The Effect of Informal Gatherings on Sustainable Participation in Leisure Sport Activities: The Case of South Korea

Abstract

Do informal gatherings with members in group leisure sports increase satisfaction and sustainable participation? The purpose of this study was to investigate the moderating role of informal gatherings with regard to self-determined motivation (SDM), leisure satisfaction (LS), and intention to participate sustainably (INPS) in the context of group leisure sports. An empirical analysis of a sample of 423 adults participating in group leisure sports was conducted. Structural equation modeling revealed that SDM exerted a statistically significant impact on LS, and, in turn, LS was conducive to INPS. The multigroup analysis indicated that informal gatherings played a significant moderating role in the relationship between non-self-determined extrinsic motivation and LS. Certain sociodemographic characteristics and sports participation behavior were ascertained as control variables. This study theoretically contributes to the expansion of self-determination theory, affirming the effect of informal gatherings as a unique variable and suggesting practical implications for sustainable leisure participation.

1. Introduction

Leisure sports represent a superior discretionary activity. Participating in leisure sports is conducive to achieving several positive health benefits, such as improved physical, mental, and psychosocial health [1,2] and reduced depression levels [3]. Nevertheless, a small proportion of the population in Europe [1], North America [4], and Asia [5] participates in appropriate physical activity to acquire these health benefits or continues to participate in leisure sports long term [6]. For instance, in the United States, the average time spent per day on a sports activity in the civilian population generally accounted for 0.31 h [7].

According to self-determination theory (SDT), people are satisfied with and engage in sustainable participation in activities when they have the autonomy (self-determined) to discretionarily choose the activities, as well as the desire to enhance competence and relatedness through interaction [8,9]. In addition, higher self-determined motivation (SDM) to participate in an activity leads to greater participants’ sense of achievement, leading to positive emotions [10]. According to SDT, analyzing the motivation to participate in leisure sports can offer increased insight into leisure satisfaction (LS) and sustainable participation [11,12]. Given the utility of SDT in elucidating exercise behavior [13], previous studies have immensely applied SDT to sports activity research [8,12,14].

In addition to SDM, scholars have probed the factors facilitating continual leisure sports participation, such as drivers as antecedents [15,16], moderators [17,18], or different sports settings (e.g., club-organized or not [19,20]). As a way to increase participation in leisure sports, joining sports clubs is deemed recommendable, showing that active participation in leisure sports clubs, i.e., playing group sports, provides more advantages or benefits than playing sports individually [21,22]. In previous studies, participants in group or team sports were happier [23], reported enhanced leisure satisfaction or wellbeing [24], or exhibited a better quality of life [25]. In addition, friends or peers can serve as accelerating factors sustainably boosting leisure participation. In a previous study, Salvy et al. [26] found that the presence of peers and friends increased the motivation to be physically active among youth. These studies accentuated that the presence of friends impacts playing sports.

Informal gatherings with members in group leisure sports likely influence leisure continuation or satisfaction. Rodgers et al. [27] reported that social capital could be a critical determinant for some physical health output, reviewing the systematic literature for two decades. Having a get-together during or after a leisure sport activity can be an opportunity for members who are not close to each other to become better acquainted, and such activities are often an important factor necessary to enjoy more leisure sports activities in the future, which frequently occur in Korean society. However, knowledge regarding behaviors that generate persistent participation, such as informal gatherings (e.g., leisure sport class-end parties, regular meetings, or casual irregular gatherings after group workouts), is relatively limited. Due to a new conceptualization of informal gathering, our understanding of its association with leisure sports remains unclear. Therefore, the aim of this study was to investigate the moderating role of informal gatherings as a new, hitherto unthinkable concept. Specifically, this study addresses the following four major research questions: (1) What is the role of the participants’ SDM (intrinsic motivation (IM), self-determined extrinsic motivation (SDEM), and non-SDEM (NSDEM)) in affecting LS? (2) What is the connection between LS and intention to participate sustainably (INPS)? (3) What is the moderating role of informal gatherings after leisure sports in the relationships among SDM, LS, and INPS? (4) What is the control effect of sociodemographic characteristics and sports participation behavior (duration, frequency, and hours) on leisure satisfaction? These findings add comprehensive insight into the existing SDM literature, highlighting informal and social gatherings as a key resource in the initiation and maintenance of sports participation.

2. Literature Review

2.1. Self-Determined Motivation and Leisure Satisfaction

Motivation involves the initiation (triggers behavior), direction (guides the individual toward specific behavior), and maintenance (drives the individual to pursue an activity) of behavior [15]. Gaining insight into motivation in the context of leisure sports is likely the ideal way to understand avid leisure activity participants, such as sports club members. Research concerning leisure motivation has been pursued since the development of Beard and Ragheb’s [28] Leisure Motivation Scale (LMS). Extant studies have been conducted by listing various leisure-related needs and then categorizing these needs through a factor analysis. However, leisure motivation based on individual leisure needs is limited in explaining all various leisure needs. One academic approach to human motivation is Deci and Ryan’s [9] SDT, which provides a pertinent motivational framework for elucidating individual differences in exercise due to external factors that either facilitate or hinder participation in exercise [16].

SDT proposes that motivation varies depending on whether it is self-determined or controlled [12]. Behaviors that are self-determined are not only initiated freely but also emanate from within oneself [29]. In contrast, behaviors that are controlled are regulated by external pressure. Given these distinctions, Deci and Ryan [9] distinguished among the following three forms of motivation: IM, extrinsic motivation (EM), and amotivation (AM). IM reflects engaging in a behavior because it is enjoyable by nature, i.e., the activity is gratifying [15]. Given that not all human behaviors are considered enjoyable, EM is another way to induce behaviors. EM occurs when an activity is performed to elicit or avoid external consequences. These outcomes may be irrelevant to the activity, such as social comparison [30]. Lastly, AM has been defined as representing ‘a state of lacking any intention to engage in behavior’ [31] (p. 191) and is an entirely non-self-determined type of motivation.

SDT proposes that internalized EM can vary depending on the extent [12]. A multidimensional conceptualization of EM comprising external, introjected, identified, and integrated regulations can be hypothesized to exist. These regulations are located on a continuum ranging from lower to higher self-determination and show the level of the internalization process [9]. Integrated regulation represents the most self-determined form of EM, identified regulation is slightly less autonomous than integrated regulation, introjected regulation reflects less self-determined than identified regulation, and external regulation is the least self-determined form of EM [32]. Vallerand and O’Connor [33] developed a motivation scale based on SDT and divided motivation into four motives (IM, SDEM, NSDEM, and AM). Leisure sports participants engage in at least some types of exercise; hence, these individuals are not amotivational because sports participation necessitates motivation. Accordingly, amotivation is not discussed in the current study, and individuals’ motivation is classified as IM, SDEM, or NSDEM.

Satisfaction is a consequence of the response to consumption activities and an emotional state in response to products, services, and performance, such as pleasure [34]. Leisure satisfaction is the degree of positive emotion or cognition formulated after participation in leisure activities [35]. Oliver [36] proposed the expectancy-disconfirmation theory, which compares the difference between expectation and performance of service. Leisure participants are satisfied if their experience is higher than the level of the previous expectation and vice versa.

Intrinsically motivated athletes have better concentration during practice [37]. Owen, Astell-Burt, and Lonsdale’s [38] results showed that individual motivation may be important for leisure time, physical education, and satisfaction with physical activity. Alexandris et al. [11] revealed that IM and EM had a positive effect on leisure participation (frequency of participation), whereas AM did not. Chen, Li, and Chen [39] substantiated that leisure motivation had a positive impact on LS, indicating leisure involvement’s mediating role. Hence, IM and SDEM are more likely to have a positive impact on LS, while NSDM is likely to have a negative impact on LS. According to the aforementioned literature, the first hypothesis was developed as follows:

Hypothesis 1 (H1a).

IM in leisure sports has a positive impact on LS.

Hypothesis 1 (H1b).

SDEM in leisure sports has a positive impact on LS.

Hypothesis 1 (H1c).

NSDEM in leisure sports has a positive impact on LS.

2.2. Relationship between Leisure Satisfaction and Sustainable Participation

Importantly, high satisfaction with leisure sports leads to a more positive influence on sustainable participation activities [40]. The conceptualization of leisure participation and satisfaction has been researched for a long time since Beard and Ragheb’s [34] early study. Positive emotions stemming from participation in leisure sports enable individuals to be satisfied, further leading to sustainable participation [40]. Behavioral intention is defined as an individual’s desire to engage in a particular behavior in a rather short and expected time [41]. To ensure that behavioral intentions succeed in actual behavior, satisfaction should be precedent. Therefore, INPS in leisure sports depends on LS; thus, it may be necessary to conduct research exploring LS among existing leisure sports participants [40]. Consistent with the findings in the above literature, we suggest the following second hypothesis:

Hypothesis 2 (H2).

The LS of leisure sports participants has a positive impact on INPS.

2.3. Informal Gatherings with Members in Group Leisure Sports

Informal gatherings with members in group leisure sports can likely be explicable in three aspects. First, such gatherings are associated with social networking and are somewhat explained by social capital theory. The concept of social capital stems from either Pierre Bourdieu or Robert Putnam. Putnam conceptualized social capital in the form of ‘bridging’ and ‘bonding’ [42]. Social capital is defined as the resources and advantages that individuals and groups receive through interaction with others and is associated with both shared standards and values that promote collaboration and actual social relationships [43]. Scholars have ascertained a positive relationship between social capital and health [44] or health behaviors, including physical activity [44]. Poortinga [45] revealed that social capital was associated with self-rated health and different health behaviors. Wiltshire and Stevinson [46] explored the role of social capital in community-based physical activity and found that community-based running initiatives had the capacity to mobilize resources through social networks. According to these authors, the sports participants benefitted from the wider community (their network of social relations); hence, they are privy to crucial managerial and affective support.

Second, friends or peers can serve as accelerating factors to sustainably enhance leisure participation. For instance, Salvy et al. [26] found that the presence of peers and friends increased the motivation to be physically active and actual physical activity among youth. However, in their study, the outcome indicated that the presence of a peer motivated overweight, but not lean, participants to exercise, limiting generalization to all individuals. In a recent study, Lee and Lee [17] revealed that a high level of exercise participation by close friends was related to youths’ increased intention to exercise and their actual exercise behavior. The authors reported that the exercise participation of close friends was a significant mediator that enables youths’ physical activity to be promoted.

Third, informal gatherings are technically associated with eating together, and food signals the meaning of social relations. Sylow and Holm [47] identified the role of food in social interaction in sports centers in Denmark. According to the authors, food functions as a mediator that introduces occasions for getting together.

Informal gatherings are predicted to have a positive effect on leisure satisfaction or continuation by stimulating motivation and generating relatedness, which is the basic psychological need of SDT. Given the relationship between leisure participation and close people, such as neighborhoods, close friends, or people who exercise together, we assume that informal gatherings are likely to affect sports participation behavior. To the best of our knowledge, the moderating role of informal gatherings after leisure sport activities has not been investigated in the prior literature. However, it can be expected that the impact of IM, SDEM, and NSDEM on LS and the impact of LS on INPS may differ on the basis of individuals’ behavior in informal gatherings. Accordingly, the following hypothesis is proposed:

Hypothesis 3 (H3a).

Informal gatherings positively moderate the relationship between IM and LS.

Hypothesis 3 (H3b).

Informal gatherings positively moderate the relationship between SDEM and LS.

Hypothesis 3 (H3c).

Informal gatherings positively moderate the relationship between NSDEM and LS.

Hypothesis 3 (H3d).

Informal gatherings positively moderate the relationship between LS and INPS.

3. Materials and Methods

3.1. Research Model

Based on previous research, this study presents a research model to analyze the relationship among SDM, LS, and INPS in leisure sports and the moderating effects of gathering vs. non-gathering as shown in Figure 1.

3.2. Instruments

The measurement items were divided into the following five categories: leisure sports participation behavior, SDM, LS, INPS, and demographic characteristics. First, the respondents were asked to report the type of sports they mainly performed. Leisure sports participation behavior consisted of the duration, participation frequency, participation hours, and informal gathering. Questionnaire completion was discontinued if it was less than 1 year since the respondent began the leisure sports activity or if their frequency of participation was less than once a month as assessed by filtering questions.

Second, according to Deci and Ryan [9], Pelletier et al. [37], and Vallerand and O’Connor [33], SDM was assessed with the following question: ‘Why are you currently participating in this leisure sports activity?’. There were three IM, three SDEM, and three NSDEM items, accounting for a total of nine items. An example of an IM response was ‘for the pleasure of doing it’; an example of an SDEM response was ‘I choose to do it for my own good’; an example of an NSDEM response was ‘because I am supposed to do it’. Third, LS was measured with three questions developed by Oliver [36] and applied by Laverie and Arnette [48] that were adapted to the sports context. Fourth, INPS was described on the basis of the studies by Bum et al. [40] and Kim et al. [10] and evaluated via three questions using a Likert scale ranging from 1 (not at all) to 5 (very likely). Lastly, demographic variables (gender, age, marital status, educational background, monthly income, and occupation) were collected.

3.3. Participants and Data Collection

The data were collected over 3 weeks. As a pilot survey, from 10 to 15 January, 100 members of golf, badminton, baseball, tennis, table tennis, dance, and bowling sports clubs located in Seoul were surveyed via a self-report questionnaire. Then, the initial items were revised according to the feedback received from two professors. Subsequently, an online questionnaire was prepared for the main survey and distributed from 23 to 31 January. The questionnaires were collected via Naver Form (office.naver.com/accessed on 23 January 2018), which is a widely used tool in South Korea hosted on an online platform. Only one questionnaire could be completed per IP (Internet Protocol) address, and the response rate was increased by providing a 2 USD mobile emoticon gift to the respondents. A total of 850 adults aged 20–80 years were selected from the Korean population. Of the collected questionnaires, those with incomplete data and outliers (n = 258) and those who participated in individual sports (n = 169) were excluded, and a final sample of 423 was used for the analysis. The sample size was applicable for testing structural relationships, which included 15 latent variables, according to the criterion that at least 10 cases per variable were required for SEM [49].

3.4. Data Analysis Procedures

The data were analyzed with Windows SPSS and AMOS version 23.0. First, the descriptive statistics were analyzed, and a confirmatory factor analysis (CFA) was conducted to verify the validity of the latent variables. The internal consistency of the derived factors was verified using Cronbach’s alpha. Subsequently, to test the proposed hypotheses, structural equation modeling (SEM) was performed to examine the relationships among SDM, LS, and INPS. Then, the moderating role of gathering vs. non-gathering was tested using a multigroup analysis. Lastly, a hierarchical regression analysis was performed.

4. Results

4.1. Descriptive Statistics

Table 1. Sociodemographic profile of the sample (n = 423).

Variables	Items	n (%)	Variables	Items	n (%)
Gender	Man	191 (45.2)	Marital status	Married	292 (69.0)
	Woman	232 (54.8)		Other	132 (31.0)
Age	20–29	76 (18.0)	Education level	Highschool degree	53 (12.5)
	30–39	66 (15.6)		Bachelor’s degree	273 (64.5)
	40–49	106 (25.1)		Over Bachelor’s degree	97 (22.9)
	50–59	135 (31.9)	Occupation	Office worker	152 (35.9)
	60 or older	40 (9.5)		Self-employed	48 (11.3)
Income per household (Monthly)	Under 3000 USD	103 (24.3)		Professional/researcher	65 (15.4)
	3000–4999 USD	69 (16.3)		Housewife	98 (23.2)
	5000–6999 USD	70 (16.5)		Student	22 (5.2)
	7000–8999 USD	57 (13.5)		Other	38 (9.0)
	9000 USD or over	124 (29.3)

presents the respondents’ demographic profile. The sample included slightly more women (54.8%) than men (45.2%). Those in their 50s accounted for the greatest percentage of the respondents (31.9%), followed by those in their 40s (25.1%), 20s (18.0%), 30s (15.6%), and 60s or older (9.5%). More respondents were married (69.0%) than not married, and 64.5% of the respondents completed a bachelor’s degree, followed by over bachelor’s degree (22.9%). Office workers constituted the largest percentage of the respondents (35.9%), followed by full-time housewives (23.2%) and professionals/researchers (15.4%). The monthly income followed the order of more than 9000 USD (29.3%), less than 3000 USD (24.3%), 5000–6999 USD (16.5%), and 3000–4999 USD (16.3%).

As shown in

Table 2. Leisure sports participation behavior (n = 423).

Types of Sports	n (%)	Leisure Participation		n (%)
Golf	116 (27.4)	Duration of participation (year) (M = 7.44, SD = 8.41)	Less than 1	78 (18.4)
Racket sports	77 (18.2)		1 < D ≤ 3	100 (23.6)
Basketball/baseball/soccer	36 (8.5)		3 < D ≤ 5	64 (15.1)
Billiard/bowling/table tennis	21 (5.0)		5 < D ≤ 10	93 (22.0)
Mountain climbing	38 (9.0)		Over 10	88 (20.8)
Dance	32 (7.6)	Frequency of participation per month (M = 8.28, SD = 6.44)	Less than 2	80 (18.9)
Yoga/Pilates	103 (24.3)		2 < F ≤ 5	106 (25.1)
Informal gatherings (n = 423)	N (%)		5 < F ≤ 9	90 (21.3)
Yes	248 (58.6)		9 < F ≤ 14	68 (16.1)
No	175 (41.4)		Over 14	79 (18.7)
Frequency of informal gatherings per month (n = 337) (M = 3.26, SD = 3.75)	1~25	Hours of participation per month (M = 17.44, SD = 16.26)	Less than 7	80 (18.9)
			7–10	81 (19.1)
Duration of informal gatherings (year) (n = 341) (M = 5.13, SD = 4.92)	1~30		10–15	73 (17.3)
			15–24	104 (24.6)
			Over 24	85 (20.1)

, the types of leisure sports in which the respondents participated were classified according to the national leisure activities of the Ministry of Culture, Sports, and Tourism in South Korea. Golf was the most common, with 116 participants (27.4%), followed by 103 yoga/Pilates participants (24.3%) and 77 racket sports participants (18.2%). Regarding follow-up questions regarding whether the respondents had informal gatherings after leisure sport activities, 58.6% responded ‘yes’, and 41.4% responded ‘no’. Among the respondents who said yes, 337 responded to the frequency of informal gatherings, which ranged from one to 25 times per month (M_times = 3.26, SD = 3.75). In addition, 341 responded to the question regarding the duration of informal gatherings, which ranged from one to 30 years (M_years = 5.13, SD = 4.92). The degree of sports participation was divided into quartiles based on the median. The participation period ranged from one to 72 years (M_years = 7.44, SD = 8.42). The frequency of participation ranged from one to 30 times per month (M_times = 8.28, SD = 6.44). The participation time per session ranged from 30 min to 15 h (M_hours = 2.84, SD = 2.17). Since there were various types of leisure sports, it was deemed suitable to calculate the monthly participation time multiplied by the participation frequency and participation time and use the result for a comparison (M_hours = 17.44, SD = 16.26). This process was reported elsewhere [2].

4.2. Confirmatory Factor Analysis and Internal Consistency

A confirmatory factor analysis (CFA) was carried out to ensure the unidimensionality of the scales measuring the concepts and validate the measurement scale used in the research. The results indicated an adequate model fit. The CFA chi-square value was 285.699 with 80 degrees of freedom (p < 0.001, GFI = 0.923, IFI = 0.948, TLI = 0.931, CFI = 0.948, and RMSEA = 0.078). The TLI, IFI, and CFI values ranged from 0–1, with an estimate close to 1.0, indicating a good model fit [50]. An RMSEA value below 0.10 is acceptable, but values between 0.04 and 0.08 are considered good [51].

Table 3. Confirmatory factor analysis and internal consistency.

Item	Item	Std. Estimate	Estimate	C.R.	Cronbach’s α
IM	IM 3	0.819	1.000	-
	IM 2	0.814	1.078	17.306 ***	0.840
	IM 1	0.762	0.955	16.188 ***
SDEM	SDEM 3	0.898	1.000	-
	SDEM 2	0.733	0.944	14.337 ***	0.802
	SDEM 1	0.646	0.805	12.843 ***
NSDEM	NSDEM 3	0.789	1.000	-
	NSDEM 2	0.619	0.836	9.343 ***	0.713
	NSDEM 1	0.615	0.776	9.313 ***
LS	LS 3	0.793	1.000	-
	LS 2	0.835	1.094	18.779 ***	0.864
	LS 1	0.853	1.034	19.268 ***
INPS	INPS 3	0.926	1.000	-
	INPS 2	0.951	1.035	36.349 ***	0.942
	INPS 1	0.887	1.005	29.864 ***

Notes: All factor loadings are significant at *** p < 0.001. Goodness-of-fit statistics of the total model: χ2 = 285.699, df = 80, Q(CMIN/df) = 3.571, p < 0.001, GFI = 0.923, IFI = 0.948, TLI = 0.931, CFI = 0.948, RMSEA = 0.078, total α = 0.770. GFI = goodness-of-fit index, NFI = normed fit index, IFI = incremental fit index, CFI = comparative fit index, TLI = Tucker–Lewis index, RMSEA = root-mean-square error of approximation. IM: intrinsic motivation, SDEM: self-determined extrinsic motivation, NSDEM: non-self-determined extrinsic motivation, LS: leisure satisfaction, INPS: intention to participate sustainably.

shows the specific measurement constructs used in this study and their standard estimate values. The factor loadings were equal to or greater than 0.615, and all factor loadings were statistically significant at the p < 0.001 level, with t-values ranging from 9.313 to 36.349. The coefficients of internal consistency of each subscale (Cronbach’s alpha) were confirmed, with the criterion for an acceptable level of 0.60 [48]. The findings showed that the internal consistency was as follows: IM (α = 0.840), SDEM (α = 0.802), NSDEM (α = 0.713), LS (α = 0.864), and INPS (α = 0.942). Each factor met the criterion for acceptable internal reliability with a total α = 0.770.

As shown in

Table 4. Descriptive statistics and associated measures.

	Number of Items	Mean (Std. Deviation)	AVE	(1)	(2)	(3)	(4)	(5)
(1) IM	3	4.23 (0.64)	0.638	0.841 a	0.381 b	−0.230	0.623	0.551
(2) SDEM	3	4.07 (0.67)	0.587	0.145 c	0.807	−0.162	0.418	0.368
(3) NSDEM	3	2.23 (0.79)	0.460	0.052	0.026	0.716	−0.307	−0.325
(4) LS	3	4.30 (0.52)	0.685	0.388	0.174	0.094	0.867	0.744
(5) INPS	3	4.38 (0.64)	0.850	0.304	0.135	0.106	0.554	0.944

Notes: IM: intrinsic motivation, SDEM: self-determined extrinsic motivation, NSDEM: non-self-determined extrinsic motivation, LS: leisure satisfaction, INPS: intention to participate sustainably. AVE = average variance extracted. ^a Composite reliabilities are found along the diagonal line of bold entries. ^b Correlations are found above the diagonal line. ^c Squared correlations are found below the diagonal line.

, as existing research recommends (e.g., [52]), the discriminant validity and convergent validity of the scales were assessed using a CFA. The average variance extracted (AVE) value was above the 0.5 cutoff [53]. The high factor loadings of the intended items and the AVE estimates confirmed that the items had convergent validity [54]. All squared correlations (R²) between pairs of items were lower than the AVE of each latent variable [54], as shown in Table 4.

4.3. Structural Model

As a result of the research hypothesis testing, the model fit index was Q(CMIN/df) = 3.483, p = 0.000, GFI = 0.922, CFI = 0.948, IFI = 0.948, TLI = 0.934, and RMSEA = 0.077. The results showed that all values were consistent with the statistical criteria; thus, the dimensionality of the measurement items was adequate. According to the squared multiple correlation (SMC) value, the variance explained by SDM for LS was 60.3%, and the variance explained by LS for INPS was 72.2%.

The path model analysis showed that H1 was supported; specifically, hypotheses H1a, H1b, and H1c were supported. IM and SDEM had positive impacts on LS. The path coefficient of IM and LS was 0.578 (p < 0.001), and that of SDEM and LS was 0.184 (p < 0.001). NSDEM had a negative effect on LS. The path coefficient of NSDEM and LS was −0.207 (p < 0.001). H2 was confirmed as LS had a positive (+) effect on INPS. The path coefficient between LS and INPS was high at 0.849 (p < 0.001). The hypothesis testing results are shown in

Table 5. Results of the structural model.

H		Hypothesized Association	Std. Estimate	t-Value	Status
H1	H1a	IM → LS	0.578	9.966 ***	Supported
	H1b	SDEM → LS	0.184	3.698 ***	Supported
	H1c	NSDEM → LS	−0.207	−4.233 ***	Supported
H2		LS → INPS	0.849	19.336 ***	Supported

Notes: *** p < 0.001. Goodness-of-fit statistics of the total model: χ2 = 289.048, df = 83, Q(CMIN/df) = 3.483, p < 0.001, GFI = 0.922, NFI = 0.928, IFI = 0.948, TLI = 0.934, CFI = 0.948, RMSEA = 0.077. GFI = goodness-of-fit index, NFI = normed fit index, IFI = incremental fit index, CFI = comparative fit index, TLI = Tucker–Lewis index, RMSEA = root-mean-square error of approximation. IM: intrinsic motivation, SDEM: self-determined extrinsic motivation, NSDEM: non-self-determined extrinsic motivation, LS: leisure satisfaction, INPS: intention to participate sustainably.

.

Bootstrapping is acknowledged as a superior approach compared to other methods (e.g., the Sobel test) in testing the indirect effect among independent, mediating, and dependent constructs [55]. The mediating effects of LS on the relationship between SDM and INPS were examined using the bootstrapping method. The results (shown in

Table 6. Standardized direct, indirect, and total effects.

Paths			Direct Effects (β)	Indirect Effects (β)	Total Effects (β)
IM	→	LS	0.578 ***	-	0.578 ***
		INPS	-	0.491 ***	0.491 ***
SDEM	→	LS	0.184 ***	-	0.184 ***
		INPS	-	0.157 ***	0.157 ***
NSDEM	→	LS	−0.207 ***	-	−0.207 ***
		INPS	-	−0.176 ***	−0.176 ***
LS	→	INPS	0.849 ***	-	0.849 ***

Notes: *** p < 0.001. IM: intrinsic motivation, SDEM: self-determined extrinsic motivation, NSDEM: non-self-determined extrinsic motivation, LS: leisure satisfaction, INPS: intention to participate sustainably.

) revealed that LS significantly mediates the relationships among the four factors of IM, SDEM, NSDEM, and INPS.

4.4. Multigroup Analysis of the Moderating Effects

A multigroup analysis was conducted to analyze the moderating effects of gatherings on the relationships among SDM, LS, and INPS. The respondents were divided into two groups (gathering = 1, n = 248; non-gathering = 0, n = 175). The cross-group equality constraints method was used to test the hypotheses on the basis of the chi-square (χ²) difference between the constrained and unconstrained models by examining the difference in the degrees of freedom.

Then, the moderating role of gathering vs. non-gathering in the relationship between IM and LS was assessed (H3a), and the coefficient of the path between IM and LS was compared between the gathering and non-gathering groups. The chi-square difference between the constrained and unconstrained models was not significant at the 0.05 level (χ² = 1.035 < χ²₀.₅ (1) = 11.849, p = 0.309). This result indicates that the difference between the gathering and non-gathering groups was not significant with regard to the groups with IM; therefore, hypothesis H3a was not supported.

Subsequently, the moderating function of gathering vs. non-gathering in the relationship between SDEM and LS was assessed (H3b), and the coefficient of the path between SDEM and LS was compared between the gathering vs. non-gathering groups. The chi-square difference between the constrained and unconstrained models was not significant at the 0.05 level (χ² = 3.442 < χ²₀.₅ (1) = 8.442, p = 0.062). This result indicates that the difference between the gathering and non-gathering groups was not significant with regard to the groups with SDEM; therefore, hypothesis H3b was not supported.

Then, the moderating role of gathering vs. non-gathering in the relationship between NSDEM and LS was assessed (H3c), and the coefficient of the path between NSDEM and LS was compared between the gathering and non-gathering groups. The chi-square difference between the constrained and unconstrained models was significant at the 0.001 level (χ² = 8.966 > χ²₀.₅ (1) = 3.918, p = 0.003). This finding proves that the effect of NSDEM on LS significantly differed between the gathering and non-gathering groups. In the case of gathering, the path coefficient between NSDEM and LS was −0.053 (p < 0.001), whereas, in the case of non-gathering, the path coefficient was −0.317 (p < 0.001). In brief, gatherings had a significant effect on the participants with high NSDEM compared to those who had low NSDEM. Therefore, individuals with high NSDEM should have gatherings to elevate leisure satisfaction. Accordingly, hypothesis H3c was supported.

Lastly, the moderating role of gathering vs. non-gathering in the relationship between LS and INPS was assessed (H3d), and the coefficient of the path between LS and INPS was compared between the gathering and non-gathering groups. The chi-square difference between the constrained and unconstrained models was not significant at the 0.05 level (χ² = 0.735 < χ²₀.₅ (1) = 12.149, p = 0.391). This result indicates that the difference between the gathering and non-gathering groups was not significant, and hypothesis H3d was not supported. Therefore, H3 was partially supported as shown in

Table 7. Results of the moderating role of gatherings.

Hypothesis 3		Gathering (β)	Non-Gathering (β)	Constraints	∆χ²	Status
H3a	IM → LS	0.705	0.513	χ²(167) = 534.721	∆χ²(1) = 1.035	Not supported
H3b	SDEM → LS	0.052	0.262	χ²(167) = 538.129	∆χ²(1) = 3.442	Not supported
H3c	NSDEM → LS	−0.053	−0.317	χ²(167) = 542.653	∆χ²(1) = 8.966 **	Supported
H3d	LS → INPS	0.902	0.817	χ²(167) = 534.422	∆χ²(1) = 0.735	Not supported

Notes: ** p < 0.01. χ²(166) = 533.687 (p = 0.000), χ²/df = 3.215, RMR = 0.039, GFI = 0.861, IFI = 0.930, CFI = 0.911, RMSEA = 0.073.

.

4.5. Hierarchical Regression Analysis

Additionally, a moderated hierarchical regression analysis with three steps was performed to elucidate the specific process between NSDEM and LS, including control variables, such as sociodemographic variables (gender, age, marital status, education level, and monthly income level) and sports participation behavioral variables (duration, frequency, and hours). Several studies have proven that gender, age, marital status [18], education, income [2], frequency, and time spent [19,20] affect the satisfaction or continuation of leisure sports. First, NSDEM was converted to a Z-score, and the values were mean-centered to avoid potential multicollinearity and entered as an independent variable in Step 1. Second, the moderator (coded: 0 = non-gathering, 1 = gathering) was entered in Step 2. Lastly, the interaction variable was entered in Step 3. LS was the dependent variable.

As shown in

Table 8. Results of the moderating role of gatherings and the control variables.

Variable	Step 1			Step 2			Step 3
	β	t-Value	p	β	t-Value	p	β	t-Value	p
Control variables		24.404	0.000 ***		24.407	0.000 ***		24.530	0.000 ***
Gender	0.044	0.829	0.407	0.036	0.677	0. 499	0.035	0.668	0.505
Age	0.144	2.072	0.039 *	0.136	1.950	0.052	0.109	1.558	0.120
Marital status	0.166	2.398	0.017 *	0.169	2.442	0.015 *	0.145	2.081	0.038 *
Education	0.025	0.486	0.627	0.023	0.449	0.654	0.023	0.456	0.648
Income	0.091	1.571	0.117	0.082	1.400	0.162	0.079	1.355	0.176
Participation duration	−0.011	−0.203	0.839	−0.021	−0.388	0.698	−0.016	−0.286	0.775
Participation frequency	−0.053	−0.831	0.406	−0.053	−0.829	0.407	−0.059	−0.937	0.349
Participation hours	0.241	3.985	0.000 ***	0.239	3.953	0.000 ***	0.267	4.361	0.000 ***
NSDEM	−0.328	−7.090	0.000 ***	−0.337	−7.224	0.000 ***	−0.546	−6.607	0.000 ***
Gathering				0.069	1.402	0.162	0.082	1.655	0.099
NSDEM × gathering							0.170	2.405	0.017 *
R²	0.174			0.178			0.189
ΔR²				0.004			0.011
F-value	9.674			8.923			8.732
ΔF				1.966			5.783
p	0.000 ***			0.000 ***			0.000 ***

Notes: * p < 0.05, *** p < 0.001. Gender was coded as 0 (female) and 1 (male). Marital status was coded as 0 (married) and 1 (single and others). The interaction variable was formed by multiplying gatherings by the mean-centered NSDEM. The dependent variable was leisure satisfaction.

, when NSDEM was used as an independent variable only in Step 1, the total variance explained (R²) was 17.4%. NSDEM exerted a significant effect on LS (p < 0.001). In Step 2, the addition of the moderator increased the total variance to 17.8%, and the increase in the variance was significant (F = 9.674, p < 0.001); therefore, gathering exerted a significant moderating impact. As a result of applying the interaction variable in Step 3, the total variance increased to 18.9%, and NSDEM had a significant effect on LS (F = 8.923, p < 0.001). Therefore, the regression model was statistically significant. Because the interaction variable ‘NSDEM × gathering’ had a t-value of (+)2.405, informal gatherings had a positive effect on the participants with high NSDEM compared to those who had low NSDEM with regard to LS. Therefore, this result can be interpreted to indicate that the effect of gathering is high in the group with high NSDEM. The results of the analysis of the impact of NSDEM on LS are presented in Table 8 and Figure 2.

As reported in Table 8, marital status had an effect on LS at the 0.05 significance level (β = 0.145, t = 2.081, p = 0.038), and a single status with NSDEM had a greater effect on LS. The only sports participation control variable that affected LS was participation hours, which had a strong effect on LS at the 0.001 significance level (β = 0.267, t = 4.361, p = 0.000), and the more time individuals invested in group sports, the more satisfied they were. The other control variables, such as gender, age, education level, income, and the sports participation behavioral variables (duration and frequency) did not exert significant impacts on LS.

5. Discussion and Conclusions

The primary purpose of this article was to examine the moderating effect of gathering on the sustainable participation of leisure sports in the leisure sports domain in South Korea. To achieve the objective of the study, we first identified the structural model of SDM, LS, and INPS. First, SDM had a significant influence on LS. Specifically, IM and SDEM had a positive impact on LS, whereas NSDEM had a negative impact on LS. Most importantly, the variable with the highest coefficient value that affected LS was IM (β = 0.578). This finding is consistent with the results reported by Cox and Williams [56] and Ryan and Deci [31]. Specifically, the more leisure sports participants think that the activities are fun, enjoyable, healthful, and important (i.e., IM), the higher their leisure satisfaction is, while the more people engage in activities because of others or out of a sense of duty, the lower their LS is (i.e., EM). Therefore, to increase LS, it is necessary for activity programs to attract voluntary participation from participants and instructors to provide information to help the participants realize the benefits of leisure sports. Puente and Anshel’s [16] research showed that instructors’ interacting style affected self-determined regulation to exercise, accentuating the role of instructors. Similarly, exercise class leaders were positively related to SDM [12].

Second, the relationship between LS and INPS was found to have a positive influence. This model was theoretically supported in DeFreese and Smith’s [57] study, which suggested that SDM played a key role as an antecedent of satisfaction, achievement, and intention to continue an activity. Thus, it is important to satisfy leisure sports participants to realize sustainable participation in the activity. The indirect effect of LS played a significant mediating role in the relationship between SDM and INPS. Therefore, the structural model of SDM, LS, and INPS fully supported the hypotheses.

Third, the moderating effect of informal gatherings vs. non-gatherings with members in group leisure sports was examined. According to the results, gatherings moderated the NSDEM–LS relationship, and no other paths exerted a significant impact. Interestingly, this study revealed that those with high NSDEM when having informal gatherings after leisure sport activities had greater LS than those with non-gatherings, indicating that ‘gathering’ is indeed of importance to individuals with less autonomy motivation (i.e., NSDEM). In contrast to our expectation, gatherings did not have any moderating effect on the relationship between IM and LS, SDEM and LS, or LS and INPS. Therefore, motivation with autonomy heightens satisfaction or sustainable participation regardless of gatherings.

Lastly, the findings of this study revealed that LS differs depending on certain demographic variables (such as marital status) and sports-related characteristics. More specifically, the effect on LS was greater among the singles than among the married. Hours of participation per month served as a control variable, and differences were found, partially supporting Borgers et al.’s [19] study, which showed that frequency and time spent on sports significantly increased sport participation when engaging in club-organized sports. Moreover, this outcome is inconsistent with the findings by [2], who demonstrated that more exercise was not always better. However, gender, age, education level, income, duration, and frequency were not associated with heightened LS.

6. Implications

The results of this study provide a theoretical contribution in the context of the leisure literature in general and SDT literature in particular. The originality of this study lies in the attempt to adapt a unique variable, i.e., informal gatherings, to amalgamate leisure studies. This early study substantiates the moderating role of gathering vs. non-gathering in general leisure studies. As previously mentioned, prior studies identified the importance of social relations with leisure participants in a social capital domain. The findings of this study show the adaptability of the effect of gatherings in the context of SDT, and more research regarding social networks can be performed in the future.

The findings of this investigation also offer a few managerial implications. First, the findings are practical for sports instructors and practitioners. The empirical results imply that the participants of leisure sports who had IM and SDEM had a positive LS, which is conducive to INPS. This finding may be useful for sports instructors. When training practitioners, instructors should teach by the principles of SDT. SDT has already been applied to educational settings [29] and exercise settings with success [12]. Additionally, by informing and educating sports club members about the benefits of performing sport activities in detail and explaining the actual changes that could occur when the sports activities have been completed, the members may achieve a higher motivation level to continue the activities for a sustained period with more satisfaction and confidence. Providing such instruction may be critical for exercise novices, who are more likely to experience boredom [58]. By accentuating voluntary participation and its advantages, the leisure sports business could also benefit from continued support and patronage from its participants.

Second, the findings are helpful for online and offline marketers of leisure-related industries. The results showed that informal gatherings after leisure sport activities played an important moderating role in the relationship between NSDEM and LS, which is interpreted as casual informal gatherings leading to a smaller decrease in LS, especially among people with high NSDEM. Individuals with NSDEM are likely to quit attending group sports unless other members encourage continued participation. Therefore, fitness club organizers and leisure industry marketers need to investigate club members’ motivation on the first day of exercise class and attempt to pay more attention to individuals with NSDEM. Creating a welcoming environment might allow those with more or less NSDEM to feel more at ease in sports centers or leisure sport domains. Due to the strong measures against coronavirus 2019 (COVID-19), it has recently been uncomfortable to hold informal gatherings, which is worse for those with NSDEM. However, online gatherings (e.g., Google Meet, Zoom, and WebEx online meetings) enable individuals to participate anytime and anywhere. Informal gatherings do not always imply eating and drinking together and can be any type of relationship building. Gatherings embody spending time together online or offline, i.e., listening to other members’ techniques regarding exercise and holding a seminar or even online marketing related to exercise-related equipment in the health industry. Therefore, marketers should not be indolent in the advent of related products and concomitant changes in group sports by building up momentum.

Several limitations should be considered for future research. The data in this study were collected from South Korea only; therefore, this sample limits the generalizability. The sample consisted of a relatively larger female group; however, this corresponded to the statistics of the general Korean adult population of sports participation [59], and, because we controlled for gender, selection bias toward gender is unlikely. As another limitation, we raked various types of group sports; therefore, heterogeneous bias should be considered in the future. Personality might affect LS; hence, future studies can treat this variable as another antecedent. These issues remain, and we await future research. Data collection in other countries may provide valuable information for cultural difference comparisons in leisure sports activities.