the respondents in our sample.

5.3. Interpretation and Relation to Existing Studies

Our paper contributes to the literature by examining the long-term impact of environmental volunteerism—specifically, membership in environmental and conservation organizations earlier in life—on various outcomes, including subjective well-being, mental and physical health, and continued volunteerism in later life. Our results support existing studies that report a positive relationship between environmental concerns and subjective well-being (e.g., [5,6,7,20,21,22,26]). However, our contribution lies in demonstrating how concrete pro-environmental behaviors—such as volunteering in environmental organizations—impact well-being and foster sustained environmental engagement.

Our study explicitly distinguishes between holding environmental concerns and engaging in environmental volunteering. The latter aligns more closely with the concept of pro-environmental behavior, which refers to deliberate actions aimed at reducing individuals’ negative environmental impact [2]. By focusing on environmental volunteerism, we address the value–action gap, emphasizing behavior over stated concern. Moreover, unlike studies that rely on broadly defined measures of volunteering due to data limitations (e.g., [7]), our analysis draws on specific data on environmental organization membership, allowing for a more targeted investigation. Finally, our study evaluates the effects of volunteering on well-being over a 16-year period, offering a long-term perspective that is especially relevant for outcomes—such as mental and physical health—that may develop gradually over time.

6. Concluding Remarks

Our study is the first using a nationally representative dataset of US residents to investigate our associations using quasi-experimental methods and matching techniques. Our study contributes to understanding the link between volunteerism and individual well-being using a unique dataset. We find that environmental volunteerism is consistently associated with improvements in subjective, mental, and physical health outcomes.

According to our interpretation, these results matter for environmental policy designed to change people’s behavior. The existence of a value–action gap implies that pro-environmental behavior is not an automatic result of environmental concerns, and there is a gap between people’s perception of environmental problems and possible actions [2,62,63,64]. The volunteering in environmental organizations variable that we used in our study is more aligned with a pro-environmental behavior concept describing activities aimed at reducing the negative impact individuals confer on the environment. Consequently, it might be plausible to think that the value–action gap can shrink if more people understand the positive association of environmental community service and volunteerism with their own well-being. If a person knows that volunteering in environmental organizations is good for their own well-being and the environment, we can expect to see more behavioral change towards environmental volunteerism. To promote sustained engagement in voluntary work, policymakers can complement structural opportunities with targeted informational and educational campaigns that emphasize the personal benefits of volunteering, including improvements in well-being. Policy instruments such as public service announcements, social marketing, awards, and recognition programs can be strategically employed to encourage participation. Even when the primary policy objective is to increase volunteer rates rather than enhance individual well-being per se, highlighting the positive psychological effects of volunteering can serve as an effective strategy for encouraging engagement.

We end by considering three limitations to our study. First, any observed behavior can stem from selection based on unobserved personality traits that may be unaccounted for by covariates. For example, if people with higher levels of life satisfaction are more likely to participate in voluntary activities (reverse causality), then positive effects found in our and other studies may be overstated. However, ref. [39] presented robust evidence that volunteering has a significantly positive impact on well-being, and found no evidence that this positive relationship was driven by reversal causality, when examining the well-being impact of volunteering. This was also the main assumption in earlier studies as it is unclear why happier individuals should also worry more about the environment [31,33]. In the absence of experimental data, it is more difficult to identify causality between environmental volunteerism and individual well-being. While we controlled for possible confounders and individual-specific heterogeneity, observational studies come with some skepticism on the identification of causal effects. However, we note that unless it is a natural experiment, experimental approaches in this context come with their own problems. While our study does not establish causality, it nonetheless advances a step further and adjusts the difference in outcomes for effects of several important covariates that might cause selection bias. Furthermore, this is the only longitudinal dataset for the US asking about environmental volunteering and measures of well-being years after.

Second, the fact that our measure of volunteering for environmental groups was self-reported may be concerning. However, its meaning is straightforward and behaviorally specific because it is based on a discrete yes/no question querying involvement with conservation, recycling, or environmental groups, such as the Sierra Club or Nature Conservancy. While being a relatively more specific and relevant measure of pro-environmental behavior as opposed to questions asking about general levels of care about the environment, it may still be subject to measurement error. However, this binary measure is less prone to measurement error compared to other measures differentiating between levels of intensity of volunteering.

Third, some studies pointed out issues with comparing values of ordinal variables across people and time, and researchers do not have a full consensus on how to interpret these subjective well-being measures [65,66]. However, our study contributes to understanding the link between environmentalism and well-being by using a unique dataset and established psychological measures. Future research should examine how the intensity of volunteerism influences subjective well-being. Survey designs that capture variation in volunteering intensity would offer deeper insights into the relationship between environmental volunteerism and well-being.