Mindfulness and social identity: Predicting well-being in a high-stress environment

Research has shown that group memberships, and associated social identities, contribute to positive well-being (Haslam, Cruwys, Haslam, Dingle, & Chang, 2016; Haslam, Jetten, Cruwys, Dingle, & Haslam, 2018; Haslam, Haslam, Jetten, Cruwys, & Dingle, in press; Jetten, Haslam, Haslam, Dingle, & Jones, 2014; Jetten et al., 2017). Social identities can make people more resilient, particularly when challenged (Postmes & Jetten, 2006). In parallel, a burgeoning body of empirical evidence shows that mindfulness enhances mental well-being and resilience (eg, Brown & Ryan, 2003; Christopher et al., 2016; Jha, Morrison, Parker, & Stanley, 2017; Nila, Holt, Ditzen, & AguilarRaab, 2016; Shonin, Van Gordon, Dunn, Singh, & Griffiths, 2014; Spinelli, Wisener, & Khoury, 2019). Yet, in spite of predicting the same psychological outcomes, there is no research examining the effects of social identity and mindfulness simultaneously. Our paper aims to fill Received: 18 December 2019 | Revised: 29 May 2020 | Accepted: 1 August 2020 DOI: 10.1111/jasp.12708

this gap and explore the combined and differential role of both constructs when predicting well-being.
The current research has three main goals. First, our research aims to expand the empirical evidence on the social identity approach to health and well-being, in particular, the "social cure" literature (Jetten, Haslam, & Haslam, 2012) in an applied context. Second, our research aims to contribute to the mindfulness literature by adding to the emergent body of work on mindfulness in high-stress contexts such as prisons. Third, we aim to combine the fields of social identity with mindfulness and explore the contribution of social identification and dispositional mindfulness to well-being outcomes.
The two studies presented in the current paper were conducted after prison inmates took part in a group-based mindfulness training program (for information about the training see Mahdon, Heilbrunn, Adarves-Yorno, & Jones, 2017). The training aimed to enhance inmates' mental well-being and resilience by increasing their mindfulness capability and their social identification with a mindful group.
In this prison setting, we examined the predictive power of mindfulness and identification with a mindfulness identity. The outcome variables were two psychological measures: mental well-being and resilience (Study 1 & 2) and self-reported substance use, which served as a proxy behavioral indicator of well-being (Study 2).

| MENTAL WELL-B EING , RE S ILIEN CE , AND SUBS TAN CE US E
Mental well-being, resilience, and a reduction in substance use are priority areas for health in prisons according to the World Health Organization (Enggist, Møller, Galea, & Udesen, 2014). This assessment reflects the higher percentage of prisoners suffering from mental health problems and substance use disorder than the nonprison population (Fazel, Hayes, Bartellas, Clerici, & Trestman, 2016). Furthermore, promoting mental well-being, resilience, and a reduction in substance use in prisons has been shown to improve reintegration into society and reduce re-offending (e.g., Koehler, Humphreys, Akoensi, de Ribera, & Loesel, 2014).
Problematic substance use has high comorbidity rates with mental health disorders, particularly depression and anxiety (Lai, Cleary, Sitharthan, & Hunt, 2015).
Mental well-being is a complex construct, which covers concepts of hedonic and eudemonic well-being as well as both affect and psychological functioning (Ryan & Deci, 2001). Mental well-being has been found to have major consequences for wider health and social outcomes (Huppert, 2005). Resilience refers to the positive pole of individual differences in people's response to challenges and adversity (Rutter, 1987). Researchers have argued that resilient individuals have the capacity to recover more quickly from stressful events (Ryff & Singer, 2003). Resilience is found to be positively correlated with positive affect and life satisfaction, and inversely related to negative affect (Hu, Zhang, & Wang, 2015;Liu, Wang, Zhou, & Li, 2014).
Mental well-being, resilience, and the reduction of substance use have clear relevance for public health, and have therefore, been a focus of research in both the social identity literature and the mindfulness literature. However, relatively little research in each area has been conducted in high-stress environments, and none to our knowledge has examined the combined effects of social identification and mindfulness.

| SO CIAL IDENTIT Y
Social identity refers to "the individual's knowledge that he [or she] belongs to certain social groups together with some emotional and value significance to him [or her] of the group membership" (Tajfel, 1972, p. 31). Social identities are an integral part of one's own identity (Tajfel & Turner, 1979). There is a substantial body of empirical evidence that demonstrates that group memberships (and the social identities derived from belonging to these groups) affect health and well-being Jetten et al., 2014Jetten et al., , 2017. When groups provide a person with direction, meaning, and purpose, then, this will typically have a positive effect for that individual's well-being (Haslam, Jetten, Postmes, & Haslam, 2009). Such findings have led to group memberships being described as potential "social cures" (Jetten et al., 2012).
From a social identity perspective, group membership serves as a psychological resource that can benefit well-being and resilience Scarf et al., 2016). Indeed, research has shown that simply reminding individuals of their group memberships can increase resilience to stress (Jones & Jetten, 2011). Furthermore, social identity seems to be particularly critical when facing stressors such as social stigma (Branscombe, Schmitt, & Harvey, 1999;Jetten, Haslam, Cruwys, & Branscombe, 2018). However, it is also the case that individuals need to experience a sense of meaningful connection to the group in order to reap the benefits of that membership: that is, they need to identify with it . Social identification has been defined as the extent to which a given group is valued and self-involving, such that individuals consider group membership to be an important part of their self-definition (e.g., Doosje & Ellemers, 1997;Doosje, Ellemers, & Spears, 1995;Turner, 1999).
When someone identifies strongly with a given group, he or she will be more likely to interpret reality and behave in a manner consistent with that group's values, ideology, and norms (Turner, 1991). And as such, identification determines the nature of people's social behavior (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987;Turner & Oakes, 1997) and their involvement in group efforts (Ellemers, de Gilder, & Haslam, 2004;Ellemers, Kortekaas, & Ouwerkerk, 1999).
Hence, social identities that provide positive meaning and worth are particularly beneficial (Haslam et al., in press). Thus, it is identification with a social group, rather than group membership per se, that benefits mental well-being (Cruwys et al., 2014;Miller, Wakefield, & Sani, 2015;Sani, Herrera, Wakefield, Boroch, & Gulyas, 2012).
In a similar way, research into substance use has shown that social identification with a nonaddiction identity is linked with a reduction in behaviors such as drug and alcohol use (Bathish et al., 2017;Best et al., 2016;Dingle, Stark, Cruwys, & Best, 2015). The literature examining the content and evaluation of particular identities offers useful insights (Dingle, Stark, et al., 2015;Haslam et al., 2019). For instance, Dingle and colleagues suggest that a shift in identity over time, toward a positive nonsubstance abuse identity, is a crucial step in recovery from addiction. In the same way that defining oneself as a "drinker" or "junkie" has negative implications for substance behavior (Schofield, Pattison, Hill, & Borland, 2001). Consequently, we argue that defining oneself as a "mindful champion" (i.e., someone who uses mindfulness to lead their life and inspire others to do the same) could have positive implications for well-being and would lead to a reduction of substance use. Indeed, a previous study showed that, in a prison context, the mindful leader identity provided group members with a positive sense of belongingness as well as a set of values and norms that promote healthy behaviors and a reduction in substance use (Thompson, 2018).

| MINDFULNE SS
Mindfulness is defined as an "awareness that emerges through paying attention on purpose, in the present moment, and nonjudgmentally, to the unfolding of experience moment by moment" (Kabat-Zinn, 2003, p. 145). The skill of mindfulness refers to the ability to willfully focus the attention on the phenomena occurring internally and externally at the present moment, while maintaining a nonjudgmental and curious posture toward those perceived experiences (Baer, 2003). Tang, Hoelzel, and Posner (2015) suggest that mindfulness mediation exerts its effects by enhancing self-regulation, including attention control, emotion regulation, and self-awareness, with measurable changes in relevant brain regions. More specifically, they suggest that mindfulness practice affects self-referential processing and present-moment awareness. As a result, mindfulness can improve mental well-being by reducing stress and regulating negative emotional states. Better stress and emotion regulation is a key aspect of resilience. Similarly, better self-regulation capacity may help reduce problematic substance use by increasing alternative emotional and behavioral coping mechanisms during stressful events.

In prison contexts, participation in a Mindfulness-Based Stress
Reduction (MBSR) course has been found to have positive well-being benefits for inmates (Samuelson, College, Carmody, Kabat-Zinn, & Bratt, 2007), and a recent meta-analysis showed that mindfulness and meditation can bring positive well-being benefits to prisoners (Auty, Cope, & Liebling, 2017). Mindfulness is also associated with improved resilience (Chamberlain et al., 2016;Nila et al., 2016). Bajaj and Pande (2016) concluded that mindfulness promotes emotional balance and may assist in faster recovery from challenges. They argue that higher levels of mindfulness mean people can foster resilience as they are more likely to respond skillfully to difficulties instead of reacting automatically in nonadaptive ways.

Mindfulness research mostly focuses on mindfulness interventions
on individuals. However, many mindfulness programs are delivered in group settings (e.g., MBSR, MBRP). We are interested in exploring what happens when mindfulness recipients see themselves in terms of a shared social identity-as members of the mindfulness group.

| R ATI ONALE OF S TUD IE S
This brief review has established that social identification and mindfulness are both positively related to well-being. However, research has yet to consider the contributory utility of these two constructs when examined together in the same context. This is important because mindfulness training is often delivered in group settings (e.g., Imel, Baldwin, Bonus, & MacCoon's, 2008), although the role of the group gets overlooked. To the extent that the group processes emerging in such settings underpin development of a shared social identity among training recipients (i.e., as members of a "mindfulness group": see Tarrant, Haslam, Carter, Calitri, & Haslam, 2020), it is possible that the positive health and well-being outcomes of mindfulness training may be-at least in part-be attributable to that social identity. The current research tested this suggestion.

| THE PRE S ENT S TUD IE S
Our studies investigated the contributory potential of both mindfulness and social identification in a context where stressors are multiple and chronic: the largest male maximum security prison in Kenya. The prison has around 3,000 inmates. The jail is overcrowded, inmates lack access to basic food (e.g., milk and bread) and hygiene products (e.g., soap), and there are high levels of violence and aggression (Adarves-Yorno & Mahdon, 2017; Adarves-Yorno, Mahdon, Lee, & Haslam, 2019). This is a highly stressful environment where group relations matter (e.g., Kellezi, Bowe, Wakefield, McNamara, & Bosworth, 2018b) but remain understudied. Wider evidence demonstrates that prison populations around the world suffer from substantial psychological disorders, together with high comorbidity of mental illness and substance misuse (Penal Reform International, 2018). A recent meta-analysis showed that mindfulness interventions in prison contexts have small to moderate effects on well-being (Auty et al., 2017). Improving well-being not only helps inmates cope with the challenges of prison life, but could also aid positive engagement with rehabilitation activities. In our research project, inmates took part in a "Mindful leaders" training. The program had two core elements: mindfulness training and the inductive creation of a "mindful leader" identity (see Mahdon et al., 2017, for details on the training methods and content). In this context, a "mindful leader/champion" is someone who leads his life mindfully and inspires others to do the same. Inmates kept their mindful leader identity salient by inspiring and training other inmates. Here, we examine the impact of the identification with the group of "mindful leader" rather than evaluate the training itself.
The concept of mindfulness was completely new to participants.
Training was initially provided by a mindfulness trainer over 2 days (14 hr in total). From that training, an 8 weeks training program (minimum of 24 hr) was cocreated with mindful leaders and delivered by mindful leaders themselves (inmates), and supervised by the initial trainer. Mindfulness became common practice among those trained and those who were residing with the trainees.
The studies in this paper aim to examine the combined and differential effects of dispositional mindfulness and identification with a mindful identity on mental well-being and resilience (Study 1 & 2) and reduction of substance use (Study 2). Both studies received ethical clearance from the [blinded for review] Ethics Committee.

| Hypothesis
We predict that social identification with a mindfulness group will be positively related to mental well-being (H1a) and resilience (H1b), and negatively related to substance use (H1c).
We predict that mindfulness will be positively related to mental well-being (H2a) and resilience (H2b), and negatively related to substance use (H2c).
We hypothesize that social identity related to mindfulness would add further explanation of the variance in mental well-being (H3a), resilience (H3b), and substance abuse (H3c) than mindfulness alone.

| Sample and procedure
To test our hypotheses, data were collected from the largest maximum security prison in Kenya. Our opportunity sample was composed of all participants from the Mindful Leaders program, which started a year before. This sample was composed of initial participants selected by the Prison Governor, and trained by the founder of the program, together with the first group of mindful leaders' trainees that inmates recruited themselves. Participants were invited to complete a paper-and-pencil questionnaire at the start of a mindfulness session they had with the founder of the program. The prison program coordinator noted that participants have completed at least two-thirds of their formal training.
All surveys were distributed in English as participants were fluent in this language. A total of N = 82 male participants completed the questionnaire-according to G*Power (Faul, Erdfelder, Buchner, & Lang, 2009), this gave us enough power to detect a medium-sized effect (f 2 = .10/ partial R 2 = .09, for 1−β = .80, α = .05). Participants were all aged between 21 and 71 with a mean age of 40 years. Their prison sentences ranged from 10 years to life or condemned to death and they had spent between 2 months and 27 years in prison (M = 7.59 years, SD = 6.22).
Their participation in the research was completely voluntary. No individually identifying data were collected. A member of our research team administered the questionnaire. Participants were made aware that their participation in the research would have no effect on their sentence or conditions in prison.

| Measures
The response scales for all the measures used was a Likert-scale of 1-7, where 7 represented a high score on that construct. The measures appeared in the questionnaire in the order presented below.

| Mindfulness
A 24-item short version of the Five Facet Mindfulness Questionnaire (Bohlmeijer, ten Klooster, Fledderus, Veehof, & Baer, 2011 adapted from the original version by Baer et al., 2006) was used. The items in the scale group into five facets which were: Acting with awareness, Describing, Non-reacting, Non-judging, Observing. Example items include "I pay attention to sounds" and "I notice smells," both these are from the "observing" facet. Items were presented in a mixed order in the questionnaire and not within facet groups. Reliability analysis revealed that items from the "non-judging" facet should be removed.
Probably that low reliability is because non-judging could be perceived as oxymoron to remorse and some inmates have committed crimes that require repentance. The final scale was composed of a total of 21 items. Internal consistency of the scale was acceptable with Cronbach's α = .70.

| Mental well-being
The Short Warwick Edinburgh Mental Well-being Scale consisting of seven items was used (Stewart-Brown, Tennant, Tennant, Platt, Parkinson, Weich, 2009). Example items included whether participants feel "optimistic about the future." Internal consistency of the scale was good, with Cronbach's α = .85. | 5 ADARVES-YORNO Et Al.

| Resilience
The short version of the Connor-Davidson Resilience scale, containing 10 items, was used (Campbell-Sills, Forde, & Stein, 2009;Connor & Davidson, 2003). Example items included whether they see themselves as able to "deal with whatever comes" (for copyright reasons we can only include paraphrased examples). Internal consistency of the scale was good, with Cronbach's α = .80.

| Social identification
A six-item scale to measure social identification with a group was used. Selected items from Ellemers et al. (1999) social identification scale were used. Items were adjusted to include the relevant group "Mindful Leadership (ML) group." Items were as follows: "Being a member of the Mindful Leadership group is important to me"; "I identify with other members of the ML group"; "I am like other members of the ML group"; "The ML group is a reflection of who I am"; "I would like to continue working with the ML group"; "I feel good about the ML group." Internal consistency of the scale is good, with Cronbach's α = .80.

| Results
To test the relationship between social identification and the two outcome variables, correlational analyses were conducted. In line with Hypotheses H1a and H1b, social identification was significantly positively correlated with both mental well-being (r(75) = .41, p < .001) and resilience (r (77) = .33, p = .003). See Table 1 for full correlation matrix.
Equally, to test the relationship between mindfulness and the two outcome variables, correlational analyses were conducted. In line with Hypotheses H2a and H2b, mindfulness was significantly positively correlated with both mental well-being (r(75) = .38, p = .001) and resilience (r(77) = .36, p = .001).
In order to test whether social identification predicted mental well-being (H3a) and resilience (H3b) in addition to variance explained by mindfulness alone, we conducted a hierarchical linear regression. The two predictors, mindfulness and social identification, were not significantly correlated (r(77) = .16, p = .153), indicating that multicollinearity would not pose an issue. The regression results supported Hypothesis 3a that social identification is a significant predictor (β = .36, p = .001) of mental well-being after accounting for the relationship with mindfulness (β = .33, p = .002). Both factors together accounted for 25% of the variance in mental well-being, with social identification accounting for 13% of variance in addition to mindfulness (ΔF(1, 74) = 12.76, p = .001). 1 See Table 2.

| Study 1 discussion
Results of Study 1 supported our hypotheses: both social identification with a mindful identity and individual mindfulness predicted mental well-being and resilience, respectively. More importantly, when analyzed together, social identification and mindfulness were able to explain more variance than mindfulness alone. Thus, mental well-being and resilience were improved by combining the positive 1 When the length of time in prison and length of sentence were controlled for, results did not significantly change (Study 1).

TA B L E 2
Regression analysis of mindfulness and social identification on mental well-being (Study 1)

b SE b β
Step effects of social identification and dispositional mindfulness. To our knowledge this was the first study to explore social identification with a mindful identity and the dispositional mindfulness together.
As such Study 2 aimed to replicate these findings. In addition, Study 2 aimed to address two potential shortcomings of this study. First, the mindfulness scale was relatively long and the reliability could have been higher. Second, the social identification measure should use a full scale instead of a subset from a longer scale.

| S TUDY 2
The design and rationale of this study was almost identical to that of Study 1. However, we made three improvements. First, it includes a related but different mindfulness scale as an attempt to improve its reliability. Second, instead of using a shorter version of a long social identification scale (Ellemers et al., 1999), it used the whole scale of one of the most widely used measures in social identification research (Doosje et al., 1995). In addition to well-being and resilience, we examined a reduction in self-reported alcohol and substance use.
The intervention our sample participated on had no components on drugs and alcohol. Due to the clandestine use of drugs and alcohol, self-report was the only option to measure substance use in this research.

| Sample and procedure
As in Study 1, we invited all inmates who were enrolled in the mind-

| Measures
The response scales for all the measures used was a Likert-scale of 1-7, where 7 represented a high score on that construct. Items in the questionnaire were presented in the order discussed below.

| Mindfulness
The items on mindfulness were reduced to 19 in line with a subsequently shorter published version of the five facet questionnaire

| Social identification
A five-item scale was used, with four items from an extensively used social identification scale (Doosje et al., 1995). Example item: "I feel strong ties to the Mindful Leaders group." In addition, one item from the Study 1 scale ("Being a mindful leader/trainee is an important part of who I am") was also included. All items were worded so that the identity referred to "mindful leader/trainee group." Reliability analysis revealed an acceptable internal consistency, with Cronbach's α = .79.

| Resilience and mental well-being
The same scales were used as in Study 1 for resilience--with a good internal consistency, with Cronbach's α = .82, M = 5.68, SD = 0.95and mental well-being-also with a good internal consistency, with Cronbach's α = .82.

| Reduction in alcohol and drug consumption
We also measured self-reported reduction in alcohol consumption and drug use since the beginning of the mindfulness program on a 1-7 Likert-scale from "not at all reduced" (1) to "very much reduced" (7). Note that the question also had a "not applicable" response category as not all inmates may have been involved in such activities before the program. The items were: "The amount of alcohol I was consuming has been reduced" and "The amount of drugs I was consuming has been reduced" compared to before the training. A total of N = 97 participants provided valid responses to both questions. The two items correlated positively and significantly, with r(95) = .76, p < .001, and were therefore, combined to form one scale of reduction in substance use, with a higher score indicating a larger reduction in use. Reliability analysis revealed a Cronbach's α = .86.

| Study 2 discussion
The results of Study 2 replicated those of Study 1 and supported the hypotheses that social identification and mindfulness are positively associated with mental well-being and resilience. In addition, social identification accounted for additional variance in mental well-being and resilience over and above that accounted for by mindfulness. Extending the work from Study 1, Study 2 revealed that in relation to substance use, both social identification with the mindful group and dispositional mindfulness individually explained significant amounts of variance in the reduction of substance use compared to before the training. However, when analyzed together, the effect of mindfulness was reduced 2 When the length of time in prison and length of sentence were controlled for, results did not significantly change (Study 2).

B SE b β
Step

TA B L E 4 Correlations amongst all variables (Study 2)
substantially, whereas social identification remained as a significant predictor.

| G ENER AL D ISCUSS I ON
The current research aimed to investigate the combined and differential effects of social identification and mindfulness on well-being.
In particular, the research examined the role of social identification with a mindfulness group in a context characterized by multiple and prevalent stressors: a maximum security prison in Kenya. Consistent with the predictions, we found that both identification with a mindfulness group and dispositional mindfulness were positively correlated with mental well-being and resilience. These findings were consistent across two studies. Further, our findings indicated that, when predicting mental well-being and resilience, social identification accounted for variance in addition to variance accounted for by mindfulness alone. In the case of substance use, results showed that both mindfulness and social identification were correlated with a reduction in consumption. However, when combined, identification with a mindfulness group provided all the explanatory power. That is, identification with a mindful identity was a stronger predictor of reduction of substance use. The theoretical and practical implications of these findings are discussed below.

| Theoretical implications
Our results have several theoretical implications. First, our research contributes to the emergent body of work on mindfulness in highstress contexts such as prisons. Previous research has identified that mindfulness can improve two key psychological well-being outcomes, that of mental well-being (Brown & Ryan, 2003;Carmody & Baer, 2008) and resilience (Nila et al., 2016;Pidgeon & Keye, 2014).
However, most of the research on mindfulness has been conducted in clinical populations (e.g., Kuyken et al., 2016), and research conducted in prisons has mostly focused on exploring the impact of meditation-based mindfulness interventions (for a review, see Auty et al., 2017). Our findings are in line with previous research, which has found positive links between dispositional mindfulness, mental well-being, and resilience in low-stress environments (e.g., Gu et al., 2015). Second, we contribute to the social identity approach to health by providing evidence that mindfulness groups can provide a positive social identity in a real prison population. Previous research has been conducted in a simulated prison environment (Haslam & Reicher, 2006) and with detainees who were not incarcerated (Kellezi, Bowe, Wakefield, McNamara, & Bosworth, 2018a).
Third, our research suggests that combining individual mindfulness with identification with a mindful group provides significant benefits for well-being above the benefits of mindfulness alone.
The positive impact of social identification on well-being measures has been explained in the "social cure" literature by group belongingness (Jetten et al., 2012). Social groups are a psychological resource, which individuals can draw on, and which have been shown to enhance psychological well-being and resilience (Haslam, Jetten, Cruwys, Dingle, & Haslam, 2018). Yet, the psychological resource is not group membership per se. Jetten and colleagues (2014) argue that a social group can only help mitigate the impact of negative stressors when these groups are important and meaningful for us and form part of our sense of self. Furthermore, Haslam and colleagues (2018) propose that it is the meaning of the identity that holds the power. In our studies, the content of what it means to be a mindful leader/trainee was closely linked with mindfulness itself. Thus, for the inmates, the creation of a new identity that centered on being a mindful leader/trainee was potentially the key to its effect on well-being (see Thompson, 2018). Importantly, when behavior becomes infused with what an identity means, engaging in that behavior allows people to demonstrate their identity and affirm their membership of the group (Smith, Louis, & Tarrant, 2017).
For our participants, it is possible that reducing substance use was aligned with the identity of a mindful leader/trainee, and therefore, an expression of a valued identity. In line with this, Klein, Spears, and Reicher (2007) emphasize the importance of identity expression for an identity to be maintained and lived. Similarly, Gallagher, Muldoon, and Pettigrew (2015) discuss affiliative social identification where embodying norms and behaviors allows members to demonstrate that they meet group expectations. In contrast, consuming drugs and alcohol is misaligned with what it means to be a mindful leader in that prison (Thompson, 2018). Thus, the mere consumption of substances among inmates could threaten the expression of a mindful identity.

| Practical implications
Our research suggests that social identity established within the context of group-based mindfulness programs may be important for well-being. Mindfulness training has increased in popularity in recent years, including in prison settings. And although many training programs and health interventions are delivered in group settings (see Tarrant et al., 2020), they are approached as individual training TA B L E 7 Regression analysis of mindfulness and social identification on reduced substance use (Study 2)

B SE b β
Step sessions conducted in groups. This approach misses the potentially valuable opportunity of harnessing the benefits of social group membership-and associated social identity-that can be aligned with mindfulness programs. Indeed our findings provide support to Imel et al. (2008) idea that while mindfulness practice is an individual endeavor, the training of it in a group has the potential to offer something beyond that of simple delivery of intervention content (i.e., specific mindfulness techniques). In fact, research shows that in some cases it might be the group setting rather than the intervention itself that yields benefits for well-being (Gleibs, Haslam, Haslam, & Jones, 2011).
The importance of considering social identity processes in interventions is also reflected in our findings where a mindful disposition failed to account for self-reported reduction in substance use when participants' social identity was accounted for. This suggests that the well-being benefits of mindfulness training could potentially be increased by attending to social identity processes when this training is delivered in group settings. The importance of this potential is especially marked in prison contexts when there are few opportunities for development of other positive and meaningful (and health-promoting) social identities.
In the current research, participants reported identifying with a mindful leader group (in the sense of being "champions" of mindfulness). This has a proactive element that may make this identity particularly beneficial. Both "leader" and "champion" portray individuals in a positively distinct way. For instance, the leader/champion aspect of the identity may have conferred a sense of importance and empowerment to the participants, thereby increasing self-esteem and confidence beyond the effects of a mindfulness group membership.
It is important to acknowledge that proactive identities whereby individuals become leaders/champions may lead to more benefits than the association with a group per se. In this sense, a "mindful prisoner group" may not be as beneficial as a "mindful leader group." Future research is needed to better understand the effects that such identity labels have on psychological outcomes.

| Limitations and future directions
Conducting this research program in a prison environment in Kenya brought with it many practical and cultural challenges that affected the way we were able to conduct our research. A longitudinal study was not possible, and given the cross-sectional nature of our studies we cannot claim causality. Furthermore, the voluntary nature of the program and the study meant that participants may have selfselected toward the study. This means that it is currently unclear whether a wider intervention in prison contexts would be similarly successful. However, mindfulness programs-and particularly the continued practice of mindfulness-rely on an element of voluntariness and self-motivation. We also did not assess the length of time that each participant had taken part in the mindfulness program, and could therefore, not control for length of training effects. In addition, the data were derived from self-reported measures only, which may underestimate real substance use (e.g., Johnson & Fendrich, 2005).
Another limitation refers to the mindfulness measure, the "non-judging" subscale had very low internal reliability and had to be removed from the scale. As discussed previously, this low reliability may be due to the type of sample used (condemned prisoners). Future studies with different samples should use the full mindfulness scale.
Our research was conducted in a highly stressful context. In order to determine whether our findings and practical implications extend to other more "everyday" situations, future research could explore whether our findings hold true in less stressful contexts, for example, the delivery of mindfulness trainings within organizations or in traditional MBSR courses.
The correlational design of the current research means that there is some way to go to understanding the mechanisms by which mindfulness interventions and group identities can be used to enhance well-being in high-stress environments, particularly in relation to recovery from substance abuse (Dingle, Cruwys, & Frings, 2015).
Future research needs to look at the mechanism by which a mindfulness identity impacts well-being outcomes. can result from structured group interventions (e.g., self-help or therapy groups), which are thought to underpin the maintenance of substance use cessation. In addition, SIMR points out the role that the wider social network, including informal groups and friendships, can play in recovery from substance abuse. Importantly, both models emphasize the key role of norms that discourage substance use and aide recovery and well-being, and see the creation of new, "positive" identities as the best way to create and internalize such normative behavior. Along similar lines, the development of a social identity as a mindful "champion" may have signaled for our participants a normative sense of responsibility and visibility, as distinct from identifying only as a prisoner (which likely conveys different, possibly maladaptive social norms). Future research is needed to better understand what mindfulness identity, in each context, can serve as a particularly fruitful pathway to enhancing well-being.
The third pathway reflects concerns raised in the literature about identities that may act as a "social curse" rather than a "social cure" (e.g., Kellezi & Reicher, 2012). Stigmatized identities and those of low-status groups have been found to undermine self-esteem and mental health (Crabtree, Haslam, Postmes, & Haslam, 2010;Kellezi et al., 2018b). The prison system is likely to disrupt positive social group memberships that prisoners may have had before their incarceration, and enforces a highly stigmatized identity as "prisoner." Kellezi and colleagues (2018b) studied participants in a detention center and found that social identities in that context added to feelings of distress and isolation. Importantly, a shared identity with fellow prisoners does not automatically lead to negative well-being and indeed can mitigate against such negative impacts. For instance, Haslam and Reicher (2006) conducted an experiment in a simulated prison environment and found that prisoners developed a sense of social identity and solidarity that was related to a reduction in stress and better coping. Future research is needed to examine the different processes by which a mindful identity increases well-being and reduces substance use.

| CON CLUS ION
In this paper, we have provided an examination of the combined and differential effects of mindfulness and identification with a mindfulness group on well-being. Both mindfulness and social identification were found to be positively related to, as well as significant predictors of, well-being outcomes: mental well-being and resilience. The paper also looked at reduction of substance use. Results showed that identification with a mindfulness group did not only account for additional variance in the reduction of substance use, but also reduced the variance that mindfulness explained. This paper shows how both individual mindfulness and identification with a mindfulness group can be usefully combined to create better well-being, resilience, and lower substance use in a high-stress environment. Our findings contribute to the understanding of how well-being, resilience, and lower substance use in a high-stress environment may be enhanced through mindfulness programs.

ACK N OWLED G M ENTS
The authors would like to acknowledge Emma Jones for her valuable contribution in conducting this research. We would also like to ac-