Determinants of Sports Participation in Japan: The Interplay of Sociodemographic Factors, Social Roles, and Behavioral Change

Abstract

Sports participation is a widely recognized facilitator of physical health, mental well-being, and social inclusion, but persistent and substantial disparities have been observed across socioeconomic groups. Focusing on Japan, this study examined the socioeconomic determinants of sports participation, particularly the roles of gender, age, employment, and caregiving responsibilities. It used nationally representative repeated cross-sectional data to analyze participation rates and annual participation days across multiple sports at the population-segment level, defined by combinations of demographic and social attributes. Results revealed prominent sport-specific gender differences, heterogeneous age effects across sports, significant age–gender interaction effects, and distinctive behavioral changes during the COVID-19 pandemic. During the pandemic, participation in competitive and group sports declined with age, but walking increased among middle-aged and older adults. In addition, constraints in employment and caregiving had limited overall effects but significantly reduced engagement in walking. These findings suggest the crucial influence of the interaction among social roles, life-stage transitions, and historical context, rather than biological sex differences alone, on sports participation patterns, highlighting the urgency of designing sports policies as inclusive social interventions that consider diverse motivations and limitations across population groups.

1. Introduction

Sports and physical activity constitute an important lifestyle factor that helps enhance not only physical health but also mental well-being and social connections. However, research has observed substantial individual differences in sports participation and exercise frequency and has recognized the wide range of socioeconomic factors underlying these disparities. Studies have also documented the crucial roles played by socioeconomic status (SES), gender, age, occupation, caregiving responsibilities, and family circumstances, as well as the impact of the coronavirus disease 2019 (COVID-19) pandemic and household and community environments, in the formation and maintenance of sports and physical activity habits.

1.1. Key Terminology

Although related, the terms “sport”, “physical activity”, and “exercise” refer to distinct concepts. “Sport” refers to structured, rule-governed activities conducted in organized or competitive settings, such as baseball, badminton, and swimming. “Physical activity” is a broader term encompassing any bodily movement that results in energy expenditure, including everyday activities such as walking to work or climbing stairs. “Exercise” denotes planned, structured, and repetitive activity undertaken specifically to improve or maintain fitness, such as going to the gym or following a training program. These terms are distinguished where appropriate throughout this manuscript, though some cited studies use them interchangeably, and such usage is retained in the discussion of those works.

1.2. SES and Sports Participation

Scholars have identified SES as one of the most consistent determinants of sports participation. Tanaka et al. (2025) examined data on Japanese elementary school students and demonstrated that maternal educational attainment, employment status, and regional income levels are significantly linked to children’s engagement in sports activities. Yang et al. (2023), who investigated older adults in China, found that those with higher levels of education and income tend to engage in walking and light physical activity.

Research on Japanese adults has also highlighted the role of SES during the COVID-19 pandemic. Kyan and Takakura (2022) reported a pronounced decline in physical activity during that period, particularly among individuals with lower income and educational attainment. Furthermore, Zhao et al. (2025)’s analysis of the interaction between SES, region (urban versus rural), and age using nationally representative data from China revealed higher physical activity levels among urban residents, high-income groups, and younger cohorts.

These findings collectively indicate the significant influence of SES in sports and exercise habits as well as its complex effects moderated by social context and individual characteristics.

1.3. Gender Differences in Sports Participation

Studies have consistently observed gender differences in sports and physical activity participation from childhood to adulthood. Substantial international research has found that males tend to show higher engagement in sports participation, particularly in competitive and team activities, whereas females generally exhibit lower overall activity levels and a stronger preference for health-oriented and aesthetic individual sports.

Findings from Taiwan showed that 27% of boys maintained regular exercise habits compared with only 10% of girls, with particularly pronounced dropout rates among girls after adolescence (Wu et al. 2020). In a large-scale cross-sectional survey involving 28 European countries and Australia, Owen et al. (2025) reported that the male advantage in sports participation has widened in recent years despite the consistently high prevalence of insufficient physical activity for both genders.

Studies have also observed clear gender differences in sport selection, closely linked to cultural gender stereotypes, with females preferring individual and aesthetic-oriented activities such as dance, rhythmic gymnastics, skating, and swimming and males selecting team-based and contact sports such as soccer, basketball, baseball, and combat sports (Peral-Suárez et al. 2020). Research in China, Spain, and Germany has demonstrated the influence of socially constructed “masculine” and “feminine” images on sports participation motivation, with females more likely to experience anxiety and exclusion in sports perceived as masculine. Meanwhile, gender differences tend to be narrower in sports regarded as gender-neutral, such as badminton, running, and swimming (Ma et al. 2025).

1.4. Age and Sports Participation

Age is a significant driver of sports and physical activity habits. A large body of international research has reported a general age-related decline in sports and exercise participation rates, but its magnitude and pattern vary across regions and social contexts, with women and socially isolated older adults showing particularly low participation. Social survey data from Japan suggest that economic status, educational attainment, and employment status strongly influence participation, with higher-income, higher-educated, and unemployed individuals such as homemakers or retirees being more likely to maintain regular exercise habits (Shimokubo 2020). Longitudinal studies of community-dwelling older adults have also observed that women, individuals already performing light physical activity, and those with broader social networks are more likely to start participating in sports groups (Ejiri et al. 2022).

Distinct age-related changes in sports participation have also been identified across life stages. Individuals in their 50s and early 60s identified time constraints associated with work demands and workplace culture as critical factors that limit their sports participation opportunities. Meanwhile, despite the greater time availability created by retirement for those in their late 60s and beyond, participation trajectories tend to diverge depending on whether they encounter specific “triggers” or opportunities to initiate or reengage in physical activity (Jenkin et al. 2017).

Stenner et al. (2020) found that motivations for continued participation extend beyond health maintenance to include social relationships, self-fulfillment, and a sense of role or purpose. Individuals have frequently reported socially oriented motivations, such as community belonging and peer interactions, alongside factors associated with enjoyment, achievement, and event-based engagement. Concurrently, research has identified multifaceted barriers to sports participation. Individual-level factors include chronic health conditions, declining physical capacity, fear of falling, limited exercise knowledge, and psychological barriers such as the belief that “sports are for the young.” Social factors include isolation and a lack of friends as well as the prioritization of domestic and caregiving roles, particularly among older women (Suryadi et al. 2024).

1.5. The Impact of COVID-19 on Sports Participation

The impact of the COVID-19 pandemic on sports participation and physical activity levels worldwide has been profound. Dunton et al. (2020) observed that the suspension of schools, extracurricular activities, and sports clubs in many countries caused a marked decline in physical activity among children. In Italy, Maugeri et al. (2020) reported an approximately 35% decline in total physical activity after the onset of the pandemic, with reductions observed across all age groups. In Australia, a severe disruption in club-based sports participation was documented, with fewer than half of participants returning and particularly low reengagement rates among women even after the resumption of sports and physical activities (Eime et al. 2024).

Other studies have shown evidence of adaptive responses among certain population groups, who maintained or even increased their activity levels by shifting to home-based exercise, online fitness classes, or outdoor activities such as walking and running (Park et al. 2022). These findings highlight the pandemic’s heterogeneous effects on sports and exercise habits, driven by individual resources, access to alternatives, and social context.

1.6. Household and Community Environments and Sports Participation

Sports and physical activity habits are also shaped by immediate environments such as the household, community, and school, and such influence persists across the life course and not only during childhood and adolescence. Timperio et al. (2013) showed that children consistently increase their participation in sports through parental support, including encouragement, transportation, financial assistance, and joint participation, as well as through the availability of sports equipment. Similarly, Strandbu et al. (2020) demonstrated the strong influence of a family sports culture, characterized by parents’ exercise habits and values emphasizing the importance of sports, on children’s club participation through junior high and high school, with effects remaining largely stable as they age.

Studies have also reported a positive association between access to parks, sports facilities, and playgrounds, as well as safe walking environments, and sports participation and outdoor play among children and adolescents (Yang et al. 2023). Wang et al. (2024) also observed that all components of the “micro-environment,” such as school, household, and community, significantly influence adolescents’ physical activity behaviors and that positive household–school relations enhance expectations and further promote physical activity engagement.

1.7. Research Gaps and Objectives

Although sports and physical activity habits are closely linked to a wide range of socioeconomic attributes—such as gender, age, occupation, SES, household environment, and broader social changes—existing studies have largely examined these factors in isolation or in limited combinations. As a result, the relative contributions of these attributes and, critically, how they interact with one another to shape sports participation remain poorly understood.

To address this gap, this study performs a multidimensional analysis that simultaneously examines multiple socioeconomic attributes and their interactions as determinants of sports participation. By explicitly modeling interaction effects, for example, how the association between gender and participation is moderated by caregiving status or employment, this study moves beyond additive models to capture the complex, intersecting nature of participation disparities. The findings are intended to generate evidence that benefits policy interventions aimed at reducing sports participation disparities and promoting sustained engagement across the population.

2. Materials and Methods

To identify the determinants of sports participation and their interactions, we employ a Bayesian regression framework with horseshoe priors that accommodates multiple interaction effects and differences in sampling precision across survey waves. The following subsections describe the data source and sample construction, followed by definitions and descriptive statistics of the dependent and explanatory variables.

2.1. Data

2.1.1. Data Source and Sample Construction

This study used attribute-level aggregated data from the Survey on Time Use and Leisure Activities by the Statistics Bureau of Japan in 2006, 2011, 2016, and 2021. Its unit of analysis was a population segment defined by a combination of gender, age group, employment status, and caregiving status. This study observed segment-level sports participation outcomes for each survey year.

To ensure statistical stability, this study excluded population segments with extremely small sample sizes or extremely small estimated population counts. Specifically, it removed segments with fewer than 10 survey respondents or an estimated population of 10,000 or less.

Sampling weights were assigned to each segment–year observation, as the effective number of respondents differed across survey waves. These weights were based on each segment’s effective sample size within a given survey year and were normalized within each year. These resulting weights were used throughout the analysis to explain differences in sampling precision across survey waves.

After applying these exclusion criteria, this study obtained a final dataset of 192 segment–year observations of sports participation outcomes for the empirical analysis.

2.1.2. Variables and Descriptive Statistics

The analysis focused on sports participation behavior across eight activities in the following order: baseball, badminton, golf, bowling, fishing, swimming, walking, and gym use. These eight activities were selected based on two criteria. First, they represent some of the most widely practiced sports in Japan, ensuring sufficient variation in participation rates across demographic groups. Second, they collectively span a diverse range of activity types, encompassing both indoor and outdoor settings as well as individual and team-based formats, thereby allowing for a broad examination of how socioeconomic attributes shape participation across different sporting contexts. Two outcome measures were analyzed separately for each sport: annual participation rate and annual number of participation days.

Table 1. Definitions of Dependent and Explanatory Variables with Weighted Proportions.

Category	Variable	Description	Proportion
Dependent variable	Sports participation	Participation rate or annual number of participation days	–
Basic attributes
	Male	1 if male, 0 otherwise	0.475
	Work	1 if employed, 0 otherwise	0.621
	Care	1 if providing care, 0 otherwise	0.064
Age groups (reference: 20s)
	Age30	1 if aged 30–39	0.133
	Age40	1 if aged 40–49	0.157
	Age50	1 if aged 50–59	0.166
	Age60	1 if aged 60–69	0.172
	Age70	1 if aged 70 or older	0.217
Survey years (reference: 2006)
	Year2011	1 if observed in 2011	0.250
	Year2016	1 if observed in 2016	0.250
	Year2021	1 if observed in 2021	0.250
Interaction terms
	Work_Care	Employed and providing care	0.037
	Male_Work	Male and employed	0.341
	Male_Care	Male and providing care	0.025
	Age30_Male	Male aged 30–39	0.065
	Age40_Male	Male aged 40–49	0.076
	Age50_Male	Male aged 50–59	0.080
	Age60_Male	Male aged 60–69	0.083
	Age70_Male	Male aged 70+	0.093

summarizes the dependent and explanatory variables along with their definitions and weighted proportions for binary explanatory variables. The dependent variables capture sports participation behavior measured along both the extensive (participation rate) and intensive margins (participation days).

Explanatory variables included dummy indicators for gender, employment status, caregiving status, age group, and survey year. Interaction terms were also constructed between gender and other key attributes, such as employment status, caregiving status, and age group, to allow for heterogeneous associations across demographic groups. All explanatory variables were binary indicators. Interaction terms were denoted by an underscore “_” in the variable names (e.g., Age_Male or Work_Care) to indicate multiplicative combinations of the corresponding factors.

Descriptive statistics for the dependent variables are reported in

Table 2. Weighted Descriptive Statistics of Participation Rate and Annual Participation Days.

Sport	Participation Rate				Annual Participation Days
	Mean	SD	Min	Max	Mean	SD	Min	Max
Baseball	0.054	0.065	0.000	0.274	5.374	6.495	0.000	27.400
Badminton	0.047	0.049	0.000	0.205	4.707	4.877	0.000	20.500
Golf	0.067	0.064	0.000	0.245	6.670	6.402	0.000	24.500
Bowling	0.103	0.103	0.004	0.456	10.256	10.273	0.400	45.600
Fishing	0.077	0.061	0.000	0.398	7.746	6.065	0.000	39.800
Swimming	0.038	0.043	0.000	0.256	3.805	4.299	0.000	25.600
Walking	0.508	0.186	0.137	0.862	50.799	18.599	13.700	86.200
Gym equipment	0.086	0.096	0.000	0.414	8.624	9.608	0.000	41.400

All statistics are calculated using survey weights. Participation rate is defined as the weighted proportion of individuals who participated in each sport. Annual participation days are reported on the original scale for descriptive purposes; in the regression analysis, the logarithm of (annual participation days + 1) is used as the dependent variable.

, including their means, standard deviations, and minimum and maximum values. All descriptive statistics were computed using sampling weights.

2.2. Statistical Model

For each sport, the same statistical model was estimated independently, and notational simplicity was observed by omitting the sport index throughout this section.

Let $y_{gt}$ denote the sports participation outcome for population segment g in survey year t. Let ${\mathbf{x}}_{gt}$ be a k-dimensional row vector consisting of attribute dummy variables and interaction terms. All observations across segments and survey years were stacked vertically and reindexed by $i=1,\dots ,n$.

2.2.1. Model Specification

For each observation i, the participation outcome was modeled as a linear function of demographic and social attributes:

$$y_i=\alpha +{\mathbf{x}}_i\mathit{\beta }+{\epsilon }_i,{\epsilon }_i\sim \mathcal{N}(0,{\sigma }^2).$$

(1)

Here, $y_i$ represents the observed sports participation outcome for segment i, defined as the participation rate (in percentage points) or as the logarithm of annual participation days. The scalar $\alpha$ denotes the intercept term representing the baseline participation level. The vector ${\mathbf{x}}_i$ contains the explanatory variables, including gender, age group, employment status, caregiving status, survey-year dummies, and their interaction terms. The parameter vector $\mathit{\beta }$ includes the corresponding regression coefficients, and ${\epsilon }_i$ is a normally distributed error term with variance ${\sigma }^2$.

Stacking all observations results in the vectors and matrices $\mathbf{y}={(y_1,\dots ,y_n)}^{\top }$, $\mathbf{X}={({\mathbf{x}}_1^{\top },\dots ,{\mathbf{x}}_n^{\top })}^{\top }$, and the $n\times 1$ vector of ones $\mathbf{1}$, which can be expressed in matrix form as

$$\mathbf{y}=\mathbf{1}\alpha +\mathbf{X}\mathit{\beta }+\mathit{\epsilon },\mathit{\epsilon }\sim \mathcal{N}(\mathbf{0},{\sigma }^2{\mathbf{I}}_n).$$

(2)

2.2.2. Likelihood

To account for sampling precision differences across survey waves, each observation i was assigned a sampling weight $w_i$, with $\mathbf{W}=\mathrm{diag}(w_1,\dots ,w_n)$ denoting the diagonal weight matrix.

Assuming normally distributed errors with variance ${\sigma }^2$, the weighted likelihood was given by

$$p(\mathbf{y}\mid \alpha ,\mathit{\beta },{\sigma }^2)\propto {\left({\sigma }^2\right)}^{-n/2}{\left|\mathbf{W}\right|}^{1/2}\exp \left[-{\displaystyle \frac{1}{2{\sigma }^2}}{(\mathbf{y}-\mathbf{1}\alpha -\mathbf{X}\mathit{\beta })}^{\top }\mathbf{W}(\mathbf{y}-\mathbf{1}\alpha -\mathbf{X}\mathit{\beta })\right].$$

(3)

2.2.3. Prior Distributions

To complete the Bayesian specification, prior distributions were assigned to the abovementioned model parameters.

The intercept $\alpha$ was assigned a nonshrinkage normal prior,

$$\alpha \sim N({\mu }_{\alpha },{\sigma }_{\alpha }^2).$$

Regression coefficients ${\beta }_j$ followed a horseshoe prior with local–global shrinkage (Gelman 2006; Polson and Scott 2012):

$$\begin{array}{cc}\hfill {\beta }_j\mid {\lambda }_j,\tau ,\sigma & \sim N(0,{\sigma }^2{\lambda }_j^2{\tau }^2),\hfill \end{array}$$

(4)

$$\begin{array}{cc}\hfill {\lambda }_j& \sim C^{+}(0,s_{\lambda }),\hfill \end{array}$$

(5)

$$\begin{array}{cc}\hfill \tau & \sim C^{+}(0,s_{\tau }),\hfill \end{array}$$

(6)

where $C^{+}(0,s)$ denotes the half-Cauchy distribution with scale parameter s, defined as the positive-truncated Cauchy distribution. This prior specification induces strong shrinkage toward 0 while allowing a few large signals to remain unshrunk.

The error variance followed an inverse-gamma prior,

$${\sigma }^2\sim IG(a_{\sigma },b_{\sigma }).$$

2.2.4. Posterior and Full Conditional Distributions

Posterior inference was conducted via Gibbs sampling. Following Makalic and Schmidt (2016), the half-Cauchy priors were expressed as inverse-gamma scale mixtures, which yields closed-form full conditional distributions.

Let

$$\mathit{\theta }={(\alpha ,{\mathit{\beta }}^{\top })}^{\top },\mathbf{Z}=[\mathbf{1},\mathbf{X}],$$

and define the weighted sum of squared residuals as

$$SS{R}_w={(\mathbf{y}-\mathbf{1}\alpha -\mathbf{X}\mathit{\beta })}^{\top }\mathbf{W}(\mathbf{y}-\mathbf{1}\alpha -\mathbf{X}\mathit{\beta }).$$

(7)

The full conditional distribution of $\mathit{\theta }={(\alpha ,{\mathit{\beta }}^{\top })}^{\top }$ is multivariate normal:

$$\mathit{\theta }\mid \dots \sim N({\mathit{M}}^{*},{\mathit{V}}^{*}),$$

(8)

where

$$\begin{array}{cc}\hfill {\mathit{V}}^{*}& ={\left({\sigma }^{-2}{\mathbf{Z}}^{\top }\mathbf{W}\mathbf{Z}+{\mathsf{\Sigma }}_0^{-1}\right)}^{-1},\hfill \end{array}$$

(9)

$$\begin{array}{cc}\hfill {\mathit{M}}^{*}& ={\mathit{V}}^{*}\left({\sigma }^{-2}{\mathbf{Z}}^{\top }\mathbf{W}\mathbf{y}+{\mathsf{\Sigma }}_0^{-1}{\mathit{\mu }}_0\right).\hfill \end{array}$$

(10)

where ${\mathit{\mu }}_0$ and ${\mathsf{\Sigma }}_0$ denote the prior mean vector and prior covariance matrix, with ${\mathsf{\Sigma }}_0$ being diagonal and reflecting the nonshrinkage prior for $\alpha$ and the horseshoe-induced shrinkage for $\mathit{\beta }$.

The full conditional distribution of the error variance is

$${\sigma }^2\mid \dots \sim IG\left({\displaystyle \frac{N+k}{2}}+a_{\sigma },{\displaystyle \frac{SS{R}_w+{\sum }_{j=1}^k\frac{{\beta }_j^2}{{\lambda }_j^2{\tau }^2}}{2}}+b_{\sigma }\right),$$

(11)

where $N={\sum }_{i=1}^n{w}_i$.

For each regression coefficient ${\beta }_j$, the local shrinkage parameter ${\lambda }_j^2$ has the full conditional distribution

$${\lambda }_j^2\mid \dots \sim IG\left(1,{\displaystyle \frac{1}{{\nu }_j}}+{\displaystyle \frac{{\beta }_j^2}{2{\sigma }^2{\tau }^2}}\right),$$

(12)

and the corresponding auxiliary variable follows

$${\nu }_j\mid \dots \sim IG\left({\displaystyle \frac{1}{2}},{\displaystyle \frac{1}{s_{\lambda }^2}}+{\displaystyle \frac{1}{{\lambda }_j^2}}\right).$$

(13)

The full conditional distribution of the global shrinkage parameter is

$${\tau }^2\mid \dots \sim IG\left({\displaystyle \frac{k+1}{2}},{\displaystyle \frac{1}{\xi }}+{\displaystyle \frac{1}{2{\sigma }^2}}\sum _{j=1}^k{\displaystyle \frac{{\beta }_j^2}{{\lambda }_j^2}}\right),$$

(14)

with the auxiliary variable

$$\xi \mid \dots \sim IG\left({\displaystyle \frac{1}{2}},{\displaystyle \frac{1}{s_{\tau }^2}}+{\displaystyle \frac{1}{{\tau }^2}}\right).$$

(15)

2.2.5. Variable Selection via Horseshoe Shrinkage

The horseshoe prior allows the relevance of each regression coefficient to be summarized using a shrinkage factor. Following the formulation proposed by Piironen and Vehtari (2017), we defined the shrinkage factor for each coefficient as

$${\kappa }_j^{\left(s\right)}={\displaystyle \frac{1}{1+\frac{1}{4}n\overline{{\tau }^2}{\left({\lambda }_j^{\left(s\right)}\right)}^2}}.$$

(16)

Values of ${\kappa }_j$ close to 1 indicate strong shrinkage toward 0, whereas those close to 0 suggest a substantial effect.

2.2.6. MCMC and Hyperparameter Settings

A total of 20,000 MCMC iterations were generated, with the first 10,000 discarded as burn-in and the remaining 10,000 used for posterior summaries.

Hyperparameters are fixed as

$${\mu }_{\alpha }=0,{\sigma }_{\alpha }^2=100,s_{\lambda }=s_{\tau }=1,a_{\sigma }=b_{\sigma }=1.$$

3. Results

For participation rates, coefficients were interpreted directly as changes in participation probability (in percentage points). For annual participation days, outcomes were estimated on a logarithmic scale, and coefficients were interpreted as relative changes in participation days. Appendix A (see

Table A1. Posterior Summaries for Baseball.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$1.68\times {10}^{-1}$ *,†	$1.52\times {10}^{-1}$	$1.84\times {10}^{-1}$	$0.001$	$1.02$ *,†	$7.90\times {10}^{-1}$	$1.32$	$0.002$
Work	$-1.24\times {10}^{-3}$	$-9.30\times {10}^{-3}$	$6.84\times {10}^{-3}$	$0.183$	$-1.87\times {10}^{-2}$	$-1.48\times {10}^{-1}$	$1.12\times {10}^{-1}$	$0.148$
Care	$-1.45\times {10}^{-3}$	$-1.62\times {10}^{-2}$	$1.39\times {10}^{-2}$	$0.113$	$-7.27\times {10}^{-2}$	$-2.62\times {10}^{-1}$	$1.17\times {10}^{-1}$	$0.083$
Age groups (reference: 20s)
Age30	$-5.27\times {10}^{-3}$	$-1.94\times {10}^{-2}$	$8.92\times {10}^{-3}$	$0.095$	$-9.40\times {10}^{-2}$	$-3.12\times {10}^{-1}$	$1.67\times {10}^{-1}$	$0.058$
Age40	$-2.17\times {10}^{-2}$ *,†	$-3.53\times {10}^{-2}$	$-8.21\times {10}^{-3}$	$0.024$	$-4.51\times {10}^{-1}$ *,†	$-6.49\times {10}^{-1}$	$-2.04\times {10}^{-1}$	$0.010$
Age50	$-4.66\times {10}^{-2}$ *,†	$-6.06\times {10}^{-2}$	$-3.35\times {10}^{-2}$	$0.008$	$-1.36$ *,†	$-1.57$	$-1.12$	$0.002$
Age60	$-5.04\times {10}^{-2}$ *,†	$-6.34\times {10}^{-2}$	$-3.57\times {10}^{-2}$	$0.007$	$-1.58$ *,†	$-1.81$	$-1.32$	$0.001$
Age70	$-5.18\times {10}^{-2}$ *,†	$-6.59\times {10}^{-2}$	$-3.83\times {10}^{-2}$	$0.007$	$-1.73$ *,†	$-1.95$	$-1.44$	$0.001$
Survey years (reference: 2006)
Year2011	$-7.12\times {10}^{-3}$ *	$-1.39\times {10}^{-2}$	$-3.05\times {10}^{-4}$	$0.098$	$-2.10\times {10}^{-2}$	$-1.30\times {10}^{-1}$	$8.40\times {10}^{-2}$	$0.161$
Year2016	$-3.47\times {10}^{-3}$	$-9.89\times {10}^{-3}$	$3.84\times {10}^{-3}$	$0.156$	$8.61\times {10}^{-2}$	$-1.31\times {10}^{-2}$	$2.01\times {10}^{-1}$	$0.092$
Year2021	$-1.03\times {10}^{-2}$ *	$-1.73\times {10}^{-2}$	$-3.42\times {10}^{-3}$	$0.062$	$2.01\times {10}^{-2}$	$-8.82\times {10}^{-2}$	$1.21\times {10}^{-1}$	$0.189$
Interaction terms
Work_Care	$-5.68\times {10}^{-4}$	$-1.84\times {10}^{-2}$	$1.83\times {10}^{-2}$	$0.099$	$-4.74\times {10}^{-2}$	$-2.67\times {10}^{-1}$	$2.05\times {10}^{-1}$	$0.085$
Male_Work	$9.51\times {10}^{-3}$	$-1.70\times {10}^{-3}$	$2.20\times {10}^{-2}$	$0.071$	$3.34\times {10}^{-1}$ *,†	$1.81\times {10}^{-1}$	$5.05\times {10}^{-1}$	$0.015$
Male_Care	$-9.90\times {10}^{-4}$	$-2.00\times {10}^{-2}$	$1.68\times {10}^{-2}$	$0.097$	$9.52\times {10}^{-2}$	$-1.35\times {10}^{-1}$	$3.23\times {10}^{-1}$	$0.070$
Age30_Male	$-7.25\times {10}^{-2}$ *,†	$-9.12\times {10}^{-2}$	$-5.38\times {10}^{-2}$	$0.004$	$-3.13\times {10}^{-1}$ *,†	$-5.85\times {10}^{-1}$	$-4.43\times {10}^{-2}$	$0.019$
Age40_Male	$-7.08\times {10}^{-2}$ *,†	$-8.87\times {10}^{-2}$	$-5.27\times {10}^{-2}$	$0.004$	$-7.72\times {10}^{-2}$	$-3.49\times {10}^{-1}$	$1.71\times {10}^{-1}$	$0.085$
Age50_Male	$-1.24\times {10}^{-1}$ *,†	$-1.41\times {10}^{-1}$	$-1.05\times {10}^{-1}$	$0.002$	$1.08\times {10}^{-1}$	$-1.63\times {10}^{-1}$	$3.63\times {10}^{-1}$	$0.053$
Age60_Male	$-1.46\times {10}^{-1}$ *,†	$-1.64\times {10}^{-1}$	$-1.28\times {10}^{-1}$	$0.001$	$-1.05\times {10}^{-1}$	$-3.82\times {10}^{-1}$	$1.59\times {10}^{-1}$	$0.076$
Age70_Male	$-1.57\times {10}^{-1}$ *,†	$-1.75\times {10}^{-1}$	$-1.39\times {10}^{-1}$	$0.001$	$-3.35\times {10}^{-1}$ *,†	$-6.28\times {10}^{-1}$	$-6.67\times {10}^{-2}$	$0.018$
Other parameters
Intercept ($\alpha$)	$3.54\times {10}^{-3}$	$-4.83\times {10}^{-1}$	$5.78\times {10}^{-1}$	–	$5.21\times {10}^{-1}$	$-5.34\times {10}^{-1}$	$2.14$	–
Error s.d. ($\sigma$)	$1.58\times {10}^{-2}$	$1.38\times {10}^{-2}$	$1.78\times {10}^{-2}$	–	$2.05\times {10}^{-1}$	$1.70\times {10}^{-1}$	$2.47\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A2. Posterior Summaries for Badminton.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$-3.30\times {10}^{-2}$ *,†	$-5.00\times {10}^{-2}$	$-1.47\times {10}^{-2}$	$0.010$	$-2.74\times {10}^{-1}$ *,†	$-4.77\times {10}^{-1}$	$-6.66\times {10}^{-2}$	$0.016$
Work	$-7.92\times {10}^{-3}$	$-1.62\times {10}^{-2}$	$-7.51\times {10}^{-5}$	$0.061$	$-2.85\times {10}^{-2}$	$-1.47\times {10}^{-1}$	$8.43\times {10}^{-2}$	$0.123$
Care	$-8.16\times {10}^{-3}$	$-2.05\times {10}^{-2}$	$4.88\times {10}^{-3}$	$0.063$	$-1.50\times {10}^{-1}$ *,†	$-3.12\times {10}^{-1}$	$2.00\times {10}^{-2}$	$0.042$
Age groups (reference: 20s)
Age30	$-4.67\times {10}^{-2}$ *,†	$-6.03\times {10}^{-2}$	$-3.10\times {10}^{-2}$	$0.006$	$-3.42\times {10}^{-1}$ *,†	$-4.87\times {10}^{-1}$	$-2.01\times {10}^{-1}$	$0.012$
Age40	$-5.93\times {10}^{-2}$ *,†	$-7.29\times {10}^{-2}$	$-4.54\times {10}^{-2}$	$0.004$	$-4.34\times {10}^{-1}$ *,†	$-6.01\times {10}^{-1}$	$-2.58\times {10}^{-1}$	$0.009$
Age50	$-1.14\times {10}^{-1}$ *,†	$-1.28\times {10}^{-1}$	$-9.98\times {10}^{-2}$	$0.001$	$-1.31$ *,†	$-1.47$	$-1.13$	$0.001$
Age60	$-1.26\times {10}^{-1}$ *,†	$-1.40\times {10}^{-1}$	$-1.11\times {10}^{-1}$	$0.001$	$-1.74$ *,†	$-1.92$	$-1.55$	$0.001$
Age70	$-1.39\times {10}^{-1}$ *,†	$-1.53\times {10}^{-1}$	$-1.23\times {10}^{-1}$	$0.001$	$-2.31$ *,†	$-2.50$	$-2.10$	$0.000$
Survey years (reference: 2006)
Year2011	$-9.97\times {10}^{-3}$ *,†	$-1.65\times {10}^{-2}$	$-3.67\times {10}^{-3}$	$0.049$	$-7.92\times {10}^{-2}$	$-1.72\times {10}^{-1}$	$8.67\times {10}^{-3}$	$0.077$
Year2016	$6.72\times {10}^{-3}$ *	$5.68\times {10}^{-4}$	$1.32\times {10}^{-2}$	$0.088$	$1.98\times {10}^{-1}$ *,†	$1.05\times {10}^{-1}$	$2.86\times {10}^{-1}$	$0.028$
Year2021	$2.95\times {10}^{-3}$	$-3.39\times {10}^{-3}$	$9.40\times {10}^{-3}$	$0.159$	$1.64\times {10}^{-1}$ *,†	$6.88\times {10}^{-2}$	$2.53\times {10}^{-1}$	$0.036$
Interaction terms
Work_Care	$4.68\times {10}^{-3}$	$-1.07\times {10}^{-2}$	$2.09\times {10}^{-2}$	$0.080$	$5.21\times {10}^{-2}$	$-1.57\times {10}^{-1}$	$2.53\times {10}^{-1}$	$0.079$
Male_Work	$3.19\times {10}^{-3}$	$-7.06\times {10}^{-3}$	$1.47\times {10}^{-2}$	$0.108$	$9.78\times {10}^{-2}$	$-5.95\times {10}^{-2}$	$2.62\times {10}^{-1}$	$0.058$
Male_Care	$4.20\times {10}^{-3}$	$-1.12\times {10}^{-2}$	$1.98\times {10}^{-2}$	$0.083$	$6.37\times {10}^{-2}$	$-1.33\times {10}^{-1}$	$2.65\times {10}^{-1}$	$0.078$
Age30_Male	$2.81\times {10}^{-5}$	$-1.82\times {10}^{-2}$	$1.86\times {10}^{-2}$	$0.093$	$-6.59\times {10}^{-2}$	$-1.57\times {10}^{-1}$	$1.98\times {10}^{-2}$	$0.105$
Age40_Male	$7.43\times {10}^{-3}$	$-9.97\times {10}^{-3}$	$2.48\times {10}^{-2}$	$0.056$	$-1.47\times {10}^{-1}$ *,†	$-3.38\times {10}^{-1}$	$5.16\times {10}^{-2}$	$0.049$
Age50_Male	$3.04\times {10}^{-2}$ *,†	$1.35\times {10}^{-2}$	$4.79\times {10}^{-2}$	$0.012$	$8.11\times {10}^{-2}$	$-1.02\times {10}^{-1}$	$2.70\times {10}^{-1}$	$0.067$
Age60_Male	$2.94\times {10}^{-2}$ *,†	$1.25\times {10}^{-2}$	$4.82\times {10}^{-2}$	$0.012$	$8.71\times {10}^{-2}$	$-1.14\times {10}^{-1}$	$2.86\times {10}^{-1}$	$0.061$
Age70_Male	$3.45\times {10}^{-2}$ *,†	$1.54\times {10}^{-2}$	$5.23\times {10}^{-2}$	$0.009$	$3.48\times {10}^{-1}$ *,†	$1.52\times {10}^{-1}$	$5.84\times {10}^{-1}$	$0.012$
Other parameters
Intercept ($\alpha$)	$8.88\times {10}^{-3}$	$-5.58\times {10}^{-1}$	$5.42\times {10}^{-1}$	–	$1.01$	$-3.36\times {10}^{-1}$	$3.08$	–
Error s.d. ($\sigma$)	$1.33\times {10}^{-2}$	$1.15\times {10}^{-2}$	$1.52\times {10}^{-2}$	–	$1.77\times {10}^{-1}$	$1.49\times {10}^{-1}$	$2.08\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A3. Posterior Summaries for Golf.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$1.83\times {10}^{-2}$ *,†	$2.07\times {10}^{-3}$	$3.60\times {10}^{-2}$	$0.033$	$6.35\times {10}^{-1}$ *,†	$3.71\times {10}^{-1}$	$9.68\times {10}^{-1}$	$0.006$
Work	$1.54\times {10}^{-2}$ *,†	$7.04\times {10}^{-3}$	$2.39\times {10}^{-2}$	$0.038$	$4.19\times {10}^{-1}$ *,†	$2.94\times {10}^{-1}$	$5.70\times {10}^{-1}$	$0.011$
Care	$8.01\times {10}^{-4}$	$-1.58\times {10}^{-2}$	$1.85\times {10}^{-2}$	$0.102$	$-7.78\times {10}^{-2}$	$-2.73\times {10}^{-1}$	$1.24\times {10}^{-1}$	$0.082$
Age groups (reference: 20s)
Age30	$-1.94\times {10}^{-3}$	$-1.68\times {10}^{-2}$	$1.26\times {10}^{-2}$	$0.118$	$5.56\times {10}^{-3}$	$-2.24\times {10}^{-1}$	$2.64\times {10}^{-1}$	$0.105$
Age40	$-2.84\times {10}^{-3}$	$-1.71\times {10}^{-2}$	$1.10\times {10}^{-2}$	$0.111$	$4.98\times {10}^{-3}$	$-2.03\times {10}^{-1}$	$2.47\times {10}^{-1}$	$0.113$
Age50	$9.84\times {10}^{-4}$	$-1.31\times {10}^{-2}$	$1.53\times {10}^{-2}$	$0.126$	$1.08\times {10}^{-1}$	$-1.11\times {10}^{-1}$	$3.54\times {10}^{-1}$	$0.086$
Age60	$-5.77\times {10}^{-3}$	$-1.98\times {10}^{-2}$	$8.81\times {10}^{-3}$	$0.090$	$-4.21\times {10}^{-2}$	$-2.61\times {10}^{-1}$	$2.47\times {10}^{-1}$	$0.083$
Age70	$-1.52\times {10}^{-2}$ *,†	$-2.89\times {10}^{-2}$	$-2.32\times {10}^{-4}$	$0.040$	$-4.89\times {10}^{-1}$ *,†	$-7.16\times {10}^{-1}$	$-1.80\times {10}^{-1}$	$0.009$
Survey years (reference: 2006)
Year2011	$-7.11\times {10}^{-3}$	$-1.45\times {10}^{-2}$	$1.61\times {10}^{-4}$	$0.089$	$-2.56\times {10}^{-2}$	$-1.36\times {10}^{-1}$	$8.15\times {10}^{-2}$	$0.156$
Year2016	$-8.81\times {10}^{-3}$ *	$-1.61\times {10}^{-2}$	$-1.54\times {10}^{-3}$	$0.074$	$-4.62\times {10}^{-2}$	$-1.56\times {10}^{-1}$	$6.04\times {10}^{-2}$	$0.128$
Year2021	$-1.91\times {10}^{-2}$ *,†	$-2.66\times {10}^{-2}$	$-1.21\times {10}^{-2}$	$0.026$	$-1.28\times {10}^{-1}$ *	$-2.38\times {10}^{-1}$	$-2.25\times {10}^{-2}$	$0.055$
Interaction terms
Work_Care	$-3.64\times {10}^{-3}$	$-2.46\times {10}^{-2}$	$1.72\times {10}^{-2}$	$0.086$	$-1.93\times {10}^{-2}$	$-2.60\times {10}^{-1}$	$2.12\times {10}^{-1}$	$0.088$
Male_Work	$5.77\times {10}^{-2}$ *,†	$4.45\times {10}^{-2}$	$7.04\times {10}^{-2}$	$0.005$	$2.98\times {10}^{-1}$ *,†	$1.20\times {10}^{-1}$	$4.70\times {10}^{-1}$	$0.019$
Male_Care	$5.41\times {10}^{-4}$	$-2.09\times {10}^{-2}$	$2.15\times {10}^{-2}$	$0.084$	$4.34\times {10}^{-2}$	$-1.96\times {10}^{-1}$	$2.84\times {10}^{-1}$	$0.086$
Age30_Male	$3.81\times {10}^{-2}$ *,†	$1.81\times {10}^{-2}$	$5.80\times {10}^{-2}$	$0.010$	$2.81\times {10}^{-1}$†	$-1.85\times {10}^{-2}$	$5.54\times {10}^{-1}$	$0.022$
Age40_Male	$5.32\times {10}^{-2}$ *,†	$3.31\times {10}^{-2}$	$7.23\times {10}^{-2}$	$0.006$	$3.70\times {10}^{-1}$ *,†	$9.70\times {10}^{-2}$	$6.50\times {10}^{-1}$	$0.014$
Age50_Male	$7.40\times {10}^{-2}$ *,†	$5.47\times {10}^{-2}$	$9.27\times {10}^{-2}$	$0.003$	$4.14\times {10}^{-1}$ *,†	$1.35\times {10}^{-1}$	$6.75\times {10}^{-1}$	$0.012$
Age60_Male	$9.33\times {10}^{-2}$ *,†	$7.38\times {10}^{-2}$	$1.12\times {10}^{-1}$	$0.002$	$7.41\times {10}^{-1}$ *,†	$4.39\times {10}^{-1}$	$1.02$	$0.005$
Age70_Male	$5.23\times {10}^{-2}$ *,†	$3.35\times {10}^{-2}$	$7.12\times {10}^{-2}$	$0.006$	$9.08\times {10}^{-1}$ *,†	$5.77\times {10}^{-1}$	$1.20$	$0.003$
Other parameters
Intercept ($\alpha$)	$-1.89\times {10}^{-3}$	$-4.60\times {10}^{-1}$	$5.07\times {10}^{-1}$	–	$2.28\times {10}^{-1}$	$-6.37\times {10}^{-1}$	$1.41$	–
Error s.d. ($\sigma$)	$1.81\times {10}^{-2}$	$1.55\times {10}^{-2}$	$2.09\times {10}^{-2}$	–	$2.11\times {10}^{-1}$	$1.77\times {10}^{-1}$	$2.50\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A4. Posterior Summaries for Bowling.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$7.37\times {10}^{-2}$ *,†	$2.76\times {10}^{-2}$	$1.24\times {10}^{-1}$	$0.014$	$3.37\times {10}^{-1}$ *,†	$1.03\times {10}^{-1}$	$5.45\times {10}^{-1}$	$0.021$
Work	$3.54\times {10}^{-2}$ *,†	$1.31\times {10}^{-2}$	$5.75\times {10}^{-2}$	$0.039$	$2.66\times {10}^{-1}$ *,†	$1.30\times {10}^{-1}$	$3.83\times {10}^{-1}$	$0.027$
Care	$5.39\times {10}^{-3}$	$-3.03\times {10}^{-2}$	$4.06\times {10}^{-2}$	$0.111$	$-9.51\times {10}^{-2}$	$-3.02\times {10}^{-1}$	$1.31\times {10}^{-1}$	$0.082$
Age groups (reference: 20s)
Age30	$-8.55\times {10}^{-2}$ *,†	$-1.25\times {10}^{-1}$	$-4.45\times {10}^{-2}$	$0.011$	$-3.85\times {10}^{-1}$ *,†	$-5.63\times {10}^{-1}$	$-2.28\times {10}^{-1}$	$0.016$
Age40	$-1.12\times {10}^{-1}$ *,†	$-1.49\times {10}^{-1}$	$-7.38\times {10}^{-2}$	$0.007$	$-5.97\times {10}^{-1}$ *,†	$-7.93\times {10}^{-1}$	$-4.19\times {10}^{-1}$	$0.008$
Age50	$-1.77\times {10}^{-1}$ *,†	$-2.16\times {10}^{-1}$	$-1.38\times {10}^{-1}$	$0.003$	$-1.23$ *,†	$-1.43$	$-1.04$	$0.003$
Age60	$-1.93\times {10}^{-1}$ *,†	$-2.32\times {10}^{-1}$	$-1.52\times {10}^{-1}$	$0.003$	$-1.60$ *,†	$-1.79$	$-1.41$	$0.002$
Age70	$-1.99\times {10}^{-1}$ *,†	$-2.41\times {10}^{-1}$	$-1.56\times {10}^{-1}$	$0.003$	$-2.17$ *,†	$-2.38$	$-1.99$	$0.001$
Survey years (reference: 2006)
Year2011	$-4.25\times {10}^{-2}$ *,†	$-6.06\times {10}^{-2}$	$-2.42\times {10}^{-2}$	$0.028$	$-2.13\times {10}^{-1}$ *,†	$-3.24\times {10}^{-1}$	$-1.11\times {10}^{-1}$	$0.036$
Year2016	$-4.15\times {10}^{-2}$ *,†	$-5.97\times {10}^{-2}$	$-2.41\times {10}^{-2}$	$0.029$	$-1.58\times {10}^{-1}$ *	$-2.67\times {10}^{-1}$	$-5.62\times {10}^{-2}$	$0.052$
Year2021	$-1.10\times {10}^{-1}$ *,†	$-1.27\times {10}^{-1}$	$-9.13\times {10}^{-2}$	$0.007$	$-8.95\times {10}^{-1}$ *,†	$-1.00$	$-7.91\times {10}^{-1}$	$0.004$
Interaction terms
Work_Care	$-6.96\times {10}^{-3}$	$-5.04\times {10}^{-2}$	$3.83\times {10}^{-2}$	$0.089$	$5.96\times {10}^{-2}$	$-2.05\times {10}^{-1}$	$3.17\times {10}^{-1}$	$0.089$
Male_Work	$-7.43\times {10}^{-3}$	$-3.89\times {10}^{-2}$	$2.28\times {10}^{-2}$	$0.121$	$1.91\times {10}^{-2}$	$-1.53\times {10}^{-1}$	$2.04\times {10}^{-1}$	$0.128$
Male_Care	$5.27\times {10}^{-3}$	$-3.62\times {10}^{-2}$	$5.17\times {10}^{-2}$	$0.092$	$7.47\times {10}^{-2}$	$-2.01\times {10}^{-1}$	$3.45\times {10}^{-1}$	$0.082$
Age30_Male	$-5.13\times {10}^{-2}$ *,†	$-1.02\times {10}^{-1}$	$-3.66\times {10}^{-3}$	$0.028$	$-2.53\times {10}^{-1}$ *,†	$-4.50\times {10}^{-1}$	$-3.70\times {10}^{-2}$	$0.032$
Age40_Male	$-4.66\times {10}^{-2}$ *,†	$-9.28\times {10}^{-2}$	$2.19\times {10}^{-3}$	$0.031$	$-2.61\times {10}^{-1}$ *,†	$-5.13\times {10}^{-1}$	$-6.51\times {10}^{-3}$	$0.034$
Age50_Male	$-3.78\times {10}^{-2}$†	$-8.50\times {10}^{-2}$	$9.85\times {10}^{-3}$	$0.042$	$-5.10\times {10}^{-3}$	$-2.42\times {10}^{-1}$	$2.56\times {10}^{-1}$	$0.101$
Age60_Male	$-5.43\times {10}^{-2}$ *,†	$-1.01\times {10}^{-1}$	$-4.53\times {10}^{-3}$	$0.024$	$-5.69\times {10}^{-3}$	$-2.56\times {10}^{-1}$	$2.49\times {10}^{-1}$	$0.106$
Age70_Male	$-6.44\times {10}^{-2}$ *,†	$-1.16\times {10}^{-1}$	$-1.40\times {10}^{-2}$	$0.019$	$5.69\times {10}^{-2}$	$-1.80\times {10}^{-1}$	$3.27\times {10}^{-1}$	$0.086$
Other parameters
Intercept ($\alpha$)	$2.18\times {10}^{-2}$	$-6.43\times {10}^{-1}$	$6.63\times {10}^{-1}$	–	$1.71$	$-2.36\times {10}^{-1}$	$3.89$	–
Error s.d. ($\sigma$)	$3.77\times {10}^{-2}$	$3.21\times {10}^{-2}$	$4.34\times {10}^{-2}$	–	$2.40\times {10}^{-1}$	$2.03\times {10}^{-1}$	$2.77\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A5. Posterior Summaries for Fishing.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$6.80\times {10}^{-2}$ *,†	$5.29\times {10}^{-2}$	$8.25\times {10}^{-2}$	$0.002$	$8.85\times {10}^{-1}$ *,†	$6.81\times {10}^{-1}$	$1.12$	$0.002$
Work	$3.02\times {10}^{-3}$	$-4.20\times {10}^{-3}$	$1.02\times {10}^{-2}$	$0.147$	$6.44\times {10}^{-2}$	$-3.32\times {10}^{-2}$	$1.85\times {10}^{-1}$	$0.111$
Care	$1.06\times {10}^{-3}$	$-1.18\times {10}^{-2}$	$1.45\times {10}^{-2}$	$0.103$	$-8.77\times {10}^{-2}$	$-2.32\times {10}^{-1}$	$5.18\times {10}^{-2}$	$0.064$
Age groups (reference: 20s)
Age30	$2.31\times {10}^{-2}$ *,†	$1.08\times {10}^{-2}$	$3.62\times {10}^{-2}$	$0.015$	$4.26\times {10}^{-1}$ *,†	$2.57\times {10}^{-1}$	$6.27\times {10}^{-1}$	$0.008$
Age40	$3.64\times {10}^{-3}$	$-8.47\times {10}^{-3}$	$1.58\times {10}^{-2}$	$0.114$	$1.41\times {10}^{-1}$	$-1.56\times {10}^{-2}$	$3.35\times {10}^{-1}$	$0.049$
Age50	$-2.74\times {10}^{-2}$ *,†	$-3.98\times {10}^{-2}$	$-1.56\times {10}^{-2}$	$0.012$	$-4.51\times {10}^{-1}$ *,†	$-5.93\times {10}^{-1}$	$-2.67\times {10}^{-1}$	$0.007$
Age60	$-3.37\times {10}^{-2}$ *,†	$-4.58\times {10}^{-2}$	$-2.13\times {10}^{-2}$	$0.008$	$-7.41\times {10}^{-1}$ *,†	$-9.01\times {10}^{-1}$	$-5.12\times {10}^{-1}$	$0.003$
Age70	$-4.34\times {10}^{-2}$ *,†	$-5.58\times {10}^{-2}$	$-3.05\times {10}^{-2}$	$0.006$	$-1.28$ *,†	$-1.45$	$-1.04$	$0.001$
Survey years (reference: 2006)
Year2011	$-1.47\times {10}^{-2}$ *,†	$-2.08\times {10}^{-2}$	$-8.90\times {10}^{-3}$	$0.027$	$-1.27\times {10}^{-1}$ *,†	$-2.09\times {10}^{-1}$	$-4.55\times {10}^{-2}$	$0.040$
Year2016	$-8.10\times {10}^{-3}$ *	$-1.42\times {10}^{-2}$	$-2.22\times {10}^{-3}$	$0.060$	$-2.44\times {10}^{-2}$	$-1.05\times {10}^{-1}$	$5.71\times {10}^{-2}$	$0.143$
Year2021	$-1.49\times {10}^{-2}$ *,†	$-2.11\times {10}^{-2}$	$-9.22\times {10}^{-3}$	$0.026$	$-9.18\times {10}^{-2}$ *	$-1.78\times {10}^{-1}$	$-1.41\times {10}^{-2}$	$0.056$
Interaction terms
Work_Care	$-6.75\times {10}^{-3}$	$-2.22\times {10}^{-2}$	$1.05\times {10}^{-2}$	$0.070$	$-3.88\times {10}^{-2}$	$-2.08\times {10}^{-1}$	$1.38\times {10}^{-1}$	$0.091$
Male_Work	$2.65\times {10}^{-2}$ *,†	$1.62\times {10}^{-2}$	$3.67\times {10}^{-2}$	$0.012$	$2.34\times {10}^{-1}$ *,†	$9.03\times {10}^{-2}$	$3.75\times {10}^{-1}$	$0.018$
Male_Care	$2.23\times {10}^{-3}$	$-1.49\times {10}^{-2}$	$1.77\times {10}^{-2}$	$0.086$	$9.61\times {10}^{-2}$	$-6.83\times {10}^{-2}$	$2.60\times {10}^{-1}$	$0.058$
Age30_Male	$-4.99\times {10}^{-3}$	$-2.16\times {10}^{-2}$	$1.14\times {10}^{-2}$	$0.083$	$-4.02\times {10}^{-1}$ *,†	$-6.02\times {10}^{-1}$	$-2.08\times {10}^{-1}$	$0.009$
Age40_Male	$1.84\times {10}^{-2}$ *	$3.04\times {10}^{-3}$	$3.48\times {10}^{-2}$	$0.022$	$-1.13\times {10}^{-1}$	$-3.10\times {10}^{-1}$	$6.88\times {10}^{-2}$	$0.062$
Age50_Male	$2.01\times {10}^{-2}$ *	$4.52\times {10}^{-3}$	$3.58\times {10}^{-2}$	$0.021$	$2.62\times {10}^{-1}$ *,†	$1.19\times {10}^{-1}$	$4.09\times {10}^{-1}$	$0.016$
Age60_Male	$2.56\times {10}^{-2}$ *,†	$9.25\times {10}^{-3}$	$4.11\times {10}^{-2}$	$0.014$	$6.31\times {10}^{-1}$ *,†	$4.17\times {10}^{-1}$	$8.26\times {10}^{-1}$	$0.004$
Age70_Male	$-1.04\times {10}^{-2}$	$-2.67\times {10}^{-2}$	$6.16\times {10}^{-3}$	$0.054$	$7.40\times {10}^{-1}$ *,†	$5.05\times {10}^{-1}$	$9.44\times {10}^{-1}$	$0.004$
Other parameters
Intercept ($\alpha$)	$4.53\times {10}^{-3}$	$-4.70\times {10}^{-1}$	$5.46\times {10}^{-1}$	–	$3.83\times {10}^{-1}$	$-5.09\times {10}^{-1}$	$1.82$	–
Error s.d. ($\sigma$)	$1.38\times {10}^{-2}$	$1.18\times {10}^{-2}$	$1.59\times {10}^{-2}$	–	$1.41\times {10}^{-1}$	$1.20\times {10}^{-1}$	$1.65\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A6. Posterior Summaries for Swimming.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$1.19\times {10}^{-2}$	$-1.37\times {10}^{-2}$	$3.73\times {10}^{-2}$	$0.076$	$7.82\times {10}^{-2}$	$-1.57\times {10}^{-1}$	$3.04\times {10}^{-1}$	$0.096$
Work	$-6.59\times {10}^{-3}$	$-1.88\times {10}^{-2}$	$5.45\times {10}^{-3}$	$0.101$	$-6.32\times {10}^{-2}$	$-1.76\times {10}^{-1}$	$4.47\times {10}^{-2}$	$0.091$
Care	$-2.21\times {10}^{-3}$	$-2.37\times {10}^{-2}$	$1.85\times {10}^{-2}$	$0.099$	$-9.17\times {10}^{-2}$	$-2.68\times {10}^{-1}$	$1.02\times {10}^{-1}$	$0.069$
Age groups (reference: 20s)
Age30	$2.42\times {10}^{-2}$ *,†	$3.18\times {10}^{-3}$	$4.66\times {10}^{-2}$	$0.029$	$1.98\times {10}^{-1}$ *,†	$6.86\times {10}^{-3}$	$3.91\times {10}^{-1}$	$0.032$
Age40	$-1.25\times {10}^{-2}$	$-3.40\times {10}^{-2}$	$7.10\times {10}^{-3}$	$0.053$	$-6.08\times {10}^{-2}$	$-2.52\times {10}^{-1}$	$1.29\times {10}^{-1}$	$0.075$
Age50	$-6.13\times {10}^{-2}$ *,†	$-8.21\times {10}^{-2}$	$-3.94\times {10}^{-2}$	$0.006$	$-6.41\times {10}^{-1}$ *,†	$-8.39\times {10}^{-1}$	$-4.57\times {10}^{-1}$	$0.005$
Age60	$-6.03\times {10}^{-2}$ *,†	$-8.15\times {10}^{-2}$	$-3.78\times {10}^{-2}$	$0.006$	$-6.02\times {10}^{-1}$ *,†	$-7.92\times {10}^{-1}$	$-4.02\times {10}^{-1}$	$0.005$
Age70	$-9.04\times {10}^{-2}$ *,†	$-1.12\times {10}^{-1}$	$-6.66\times {10}^{-2}$	$0.003$	$-1.18$ *,†	$-1.38$	$-9.72\times {10}^{-1}$	$0.002$
Survey years (reference: 2006)
Year2011	$-2.58\times {10}^{-2}$ *,†	$-3.55\times {10}^{-2}$	$-1.58\times {10}^{-2}$	$0.022$	$-1.82\times {10}^{-1}$ *,†	$-2.79\times {10}^{-1}$	$-9.34\times {10}^{-2}$	$0.030$
Year2016	$-1.89\times {10}^{-2}$ *,†	$-2.90\times {10}^{-2}$	$-8.85\times {10}^{-3}$	$0.033$	$-1.22\times {10}^{-1}$ *	$-2.19\times {10}^{-1}$	$-3.38\times {10}^{-2}$	$0.050$
Year2021	$-6.32\times {10}^{-2}$ *,†	$-7.29\times {10}^{-2}$	$-5.32\times {10}^{-2}$	$0.005$	$-6.62\times {10}^{-1}$ *,†	$-7.54\times {10}^{-1}$	$-5.69\times {10}^{-1}$	$0.004$
Interaction terms
Work_Care	$8.73\times {10}^{-3}$	$-1.68\times {10}^{-2}$	$3.50\times {10}^{-2}$	$0.070$	$1.71\times {10}^{-1}$†	$-5.94\times {10}^{-2}$	$4.05\times {10}^{-1}$	$0.040$
Male_Work	$3.63\times {10}^{-3}$	$-1.31\times {10}^{-2}$	$2.14\times {10}^{-2}$	$0.115$	$5.19\times {10}^{-2}$	$-1.07\times {10}^{-1}$	$2.14\times {10}^{-1}$	$0.094$
Male_Care	$-1.53\times {10}^{-3}$	$-2.63\times {10}^{-2}$	$2.51\times {10}^{-2}$	$0.083$	$-4.37\times {10}^{-2}$	$-2.68\times {10}^{-1}$	$1.71\times {10}^{-1}$	$0.079$
Age30_Male	$-1.59\times {10}^{-2}$	$-4.39\times {10}^{-2}$	$1.27\times {10}^{-2}$	$0.053$	$-1.19\times {10}^{-1}$	$-3.47\times {10}^{-1}$	$1.37\times {10}^{-1}$	$0.066$
Age40_Male	$2.36\times {10}^{-2}$†	$-4.13\times {10}^{-3}$	$4.94\times {10}^{-2}$	$0.028$	$1.48\times {10}^{-1}$†	$-9.64\times {10}^{-2}$	$3.81\times {10}^{-1}$	$0.040$
Age50_Male	$-3.64\times {10}^{-3}$	$-3.04\times {10}^{-2}$	$2.31\times {10}^{-2}$	$0.090$	$5.74\times {10}^{-2}$	$-1.62\times {10}^{-1}$	$3.04\times {10}^{-1}$	$0.071$
Age60_Male	$-1.83\times {10}^{-2}$†	$-4.57\times {10}^{-2}$	$9.06\times {10}^{-3}$	$0.045$	$-1.79\times {10}^{-1}$†	$-4.11\times {10}^{-1}$	$6.60\times {10}^{-2}$	$0.044$
Age70_Male	$-9.91\times {10}^{-3}$	$-3.84\times {10}^{-2}$	$1.75\times {10}^{-2}$	$0.078$	$-4.56\times {10}^{-2}$	$-2.79\times {10}^{-1}$	$2.04\times {10}^{-1}$	$0.097$
Other parameters
Intercept ($\alpha$)	$1.33\times {10}^{-2}$	$-5.84\times {10}^{-1}$	$5.98\times {10}^{-1}$	–	$1.20$	$-3.48\times {10}^{-1}$	$3.30$	–
Error s.d. ($\sigma$)	$2.26\times {10}^{-2}$	$1.88\times {10}^{-2}$	$2.79\times {10}^{-2}$	–	$2.01\times {10}^{-1}$	$1.61\times {10}^{-1}$	$2.46\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

,

Table A7. Posterior Summaries for Walking.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$-1.13\times {10}^{-1}$ *,†	$-1.60\times {10}^{-1}$	$-5.35\times {10}^{-2}$	$0.007$	$-3.59\times {10}^{-1}$ *,†	$-4.40\times {10}^{-1}$	$-2.84\times {10}^{-1}$	$0.004$
Work	$-5.59\times {10}^{-2}$ *,†	$-8.07\times {10}^{-2}$	$-3.23\times {10}^{-2}$	$0.018$	$-1.19\times {10}^{-1}$ *,†	$-1.63\times {10}^{-1}$	$-7.77\times {10}^{-2}$	$0.023$
Care	$-1.40\times {10}^{-2}$	$-5.04\times {10}^{-2}$	$2.08\times {10}^{-2}$	$0.084$	$-3.67\times {10}^{-2}$	$-1.15\times {10}^{-1}$	$4.47\times {10}^{-2}$	$0.088$
Age groups (reference: 20s)
Age30	$5.71\times {10}^{-2}$ *,†	$1.46\times {10}^{-2}$	$1.03\times {10}^{-1}$	$0.020$	$1.17\times {10}^{-1}$ *,†	$4.42\times {10}^{-2}$	$1.83\times {10}^{-1}$	$0.026$
Age40	$7.48\times {10}^{-2}$ *,†	$3.72\times {10}^{-2}$	$1.19\times {10}^{-1}$	$0.013$	$1.59\times {10}^{-1}$ *,†	$8.98\times {10}^{-2}$	$2.22\times {10}^{-1}$	$0.016$
Age50	$9.60\times {10}^{-2}$ *,†	$5.71\times {10}^{-2}$	$1.43\times {10}^{-1}$	$0.009$	$2.01\times {10}^{-1}$ *,†	$1.31\times {10}^{-1}$	$2.65\times {10}^{-1}$	$0.011$
Age60	$1.11\times {10}^{-1}$ *,†	$6.83\times {10}^{-2}$	$1.60\times {10}^{-1}$	$0.007$	$2.24\times {10}^{-1}$ *,†	$1.55\times {10}^{-1}$	$2.88\times {10}^{-1}$	$0.009$
Age70	$-5.05\times {10}^{-2}$ *,†	$-9.10\times {10}^{-2}$	$2.87\times {10}^{-3}$	$0.020$	$-1.52\times {10}^{-1}$ *,†	$-2.24\times {10}^{-1}$	$-9.53\times {10}^{-2}$	$0.016$
Survey years (reference: 2006)
Year2011	$5.59\times {10}^{-3}$	$-1.34\times {10}^{-2}$	$2.43\times {10}^{-2}$	$0.181$	$1.06\times {10}^{-2}$	$-2.68\times {10}^{-2}$	$4.75\times {10}^{-2}$	$0.198$
Year2016	$7.06\times {10}^{-2}$ *,†	$5.18\times {10}^{-2}$	$8.97\times {10}^{-2}$	$0.013$	$1.79\times {10}^{-1}$ *,†	$1.40\times {10}^{-1}$	$2.15\times {10}^{-1}$	$0.013$
Year2021	$1.02\times {10}^{-1}$ *,†	$8.36\times {10}^{-2}$	$1.21\times {10}^{-1}$	$0.007$	$2.61\times {10}^{-1}$ *,†	$2.23\times {10}^{-1}$	$2.97\times {10}^{-1}$	$0.007$
Interaction terms
Work_Care	$1.86\times {10}^{-2}$	$-2.63\times {10}^{-2}$	$6.23\times {10}^{-2}$	$0.066$	$4.82\times {10}^{-2}$	$-5.44\times {10}^{-2}$	$1.50\times {10}^{-1}$	$0.068$
Male_Work	$-1.82\times {10}^{-2}$	$-5.03\times {10}^{-2}$	$1.38\times {10}^{-2}$	$0.086$	$-6.02\times {10}^{-2}$	$-1.29\times {10}^{-1}$	$5.23\times {10}^{-3}$	$0.061$
Male_Care	$-9.67\times {10}^{-3}$	$-5.32\times {10}^{-2}$	$3.23\times {10}^{-2}$	$0.085$	$-2.40\times {10}^{-2}$	$-1.33\times {10}^{-1}$	$7.58\times {10}^{-2}$	$0.087$
Age30_Male	$-5.51\times {10}^{-3}$	$-5.79\times {10}^{-2}$	$4.87\times {10}^{-2}$	$0.088$	$3.07\times {10}^{-2}$	$-6.32\times {10}^{-2}$	$1.30\times {10}^{-1}$	$0.087$
Age40_Male	$1.70\times {10}^{-2}$	$-3.55\times {10}^{-2}$	$6.92\times {10}^{-2}$	$0.062$	$1.28\times {10}^{-1}$†	$3.44\times {10}^{-2}$	$2.25\times {10}^{-1}$	$0.023$
Age50_Male	$4.61\times {10}^{-2}$†	$-4.88\times {10}^{-3}$	$9.81\times {10}^{-2}$	$0.024$	$2.20\times {10}^{-1}$ *,†	$1.31\times {10}^{-1}$	$3.15\times {10}^{-1}$	$0.010$
Age60_Male	$9.01\times {10}^{-2}$ *,†	$3.60\times {10}^{-2}$	$1.43\times {10}^{-1}$	$0.009$	$3.24\times {10}^{-1}$ *,†	$2.35\times {10}^{-1}$	$4.18\times {10}^{-1}$	$0.005$
Age70_Male	$1.90\times {10}^{-1}$ *,†	$1.32\times {10}^{-1}$	$2.45\times {10}^{-1}$	$0.003$	$5.57\times {10}^{-1}$ *,†	$4.68\times {10}^{-1}$	$6.49\times {10}^{-1}$	$0.002$
Other parameters
Intercept ($\alpha$)	$2.90\times {10}^{-2}$	$-5.81\times {10}^{-1}$	$6.60\times {10}^{-1}$	–	$2.42$	$-1.11\times {10}^{-1}$	$4.55$	–
Error s.d. ($\sigma$)	$3.71\times {10}^{-2}$	$3.15\times {10}^{-2}$	$4.29\times {10}^{-2}$	–	$9.02\times {10}^{-2}$	$7.85\times {10}^{-2}$	$1.02\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

and

Table A8. Posterior Summaries for Gym equipment.

Variable	Participation Rate				Annual Participation Days
	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$	Mean	2.5% HPD	97.5% HPD	$\overline{\mathit{\kappa }}$
Gender, work, and caregiving
Male	$8.51\times {10}^{-2}$ *,†	$6.18\times {10}^{-2}$	$1.07\times {10}^{-1}$	$0.004$	$5.09\times {10}^{-1}$ *,†	$2.90\times {10}^{-1}$	$7.67\times {10}^{-1}$	$0.006$
Work	$3.34\times {10}^{-4}$	$-1.09\times {10}^{-2}$	$1.09\times {10}^{-2}$	$0.175$	$5.48\times {10}^{-3}$	$-1.12\times {10}^{-1}$	$1.10\times {10}^{-1}$	$0.167$
Care	$7.90\times {10}^{-3}$	$-1.12\times {10}^{-2}$	$2.75\times {10}^{-2}$	$0.081$	$5.14\times {10}^{-2}$	$-1.15\times {10}^{-1}$	$2.24\times {10}^{-1}$	$0.089$
Age groups (reference: 20s)
Age30	$-1.50\times {10}^{-2}$†	$-3.43\times {10}^{-2}$	$4.80\times {10}^{-3}$	$0.044$	$-9.01\times {10}^{-2}$	$-2.79\times {10}^{-1}$	$1.02\times {10}^{-1}$	$0.053$
Age40	$-7.75\times {10}^{-3}$	$-2.63\times {10}^{-2}$	$1.06\times {10}^{-2}$	$0.079$	$-2.46\times {10}^{-2}$	$-2.21\times {10}^{-1}$	$1.64\times {10}^{-1}$	$0.093$
Age50	$-6.78\times {10}^{-3}$	$-2.61\times {10}^{-2}$	$1.16\times {10}^{-2}$	$0.083$	$-3.77\times {10}^{-2}$	$-2.27\times {10}^{-1}$	$1.70\times {10}^{-1}$	$0.085$
Age60	$-2.03\times {10}^{-2}$ *	$-3.97\times {10}^{-2}$	$-1.49\times {10}^{-3}$	$0.031$	$-1.53\times {10}^{-1}$†	$-3.44\times {10}^{-1}$	$6.43\times {10}^{-2}$	$0.030$
Age70	$-7.18\times {10}^{-2}$ *,†	$-9.21\times {10}^{-2}$	$-5.25\times {10}^{-2}$	$0.005$	$-7.09\times {10}^{-1}$ *,†	$-9.01\times {10}^{-1}$	$-4.66\times {10}^{-1}$	$0.004$
Survey years (reference: 2006)
Year2011	$-9.28\times {10}^{-3}$ *	$-1.86\times {10}^{-2}$	$-1.76\times {10}^{-4}$	$0.079$	$-3.74\times {10}^{-2}$	$-1.32\times {10}^{-1}$	$5.01\times {10}^{-2}$	$0.127$
Year2016	$4.20\times {10}^{-2}$ *,†	$3.23\times {10}^{-2}$	$5.06\times {10}^{-2}$	$0.011$	$3.75\times {10}^{-1}$ *,†	$2.83\times {10}^{-1}$	$4.62\times {10}^{-1}$	$0.009$
Year2021	$2.64\times {10}^{-2}$ *,†	$1.73\times {10}^{-2}$	$3.54\times {10}^{-2}$	$0.022$	$2.71\times {10}^{-1}$ *,†	$1.82\times {10}^{-1}$	$3.64\times {10}^{-1}$	$0.015$
Interaction terms
Work_Care	$2.28\times {10}^{-3}$	$-2.18\times {10}^{-2}$	$2.44\times {10}^{-2}$	$0.089$	$9.25\times {10}^{-3}$	$-2.08\times {10}^{-1}$	$2.12\times {10}^{-1}$	$0.086$
Male_Work	$-9.40\times {10}^{-3}$	$-2.52\times {10}^{-2}$	$5.97\times {10}^{-3}$	$0.085$	$-1.01\times {10}^{-1}$	$-2.50\times {10}^{-1}$	$5.24\times {10}^{-2}$	$0.070$
Male_Care	$-7.99\times {10}^{-3}$	$-3.18\times {10}^{-2}$	$1.56\times {10}^{-2}$	$0.076$	$-7.83\times {10}^{-2}$	$-2.80\times {10}^{-1}$	$1.22\times {10}^{-1}$	$0.071$
Age30_Male	$-3.99\times {10}^{-2}$ *,†	$-6.41\times {10}^{-2}$	$-1.44\times {10}^{-2}$	$0.013$	$-1.69\times {10}^{-1}$†	$-4.04\times {10}^{-1}$	$5.46\times {10}^{-2}$	$0.042$
Age40_Male	$-6.46\times {10}^{-2}$ *,†	$-8.97\times {10}^{-2}$	$-4.13\times {10}^{-2}$	$0.006$	$-3.47\times {10}^{-1}$ *,†	$-5.70\times {10}^{-1}$	$-1.10\times {10}^{-1}$	$0.012$
Age50_Male	$-8.95\times {10}^{-2}$ *,†	$-1.14\times {10}^{-1}$	$-6.58\times {10}^{-2}$	$0.003$	$-5.18\times {10}^{-1}$ *,†	$-7.37\times {10}^{-1}$	$-2.74\times {10}^{-1}$	$0.006$
Age60_Male	$-8.82\times {10}^{-2}$ *,†	$-1.13\times {10}^{-1}$	$-6.41\times {10}^{-2}$	$0.003$	$-5.21\times {10}^{-1}$ *,†	$-7.51\times {10}^{-1}$	$-2.74\times {10}^{-1}$	$0.006$
Age70_Male	$-6.71\times {10}^{-2}$ *,†	$-9.26\times {10}^{-2}$	$-4.29\times {10}^{-2}$	$0.005$	$-2.61\times {10}^{-1}$ *	$-5.23\times {10}^{-1}$	$-1.91\times {10}^{-2}$	$0.022$
Other parameters
Intercept ($\alpha$)	$8.92\times {10}^{-3}$	$-5.84\times {10}^{-1}$	$6.02\times {10}^{-1}$	–	$9.38\times {10}^{-1}$	$-3.83\times {10}^{-1}$	$2.94$	–
Error s.d. ($\sigma$)	$2.00\times {10}^{-2}$	$1.72\times {10}^{-2}$	$2.30\times {10}^{-2}$	–	$1.80\times {10}^{-1}$	$1.49\times {10}^{-1}$	$2.17\times {10}^{-1}$	–

Note: * indicates that the 95% HPD interval does not include zero. ^† indicates variables with $\overline{\kappa }<0.05$.

) shows posterior summary statistics for each sport.

3.1. Effects of Gender, Employment, and Caregiving

Gender is strongly associated with participation behavior in many sports. For baseball, fishing, gym use, and golf, the Male coefficient was positive, with participation rates higher by 5 to 17 percentage points. Annual participation days also showed positive effects, suggesting more frequent participation among men than among women.

Walking showed a different pattern. The Male coefficient for participation rate was negative, implying that men are about 11 percentage points less likely to participate than women. Annual participation days for walking were also lower for men.

Across most sports, the main effects of Work and Care were relatively small. However, for walking, both Work and Care were associated with lower participation rates and fewer participation days. For other sports, limited effects of employment and caregiving were observed.

3.2. Effects of Age Groups

Age effects varied substantially across sports. For baseball, swimming, bowling, and fishing, consistently negative age-group dummy variables were found from the 30s onward, indicating that participation declines with age. An example can be observed in baseball, where the participation rate for Age50 was about 5 percentage points lower than that for individuals in their 20s, with further declines seen for Age60 and older groups. Annual participation days showed similar patterns, with pronounced reductions among older age groups.

Meanwhile, for walking, increasing participation rates were found from Age30 to Age60. For Age60, the participation rate was approximately 11 percentage points higher than that for individuals in their 20s, with annual participation days also increasing. However, a reduction in participation rate and participation days was found for Age70.

3.3. Effects of Survey Years

Survey-year dummy variables captured temporal changes in sports participation. For Year2011 and Year2016, the magnitude of changes was generally modest despite differing effects by sport.

In contrast, Year2021 was associated with pronounced changes across sports. For baseball, swimming, bowling, fishing, and golf, participation rates declined by approximately 1 to 11 percentage points compared with 2006. A substantial decrease in annual participation days, by roughly 10–60%, was also observed. Conversely, for walking, participation rates were about 10 percentage points higher in Year2021 compared with 2006, with annual participation days also increasing.

3.4. Interaction Effects

Age–gender interaction terms (Age_Male) were statistically meaningful for many sports. For baseball, swimming, and bowling, older age groups consistently showed negative coefficients, suggesting a weakening of the positive main effect of Male with age. In baseball, for instance, a reduction in participation rates and annual participation days for men was observed from Age50_Male onward.

Meanwhile, positive Age_Male coefficients were seen for golf, with higher participation rates and participation days among men aged 50 and older. For walking, Age_Male coefficients were negative for younger age groups but became positive for middle-aged and older groups.

For other interaction terms, such as Work_Care, Male_Work, and Male_Care, small coefficients were found across sports, suggesting limited interaction effects.

4. Discussion

This study examined how socioeconomic attributes such as gender, age, employment status, and caregiving responsibilities influence participation rates and annual participation days across sports disciplines in Japan. The above findings revealed pronounced gender differences in participation behavior, varying age effects by sport type, distinctive changes in 2021 corresponding to the COVID-19 pandemic, and age–gender interaction effects. The following sections discuss these results in relation to extant research.

4.1. Gender Differences in Participation Behavior

For sports such as baseball, fishing, gym use, and golf, participation rates and annual participation days were higher among men than among women, suggesting a strong link between gender and participation behavior in these activities. Meanwhile, a different pattern was observed for walking, where participation rates and participation days were lower among men than among women. These findings are consistent with international evidence regarding men’s tendency to participate more in competitive and leisure-oriented sports as opposed to women’s preference for health-oriented and aesthetic forms of physical activity (Peral-Suárez et al. 2020).

The advantage observed in walking among women specifically suggests their higher likelihood of selecting health-oriented and socially engaging forms of physical activity, a pattern that aligns with the literature (Ma et al. 2025). Meanwhile, men are often more strongly oriented toward competition; hence, once they disengage from sports because of retirement or health reasons, they may find it more challenging to find alternative forms of physical activity. This is consistent with longitudinal evidence obtained in Poland by Biernat et al. (2019), who observed a decline in men’s sports and exercise participation two to four years after retirement and found that being concerned about one’s health did not necessarily translate into behavioral change. However, studies involving Japanese cohorts have reported an association between retirement and increased leisure-time physical activity (Oshio and Kan 2017).

These gender differences in sports and exercise habits are less likely influenced by individual preferences than by gender norms and socially constructed roles. Such patterns may also vary across countries, occupations, and social security systems, highlighting that gender disparities in sports participation must be understood through institutional and cultural contexts.

4.2. Effects of Employment and Caregiving

This study revealed relatively small main effects of employment (Work) and caregiving (Care) across most sports. For walking, however, both factors significantly reduced participation rates and annual participation days, indicating that for individuals under strong time constraints, such as workers and caregivers, even walking can be difficult to maintain despite being a low-intensity and easily accessible form of physical activity.

This interpretation aligns with studies involving caregivers of patients with dementia, who consider the time scarcity, physical and mental fatigue, and sense of self-sacrifice associated with caregiving to be major barriers to sustained physical activity (Sewerbridges-Williams et al. 2025). Therefore, developers of interventions promoting sports and physical activity among workers and caregivers may find it more effective to frame participation not solely in terms of “exercise for health” but also in terms of incorporating goals such as short-duration activities, stress relief, and opportunities for social interaction.

4.3. Age Effects and Sport-Specific Withdrawal

The direction of the observed age effects in this study differed depending on the sport. Whereas participation rates declined with age in competitive and group sports such as baseball, swimming, bowling, and fishing, participation in walking increased among middle-aged and older adults. These are consistent with systematic reviews by Jenkin et al. (2017) and Stenner et al. (2020), who found a polarization between retirees who maintained an exercise habit and those who did not. The current results also align with Ejiri et al. (2022), who observed that new participation in sports groups is more common among women, individuals already performing light physical activity, and those with broader social networks.

This study also found a decline in both participation rates and participation days among individuals in their 70s and older, corresponding with the influence of physical anxiety and social isolation observed by Suryadi et al. (2024). These findings suggest that age-related changes in sports and exercise habits are not merely driven by the decline in one’s physical capacity but are also strongly influenced by their life-course transitions, such as employment, retirement, and caregiving, and by reconfigured social ties.

4.4. The Impact of COVID-19

In 2021, a reduction in participation rates and annual participation days was seen across many sports, including baseball, swimming, bowling, fishing, and golf; walking was the only activity that showed an increase. This aligns with Maugeri et al. (2020) and Park et al. (2022), who documented a reduction in group and indoor sports during the pandemic alongside a shift toward outdoor and individual forms of physical activity.

These changes reflect some behavioral flexibility under crises and suggest that easily accessible and noncompetitive activities, such as home-based exercise, online fitness classes, and active video games, may provide more socially inclusive options for maintaining physical activity during periods of social disruption.

4.5. Age–Gender Interactions

Statistically significant interactions were revealed between age and gender across many sports. For baseball, swimming, and bowling, males’ advantage in participation diminished as they aged, whereas for golf, gender disparities widened among middle-aged and older men. These findings are consistent with Roh and Chang (2025), who observed a higher likelihood among older men to maintain sports participation using their economic resources and social networks.

With regard to walking, participation rates were relatively low among younger men but increased in middle-aged and older groups to levels comparable with, or even exceeding, those of women. This likely reflects a life-stage transition effect in which motivations for physical activity shift from competition to health maintenance in response to retirement or growing health concerns (Inui et al. 2022).

Studies on Japanese populations have also shown that health is positively associated with self-esteem and psychological well-being (Nakakita et al. 2025a, 2025b). Hence, promoting sports and physical activity during changes in employment, caregiving responsibilities, aging, and other key life transitions can play a vital role in supporting both physical health and psychological well-being. The patterns identified in this study suggest that continued sports engagement must be ensured across these transitions so that individuals can maintain their self-esteem and overall well-being. These findings highlight not only the dynamic nature of gender differences in sports participation across the life course but also their broader significance from a health and well-being perspective.

4.6. Limitations, Policy Implications, and Future Research

4.6.1. Limitations

This study incorporated four waves of aggregated data from the Statistics Bureau of Japan’s Survey on Time Use and Leisure Activities. It highlighted population segments as its unit of analysis, defined by combinations of gender, age group, employment status, and caregiving status. Despite the advantages of this research design in terms of privacy protection and population representativeness relative to individual-level microdata, several methodological limitations must be discussed.

First, because the unit of analysis was a population segment (gender × age × employment × caregiving × year) rather than individuals, the estimated results reflected average segment-level tendencies. This prevented a direct analysis of individual-level variability and heterogeneity, such as differences in health status or social relationships within the same age group.

Second, the data were composed of repeated cross-sectional aggregates rather than a true panel structure. Despite the inclusion of year dummy variables to capture temporal trends, this design limited causal inference and did not allow for a rigorous identification of the impact of specific social shocks such as the COVID-19 pandemic.

Third, to ensure statistical stability, this study excluded segments with fewer than 10 respondents or with an estimated population size below 10,000. As a result, extremely small or highly specific social groups (e.g., older men who perform intensive caregiving) were not represented in the analysis. Therefore, the findings reflected representative patterns among statistically stable population segments and may underrepresent minority groups or atypical behavioral patterns.

Fourth, this study normalized sampling weights within each survey year to adjust for differences in sampling precision, which improves comparability across segments. However, this approach may also homogenize variance structures across years and underestimate true year-to-year population fluctuations.

Taken together, the current findings should be interpreted as illustrative of medium-term sports participation patterns by age, gender, and social roles rather than demonstrative of causal mechanisms or individual behavioral heterogeneity.

4.6.2. Policy Implications

Four main directions for sports and health policy may be deduced from the results.

First, policy design should be sensitive to life-stage differences. Sports participation determinants vary by age, gender, and social responsibilities: younger men tend to be motivated by competitive achievement and recognition whereas middle-aged and older women tend to be driven by health maintenance and social connections. However, it is important to note that these observed gender differences likely reflect socially constructed gender roles and cultural expectations rather than innate biological differences in preference. Sports policies should therefore avoid reinforcing existing gender stereotypes and instead focus on dismantling structural and cultural barriers that constrain participation choices. These findings suggest the need to develop targeted approaches that account for age, gender, and life circumstances. For instance, male retirees and female childcare practitioners or caregivers may benefit from flexible and low-burden programs.

Second, structural support must be ensured to reduce social constraints on participation. Employment, caregiving, and similar social roles often obstruct sustained sports engagement. Therefore, continued participation must be driven by policies that minimize these constraints, including workplace-based sports opportunities, government-sponsored short-duration exercise programs for family caregivers, and improved access to nearby facilities.

Third, support mediated by social connections is central to sustained sports participation, which is strongly facilitated by shared activities, interpersonal relationships, and a sense of role or belonging. Hence, sports policies should move beyond health education alone and be restructured as social participation initiatives that foster interpersonal connections. Support for community sports clubs and volunteer-based activities may also help reduce health inequalities.

Finally, beyond initial participation, re-engagement of lapsed participants represents an important and underutilized policy lever. The present findings—specifically the pronounced decline in participation with age and the dropout of female adolescents from competitive sports—highlight the need for targeted re-engagement initiatives. older adults can be effectively re-engaged through age-appropriate programs that prioritize social interaction and health improvement (Jenkin et al. 2021). For female adolescents, re-engagement strategies must be clearly distinguished from general recruitment efforts and designed around the specific barriers young women face, including social pressures, limited access to same-gender role models, and exclusionary club cultures (Kay et al. 2025). Developing such targeted re-engagement pathways rather than relying solely on retention of current participants may prove essential for reducing long-term sports participation disparities in Japan.

4.6.3. Future Research

This study recommends three main directions for future research.

First, studies must conduct longitudinal analysis using individual-level data to better understand changes in sports habits in relation to retirement, caregiving onset, health transitions, and other life events. Specifically, researchers may investigate interactions among employment, caregiving responsibilities, and health status to substantially strengthen the empirical basis for evidence-informed policy design.

Second, research must include detailed residential or geographic information to increase analytical precision. By incorporating data on individuals’ living environments, such as neighborhood characteristics, accessibility to sports facilities, and local social resources, studies can provide a more comprehensive analysis of how contextual factors shape sports participation behaviors beyond individual-level attributes.

Third, scholars must examine why behavioral changes have persisted in the post-COVID-19 period. A key challenge here is determining whether the increase in walking and the decline in competitive sports observed in this study represent temporary adaptations or lasting transformations. This may be addressed by developing behavioral models that integrate evolving risk perceptions, the diffusion of remote work, and the reorganization of community structures, which will be crucial for understanding long-term shifts in sports participation.

5. Conclusions

Using data from Japan’s Survey on Time Use and Leisure Activities, this study examined sports participation behavior at the population-segment level to identify its socioeconomic drivers. The results highlighted the combined influences of gender, age, social roles, and historical context on the sports participation structure.

First, gender effects were clearly sport-specific. Men displayed higher participation rates and annual participation days in competitive and leisure-oriented activities such as baseball, gym use, golf, and fishing, whereas women showed higher participation rates in walking. These findings align with international evidence characterizing men as more competition-oriented and women as more health- and socially oriented in their activity preferences. In addition, despite the sharp decline in men’s participation in mid- to later life, older men tend to show continued or increased participation in activities such as golf, which are often embedded in social capital. This indicates that sports participation is not merely an individual health behavior but a selective activity closely associated with social and economic resources.

Second, age effects differed substantially across sports. Participants’ engagement in organized sports such as baseball, swimming, and bowling declined as they aged, whereas participation in walking increased markedly among middle-aged and older adults. This pattern corresponds with the observed polarization of sports participation around the retirement transition. The higher participation rate in walking among older adults reflects a shift toward noncompetitive activities focusing on health maintenance and social interaction, suggesting an evolution of the social meaning of sports participation over the life course.

Third, despite the limited overall main effects of employment and caregiving, significant negative effects were observed for walking, suggesting that time constraints linked to paid work and family caregiving can inhibit participation even in low-intensity activities. Specifically, women with caregiving responsibilities experience fewer opportunities for physical activity because of their prioritization of household roles, reflecting the consistently salient gendered cultural norms in the Japanese context.

Fourth, in 2021, which corresponds to the COVID-19 pandemic period, participation rates and participation days declined across many competitive and group sports, while walking increased. This divergence illustrates a behavioral restructuring process aimed at balancing infection risk avoidance with health maintenance, highlighting the adaptability of physical activity behaviors during social crises.

Overall, sports participation patterns are influenced not solely by biological sex differences but also by the interaction among social attributes such as gender roles, employment, caregiving, and life-stage transitions within specific cultural and institutional contexts. This highlights the urgency of extending sports promotion policies beyond narrow health-promotion objectives toward comprehensive social policies that limit social constraints and support diverse motivations for participation across different life stages.

To allow for a causal analysis of dynamic changes in sports participation, studies should integrate individual-level longitudinal data with information on regional social resources. Research should also assess the persistence of behavioral changes observed during the post-COVID-19 period to clarify how new forms of sports participation may enhance social inclusion. By using nationally representative data to empirically demonstrate the relation between socioeconomic attributes and sports participation, this study offers a novel perspective that conceptualizes sports not only as a health behavior but also as a form of social participation. In doing so, it provides foundational evidence for integrating sports policy with broader societal goals such as health equity, social welfare, gender equality, and community cohesion.