Too busy or too far away? The importance of subjective constraints and spatial factors for sports frequency

Abstract

Previous studies on the association between the availability of sports facilities and sports participation have neglected the influence of subjective constraints that individuals experience with regard to sports participation. This paper investigates to what extent constraints experienced by sports participants are associated with their spatial circumstances and whether these subjective constraints or objective spatial circumstances have a greater impact on sports frequency. Based on a survey among 776 adults in urban and rural municipalities in the Netherlands, regression analyses revealed that constraints were related to neighbourhood liveability and distance to indoor sports facilities and swimming pools. Time constraints had a strong negative effect on sports frequency, but the effect of distance to indoor facilities and swimming pools was even more important. Our results furthermore indicate a growing need for flexibility in the spatiotemporal organization of sports activities and an increased importance of the public space for sports participation.

Keywords:

• sports participation
• constraints
• sports facilities
• spatial factors
• neighbourhood characteristics

View correction statement:

Correction

INTRODUCTION

The provision of a varied, safe and accessible sports infrastructure for all groups in society is a key policy focus in many Western countries in order to promote participation in sports and physical activity (Christiansen, Kahlmeier, & Racioppi, 2014; Ministerie van Volksgezondheid Welzijn en Sport, 2011; World Health Organization, 2011) . This policy objective is particularly challenging in the context of changes in the organization of sports facilities and sports activities in the Netherlands. As the market share of memberships of organized sports clubs is decreasing, participation in flexible forms of sports, often taking place in public space is increasing (Borgers, Pilgaard, Vanreusel, & Scheerder, 2016; Van der Roest, 2015). These developments raise questions about the importance of heavily subsidized official sports facilities for sports participation, relative to informal sports facilities such a public spaces (Hoekman, Breedveld, & Kraaykamp, 2015; Scheerder & Breedveld, 2004; Tiessen-Raaphorst, 2015).

The issue to what extent official and informal sports facilities stimulate participation in sports has been addressed from different angles. Most prominently, one strand of research has addressed effects of objectively measured accessibility and availability of sports facilities on sports participation and sports frequency (Hallmann, Wicker, Breuer, & Schönherr, 2012; O’Reilly, Berger, Hernandez, Parent, & Séguin, 2015; Wicker, Breuer, & Pawlowski, 2009). In addition, studies have shown the importance of social neighbourhood characteristics such as socio-economic status and safety in explaining differences in sports participation (Beenackers, Kamphuis, Burdorf, Mackenbach, & van Lenthe, 2011; Eime, Charity, Harvey, & Payne, 2015; Hoekman, Breedveld, & Kraaykamp, 2016; Kramer, Stronks, Maas, Wingen, & Kunst, 2015). However, as discussed in the next section, the effects of accessibility of sports facilities and neighbourhood characteristics that are found differ significantly between different geographical contexts.

An alternative approach to analyse participation in sports has focused on subjective factors such as the experience of constraints that withhold individuals from participation in sports (Lim et al., 2011). Constraints are experienced on multiple levels and may be intrapersonal (e.g. lack of skills or self-confidence), interpersonal (e.g. lack of social support) or structural (e.g. lack of appropriate accommodation) (Godbey, Crawford, & Shen, 2010). Those subjective constraints to sports participation have been found to play a significant negative role on sports participation and sports frequency (Alexandris & Carroll, 1997a; Casey et al., 2016; Downward, 2007).

While both objective spatial factors and subjective constraints may influence participation in sports, little is known about how they interact. First, insight is lacking how subjective constraints differ between places with different neighbourhood characteristics and if constraints are related to variations in individual travel distance to sports facilities. For instance, do participants living in neighbourhoods with specific socio-demographic composition experience more intra- or interpersonal constraints, and how is the experience of structural constraints related to travelled distance and type of sports facilities? Insight into such issues would facilitate the development of place specific policies aimed at relieving certain constraints. Second, insight is lacking into the importance of subjective constraints for sports participation relative to objective spatial factors. In particular, if constraints are experienced, do they indeed lead to lower levels of sports participation, and to a larger extent than objective spatial factors? Such insight is relevant when deciding whether policies targeting physical sports infrastructure or policies aiming at relieving constraints would be a better option to promote sports participation.

This paper addresses these research gaps by analysing the relative role of objective spatial factors, such as travelled distance to sports facilities and socio-spatial neighbourhood characteristics, and subjective constraints for sports participation based on a dataset of sports participation from the Netherlands. The paper investigates to what extent various types of constraints are experienced with respect to sports participation, how constraints are related to spatial factors as well as personal characteristics, and to what extent these constraints as well as spatial factors influence sports frequency. Given the increasing role of the public space as a facility for sports, we investigate how objective spatial factors and subjective constraints may have different effects on sports in official facilities (indoor-, outdoor facilities and swimming pools) than on sports in public space.

LITERATURE REVIEW AND THEORY

Distance to sports facilities, neighbourhood characteristics and sports participation

Despite theoretical agreement on the positive effect of sports facilities being available at an acceptable distance (Giles-Corti, Timperio, Bull, & Pikora, 2005; Sallis et al., 2006), empirical studies have shown mixed results regarding the influence of objectively measured accessibility and availability of sports infrastructure on sports participation, partly due to the use of different definitions (Hallmann et al., 2012; Hoekman et al., 2016; O’Reilly et al., 2015). Some studies have shown a positive association between the accessibility of sports facilities and sports participation in adults. However, differences were found between different types of sports (Eime et al., 2015; Hallmann et al., 2012; Karusisi, Thomas, Méline, & Chaix, 2013), age groups and gender (Limstrand & Rehrer, 2008; Wicker et al., 2009). Karusisi et al. (2013) found a significant association between access to and engagement in sports activities for swimming but not for team sports, racket sports or workouts in gyms in the Paris Ile-de-France region in France. Steinmayr, Felfe, and Lechner (2011) reported that distance to the nearest sports facility did not influence the engagement in sports activities of children in and outside sports clubs in larger towns and cities, but this distance did matter in smaller towns in the countryside in Germany. In contrast, Prins et al. (2011) found no association between the availability of sports facilities and sports participation in Dutch adolescents.

Besides the effect of travel distance to or availability of sports facilities, also other factors of the spatial environment do affect sports participation. For instance, various characteristics of the neighbourhood have been shown to influence sports participation behaviour. Recent studies in the Netherlands have shown that weekly sports participation rates in rural areas are higher than those in urban areas (Hoekman et al., 2016), although in rural areas individuals have to travel further, and the density of sports facilities available is lower (Hoekman et al., 2015). Hoekman et al. (2016) conclude that the variety and accessibility of sports facilities is hardly related to this ‘rural-urban’ divide in weekly sports participation, but instead is attributed to neighbourhood socio-economic status and safety. Although there is inconsistency in the literature about effects of perceptions of neighbourhood safety or crime on physical activity (Foster & Giles-Corti, 2008; Shenassa, Liebhaber, & Ezeamama, 2006; Van Dyck et al., 2011), several Dutch studies have confirmed a positive association of perceived neighbourhood social safety, social capital and higher socio-economic status with the likelihood of sports participation (Beenackers et al., 2011; Kamphuis et al., 2008). The association between perceived neighbourhood safety and sports participation was found for both ‘neighbourhood oriented’ sports and organized types of sports, although the effect was stronger for those who participated in ‘neighbourhood oriented’ sports (Beenackers et al., 2011). Another Dutch study also emphasized the importance of neighbourhood safety for participation in indoor sports clubs, suggesting the importance of distinguishing between various types of sports locations (Kramer et al., 2015).

To conclude, the literature suggests that travel distance to sports facilities may affect sports participation, but the extent to which this is the case seems to vary significantly between geographical settings and types of sports. However, it is still unclear what the impact is of the actual travel distance to different types of sports facilities for sports participants who already have chosen a particular sports location. Furthermore, objectively measured characteristics regarding the liveability of the neighbourhood may have a significant impact on sports frequency, potentially offsetting the impact of distance to facilities. However, to what extent the objective spatial factors of individual travel distance and neighbourhood characteristics affect subjective constraints, is still unknown.

Subjective constraints

A useful framework to study subjective constraints affecting sports participation is the hierarchical leisure constraints theory (Crawford & Godbey, 1987; Crawford, Jackson, & Godbey, 1991; Godbey et al., 2010). According to this theory, the decisions to undertake leisure activities—including sports participation—are influenced by three types of constraints: intrapersonal, interpersonal and structural constraints. These constraints are experienced in a sequential hierarchical order. Intrapersonal constraints are the most powerful or proximal, whereas structural constraints are the least powerful or distal (Crawford et al., 1991; Crawford & Godbey, 1987; Godbey et al., 2010). Intrapersonal constraints refer to individual physical or psychological constraints such as fatigue, health problems, self-confidence, stress or anxiety and to constraints regarding lack of skills or knowledge for sports participation. Interpersonal constraints relate to the constraints stemming from the dependency on other people, such as the inability to find a partner to take part in a sports activity. Structural constraints correspond to spatial determinants and include problems related to the accessibility of sports facilities, transport, financial resources and the quality of sports facilities or supplied activities but also time constraints (Alexandris & Carroll, 1999, 1997a, 1997b). Structural constraints thus refer to the perception of objective external issues such as travel time, availability of sports activities and time. Godbey et al. (2010) state that if intrapersonal constraints are not overcome, the desire or preference for a sports activity might not even come into being or will disappear or diminish. In turn, a higher demand to participate will increase structural constraints. The importance of constraint factors, however, varies depending on individual characteristics and context (Godbey et al., 2010). Furthermore, Jackson, Crawford, and Godbey (1993) suggest that leisure behaviour depends on the successful negotiation of constraints. Individuals use negotiation strategies such as time management, skill acquisition, interpersonal coordination and financial strategies to overcome constraints (Hubbard and Mannell (2001).

Empirical studies that apply the hierarchical leisure constraints theory to sports participation are scarce. Alexandris and Carroll (1997a) found in a study in Greece in which nonparticipants experienced significantly more intrapersonal constraints than participants. Constraints were experienced most frequently by women, lower educated people, the elderly and people with a minority background, indicating a relationship between the experience of constraints and social class (Alexandris & Carroll, 1997b; Casper, Bocarro, Kanters, & Floyd, 2011; Shores, Scott, & Floyd, 2007). Alexandris, Kouthouris, Funk, and Chatzigianni (2008) stressed the importance of intrapersonal constraints for involvement and loyalty among skiers. In addition, Alexandris (2013) found associations of intrapersonal constraints with lower levels of involvement and motivation in recreational tennis players. As far as we know, only one study investigated spatial aspects of leisure constraints. Jackson (1994) reported that geographical constraints (distinguished by a lack of opportunities near home, transportation costs and a lack of transportation) were less influential than other types of structural constraints (e.g. lack of time and quality of facilities) in starting a new leisure activity. However, a more explicit linking of subjective constraints to objective spatial factors, by specifying how constraints experienced by sports participants depends on these objective factors, is lacking to date.

Overall, this literature review suggests the importance of objectively measured spatial factors of travel distance to sports facilities and socio-spatial neighbourhood characteristics on the one hand, and subjective constraints at different intrapersonal, interpersonal and structural levels on the other, in explaining sports participation and frequency of participants. Because an integrative approach of spatial factors and constraints is still missing in empirical studies, our study investigates this interrelatedness by examining the conceptual relationships as illustrated in Figure 1. We address the following research questions:

1. To what extent are various subjective constraints experienced with respect to participation in sports, and does the experience of constraints depend on spatial factors and personal characteristics?

2. What is the impact of constraints on the actual sports frequency of participants, compared to spatial factors and personal characteristics?

3. To what extent are the effects of constraints and objective spatial factors different for official sports facilities (indoor facilities, outdoor facilities and swimming pools), as compared to sports in public space?

Fig. 1. Conceptual model

METHODS

Sample

The role of constraints and spatial factors is investigated based on an online survey conducted in September 2014 that recorded information about sports participation and sports facilities used, and was enriched with geodata. Data were collected in six municipalities in the Netherlands that varied in urban density levels: Amsterdam and Utrecht (highly urban and densely populated), Alphen aan den Rijn and Heerlen (mid-sized and moderately densely populated) and Berkelland and Roerdalen (small and rural). Eighteen thousand adults (3000 per municipality), aged 18–80 years old, were randomly drawn from municipal population registers and invited to fill out an online questionnaire by postal letter, sent from the corresponding municipalities. Complete questionnaires were obtained from 1663 respondents (9.2% response rate). From this sample, we included respondents who participated at least once a month in one vigorous type of sports during the past year (70.1%) and primarily used official indoor sports facilities (e.g. sport hall and gym), official outdoor sports facilities (e.g. soccer field, athletics track and tennis court), swimming pools or public spaces (park, forest, beach, road or sports court in the neighbourhood) for their sports activities. We excluded respondents engaging in non-active forms of sports, elite athletes and participants travelling more than 55 km to reach their sports activity. The final sample was comprised of 776 participants.

Measures

Sports frequency

Participants were asked about their average sports participation during the 12 months prior to the survey. The self-reported sports frequency of the main sports activity of the participants was measured as a categorical variable ranging from 1 to 3 times a month to at least 3 times a week.

Constraints to sports participation

We used the Leisure Constraints Scale to measure constraints to participate in sports. This scale was developed and applied to sports participation by Alexandris and Carroll (1997a, 1997b). Sports participants were asked to indicate the extent to which they had experienced the 29 formulated items as constraining factors when participating in their main type of sports during the past year using a 7-point Likert scale ranging from 1 (very unimportant) to 7 (very important; see Table 2). The ‘results’ section details how we distinguished between the following six factors of constraints: ‘physical/psychological’ and ‘skills/knowledge’ constraints (intrapersonal constraints), ‘partner’ constraints (interpersonal constraints), and ‘accessibility’, ‘time’ and ‘sports facility/supply’ constraints (structural constraints).

Objectively measured spatial factors

Individual travel distance to the main sports facility. Because we focus on sports participants who already have made a choice for a sports location, travel distance to the sports facility was expressed as the travelled distance between the respondent’s home and the sports facility that was principally used for the main sports activity of the respondent. It was objectively measured using the Model Builder Analyst Tool of Geographic Information Systems (GIS) software, ArcMap version 10.1 taking into account the transport mode (motorized or active transport) that was primarily used to reach the sports facility. The individual travel distance was categorized per type of sports facility (indoor, outdoor and swimming pools) into the following more or less equally sized groups: 1–1.7 km; 1.7–4.5 km; >4.5 km. In the analyses, these categories were included as dummy variables with public space users being the reference category.

Socio-spatial neighbourhood characteristics. Liveability variables on the 6-digit zip code level were obtained from the ‘Leefbaarometer’ (2012) (Liedelmeijer et al., 2008). The Leefbaarometer distinguishes the following six dimensions: composition and quality of the housing stock (in the rest of this paper referred to as ‘housing stock’), public space, facility level, demographics, social cohesion and safety. These dimensions were derived from 49 objectively measured items. Based on multivariate analysis of variance, we included only the separate liveability dimensions (measured as continuous scores) that were associated with sports frequency. These were housing stock (e.g. dominance of houses regarding type, price, date of construction, ownership, density and % social housing), demographics (e.g. % unemployed jobseekers, income, non-western immigrants and % highly educated) and safety (e.g. reported demolitions, crime and theft). We also added an indicator of nature (proximity to nature reserves, forests and coasts). Higher scores on these dimensions of liveability should be interpreted as better neighbourhood liveability. For instance, a higher score on safety means less reported demolitions, crime and theft.

Urban density levels are based on the average number of addresses within a radius of 1 km and were derived from address density classification data from Statistics Netherlands (2014). Residential locations were classified into five categories: rural (<500 addresses per km²), hardly urbanized (500–1000 addresses per km²), moderately urbanized (1000–1500 addresses per km²), strongly urbanized (1500–2500 addresses per km²) and extremely urbanized (>2500 per km²).

Control variables

We controlled for several individual characteristics. With regard to sports membership, we distinguished members of sports clubs or sports unions, participants in other organized forms (informal sports or running groups, healthcare or socio-cultural work), participants of gyms/fitness centres and unorganized participants (those who participate individually or together with friends, family or colleagues). The flexibility of conducting sports activities was measured to gain insight into the importance of flexibility in time allocation. ‘Type of athlete’ was measured by asking respondents the extent to which they saw themselves as recreational/novice or more competitive/experienced types of athletes.

Furthermore, we controlled for the following (dummy-coded) socio-demographic variables: gender, age, attained education level, self-reported health status, youngest child living at home (age categories < 4, 5–11 and 12–17 years old), living together with a partner and having a partner with a paid job. Self-reported health status was measured by asking respondents to indicate how they would describe their health. Individual (net) income level was excluded because of the large share of the sports participants (25.7%, N = 197) that answered ‘don’t know/I prefer not to mention’.

Analytical approach

First, we performed principal component analyses on the scores of the items of the Leisure Constraints Scale (Alexandris & Carroll, 1997a, 1997b) to identify underlying factors representing types of constraints. Because of theoretical assumptions about existing correlations between the constraint factors and the underlying dimensions (Godbey et al., 2010), we decided to use oblique rotation (direct oblimin). Second, hierarchical regression analyses were carried out to investigate how the constraint factors are explained by objectively measured spatial factors (model 1) and individual sports participation and socio-demographic characteristics as control variables (model 2). We used separate models to test the associations of the included variables with each of the six constraint factors as outcome measures. We tried to include the municipality of the respondents into the multivariate analyses. However, the use of dummy variables with one of the municipalities functioning as the reference category, made the results difficult to interpret. In addition, within some municipalities significant differences in urban density level do exist, which made it less meaningful to make statements on the municipal level. We therefore chose to use urban density level as a measure to differentiate between respondents’ residential locations. However, because of problems with the multicollinearity (VIF rate > 8) of the urban density level with scores on the liveability dimensions of safety and housing stock, we had to exclude urban density level from the regression analysis. Finally, ordinal regression analyses were performed to explain how the sports frequency of participants was determined by objectively measured spatial factors, individual sports participation and socio-demographic characteristics (model 1) and constraints (model 2).

RESULTS

Sample description

The sample included 776 sports participants, distributed across the municipalities of Roerdalen (22.3%), Alphen aan den Rijn (21%), Berkelland (17.9%), Heerlen (16.1%), Amsterdam (11.3%) and Utrecht (11.3%). The mean age was 50.6 years (SD = 15.6). As presented in Table 1, 54.3% were female, 52.4% had attained higher levels of education, and 39% lived in households with children. Of the original sample, 63.8% was participating at least once a week in sports, which is higher than the Dutch average (53%; GGD, CBS, RIVM, 2016). Working out individually in a gym was the most popular type of sports in the sample (17.8%), followed by running (14.9%) and cycling, race cycling or mountain biking (12.9%). Most participants practised their sports in public space (39.1%), followed by indoor sports facilities (37.5%), outdoor sports facilities (15.5%) and swimming pools (7.9%).

Table 1. Sample descriptions (N = 776)

	%
Age, %
18–25	7.3
26–45	29.1
46–65	44.1
66–80	19.5
Gender, % female	54.3
Household characteristics, %
Youngest child at home < 4 years old	5.7
Youngest child at home 5–11 years old	10.2
Youngest child at home 12–17 years old	8.3
Living together with partner	69.8
Employed partner	48.3
Education, %
Low	11.1
Middle	36.5
High	52.4
Weekly working hours, %
0–18 h	7.6
19–35 h	22.4
36–40 h	21
>41 h	12.5
Not working	36.5
Health status, %
Bad/very bad	2
Moderate	19.4
Good	64.9
Very good	13.7
Sports frequency, %
1–3 times a month	8.6
Once a week	31.4
Twice a week	33.2
At least 3 times a week	26.7
Sports facility, %
Indoor sports facility	37.5
Outdoor sports facility	15.5
Swimming pool	7.9
Public space	39.1
Travel distance to sports facility, Mean (SD)
Indoor	4.3 (5.1)
Outdoor	4.1 (5.0)
Swimming pool	4.3 (3.7)
Public space	0
Type of sports, %
Individual workout in gym	17.8
Running	14.9
Other individual sports	10.3
Team (ball) sports	9
Exercise classes and dance	5.5
Racket sports	11.5
Swimming	7.5
Cycling/Race cycling/mountain biking	12.9
Walking	10.6
Sports membership, %
Organized—member sports club/union	27.2
Organized—gym/fitness centre	19.7
Organized—other	5
Unorganized—group	22.9
Unorganized—individual	25.1
Type of athlete, %
Recreational—novice	18.8
Recreational—experienced	60.2
Competition—pleasure oriented	6.2
Competition—achievement oriented	6.4
Don’t know	8.4
Flexibility of sports activity, %
Fixed days or fixed times in week	23.7
Fixed days and times	49.5
Flexible	26.8

Constraint factors to sports participation

An initial principal component analysis resulted in five factors of constraints. However, the items concerning the accessibility of sports facilities and sports facility/supply characteristics were loaded onto the same factor, which hampered the interpretation. Therefore, analyses were re-run with a fixed number of factors (6). We deleted the items ‘I am not interested in my sport’, ‘I practised my sport but I did not like it’ and ‘I do not like social situations’ from further analysis because of low factor scores (<0.45). In the final analysis, six constraint factors remained: accessibility (4 items), physical/psychological (6 items), time (5 items), partners (3 items), skills/knowledge (3 items) and sports facility/supply (5 items), which accounted for 72.9% of the variance (Table 2).

Table 2. Principal component analysis of the constraints experienced by participants to participate in sport (N = 776), based on the Leisure Constraints Scale

Construct/item	Accessibility constraints	Physical/psychological constraints	Time constraints	Partner constraints	Skills/knowledge constraints	Sports facility/supply constraints	Total
Travelling takes too much time	0.85	–
I do not have access to transportation	0.82
I cannot afford it	0.75
There are no opportunities to participate in my sport near my home	0.75
Health problems are hindering me		0.85
I am not fit enough		0.81
I feel too tired to participate in my sport		0.64
Practicing my sport makes me feel tired		0.61
I am afraid of getting hurt		0.57
I do not feel confident		0.54
I experience time constraints due to work or study			0.9
I experience time constraints due to social commitments			0.88
I experience time constraints due to family			0.86
Participating in my sport does not fit with my timetable			0.62
Participating in my sport interrupts my daily routine			0.53
My friends do not like my sport				0.93
My friends do not have time to participate in my sport				0.93
I do not have anyone to participate with				0.91
I do not have anyone to teach me my sport					0.87
I am not skilled enough in my sport					0.82
I do not know where to participate					0.78
Sports facilities of my sport are poorly kept						0.93
Sports facilities of my sport are inadequate						0.9
Sports facilities of my sport do not meet my requirements						0.9
Sports facilities of my sport are too crowded						0.73
I do not like the supplied sports activities (trainings. lessons) of my sport						0.66
Eigenvalues	9.61	3.47	2.03	1.56	1.39	0.94
Explained variance, %	36.97	13.35	7.8	6	5.11	3.62
Explained variance, cumulative %	36.97	50.32	58.12	64.12	69.23	72.85
Cronbach’s alpha (based on standardized items)	0.89	0.82	0.87	0.92	0.91	0.93	0.93
Scale mean	2.25	2.5	2.83	2.1	1.8	2.62	2.35

Notes. Although the Leisure Constraints Scale of Alexandris and Carroll (1997a, 1997b) refers to ‘it’ and ‘sport’, we have chosen to refer to ‘my sport’. In the survey the actual name of the main type of sports of the respondent was filled in. Factor loadings < 0.5 are not displayed.

Accessibility constraints accounted for the largest share of the total variance (37.0%), whereas the sports facility/supply constraints accounted for only 3.6%. The eigenvalue of this last factor was 0.94, which is below Kaiser’s criterion of 1 (Stevens, 2002). However, we decided not to delete this factor because it was well interpretable and distinct from the other factors. To measure internal consistency for both the whole scale and each subscale, we estimated Cronbach’s alpha coefficients (Table 2). These ranged from α = 0.82 to α = 0.92, with α = 0.93 for the total scale, which is relatively high (Field, 2009). Mean scores per subscale were calculated, and they showed that time constraints were on average the most important constraints (M = 2.8), followed by constraints regarding sports facilities/supply (M = 2.6) and physical/psychological constraints (M = 2.5; Table 2).

Associations of spatial factors and individual characteristics with constraint factors

Results of hierarchical regression analyses (Table 3) showed that the baseline models, which included travel distance to sports facilities and neighbourhood liveability characteristics, turned out to be significant for the following subjective constraints: accessibility, physical/psychological, skills/knowledge and sports facilities/supply. Spatial factors contributed with a larger extent to explaining sports facility/supply (adj. R² = 0.06) and accessibility constraints (adj. R² = 0.04) than to physical/psychological (adj. R² = 0.02) and skills/knowledge constraints (adj. R² = 0.02). Adding the individual sports participation and socio-demographic control variables to the models led to a significant increase in the adjusted R²s of all models. Models for all constraint factors turned out to be significant. Time (adj. R² = 0.24), physical/psychological (adj. R² = 0.17) and to a lesser extent skills/knowledge constraints (adj. R² = 0.10) were well explained from both individual and spatial factors.

Table 3. Multiple linear regression on constraints factors (N = 763)

	Accessibility constraints		Physical/psychological constraints		Time constraints		Partner constraints		Skills/knowledge constraints		Sports facilities/supply constraints
	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2	Model 1	Model 2
	β (stand.)	β	β	β	β	β	β	β	β	β	β	β
Constant (B)	−0.2	−0.48	0.02	−0.71	0.15	−0.89	0.07	0.6	−0.07	0.34	−0.35	−0.01
Neighbourhood liveability
Housing stock	−0.13*	−0.14*	−0.01	0.03	−0.08	−0.06	0.1	0.11	0.14*	0.13*	−0.01	0.01
Demographics	−0.05	−0.01	−0.13**	−0.07	−0.05	−0.06	0.02	−0.01	0.11*	0.05	−0.07	−0.04
Safety	0.16*	0.1	0.05	−0.05	0.03	0.07	−0.17**	−0.12*	−0.23***	−0.15*	0.04	−0.02
Nature	−0.04	−0.02	−0.06	−0.05	0.02	0.03	0.04	0.04	0	−0.02	−0.10*	−0.08*
Sports facility × distance (ref = public space)
Indoor 0–1.7 km	0.07	0.01	−0.01	−0.06	−0.07	−0.08	0.04	0.02	0.03	0.01	0.15***	0.14**
Indoor 1.7–4.5 km	0.13**	0.07	0.09*	0.03	0.02*	0.00*	−0.06	−0.06	0	−0.03	0.15***	0.15**
Indoor > 4.5 km	0.17***	0.1*	0.04	−0.02	−0.08	−0.09	0.01	−0.02	0.02	−0.02	0.17***	0.17***
Outdoor 0–1.7 km	0	0	−0.07	−0.03	−0.07	0.01	−0.01	−0.03	0.01	−0.05	0.06	0.07
Outdoor 1.7–4.5 km	−0.05	−0.05	−0.03	−0.02	−0.02	−0.03	0.01	−0.01	0.09*	0.04	0.03	0.05
Outdoor > 4.5 m	0.06	0.06	−0.07	−0.05	0.01	0.03	0.02	−0.02	0.04	−0.02	0.06	0.08
Swimming pool 0–1.7 km	0	−0.02	0.02	0.01	−0.01	0.01	−0.01	−0.01	−0.04	−0.04	0.12**	0.10**
Swimming pool 1.7–4.5  km	0.10**	0.09*	−0.02	−0.03	−0.03	0.01	−0.03	−0.03	−0.06	−0.05	0.13***	0.12**
Swimming pool > 4.5 km	0.02	0	−0.04	−0.05	−0.03	0.01	0	0	−0.04	−0.03	0.05	0.04
Sports membership (unorganized individual = ref)
Organized—member sports club/union		0.04		0.03		−0.02		−0.11		0.02		−0.03
Organized—gym/fitness centre		0.11		0.07		0.09		−0.01		0.11*		−0.02
Organized—other		−0.01		−0.02		0		−0.06		0.03		−0.02
Unorganized—group		0.01		−0.03		−0.02		−0.1*		0.01		−0.03
Flexibility of sports activities (flexible = ref)
Fixed days or times in week		0.02		0.02		−0.07		−0.03		0		0.01
Fixed days and times in week		0.03		−0.02		−0.12**		0.07		−0.01		0.02
Type of athlete (competition—achievement oriented = ref)
Recreational—novice		−0.03		0.22**		0.2**		−0.16*		−0.22**		−0.14
Recreational—experienced		−0.03		0.13		0.19**		−0.11		−0.07		−0.20*
Competition—pleasure oriented		−0.03		0.09*		0.02*		−0.02		0		−0.16**
Don’t know		0.08		0.15**		0.1		−0.15*		−0.13*		−0.03
Gender (female = ref)
Male		0.01		−0.11**		−0.01		−0.05		0.04		−0.01
Age (66–80 = ref)
18–25		−0.03		0.04		0.14**		0.02		0.09		−0.11*
26–45		0.04		−0.05		0.14*		−0.04		0.03		−0.05
46–65		0.02		–0.06		0.03		0		0		0.01
Education (high = ref)
Low		0.15***		0.13		−0.01		−0.11**		−0.19***		0.09*
Middle		0.15***		0.09		−0.03		−0.08		−0.12**		0.07
Household characteristics
Youngest child at home < 4 years old (no child or youngest child at home > 4 years old = ref)		0.02		0.05		0.16***		−0.12**		0.01		0.06
Youngest child at home 5–11 years old (no child or youngest child at home > 12 years old = ref)		−0.01		0.05		0.11**		0.01		0.06		−0.02
Youngest child at home 12–17 years old (no child or youngest child at home > 18 years old = ref)		0.06		0.05		0.11**		−0.06		0		−0.02
Living together with partner (no partner/not living together with partner = ref)		0.02		0.01		0.01		−0.02		−0.05		0.03
Employed partner (no partner/partner is not employed = ref)		0.01		0.01		0.07		0.07		0.04		−0.06
Weekly working hours (no work = ref)
0–18		0.02		−0.07		0.12**		0.01		0.02		0.07
19–35		−0.04		−0.05		0.17***		0.04		−0.04		−0.01
36–40		−0.04		−0.01		0.14**		0.02		−0.03		−0.02
>41		−0.01		0.01		0.25***		0.05		−0.01		0.04
Health status (very good = ref)
Very bad/bad		0.05		0.2***		0.06		−0.02		−0.04		0.07
Moderate		0.03		0.35***		0.11*		−0.09		−0.07		0.02
Good		0.03		0.19***		0.11*		0.02		−0.02		0.01
Adjusted R^2	0.04	0.07	0.02	0.17	0.01	0.24	0	0.06	0.02	0.1	0.06	0.08
P model	<.001	<.001	<.01	<.001	0.163	<.001	0.248	<.001	<.01	<.001	<.001	<.001

*p < .05; **p < .01; ***p < .00.

Concerning accessibility constraints, the baseline model showed that participants using indoor sports facilities further than 1.7 km and swimming pools at a 1.7–4.5 km distance were more constrained than participants who used public space. Better neighbourhood safety, as well as a more liveable neighbourhood housing stock were associated with the experience of fewer accessibility constraints. Adding the individual control variables led to insignificant effects of neighbourhood safety and indoor facility users at a 1.7–4.5 km distance. Higher educated individuals experienced fewer accessibility constraints.

Physical/psychological constraints were initially negatively associated with neighbourhood demographics and positively associated with indoor sports facility users at a 1.7–4.5 km distance in the baseline model. After controlling for individual characteristics, these effects became insignificant. Male participants, competitive and achievement-oriented athletes, and participants with a very good health status experienced significantly fewer physical/psychological constraints.

Regarding time constraints, also after adjustment, only participants who used indoor sports facilities at a 1.7–4.5 km distance experienced more time constraints than participants who used public space. However, the second model showed that participants with young children living at home, younger people, participants who worked many hours, less competitive and less experienced athletes, and participants with a moderate and good health status, experienced more time constraints. In addition, the flexible scheduling of sports activities (the more fixed, the less constrained) was negatively associated with time constraints.

For partner constraints, we found, in both models, that a safer neighbourhood was associated with more partner constraints. Being a novice athlete, being a member of a gym/fitness centre, being lower educated and having a child younger than 4 years old were significantly associated with the experience of more partner constraints.

Concerning skills/knowledge constraints, the baseline model showed that neighbourhood liveability characteristics (housing stock, demographics and safety) and the use of outdoor sports facilities at a 1.7–4.5 km distance were associated with skills/knowledge constraints. After adding the individual control variables, only the effects of housing stock and safety remained significant. Both a higher score on housing stock liveability and a higher score on safety were associated with fewer skills/knowledge constraints. Lower and middle educated participants, novice recreational participants and participants visiting gyms/fitness centres experienced more skills/knowledge constraints.

Finally, indoor sports facility and swimming pool users that travelled less than 4.5 km experienced more sports facilities/supply constraints than public space participants. Furthermore, proximity to nature led to fewer constraints. Although decreasing, these effects remained significant after adjusting for individual factors. More recreationally oriented athletes and younger and higher educated people were less constrained with regard to the sports facilities/supply.

Associations of constraints and spatial factors with sports frequency

Table 4 shows the results of ordinal logit regressions in which spatial factors and constraint factors explain sports frequency. A baseline model with a Nagelkerke R² of 0.18 suggests that type of sports facility and travel distances of participants significantly influenced sports frequency (X² (41) = 139.99; p < .001). The model indicated that sports frequency was higher among outdoor sports facility and public space users. With regard to individual characteristics, we found that competitive achievement-oriented athletes participated more frequently in sports than less competitive/less experienced participants. Very good health status and being a non-organized individual participant were associated with a higher sports frequency. Working more than 41 h a week was associated with a lower sports frequency.

Table 4. Ordinal Regression on Sports frequency (N = 776)

	Model 1		Model 2
	Estimate	SE	Estimate	SE
Threshold parameters
Sports frequency (at least 3 times a week = ref)
1–3 times a month
Once a week
Twice a week
Sports facility × distance (public space = ref)
Indoor 0–1.7 km	−0.79**	0.27	−0.87**	0.28
Indoor 1.7–4.5 km	−1.11***	0.27	−1.12***	0.27
Indoor > 4.5 km	−0.65*	0.27	−0.83**	0.28
Outdoor 0–1.7 km	−0.13	0.36	−0.04	0.36
Outdoor 1.7–4.5 km	−0.11	0.4	−0.09	0.4
Outdoor > 4.5 m	−0.11	0.4	−0.09	0.4
Swimming pool 0–1.7 km	−0.65	0.53	−0.43	0.54
Swimming pool 1.7–.5 km	−1.07**	0.41	−1.06*	0.42
Swimming pool > 4.5 km	−1.71***	0.45	−1.60***	0.46
Neighbourhood liveability
Housing stock	0	0.01	0	0.01
Demographics	0	0.01	0	0.01
Safety	0	0	0	0
Nature	0	0.01	0	0.01
Constraints factors
Accessibility constraints			0.24*	0.1
Physical/psychological constraints			0.14	0.09
Time constraints			−0.47***	0.09
Partner constraints			0.17	0.09
Skills/knowledge constraints			0.1	0.09
Sports facility/supply constraints			−0.13	0.09
Sports membership (unorganized individual = ref)
Organized—sports club/union	−0.62*	0.27	−0.64*	0.28
Organized—gym/fitness centre	0.15	0.27	0.2	0.27
Organized—other	−0.63	0.36	−0.63	0.36
Unorganized—group	−0.72**	0.21	−0.68**	0.21
Flexibility of sports activity (flexible = ref)
Fixed days or times in week	0.24	0.21	0.18	0.21
Fixed days and times in week	0.36	0.19	0.25	0.19
Type of athlete (competition—achievement oriented = ref)
Don’t know	−2.00***	0.43	−1.94***	0.43
Recreational—novice	−2.01***	0.38	−1.86***	0.39
Recreational—experienced	−1.54***	0.35	−1.45***	0.36
Competition—pleasure oriented	−1.08**	0.41	−1.22**	0.42
Gender (female = ref)
Male	−0.27	0.16	−0.26	0.16
Age (66–80 = ref)
18–25	−0.32	0.37	−0.23	0.38
26–45	−0.58	0.3	−0.5	0.3
46–65	−0.17	0.24	−0.17	0.24
Education (high = ref)
Low	−0.35	0.24	−0.39	0.25
Middle	0.14	0.16	0.08	0.16
Household characteristics
Youngest child at home < 4 years old
(no child or youngest child at home > 4 years old = ref)	0.51	0.34	0.08	0.35
Youngest child at home 5–11 years old
(no child or youngest child at home > 12 years old = ref)	0.24	0.26	0.09	0.27
Youngest child at home 12–17 years old
(no child or youngest child at home > 18 years old = ref)	−0.2	0.27	−0.36	0.27
Living together with partner (no partner/not living together with partner = ref)	0.04	0.18	0.05	0.18
Employed partner (no partner/partner is not employed = ref)	0.01	0.17	0.05	0.18
Weekly working hours (no work = ref)
0–18	0.42	0.33	0.68*	0.33
19–35	−0.44	0.23	−0.23	0.23
36–40	−0.25	0.24	−0.06	0.24
>41	−0.56*	0.27	−0.21	0.28
Health status (very good = ref)
Very bad/bad	−1.95***	0.56	−1.93**	0.57
Moderate	−0.71**	0.25	−0.73**	0.27
Good	−0.71**	0.21	−0.75**	0.22
Nagelkerke R^2	0.18		0.24
P model	<.001		<.001
Pearson Chi-Square Model 1	2358.81*		2262.42
Test of Parallel Lines	Test of Parallel Lines
−2 Log Likelihood: 1673.55	−2 Log Likelihood: 1631.86
Chi-Square: 154.56	Chi-Square: 150.43
Sign: <.001	Sign: <.001

*p < .05; **p < .01; ***p < .0.

Finally, adding the six constraint factors to the model led to a significant increase in the Nagelkerke R² to 0.24 (X² (47) = 185,81; p < .001). Only time and accessibility constraints significantly influenced sports frequency, with time constraints showing the strongest (negative) effect on sports frequency. Accessibility constraints positively affected sports frequency. Working 0–18 h a week now turned out to have a positive effect on sports frequency. Although all other effects remained significant, they decreased slightly, excluding the effect of health status.

DISCUSSION

Interpretation of the main findings of spatial determinants of constraint factors

Our results showed that spatial factors were associated significantly with sports facility/supply and accessibility constraints and to a lesser extent with skills/knowledge constraints. Constraints concerning the sports facility/supply were significantly more experienced by users of indoor sports facilities and swimming pools (except for swimming pool users in the farthest distance category), compared to public space and outdoor facility users. This finding might be related to requirements regarding the quality of equipment, dressing rooms, atmosphere, crowdedness or opening hours that might be more critical to the perception of indoor sports facility and swimming pool users. For indoor sports facilities, our results showed an effect of distance: participants who travelled further to an indoor sports facility, have experienced more sports facility/supply constraints. Individuals might experience constraints regarding the quality of indoor sports facilities, such as higher costs and crowdedness, as more serious when they have to make a greater effort by travelling further. In addition, indoor sports facility users who travelled more than 4.5 km experienced significantly more accessibility constraints than public space participants. We noticed a similar trend for swimming pool users at a 1.7–4.5 km travel distance. These results confirm that travelling longer distances translates into experiencing accessibility constraints to a higher degree and that this occurs for relatively short distances.

Neighbourhood liveability characteristics also contributed, after adjusting for individual characteristics, to the explanation of constraint factors. Housing stock (e.g. dominance of houses regarding type, price, ownership and density) was significantly associated with accessibility and skills/knowledge constraints, indicating that participants living in a more liveable neighbourhood regarding the housing stock experienced significantly less accessibility and skills/knowledge constraints. Because the housing stock indicator includes several items including density, further research is needed to investigate which characteristics of the housing stock composition and quality are associated with accessibility and skills/knowledge constraints. Furthermore, our results showed that living in a safer neighbourhood was associated with the experience of fewer partner and skills/knowledge constraints. Because our data showed that a higher score on neighbourhood safety was correlated with rural environments, it is likely that it is less difficult for rural participants to find likeminded sports partners and sports opportunities in their neighbourhood. In addition, we found that proximity to nature reserves, forests or coasts was negatively associated with sports facility/supply constraints. People who live closer to natural environments will likely make more use of these natural environments for their sports activities instead of using sports facilities and experience no constraints in doing so.

Individual socio-demographic and sports participation characteristics were important confounders in explaining most constraints, which corresponds to findings reported by Godbey et al. (2010) and Shores et al. (2007). In particular, physical/psychological constraints and to a lesser extent partner and time constraints were not significantly associated with spatial determinants but could to a large degree be explained by self-reported health status, type of athlete and gender. Although all constraint factors were affected by individual characteristics, the extent differed per type of constraint. Specifically, type of athlete was an important confounder, except for accessibility constraints. Attained education level was particularly important for explaining accessibility, partner, skills/knowledge and sports facility/supply constraints. A possible explanation is that higher educated individuals have better access to social capital, which provides them with knowledge about sports activities and partners for sports participation (Coalter, 2007; Lindström, Hanson, & Ostergren, 2001; Wilson, 2002). Time constraints were associated with having children living at home and working hours but not with (disadvantaged) social class, which was also found in adolescents by Shores et al. (2007). Because these variables are highly significant and have relatively large effect sizes, these findings confirm outcomes of previous studies indicating the existence of time pressure for these groups of individuals in relation to participation in leisure activities (Ettema, Schwanen, & Timmermans, 2007; Bianchi & Mattingly, 2003; Crompton & Lyonette, 2006; Portegijs, Cloïn, Roodsaz, & Olsthoorn, 2016).

Interpretation of the main findings of effects on sports frequency

After adjusting for spatial and individual variables, we found that only time and accessibility constraints significantly accounted for the explanation of sports frequency. The strongest effect was found for time constraints, which affected sports frequency negatively. This major effect of time constraints is plausible because it is hard to make time for sports activities when handling busy work- and private-life schedules with multiple responsibilities and interests. Again, this finding confirms earlier studies indicating time pressure for these population segments (Ettema, Schwanen, & Timmermans, 2007; Bianchi & Mattingly, 2003; Crompton & Lyonette, 2006; Portegijs et al., 2016). Our results did not support the hierarchical proposition of the leisure constraints theory (Crawford et al., 1991; Crawford & Godbey, 1987; Godbey et al., 2010), which states that intrapersonal constraints (physical/psychological and skills/knowledge constraints), and subsequently interpersonal constraints (partner constraints), are more important in determining sports frequency than structural constraints (time, accessibility and sports facility/supply constraints). However, it is argued that time constraints can be internalized as intrapersonal constraints by participants, instead of structural constraints (Godbey et al., 2010; Shores et al., 2007). Participants might perceive their time constraints as personal, autonomous constraints, which might explain their effect on the frequency of sports participation. In addition, we have to note that our participants already participated in sports. Time constraints might have more effect on the frequency of engaging in sports activities than on the decision to participate in sports. Therefore, nonparticipants might experience a different hierarchy of importance of constraints. This is in line with Godbey et al. (2010), who state that constraints can be very personal and related to all types of individual characteristics, including stage and level of participation.

Unexpectedly, we found that participants who to a large extent experienced accessibility constraints had a higher sports frequency. It is likely that a higher sports frequency leads to both more travelling and more expenses at transport and sports activities. Experiencing more accessibility constraints might therefore not cause a higher sports frequency but rather be an effect of having a higher sports frequency. The successful use of negotiation strategies (e.g. managing time, combining sports with other activities or using public space to save travel time and costs) might explain why people who experienced more constraints may still participate and may actually participate more than people with fewer constraints (Hubbard & Mannell, 2001; Son, Mowen, & Kerstetter, 2008). Both the effects of time and accessibility constraints might apply even more strongly for participants combining multiple types of sports at multiple locations.

Regarding the relative influence of spatial factors on sports frequency, we found a significant effect only of travel distance to swimming pools. Participants who had to cover the biggest distance to their swimming pool, engaged less frequently in swimming. This finding is in line with the findings of Karusisi et al. (2013) and Wicker, Hallmann, and Breuer (2013), who reported positive significant associations between access to swimming pools and participation in swimming and general sports frequency in general. Furthermore, we found that travel distance to indoor sports facilities also negatively affected sports frequency, compared to participants using public space.

Despite the influence of liveability characteristics on several constraint factors, those neighbourhood characteristics did not significantly affect sports frequency. Additionally, we did not find any direct effect of individual socio-demographic characteristics of education, age, gender and household characteristics on sports frequency. Theoretically, it is possible that sports frequency is influenced only indirectly by individual characteristics, via subjective constraints. However, to investigate these indirect relations, further research is needed. The only indication we found for an indirect effect is the result that participants who had been working more than 41 h a week showed a significantly lower sports frequency. After adjusting for the constraint factors this effect decreased, showing an indirect effect of working hours via time constraints on sports frequency. Strikingly, adjustment for the constraint factors now led to a significant effect of working 0–18 h a week on sports frequency, compared to non-working. Furthermore, health status was an important determinant of physical/psychological constraints. However, the negative effect of health status on sports frequency, which was also found by Downward (2007), remained significant after adding the physical/psychological constraints and other constraint factors to the model. Apparently, sports frequency was determined not by the constraints with respect to sports participation but rather by the direct measurement of health status. It is possible that the constraints were experienced when participating in sports but did not play a role when deciding about the frequency of participating in sports.

Finally, our finding that a more competitive attitude or experience in sports was associated with a higher sports frequency seems to be related to the importance of motivation and negotiation strategies. However, we cannot explain why members of sports clubs and unofficial (self-organized) participants that engaged in sports with other(s) participated less frequent in sports than non-organized individual participants.

Strengths and limitations

This study is among the first to investigate the relative importance of subjective constraints, compared to objectively measured spatial factors and individual characteristics on sports frequency. By using a geographical approach and applying insights from leisure constraints theory studies, we contribute to a further understanding of the determinants of sports participation. Focusing on sports participants, rather than comparing participants and nonparticipants, allowed us to deepen our knowledge of variations in the constraints of participants with different frequencies of sports participation. In contrast to existing studies, we also included participants that participated in sports activities in public space. Furthermore, our study design allowed us to include respondents living in urban, moderately dense and rural municipalities.

A limitation of our study was that more than half of the sample (466 out of 776 respondents) was living in a neighbourhood with a very positive general liveability score. In addition, higher educated respondents were overrepresented, and respondents with a non-Dutch origin were underrepresented, leading to a sample of the population that was not fully representative. Although sports participation rates differed by municipality, these rates were largely in accordance with data on the population level, and the distribution of participants per education level corresponded to data on the national level. However, the two rural municipalities of Berkelland and Roerdalen had high response rates (probably due to a greater willingness to participate in the survey), with a relatively large amount of sports participants compared to averages at the municipal level. Furthermore, in this study we focussed only at constraints regarding the main sports activity of participants. The results were not controlled for a potential second sport of participants or for participation in the same sport in different geographical settings. Participation in multiple types of sports or the use of multiple locations may, however, have impact on the daily activity patterns, time use, the experience of constraints, and total sports frequency of participants. It may even have impact on the choice for a sports location and willingness to travel. Finally, no data was available on the travel distance of participants that use the public space for their sports. This is a limitation as not only neighbourhood parks or streets are used for sports participation, but also more remote locations such as natural areas or city parks. Also, whereas some start using the public space directly outside their front door (e.g. for running), others may deliberately travel to public spaces for specific forms of sports (e.g. team sports or skating). Further research should address the question which types of public spaces for different sports activities are used, and what this implies for travel distances.

CONCLUSIONS AND IMPLICATIONS

This study has shown that spatial factors, including both socio-spatial factors (e.g. neighbourhood liveability characteristics) and physical spatial factors (e.g. travel distance to sports facilities), are significantly associated with subjective constraints with regard to sports participation. Whereas the effect of distance seems logical and straightforward, further research is needed to assess how neighbourhood liveability characteristics such as social composition, safety and urban form are associated with constraints regarding acquiring knowledge and skills about sports facilities, finding partners, and the quality and accessibility of facilities.

In explaining the participants’ frequency of engaging in sports, time and accessibility constraints are the only constraint factors that have a significant effect on sports frequency. Travel distance to swimming pools and indoor sports facilities appear to affect sports frequency significantly. Furthermore, health status, type of athlete and to a lesser extent type of sports membership affects sports frequency. Time constraints appear to be a major limiting factor of sports participation. Geographical theories (Ettema, Schwanen, & Timmermans, 2007; Hägerstrand, 1970) suggest that the distance between locations plays a major role in the possibility of spending time participating in activities and therefore also affects time pressure. However, our results do not support this assumption because time constraints are not associated with travel distance to sports facilities. Apparently, in the Dutch context, with a high accessibility and density of facilities even in rural areas (Hoekman et al., 2015), distance is not decisive, whereas personal factors such as work hours and household responsibilities play a key role in the experience of time pressure. Unexpectedly, accessibility constraints were associated with a higher sports frequency. Because both time and accessibility constraints are regarded as structural constraints in the hierarchical leisure constraints perspective, we cannot support that intrapersonal constraints have the strongest effect on the sports frequency of participants despite theoretical assumptions. The positive effect of accessibility constraints on sports frequency makes it plausible that accessibility constraints are easier to overcome and negotiate than other constraints (including time constraints), particularly for participants who prioritize their sports activities. In addition, the effect of time constraints and other structural constraints, which is often related to a higher demand for activities, might be more important for adults who already engage in sports than for nonparticipants. It is also likely that the impact of constraints and distance differs per day and per participant, and that those factors are not always having impact on their sports frequency. However, for nonparticipants, the decision to participate in sports may be accompanied by personal reasons, and those intrapersonal constraints might be more difficult to overcome (Godbey et al., 2010).

Further research could focus on the interplay between direct and indirect effects of determinants of sports frequency, including motivation, negotiation techniques, constraints, and individual and spatial factors. In addition, further research is needed to interpret the influence of neighbourhood liveability characteristics on constraint factors. Another avenue for further research concerns objective and perceived neighbourhood characteristics of the sports (destination) location, as well as the travel route between home and the sports location. Furthermore, both participants and nonparticipants should be taken into account in future research while paying attention to people participating in multiple types of sports or using multiple sports locations. In particular, nonparticipants with an indicated intention to participate in sports are important from both a sport marketing and policy perspective.

Implications for managers

Our results have various implications for sports managers and policy makers. First, they suggest that the extent to which sports participants experience constraints may differ between neighbourhoods and type of sports facility. In majority, better liveability scores of a neighbourhood correspond with the experience of fewer constraints of all kinds. Even if these constraints do not translate into lower sports frequencies, this suggests that area specific interventions would be useful to lower these constraints in order to avoid drop out of sports on the longer term. Such interventions might be aimed at infrastructural improvements or modifications of the public space in case inhabitants face accessibility or sports facility constraints. If inhabitants are lacking skills or knowledge about where to participate in sports, solutions might be found in the social environment: for instance by the attainment of community sports coaches. A key characteristic of such an approach is that area specific interventions are needed based on the characteristics of the neighbourhoods involved. However, such approaches require a thorough investigation of constraints experienced by different groups of inhabitants (regarding for instance age, gender and ethnic background) on the neighbourhood level.

Second, our results suggest that travel distance and facility type have impact on the experience of accessibility and sports facility/supply constraints. In particular, users of indoor facilities and swimming pools at larger distances are associated with these constraints. Again, even if these constraints do not influence sports frequency directly, it implies that the quality of indoor facilities and distance remain a concern for policy makers and sports providers. Also, travel distance to these sports facilities was found to have a direct effect on sports frequency. Sports managers and urban planners may respond to this by facilitating sport facilities on strategic places such as work places, and take care of the accessibility of sports facilities, in particular by bicycle and car. For swimming pools, the effect of distance becomes increasingly important in the context of considerations about prospective closing down of swimming pools due to the population decline in (mostly) rural areas. Closing down swimming pools in such areas might result in excessive travel distances and increase the risk of drop out of participants in swim related sports, and for nonparticipants it might lead to even more constraints to participate in sports.

Third, we found that especially time constraints have a large impact on sports frequency, and that time constraints were associated with the number of working hours and having children. To promote sports participation among time pressed groups, sports providers and policy makers can respond to the increasing demand for flexibility in the temporal organization of sports activities and locations. For instance, by stimulating innovations and collaboration between traditional sports organizations (e.g. voluntary sports clubs), commercial flexible initiatives and societal initiatives. Relevant examples are ‘open club’ initiatives, whereby sports clubs are stimulated to increase their public role (Waardenburg, 2016), for instance by sharing (public) sports facilities with multiple sports providers or other parties, or by offering sports activities in the neighbourhoods and strengthening the relationship with neighbours.

Fourth, we found that recreational sports participants and people with poorer health appear to be vulnerable groups, as they experience more constraints and have lower sports frequencies. This suggests that sports managers might develop interventions to relieve constraints for these groups, for instance by investigating and developing forms of sports that reflect the needs of these participants.

Finally, our results stress the importance of the public space as a sports facility. With 39% of the participants primarily using public space for their sports activities, and public space being associated with less accessibility and supply related constraints and higher sport frequencies, public space is a potentially attractive sports facility for many, including both participants and nonparticipants. In addition, due to its flexibility in accessibility regarding time and costs, the public space is also popular among sports participants that combine multiple types of sports. Planners should therefore optimise the attractiveness of the public space for sports for different groups of participants. Yet, our insight in how to do this (and how participants are using the public space for sports) is only in its infancy. This calls for extensive research in this domain. In the context of the urgency of insight in this area, we believe that learning from planned interventions in the public space such as fitness equipment or bark running tracks (Borgers, Vanreusel, Vos, Forsberg, & Scheerder, 2016) and innovations such as real time and personalized feedback for recreational sports participations via smartphone apps, and connecting sports accommodations by attractive routes for runners, cyclists and hikers would be a fruitful approach. Collaboration between scientists, urban planners and sports managers in living labs might deliver both scientifically and policy relevant insights that lead to public spaces that are conducive to sports and physical activity.

Acknowledgements

We thank Hugo van der Poel and Remco Hoekman for their comments and discussions of earlier versions of this paper, which have significantly added to its quality. We also thank Sjors Hoek for helping with the GIS analysis.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCiD

Ineke Deelen http://orcid.org/0000-0001-7233-6054

Additional information

Funding

This work was supported by the Netherlands Organization for Scientific Research under NWO grant number 328-98-008.

Notes

This article was originally published with errors, which have now been corrected in the online version. Please see Correction (http://dx.doi.org/10.1080/23750472.2019.1565062)