Environmental volunteer well-being: Managers’ perception and actual well-being of volunteers

Volunteer well-being

Many studies have shown that volunteering is closely linked to increased well-being of volunteers (Binder & Freytag, 2013; Borgonovi, 2008; Greenfield & Marks, 2004; Jenkinson et al., 2013; Koss & Kingsley, 2010; O’Brien et al., 2010; Son & Wilson, 2012; Stukas et al., 2016; Thoits & Hewitt, 2001; Townsend, 2006; Van Willigen, 2000; Wheeler et al., 1998; Wilson, 2000). However, studies have used different definitions of well-being, and have therefore measured different constructs, which have often included only some aspects of well-being instead of taking a holistic approach. Two main approaches to conceptualising well-being prevail: hedonism and eudaimonia. Hedonism is the idea that maximisation of pleasure is the goal and the way to happiness for all humans, whereas eudaimonia proposes that striving to lead a meaningful life and achieve optimum functioning is the way to happiness (Aristotle, 2009; Diener, 2000; Ryan & Deci, 2001; Ryff, 1989). The two approaches have informed research into human well-being with different methods proposed for the study of well-being. Methods based on the study of ‘subjective well-being’ includes measures of positive affect, negative affect and life satisfaction, a mixture of both hedonic and eudaimonic well-being (Bradburn, 1969; Diener, 1984; Diener, 1994; Diener et al., 1999). The study of ‘psychological well-being’ on the other hand measures only eudaimonic elements of life, such as self-acceptance, positive relations with others, autonomy, environmental mastery, purpose in life and personal growth, leaving out the hedonic focus on pleasures (Ryff, 1989; Ryff, 1995; Ryff, 2014).

Though some aspects of volunteer well-being have been studied in depth, no previous studies have investigated volunteer managers’ perceptions of the well-being of their volunteers. As volunteer managers are responsible for the well-being of their volunteers, and as improved volunteer well-being is often an important outcome for volunteers, organisations and society (O’Brien et al., 2011), it is vital that managers’ perceptions of the well-being of their volunteers correspond to actual volunteer well-being. The cumulative evidence from a broad range of studies (see meta-analyses and reviews in Jenkinson et al., 2013; Musick & Wilson, 2008; Wilson, 2000; Wheeler et al., 1998) is that volunteering has a positive relationship with a wide range of elements within the concept of well-being, though causation can be difficult to determine (Greenfield & Marks, 2004). Previous studies have investigated the effect of volunteering on subjective well-being (e.g. Binder & Freytag, 2013; Harlow & Cantor, 1996; Windsor et al., 2008) or psychological well-being (e.g. Ho, 2015), or a combination of one of these along with other elements of well-being, such as social well-being, trust, self-esteem, depression or physical health (e.g. Greenfield & Marks, 2004; Koss & Kingsley, 2010; O’Brien et al., 2010; Son & Wilson, 2012; Stukas et al., 2016; Thoits & Hewitt, 2001; Townsend, 2006). Some studies show that volunteering leads to increased well-being (Borgonovi, 2008; Piliavin, 2009; Piliavin & Siegl, 2007), while other studies show that people higher in well-being are also more likely to volunteer (Gimenez-Nadal & Molina, 2015; Greenfield & Marks, 2004) and to volunteer more hours (Son & Wilson, 2012; Thoits & Hewitt, 2001). Most likely the causality runs both ways between volunteering and well-being (Binder & Freytag, 2013; Gimenez-Nadal & Molina, 2015) in a ‘virtuous cycle’ where happy and healthy people volunteer more and volunteers are happier and healthier (Brooks, 2007). Environmental volunteering could further enhance this virtuous cycle, as spending time in nature has been linked to increased well-being (Frumkin, 2001).

Environmental volunteer well-being

Only a few studies have focused specifically on the relationship between environmental volunteers and their well-being (e.g. Koss & Kingsley, 2010; O’Brien et al., 2010; Townsend, 2006), as many studies have used cohort datasets where volunteering type was often heterogeneous or not described (Jenkinson et al., 2013). Volunteering in nature has been linked to well-being benefits for volunteers, including improved social networks (Bell et al., 2008; Gooch, 2005; Koss & Kingsley, 2010; Muirhead, 2011; O’Brien et al., 2010), increased personal satisfaction and feelings of enjoyment (Koss & Kingsley, 2010; Muirhead, 2011), and improved health and well-being (Koss & Kingsley, 2010; O’Brien et al., 2010; Pillemer et al., 2010). Environmental volunteering can have a positive effect, not only by increasing positive indices of well-being, but also by reducing negative indices such as reducing stress (Guiney & Oberhauser, 2009; O’Brien et al., 2010) and depression (Pillemer et al., 2010). Furthermore, environmental volunteering offers the added benefit of providing opportunities for volunteers to spend time in nature, which can lead to a better connection or re-connection with nature for the volunteers (Bell et al., 2008; Guiney & Oberhauser, 2009). It can also lead to volunteers gaining an increased understanding of the natural environment (Koss & Kingsley, 2010) and thereby also an enhanced sense of place (Evans et al., 2005; Gooch, 2005). A closer connection to nature has been shown to enhance people’s well-being (Bowler et al., 2010; Kellert & Wilson, 1993), and therefore it could be expected that environmental volunteers would benefit more from their volunteering than other types of volunteers. Practical conservation volunteering requires stamina and physical strength and it provides a way to exercise and gain improved fitness (Guiney & Oberhauser, 2009; O’Brien et al., 2010), which can also reinforce positive well-being (Pretty et al., 2005).

To better understand these relationships between volunteering and well-being, a more holistic and multidimensional approach to well-being, including both hedonic and eudaimonic elements, as well as social elements, would be well suited (Piliavin, 2009). Such a holistic approach to well-being is gaining acceptance (Forgeard et al., 2011; Keyes, 2002; Ryan & Deci, 2001), and one proposed multidimensional model of well-being is Seligman’s (2011) PERMA model. It is a construct with five contributing elements (PERMA): 1) ‘Positive emotion’, which encompass present positive feelings, life satisfaction and positive emotions about the future; 2) ‘engagement’, which is employing one’s strengths to a task, becoming fully absorbed in the task and therefore completely losing track of time, also referred to as getting into ‘flow’ (Csikszentmihalyi, 1975; Csikszentmihalyi, 1991; Seligman, 2011); 3) ‘positive relationships’, which are fundamental to a good life according to Seligman (2011), and Baumeister & Leary (1995) have also defined it as a basic human need that is essential for well-being; 4) ‘meaning’, which includes feelings of doing something worthwhile and having a purpose and direction in life, something which is crucial to well-being as, according to Seligman (2011), most people have a need to belong to or serve something they believe is larger than themselves, e.g. their family, an organisation or a religious group; and 5) ‘achievement’, often pursued for its own sake by individuals setting their own personal goals or striving to achieve recognition in the wider world, e.g. winning an award or accumulating wealth. Seligman (2011) did not propose a measure for his PERMA model but Butler & Kern (2016) subsequently developed the PERMA-Profiler (PERMA-P), a scale based on the PERMA model, which also includes additional elements of well-being. The additional elements in the PERMA-P are 1) ‘negative emotion’ from the concept of subjective well-being acknowledging the importance of both positive and negative aspects of well-being; 2) ‘health’, which can be considered a core part of well-being; 3) ‘loneliness’, which is a strong predictor of many negative life outcomes; and 4) ‘overall happiness’, which allows an overall assessment after reflecting on specific elements of well-being (Butler & Kern, 2016).

Aims and research questions

This paper aims to use a multidimensional approach to well-being to explore the immediately experienced and later remembered well-being of environmental volunteers, as well as their general well-being and to compare this to the well-being of participants in other types of nature-based activities and volunteering. It also aims to compare volunteer managers’ perception of their volunteers’ well-being with the self-reported well-being of the volunteers. These aims were addressed through the following research questions: 1) How does environmental volunteering immediately affect participants’ sense of well-being, and how does that compare to the immediate effect of other types of nature-based activities on participants’ sense of well-being? 2) How well do volunteers sustain the memory of this immediately experienced sense of well-being after they have gone home? 3) How do volunteer managers perceive the effect of volunteering on the well-being of their volunteers? 4) How does the volunteer managers’ perception of volunteer well-being compare to volunteers’ actual sense of volunteering-related well-being?

Participants

Ethics. This research project was approved through the ethics approval process at Bournemouth University (ref ID 2419). All participants provided written informed consent for participation.

Study 1. The onsite study was conducted between October 2014 and November 2015 and involved participants from 13 organisations from Southern England, divided into four types of activities: Biodiversity monitoring, practical conservation volunteering, walking, and students conducting fieldwork as part of their university course (Table 2). Environmental organisations were invited to participate in the study based on them conducting volunteer activities in groups. Control groups were invited based on their group activity being conducted in the same natural environments as the volunteer activities of the environmental organisations. To determine if environmental volunteering had a different effect on well-being compared to other non-altruistic activities performed outdoors, students and walkers were surveyed in addition to environmental volunteers. Students were chosen as the control group to the biodiversity monitoring volunteers, as both groups were conducting ecological fieldwork in similar areas, but whereas volunteering is often seen as altruistic (Smith, 1981; Unger, 1991), students did the fieldwork because it was a requirement of their university courses. Walking groups were chosen as the control group for the practical conservation volunteers as both activities were performed outdoors in similar areas and were somewhat physically demanding, but the purpose of the activities were again different, with volunteering being partly altruistic and walking only benefitting the walkers themselves. Also, walking is the most popular activity in the natural environment in England (Natural England, 2015) and walking programmes are promoted as health interventions to decrease negative affect and mental illness and increase well-being in participants (Iwata et al., 2016; Marselle et al., 2014). The survey was designed as a paired before-activity and after-activity survey to measure general level of well-being and experienced level of well-being during an activity, respectively. Activity participants only completed questionnaires once to ensure independent samples even if they participated in activities later where other activity participants completed questionnaires.

Studies 2 and 3. Both online surveys were open to anyone with the link between September and December 2015. Environmental organisations involved in study 1 as well as other worldwide environmental organisations and volunteer centres in the UK were contacted directly and asked to invite their volunteers and volunteer managers to participate and the surveys were also sent out more widely through professional networks. Study 2 investigated the general level of well-being of former and potential volunteers as well as the remembered level of well-being during volunteering of current volunteers. In Study 2, a total of 417 responses were received with completed questions about well-being. This sample comprised 53% females and 47% males. Age ranged from 18 to 94 years old (mean=54.86, SD=16.10). Most respondents had at least one university degree (65.23%) and many were retired (48.68%), some were in full-time (21.10%) or part-time (13.19%) employment and few were students (6.95%), not currently employed (5.28%) or homemakers (1.20%). Respondents were from 11 different countries, with the majority residing in the United Kingdom (88.49%). They named 118 different organisations they previously or currently volunteer for or would like to volunteer for in the future. Respondents included people from three different periods: former volunteers (18%), current volunteers (70%) and potential future volunteers (12%). They were grouped into four types of volunteers: biodiversity monitoring volunteers (BM; 21%), practical conservation volunteers (PC; 34%), biodiversity monitoring volunteers also performing practical conservation work (BMPC; 25%), and all other types of volunteers (19%) (Table 3).

Study 3 investigated the perceived level of well-being of volunteers by former and current volunteer managers. A total of 96 responses were received with completed questions about well-being. This sample comprised 61% females and 39% males. Age ranged from 19 to 74 years old (mean=43.01, SD=13.03). Most respondents had at least one university degree (80%) and most respondents were in full-time (69%) or part-time (13%) employment, few were retired (10%), students (2%), not currently employed (1%) or homemakers (1%). Respondents were from 10 different countries, with the majority residing in the United Kingdom (80%). Respondents included people from two different periods: former volunteer managers (14%) and current volunteer managers (86%), and they identified 62 different organisations they previously or currently manage volunteers for. They were grouped into four types of volunteering similarly to the volunteers in Study 2: BM (20%), PC (26%), BMPC (35%) and all other types of volunteering (19%) (Table 4).

Data analyses

Deriving the well-being factors. The first step in exploring well-being was to test if the structures of self-reported well-being and managers’ perception of volunteer well-being were consistent with the proposed seven-factor PERMA-Profiler (PERMA-P) model (Butler & Kern, 2016). This was done by performing exploratory factor analysis (EFA) on a subsample of self-reported well-being data to generate a best fit model. The generated model and the original seven-factor PERMA-P model were subsequently tested for best fit through confirmatory factor analysis (CFA) using the other subsample of collected data from participants, and the total combined sample. EFA was also performed on the volunteer manager data sample to generate a best fit model and confirmatory factor analysis was run on the generated model, the model generated from the self-reported subsample and the original seven-factor PERMA-P model to determine the best fit model.

Self-reported well-being: Only complete responses were used for factor analysis (n=1157) (Figure 1). The data were split in two subsamples to develop (n=645) and test (n=512) the factor model. The development sample consisted of all onsite and online respondents to questionnaires measuring activity-related well-being, which included volunteers and control activity participants from Study 1 (‘after-activity survey’) and current volunteers from Study 2. The test sample consisted of all onsite and online respondents to questionnaires measuring general well-being which included volunteers and control activity participants from Study 1 (‘before-activity survey’) and former and potential volunteers from Study 2. The largest subsample was used as the development sample for the EFA.

Figure 1. Analysis flowchart for determining the best fit model for self-reported well-being factors.

The first step in determining the best fitting model was to test the factorability of the items in the development subsample with the Kaiser-Meyer-Olkin measure of sampling adequacy, recommended to be >0.60, and with Bartlett’s test of sphericity, where significance indicates the data are suitable for factor analysis (Dziuban & Shirkey, 1974). The first step in EFA is to determine the number of factors to extract. There is no set formula for determining this number and it is determined by using a variety of methods and interpretation of the data (Matsunaga, 2010). Several methods were used to determine the number of factors to extract, including parallel analysis (Horn, 1965), the Kaiser-Guttman criterion (counting only Eigenvalues above one, Kaiser, 1960), Velicer’s minimum average partial (MAP) test (Velicer, 1976) and visual inspection of the scree plot (Cattell, 1966). EFA using ordinary least squares to find the minimum residual (minres) solution with oblique (promax) rotation, which allows factors to be correlated, were performed for relevant models. To determine overall best fit model, results were evaluated using the root mean square error of approximation (RMSEA). RMSEA <0.05 indicate a good fit and between 0.05 and 0.08 indicate a fair fit (MacCallum et al., 1996). Cronbach’s α (Cronbach, 1951) was calculated for each factor to test internal reliability of factors. Cronbach’s α values >0.70 are considered acceptable (Nunnally, 1978), though for scales with 6 or fewer items lower α values may be acceptable (Cortina, 1993). Items with factor loadings <0.04 or loading on two factors with the difference between primary and secondary loadings <0.03 were removed from the dataset before further analyses, a suggested way of dealing with inconclusive factor loadings (Matsunaga, 2010). The best factor model was determined by choosing the model with optimal model fit indices, high internal reliability of factors and best interpretability of the data. CFA is a method to test if a certain predetermined model is a good fit for a data sample. CFA was performed for the best fit model developed from the EFA, the original seven-factor PERMA-P model and a generic one-dimensional control model using the test sample and the combined development and test sample. Model fits were evaluated using RMSEA, the standardised root mean residual (SRMR), comparative fit index (CFI) and the Tucker Lewis Index (TLI), and models were compared for best fit using χ² difference tests. SRMR below 0.08 is considered a good fit, and TLI and CFI values >0.90 are considered acceptable and close to or above 0.95 are considered good fits (Hu & Bentler, 1999).

Volunteer managers’ perception of volunteers’ well-being: Only complete responses from former and current volunteer managers were used for factor analysis (n=96) (Figure 2). Due to the limited sample size, it was not possible to split the data into a development and a test sample, as sample size should be at least 100–200 per subsample to perform the analysis (MacCallum et al., 1996). EFA was performed on the complete sample, following the method described above, including testing factorability of items, determining number of factors to extract and using oblique (promax) rotation for the EFA. The best fit model was determined also following the described method above by evaluating RMSEA, interpretability and Cronbach’s α. Items with inconclusive factor loadings were removed. CFA was then performed on the volunteer manager data sample using the best-fitting model from the EFA, the model developed from the self-reported well-being sample EFA described above, the original seven-factor PERMA-P model and a one-dimensional control model. Model fit for all models were evaluated using RMSEA, SRMR, CFI and TLI, and models were compared for best fit using χ² difference tests.

Figure 2. Analysis flowchart for determining best fit model of perceived volunteer well-being factors by volunteer managers.

Influence of volunteering type and other variables on well-being scores. As data were non-normally distributed, non-parametric tests were used in all analyses. As samples in the onsite survey (Study 1) had subject replication, Wilcoxon signed-rank tests were used to test for differences in the level of general well-being and level of activity-related well-being within the four groups of activity participants. For all other comparisons without subject replication, Wilcoxon rank sum tests were used to test for differences in levels between general and activity-related well-being. Kruskal-Wallis tests with post hoc Dunn’s test with Bonferroni correction were used to identify significantly different levels of actual and perceived well-being between the four different types of volunteers (Studies 2 and 3) and between managers in the four different types of volunteering (Study 3), respectively.

Stepwise multiple regression was performed to examine if there were any effects of external variables on overall mean well-being, calculated as the mean of all well-being items (23 items) with negative items, i.e. negative emotions and loneliness, reverse scored. Variables included in Study 1 were volunteer frequency, tenure and hours per month volunteered, and specific variables on the day: weather, group size, hours volunteered, volunteer manager experience and type of volunteering. In Study 2, variables included were volunteering type, as well as demographic variables (age, gender, education, country). Variables included in Study 3 were volunteering type, period and manager tenure, as well as demographic variables (age, gender, education, country).

Statistical analysis. All statistical analyses were completed using RStudio v3.2.3 (RStudio Team, 2015). The nFactor package v.2.3.3 (Raiche, 2010), psych package v.1.5.8 (Revelle, 2016) and the GPArotation package v.2014.11-1 (Bernaards & Jennrich, 2005) were used for exploratory factor analysis, the lavaan package v.0.5-20 for R was used for confirmatory factor analysis (Rosseel, 2012) and the ggplot2 package v.2.0.0 was used to create graphs (Wickham, 2009).

Studies 1 and 2: Immediate and remembered effects of environmental volunteering, other nature-based activities and other types of volunteering

Deriving the self-reported well-being factors. Factorability of the items in the development sample was supported by a Kaiser-Meyer-Olkin measure of 0.94 and a significant Bartlett’s test of sphericity (χ²(210)=8448.17; p<0.001), indicating the data were fit for factor analysis. The number of factors to extract was determined by evaluating several factor extraction results: parallel analysis suggested six factors, the Kaiser-Guttman criteria suggested four factors, Velicer’s minimum average partial test identified three factors and visual inspection of the scree plot suggested between two and five factors. Three-, four-, five- and six-factor models were evaluated through exploratory factor analysis and Cronbach’s α for individual factors for each model were also evaluated. The five-factor model provided the clearest structure with a good fit (RMSEA = 0.056 [90% confidence interval = 0.048, 0.062]). Five of the seven factors could be interpreted as factors from the PERMA-P (Table 5): ‘Engagement’ (four items, α = 0.79), ‘relationships’ (three items, α = 0.77), ‘meaning’ (two items, α = 0.88), ‘negative emotions’ (three items, α = 0.64) and ‘health’ (three items, α = 0.92). One ‘positive emotion’ item, ‘In general, how often do you feel joyful?’, loaded on the ‘engagement’ factor. One ‘achievement’ item, ‘How often do you achieve the important goals you have set for yourself?’ loaded on the ‘meaning’ factor, but was dropped to substantially improve internal reliability of factor and overall model fit. Five items failed to load conclusively on any one factor and were left out of the subsequent confirmatory factor analysis.

CFA was run on the test sample and the combined development and test sample with the five-factor model developed from the EFA. Model fit was acceptable for the test sample based on all fit indices (RMSEA (0.076 [0.067; 0.085]), SRMR (0.066), CFI (0.938) and TLI (0.918)). Model fit was good for the combined development and test sample based on SRMR (0.055), CFI (0.955) and TLI (0.940) indices and was acceptable based on RMSEA (0.069 [0.064; 0.075]). The five-factor model from the EFA fitted the test sample significantly better than the original seven-factor PERMA-P model (Δχ²(88) = 530; p<0.001) or a generic one-factor model (Δχ²(109) = 1565; p<0.001). The five-factor model also fitted the combined development and test sample significantly better than the original seven-factor PERMA-P model (Δχ²(88) = 788; p<0.001) or a generic one-factor model (Δχ²(109) = 3717; p<0.001). Factor correlations based on the combined test and development sample are summarised in Table 6, and show that all factors were significantly correlated.

External factors and volunteer well-being. Volunteers spending more hours volunteering per month, and for Study 2 also spending more time volunteering outdoors, reported higher levels of overall well-being. For volunteers in Study 1, this result came from stepwise multiple regression, which reduced the model for predicting the overall mean volunteering-related well-being score to only include the number of hours spent volunteering per month as a significant factor (F_1,164 = 5.55; p<0.05; R² = 0.03). For the current volunteers in Study 2, stepwise multiple regression reduced the model for predicting the overall mean volunteering-related well-being score to include the number of hours spent volunteering per month (p<0.001) and the amount of time spent outdoors while volunteering (p<0.001) as significant factors (F_2,225 = 11.69; p<0.001; R²_adj = 0.09). The number of hours spent volunteering per month (r=0.22; p<0.001) and the amount of time spent outdoors while volunteering (r=0.21; p<0.01) were both significantly positively correlated with the overall mean volunteering-related well-being score.

Study 1: How does environmental volunteering immediately affect well-being? Mean scores were calculated for each well-being element for both general well-being and activity-related well-being in the four participating groups: Biodiversity monitoring volunteers, practical conservation volunteers, students and walkers (Table 7). All groups rated most of their activity-related well-being significantly better than their general well-being with the positive indices, ‘engagement’, ‘relationship’, ‘meaning’, ‘health’ and ‘happiness’, rated significantly higher and the negative indices, ‘negative emotions’ and ‘loneliness’, rated significantly lower for activity-related well-being than for general well-being (Wilcoxon signed-rank test; p<0.05 for all; Figure 3). The only exceptions were students’ rating of ‘meaning’, which was not significantly different between generally in life and during their fieldwork, and their rating of ‘engagement’, which was significantly lower for activity-related well-being than generally in life.

Figure 3. Differences between paired general well-being scores and activity well-being scores of participants in nature-based activities.

Differences between paired general well-being scores (light grey) and activity well-being scores (dark grey) for biodiversity monitoring volunteers, practical conservation volunteers, students and walkers (±SE bars). ‘Engagement’, ‘relationship’, ‘meaning’, ‘negative emotion’ and ‘health’ factor scores were means of factor item aggregates. ‘Loneliness’ and ‘happiness’ were single item measures (Wilcoxon signed-rank tests;*p< 0.05, **p<0.01, ***p<0.001).

Comparing biodiversity monitoring volunteers to their student control group for general well-being there was one significant difference, as volunteers rated their ‘health’ significantly higher than students did (Wilcoxon rank sum test; p<0.05; Figure 4). Volunteers also rated their ‘negative emotions’ slightly lower than students did (Wilcoxon rank sum test; p<0.06). When comparing their activity-related well-being, however, there were significant differences in all elements of well-being, except ‘loneliness’, as volunteers consistently rated positive indices significantly higher and ‘negative emotions’ significantly lower than students did (Wilcoxon rank sum tests; p<0.01 for all).

Figure 4. Differences between biodiversity monitoring volunteers and students in their level of general and activity well-being.

Differences between biodiversity monitoring volunteers (BM) and students (S) in their level of general well-being (light grey) and activity well-being (dark grey) (±SE bars). ‘Engagement’, ‘relationship’, ‘meaning’, ‘negative emotion’ and ‘health’ factor scores were means of factor item aggregates. ‘Loneliness’ and ‘happiness’ were single item measures (Wilcoxon rank sum test; ·p<0.06, *p<0.05, ***p<0.001).

Comparing practical conservation volunteers to their walker control group for their general level of well-being there was one significant difference, as volunteers rated ‘relationships’ significantly higher than walkers did (Wilcoxon rank sum test; p<0.01; Figure 5). This difference in ‘relationship’ ratings was even more significant when comparing their activity-related well-being (Wilcoxon rank sum test; p<0.001). Also negative indices showed differences in activity-related well-being with volunteers rating their ‘negative emotions’ significantly lower than walkers (Wilcoxon rank sum test; p<0.05) and rating their ‘loneliness’ lower than walkers.

Figure 5. Differences between practical conservation volunteers and walkers in their level of general and activity well-being.

Differences between practical conservation volunteers (PC) and walkers (W) in their level of general well-being (light grey) and activity well-being (dark grey) (±SE bars). ‘Engagement’, ‘relationship’, ‘meaning’, ‘negative emotion’ and ‘health’ factor scores were means of factor item aggregates. ‘Loneliness’ and ‘happiness’ were single item measures (Wilcoxon rank sum tests; · p<0.06, *p<0.05, **p<0.01, ***p<0.001).

Comparing the two different types of environmental volunteers, the biodiversity monitoring volunteers and the practical conservation volunteers, there were no significant differences in their levels of general (Wilcoxon rank sum tests; p>0.07 for all) or activity-related (Wilcoxon rank sum tests; p>0.30 for all) well-being, suggesting that irrespective of the type of environmental volunteering performed, the effect on well-being is equally positive.