Market segmentation

Sports organizations that include managers of physical activity programmes for populations increasingly face a saturated market with huge competition to attract citizens. This competition has led to the offering of a wide range of physical activities to various groups of citizens, forgetting that among the mass of existing participants it is possible to find smaller subsets of individuals (^{Ross, 2007}). Sports managers cannot afford to stop working to identify characteristics and subdivisions of individuals from among all participants in their sports activities and events.

Therefore, the practice of locating groups within the base of current runners is necessary for success, because a sports organization or a public programme to promote physical activity cannot gain a strong position in the market if these individuals are not properly identified and managed. This process of identifying well-defined groups of individuals is known as market segmentation.

Market segmentation is the process of dividing a large, heterogeneous market into more homogeneous groups of individuals, who have similar needs, motivations, or demographic profiles, to whom a product may be targeted (e.g., ^{Mullin et al., 2000}).

It provides a conceptual framework upon which a sports programme manager builds promotional strategies to achieve organizational objectives. The segmentation criteria are not limited to demographic and socioeconomic characteristics, values and lifestyle characteristics (psychographics), product usage patterns, or attitudes towards products or benefits sought (^{Ross, 2007}). For the purposes of this research, loyalty to running is held to correspond to product usage patterns, and running involvement is believed to influence positively attitudes towards products.

Cluster analysis

Cluster analysis through the segmentation process has become a common tool for marketing research in both academia and in the professional sector with the aim of creating homogeneous sets of consumers. The primary purpose of this research is to use cluster analysis to identify segments of runners based upon the level of connection with running. Using the variables of running involvement, loyalty to running and subjective well-being to describe and distinguish the market segments can result in greater effectiveness in managing marketing campaigns.

These campaigns should pursue the objective of raising the level of sports participation. The level of sports participation represents one of the key results of sports systems and policies, obtained by using the criteria of running involvement, loyalty to running and subjective well-being with a cluster analytic approach. In light of the theoretical rationale of Self-determination theory and the Psychological Continuum Model, as well as empirical evidence, we expect to create several clusters composed of runners with high levels of running involvement, loyalty to running and subjective well-being, and other clusters whose levels of running involvement, loyalty to running and subjective well-being are moderate or low.

Running involvement

A number of studies have examined the motives of marathoners (^{Funk et al., 2011}; ^{Masters et al., 2003}), but limited attention has been given to the concept of leisure involvement with running (^{Cantón, 2016}; ^{Ridinger et al., 2012}). The concept of ‘involvement' was first introduced in psychology as part of social-judgment theory (^{Sherif & Cantril, 1947}; ^{Sherif & Hovland, 1961}). Since that time, the generic appeal of involvement in social psychology has resulted in a proliferation of involvement research (^{Beaton et al., 2011}). However, the notable exception to this proliferation is in the literature of sports management (^{Beaton et al., 2011}).

Activity involvement is an expression of an individual's interest in a specific sports activity and has been defined as “an unobservable state of motivation, arousal or interest toward a recreation activity or associated product” (^{Havitz & Dimanche, 1997}, p. 246).

Sports involvement is conceptualised as a multifaceted construct representing the degree to which participation in a sports activity becomes a central component of a person's life and provides both hedonic and symbolic value (^{Beaton et al., 2011}), or a passion for sport (^{Méndez-Giménez et al., 2016}).

Involvement is an important construct because of its potential influence on people's attitudes and behaviour relevant to a product or activity (Havitz & Dimanche, 1990). Interest in involvement gained strength in consumer behaviour and marketing literature in the area of developing marketing campaigns aimed at increasing intent to buy and purchasing behaviour (^{Bachleda et al., 2016}).

The empirical analysis demonstrates the utility of the sports involvement construct in the context of a marathon event (^{Beaton et al., 2011}). The runner's involvement has been understood as a motivation to process information about the context of the race, and the motivational state is governed by the perceived relevance that the race has for the runner.

Behavioural differences have been found, suggesting that runners with stronger psychological connections are increasingly involved in the frequency, depth and breadth of running-related behaviours (^{Beaton et al., 2011}; ^{López et al., 2017}), such that ^{Clevinger et al. (2020)} confirmed that sports involvement was related significantly to improvements in physical activity self-efficacy and physical self-concept.

According to Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷), a connection with the social context of an athletics race can satisfy psychological needs for competence, autonomy and relatedness, which in turn, sustain persistence in running practice. The Psychological Continuum Model (^{Funk & James, 2001}) establishes that the second level, ‘attraction' - personal importance and meaning - can adequately express the concept of running involvement, as shown in Figure 1.

Figure 1.  Psychological Continuum Model. Adapted from ^{Funk & James, 2001, p. 122}  

Therefore, considering the theoretical references of Self-determination theory and the Psychological Continuum Model, we hypothesise that a cluster can be created that contains the group of runners with the highest levels of running involvement.

Loyalty to running

Since 1960, the definitions of loyalty have incorporated two dimensions: one of attitude and the other of behaviour (^{Russell-Bennett et al., 2007}). Original and historical consumer loyalty, however, was viewed only in a behavioural and observable way.

The concept was expressed by a variety of types of behaviour, which may involve: the repetition of purchase of a particular product or service, the frequency of consumption, the proportion of purchases or the time dedicated to an activity, recommendation behaviour and the degree of relationship with a brand (^{Bodet & Bernache-Assollant, 2011}). In other words, behavioural loyalty places the emphasis on observable behaviour recorded in the past.

Despite this vision oriented towards the observable component, there is agreement (^{Bodet & Bernache-Assollant, 2011}, ^{Mahony et al., 2000}) that sports consumer loyalty is defined as a two-dimensional concept that brings together an attitudinal component and a behavioural component.

Referring to loyalty as the individual's commitment to a sporting activity that is reflected in a persistent attitude that guides behaviour, ^{Funk and James (2001)} put forward the Psychological Continuum Model which maintains that this commitment is represented by four levels of psychological connection with, and response by, participants to a sporting object (e.g., an athletics race), through stages or levels along a vertical psychological continuum, as shown in Figure 1.

Levels 2 (Attraction), 3 (Attachment) and 4 (Allegiance), which this study focuses on, culminate in a behaviour of fidelity to the practice of running. The idea of loyalty by participants refers to a process of psychological connection to the sporting object that results in positive, consistent and lasting attitudes, as well as the behaviour that is associated with it (^{Funk & James 2001}).

Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) argues that behavioural persistence - in running - can be explained by a connection with a social context - running - which ensures the satisfaction of needs competence, autonomy and relatedness, which in turn support persistence. The Psychological Continuum Model (^{Funk & James, 2001}), argues that the fourth level of connection is allegiance, in which individuals express a devotion to some sport - running - translated into a behaviour of loyalty.

Thus, according to Self-determination theory and the Psychological Continuum Model we hypothesise that a cluster may be created that contains the group of runners with the highest levels of loyalty to running.

Subjective well-being

Quality of life and well-being have been analyzed from different points of view over time. During recent years, several studies have used the most diverse terms such as happiness and life satisfaction when aiming to study the way people evaluate their own lives (^{Diener & Diener, 1996}; ^{Rodríguez-Bravo et al., 2020}). Definitions of subjective well-being can be grouped into three domains (^{Diener, 1984}). Throughout this study, the adopted definition was the one formulated by social research, according to which subjective well-being depends on how and why individuals positively evaluate their lives, i.e., individual's life satisfaction. This dimension has been attracting sociologists' attention and stands out as the main indicator of well-being (^{Giacomoni, 2004}). In this study “satisfaction with life” and “subjective well-being” are used synonymously.

According to ^{Diener and colleagues (1999)}, subjective well-being is “a broad category of phenomena that includes people's emotional responses, domain satisfactions, and global judgments of life satisfaction”, which expresses an individual's current assessment of their own happiness (^{Schwartz & Strack, 1999}).

It is worth mentioning three characteristics of subjective well-being (^{Diener, 1984}): 1) well-being is subjective, depending on the individual's perception and personal experience; 2) well-being implies not only the absence of negative factors, but also the presence of positive factors; 3) well-being must be seen as an overall measure rather than a measure of a single aspect of life. We include self-rated well-being and symptoms as indicators of subjective well-being.

We agree with the idea of ^{Testoni et al. (2018)} that, in spite of all the facts of well-being, only subjective well-being can encapsulate whether running (or anything else) really matters to people. Indeed, subjective well-being is not observed directly; it can only be measured by asking people to report on how well they feel. The very way we live our lives and what we value in it are subjective, and what makes life good or bad is precisely the subjective character of our experiences.

Leisure and physical activity experiences are a crucial part of a balanced lifestyle and are often described as essential to well-being. (e.g., ^{Wiese et al., 2017}). Leisure activities, such as running, are held to promote subjective well-being through a variety of psychological and physical mechanisms. For example, ^{Newman et al. (2014)} argue that leisure activities improve subjective well-being by allowing individuals to turn off and relax from the stresses of everyday life as well as by engaging in activities - running - that fulfil several psychological needs.

It has been verified that the derived approaches of Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) promote sustained long-term behavioural changes in diverse areas such as increased physical activity (^{Flannery, 2017}). Central to Self-determination theory is the concept of intrinsic motivation, which is based on the satisfactions of behaving “for its own sake” (^{Self-Determination Theory, 2017}).

Self-determination theory is a highly appropriate conceptual framework from which to study sports persistence and dropout (^{Calvo et al., 2010}), so it was used as the theoretical framework for this study. This theory examines whether sustained participatory involvement, defined as continued participation in a sport throughout the following year, was influenced by individuals' self-determined motivation or by the fulfilment of the three basic psychological needs of autonomy, competence, and relatedness. When the social context facilitates these three basic needs, a person is more likely to engage in intrinsically motivated types of behaviour. If the situational context promotes these three needs, then a person's natural behaviour is more likely to be self-motivated.

Our approach considers subjective well-being perspectives that focus on how people feel (over the short and long term) as a result of their actions and situations (e.g., ^{Kahneman et al., 1999}).

Figure 2 below considers running traditionally seen as a widely accessible alternative to physical activity and sport and shows how running (running involvement and loyalty to running) can support subjective well-being.

Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) holds that psychological well-being can be nourished by social contexts - athletics races - in which individuals satisfy the psychological needs for competence, autonomy, and relatedness.

Figure 2.  Model of how running enhances subjective well-being via satisfaction of psychological needs 

An important proportion of the literature has focused on identifying the sociodemographic and economic determinants of individuals' subjective well-being (^{Dolan et al., 2008}; ^{Gallagher et al., 2015}; ^{Kahneman et al., 1999}; ^{Scorsolini-Comin & Santos, 2010}) such as income, work situation, academic degree, gender and race.

However, in the sports field, few studies have attempted to examine whether physical and sporting activity impacts individuals' overall well-being. ^{Pawlowski et al. (2011)} argue that practising physical activity and sports is a personal and rational decision that maximizes an individual's rational utility, and must therefore, be logically associated with increased subjective well-being.

Some results show convergence of positive and significant effects of sports practice on subjective well-being (^{Chirivella, 2001}; ^{Kim & James, 2019}; ^{Ruseski et al., 2014}; ^{Wiese et al., 2017}) and stress perception (^{Da Silva et al., 2020}); ^{Lechner (2009)} observed significant effects of sports participation on men's subjective well-being but not on women's; Rasciute and ^{Downward (2010)} concluded that walking and recreational cycling had a positive effect on individuals' happiness; ^{Downward and Rasciute (2011)} verified different effects on subjective well-being when sports practice involved a higher level of social interaction (team sports performed with partners) and higher monthly frequency; frequency and intensity of exercise (^{Moradi et al., 2014}) and active lifestyle had positive effects on life satisfaction (^{Gutiérrez-Caballero et al., 2019}); in turn, ^{Silva et al. (2020)} showed that sports participation - weekly practising of vigorous physical activity - positively affected an individual's subjective well-being.

^{Panza et al. (2019)} highlighted that the effect of physical activity intensity on subjective well-being has not been well established. There are contradictory results. For example, ^{Kekäläinen et al. (2020)} found that different types of physical activities were related to different dimensions of well-being. Walking had positive associations with psychological and social well-being and endurance training with subjective health. Rambling in the countryside was also positively related to subjective health but only among men; ^{Salama-Younes (2018)} proved that harmonious passion was a positive predictor of life satisfaction, whereas obsessive passion - extreme loyalty to running - was not a direct negative predictor of life satisfaction. ^{Wicker and Frick (2017)}, in a study that assessed the effect of different intensities of physical activity and the recommendations of the World Health Organization (WHO) on the subjective well-being, found that for 18- to 64-year-olds, walking (minutes and days/week) and vigorous activity (minutes/week) significantly added to subjective well-being, while moderate activity (minutes/week) had a negative effect.

Surprisingly, running involvement, frequency and number of runs have been neglected in previous studies. Based on these gaps in the running activity literature, we offer a description of the runner segments based on running involvement, loyalty to running and subjective well-being. Thus, based on empirical evidence and the terms of Self-determination theory and the Psychological Continuum Model, we hypothesise that a cluster can be created containing the group of runners with the highest levels of subjective well-being.

Research design

The study was empirical in type, transversal, using an associative strategy in a comparative process (^{Ato et al., 2013}).

Participants

The respondents were selected from the runners of the Scalabis Night Race athletics race, from which was taken a convenience sample. The selection of this race was carried out according to the following criteria: (1) all runners can be considered with leisure and competitive goals, but there are no cash prizes; and (2) the event shows a good participation rate, with more than 4,000 runners, which is necessary to obtain a good sample. The research was conducted in May 2018, after the completion of the race.

The data was collected by supplying a questionnaire to runners in the "Scalabis Night Race" (https://scalabisnightrace.pt/) that has been held for over 12 years. The race has two distances, 10 km, and 5 km. An email containing a link to a self-administered questionnaire was sent to all runners who participated in the 2018 event who provided their email address on the registration form. The questionnaire remained active for two weeks, by which time 567 usable questionnaires were collected, out of a total of 3,305 questionnaires sent, with a response rate of 17.1%.

The sample shows a 98% confidence level and a sampling error of up to 4%. Therefore, the sample was very representative of the runners participating in the race. The questionnaires were supplied only after obtaining informed consent.

The search software allowed only one response to be recorded for each IP address, preventing participants from answering the questionnaire several times. It is important to highlight that the use of questionnaires commonly posted on web pages allows the collection of large samples in a short period of time, but can limit the representativeness of the sample. Participants consisted of a total of 567 individuals, 267 females (47%) and 300 males (52.9%) aged 41.21±9.29. One hundred and forty-five took part in the 5km and 422 ran in the10km. Twenty eight percent were single, 46.38% married, 8.2% divorced, 16% non-marital and 0.88% widowed.

Measurements

For translating and adapting the instruments from the original language to Portuguese, we followed the recommended methodological prodecures (^{Vallerand, 1989}). Participants respondend to several questionnaires related to all the theoretical concepts under analysis, see Table 1.

Table 1.  Variable definition and measurement 

^(*)Standard Deviation

Statistical analysis

To characterise the subjects and identify different profiles, we used an unsupervised machine learning method for the clustering. The adopted approach was random forest clustering, an ensemble approach (^{Breiman, 2001}), with the following advantages (^{Cutler et al., 2012}): they naturally handle both regression and (multiclass) classification; are relatively fast to train and to predict; depend only on one or two tuning parameters; have a built-in estimate of generalisation error; can be used directly for high-dimensional problems; can easily be implemented in parallel. Handling also mixed variable types well, is invariant to monotonic transformations of the input variables, and is robust to outlying observations (^{Shi & Horvath, 2006}). The number of clusters was determined using Bayesian Information Criterion (BIC), giving the minimum BIC score which can be selected as the best model (^{Schwarz, 1978}), simplifying the problem related to choosing the number of components and identifying the structure of the covariance matrix, based on modelling with multivariate normal distributions for each component that forms the data set (^{Akogul & Erisoglu, 2016}). The calculation of the number of clusters used the package mclust (^{Scrucca et al., 2016}) developed in R (^{R Core Team, 2020}). The number of clusters identified was four considering the lower BIC applying the VVI (diagonal, varying volume and shape) model with the value -10,404.19. he multivariate normality test using Shapiro-Wilk showed non-normal distribution (W=0.776, p=0.01). Thereafter, Kruskal-Wallis (^{Kruskal & Wallis, 1952}) and Dunn's Post Hoc (^{Dunn, 1964}) were used to examine whether differences existed among the clusters in subjective well-being, loyalty to running and running involvement (centrality and attraction). The analysis was developed using JASP (^{Love et al., 2019}), and the cluster visualisation with t-SNE (^{Maaten & Hinton, 2008}), which allows significantly better visualisation, reducing the tendency to crowd points together in the centre of the map. t-SNE is better than existing techniques at creating a single map that reveals structure at many different scales (^{Maaten & Hinton, 2008}).

Running involvement

Our first research question has been confirmed. It is possible to state affirmatively that runners with a higher level of running involvement belong to the group of runners with a higher level of subjective well-being.

The involvement with an activity as an expression of motivation and high interest in a specific sports activity was evidenced by cluster 4 (n = 25; 4.4% of the sample); this was labelled the High Group as runners in this cluster showed high levels of subjective well-being, running involvement and high/moderate levels of loyalty to running.

This set of runners were those who consider that running has a central role in their lives, are highly interested in running with much of their time organized around running, which manifests a high level of running involvement. In turn, this group manifested the highest levels of subjective well-being.

In empirical terms, this result agrees with behavioural differences found by ^{Beaton et al. (2011)}, suggesting that runners with stronger psychological connections are increasingly involved in the frequency, depth, and breadth of race-related behaviour. In turn, it was confirmed that involvement in sports was significantly related to improvements in the self-efficacy of physical activity and physical self-concept (^{Clevinger et al., 2020}), elements that may suggest a link to subjective well-being.

This result seems to be explained by the theoretical contributions of the Psychological Continuum Model (^{Funk & James, 2001}) and Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷). Considering the Psychological Continuum Model (^{Funk & James, 2001}), the runners that are at the upper levels of the model correspond to levels 3 (Attachment) and 4 (Allegiance), which culminate in persistence, loyalty and regular running behaviour.

Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) states that a connection with the social context of an athletics race can satisfy psychological needs, which in turn, sustains persistence in running practice. This was the group of runners with the high/moderate level of loyalty to running, who completed on average 23.88 races per year. It was also the group that showed the highest levels of subjective well-being, which under the terms of Self-determination theory seem to have satisfied the basic needs of competence, autonomy and relatedness, thereby providing it with a high satisfaction with life.

Cluster 2, called the Low Group, consists of runners with the lower levels of running involvement and also with the lower levels of subjective well-being. This result is relevant from a theoretical and practical point of view. They seem to confirm the differences in runners' behaviour, suggesting that runners with weaker psychological connections are less involved in the frequency, depth and breadth of running-related types of behaviour (^{Beaton et al., 2011}; ^{López et al., 2017}). This group of runners (76.81% are women) should be targeted with appropriate strategies in order to raise their levels of running involvement. In the social context of the race, activities that lead to greater satisfaction of basic needs - competence, autonomy and relatedness - as supported by Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷), should be developed. The implementation of these strategies will help to achieve the goals of satisfying, retaining and ensuring persistent runner behaviour.

Loyalty to running

The number of races completed per year is the best indicator of race loyalty. The research question we asked has been confirmed: Cluster 3 emerged (n = 93; 16.4% of the sample) and was labelled the Moderate/Low Group, as runners in this cluster showed moderate levels of subjective well-being, moderate levels of running involvement and the highest levels of loyalty to running, with 56.84 ± 31.40 races per year.

The highest levels of loyalty to running refer to an individual's commitment to sports activity that is reflected in a persistent attitude that guides frequent and enduring behaviour.

Strangely, this group manifested a low level of subjective well-being (the third lowest of the four clusters), a situation that seems contradictory for the following reason. According to Self-determination theory (^{Ryan & Deci, 2000}; ²⁰¹⁷), high levels of running involvement, according to the in the social context of the race, would have led to the satisfaction of basic needs (competence, autonomy and relatedness), a factor that would lead to persistent behaviour. But this persistent behaviour may have "crossed the line" and have become an obsessive behaviour in such a way that it may not lead to a higher feeling of subjective well-being, as one would expect. A similar situation has been reported (^{Salama-Younes, 2018}), where a harmonious passion was a positive predictor of life satisfaction, but obsessive passion - extreme loyalty to running - was not a direct negative predictor of life satisfaction. In addition, Self-determination theory is a highly appropriate conceptual framework from which to study sports persistence and dropout (^{Calvo et al., 2010}); it explains loyalty to running, since a full satisfaction of the needs of competence, autonomy and relatedness leads to persistent behaviour observed by the highest number of races held per year.

Considering the Psychological Continuum Model, this group shows very high results at level 4 (allegiance), which can mean loyalty or devotion to a cause or similar (^{Funk & James, 2001}). These runners seem to receive a high intrinsic value which creates a persistent attitude translated into consistent evaluative responses elicited by the sports object over an extended period of time, which may even assume an obsessive characteristic.

Cluster 2, designated the Low Group, consists of runners with the lowest levels of loyalty to running (M = 1.9 races per year) and also with the lowest levels of running involvement. From a theoretical point of view, it can be explained by an insufficiently assigned importance to a race, perhaps by the reduced satisfaction of the needs of competence, autonomy and relatedness - supported by Self-determination Theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) - factors that may have led to non-persistent behaviour. Appropriate strategies are required for this group of runners (76.81% are women). Segmentation requires appropriate strategies with the aim of raising the levels of running involvement with the higher objective of retaining and ensuring a persistent and frequent behaviour in this group of runners.

Subjective well-being

As expected, a cluster was formed with the highest levels of subjective well-being: Cluster 4 (n = 25; 4.4% of the sample) was labelled the High Group as runners in this cluster showed high levels of subjective well-being, running involvement and high/moderate levels of loyalty to running. Also, with relevant levels of subjective well-being we created the cluster with the largest number of runners. Cluster 1 (n = 380; 67.00% of the sample) was labelled the Moderate Group as runners in this cluster showed the second highest level of subjective well-being, but low levels of running and loyalty to running.

Cluster 4 was labelled the High Group because it was the group of runners that showed the greatest feeling of subjective well-being, which probably suffered influence from the high levels of running involvement (centrality and attraction). These runners attach high importance to running; much of their life is organized according to running and running is one of the most fun activities for them. However, this group was not the one that manifested the highest levels of loyalty to running (M = 23.88 ± 5.78 vs 56.84 ± 31.40 of the High Group). It was not a high level of loyalty to running that characterised the group with the highest levels of subjective well-being. On the contrary, it was a moderate level of loyalty to running that showed high levels of subjective well-being. This result reinforces two central ideas in the concept of subjective well-being. The first idea (^{Testoni et al., 2018}) is that the way we live our lives and what we value in it are subjective, and what makes life good or bad is precisely the subjective character of our experiences (running and number of races completed). The second idea (^{Diener, 1984}) highlights that well-being implies not only the absence of negative factors, but also the presence of positive factors (a moderate number, and perhaps not an excessive number, of races completed). These ideas can also support the results obtained in the creation of cluster 1 that consist of the large majority (two-thirds) of runners, which despite having low levels of running involvement (centrality: M = 3.51; attraction: M = 4.89) and loyalty to running (M =1 3.02 races per year) was the group that had the second highest level of subjective well-being (M =7 .60 vs the highest C4: M = 9.20)

This result in particular diverges from the position of ^{Pawlowski et al. (2011)} who argue that practising physical activity and sports is a personal and rational decision that maximizes an individual's rational utility and must therefore be logically associated with increased subjective well-being. Perhaps it is a moderate, but vigorous, level of physical activity that shows better results, as evidenced in several investigations: walking and recreational cycling had a positive effect on individuals' happiness (^{Rasciute & Downward, 2010}); sports participation - the weekly practising of vigorous physical activity - positively affected individuals' subjective well-being/life satisfaction (^{Moradi et al., 2014}; ^{Silva et al., 2020}). On the other hand, ^{Downward and Rasciute (2011)} verified different effects on subjective well-being when sports practice involved a higher level of social interactions (team sports performed with partners) - and running as an individual sport may not provide a sufficient level of social interaction.

As ^{Panza et al. (2019)} highlighted, the effect of physical activity intensity on subjective well-being has not been well established, such that (1) ^{Wicker and Frick (2017)}, in a study that assessed the effect of different intensities of physical activity and the recommendations of the World Health Organization (WHO) on subjective well-being, found that for 18- to 64-year-olds, walking (minutes and days/week) and vigorous activity (minutes/week) significantly added to SWB, while moderate activity (minutes/week) had a negative effect; (2) ^{Salama-Younes (2018)} proved that harmonious passion was a positive predictor of life satisfaction, whereas obsessive passion - extreme loyalty to running - was not a direct negative predictor of life satisfaction, and (3) ^{Kekäläinen et al. (2020)} showed that rambling in the countryside was also positively related to subjective health but only among men.

In theoretical terms, Self-determination theory is a highly appropriate conceptual framework from which to study sports persistence and dropout (^{Calvo et al., 2010}), when the social context - running alone or in a group - facilitates the satisfaction of the three basic needs of autonomy, competence and relationship; then the runner will be more likely to engage in intrinsically motivated behaviour, but it will not have provided the highest levels of subjective well-being, as was expected. It seems that these needs are not enough to find satisfaction in the race, and there may be other factors that exert influence on the subjective well-being that need further study.

The lowest levels of subjective well-being were found in cluster 2 (n = 69; 12.1% of the sample; M = 7.19 ± 1.7 vs 9.2 ± 0.65 of the High Group). This set of runners was also the one that manifested 1) the lowest levels of running involvement (centrality and attraction), and 2) the lowest levels of loyalty to running (M = 2.13 ± 0.78 races per year). Considering the Psychological Continuum Model, this group of runners seems to present a residual connection with running. Maybe they are at the base of level 2 of the model - attraction: entertainment, escape, excitement (^{Funk & James, 2001}). In turn, the interpretation of this result in the light of Self-determination theory (^{Ryan & Deci, 2000}; 2017) shows that it is possible to deduce that the social context of running, for these individuals, may not be sufficient for a full satisfaction of the needs of autonomy, competence and relationship, and as such, persistent and lasting behaviour is not observed. Finally, it is possible to assume that the second level: attraction (^{Funk & James, 2001}) does not present a level of psychological connection sufficient to raise the level of subjective well-being.

Limitations and future research

As in any research, there are a number of limitations to the results, which should be mentioned and grouped into conceptual and methodological limitations. First, the assessment of the connection of runners with an activity was structured around the running involvement construct. This theoretical position was obviously very pertinent in evaluating the responses of runners. However, other perspectives could be admitted, such as runner identity (Runner Identity Scale) (^{Lough et al., 2014}). As regards methodology, the use of a convenience sample may have been a limitation.

The results obtained and the methodology that led to the study, in addition to the limitations identified, provide opportunities for further research that are ready to be developed.

Regarding the constructs used, it is clear that the idea of the existence of other constructs, such as motivational or sought benefits, is not excluded. Whereas Self-determination Theory (^{Ryan & Deci, 2000}; ²⁰¹⁷) maintains, to a large extent, that the persistence of behaviour is driven by the satisfaction of the needs of competence, autonomy and relatedness, it would be interesting to evaluate the level of satisfaction of these needs in different runners, in relation to levels of loyaty to running, in order to understand better the attitudes and behaviour of runners. The results obtained showed that the component of behavioural loyalty did not manifest in a cluster where the levels of subjective well-being were the highest, so it is necessary to deepen the studies of the relationship between these two constructs.

In terms of context, it would seem appropriate to investigate at various race distances: 5 km, 10 km, half marathon and marathon, or even trail races.