Abstract
Community choirs can contribute to health and wellbeing, but there is less clarity about the mechanisms through which choir participation promotes positive change, or whether mechanisms are shared with other leisure activities. This research explores two possible types of mechanism: Mechanisms relating to individual characteristics pertaining to mindset and motivation, including experiences of flow, competence, autonomy, and relatedness (Study 1); and mechanisms relating to the activity experience, which include music, movement, and social opportunities (Study 2). In Study 1, middle and older aged members of choirs, exercise groups and other kinds of social groups (N = 190) completed surveys on their experience of emotional wellbeing (operationalized as positive and negative affect scores), mental wellbeing, and social cohesion (outcomes) pertaining to their group activity as well as experiences of motivation, flow, and the components of Self-Determination Theory (potential mediators). Multiple regression analyses revealed that participation in Choir or Exercise groups predicted positive emotional wellbeing, but not social cohesion. Underlying mechanisms differed, with positive affect mediated by intrinsic motivation for choir members, and by intrinsic motivation, identified regulation, and flow for Exercise group members. Mental wellbeing was correlated only with exercise group participation and was mediated by flow. Study 2 used an experience sampling methodology conducted with a sub-group from Study 1 (N = 59), which asked daily questions about wellbeing (happiness, sense of social connection, and energy levels) and participation in activities (music engagement, exercise, and social activities) experienced in their everyday lives and not directly associated with any leisure group participation. Repeated-measures t-tests revealed that participants were more likely to report higher levels of social connection on days in which they participated in music activities than on days in which they did not engage in music activities. Engaging in exercise or social activities was also associated with a greater sense of social connection, as well as higher levels of happiness and energy. In sum, the activity characteristics and individual differences of motivation and mindset towards participation correlated with greater wellbeing, reflecting an ecological model of person-activity fit, with no indication of superior benefits associated with group singing. Findings are discussed in terms of social prescribing and other settings where social opportunities are organized.
Keywords
Introduction
Evolutionary theorists suggest that the longevity and universality of group singing may be attributable to a unique capacity for creating shared positive emotional states and increasing social cohesion (Harvey, 2018; Dunbar et al., 2012; Schulkin & Raglan, 2014; Weinstein et al., 2015). There has, therefore, been growing interest in community choir membership as an effective way to improve social and emotional wellbeing (Clift et al., 2010; Dingle et al., 2019; Maury & Rickard, 2016; Williams et al., 2018). Many studies have reported superior benefits for participation in choirs than in control conditions, including studies that compare singing to a non-singing condition (Bullack et al., 2018; Kreutz, 2014; Sanal & Gorsev, 2014) or to another activity (Johnson et al., 2017; Kirschner & Tomasello, 2010). However, other studies that use similarly engaging leisure group activities as controls report comparable benefits for wellbeing for both singing and non-singing conditions, including cooperative Lego-building (Allpress et al., 2012), creative writing classes (Dingle et al., 2017), exercise (Maury & Rickard, 2018, 2020; Valentine & Evans, 2001), and solo or group music and sporting activities (Good & Russo, 2021; Lonsdale & Day, 2020). Such diverse findings require further elucidation.
Understanding how group singing contributes to wellbeing would contribute to addressing this knowledge gap. It could be that the underlying mechanisms that promote wellbeing are attributable to activity characteristics inherent in the choir experience – whether unique or shared with other engaging activities. Alternatively, wellbeing improvements may be best explained through interindividual differences of mindset or motivation towards participation, which may play a more critical role than the activity characteristics. Research to date, however, tends to have focused more on establishing whether differences exist between choirs and other kinds of social groups, with less attention paid to the mechanisms which might underlie improved wellbeing (and whether they are shared by other kinds of group activities). Given the diversity of physical and mental activities that have provided a comparison to group singing, it is possible that leisure group activities benefit wellbeing through characteristics that are not unique to choir participation, such as social interaction and engagement. While comparing choir participation with control activities that are equally engaging is an important first step, ensuring comparisons share some activity components, such as music exposure, movement, and social opportunities, will further elucidate the mechanisms that contribute to wellbeing changes. Several key potential contributors to the benefits of choir participation (social interaction, emotion regulation, individual disposition, and mindset) are reviewed next.
Social Interaction, Health, and Wellbeing in Older Adults
Interpersonal relationships are a significant determinant of physical and mental health (Ong et al., 2016). A large population study of older adults found that being socially active was a primary predictor of healthy older age with few chronic diseases, alongside not smoking (Barak et al., 2020). A meta-analysis revealed that people with stronger social relationships increased their likelihood of survival by 50%, and that these effects were greatest for people with strong social networks (Holt-Lunstad et al., 2010). The loss of social opportunities and roles resulting from retirement can be a key contributor to reductions in health and wellbeing for older adults (Heybroek et al., 2015). For example, a longitudinal study of retirees found that those who maintained activity in at least two social groups had a 2% risk of death in the first six years of retirement compared with a 12% risk for those who ceased social group activity, while also recording a 10% drop in quality of life scores for each social group that they lost (Steffens et al., 2016). These changes suggest that organized leisure groups are increasingly important for healthy ageing in older adults, which may require proactive management even while other capabilities (e.g., mobility or cognitive capabilities) may be in decline (Steverink et al., 2005). Older adults also often have more time to invest in leisure activities and socializing (Cornwell et al., 2008), which creates an opportunity for maintaining and improving health and wellbeing.
Organized leisure group activities have been identified as an important facilitator of diverse social networks for older adults (Chang et al., 2014), where “[p]erceptions of positive social relationships were associated with greater involvement in leisure activities, and greater involvement in leisure activities was associated with better health in older age” (p. 516). These findings have been corroborated by Fiori et al. (2006) who report that diverse networks – which include social group attendance – are most predictive of mental wellbeing, while family-intensive networks were least predictive, implying that the breadth of social contact may provide important contributions to improved wellbeing. Beyond the social opportunities that all leisure groups provide, particular characteristics of certain leisure groups, and common in community choirs, may also contribute to improved wellbeing by specifically enhancing social connections. For example, creating a shared emotional state, a common experience in joint musical experiences (Egermann & McAdams, 2013; Peltola, 2017), has been shown to increase a sense of social bonding (Páez et al., 2015). Indeed, music engagement has been linked to an increased sense of social cohesion (Cross, 2007; Loersch & Arbuckle, 2013; Schäfer & Eerola, 2018) and empathy (Fukui & Toyoshima, 2014; Greenberg et al., 2015). Increased social cohesion is therefore likely to be a key component of wellbeing benefits from choir membership.
Several empirical studies corroborate the claim that choir participation increases social cohesion. Dunbar et al. (2012) demonstrated that active music co-creation, including group singing, increased pain tolerance to a similar level to that experienced with a dance condition, while passive music listening did not. Pain tolerance is often used as a proxy measure of oxytocin levels, since elevated levels of oxytocin are implicated in both increased pain tolerance and sense of social bonding (Johnson & Dunbar, 2016). Weinstein et al. (2015) reported that singing in both small community choirs and a larger combined choir increased pain tolerance and self-reported ratings of social bonding. Coordinated movement is another aspect of group singing (Himberg & Thompson, 2009; Müller & Lindenberger, 2011; Phillips-Silver & Keller, 2012) known to increase a sense of social cohesion (Codrons et al., 2014; Jackson et al., 2018; Müller & Lindenberger, 2011; Valdesolo & DeSteno, 2011).
Many exercise groups share characteristics with choirs, including music exposure, coordinated movement, and social opportunities. Exercise groups have also been found to improve subjective wellbeing and quality of life in older populations (Rennemark et al., 2009), with participation in exercise correlated with reduced depressive symptoms (Lindwall et al., 2007), and overall psychological wellbeing and mood (Hassmen et al., 2000; McIntyre et al., 2020). Similar to choirs, exercise groups have also been found to increase a sense of social cohesion (Dunlop & Beauchamp, 2011; Tunçgenç & Cohen, 2016) and expression of pro-social behaviours (Di Bartolomeo & Papa, 2019). There are also indications that group exercise increases oxytocin levels in a similar manner to group singing (Rassovsky et al., 2019), which may therefore be expected to have a similar effect on pro-social behaviours (MacDonald & MacDonald, 2010). Furthermore, while choir participation incorporates what is typically low-level physical activity through the act of singing, exercise classes provide heightened experiences of movement which is often synchronized with other group members. There are also indications that the intensity of exercise may influence these effects (Davis et al., 2015). While researchers have at times expressed a reluctance to compare group singing with exercise classes due to such similarities, using a control group that shares some of the potential mechanisms for improving wellbeing could provide greater insights than when comparing to more sedate activities such as craft groups or creative writing. Reducing the contrast between the group singing and comparison conditions would enable a less confounded examination of the proposed underlying mechanisms; that is, enabling an assessment of whether they yield similar positive benefits outside of the group singing experience, rather than merely seeking to highlight stark differences with leisure activities that are dissimilar.
Music-based Emotion Regulation
The use of music for emotion regulation (Juslin & Sloboda, 2010; Moore, 2013; Schäfer et al., 2013; Thoma et al., 2012) could also account for improvements derived from choir participation. Music has a powerful impact on mood (Menon & Levitin, 2005; Rickard, 2012; Schäfer et al., 2013), which appears to be heightened when shared with others (Weinberg & Joseph, 2017). A growing body of evidence indicates that choir participation improves emotional wellbeing (Clift & Hancox, 2010; Daykin et al., 2018; Sandgren, 2009), including for older adults (Coulton et al., 2015; Lamont et al., 2017; Lee et al., 2016). Most exercise classes also incorporate music, which can provide music-related wellbeing effects, provided individuals enjoy and are engaged in the music (Hallam, 2012; Van Den Bosch et al., 2013). A review on the effects of music listening during exercise found that music enhances affect and psychological wellbeing amongst other benefits (Karageorghis & Priest, 2012).
Individual Disposition and Mindset
While there is great variation in how individual groups are structured, both choirs and exercise groups can offer opportunities for social interaction and cohesion, music-related emotion regulation and coordinated movement, each of which may explain benefits for wellbeing. However, several comparison studies have shown that choir participation does not necessarily yield wellbeing benefits superior to leisure activities that do not comprise these key elements. For example, when choirs have been compared to cooperative Lego-building (Allpress et al., 2012), creative writing classes (Dingle et al., 2017), and listening to a choir (Unwin et al., 2002), no discernible differences between groups on wellbeing measures were found. Similarly, a meta-analysis identified similar reductions in perceived stress for both exercise group participants and those who practiced mindfulness (Díaz-Silveira et al., 2020).
The wellbeing benefits arising from these more sedate group activities provide insight into an additional important factor that could underlie choir participation benefits also; that of individual mindset towards participation. We use the term “mindset” here to encompass a range of individual differences in disposition towards participating in a social group, including motivation, experiences of flow, and autonomy. One way to explain these differences is through Self-Determination Theory (SDT), which argues that perceived autonomy, feeling competent, and a sense of relatedness (that is, that an individual's contribution is meaningful and valued by others) is fundamental to wellbeing across the life course (Ryan, 2009). Employing a Self-Determination Theory frame highlights the critical role that preference and choice may play in wellbeing improvements for group activity participants. Wellbeing is predicted to be higher when an individual is intrinsically motivated, or at least choosing to engage in an activity (e.g., in cases of identified regulation, in which the outcomes of participation are the motivator), than when extrinsically motivated or not motivated at all (e.g., amotivation) (Deci & Ryan, 2008; Guay et al., 2000). A sense of being “forced” into participation, or being motivated extrinsically to participate, may undermine a sense of autonomy, competence, and relatedness, thereby countering any benefits for wellbeing. There are indications that autonomy may be an under-explored mechanism for wellbeing improvements related to music; for example, two recent studies identify perceived agency over music selection as a contributor to reductions in pain (Howlin & Rooney, 2021; Fritz et al., 2018).
Intrinsic motivation is also associated with engagement in an activity, and, at its best, can lead to a sense of “flow”, characterized by enjoyment of the activity, intense concentration, a sense of mastery, and a distortion of time passing, which can be particularly positive (Nakamura & Csikszentmihalyi, 2004). Flow is often reported by individuals engaged in music production (Chirico et al., 2015; De Manzano et al., 2010), and this experience has been linked to wellbeing improvements (Fritz & Avsec, 2007). It is also commonly experienced in exercise settings (Jackson et al., 2001; Swann, 2016), and may be enhanced in group settings as opposed to individual endeavors (Decloe et al., 2009; Walker, 2010). It could therefore be that individual motivation for participating in an activity influences wellbeing outcomes, with those who are intrinsically motivated gaining greater benefits. Hence, an individual's disposition towards the activity may be a critical factor for consideration, rather than just the choice of a particular leisure activity (such as “choir” or “exercise” group). Comparison of choir members with solo singers, solo instrumentalists, solo exercise, ensemble instrumentalists and members of sports teams found that self-reported wellbeing measures did not differ across any of the activities, but that using a Self-Determination Theory framework highlighted that choir members had lower ratings of autonomy than those participating in other activities (Lonsdale & Day, 2020). The authors suggest that individuals may be willing to forego autonomy when choosing to participate in a social group activity that appeals to them.
Potential mechanisms for facilitating wellbeing therefore may be related to the activity components (such as social interactions, music engagement, and movement), or may be related to individual disposition or mindset towards participation (as described by Self-Determination Theory, types of motivation, and/or experiences of flow). A third option is that wellbeing improves when there is a good fit between the activity's characteristics and individual attitudes or motivation. The concept of person-activity fit (Lyubomirsky & Layous, 2013) accounts for both the components of the activity as well as the motivation, engagement, and enjoyment experienced by the individual. In this model, wellbeing is not mediated solely by the activity nor by individual attitudes, but rather is dependent on the congruence, or fit, of the two. Person-activity fit was developed to consider specific wellbeing interventions such as gratitude journals, savoring life's joys and acts of kindness (Thompson et al., 2015), but may also be a useful construct to understand how wellbeing might be influenced by an interaction between social group activities and individual motivation.
The Current Study
The aim in the current study was to explore two kinds of mechanisms that might explain improved wellbeing as a result of choir membership: mechanisms of individual mindset (Study 1) and mechanisms of the group activity (Study 2). Choir participation was compared to other non-singing social groups to help clarify whether there is anything unique about this particular leisure activity with regard to its impact on wellbeing.
Study 1 examined motivational and mindset factors (specifically self-reported sense of flow, autonomy, competence, relatedness, and type of motivation for participation) as potential mediators of wellbeing during participation in one of three types of social group. All groups observed were self-selected social leisure groups, with choir participation (Choir group) compared with a similarly social, movement-based leisure group that also incorporates music listening (Exercise group), as well as more sedentary, non-musical social leisure groups (Other group). Study 2 examined the potential effects of the mechanisms of music, movement/exercise and socializing. Rather than focusing on the group type, the key features of interest (music, movement/exercise, social interaction) were examined in participants’ daily activities independent of any social group affiliation across a two-week period to determine the importance of each element, or combination of elements, to participant wellbeing. It was hypothesized that participation in social leisure group activities (of any type on any given day) would predict increased subjective wellbeing ratings, reflecting changes in emotional state, energy levels, and sense of social connection. It was further hypothesized that motivation factors (subjective ratings of intrinsic motivation, identified regulation, experience of flow, autonomy, competence and relatedness) would more strongly predict self-reported wellbeing than would the type of group activity.
This research incorporates two components. Study 1 involved completing an online survey concerning experiences in one of three kinds of social groups: A singing group, an exercise group, or another kind of leisure group. An optional Study 2 required responses to two online questions every evening for 14 days, focused on types of daily activities engaged in without regard for the group affiliation identified for Study 1.
Study 1
Method
Study 1 examined wellbeing differences (emotional state as operationalized by positive and negative affect, mental wellbeing, and social cohesion) by group affiliation (Choir, Exercise, or Other) as mediated by differences in mindset and motivational factors (specifically flow, the components of Self-Determination Theory, and motivation).
Participants
Social media advertisements (paid) and posts (unpaid) were used to target individuals who belong to and enjoy a range of organized social groups, with a particular focus on choirs and exercise groups. In addition, the researchers directly contacted individuals who were known to participate in social groups, with a request to share the study with others. The study was described as exploring how participation in a social group improves individual wellbeing. Participants were asked to select the one social group that they belonged to which helped to improve their overall sense of wellbeing the most, and keep it in mind when answering the survey. Targeted advertising focused on an older demographic (45 years + ) in the English-speaking countries of Australia, New Zealand, the United Kingdom, and the United States. Individuals who were likely to be a member of a choir (Choir), an exercise group (Exercise), or another kind of social group (Other) were targeted through the social media accounts of various kinds of social groups (specifically choirs, exercise classes, and other kinds of groups such as book clubs or knitting groups) or if they had expressed an interest in these activities. Social media posts encouraged people to click on the URL link to read more about the study, which included a plain language description of the research. All cases included in the analysis provided informed consent to participate.
Of a sample of 424 individuals who navigated to the survey page, 273 agreed to participate (see Figure 1 for details of participant attrition). The final number of included cases was 202, which is above the minimum advised by Tabachnick and Fidell (2013) for multiple regression analyses, in which N > 50 + 8 m (m = number of variables), totaling 154.
Flowchart of attrition in study 1 sample. Figure created using Lucid.
The mean age for the Choir group was 64.9 (SD = 13.5), for the Exercise group it was 56.2 (SD = 10.0), and for the Other activity group it was 59.0 (SD = 18). A one-way ANOVA showed that there were no significant differences in age between the Exercise and Other groups, but that Choir members were significantly older than both the Exercise and Other groups, F (2, 197) = 5.98, p = .003. The sample was therefore a mix of middle and older aged adults, which will be referred to as “older” for the sake of brevity. The Choir had a total of 64 participants (male = 14), with an approximate ratio of female: male of 4:1. The Exercise group also had 64 participants (male = 12), with a female: male ratio of 4:1. The Other activities group had a total of 74 participants (male = 16, non-binary = 2), with female: male: non-binary ratios of 3:1:.05. Chi-square tests indicated the gender ratio was not significantly associated with group. Educational attainment and employment status are displayed in Table 1. A chi-square test indicated the group was not associated with educational attainment level, but was associated with employment level, with Choir members more likely to be retired than other group members, X2 [df = 10] = 27.6, p = .002.
Demographic frequencies for Choir, Exercise, and Other groups.
Because individuals were specifically recruited based on enjoyment of their leisure group activity, all leisure group types were included provided they met the descriptive criteria of “organized”, which was described as “a community group of people who meet regularly (e.g., weekly) to participate in a particular activity.” There was, therefore, also a range of choir types (e.g., community choirs and church choirs) and exercise group types (e.g., aerobics or running clubs) within these broader categories. Activities categorized as “Other” (that is, neither Exercise nor Choir) were quite diverse and included church-associated groups, voluntary associations, craft groups, gaming, Pokémon, professional associations, book clubs, acting or dance clubs, meditation, and collector's clubs. While the diversity of the “Other” group types is quite varied, the decision was made to keep recruitment open to all leisure group types, for the pragmatic reason of meeting recruitment target numbers. However, for Study 1, participants reporting activities involving strenuous exertion were excluded from the “Other” category to avoid overlap with the Exercise group and to increase the homogeneity of this category to include more sedate activities. Eleven cases were removed from analysis, including dance and membership in sports teams. One case was excluded due to identifying both a choir and an exercise group affiliation. Finally, one “Other” case which identified as a member of a Pilates group was recoded as an Exercise group. This reduced the number in the Other group to 62, which brought it closer in size to the Choir and Exercise groups (n = 64 for each). Arts-based or creative activities, which may have some overlap with the choir experience, were maintained within the “Other” grouping. While the assumption is that exercise classes are likely to incorporate music, while other forms of exercise groups (e.g., running clubs) and activities included in the Exercise category were less likely to do so, the survey did not specifically ask whether activities included a music component in any form. In sum, for analysis the Choir group included any kind of choir, the Exercise group included exercise classes of various types and running clubs, and the Other group included a diverse range of more sedentary activities such as discussion groups, crafting, book clubs, and similar.
Materials
Three scales were used to measure the wellbeing constructs of emotional state (operationalized as subjective positive and negative affect scores), mental wellbeing, and social cohesion in Study 1. Three scales were also used to measure personal motivation and mindset towards group participation, incorporating experiences of flow, the components of Self-Determination Theory (competence, autonomy, and relatedness), and type of motivation (intrinsic, external, and identified regulation). Amotivation (not being motivated at all) was dropped from analysis since it would neither be expected to correlate to wellbeing, nor be present in voluntary social group participation. Further information on all scales is provided in Table 2.
Measures of wellbeing and motivational states.
Procedure
A plain language statement explained that the research was interested in how membership in organized social groups may improve wellbeing. Participants were informed that participation was voluntary and that they could withdraw at any time. Only those who provided consent to participate were directed to the surveys. Surveys were administered via the Qualtrics© software platform and were completed at a time and place of the participants’ choosing and took an estimated 10–15 min to complete.
An organized social group was defined as “a community group of people who meet regularly (e.g., weekly) to participate in a particular activity. This could be an exercise group, a singing group, a book club, a crafting group, a sports club, a discussion group or another activity-focused group.” Because some participants may belong to more than one organized social group, they were asked at the start of the survey to select the one group that they think improves their overall sense of wellbeing most, and answer the survey with their experiences of that particular group in mind. They were then asked to identify what kind of group this was: a singing group, an exercise group, or another kind of group (specified). They were also asked how often their group generally met: weekly, fortnightly, monthly, or other (specified).
Data Analysis
All analyses were conducted using SPSS version 26. First, preliminary analyses were conducted using one-way ANOVAs to identify any differences by group in self-rating for wellbeing outcomes (positive affect, negative affect, social cohesion, and mental wellbeing) and also for mediators (flow, intrinsic motivation, identified regulation, external motivation, competence, autonomy, and relatedness). Second, a series of mediated multiple linear regression analyses were run using the PROCESS version 3.4 add-in (Hayes, 2019), which explored whether: a) any of the wellbeing outcomes – positive affect, negative affect, social cohesion, and mental wellbeing – were mediated by any of the proposed mechanisms – motivation, flow and the components of Self-Determination Theory; and b) whether the mechanisms differed by group. Due to the nominal nature of the Group Participation variable, three sets of analyses were conducted; the first compared Choir to Exercise, the second compared Choir to Other activity, and the third compared Exercise to Other activity. Comparing both Choir and Exercise groups against the Other group allowed for a contrast of groups that incorporate music listening and movement to groups that are, taken as a whole, more sedentary, and less likely to incorporate music and movement as part of the core activity. See Figure 2 for the testing model.
Model for testing mediators of wellbeing outcomes by social group participation type. a = relationship between group participation and mediators, b = relationship between mediators and outcomes, c' = direct effect of group participation on outcomes, and c = total effect (c' + a x b) (Meule, 2019). Figure created using Lucid.
The analyses tested for the total effect of Group (Choir, Exercise, Other) on the outcome variables (wellbeing markers of positive affect, negative affect, mental wellbeing and social cohesion scores), the direct effect of Group on outcome variables (that is, unmediated) and the indirect effect of Group on outcomes variables (that is, mediated by one of the proposed mechanisms of flow, motivation and the components of Self-Determination Theory), employing bootstrap confidence intervals. Indirect effect tests also determined whether the mediation was explained by an unidentified mechanism that was not included in the study design. This process aligns with contemporary approaches to testing for mediation effects (Hayes, 2009; Meule, 2019; Zhao et al., 2010).
Results
All analyses were conducted using SPSS version 26. First, preliminary analyses were conducted using one-way ANOVAs to identify any differences in self-rating for wellbeing outcomes (positive affect, negative affect, social cohesion, and mental wellbeing) and also for mediators (flow, intrinsic motivation, identified regulation, external motivation, competence, autonomy, and relatedness). Table 3 shows differences in wellbeing measures by group: Choir, Exercise, and Other.
One-way ANOVA, reported means by group on wellbeing outcomes: emotional state (positive and negative affect), social cohesion, and mental wellbeing.
*significant at ≤.05 **significant at ≤.005.
Because the assumption of homogeneity of variances was not met for this measure, the F statistic, degrees of freedom and significance are reported using Welch.
Post-hoc tests (Tukey's HSD) reveal that there were no significant differences between the Choir Group and the Exercise Group on any measure, and no differences between any of the groups in ratings of mental wellbeing. Members of the Other Group had significantly lower PA ratings when compared to both the Choir (p = .002) and Exercise (p = .001) groups, higher ratings of NA compared to both Choir (p = .004) and Exercise (p = .002), and lower ratings of social cohesion than the Choir group (p = .026). These analyses confirm that, as predicted, the wellbeing outcomes for the Choir and Exercise groups were not significantly different, while the Other group (of more sedentary activities) demonstrated significantly different self-ratings that would allow for greater contrast for the multiple linear regression analyses.
Table 4 reports differences in self-reported experiences of individual mindset measures, including flow, motivation, and the components of Self-Determination Theory, by group affiliation.
One-way ANOVA, reported means by group on wellbeing mediators: flow, motivation, and Self-Determination Theory.
* Significant at ≤.05 **Significant at ≤.005.
Because the assumption of homogeneity of variances was not met for this measure, the F statistic, degrees of freedom and significance are reported using Welch.
Post-hoc tests (Tukey's HSD) revealed that significant differences in motivation was driven by differences in intrinsic motivation ratings between Other group members, M = 5.69 (SD = 1.35) and both Choir group members, M = 6.32 (SD = .777), p = .001, and Exercise group members, M = 6.16 (SD = .683), p = .025. Differences in identified regulation were between Exercise group members, M = 6.41 (SD = .586) and Other group members, M = 5.89 (SD = 1.19), p = .002. Differences in external regulation were between Choir group members M = 1.45 (SD = .813) and both Exercise group members, M = 2.13 (SD = 1.11), p = .002 and Other group members, M = 2.01 (SD = 1.17), p = .008. For the sub-scales of Self-Determination Theory, differences in competence were between Choir group members M = 4.23 (SD = .651), and Other group members, M = 3.86 (SD = .773), p = .005; and autonomy by differences in Exercise group members, M = 4.34 (SD = .518), and Other group members, M = 4.02 (SD = .869), p = .026. There were no significant differences between groups on ratings of flow, competence, or relatedness. These differences provide an indication that the same wellbeing outcomes may be achieved through different pathways, dependent on group affiliation.
Second, a series of mediated linear regression analyses were conducted. Proposed mediators of mindset included in the model were flow, competence, autonomy, relatedness, intrinsic motivation, identified regulation, and extrinsic motivation. This analysis tested whether any of these mediators may be predictors of wellbeing outcomes, specifically emotional state (positive and negative affect), mental wellbeing, and social cohesion. Three sets of mediated linear regression analyses were run to compare differences between all groups: Choir and Exercise, Choir and Other, and Exercise and Other.
The data set was checked to ensure the assumptions of multicollinearity were not violated. The independent variables of intrinsic motivation and identified regulation were highly correlated, r (197) = .71, p <.001 as were autonomy and competence, r (197) = .77, p <.001. This is below the suggested threshold for multicollinearity of r = .9 (Pallant, 2016), and the Tolerance and Variance Inflation Factor (VIF) were also found to be normally distributed. There were no instances of singularity.
All variables demonstrated normality and linearity with the exception of negative affect, which exhibited non-normal distribution in scatterplots and probability. Attempts to normalize the distribution via logarithmic transformations were unsuccessful (Tabachnick & Fidell, 2013), so the raw data were maintained. While caution should be taken in interpreting results for negative affect, it should be noted that in non-clinical populations low levels of negative affect are normal (Crawford & Henry, 2004; Diener & Diener, 1996). One case from the Other activity group was found to be an extreme outlier on several measures and exceeded the Mahalanobis distance critical value of 27.9 and was therefore excluded from the analysis. Preliminary checks indicated there was no correlation between any outcome variable and gender, but age was found to correlate with negative affect and mental wellbeing. Age was therefore entered as a covariate for these analyses only.
The mediated regression analysis comparing Choir to Exercise group members showed no direct effect of group on any outcome. A weak indirect effect of relatedness to social cohesion was found, b = .21, [CI = .02, .42]; see Table 5.
Mediated linear regression analysis full results: Choir compared to Exercise.
†Mediator contributes to total effect.
When the Choir was compared to the Other group, membership in a Choir predicted scores for emotional state only, with no effects for either mental wellbeing or social cohesion. There was a significant total effect of choir participation and positive affect scores, b = 3.2, p = .007, CI [0.88, 5.52], but no direct effect. Choir participation predicted scores for positive affect through a weakly significant indirect effect of intrinsic motivation, b = .92, CI [0.03, 2.35]. Choir membership also predicted negative affect scores, with a significant total effect, b = −1.13, p = .041, CI [−2.13, −0.05]. No significant direct effect of choir membership on negative affect was observed. The mediated multiple regression analysis revealed no significant indirect effects of choir membership with any of the tested mediators to explain the association with negative affect scores. Since the direct effect of choir membership on negative affect was not significant, the significant total effect may be explained by an unidentified mediator which was not measured in this study. There was no mediation through flow, competence, autonomy, relatedness, or identified regulation. The Other group did not reveal either a direct or indirect (mediated) relationship with the outcome variables. See Table 6.
Mediated linear regression analysis full results: Choir compared to Other.
*significant at ≤.05 **significant at ≤.005 †Mediator contributes to total effect.
The analyses comparing Exercise group to Other group membership also revealed significant relationships for emotional state outcomes. The total effect for positive affect was significant, b = 3.55, p = .003, CI [1.23, 5.86]. However, there was no direct effect between Exercise group membership and positive affect. Exercise group membership predicted positive affect scores through significant indirect effects of flow, b = .45, CI [.00, 2.37], intrinsic motivation, b = .69, CI [.02, 1.79], and identified regulation, b = 0.99, CI [.00, 2.37], although these effects were weak.
The total effect for Exercise group membership was also significant for negative affect, b = −1.46, p = .007, CI [−2.51, −0.41]. There was a significant but weak direct effect of group membership on negative affect, b = −1.19, p = .046, CI [−2.66, −0.02]. The mediated regression analysis revealed no significant indirect effects of exercise group membership with any of the tested mediators to explain the association with negative affect scores. Some of the lower negative affect ratings associated with Exercise group membership may therefore be explained by a mediator not contained in our model.
Exercise group membership also predicted a correlation with mental wellbeing scores with a significant total effect, b = 1.58, p = .034, CI [.12, 3.05]. There was no direct effect, but flow mediated this relationship, b = .54, CI [.07, 1.30]. See Table 7.
Mediated linear regression analysis full results: Exercise compared to Other.
*significant at ≤.05 **significant at ≤.005 †Mediator contributes to total effect.
No total or direct effects were observed for social cohesion. However, when comparing Choir to Other groups, intrinsic motivation was found to negatively predict social cohesion scores, b = –.09, CI [−.21, −.01] while autonomy was found to positively predict social cohesion scores, b = .11, CI [.00, –.27]. When comparing the Exercise group to Other groups, identified regulation was predictive of social cohesion, b = 0.08, CI [.01, .18]. These effects did not, however, contribute significantly to overall social cohesion scores for either group. The Other group revealed neither a direct nor an indirect (mediated) relationship with the outcome variables when compared to the Choir and Exercise groups.
Figure 3 summarizes the significant outcomes for both Choir (A) and Exercise group (B) membership.
Significant pathways of mediated wellbeing changes for choir participation (A) and exercise group participation (B). (a) = relationship between group and mediator, (b) = relationship between mediator and outcome variable, (c) = direct effect of group on outcome variable, and (c') = total effect. Coefficients are reported for all significant pathways, with asterisks indicating the strength of significance: *significant at <.05**significant at <.01. Age was included as a covariate for negative affect analyses only, and was a significant contributor only for Exercise group associations. Figure created using Lucid.
Study 2
Study 2 examined daily fluctuations in wellbeing as a result of participation in the core activities of interest – music, exercise, and socializing. Group affiliation was therefore not a consideration for this study. The Experience Sampling Methodology (ESM) enabled wellbeing to be assessed as a function of the activity characteristics for each day as they varied across the 14-day period, considering music (or no music), exercise (or no exercise) and socializing (or no socializing).
Method
Participants
At the completion of the Study 1 surveys, participants were invited to participate in a second study. One hundred people indicated interest and provided an email address. Eligibility for the study required at least three daily responses to be completed, resulting in a sample of 59 participants; nine males, 49 females, one non-binary, mean age 58.95 (SD = 9.94).
Materials
Fourteen individual daily surveys were created for this study. The survey for Day 1 started with two demographic questions of sex and age. The two survey questions that followed were identical for all 14 days. Question 1 asked about mood using three indices, responding using a drop-down Likert scale: “What's your mood now?” (1 = low, 7 = high): Happy, Socially Connected, Active. The second question asked about activity during the day: “What have you been doing today? Check all that apply (or none).” Options were:
I’ve exercised by myself I’ve been singing by myself I’ve listened to or made music by myself (e.g., listening to the radio or practicing an instrument) I’ve attended an exercise class I’ve been singing in a choir I’ve listened to or made music with other people (e.g., music was playing in an exercise class or I played an instrument in a band) I’ve participated in another kind of group activity (please specify): I’ve not done any of these things today.
In this way, the components of music engagement, exercise, and socializing were considered within the context of everyday life rather than as discrete experiences associated solely with leisure group activities. For example, individuals who attended an exercise class with music playing in the background would indicate a) the exercise; b) the music; and c) the social nature of the experience. Conversely, an individual who had exercised alone and without music only noted the exercise.
Procedure
Participants were asked to provide their email address and country of residence. Email addresses were entered into an automated email system on the Qualtrics© platform, grouped by region: Australia/New Zealand (40 participants), United Kingdom (32), and North America (29). Each email address was assigned a number code to anonymize responses. Once a day for two weeks an automated email was sent out to all participants, timed to coincide with early evening for their region (between 4–8 pm). Each email contained a link for that day's survey, available via Qualtrics©. Participants were encouraged to respond to the survey in the evening when all activities were completed, were requested to not respond retrospectively, and told that if they were busy they could skip the day's survey. Participants were also told in each email that they could opt out at any time, and a link was provided to be removed from the email list. The survey was designed so that it could be completed on a smart phone, to make it easy to quickly respond in the evening.
Data Analysis
Individual daily responses were downloaded and matched using the pre-assigned codes. Due to the optional nature of each daily survey, there was variability in the number of surveys completed across individuals. There were almost twice as many high responders (those who responded for 10 or more days of the 2-week survey) than medium- to low-responders (responses for 9 days or fewer). Since state wellbeing varies by individual, it was important to control for number of responses; otherwise, individuals who were high responders may bias the findings. Therefore, individual ratings were averaged based on the activity or combination of activities that each participant had participated in each day.
Questions of activity covered musical activities, exercise, or another kind of activity. In addition, these activities could be performed alone or with an organized group. On any given day, an individual may participate in multiple activities. Therefore, there was the potential for multiple combinations of activities for any given day. To maximize information generated by these analyses, the categories were combined into three primary activities, reflecting the three activities of greatest interest to this study:
Music activities, including any music activity a participant made note of, e.g., music listening, playing an instrument, or singing alone or with others. Exercise activities. Participation in any organised social groups.
Summarizing the data in this way allowed for isolating the effect on the wellbeing markers of the three primary activity conditions: music vs. no music, exercise vs. no exercise, and group activity vs. no group activity.
A series of repeated-measures t-tests (with two-tailed alpha set at .05) were run to compare wellbeing ratings for “any” versus “none” activity involving music, exercise, and group, using SPSS version 26.
Results
Results are presented in Figure 4, displaying the Mean (Standard Deviation) for ratings of Happy, Connected, and Active, and the t-test results.
Daily ratings on happy, connected, active with the presence or absence of music (n = 44), exercise (n = 42), and group (n = 44) activities. Graphs created using Excel *p ≤ .05; ***p ≤ .0001.
Days reporting any music engagement yielded significantly higher ratings of “Connected” than days reporting no music engagement, t (47) = 2.03, p = .048. No differences were observed in “Happy” or “Active” ratings for days reporting or not reporting music. Days reporting exercise yielded significantly higher ratings for “Happy”, t (41) = 4.86, p = .001, “Connected”, t (41) = 3.95, p = .001, and “Active”, t (41) = 8.40, p = .001 than did days not reporting exercise. Days reporting group interaction also yielded significantly higher ratings for “Happy”, t (43) = 2.75, p = .009, “Connected”, t (43) = 3.94, p = .001, and “Active”, t (43) = 4.59, p = .001, than did days not reporting group interaction.
General Discussion
This research provides insight into how participation in choirs and other social groups might be associated with improving or maintaining wellbeing, which has implications for how individuals are advised concerning participation in particular types of leisure group activities. We explored two categories of mechanisms that might explain wellbeing changes as a result of choir participation: those related to individual attitudes towards participation, which included flow, motivation, and the components of Self-Determination Theory (Study 1); and those related to the group activity characteristics itself, which included the presence of music, physical movement, and socializing (Study 2). Furthermore, choirs were compared to exercise groups, incorporating some of the same activity characteristics, and other kinds of more sedentary groups, which were less likely to do so, to determine whether wellbeing benefits were achieved by a range of social group activities through similar mechanisms.
The findings from Study 1 revealed that Exercise group participation had wider-reaching positive impacts on wellbeing outcomes than Choir group participation. Both Choir and Exercise group participation predicted increases in positive affect and reductions in negative affect, while Exercise group participation also predicted mental wellbeing. While the positive impact of Choir and Exercise group participation was similar, Exercise group participation achieved these outcomes through a more diverse range of mediators. Intrinsic motivation was a mediator for positive affect for the Choir group members, while for the Exercise group members, positive affect was mediated by flow, intrinsic motivation, and identified regulation, and mental wellbeing benefits were mediated by flow. Intrinsic motivation was the only mediator that explained wellbeing associations for the Choir group, with no mediation identified through flow, competence, or relatedness. The lack of direct effect by group on any outcome when Choir and Exercise group members were compared directly to one another may reflect the similarity of activity-based mechanisms (social opportunities, music listening, and movement) as well as the effects of exercise in general on wellbeing and mental health (De Moor et al., 2006; McAuley et al., 2000; Windle et al., 2010).
Group membership was not predictive of social cohesion in Study 1. However, participation in any social group activity was found to be associated with higher levels of social connection in Study 2. Study 2 compared wellbeing ratings on days with or without music, with or without exercise, and with or without social interaction, considered within the context of everyday experiences rather than within the context of a social group. Participating in activities which included music engagement yielded significantly higher levels of social connection than did activities not containing music engagement. Participating in activities that contained exercise or any kind of social interaction, however, yielded significantly higher levels of social connection, happiness, and activation when compared to no exercise or social activity in a day. It is worth noting, however, that Study 2 incorporated music engagement in the broadest sense, rather than being confined to choir experiences. Conversely, no association between music engagement and increased happiness was identified in the ESM data of Study 2, despite mood regulation being a well-established primary function of music engagement (Moore, 2013; Saarikallio, 2011; Thoma et al., 2012). It may be the variety of settings in which music engagement was measured in Study 2 yielded more diverse impacts on mood, including increases in both positive and negative affect. In addition, because we recruited individuals who participated in social group activities, it is possible that the sample was skewed towards people who particularly enjoy social interactions. For this group, it may be that music serves as a proxy for social interactions, and as a result boosted their connectedness scores above the exercise group's scores. This hypothesis is supported by research conducted by Schäfer and Eerola (2018), which indicated that individuals use music listening, watching television and reading fiction as a kind of “social surrogacy” that serves as a proxy when there are not opportunities to socialize. Furthermore, when an individual feels sadness and loss, listening to affect-congruent music can provide an empathic proxy; listening to sad music resulted in reduced sense of loneliness, a rise in empathy, and an improvement in mood (Schäfer et al., 2020). There is, therefore, an emerging understanding of how people may use music to feel socially connected, particularly in times of isolation. For the people in the current study, who enjoyed their social group participation, music use may have become more salient on days when they were not able to attend their group.
Taken together, the findings from these studies indicate that socio-emotional wellbeing may be influenced by characteristics of both the group activity and individual attitudes towards participation – specifically self-directed motivation, whether intrinsic motivation or identified regulation. Furthermore, wellbeing benefits were not predicted solely by group singing or music engagement; rather, exercise and group interactions appeared to predict broader contributions to improving wellbeing than music engagement. This finding indicates that the primary mechanisms that underlie wellbeing improvements may be shared by a range of social group or exercise activities, rather than being specific to choirs. Wellbeing benefits of choir participation may therefore be mediated as much, or indeed more so, by individual attitudes towards participation than by the components of the activity itself. Specifically, our findings point to intrinsic motivation as a primary contributor to wellbeing for Choir members. Taken together, this research appears to support an ecological model of person-activity fit as proposed by Lyubomirsky and Layous (2013), in which both the motivation and attitude towards participation as well as the characteristics of the activity itself are considered. Future research could directly test this hypothesis.
The role of social group participation is central to building strong, diverse social networks, which is an important determinant of health. For older adults, who may be losing social opportunities through such changes as retirement or reduced mobility, indications are that organized leisure activities are a core component of healthy social networks (Chang et al., 2014; Fiori et al., 2006; Steffens et al., 2016). This research provides an important contribution to understanding the role of activity characteristics, specifically music engagement, movement, and social opportunities, in supporting the wellbeing of individuals. However, these studies highlight the importance of also considering motivation for an individual to participate in a social group. The research participants who belonged to choirs tended to be intrinsically motivated, finding enjoyment in the activity itself. In contrast, exercise group members were self-motivated to participation through intrinsic motivation and also through identified regulation, by appreciating the outcome achieved from the activity (e.g., better health), which may be linked to goal striving and sense of achievement. Exercise group members also experienced flow while Choir members did not. While there were differences in motivation, self-motivation for participation appears to be a key mediator in wellbeing outcomes achieved through social group participation.
Identifying the features of choir and other types of social group participation which are responsible for wellbeing benefits may help to optimize the efficacy of social-based interventions, such as social prescribing initiatives (Bickerdike et al., 2017). Social prescribing provides a non-medical response to address social determinants of health (Drinkwater et al., 2019), and can include “prescriptions” for activities such as arts engagement, outdoor walks, or participating in a social group (Chatterjee et al., 2018). Similarly this information could be useful for those who provide leisure activities in settings such as retirement communities, schools, or community hubs, or for individuals managing their own health and wellbeing maintenance. This is particularly true for older adults, who sometimes struggle to look after their wellbeing following such events as retirement or relocation.
Overlaid with motivation, this research found that providing social opportunities elevated ratings of emotional state, social connection, and energy. Organized leisure groups could therefore be specifically designed to maximize social interactions, rather than focusing exclusively on the core activity. This validates the importance of including social rituals in group activities, such as asking participants to exchange greetings with three other people, providing a break time at the mid-point, or providing snacks at some or all meetings. There are also indications from previous research that social ties themselves can keep people committed to leisure group activities in the absence of intrinsic motivation (Beauchamp et al., 2007; Beauchamp et al., 2018; Maury & Rickard, 2020). In settings where the core activity is expected to provide other important benefits, for example, the beneficial impact of exercise on physical health, strengthening social ties may be a strategic avenue for maintaining commitment to the activity in the absence of intrinsic motivation.
Emotional state, sense of connection, and energy levels were also raised on days in which participants reported exercising in this study. Therefore, directing people to activities that provide more rather than less movement and exertion would be recommended, bearing in mind that such exertion needs to be commensurate with individual capabilities. Choirs could also capitalize on this effect through deliberately encouraging increased movement through such additions as hand clapping or swaying during singing sessions. Music engagement provided an elevation in sense of social connection, so, where practical, social opportunities could include music as well. Where music is not a core component of the activity, however, it should be introduced with caution, since benefits of music are directly associated with preference. Playing music that is unpleasant to some group members may have the opposite intended effect (Jezova et al., 2013; Salimpoor et al., 2009).
Limitations
The online recruitment strategy meant that there was quite a lot of heterogeneity within each of the groups studied, making interpretation challenging. Clarifying the elements of each group would be helpful in controlling potential confounds (e.g., whether the group activity nominated in Study 1 contained music or not); more targeted recruitment and/or more detailed information about activity characteristics is recommended in future studies attempting to replicate the current findings.
This research deliberately recruited individuals who enjoy their social group participation, due to the focus on understanding wellbeing effects. This bias in recruitment means that findings may not be generalizable to individuals with less interest in social opportunities or leisure group participation. It is important to consider findings and implications within this limitation; for social prescribing or enforced participation (e.g., in a school or institutional setting) such considerations are paramount.
Further Research
There are several avenues for extending the findings from this study to gain a better understanding of how participation in various social groups may influence wellbeing. The ESM component of the research was limited by low rates of choir participation during the data collection timeframe, when many northern hemisphere participants were on break from choir practice. Replicating this study with more days of choir participation would allow for a more robust direct-comparison data with other kinds of social groups. This data set was also collected prior to the onset of the Coronavirus pandemic; since that time, many leisure groups have been cancelled or moved to an online format. The mechanisms identified in this research may operate differently in an online context, which warrants further exploration.
Previous research on leisure groups emphasize that quantity – the number of leisure activities – is a critical protective factor for health and wellbeing (Chang et al., 2014; Steffens et al., 2016). While, as predicted, this research found members of choirs and exercise groups reported higher levels of wellbeing associated with their group participation, further research could focus on how more sedentary social groups contribute to individual wellbeing, and whether the key to improvements is in quantity. Because this research did not consider the quantity of social groups for each participant, this omission could have been a confounding factor in Study 1, which was focused on group affiliation, despite the instructions to respond with one chosen group in mind. For individuals who participate in diverse and multiple groups, wellbeing effects may be heightened with no clear pathway attributable to one individual group.
Findings point to heightened sense of wellbeing on days when individuals exercised when compared to days with music engagement. It therefore appears that movement may be an important contributor to wellbeing. What this research did not consider is what aspect of movement is the primary contributor – physical exertion or synchronicity, both of which have been shown to improve wellbeing measures. Further research could examine how these two components of movement independently contribute to subjective wellbeing, to better understand the contributions of each.
This research tested only a limited number of possible mechanisms to understand changes in wellbeing, but future research could employ different frameworks to further elucidate the findings that are presented here. For example, another possible mechanism that was untested in the current study is social identity theory, in which an individual feels a strong sense of affiliation with the group and/or individual members (Haslam et al., 2012; Jetten et al., 2014). Social identity may explain the wellbeing correlates for Exercise group members despite the lack of intrinsic motivation, as the commitment to the group may provide a proxy motivation for continued group participation. Such a displacement of commitment from the activity to the group may have important implications for commitment to group activities such as exercise, where there is a value to participation beyond wellbeing effects. Social identity may also be implicated in wellbeing attached to choir participation, as per the findings of Pearce and colleagues (2015; 2017) that choirs bonded more quickly than other, sedentary, leisure groups; the current study did not track changes in groups over time, which may reveal unique pathways to wellbeing such as the “icebreaker effect,” and explore whether these are similar effects for both choir and exercise group membership.
Finally, the results of this research may be explained by the person-activity fit model, in which there is a match between individual mindset and motivation factors and group characteristics. An important step in confirming this conclusion is to directly test the person-activity fit model as it pertains to wellbeing outcomes for social group participation across a range of group types.
Conclusion
This study contributes to the current knowledge base in two primary ways. First, it elucidates how participation in several different types of leisure group activities may predict wellbeing of middle and older aged adults, and how benefits from group singing are likely to be shared by a range of other leisure activities for which participants experience a good person-activity-fit. Second, the findings assist programs such as social prescribing or activities provided in retirement communities or similar settings to effectively direct people to social groups which will be of maximum benefit to them, considering the optimal fit between activity characteristics and personal motivation.
The findings presented here provide an understanding of how social group participation in general, and choir membership in particular, impacts wellbeing. For practitioners, this information is valuable as it demonstrates that the activity characteristics of movement, music, and social opportunities are more likely to have a positive impact on emotional wellbeing. However, it is equally critical to understand individual motivation for participation; encouraging or forcing participation where an individual either does not enjoy or see the personal value of participation is unlikely to result in wellbeing improvements. Person-activity fit, an ecological model that considers both individual motivation and the activity characteristics, therefore best accounts for our findings. This has important implications for both practitioners and social prescribing endeavors when the goal is enhanced wellbeing, pointing to the need for meaningful consultation and access to varied opportunities.
Footnotes
Action Editor
Kelly Jakubowski, Durham University, Department of Music.
Peer Review
Claire Howlin, University of Cambridge, Autism Research Centre.
One anonymous reviewer.
Contributorship
SM and NR scoped the study and design. Material preparation, data collection, and analysis were performed by SM. The first draft of the article was written by SM, and all authors contributed to its revision and interpretation of the results. All authors have read and approved the final manuscript.
Availability of Data and Material
All data and material is available on request from the corresponding author: SusanEMaury@gmail.com
Code Availability
Consent to Participate
All participants were provided with a plain language explanatory statement and provided consent to participate.
Consent for Publication
All participants were informed that findings would be published, and agreed to having their de-identified data included.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethics Approval
All aspects of this research were approved by the Monash University Human Research Ethics Committee (project number 19544). All data are available from the corresponding author.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Prior Publication
A preliminary version of this article was published as a chapter in Susan Maury's thesis/dissertation.
Social Media
The lead author is active on Twitter @SusanMaury and on Mastodon @SusanMaury@fediscience.org.