Why are some groups physically active and others not? A contrast group analysis in leisure settings

In the last ten years, public health research has paid increased attention to determinants of physical activity in order to fight the pandemic development of non–communicable diseases [1, 2]. Most studies in this regard have analysed socioeconomic correlates of physical activity, particularly the socioeconomic status [3‐8]. The studies in unison report that a high socioeconomic position (including occupational position, income and educational level) positively correlates with a higher degree of physical activity in leisure time.

Further studies also analysed the effects of social networks on the willingness of an individual to be physically active. Ball et al. [9] note that physical activity is associated with individual social participation and neighbourhood interpersonal trust. Mc Neill, Kreuter, and Subramanian [10] identify five modifiable dimensions of the social environment that have an influence on the activity status of a person: firstly, social support and social networks; secondly, socioeconomic position and income inequality; thirdly, racial discrimination; fourthly, social cohesion and social capital; and fifthly, neighbourhood factors. Studies by Yu et al. [11] and Carlson et al. [12] confirm the impact of social networks and social support on the degree that people are physically active. However, in the latter study, this effect was particularly observable in participants with positive psychosocial attributes.

De la Haye and colleagues [13] were able to show that close friends in adolescent friendship networks have a significant influence on each other regarding a number of health related behaviours, especially the physical activity behaviour was “found to be similar” [13]. Carell, Hoekstra, and West [14] report effects of social contagion analogical to findings with regard to general health related behaviour [15]. They come to the conclusion that especially ‘poor’ fitness spreads on a person-to-person basis through a social network [14]. They conclude that their findings are “consistent with the notion that people imitate the diet or exercise habits of their least fit friends, or use those friends’ fitness as a benchmark for their own” [14]. Regarding beneficial outcomes on the willingness to be longer physically active, Scarapicchia and colleagues [16] note that it is particularly intrinsic motivation which seems to spread.

The existing data about behavioural and social factors of physical activity are mostly based on questionnaires or demographic statistics. Furthermore, most studies investigate psychosocial variables, such as attitudes, motivation, action planning, self-efficacy, or stress [3]. Observational studies are scarce. Therefore, there is still a lack of research and findings representing physical activity in ‘real life’, particularly in unsupervised contexts [17]. Finally, there is also a lack of research on which factors characterise people which are particularly prone to be physically active or inactive.

In order to close this research gap, we conducted a study on social determinants of physical activity in groups. According to Tajfel and Turner [18], we define a social group “as a collection of individuals who perceive themselves to be members of the same social category, share some emotional involvement in this common definition of themselves, and achieve some degree of social consensus about the evaluation of their group and of their membership in it” [18]. Groups are considered as a specific form of social environment that has a relevant impact on the behaviour of group members due to group’s specific behavioural rules, motives and norms [19]. Group membership influences behaviour because “the way we perceive others will influence directly how we act towards them” [19]. The underlying mechanism of this influence is imitation, as “the demeanour, the presentation, the look, the size, and the physicality of bodies is automatically and deeply perceived and read as a text at an automatic and deep level of perception” [20].

Our study was conceptualized as a multi-method participant observation study. In contrast to existing studies that are based on large data sets or self-reports, we decided to systematically observe physical activity-related behaviours “in action” [21] and thereby to consider the context in which physical activity occurs in order to fully understand activity promoting and hindering mechanisms. The basic idea of the study was to observe group behaviour in specific leisure settings, where the opportunity to be physically active is generally given. Following a methodological observation paradigm, we focused exclusively on face-to-face interacting social groups and their visible actions. Since cultural values have an influence on how people deal with their own body [22‐24], we also wanted to analyse their impact on the occurrence of very active and inactive groups. Cultural representations form the individual perception of how individual behaviour and appearance in a certain social context should be [25, 26]. Stereotypical and normative body ideals as well as somatic practices therefore symbolise the socially ‘expected’ body appearance and activity. For this reason, we decided to conduct the research in two different countries, where we expect culturally given differences regarding social expectations towards body appearance: a more multi-cultural environment: Hawai’i, and a more mono-cultural environment: Germany. For example, in both Hawai’i and Germany, obese bodies are considered not to conform with the societal body ideal [27, 28]. However, while in Germany obese bodies are generally stigmatised as unhealthy and considered to be caused by a lack of discipline [28], overweight bodies in Hawai’i are also related to Hawai‘i’s cultural traditions “that embrace food as an offering of gratitude, graciousness” [29].

We have already published an article about the qualitative data of this study which focused on group dynamics of physical activity and inactivity [30]. The results showed that physical activity is hindered and promoted by several social mechanisms, such as age-related social group norms and behaviours, so called (in) activity anchors, social contagion effects, and the way people interact in groups.

In this article, we analyse the quantitative observational data of this study. In this regard, our main focus lies on the identification of contrast groups, i.e. groups which are particularly physically active or inactive. Furthermore, we want to find out which consistent correlates of physical activity [3], such as age, gender or body appearance (BMI: Body Mass Index), characterise these contrast groups. But we also included further factors like group size or communication behaviour in order to analyse how these may influence the physical activity level of groups.

The present paper is therefore guided by the following research question:

According to the expected cultural differences, our study should also give answers to a second question:

In total we observed and recorded 907 groups with the groups’ size varying between 2 and 8 members (group size mean = 3.51, SD (standard deviation) = 1.348). 86.1% of the groups were age-homogeneous, 13.9% age mixed. 71% of the groups were categorized as adult groups, 20.9% adult groups with children and/or adolescents, 3.7% children only groups, and 4.2% adolescent only groups. 22.9% were groups consisting of female members only, 12.2% as male only groups, and 64.8% as gender mixed groups.

The main goal of our mixed-method observational study was to investigate social determinants of physical activity of groups in two leisure settings in Hawai’i and Germany [30]. This article presents the analysis of the quantitative observational data of this study with the main focus being laid on the identification of contrast groups, i.e. groups that are especially active or inactive. In this regard, the analysis was guided by two questions: which factors characterise extremely active or inactive groups? And, in which regard do the two different socio-cultural settings have an impact on the occurrence of such active and inactive groups?

Before discussing the results, we would like to discuss limitations of our findings. We addressed some of these already in the qualitative article of our mixed method observational study [30]. Firstly, the observational study is only able to give an account of visitors’ activities and behaviours during their stay at the pools. In order to stay covert, we did not speak to any of the visitors. Therefore, we were not able to collect data about the physical activity behaviour of participants outside the research setting and the observational period. By the same token, we do not have information about participants’ purpose or motivation for attending the research site. Consequently, we cannot rule out that some of our results could be explained by the participants’ intents for going to the pools [30]. The fact that smaller sized groups were more active than larger groups, for example, may be a result of motivational differences. Hence, while some larger groups perhaps come to the pool with the intention of relaxing and socialising joyfully together, some groups of pairs possibly visit the pools with the explicit motivation to swim lanes. An additional limitation of the study results from the fact that we executed the observation covertly. In order to stay unnoticed by participants, we sometimes had to record and complete description sheets for observed groups outside the observational areas. Observers had to memorize their observations for a short time in these cases. This was especially the case for the pond in Hawai’i, where several observations were executed from the water. Although we trained and tested memorizing the observations, observational errors may have occurred in these cases [30]. Finally, we should point out that we had different amounts of observational days at the research sites and more data for the Hawaiian site due to different climate and weather. This may have affected the estimates. However, since we used all possible days to collect data in Germany, the collected data should be representative for the entire season of that year.

With regards to the main research questions, our findings partially confirm previous studies. The cultural setting itself did overall not appear to have a very significant impact on the occurrence of physically very active or very inactive groups. This lack of major differences may appear surprising. Yet, it indicates that physical activity in leisure settings is significantly less influenced by cultural values than, for example, by the way in which individuals perceive and present their body [30]. However, one minor difference should be highlighted. The tree analysis showed that occasionally communicating groups that brought no additional equipment were more active in Hawai’i than in Germany. This difference may stem from a cultural difference which data collectors noticed in Hawai’i. In Hawai’i, people who spend their leisure time with others at the beach generally tend to ‘domesticate’ their site heavily with additional equipment [30]. However, in Germany, relaxing with others in leisure settings is not necessarily coupled with ‘furnishing’ the environment in a cosy way. This could also be the reason that the difference in the ways of spending ‘social’ sedentary time and ‘functional’ exercise was more distinctly observable in Hawai’i than in Germany [30]. To our knowledge, there are no studies which explain such cultural differences in spending leisure time in groups and its potential impact on physical activity. However, this may also be a result of research settings’ characteristics - we discuss this in more detail below regarding the results about additional equipment.

Although the cultural setting did not have a direct influence on the occurrence of activity, our study is mostly consistent with previous research showing that the social environment in general and the embeddedness in a social network in particular have a very relevant impact on a person’s physical activity level [10‐13, 15, 38]. Our results furthermore strengthen the hypothesis that specific group structures play a significant role regarding the activity of the group members [18]. For example, sitting together in groups and chatting, appears to be hardly compatible with physical activity for adults. In this regard, the ‘furnishing’ of the group’s site with additional equipment [30] seems to additionally reduce the willingness to be active. In contrast to adult groups, the high activity level of children-only groups shows that for them, playful, boisterous physical activity with lively verbal and non-verbal communication may be a relevant strategy of collective bonding [30].

Overall, the single most determining factor for activity/inactivity is the body appearance of group’s members. While obese groups had the lowest average activity level, athletic groups were the most active. The fact that obesity turned out to be one of the strongest activity limiting factors confirms previous epidemiological studies [3]. The observation that mixed-BMI groups were significantly less active than normal weight or athletic groups could be regarded as a social contagion effect, i.e. the spread of a low degree of physical fitness. This is consistent with findings by Carrell et al. [14] that it is especially ‘poor fitness’ and unhealthy behaviours that appear to spread in social networks. Christakis and Fowler suggested that such a spread of behaviours may be explained by ‘mirroring’ [39, 40]. A recently published work by Datar and Nicosia explains in this regard that “mirroring is more commonly observed in friends and family networks, peer networks, and cultural groups, where individuals may mirror the behaviour of significant others or those they esteem” [40]. This effect supports Turner’s theory on group behaviour: when individuals become member of a social group, it impacts their behaviour because they perceive other group member’s behaviour and this influences how they act themselves [19]. In this context, we should highlight that not all overweight visitors were inactive in general. In fact, our observational notes indicate that we did observe some obese visitors who were very active. Yet, they were often on their own and not part of groups, and, thus, not part of the analysis.

Also, the level of communication within a group was a very strongly activity influencing factor. Frequently communicating groups were clearly less active than non-communicating groups who had the second highest average level of activity. In our study on the qualitative observations [30], we provided a possible explanation of this finding in detail. We assume that the activity-hindering effect of lively communication can be attributed to the phenomenon that holding an interacting adult group together requires ‘physical discipline’, which means that members of lively communicating groups cannot come and go whenever they want to.

A closely related important factor that characterises inactive groups was a large group size. Our general statistics strongly suggests that with increasing size, the activity decreased. However, regarding the contrasting clusters, the tree analysis showed that group size had a differentiating effect only in mainly athletic and normal weight groups with a high level of verbal communication. Referring to the point raised above and our qualitative study this could be explained by the phenomenon that in large, vividly communicating groups, the ‘social norm’ is to be physically inactive in order to socialise and to not disturb the group’s conversation [30]. According to Turner, such group behavioural rules and norms play an important part in group members’ behaviour [19]. Since athletic and normal weight groups generally tended to be more physically active, this effect may have been stronger than in groups of obese people. The result that a large group size presents a barrier to activity also supports findings by Kilpatrick that physical activity and exercise is often viewed and executed as an individual activity [30, 41].

Furthermore, the gender-related composition of a group had a strong effect on physical activity. While mixed groups tend to be rather inactive, male only groups had the third highest activity level on average.

The finding that groups coming to the beach only with a towel are more active than groups bringing barbecue grills, sun-lounges, chairs etc. can be regarded as an expression of leisure motives. Bringing along additional equipment speaks for the motive of ‘domesticating’ one’s leisure setting, while the absence of additional equipment could be interpreted by a functional orientation of the visitors [30]. The observation that the absence of additional equipment is stronger correlated with the degree of a group’s physical activity in Kapoho/Hawai’i than in Entringen/Germany could be explained by the recreational nature of the observed leisure site. Without chairs, the black lava beach in Kapoho was not as convenient for a long stay in an inactive position as the lawn in Entringen. We therefore assume that in Kapoho, people who came to the beach without trucks, barbecues, chairs and sun shades may have had the primary intention to be physically active during their stay. This may have been the reason why they did not need the additional equipment.

However, our classification tree analysis shows that the level of physical activity in groups is to a relevant degree the result of a combination of different factors. This confirms the results of a friendship network analysis of obese persons’ health behaviours by De la Haye and colleagues [13] who also came to the conclusion that several factors influence health behaviour outcomes. In our study, for example, obesity turned out to be not a barrier per se; it rather hinders activity particularly under specific circumstances. According to the results of the tree analysis, such a factor is frequent and lively verbal communication. It has a particularly hindering effect on the physical activity of obese groups. Lack of verbal communication, on the other hand, is mainly a characteristic of very active groups. However, a gender effect became apparent in this regard, namely, that non-communicating male groups were significantly more active than non-communicating female groups. In fact, this group configuration was one the contrast clusters with highest physical activity level of all. We thus assume that non-communicating male groups came to the pool very often with the purpose of being physically active or doing sports. In contrast to this, in non-communicating female and mixed-gender groups, the level of activity was much higher on weekdays than on the weekends. This suggests that the degree to which these groups allow themselves to relax correlates with the amount of possible free time. The observed gender difference could be explained by the observation of Kilpatrick et al. that particularly men are “inclined to view exercise and fitness activities as an opportunity to pursue and achieve ego-related outcomes” [41]. Against this background, one could assume that physically active men who spend their free time together have a rather functional understanding of leisure activities.

Finally, it is important to highlight that ‘mixed’ groups regarding gender, body shape, and age tend to be more inactive than homogeneous groups. Hence, physical activity seems to decrease when people of different body size, gender, and age socialise. This could - again - be a special effect of social contagion in the sense that active persons mirror the behaviours of the inactive [14, 39, 40]. In this regard, the reduction of activity in mixed groups could be seen as a ‘courtesy’ or ‘protection’ habit [30]. A central role in this respect play so called ‘inactivity anchors’ [30]: for example ‘overprotective parents’ who limit their children’s activity behaviour or young individuals who limit their own activity out of consideration of the older group members. The same could be assumed about the interaction of athletic/normal weight persons with obese persons or for the interaction of gender. This habit was brought to the attention of data collectors when they discussed observations with locals in both settings [30]. When people of different age, body shape, and gender socialise, the more active individuals may adjust their behaviour as a courtesy to the rest of the group.

Our study shows that a number of social factors determine active and inactive groups. Future health promotion strategies should therefore consider these determinants in group level interventions instead of just focusing on individual motivational approaches [38, 42].

Our findings contradict somewhat the general health scientific assumption that the inclusion in a social network is one of the most beneficial health factors [43]. Although joyfully chatting with others in a relaxing environment is an important element of health promotion programmes, our results show that it can also hinder physical activity, at least in adults. To overcome this ‘health promotion paradox’, future research has to investigate how the ‘socially connecting’ effects of joyful play activities can be transferred into health related exercise programmes. A potential strategy could be to use team sports as a medium of health promotion because they are seen as more motivating than individually executed exercise and “may facilitate improved adherence to physical activity” [41], particularly because of their joyful and playful nature. A very important target group in this regard are obese people. Lecturing obese people about their potential health hazards often has counterintuitive effects on their weight management and health behaviour: they tend to overeat, avoid diets and engage in less physical activity particularly if they have been victims of weight-related stigmatization [44‐47]. Finding ways to create joyful environments which foster physical activities of obese people, without lecturing or humiliating them, is therefore a very important challenge for health research.

For future research, we recommend to employ the method of participant observation in order to study physical activity in other social settings. In this context, it may be beneficial to also include research settings such as parks, beaches, or playgrounds [30], but also educational contexts or workplace environments.