Preventive Medicine Reports 36 (2023) 102402

Contents lists available at ScienceDirect

Preventive Medicine Reports

journal homepage: www.elsevier.com/locate/pmedr

A life course perspective on determinants of discontinuance of active

participation in sports activities
Xiaoyue Chen a, *, Astrid Kemperman a, Harry Timmermans a, b
a
Department of the Built Environment, Eindhoven University of Technology, Eindhoven, The Netherlands
b
Department of Air Transportation Management, Nanjing University of Aeronautics and Astronautics, Nanjing, China

A R T I C L E I N F O A B S T R A C T

Keywords: Physical inactivity remains a global public health challenge today. Determining why people stop regularly
Life course participating in sports is significant to develop targeted intervention strategies for sports promotion and healthy
Life transitions living. As sports participation is dynamic throughout life, a life-course perspective is needed to provide a more
Neighborhood characteristics
comprehensive understanding. This study adopts a life-course perspective to explore the determinants of the
Sports discontinuance
Retrospective longitudinal data
change from active participation in sports to becoming inactive. Based on online retrospective survey data
A two-level binary logistic regression collected in the Netherlands, a two-level binary logistic regression model is estimated to capture the effects of
socio-demographics, sports motivations, life transitions, and neighborhood characteristics on sports dropout over
the lifespan. Results show that dropout from sports is age-specific, and that people are less likely to discontinue
sports participation when they have health and weight loss goals. Life transitions have different effects. The
cessation of living with physically active people appears to be the most important event to make people stop
sporting, followed by having a baby, and then owning the first car. Compared with education-related events,
work-related events are more likely to cause people to stop sporting. Moreover, the probability of sports dis­
continuance may increase when residents feel unsafe doing physical activities in their neighborhoods or when
the neighborhood has sufficient greenspace for walking. The findings have implications for supporting sports
participants to continue exercising by addressing the barriers.

1. Introduction Temple, 2014; Deelen et al., 2018; Cannella et al., 2022). Specifically,
people, especially females and lower-educated adolescents, were
With great health benefits (Downward et al., 2016), participating in reportedly more likely to stop sporting in late adolescence (Lunn, 2010;
sports is a popular way of active living, particularly in the Netherlands. Prins et al., 2013). Also, loss of interest and enjoyment, poor health, and
In 2020, 55% of the Dutch population aged over 4 participated in sports lack of time can reduce sports participation (Farrell and Shields, 2002;
weekly, one of the highest levels of participation in the world (Duij­ Rottensteiner et al., 2013; Crane and Temple, 2014; Gardner et al., 2017;
vestijn et al., 2021). Despite the popularity of sports, high drop-out rates Jenkin et al., 2017; Jenkin et al., 2018; Jenkin et al., 2021; Pedersen
were observed, especially among adolescents (Deelen et al., 2018). et al., 2021). Another factor is the lack of social support, such as parental
Specifically, 70% of Dutch youths aged 12–17, 57% of adults aged objection and low peer acceptance (Rottensteiner et al., 2013; Gardner
18–64, and 30% of older adults weekly participated in sports (Duijves­ et al., 2017; Schlesinger et al., 2018; Vojvodić et al., 2020). Despite some
tijn et al., 2021). However, compared to numerous studies on starting evidence, existing studies are insufficient to explain sports discontinu­
sports, the determinants of stops are understudied. Undeniably, for ance. Some research gaps exist.
lifelong sports promotion, understanding why people quit sports is as Firstly, studies using long-term longitudinal data are lacking. Pre­
important as understanding why they start. vious studies are predominantly cross-sectional, typically ignoring the
Most previous studies on sports discontinuation focus on adolescents dynamic nature of sports behavior throughout life. Cross-sectional
and highlight the importance of intrapersonal (i.e., psychological and studies only focus on a specific time. It’s unclear whether that time is
social-demographic characteristics) and interpersonal (i.e., social in­ representative for causality analysis. Indeed, some studies are longitu­
teractions) factors (Calvo et al., 2010; Balish et al., 2014; Crane and dinal (e.g., Sarrazin et al., 2002; Jakobsson et al., 2012), but they mostly

* Corresponding author at: Groene Loper 6, Vertigo 8.09, P.O. Box 513, Eindhoven 5600 MB, The Netherlands.
E-mail address: [email protected] (X. Chen).

https://doi.org/10.1016/j.pmedr.2023.102402
Received 27 March 2023; Received in revised form 5 September 2023; Accepted 6 September 2023
Available online 12 September 2023
2211-3355/© 2023 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

consider short periods. Studies based on long-term longitudinal data are discontinuance over the life course? Habitual sports participation here
needed to develop lifelong strategies. refers to participation in any form of sports within the scope of the World
Secondly, studies based on a life-course perspective and focusing on Sports Encyclopedia (Lipoński, 2003) on a regular basis (at least once a
multiple life transitions are limited. The life-course perspective is an week and lasting more than 6 months). According to the World Sports
interdisciplinary approach to analyzing behaviors over several years Encyclopedia (Lipoński, 2003), sports are activities that require physical
across serval life stages or affected by several life transitions (Mayer, exertion and skills, have rules and guidelines, and can be competitive.
2009; Engberg et al., 2012). Although prior studies have recognized that Based on a life-course conceptual framework (Fig. 1), the determinants
time conflicts are important for the decision to quit sports (Rottensteiner for examination include socio-demographics, sports motivations, life
et al., 2013; Eime et al., 2015; Crane and Temple, 2014; Jenkin et al., transitions, and neighborhood characteristics. The results may help
2018; Jenkin et al., 2021; Pedersen et al., 2021), few studies have develop lifelong interventions to prevent or delay people from quitting
explored the relevant life transitions behind them (Van Houten et al., sports and promote active living.
2015). According to the life-course theory, life transitions, accompanied
by changes in personal and social status or identity as well as changes in 2. Method
living environments, may induce changes in sports behavior (Mortimer
and Shanahan, 2003). However, such studies are scarce, many of which 2.1. Data collection
are qualitative (Mayer, 2009). For example, via interviews, Haycock and
Smith (2014), Haycock and Smith (2018) found that life transitions Approved by the Ethical Review Board of the TU/e (Reference
impacted sports dropout through their influence on social inequality, ERB2019BE2), this study used an online retrospective survey to collect
Aarresola et al. (2017) reported that unexpected life events (e.g., illness) life course data and sports behaviors in the Netherlands from September
affected sports discontinuance, and Gatouillat et al. (2020) identified to October 2020. For decades, retrospective designs have been one of the
friends’ decision to stop, injury, and location changes as reasons for useful methods for collecting longitudinal data to analyze the relation­
teenagers to quit sports. ship between behaviors and life transitions, particularly in the case of
As for quantitative research on related topics, it is usually less in- long-term life spans (Scott and Alwin, 1998; Moschis, 2019).
depth or only focuses on specific periods (e.g., early adulthood). For With known outcomes, data on historical events, environments,
instance, Engel and Nagel (2011) speculated that sports discontinuance characteristics, and their changes are collected from records or mem­
might be influenced by life transitions such as childbirth and job changes ories (Scott and Alwin, 1998; Ranganathan and Aggarwal, 2018).
but had no further verification. Similarly, Chen et al. (2022) only dis­ Despite a risk of memory bias, the retrospective design is comparatively
cussed whether historical sports participation affected current partici­ inexpensive, less time-consuming, and can cover a longer time span
pation, ignoring the specific roles of life transitions. Additionally, Deelen (Scott and Alwin, 1998; Hirvensalo and Lintunen, 2011). Therefore, the
et al. (2018) focused on youths aged 13–21 and found that players who authors developed a retrospective survey to know when respondents
changed schools had a higher probability of discontinuing tennis. Also, changed their participation status in sports and whether these changes
concentrating on early adulthood, Van Houten et al. (2015), Van Houten co-occurred with life transitions.
et al. (2019) showed that respondents participated less frequently in
sports when stopping full-time education, beginning to work, entering 2.2. Survey design
an intimate relationship, and becoming a parent. Due to the dynamic
occurrence of life transitions, studies from a life-course perspective The survey consisted of two sections: life trajectory and history of
involving multiple life transitions and long-term spans are needed to participation in different forms of physical activity (sports being one of
better understand sports discontinuance. them and the only focus of this study). Fig. 2 contains details of the
Thirdly, neighborhood attributes and sports motivations that are survey questions.
important determinants of starting sports have received less attention in The questions in the life trajectory section were designed around the
explaining sports dropout. It should come as no surprise that a particular major transitions mentioned in the life course theory (Mortimer and
driver can start a behavior and also end it when a change occurs. More Shanahan, 2003). The investigated transitions included 1) birth year, 2)
potential determinants should be explored. Currently, existing research marital status changes, 3) baby birth, 4) relocation and associated
has yielded inconsistent results regarding the impact of neighborhood neighborhood characteristics, 5) education background, 6) employment
attributes. Some agreed that inadequate sports facilities or unsafe en­ history, 7) diagnosis of chronic diseases, 8) living arrangements with
vironments were significantly associated with less activity or dropout physically active people, and 9) car ownership. For each life transition,
(Panter et al., 2008; Crane and Temple, 2014; Halonen et al., 2015; respondents were asked to indicate whether they experienced this
Deelen, 2019; Pedersen et al., 2021). Conversely, others believed that particular event or to answer their current status. According to their
neither perceived safe nor the availability of parks and playgrounds responses, they were asked to detail the history of the event and asso­
significantly affected sports dropout (Vella et al., 2014; Deelen et al., ciated characteristics in chronological order from their birth year to the
2018). Therefore, more research is required to verify these findings. survey date. Additionally, the investigated neighborhood characteristics
Given the above gaps, this study contributes to the current literature of each home location included greenspace for walking, cycling facil­
from three aspects. First, this study has collected longitudinal data to ities, facilities that support physical activities, and safety for physical
explore sports discontinuance over a quite long period, from birth to the activities, which were asked via the five-point Likert scale. Respondents
survey date. Second, considering the dynamics of life trajectories, this were asked to compare each neighborhood they lived in with the
study examines the effects of multiple life transitions. Third, this study average-level Dutch neighborhood and then rate it as few (well below
hypothesizes that determinants of sports engagement, including neigh­ average), below average, average-level, above average, or a lot (well
borhood attributes and motivations for sporting, may also affect sports above average).
dropout for a more comprehensive understanding. As for questions on sports, respondents were asked whether they
Overall, this study aims to analyze the drivers of discontinuing participated regularly (at least once a week) in any form of sports for
habitual sports participation from a life-course perspective using retro­ more than six months. If so, they were asked for further details of each
spective longitudinal data. Specifically, three central questions are sport they played regularly, including start time, end time, sport type,
focused on: 1) How do major life transitions and associated character­ weekly activity hours, and motivations to start. The options for sport-
istics influence the decision to quit habitual sports participation? 2) How type questions were mainly based on the report “Zo Sport Nederland”
do neighborhood attributes affect sports discontinuance over the life (NOC*NSF, 2019).
course? 3) Do the motivations to start sporting also affect the The designed survey was created, run, and managed with

2
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Fig. 1. A life-course conceptual framework for this sports discontinuance research (Adult respondents, Netherlands, 2020–2023).

LimeSurvey, a professional online survey tool. To reduce the possible lead effects, this study assumed that the impact of a baby’s birth was
recall bias and improve data quality, the created online survey system within one year before or after, while for other events, six months. Thus,
mainly contained three types of error-checking functions to guarantee this study included both anticipatory and responsive behavior. Co-
consistency in responses: 1) temporal logic check, 2) event logic check, occurrence was assumed to indicate that the life event directly or indi­
and 3) consistency check. Fig. 3 depicts explanations and examples of rectly triggered the dropout. Along with the outcome variable, the status
these functions. of other explanatory variables was recorded simultaneously. That is, this
step generated observations at the timepoints of life transitions. Thirdly,
2.3. Samples and observations added to these processed data were the data reflecting other occasions of
discontinuing sports without the occurrence of a life event. That is, this
This investigation collaborated with PanelClix, a national survey step generated additional observations at the timepoints of sports dis­
company maintaining a representative panel of the Dutch population. continuance. In total, 5029 observations were used for further analysis,
PanelClix reaches their panelists in a targeted manner (representative 10.54% of which stopped active sports participation.
and stratified) based on the characteristics their panelists have provided
them. In this study, the qualifying panelists were representative of the 2.4. Explanatory variables
Dutch population over the age of 18. Totally, 627 respondents
completed the survey, 378 (60.29%) of which participated in sports This study considered four types of explanatory variables extracted
regularly for more than six months. Only the data from these 378 re­ from the literature on sports participation and dropout: (1) socio-
spondents were used for further analysis. demographics, including age, gender, marital status, education level,
Sports participation and life transitions are dynamic throughout weekly working hours, car ownership, child age, having chronic dis­
one’s life course and may occur multiple times. To this end, this study eases, and living with physically active people, (2) sports motivations,
integrated the sports participation data with the life-course data to including sports interests, health benefits, and weight control, (3) life
explore whether a respondent stopped sporting under the influence of transitions, including marital status change, baby birth, residential
events changes. Specifically, the process of data integration and obser­ relocation, changing schools, stopping full-time education, beginning to
vation generation mainly included three steps. Firstly, the sports work, changing jobs, having the first car, and no longer living with
participation data for each respondent were integrated to identify the physically active people, and (4) neighborhood characteristics,
periods during which the respondent played sports regularly. In the including greenspace for walking, cycling facilities, supportive facilities
survey, respondents were asked to review their sports participation by for physical activities, and safety for physical activities. Table 1 lists all
type. Thus, for respondents who had two or more experiences, their the explanatory variables. Age was divided into six categories based on
answers for the time periods might overlap partially or completely. the result of the decision tree at the observation level (Fig. 4). Neigh­
Since the outcome of this study was whether to stop sporting, it was borhood characteristics were reclassified into two or three levels based
necessary to integrate temporally overlapping sports participation data on similarity in response distributions. Effect coding was used to
for each respondent. Afterward, for each life event, it was checked represent categorical variables.
whether, during some specific period before and after the occurrence of
the event, the respondent stopped sporting. Given the possible lag and

3
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Fig. 2. An overview of life trajectory and sports behavior questions in the online retrospective survey (Adult respondents, Netherlands, September-October 2020).

4
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Fig. 3. An overview of error-checking functions in the online retrospective survey (Adult respondents, Netherlands, September-October 2020).

2.5. Data analysis ( )

Pij
log = β0 + β′1 xij + uj
This study applied a two-level binary logistic regression model to 1 − Pij
analyze the determinants of sports discontinuance. This model was
where β0 is the intercept when x = 0 and u = 0. xij is a vector of
chosen for two main reasons. First, the dependent variable of this study
was whether a respondent stopped active participation in sports. The explanatory variables at the observation-level. β′1 is a vector of param­
responses were binary. Second, this study measured the same respon­ eters to be estimated, representing the effects of explanatory variables.
dent at different time points. Observations from the same respondent uj is the random effects term, capturing the unobserved respondent-level
were less likely to be independent. A two-level binary logistic regression effects. It is assumed to be normally distributed with mean of zero.
model, which considers the dependency of the data by including a In the model-building process, the intraclass correlation coefficient
group-level (here, respondent-level) random component in the inter­ (ICC) is computed to indicate the proportion of the total variance in the
cept, is appropriate for this study. outcome variable explained by the grouping structure (Hox, 1995). The
The analysis process involved three steps. Firstly, the multilevel null equation for ICC is:
model with no explanatory variables was built to serve as a reference σ2u
point for comparison with the final model to see whether the random ICC =
σ 2u + (π2 /3)
effects at the group level (here, respondent level) significantly affect the
intercept of the dependent variable (here, sports dropout) at the
observation level. Secondly, explanatory variables were filtered via 3. Results
univariate analysis. Each explanatory variable was added to the null
model for separate analysis to exclude extremely irrelevant variables. 3.1. Characteristics of observations
Only variables with a p-value less than 0.15 were considered for the final
model (Bursac et al., 2008). Finally, all the filtered variables were used Table 1 indicates descriptive statistics of the observations. To better
as input for the two-level binary logistic regression model. R software understand the difference between the group that stopped sports and the
was used for estimation (R Core Team, 2022). group that didn’t stop, a comparison between the two groups was pre­
sented. As shown, the percentage of dropouts in the 13–19 group
Let yij denote the binary response for the ith observation of respon­
( (26.78%) was the highest. The higher the education level, the lower the
dent j (yij = 1 denotes the occurrence of sports dropout). Let Pij = Pr yij proportion of discontinuing sports. As for working hours, in 41.96% of
)
observations, people worked more than 20 h/w, and their sports dropout
= 1 . Then, the functional form of the two-level binary logistic regres­
rate was lower than those who worked less. The percentage of having a
sion model is: car was 41.82%, and the ratio of stopping sports to non-stop in these
observations was much lower. Besides, people whose children were all

5
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Table 1 over 12 or whose significant others were in poor health only accounted
Descriptive statistics of socio-demographic characteristics, sports motivations, for a small proportion of the observations, but their sports dropout rate
life transitions, and neighborhood characteristics (N = 5029 observations, was much higher.
Netherlands, September-October 2020). Regarding sports motivations, the most popular reason to start
Covariates Levels PCT Ratio [Stop sports sporting was health considerations (42.79%), followed by sports in­
(%) (%)/Non-stop (%)] terests (40.78%), and then weight control (11.25%). The proportion of
Socio-demographic Characteristics stopping sports among observations that desired to be healthier or
Age ≤12 12.87 0.07 [6.65/93.35] control weight was much lower than those without these desires.
13–19 26.78 0.17 [14.55/85.45] Additionally, changing schools was the most observed life event,
20–23 18.53 0.12 [10.73/89.27]
24–26 12.17 0.06 [5.39/94.61]
accounting for about 25% of the total, while stopping living with
27–30 10.18 0.09 [8.40/91.60] physically active people was the least, only accounting for 1.07%. For
>30 19.47 0.13 [11.75/88.25] each of the other life events, the observed event changes accounted for
Gender Female 49.55 0.12 [10.51/89.49] 7% to 16% of the total observations. The percentage of discontinuing
Male 50.45 0.12 [10.56/89.44]
sports was much higher among the observations that stopped living with
Marital status Have no partner 67.67 0.12 [10.81/89.19]
Have a partner 32.33 0.11 [9.96/90.04] physically active people. Conversely, for observations that had a change
Child age No child 78.78 0.12 [10.60/89.40] in other transitions, the rates of discontinuing sports were about 4%-8%,
≤12 17.78 0.11 [9.96/90.04] lower than those observed without these transitions.
>12 3.44 0.14 [12.14/87.86] Concerning neighborhood characteristics, statistical results indi­
Education level Primary education 17.26 0.15 [12.79/87.21]
cated that the percentage of discontinuing sports in “few cycling facil­
or lower
Secondary 45.54 0.12 [10.87/89.13] ities” was 17.74%, much higher than that in the other. By contrast, for
education greenspace and physical activity facilities, this percentage in “few green
Higher education 37.20 0.10 [9.09/90.91] spaces/facilities” was lower than the other. As for safety, “below
Working hours ≤20 h/w 58.04 0.14 [12.23/87.77]
average” had the highest ratio of stopping sports to non-stop. For the
>20 h/w 41.96 0.09 [8.20/91.80]
Car ownership No 58.18 0.13 [11.59/88.41] unmentioned variables, dropout rates within categories were similar,
Yes 41.82 0.10 [9.08/90.92] implying small effects.
Live with physically No 81.29 0.12 [10.76/89.24]
active people Yes 18.71 0.11 [9.56/90.44] 3.2. Results of variables filtering
Have chronic diseases No 90.81 0.12 [10.44/89.56]
(Self) Yes 9.19 0.13 [11.47/88.53]
Have chronic diseases No 88.03 0.11 [9.87/90.13] Table 2 presents the results of univariate analysis, indicating the
(Family) Yes 11.97 0.16 [13.79/86.21] relationship between each explanatory variable and sports dropout. To
exclude extremely irrelevant variables, the cut-off p-value was set at
Sports Motivations 0.15 (Bursac et al., 2008). Estimates showed that the p-values of gender,
Interest in sports No 59.22 0.11 [9.84/90.16] marital status, having chronic diseases (self), living with physically
Yes 40.78 0.13 [11.56/88.44] active people, sports interest, cycling facilities, and physical activity
Health benefits (Be No 57.21 0.13 [11.75/88.25]
facilities were higher than 0.15 and were therefore not considered in the
healthier) Yes 42.79 0.10 [8.92/91.08]
Weight control No 88.75 0.12 [10.87/89.13] subsequent analysis. Although variables including child age, having
Yes 11.25 0.09 [7.95/92.05] chronic diseases (family), living separately from physically active peo­
ple, greenspace for walking, and safety for physical activities were
Life Transitions insignificant at conventional levels, their p-values were less than 0.15,
Marital status change No 92.58 0.12 [11.02/88.98] meeting the selection criteria. The remaining variables’ p-values were
Yes 7.42 0.05 [4.56/95.44] less than 0.05, indicating that they were significant for discontinuing
Baby birth No 92.92 0.12 [10.83/89.17]
sports.
Yes 7.08 0.07 [6.74/93.26]
Residential relocation No 84.69 0.13 [11.69/88.31] Afterward, the filtered variables were tested for multicollinearity. As
Yes 15.31 0.04 [4.16/95.84] shown in Table 3, the variance inflation factors (VIFs) of all the variables
Change schools No 75.04 0.14 [12.06/87.94] were less than 5, indicating that the associated explanatory variable was
Yes 24.96 0.06 [5.98/94.02] lowly collinear with the other variables (Vittinghoff et al., 2012). There
Stop full-time education No 91.23 0.12 [11.03/88.97]
Yes 8.77 0.06 [5.44/94.56]
were no serious multicollinearity problems.
Begin to work No 91.47 0.12 [10.87/89.13]
Yes 8.53 0.08 [6.99/93.01] 3.3. Results of the two-level binary logistic regression model
Change jobs No 85.11 0.13 [11.40/88.60]
Yes 14.89 0.06 [5.61/94.39]
A two-level binary logistic regression model was applied to analyze
Have the first car No 93.00 0.12 [10.80/89.20]
Yes 7.00 0.08 [7.10/92.90] what triggered the discontinuance of sports participation. The log-
Stop living with No 98.93 0.12 [10.43/89.57] likelihood value at convergence of the final model was − 1228.1, and
physically active Yes 1.07 0.26 [20.37/79.63] the corresponding value for the null model without explanatory vari­
people
ables was − 1647.6. The likelihood ratio test value was 839.13, much
larger than the critical chi-square value with 25 degrees of freedom at
Neighborhood Characteristics any reasonable level of significance. Additionally, the ICC value of the
Greenspace for walking Few green spaces 5.29 0.08 [7.52/92.48]
null model indicated that 17.6% of the variance in observation sports
Other levels 94.71 0.12 [10.71/89.29]
Cycling facilities Few facilities 2.47 0.22 [17.74/82.26] discontinuation was attributable to between-respondent differences.
Other levels 97.53 0.12 [10.36/89.64] Therefore, the consideration of the effects of both observation and
Physical activity Few facilities 2.74 0.07 [6.52/93.48] respondent level was necessary. Table 4 describes the model results.
facilities Other levels 97.26 0.12 [10.65/89.35] The estimated parameters of sociodemographic variables showed
Safety for physical Below average 6.98 0.16 [13.68/86.32]
activities Other levels 93.02 0.11 [10.30/89.70]
that age, education level, and working hours had significant effects on
sports discontinuance. The estimated parameters of age were positive.
Compared with children under 12, young people aged 13–19 were most
likely to discontinue sports, followed by those aged 20–23. The negative

6
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Fig. 4. Decision tree results for the explanatory variable age (N = 5029 observations, Netherlands, September-October 2020).

parameter of education level indicated that people with a primary ed­ participation. The findings may support sports participants to continue
ucation or below were more likely to quit sports than those with a higher sporting by addressing their barriers over the life course.
education. As for working hours, sports dropouts were more likely to An important finding is that people with one or more of the following
occur when working hours were less than 20 h/w. According to the sociodemographic characteristics are more likely to stop sporting: being
estimated parameters of sports motivations, individuals with health or 13–23 years old, having a low education level, and working less than 20
weight loss aspirations were less likely to stop regularly sporting. h/w. It is consistent with previous research showing a high risk of sports
Moreover, all life events variables showed significant results, albeit discontinuance in late adolescence, especially among less educated
the estimated parameters were negative. Nevertheless, the estimated groups (Lunn, 2010; Prins et al., 2013). At the age of 13–19, the most
results still illustrated the different effects of various life events on sports likely period to quit sports, youths may face more activities and heavier
dropout. Specifically, the estimated value for stopping living with academic loads, which can cause difficulties in balancing study, work,
physically active people was the largest. That is, compared with other life, and sports (Crane and Temple, 2014; Deelen et al., 2018; Persson
life events, the discontinuance of sports participation was most likely to et al., 2020). This period also coincides with a high incidence of with­
occur when people stopped living with sporty people. Having a baby was drawal from school-organized sports activities, necessitating school-
in second place. The remaining life events affecting sports discontinu­ based interventions (Deelen et al., 2018; Espedalen and Seippel,
ance, in descending order, from largest to smallest, were having the first 2022). Additionally, this study suggests that people who start sporting
car, changing jobs, changing marital status, beginning to work, moving with health or weight loss goals are more likely to stick with it, reflecting
home, stopping full-time education, and changing schools. the importance of increased awareness of the benefits of sports for sports
Furthermore, neighborhood characteristics, including greenspace for promotion.
walking and safety for physical activities, were significant. The esti­ Moreover, this study displays different effects of various life transi­
mated value for greenspace indicated that people were less likely to stop tions on sports dropout. The end of living with physically active people
sporting when few green spaces were available in the living environ­ seems to be the most important event for discontinuation, followed by
ment. Oppositely, the estimation of safety suggested that people were having a baby. The result illustrates the significant impact of social
more likely to quit sports when they felt that their neighborhood was support. Social factors can be the driving force for sporting. For example,
unsafe for doing physical activities. people are more likely to join sports when their significant others do
(Farrell and Shields, 2002; Keresztes et al., 2008). Policymakers should
4. Discussion and conclusion act to prevent this force from disappearing. As for having a baby, quit­
ting sports may be due to sports participants taking on more family re­
This study applied a two-level binary logistic regression model to sponsibilities than usual. Active living consultation and services are
analyze the determinants of sports discontinuance, including socio- necessary for families with newborns. Compared with education-related
demographics, sports motivations, life transitions, and neighborhood events, work-related events are more likely to cause people to stop
characteristics, from a life course perspective. Data were collected in the sporting. It may hint that it is more difficult for people to prioritize
Netherlands through an online retrospective survey. Based on a life- sports on a day-to-day basis when entering a new working environment.
course framework, results show that life transitions and associated Or it may also be related to the location change. As transportation costs
socio-demographic and environmental characteristics affect sports dis­ increase or sports resources decrease, people are more likely to discon­
continuance differently. Also, this study implies that starting sports with tinue sports. Policymakers could consider developing work-based
different motivations may affect the persistence of habitual sports interventions.

7
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Table 2 Table 2 (continued )

Results of univariate analysis of all the explanatory variables by two-level binary Covariates Fixed effects Random
logistic regression models (N = 5029 observations, Netherlands, September- effects:σ2respondent
October 2020). Coeff. Sig.

Covariates Fixed effects Random Socio-demographic Characteristics

effects:σ2respondent Neighborhood Characteristics
Coeff. Sig.
Greenspace for Few green − 0.243 0.708
Socio-demographic Characteristics walking spaces (Ref.)
Age ≤12 (Ref.) − 0.408 0.754 Other levels 0.243 0.076
13–19 0.572 0.000 Cycling facilities Few facilities − 0.204 0.686
*** (Ref.)
20–23 0.156 0.156 Other levels 0.204 0.183
24–26 − 0.652 0.000 Physical activity Few facilities − 0.236 0.696
*** facilities (Ref.)
27–30 − 0.066 0.661 Other levels 0.236 0.224
>30 0.398 0.000 Safety for physical Below average 0.164 0.092 0.696
*** activities Other levels − 0.164
Gender Female (Ref.) − 0.022 0.702 (Ref.)
Male 0.022 0.746 Note: 1) Significance codes: p less than 0.001***; p less than 0.01**; p less than
Marital status No partner 0.012 0.700
0.05*. 2) The intercept for each model was omitted.
(Ref.)
Have a partner − 0.012 0.827
Child age No child (Ref.) − 0.117 0.717
≤12 − 0.176 0.140 Table 3
>12 0.293 0.100 Multicollinearity test results of the filtered explanatory variables (N = 5029
Education level Primary 0.334 0.709 observations, Netherlands, September-October 2020).
education or
Explanatory variables VIF
lower (Ref.)
Secondary 0.007 0.928 Age: 13–19 2.362
education Age: 20–23 1.644
Higher − 0.341 0.004** Age: 24–26 1.930
education Age: 27–30 2.156
Working hours ≤20 h/w (Ref.) 0.198 0.664 Age: >30 2.862
>20 h/w − 0.198 0.000 Child age: ≤12 4.609
*** Child age: >12 4.149
Car ownership No (Ref.) 0.122 0.689 Education: Secondary education 1.440
Yes − 0.122 0.022* Education: Higher education 1.571
Live with No (Ref.) 0.072 0.702 Working hours: >20 h/w 1.888
physically Yes − 0.072 0.322 Car ownership: Yes 1.717
active people Have chronic diseases (Family): Yes 1.037
Have chronic No (Ref.) − 0.016 0.700 Health benefits (Be healthier): Yes 1.050
diseases (Self) Yes 0.016 0.872 Weight control: Yes 1.051
Have chronic No (Ref.) − 0.168 0.687 Marital status change: Yes 1.288
diseases Yes 0.168 0.059 Baby birth: Yes 1.913
(Family) Residential relocation: Yes 1.575
Change schools: Yes 2.626
Stop full-time education: Yes 1.511
Sports Motivations
Begin to work: Yes 1.553
Interest in sports No (Ref.) − 0.088 0.690
Change jobs: Yes 1.776
Yes 0.088 0.168
Have the first car: Yes 1.526
Health benefits No (Ref.) 0.247 0.728
Stop living with physically active people: Yes 1.050
(Be healthier) Yes − 0.247 0.000***
Greenspace for walking: Below average and above 1.114
Weight control No (Ref.) 0.264 0.718
Safety for physical activities: Below average 1.102
Yes − 0.264 0.016*

Life Transitions Besides, people are more likely to stop sporting in unsafe neighbor­
Marital status No (Ref.) 0.497 0.710 hoods, but less likely to stop sporting in neighborhoods with insufficient
change Yes 0.497 0.000***
greenspace. Consistent with many previous studies (e.g., Deelen, 2019),
−
Baby birth No (Ref.) 0.288 0.711
Yes − 0.288 0.011* safety issues may reduce people’s sports participation as they instinc­
Residential No (Ref.) 0.598 0.718 tively avoid dangers. Regarding greenspace, it can be explained that
relocation Yes − 0.598 0.000*** neighborhoods without sufficient greenspace usually provide residents
Change schools No (Ref.) 0.412 0.708
with fewer opportunities for other recreational activities (Sugiyama
Yes − 0.412 0.000***
Stop full-time No (Ref.) 0.413 0.718 et al., 2013; Feng et al., 2021). It may partly result in residents regularly
education Yes − 0.413 0.000*** playing sports at sports centers and clubs. Combined with other con­
Begin to work No (Ref.) − 0.100 0.702 clusions of this study, neighborhood-based strategies should be devel­
Yes 0.100 0.015* oped according to different environmental characteristics and
Change jobs No (Ref.) 0.402 0.706
Yes − 0.402 0.000***
population composition.
Have the first car No (Ref.) 0.257 0.708 Note that this study has some limitations. First, this study collected
Yes − 0.257 0.020* life-course and sports behavior data retrospectively. Although the
Stop living with No (Ref.) − 0.308 0.692 possible bias caused by recall has been minimized, the risk is inevitable.
physically Yes 0.308 0.098
Second, respondents were asked to assess neighborhood characteristics
active people
by comparison with a typical Dutch neighborhood. It may be a bit
general since the Netherlands has many “average-level” neighborhoods.
Such self-reported assessments may slightly reduce the sensitivity of the

8
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Table 4 understand the discontinuance of habitual sports participation. It may

Results of the multivariate analysis of the filtered explanatory variables by the assist multidisciplinary practitioners in sports research and practice to
two-level binary logistic regression model (N = 5029 observations, Netherlands, create a more supportive social and physical environment for active
September-October 2020). living. This study suggests that the dynamic nature of sports behavior
Covariates Coeff. Sig. and the role of life transitions deserve more attention for theoretical
Fixed effects development. Also, more potential determinants, such as those to start
sports, may have been overlooked. Exploring new determinants remains
Socio-demographic Characteristics
crucial to enrich existing theories. Since the decision to join or quit
Age ≤12 (Ref.) − 0.929 sports is complex, dynamic, and even personal, policies and in­
13–19 0.816 0.000***
terventions should evolve to be dynamic, lifelong, and tailored.
20–23 0.448 0.001**
24–26 − 0.274 0.156
27–30 − 0.008 0.969 CRediT authorship contribution statement
>30 − 0.054 0.766
Child age No child (Ref.) − 0.038
Xiaoyue Chen: Conceptualization, Methodology, Investigation,
≤12 0.117 0.469
>12 − 0.079 0.729 Data curation, Software, Writing – original draft. Astrid Kemperman:
Education level Primary education or 0.701 Conceptualization, Methodology, Investigation, Writing – review &
lower (Ref.) editing, Supervision. Harry Timmermans: Conceptualization, Meth­
Secondary education − 0.189 0.072
odology, Investigation, Writing – review & editing, Supervision.
Higher education (Ref.) − 0.511 0.003**
Working hours ≤20 h/w (Ref.) 0.169
>20 h/w − 0.169 0.043*
Declaration of Competing Interest
Car ownership No (Ref.) 0.165
Yes − 0.165 0.054
Have chronic diseases No (Ref.) − 0.202 The authors declare that they have no known competing financial
(Family) Yes 0.202 0.070 interests or personal relationships that could have appeared to influence
the work reported in this paper.
Sports Motivations
Health benefits (Be No (Ref.) 0.256 Data availability
healthier) Yes − 0.256 0.002**
Weight control No (Ref.) 0.325
Yes − 0.325 0.018* The data that has been used is confidential.

Life Transitions References

Marital status change No (Ref.) 1.651
Yes − 1.651 0.000*** Aarresola, O., Itkonen, H., Laine, K., 2017. Young athletes’ significant experiences in
Baby birth No (Ref.) 1.542 sport: critical sociological reflections on athlete development. European Journal for
Yes − 1.542 0.000*** Sport and Society 14, 265–285. https://doi.org/10.1080/16138171.2017.1349067.
Balish, S.M., McLaren, C., Rainham, D., Blanchard, C., 2014. Correlates of youth sport
Residential relocation No (Ref.) 1.839
attrition: A review and future directions. Psychology of Sport and Exercise 15,
Yes − 1.839 0.000***
429–439. https://doi.org/10.1016/j.psychsport.2014.04.003.
Change schools No (Ref.) 1.943
Bursac, Z., Gauss, C.H., Williams, D.K., Hosmer, D.W., 2008. Purposeful selection of
Yes − 1.943 0.000*** variables in logistic regression. Source Code for Biology and Medicine 3, 17. https://
Stop full-time education No (Ref.) 1.890 doi.org/10.1186/1751-0473-3-17.
Yes − 1.890 0.000*** Calvo, T.G., Cervelló, E., Jiménez, R., Iglesias, D., Murcia, J.A.M., 2010. Using self-
Begin to work No (Ref.) 1.691 determination theory to explain sport persistence and dropout in adolescent athletes.
Yes − 1.691 0.000*** The Spanish Journal of Psychology 13, 677–684. https://doi.org/10.1017/
Change jobs No (Ref.) 1.612 S1138741600002341.
Yes − 1.612 0.000*** Cannella, V., Serrat, R., Villar, F., 2022. “Until Body Do Us Part”: Barriers to retention in
Have the first car No (Ref.) 1.574 competitive sports among European older athletes. Top. Geriatr. Rehabil. 38, 65–72.
Yes − 1.574 0.000*** https://doi.org/10.1097/TGR.0000000000000344.
Stop living with physically No (Ref.) 0.877 Chen, G., Janke, M.C., Liechty, T., Wong, J.D., West, S.T., Son, J.S., Juris, J.,
active people Yes − 0.877 0.000*** Johnston, C., 2022. Sport participation for adults aged 50 + years: A socioecological
analysis. International Journal of Aging & Human Development 97, 354–373.
https://doi.org/10.1177/00914150221143958.
Neighborhood Characteristics Crane, J., Temple, V., 2014. A systematic review of dropout from organized sport among
Greenspace for walking Few green spaces (Ref.) − 0.339 children and youth. Eur. Phys. Educ. Rev. 21, 114–131. https://doi.org/10.1177/
Other levels 0.339 0.041* 1356336X14555294.
Safety for physical activities Below average 0.422 0.000*** Deelen, I., Ettema, D., Kamphuis, C.B.M., 2018. Time-use and environmental
determinants of dropout from organized youth football and tennis. BMC Public
Other levels (Ref.) − 0.422
Health 18, 1022. https://doi.org/10.1186/s12889-018-5919-2.
Constant − 14.527 0.000***
Deelen J.C.C., 2019. Space for sport: The interplay between the physical environment,
intrapersonal factors and sports participation. Utrecht University Repository,
Random effects Utrecht. https://dspace.library.uu.nl/handle/1874/380195.
Downward, P., Dawson, P., Mills, T.C., 2016. Sports participation as an investment in
σ2respondent : 1.224
(subjective) health: a time series analysis of the life course. J. Public Health
ICCfinal : 0.271 (Bangkok). 38, e504–e510. doi: 10.1093/pubmed/fdv164.
Marginal R2 /Conditional R2 : 0.344/0.522 Duijvestijn, M., Schurink, T., van den Berg, S., Euser, S., Wendel-Vos, W., 2021. Sport
AIC/BIC: 2510.1/2686.2 and exercise behavior report in 2020. RIVM. https://www.sportenbewegenincijfers.
nl/documenten/rapport-sport-en-beweeggedrag-in-2020 (accessed 1 November
Note: Significance codes: p < 0.001***; p < 0.01**; p < 0.05*. 2022).
Eime, R.M., Casey, M.M., Harvey, J.T., Sawyer, N.A., Symons, C.M., Payne, W.R., 2015.
Socioecological factors potentially associated with participation in physical activity
analysis. Additionally, limited by the sample size, this study didn’t and sport: A longitudinal study of adolescent girls. Journal of Science and Medicine
further subdivide life transitions when analyzing their relationship with in Sport 18, 684–690. https://doi.org/10.1016/j.jsams.2014.09.012.
sports discontinuance. Further research is needed to gain insights into Engberg, E., Alen, M., Kukkonen-Harjula, K., Peltonen, J.E., Tikkanen, H.O.,
Pekkarinen, H., 2012. Life events and change in leisure time physical activity. Sport.
subcategories of life transitions. Med. 42, 433–447. https://doi.org/10.2165/11597610-000000000-00000.
Despite these limitations, this study expands the evidence base to Engel, C., Nagel, S., 2011. Sports participation during the life course. European Journal
for Sport and Society 8, 45–63. https://doi.org/10.1080/16138171.2011.11687869.

9
X. Chen et al. Preventive Medicine Reports 36 (2023) 102402

Espedalen, L.E., Seippel, Ø., 2022. Dropout and social inequality: Young people’s reasons International Publishing, Cham, pp. 111–131. https://doi.org/10.1007/978-3-030-
for leaving organized sports. Ann. Leis. Res. 1–18 https://doi.org/10.1080/ 05008-5_6.
11745398.2022.2070512. *NOC*NSF, 2019. De ledentallenrapportage NOC*NSF 2018. NOC*NSF. https://nocnsf.
Farrell L., Shields M.A., 2002. Investigating the economic and demographic determinants nl/media/1080/ledentalrapportage-nocnsf-2018-sportonderzoek.pdf (accessed 10
of sporting participation in England. J. R. Stat. Soc. Ser. A Stat. Soc. 165, 335–348. July 2023).
doi: 10.1111/1467-985X.00626. Panter, J., Jones, A., Hillsdon, M., 2008. Equity of access to physical activity facilities in
Feng, X., Toms, R., Astell-Burt, T., 2021. Association between green space, outdoor an English city. Prev. Med. (Baltim) 46, 303–307. https://doi.org/10.1016/j.
leisure time and physical activity. Urban Forestry & Urban Greening 66, 127349. ypmed.2007.11.005.
https://doi.org/10.1016/j.ufug.2021.127349. Pedersen, M.R.L., Hansen, A.F., Elmose-Østerlund, K., 2021. Motives and barriers related
Gardner, L.A., Magee, C.A., Vella, S.A., 2017. Enjoyment and behavioral intention to physical activity and sport across social backgrounds: Implications for health
predict organized youth sport participation and dropout. J. Phys. Act. Heal. 14, promotion. International Journal of Environmental Research and Public Health 18
861–865. https://doi.org/10.1123/jpah.2016-0572. (11), 5810.
Gatouillat, C., Griffet, J., Travert, M., 2020. Navigating the circles of social life: Persson, M., Espedalen, L.E., Stefansen, K., Strandbu, Å., 2020. Opting out of youth
Understanding pathways to sport drop-out among French teenagers. Sport, sports: How can we understand the social processes involved? Sport, Education and
Education and Society 25, 654–666. https://doi.org/10.1080/ Society 25, 842–854. https://doi.org/10.1080/13573322.2019.1663811.
13573322.2019.1643300. Prins, R.G., Kamphuis, C.B.M., van Empelen, P., Beenackers, M.A., Brug, J.,
Halonen, J.I., Stenholm, S., Kivimäki, M., Pentti, J., Subramanian, S.V., Kawachi, I., Mackenbach, J.P., Oenema, A., 2013. Explaining socio-demographic differences in
Vahtera, J., 2015. Is change in availability of sports facilities associated with change disengagement from sports in adolescence. European Journal of Public Health 23,
in physical activity? A prospective cohort study. Prev. Med. (Baltim) 73, 10–14. 811–816. https://doi.org/10.1093/eurpub/cks188.
https://doi.org/10.1016/j.ypmed.2015.01.012. R Core Team, 2022. R: A language and environment for statistical computing. R
Haycock, D., Smith, A., 2014. A family affair? Exploring the influence of childhood sport Foundation for Statistical Computing, Vienna, Austria https://www.R-project.org/.
socialisation on young adults’ leisure-sport careers in north-west England. Leisure Ranganathan, P., Aggarwal, R., 2018. Study designs: Part 1–An overview and
Studies 33, 285–304. https://doi.org/10.1080/02614367.2012.715181. classification. Perspectives in Clinical Research 9, 184–186. https://doi.org/
Haycock, D., Smith, A., 2018. Adult sport participation and life transitions: The 10.4103/picr.PICR_124_18.
significance of childhood and inequality. In: Dionigi, R.A., Gard, M. (Eds.), Sport and Rottensteiner, C., Laakso, L., Pihlaja, T., Konttinen, N., 2013. Personal reasons for
Physical Activity Across the Lifespan: Critical Perspectives. Palgrave Macmillan UK, withdrawal from team sports and the influence of significant others among youth
London, pp. 195–210. https://doi.org/10.1057/978-1-137-48562-5_10. athletes. International Journal of Sports Science and Coaching 8, 19–32. https://doi.
Hirvensalo, M., Lintunen, T., 2011. Life-course perspective for physical activity and org/10.1260/1747-9541.8.1.19.
sports participation. European Review of Aging and Physical Activity 8, 13–22. Sarrazin, P., Vallerand, R., Guillet, E., Pelletier, L., Cury, F., 2002. Motivation and
https://doi.org/10.1007/s11556-010-0076-3. dropout in female handballers: a 21-month prospective study. European Journal of
Hox, J. J., 1995. Applied Multilevel Analysis. TT-publikaties, Amsterdam. http:// Social Psychology 32, 395–418. https://doi.org/10.1002/ejsp.98.
joophox.net/publist/amaboek.pdf. Schlesinger, T., Löbig, A., Ehnold, P., Nagel, S., 2018. What is influencing the dropout
Jakobsson, B.T., Lundvall, S., Redelius, K., Engström, L.-M., 2012. Almost all start but behaviour of youth players from organised football?Wodurch wird das Drop-out-
who continue? A longitudinal study of youth participation in Swedish club sports. Verhalten junger Spieler im organisierten Fußball beeinflusst?: Eine systematisierte
Eur. Phys. Educ. Rev. 18, 3–18. https://doi.org/10.1177/1356336X11430660. Übersicht. Ger. J. Exerc. Sport Res. 48 (2), 176–191.
Jenkin, C.R., Eime, R.M., Westerbeek, H., O’Sullivan, G., van Uffelen, J.G.Z., 2017. Sport Scott, J., Alwin, D., 1998. Retrospective versus prospective measurement of life histories
and ageing: A systematic review of the determinants and trends of participation in in longitudinal research. In: Giele, J.Z., Elder Jr., G.H. (Eds.), Methods of Life Course
sport for older adults. BMC Public Health 17, 976. https://doi.org/10.1186/s12889- Research: Qualitative and Quantitative Approaches. SAGE Publications Inc,
017-4970-8. pp. 98–127. https://doi.org/10.4135/9781483348919.
Jenkin, C.R., Eime, R.M., Westerbeek, H., van Uffelen, J.G.Z., 2018. Sport for adults aged Sugiyama, T., Giles-Corti, B., Summers, J., du Toit, L., Leslie, E., Owen, N., 2013.
50+ years: Participation benefits and barriers. Journal of Aging and Physical Initiating and maintaining recreational walking: A longitudinal study on the
Activity 26, 363–371. https://doi.org/10.1123/japa.2017-0092. influence of neighborhood green space. Prev. Med. (Baltim) 57, 178–182. https://
Jenkin, C.R., Eime, R.M., van Uffelen, J.G.Z., Westerbeek, H., 2021. How to re-engage doi.org/10.1016/j.ypmed.2013.05.015.
older adults in community sport? Reasons for drop-out and re-engagement. Leisure Van Houten, J.M.A., Kraaykamp, G., Breedveld, K., 2015. When do young adults stop
Studies 40, 441–453. https://doi.org/10.1080/02614367.2021.1888310. practising a sport? An event history analysis on the impact of four major life events.
Keresztes, N., Piko, B.F., Pluhar, Z.F., Page, R.M., 2008. Social influences in sports International Review for the Sociology of Sport 52, 858–874. https://doi.org/
activity among adolescents. The Journal of the Royal Society for the Promotion of 10.1177/1012690215619204.
Health 128, 21–25. https://doi.org/10.1177/1466424007085228. Van Houten, J.M.A., Kraaykamp, G., Pelzer, B.J., 2019. The transition to adulthood: a
Lipoński W., 2003. World Sports Encyclopedia. MBI, Oficyna Wydawnicza Atena, St. game changer!? A longitudinal analysis of the impact of five major life events on
Paul. sport participation. European Journal for Sport and Society 16, 44–63. https://doi.
Lunn, P.D., 2010. The sports and exercise life-course: A survival analysis of recall data org/10.1080/16138171.2019.1603832.
from Ireland. Social science and medicine 70, 711–719. https://doi.org/10.1016/j. Vella, S.A., Cliff, D.P., Okely, A.D., 2014. Socio-ecological predictors of participation and
socscimed.2009.11.006. dropout in organised sports during childhood. International Journal of Behavioral
Mayer, K.U., 2009. New directions in life course research. Annual Review of Sociology Nutrition and Physical Activity 11, 62. https://doi.org/10.1186/1479-5868-11-62.
35, 413–433. https://doi.org/10.1146/annurev.soc.34.040507.134619. Vittinghoff, E., Glidden, D.V., Shiboski, S.C., McCulloch, C.E., 2012. Regression Methods
Mortimer, J.T., Shanahan, M.J., 2003. Handbook of the life course. Springer, New York. in Biostatistics: Linear, Logistic, Survival. In: And Repeated Measures Models, 2nd
https://doi.org/10.1007/b100507. ed. Springer, New York. https://doi.org/10.1007/b138825.
Moschis, G.P., 2019. Methods of life course research. In: Moschis, G.P. (Ed.), Consumer Vojvodić, M., Simović, S., Paspalj, D., 2020. Analysis of causes which make young people
Behavior Over the Life Course: Research Frontiers and New Directions. Springer quit sports and sport activities. J Adv Sport Phys Edu, 3, 142-149. doi: /10.36348/
jaspe.2020.v03i09.001.

10