The health benefits of physical activity (PA) for children and adolescents are well established.1,2 These benefits include improved bone health, weight status, cognitive function, and cardiometabolic health.2 Additionally, PA during childhood has been shown to track to adulthood.3 Thus, increasing childhood PA is important for decreasing the risk of chronic conditions, like heart disease and many types of cancer, later in life.2 Despite this, over three-quarters of children and adolescents in the United States (U.S.), and globally, are insufficiently active based on the U.S. and World Health Organization PA guidelines.1,4–6
Improving population health PA outcomes for children requires an understanding of the community systems in which they live, learn, grow, and play. Communities can be characterized as “wellness landscapes” of behavior settings, defined as the geospatially and temporally bound social and physical environments in which people live their daily lives.7 For children, these settings include the adult-led organized groups in which they spend their time, such as school classrooms and youth club groups. Such settings are dynamic social systems, and the interactions within setting environments (eg, children interacting with other children) produce PA outcomes. Thus, an interdependency exists such that setting PA outcomes are dependent upon the interactions among individuals within group setting environments, rather than an aggregation of individuals behaving independently.8–11 Focusing on organized group settings for understanding and improving PA outcomes for children is important because of their high reach, as nearly all children participate in school,12 and estimates suggest over two-thirds of children participate in organized activities outside of school.13 Additionally, although PA can occur outside of organized group settings, such as in unstructured outdoor play, a shift to participation in organized settings has contributed to a decline in children’s unstructured time.14–17
Process control systems theory defines data monitoring and feedback as critical functions for understanding and improving system outcomes.18,19 Data collected can be used to guide decision making and aid in understanding dynamic interactions and system outcomes over time. Accordingly, the importance of collecting data on PA indicators to inform efforts for improving community population PA outcomes is increasingly being recognized.20–22 While large-scale surveillance systems can provide information on global-, national-, and state-level PA outcomes,20,23 data on more granular scales are needed to allow communities to access relevant information for understanding PA in their communities and informing decision making locally.24,25 Incorporating data from organized group settings for children into routine community monitoring efforts can help communities understand dynamic child PA outcomes and factors constraining and promoting PA within these setting systems.11
To understand variability in children’s PA and time periods when children are more and less active, previous research has examined PA by segmenting time and by social system characteristics, such as weekday and weekend PA,26–28 morning and afternoon PA,29,30 and PA during specific lessons (eg, physical education, recess).31 Within organized group settings, momentary time sampling with preestablished time intervals (eg, 10-s observation and record intervals) is often used to examine PA by social system characteristics and to determine averages across the entire group meeting, that is, duration of setting time, defined by meeting start and stop points.32–34 However, continuous time sampling, in which meetings are segmented into smaller stable units (ie, sessions) with start and stop points defined by naturally occurring changes in the social system, allows for a more granular examination of PA patterns and can illuminate interacting factors influencing PA variability across time.7,11,35,36 Studies have illustrated time segmentation of group setting meetings and variability in group PA based on social system changes in location (eg, indoors, outdoors),30,37 task/purpose (eg, academic, PA),35,37–41 and group member arrangement (eg, small group, whole group).30,35 This demonstrates that different social system structures within group meetings produce different PA outcomes. However, despite the importance of monitoring setting systems on smaller time scales to understand the influence of meeting routines on PA,11 examination of the influence of naturally occurring setting system structures on PA across multiple time scales (ie, meeting and session duration) within in-school and out-of-school setting time for children remains relatively limited.
Existing studies examining children’s PA during setting time predominantly measure group outcomes by averaging PA among children across meeting and session duration (eg, average minutes per hour of moderate to vigorous PA [MVPA]).30,38,41,42 Although beneficial metrics, using the mean and variability around the mean as the sole metrics for understanding setting PA may obscure important within-group heterogeneity43 and may mask how PA is distributed among children during setting time, such as whether all children accrue PA or whether inequalities exist and only a subset of one or a few individuals accrues PA. Additionally, PA outcomes are typically skewed44,45 and, within setting time, are highly dependent on the interactions within these environments.8–11,44 For example, a basketball game allows for a limited number of children to participate in the game at a given time (ie, limited playing time), which creates a dependency in which one child’s ability to play is impacted by whether other children are playing in the game. The game may be structured such that only a subset of children present in the setting can participate, thus accruing all of the PA while others have limited PA, creating within-group inequalities. The game may also be structured such that all children have an equal amount of playing time and PA. These 2 scenarios result in different distributions but may have the same mean PA among children in the setting. Rather than use traditional approaches, such as transforming data to fit a normal distribution and removing outliers, it is important to study these distributions and how PA outcomes are influenced by the social structure of setting time.
Measuring PA inequality, examined here as the extent to which PA is distributed unequally across individuals,46 allows for characterizing differences in group-level PA distributions and whether the structure of setting time influences inequalities in how PA is distributed. Numerous calls have been made for additional metrics that are informative about the distribution of and inequalities in outcomes within and between populations to advance public health monitoring efforts,24,47–49 and understanding population health PA outcomes of a group of individuals requires an understanding of the distribution of such outcomes within the group.50 Further, examining inequalities in the distribution of health outcomes has been highlighted as an important step for moving toward understanding unfair distributions and reducing health disparities.51
The Gini coefficient, a popular measure of distribution and inequality,52–55 can be quantified at multiple scales of granularity to characterize community and group setting system outcomes, including outcomes that do not approach a normal distribution.55 The Gini coefficient is typically used to describe income inequality,46,55 but has been used in numerous disciplines, including chemistry, biology, and ecology, and is applicable to a variety of outcomes and any size distributions.56 Related to PA, the coefficient has most commonly been used to examine the influence of income inequality on PA,57 though a few studies have used the metric to characterize inequality in PA outcomes.49,58–61 For instance, Althoff et al58 studied country-level PA inequality among adults, measured by the Gini coefficient for step counts, and found inequality in the distribution of steps within countries. They found that PA inequality was associated with higher obesity prevalence and that this inequality was a better correlate of obesity prevalence than average steps.58 Examining how outcomes are distributed within populations can provide a better understanding of system-level outcomes. Inequality metrics, such as the Gini coefficient, characterizing the distribution of PA among children, have not yet been applied to understand inequality in PA within school, youth club, and before-/after-school settings.
A greater understanding of inequalities in the distribution of PA among children and how social system structure, defined as session purpose, influences PA inequality within and across organized group setting time is needed. The primary aims of this study were to examine (1) PA inequality in organized group setting meetings for children and the influence of setting type on meeting time PA inequality and (2) the influence of time-segmented session purpose on PA inequality during group meetings. A secondary, exploratory aim was to examine whether inequalities exist within demographic subgroups by gender and socioeconomic status and the influence of session purpose on PA inequality within these subgroups. We hypothesized that PA inequality would exist in all settings and at the meeting and session timescales. We also hypothesized that PA inequality would differ by session purpose, specifically that sessions with a purpose of free play would have lower inequality compared with organized PA sessions.
Methods
Project Design
This cross-sectional study was a substudy of Wellscapes, a 2-wave staggered-start hybrid effectiveness-implementation62 community randomized controlled trial targeting whole-of-community PA promotion among children (ClinicalTrials.gov Identifier: NCT03380143). Two rural communities in Nebraska with a concentration of primarily non-Hispanic, White children were recruited for participation in wave 1 (fall 2018–spring 2020), reported in the present study. Communities were eligible if they were in a rural area at least 10 miles from an urbanized area and had only one public high school. Population and distance classifications were drawn from the Department of Education Rural and Low-Income School Program based on the National Center for Education Statistics.63,64 Study procedures were approved by the University of Nebraska Medical Center Institutional Review Board (IRB #446-18-EP, IRB #439-18-EX).
Settings and Participants
The settings, participants, and data collection procedures have been fully described elsewhere.11 Briefly, a sample of school classrooms, youth club groups, and before-/after-school programs for third- through sixth-grade children from the 2 participating communities were recruited to participate in video observation and accelerometer data collections during their usual group meetings. We used a volunteer sample of one classroom from each target grade in the public elementary/middle school organization in each community. We identified out-of-school groups through internet searching and with the assistance of community members, and leaders of identified groups were invited through direct contact to participate in the study.11 We included a maximum of 16 groups in each community each season. Data collection procedures were adopted as normal educational practice, and data were de-identified. Thus, once a group leader agreed to participate in the study, all children in their group were eligible to participate in the video observation and accelerometer data collections. All children in participating groups were also recruited to provide active parent/guardian informed consent to have their group setting data identified and linked with their demographic information from school enrollment records provided by the schools to the research team.11
Measures
Demographics
Free and reduced lunch status (FRLS) was obtained from school enrollment records for each child with parent/guardian informed consent and was dichotomized into free/reduced and full pay lunch status (FPLS). For children with consent, gender, dichotomized as male/female, was recorded from school enrollment records. For children who did not provide consent, gender was based on group make-up for single-gender groups (eg, Boy Scouts). In fewer than 15% of cases, gender was recorded based on researcher observation following previously validated observation protocols.65,66
Video Observation
Meeting routines were recorded using an Apple iPad Mini 4, while the primary adult leader wore a microphone to capture audio. Trained research assistants (RAs) coded recorded videos using the Geosocial Observation System.35,36 Meetings were time segmented into sessions, or units with defined start and stop points, based on changes in purpose, location, or group participants (ie, >50% of participants change or leave), and each session was assigned a mutually exclusive code for purpose (Table 1).
Session Purpose Codes and Definitions
| Session purpose | Definition |
|---|---|
| Academic | Session led by adult leader of the group and designed for enhancement of academic knowledge (eg, math lesson during school day) |
| Enrichment | Session led or supervised by adult leader of the group and designed to build skills and improve a particular knowledge base but not designed for enhancement of academic knowledge (eg, fire safety lesson during youth club meeting) |
| Nonactive recreation | Session led or supervised by adult leader of the group but no physical activity and no intention to build skills (eg, playing board games) |
| Organized physical activity | Session with physical activity and involving close supervision of an adult with procedural aspects like rules/boundaries/expectations (eg, adult supervises a dance activity) |
| Free play physical activity | Session with physical activity remotely supervised by adult with no rules/boundaries/expectations (eg, children self-organize into a basketball game, recess) |
| Snack | Time devoted to eating |
| Academic-physical activity | Session with a primary academic purpose and a secondary physical activity purpose (eg, children are asked to move while answering math problems) |
| Enrichment-physical activity | Session with a primary enrichment purpose and a secondary physical activity purpose |
| No purpose | Adult supervised time, but no activity or free play occurring (eg, adult leader is collecting permission forms for field trip) |
Physical Activity
PA was measured using ActiGraph wGT3X-BT accelerometers. ActiLife software (ActiGraph) was used to initialize accelerometers to collect data in 15-second epochs67,68 beginning 1 hour before each scheduled meeting start time. Children wore accelerometers on their right hip, attached by an elastic belt, for each observed meeting duration. Following data collection, cut points established by Evenson et al69 were used to determine time spent in MVPA (≥2296 counts/min), and data were paired with meeting and session time segments, as further described below.
PA Inequality
PA inequality was determined by the Gini coefficient. The Gini coefficient is derived from the Lorenz curve, which plots the cumulative percentage of an entity (eg, PA minutes) against the cumulative percentage of the population.53 A perfectly “equal” distribution is illustrated by a diagonal line, called the line of equality, where the first 20% of the population has 20% of the total PA minutes, the first 60% of the population has 60% of the total PA minutes, and so on. The Lorenz curve deviates from this line as inequality occurs, such as when the first 20% of the population has less than 20% of the total PA minutes (see Figure 1).53 The Gini coefficient is the percentage of the area between the line of equality and the Lorenz curve to the total area under the line of equality. The coefficient ranges from 0, or perfect equality in which PA is equally distributed, to 1, or complete inequality in which one individual has all of the PA.54
—Inequality in children’s PA during an organized PA session (A) and an enrichment session (B). MVPA indicates moderate to vigorous PA; PA, physical activity.
Citation: Journal of Physical Activity and Health 21, 9; 10.1123/jpah.2024-0053
The Gini coefficient was calculated using the equation proposed by Glasser54 to examine inequality during total meeting time and during each time-segmented session. Two Gini coefficients were calculated for each meeting and session: (1) a Gini coefficient from total activity counts of each child present and (2) a Gini coefficient from MVPA minutes. Additionally, to conduct demographic subgroup analyses, Gini coefficients were calculated for each time-segmented session for male, female, FRLS, and FPLS participants.
Procedures
RAs attended 3 meetings each fall and spring season for each group. The entire meeting duration was observed for out-of-school groups, while school classrooms were observed for the first half of the school day.11 Before the start of each meeting, RAs set up the iPad in a position to capture the group, gave the microphone to the primary adult leader, and recorded the video start time. They placed accelerometers on each assenting child and recorded accelerometer on time. At the conclusion of the meeting, RAs retrieved accelerometers and recorded accelerometer and video stop times. Following the observation, trained coders time-segmented the recorded meeting routine into sessions according to the method described above. A random subsample of 11% of each coders’ videos was coded by a gold standard RA to ensure at least 80% interrater reliability.
Data Reduction
Using a SAS macro, Evenson cut points69 were applied to accelerometer PA count data to determine the amount of time in MVPA. Gini coefficients from total activity counts and from MVPA minutes were calculated for each meeting and session. Therefore, each meeting had values for the Gini coefficient derived from total activity counts, the Gini coefficient derived from MVPA minutes, and setting type. Each session had values for the Gini coefficient derived from total activity counts, the Gini coefficient from MVPA minutes, and session purpose. Children predominantly had the same accelerometer wear time (ie, the observed meeting duration). However, in limited instances when a child was not present for more than 50% of a session time segment (eg, arrived late, left early), their accelerometer data for that session was removed.
Statistical Analysis
Analyses were conducted using SAS (version 9.4). Descriptive statistics summarized observed group, meeting, session, and child participant characteristics. Generalized estimating equations using a beta distribution in SAS PROC GLIMMIX were used to examine the influence of setting type on PA inequality with meeting as the unit of analysis (model 1) and the influence of session purpose on PA inequality with session as the unit of analysis (model 2). With meeting as the unit of analysis (model 1), correlation and clustering were addressed using a multilevel model with a random effect for meeting nested within group. With the session time segment as the unit of analysis (model 2), correlation and clustering for group, meeting, and time segment components were addressed using a multilevel model with a random effect for meeting-by-time segment nested within group. The session analysis was also conducted for each demographic subgroup to examine the influence of session purpose on PA inequality for male, female, FRLS, and FPLS participants. In each model, statistical significance was set at P < .05, and time-segment length and number of participants were included as covariates. The Kenward–Roger method, using a DDFM=KR statement, was used to ensure degrees of freedom were properly estimated. Mean estimates were converted from the model scale to the data scale using the ILINK function. The Benjamini–Hochberg procedure with a false discovery rate of 0.05 was applied in all analyses to adjust for multiple comparisons and control for type I errors.70
Results
Observed group, meeting, session, and participant characteristics are presented in Table 2. A total of 130 meetings, with 835 sessions nested within, were observed. As indicated in Tables 3 and 4, fewer sessions were included in MVPA minute and subgroup analyses due to missing Gini coefficients (eg, no time in MVPA during sessions, no males present in sessions). Mean meeting length was 164.2 minutes (SD = 86.7), and 6.5 sessions (SD = 3.7), on average, were observed per meeting. Mean session length was 25.7 minutes (SD = 27.5).
Observed Groups, Meetings, Sessions, and Participants
| N (%) | |
|---|---|
| Groups | |
| Overall | 30 |
| School | 16 (53.3) |
| Before-/after-school program | 5 (16.7) |
| Youth club | 9 (30.0) |
| Meetings | |
| Overall | 130 |
| School | 80 (61.5) |
| Before-/after-school | 20 (15.4) |
| Youth club | 30 (23.1) |
| Sessions | |
| Overall | 835 |
| Academic | 381 (45.6) |
| Enrichment | 62 (7.4) |
| Nonactive recreation | 35 (4.2) |
| Organized physical activity | 104 (12.5) |
| Free play physical activity | 62 (7.4) |
| Academic-physical activity | 119 (14.3) |
| Enrichment-physical activity | 17 (2.0) |
| Snack | 12 (1.4) |
| No purpose | 43 (5.1) |
| Participants | |
| Overalla | 699 |
| Consenteda | 588 (84.1) |
| Genderb | |
| Male | 269 (45.7) |
| Female | 319 (54.3) |
| Grade | |
| Third | 115 (19.6) |
| Fourth | 126 (21.4) |
| Fifth | 177 (30.1) |
| Sixth | 169 (28.7) |
| Unknown | 1 (0.2) |
| Race/ethnicity | |
| Non-Hispanic, White | 553 (94.1) |
| Hispanic and all other races | 32 (5.4) |
| Unknown | 3 (0.5) |
| Lunch status | |
| Free/reduced | 160 (27.2) |
| Full pay | 425 (72.3) |
| Unknown | 3 (0.5) |
aChildren who participated in multiple observed groups (eg, a classroom and a youth club group) are counted for each group in which they participated. bGender was recorded for all children during observations. However, as accelerometer data for children without consent remained de-identified, demographics are only reported for the sample with informed consent.
Inequality in PA for Meetings and Sessions by Type
| Time segment | Total activity counts, LS mean (95% CI) | Differences (P < .05) | MVPA minutes, LS mean (95% CI) | Differences (P < .05) |
|---|---|---|---|---|
| Meeting, Gini coefficient | N = 130 | N = 130 | ||
| Setting type | P = .0005 | P = .0005 | ||
| a. School | 0.22 (0.20–0.24) | c* | 0.34 (0.30–0.38) | b,* c* |
| b. Before-/after-school | 0.25 (0.20–0.30) | c* | 0.21 (0.16–0.27) | a* |
| c. Youth club | 0.17 (0.14–0.20) | a,* b* | 0.18 (0.14–0.23) | a* |
| Session, Gini coefficient | N = 835 | N = 774 | ||
|---|---|---|---|---|
| Session purpose | P = .0042 | P = .0042 | ||
| a. Academic | 0.30 (0.27–0.34) | d,* e | 0.55 (0.48–0.62) | d,* e* |
| b. Enrichment | 0.31 (0.27–0.36) | d,* e | 0.56 (0.47–0.65) | d,* e* |
| c. Nonactive recreation | 0.30 (0.25–0.34) | d* | 0.47 (0.37–0.56) | d,* e* |
| d. Organized PA | 0.20 (0.17–0.23) | a,* b,* c,* g, i | 0.26 (0.20–0.32) | a,* b,* c,* g,* h,* i* |
| e. Free play PA | 0.24 (0.19–0.30) | a, b | 0.28 (0.19–0.39) | a,* b,* c,* g,* h,* i |
| f. Snack | 0.19 (0.10–0.32) | 0.60 (0.19–0.90) | ||
| g. Academic-physical activity | 0.26 (0.21–0.32) | d | 0.51 (0.39–0.62) | d,* e* |
| h. Enrichment-physical activity | 0.21 (0.12–0.33) | 0.78 (0.40–0.95) | d,* e* | |
| i. No purpose | 0.28 (0.21–0.36) | d | 0.48 (0.32–0.64) | d*, e |
Abbreviations: LS, least squares; MVPA, moderate to vigorous PA; PA, physical activity.
*These P values were significant using the Benjamini–Hochberg procedure with a false discovery rate of .05.
Inequality in PA by Session Type for Gender and Socioeconomic Status Subgroups
| Session by purpose type, Gini coefficient | Males, LS mean (95% CI) | Differences (P < .05) | Females, LS mean (95% CI) | Differences (P < .05) | F/R lunch status, LS mean (95% CI) | Differences (P < .05) | FP lunch status, LS mean (95% CI) | Differences (P < .05) |
|---|---|---|---|---|---|---|---|---|
| Total activity counts | N = 794 | N = 777 | N = 784 | N = 826 | ||||
| Session purpose | P = .0899 | P = .0035 | P = .0183 | P = .0022 | ||||
| a. Academic | 0.31 (0.27–0.35) | 0.29 (0.25–0.33) | d,* e | 0.34 (0.28–0.41) | d | 0.30 (0.26–0.34) | d,* f | |
| b. Enrichment | 0.27 (0.23–0.32) | 0.31 (0.26–0.37) | d,* e | 0.37 (0.29–0.45) | d | 0.30 (0.26–0.35) | d,* f | |
| c. Nonactive recreation | 0.28 (0.23–0.33) | 0.27 (0.22–0.33) | d* | 0.37 (0.29–0.47) | d | 0.28 (0.22–0.33) | d,* f | |
| d. Organized PA | 0.20 (0.16–0.23) | 0.19 (0.16–0.23) | a,* b,* c* | 0.28 (0.22–0.34) | a, b, c, e, g,* i* | 0.20 (0.17–0.23) | a,* b,* c,* g | |
| e. Free play PA | 0.20 (0.15–0.26) | 0.21 (0.15–0.28) | a, b | 0.35 (0.28–0.43) | d | 0.25 (0.20–0.32) | ||
| f. Snack | 0.16 (0.07–0.34) | 0.17 (0.08–0.34) | 0.39 (0.27–0.52) | 0.13 (0.06–0.26) | a, b, c | |||
| g. Academic-physical activity | 0.29 (0.23–0.35) | 0.24 (0.19–0.31) | 0.36 (0.29–0.44) | d* | 0.26 (0.20–0.32) | d | ||
| h. Enrichment-physical activity | 0.19 (0.11–0.32) | 0.25 (0.03–0.80) | 0.35 (0.23–0.48) | 0.17 (0.08–0.31) | ||||
| i. No purpose | 0.26 (0.19–0.35) | 0.26 (0.18–0.36) | 0.41 (0.33–0.50) | d* | 0.23 (0.19–0.31) |
| MVPA minutes | N = 684 | N = 674 | N = 588 | N = 720 | ||||
|---|---|---|---|---|---|---|---|---|
| Session purpose | P = .0104 | P = .0186 | P = .0431 | P < .0001 | ||||
| a. Academic | 0.55 (0.48–0.62) | d,* e* | 0.55 (0.47–0.63) | d,* e,* i | 0.73 (0.61–0.81) | c, d,* e, i | 0.57 (0.48–0.65) | d,* e, i |
| b. Enrichment | 0.54 (0.45–0.62) | d,* e* | 0.57 (0.48–0.66) | d,* e,* i | 0.61 (0.48–0.73) | d | 0.59 (0.49–0.68) | d,* e, i |
| c. Nonactive recreation | 0.47 (0.38–0.56) | d,* e* | 0.50 (0.40–0.61) | d,* e | 0.58 (0.43–0.72) | a, d, g | 0.49 (0.38–0.60) | d* |
| d. Organized PA | 0.25 (0.19–0.32) | a,* b,* c,* g,* h, i | 0.24 (0.17–0.31) | a,* b,* c,* g* | 0.43 (0.32–0.55) | a,* b, c, g* | 0.27 (0.20–0.35) | a,* b,* c,* g,* h |
| e. Free play PA | 0.22 (0.13–0.35) | a,* b,* c,* g,* h, i | 0.29 (0.17–0.44) | a,* b,* c, g | 0.48 (0.29–0.68) | a, g | 0.39 (0.25–0.55) | a, b |
| f. Snack | 0.28 (0.06–0.70) | 0.27 (0.07–0.64) | 0.49 (0.12–0.87) | 0.37 (0.12–0.70) | ||||
| g. Academic-physical activity | 0.51 (0.39–0.62) | d,* e* | 0.48 (0.36–0.61) | d,* e | 0.75 (0.61–0.86) | c, d,* e, i | 0.53 (0.41–0.65) | d* |
| h. Enrichment-physical activity | 0.47 (0.27–0.69) | d, e | 0.18 (0.00–0.99) | 0.60 (0.32–0.83) | 0.54 (0.31–0.75) | d | ||
| i. No purpose | 0.44 (0.29–0.60) | d, e | 0.37 (0.22–0.55) | a, b | 0.49 (0.29–0.69) | a, g | 0.38 (0.23–0.56) | a, b |
Abbreviations: F/R, free/reduced; FP, full pay; LS, least squares; MVPA, moderate to vigorous PA; PA, physical activity.
*These P values were significant using the Benjamini–Hochberg procedure with a false discovery rate of .05.
Meetings
Across all meetings, inequality in total activity counts during the duration of the meeting ranged from 0.06 to 0.47 (mean = 0.20, SD = 0.08). Inequality in minutes of MVPA ranged from 0.07 to 0.66 (mean = 0.28, SD = 0.13).
Table 3 presents least squares means estimates of inequality in activity counts and MVPA minutes during meeting time by setting type. The main effect of setting type on inequality was significant for total activity counts and MVPA minutes. Youth club meetings had significantly lower inequality in total activity counts than school (P = .032) and before-/after-school (P = .0006) meetings. School classrooms had significantly greater inequality in MVPA minutes than before-/after-school (P = .0023) and youth club (P = .0001) meetings.
Sessions
Across all sessions, inequality in total activity counts ranged from 0.06 to 0.96 (mean = 0.31, SD = 0.13), and inequality in minutes of MVPA ranged from 0.03 to 0.98 (mean = 0.52, SD = 0.24).
Table 3 presents least squares means estimates of inequality in activity counts and MVPA minutes during time-segmented sessions. The main effect of session purpose on inequality was significant for total activity counts and MVPA minutes. Organized PA sessions had significantly lower inequality in activity counts than academic (P < .0001), enrichment (P < .0001), and nonactive recreation (P < .0001) sessions. Organized PA and free play PA sessions had significantly lower inequality in MVPA minutes than academic (P < .0001), enrichment (P < .0001 and P = .0001, respectively), nonactive recreation (P < .0001 and P = .0076, respectively), academic-PA (P < .0001 and P = .0024, respectively), and enrichment-PA (P = .0072 and P = .013, respectively) sessions. Figure 1 illustrates the Lorenz curve for activity counts and MVPA minutes of one organized PA and one enrichment session.
Demographic Subgroups
Table 4 presents estimates of inequality in activity counts and MVPA minutes by session purpose among each demographic subgroup. For males, the main effect of session purpose on inequality in activity counts was not significant. For female, FRLS, and FPLS subgroups, the main effect of session purpose on activity counts was significant. Organized PA sessions had significantly lower inequality in activity counts (P < .05) than academic, enrichment, and nonactive recreation sessions. Organized PA sessions had significantly lower inequality in MVPA minutes (P < .05) than academic and academic-PA sessions for all demographic subgroups.
Discussion
The present study examined the influence of social structure on PA inequality during time-segmented organized group setting meetings for children. Although inequality existed in all settings, as hypothesized, across total meeting time, youth club settings had lower inequality in activity counts than school and before-/after-school settings, and school had greater inequality in MVPA minutes. On the more granular session time scale, results supported our hypothesis that PA inequality would differ by session purpose. Specifically, organized PA sessions had lower PA inequality than academic, enrichment, and nonactive recreation sessions for total activity counts and MVPA minutes. These findings were consistent for activity counts for females and FPLS participants, as well as MVPA minutes for males, females, and FPLS participants. Our hypothesis that free play sessions would have lower inequality compared with organized PA sessions was not supported.
A meta-analysis of PA in organized settings found that children accrue substantial amounts of PA during setting time, and PA engagement is greater in after-school programs compared with school.42 The present study showed that the school setting has greater inequality in MVPA than the out-of-school settings. Examining setting meeting time on the more granular session time scale can aid in understanding these differences in PA inequality between settings. PA inequality differed by session purpose, such that PA sessions had lower inequality than other session types. Thus, the differences in inequality by setting type during total meeting time may be due to differences in the number and length of implemented sessions during meeting routines. For instance, a greater amount of time is likely spent in academic sessions during school compared with out-of-school settings. Additionally, as schools face challenges in promoting PA opportunities,20,42 the out-of-school settings may have inserted more PA sessions during meeting time, contributing to lower inequality during meeting time overall. Further examining meeting routines in school and out-of-school settings would contribute to a greater understanding of how children spend their time and PA inequality within these settings.
Studies have shown that social system structure within setting time influences variability in average PA outcomes.30,35,38–41 For example, in preschool settings, mean PA has been shown to be greater when children are arranged in small groups compared with whole groups.30 In youth club settings, research has shown that a greater proportion of time is spent in MVPA during PA sessions compared with curricular sessions.41 The results of the present study extend these findings by illustrating that social structure also influences inequalities in the distribution of PA among children within setting time. These findings suggest that how meeting routines are structured can influence not only how much PA children accrue, but also whether PA is accrued by all children in the setting or by only a subset of a few children in the setting.
Several studies have shown that children’s PA is higher during free play PA than organized PA sessions.35,38,39 We hypothesized that free play PA sessions would have lower inequality than organized PA sessions. Although this hypothesis was not supported, results showed PA inequality was lower during organized PA sessions than other session purpose types, such as academic and enrichment. Organized PA sessions may provide structure that promotes similar amounts of PA among all children present in the setting, whereas during other session types, only a subset of children choose to engage in PA. Fairclough et al71 investigated PA patterns of children classified as high- and low-active during school and found that children classified as high-active had greater PA than children classified as low-active during segments such as lunchtime. However, no differences in PA were found between the high- and low-active groups during recess.71 The authors suggest children had more choice regarding the types of behaviors to engage in during lunchtime, such that the high-active group may have chosen to be more active during that time segment compared with the low-active group.71 In the present study, PA inequality may be greater during sessions such as academic and enrichment because only a subset of children chooses to be active when the intended purpose is not PA. These findings suggest that inserting both structured, organized PA and free play sessions into setting routines may be a strategy for increasing both the amount and equality of PA for children during setting time.
When examining PA inequality by session purpose among demographic subgroups, PA inequality appears to be higher among FRLS than FPLS participants, indicated by larger Gini coefficients for all session purpose types, while inequality is similar between males and females. Prior research examining child PA among demographic subgroups has predominantly focused on PA differences between subgroups. For example, males have been shown to have a greater likelihood of meeting PA guidelines than females.5 The present study indicates that inequalities also exist within demographic subgroups. Widyastari et al49 recently examined PA inequality among Thai adults by demographic subgroups (eg, male/female). Using the Gini coefficient, they found that PA inequality existed in all examined subgroups. Specifically, post-COVID-19 pandemic, they found similar levels of inequality in PA among males (Gini = 0.49) and females (Gini = 0.45). By income level, PA inequality existed in all income groups, although inequality was highest among those with no income (Gini = 0.55).49 Our findings are consistent with this prior research and suggest that PA inequality overall may not be due solely to differences in PA by gender and lunch status, illustrating the importance of examining inequality both within and between populations.
Although the Gini coefficient uses information from the entire distribution to generate a single summary statistic of inequality,46,55 the metric does not differentiate between different kinds of inequalities, such that 2 populations or session time segments with similar Gini coefficient values may have different distributions of PA. Additionally, interpreting the magnitude of the Gini coefficient and differences between coefficients is challenging because it is not expressed in natural units.72 Thus, interpreting the magnitude of PA inequality in the present study is difficult. In the income inequality literature, a coefficient below 0.3 is considered low (ie, more equal) and above 0.5 is considered very unequal,59 and previous analyses suggest a significant increase in the odds of poor health outcomes with a 0.05 change in income inequality.73 Based on these values, our results indicate relatively large inequalities in MVPA minutes exist in organized group settings for children and suggest meaningful differences in inequality by session purpose. Although the Gini coefficient has been the most popular measure of inequality in the public health literature,55 other potential metrics for operationalizing inequality and understanding PA outcomes exist,46,55 such as decile ratios74 and the Index of Concentration at the Extremes.48 While each inequality metric has advantages and disadvantages, other inequality metrics, to our knowledge, have not yet been used to investigate PA, warranting further investigation to more fully understand PA inequality.
Metrics for understanding community and setting PA outcomes, including the distribution of such outcomes, are needed to inform policy and practice efforts to advance population health PA improvement for children.18–22 Data collected on granular scales, such as within organized group setting time, can aid in understanding social system structures promoting and constraining PA. Additionally, setting leaders (eg, teachers) can use such data to understand their unique setting meeting routines and restructure setting time to promote PA for all children. The collection of local inequality data can also be integrated into multicomponent PA efforts, such as the Comprehensive School Physical Activity Programs,75 to assess and intervene upon PA opportunities. The findings of the present study suggest integrating time segments for PA into setting meeting routines can improve the level of equality in how PA is distributed among children during setting time.
Study Limitations and Strengths
A limitation of the current study is that organized group settings from only 2 rural Nebraska communities were examined. Communities were predominantly non-Hispanic, White, and only a male/female dichotomy was considered in gender analyses, potentially limiting the generalizability of results. Additionally, although the Gini coefficient is commonly used to examine inequality, the metric does not indicate where inequality occurs, nor does the coefficient directly indicate whether a particular subgroup (eg, males) is more or less advantaged than another in a session where inequality exists (eg, whether males hold all of the PA and females hold no PA in a particular session). Finally, this study focused on understanding inequalities in PA within organized group setting time for children and did not measure PA across the entire day. Future research should examine PA inequality using additional metrics of outcomes and determinants to understand both within- and between-group inequalities, larger samples of communities, other time segments (eg, 24-h PA), and other potential factors influencing relationships between setting structure and PA inequality (eg, child age/grade level and group make-up, such as single-grade or multigrade). Strengths of the study include the use of a device-based measure of PA and a large sample size of meetings, sessions, and children. Additionally, we used a novel observation method to characterize organized group setting meeting routines for children and the naturally occurring social structures within meeting routines to understand how social structure influences PA outcomes. Despite the limitations of the Gini coefficient, this metric of inequality allows for characterizing the full distribution of PA within and across school, before-/after-school, and youth club setting meetings for children. The present study examines these organized group settings as dynamic social systems in which PA is dependent upon within-setting interactions, thus contributing a greater understanding of system-level PA outcomes during setting time. Examining within-group inequality, and the social structures influencing inequality, by using metrics that characterize the distribution of outcomes is an important step for moving toward a comprehensive understanding of population health PA outcomes.
Conclusions
The present study showed that PA inequality in organized group settings for children differed by setting type and time-segmented session purpose and illustrated that the Gini coefficient can aid in understanding whether the setting system results in PA for all children in the setting or inequality in PA. Community metrics are needed to inform policy and practice efforts for improving population PA outcomes for children and reducing health disparities. Researchers and practitioners should not only focus on mean PA outcomes but should also monitor inequalities by examining the distribution of PA across organized group settings and the influence of social structure on this distribution.
Acknowledgments
The authors would like to thank our community partners for participating in this study. Funding Source: Research reported in this publication was supported by a grant (R01CA215420) from the National Cancer Institute of the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This study is registered at www.clinicaltrials.gov (Identifier: NCT03380143).
References
- 1.↑
U.S. Department of Health and Human Services. Physical Activity Guidelines for Americans. 2nd ed. U.S. Department of Health and Human Services; 2018.
- 2.↑
2018 Physical Activity Guidelines Advisory Committee. 2018 Physical Activity Guidelines Advisory Committee Scientific Report. Department of Health and Human Services; 2018. Accessed January 12, 2021. https://health.gov/our-work/physical-activity/current-guidelines/scientific-report
- 3.↑
Telama R, Yang X, Viikari J, Välimäki I, Wanne O, Raitakari O. Physical activity from childhood to adulthood: a 21-year tracking study. Am J Prev Med. 2005;28(3):267–273. PubMed ID: 15766614 doi:
- 4.↑
World Health Organization. WHO Guidelines on Physical Activity and Sedentary Behaviour. World Health Organization; 2020.
- 5.↑
Friel CP, Duran AT, Shechter A, Diaz KM. U.S. children meeting physical activity, screen time, and sleep guidelines. Am J Prev Med. 2020;59(4):513–521. PubMed ID: 32863080 doi:
- 6.↑
World Health Organization. Global Status Report on Physical Activity 2022. World Health Organization; 2022.
- 7.↑
Barker RG. Ecological Psychology: Concepts and Methods for Studying the Environment of Human Behavior. Stanford University Press; 1968.
- 9.
Bates FL, Bacon L. The community as a social system. Social Forces. 1972;50(3):371–379. doi:
- 10.
Essay AM, Schlechter CR, Mershon CA, et al. A scoping review of whole-of-community interventions on six modifiable cancer prevention risk factors in youth: a systems typology. Prev Med. 2021;153:106769. doi:
- 11.↑
Essay AM, Schenkelberg MA, Von Seggern MJ, et al. A protocol for a local community monitoring and feedback system for physical activity in organized group settings for children. J Phys Act Health. 2023:20(5):385–393. PubMed ID: 36965493 doi:
- 12.↑
McFarland J, Hussar B, de Brey C, et al. The Condition of Education 2017. NCES 2017-144. National Center for Education Statistics. 2017.
- 13.↑
Child and Adolescent Health Measurement Initiative. 2021–2022 National Survey of Children’s Health (NSCH) data query. Data Resource Center for Child and Adolescent Health supported by the U.S. Department of Health and Human Services, Health Resources and Services Administration (HRSA), Maternal and Child Health Bureau (MCHB). Accessed April 4, 2024. http://www.childhealthdata.org
- 14.↑
Mullan K. A child’s day: trends in time use in the UK from 1975 to 2015. Br J Sociol. 2019;70(3):997–1024. PubMed ID: 29638001 doi:
- 15.
Bassett DR, John D, Conger SA, Fitzhugh EC, Coe DP. Trends in physical activity and sedentary behaviors of United States youth. J Phys Act Health. 2015;12(8):1102–1111. PubMed ID: 25347913 doi:
- 16.
Kyttä M, Hirvonen J, Rudner J, Pirjola I, Laatikainen T. The last free-range children? Children’s independent mobility in Finland in the 1990s and 2010s. J Transp Geogr. 2015;47:1–12. doi:
- 17.↑
Dzewaltowski DA. Community out-of-school physical activity promotion. In: Smith AL, Biddle S, eds. Youth Physical Activity and Sedentary Behavior: Challenges and Solutions. Human Kinetics; 2008:377–401.
- 18.↑
Liu YY, Barabási AL. Control principles of complex systems. Rev Mod Phys. 2016;88(3):035006. doi:
- 19.↑
Carver CS, Scheier MF. Control processes and self-organization as complementary principles underlying behavior. Pers Soc Psychol Rev. 2002;6(4):304–315. doi:
- 20.↑
Aubert S, Barnes JD, Demchenko I, et al. Global matrix 4.0 physical activity report card grades for children and adolescents: results and analyses from 57 countries. J Phys Act Health. 2022;19(11):700–728. PubMed ID: 36280233 doi:
- 21.
Pate RR, Sallis JF, Pollack Porter KM. Surveillance of physical activity: actions needed to support new federal guidelines. Am J Public Health. 2020;110(1):87–89. PubMed ID: 31800284 doi:
- 22.↑
Sallis JF, Pate RR. Creating the future of physical activity surveillance in the United States: better data for better health. J Phys Act Health. 2021;18(S1):S1–S5. PubMed ID: 34465655 doi:
- 23.↑
National Academies of Sciences, Engineering, and Medicine. Implementing Strategies to Enhance Public Health Surveillance of Physical Activity in the United States. National Academies Press; 2019. doi:
- 24.↑
National Academies of Sciences, Engineering, and Medicine, Institute of Medicine, Board on Population Health and Public Health Practice, Roundtable on Population Health Improvement. Metrics That Matter for Population Health Action: Workshop Summary. National Academies Press; 2016. doi:
- 25.↑
Schenkelberg MA, Essay AM, Rosen MS, et al. A protocol for coordinating rural community stakeholders to implement whole-of-community youth physical activity surveillance through school systems. Prev Med Rep. 2021;24:101536. doi:
- 26.↑
Steele RM, van Sluijs EM, Sharp SJ, Landsbaugh JR, Ekelund U, Griffin SJ. An investigation of patterns of children’s sedentary and vigorous physical activity throughout the week. Int J Behav Nutr Phys Act. 2010;7:88. doi:
- 27.
Fairclough SJ, Boddy LM, Mackintosh KA, Valencia-Peris A, Ramirez-Rico E. Weekday and weekend sedentary time and physical activity in differentially active children. J Sci Med Sport. 2015;18(4):444–449. PubMed ID: 25011925 doi:
- 28.↑
Brooke HL, Atkin AJ, Corder K, Ekelund U, van Sluijs EMF. Changes in time-segment specific physical activity between ages 10 and 14 years: a longitudinal observational study. J Sci Med Sport. 2016;19(1):29–34. PubMed ID: 25459234 doi:
- 29.↑
McLellan G, Arthur R, Donnelly S, Buchan DS. Segmented sedentary time and physical activity patterns throughout the week from wrist-worn ActiGraph GT3X+ accelerometers among children 7–12 years old. J Sport Health Sci. 2020;9(2):179–188. PubMed ID: 32099726 doi:
- 30.↑
Schlechter CR, Rosenkranz RR, Fees BS, Dzewaltowski DA. Preschool daily patterns of physical activity driven by location and social context. J Sch Health. 2017;87(3):194–199. PubMed ID: 28147455 doi:
- 31.↑
Tudor-Locke C, Lee SM, Morgan CF, Beighle A, Pangrazi RP. Children’s pedometer-determined physical activity during the segmented school day. Med Sci Sports Exerc. 2006;38(10):1732–1738. PubMed ID: 17019294 doi:
- 32.↑
Brown WH, Pfeiffer KA, McIver KL, Dowda M, Almeida MJCA, Pate RR. Assessing preschool children’s physical activity: the observational system for recording physical activity in children-preschool version. Res Q Exerc Sport. 2006;77(2):167–176. PubMed ID: 16898273 doi:
- 33.
Cohen A, McDonald S, McIver K, Pate R, Trost S. Assessing physical activity during youth sport: the observational system for recording activity in children: youth sports. Pediatr Exerc Sci. 2014;26(2):203–209. PubMed ID: 24277926 doi:
- 34.↑
Guagliano JM, Rosenkranz RR, Kolt GS. Girls’ physical activity levels during organized sports in Australia. Med Sci Sports Exerc. 2013;45(1):116–122. PubMed ID: 22843107 doi:
- 35.↑
Schlechter CR, Guagliano JM, Rosenkranz RR, Milliken GA, Dzewaltowski DA. Physical activity patterns across time-segmented youth sport flag football practice. BMC Public Health. 2018;18(1):226. doi:
- 36.↑
Dzewaltowski DA, Schlechter CR. CLOUDEE: child leadership observation for understanding developmental episodes and engagement. Published online 2017. Accessed August 17, 2022. http://www.drdaviddzewaltowski.com/child-leadership-observation.html
- 37.↑
Rosenkranz RR, Welk GJ, Dzewaltowski DA. Environmental correlates of objectively measured physical activity and sedentary behavior in after-school recreation sessions. J Phys Act Health. 2011;8(s2):S214–S221. PubMed ID: 28829717 doi:
- 38.↑
Coleman KJ, Geller KS, Rosenkranz RR, Dzewaltowski DA. Physical activity and healthy eating in the after-school environment. J Sch Health. 2008;78(12):633–640. PubMed ID: 19000239 doi:
- 39.↑
Trost SG, Rosenkranz RR, Dzewaltowski D. Physical activity levels among children attending after-school programs. Med Sci Sports Exerc. 2008;40(4):622–629. PubMed ID: 18317385 doi:
- 40.
Guagliano JM, Updyke NJ, Rodicheva NV, et al. Influence of session context on physical activity levels among russian girls during a summer camp. Res Q Exerc Sport. 2017;88(3):352–357. PubMed ID: 28622098 doi:
- 41.↑
Schlechter CR, Rosenkranz RR, Guagliano JM, Dzewaltowski DA. Impact of troop leader training on the implementation of physical activity opportunities in Girl Scout troop meetings. Transl Behav Med. 2018;8(6):824–830. PubMed ID: 30476320 doi:
- 42.↑
Tassitano RM, Weaver RG, Tenório MCM, Brazendale K, Beets MW. Physical activity and sedentary time of youth in structured settings: a systematic review and meta-analysis. Int J Behav Nutr Phys Act. 2020;17(1):160. doi:
- 43.↑
Subramanian SV, Kim R, Christakis NA. The “average” treatment effect: a construct ripe for retirement. a commentary on deaton and cartwright. Soc Sci Med. 2018;210:77–82. PubMed ID: 29724462 doi:
- 44.↑
Lee AH, Xiang L, Hirayama F. Modeling physical activity outcomes: a two-part generalized-estimating-equations approach. Epidemiology. 2010;21(5):626–630. PubMed ID: 20588183 doi:
- 45.↑
Baldwin SA, Fellingham GW, Baldwin AS. Statistical models for multilevel skewed physical activity data in health research and behavioral medicine. Health Psychol. 2016;35(6):552–562. PubMed ID: 26881287 doi:
- 47.↑
World Health Organization. Monitoring health inequality: An essential step for achieving health equity. Published online 2014. Accessed November 23, 2021. http://apps.who.int/iris/bitstream/10665/164530/1/WHO_HIS_HSI_2015.1_eng.pdf?ua=1&ua=1
- 48.↑
Krieger N, Waterman PD, Spasojevic J, Li W, Maduro G, Van Wye G. Public health monitoring of privilege and deprivation with the index of concentration at the extremes. Am J Public Health. 2016;106(2):256–263. PubMed ID: 26691119 doi:
- 49.↑
Widyastari DA, Khanawapee A, Charoenrom W, Saonuam P, Katewongsa P. Refining index to measure physical activity inequality: which group of the population is the most vulnerable? Int J Equity Health. 2022;21(1):123. doi:
- 50.↑
Kindig D, Stoddart G. What is population health? Am J Public Health. 2003;93(3):380–383. PubMed ID: 12604476 doi:
- 51.↑
Asada Y, Hurley J, Norheim OF, Johri M. A three-stage approach to measuring health inequalities and inequities. Int J Equity Health. 2014;13:98. doi:
- 54.↑
Glasser GJ. Variance formulas for the mean difference and coefficient of concentration. J Am Stat Assoc. 1962;57(299):648–654. doi:
- 55.↑
De Maio FG. Income inequality measures. J Epidemiol Community Health. 2007;61(10):849–852. PubMed ID: 17873219 doi:
- 56.↑
Sitthiyot T, Holasut K. A simple method for measuring inequality. Palgrave Commun. 2020;6(1):112. doi:
- 57.↑
Pabayo R, Fuller D, Lee EY, Horino M, Kawachi I. State-level income inequality and meeting physical activity guidelines; differential associations among US men and women. J Public Health. 2018;40(2):229–236. doi:
- 58.↑
Althoff T, Sosič R, Hicks JL, King AC, Delp SL, Leskovec J. Large-scale physical activity data reveal worldwide activity inequality. Nature. 2017;547(7663):336–339. PubMed ID: 28693034 doi:
- 59.↑
Chaput JP, Barnes JD, Tremblay MS, et al. Inequality in physical activity, sedentary behaviour, sleep duration and risk of obesity in children: a 12-country study. Obes Sci Pract. 2018;4(3):229–237. PubMed ID: 29951213 doi:
- 60.
Hirata RP, Oliveira JM de, Schneider LP, et al. The Gini coefficient: a new approach to assess physical activity inequality in COPD. COPD. 2020;17(6):623–626. PubMed ID: 33238759 doi:
- 61.↑
Schlechter CR. The social contexts of youth settings for influencing physical activity. Published online 2018. Accessed May 16, 2023. http://hdl.handle.net/2097/39124
- 62.↑
Curran GM, Bauer M, Mittman B, Pyne JM, Stetler C. Effectiveness-implementation hybrid designs: combining elements of clinical effectiveness and implementation research to enhance public health impact. Med Care. 2012;50(3):217–226. PubMed ID: 22310560 doi:
- 63.↑
Geverdt D. Education Demographic and Geographic Estimates Program (EDGE): Locale Boundaries User’sManual (NCES 2016-012). U.S. Department of Education; 2015. Accessed August 30, 2022. https://nces.ed.gov/pubsearch/
- 64.↑
Department of Education: Office of Elementary and Secondary Education. RLIS eligibility. 2020. Accessed August 19, 2022. https://oese.ed.gov/offices/office-of-formula-grants/rural-insular-native-achievement-programs/rural-education-achievement-program/rural-and-low-income-school-program/eligibility/
- 65.↑
McKenzie TL, Cohen DA, Sehgal A, Williamson S, Golinelli D. System for observing play and recreation in communities (SOPARC): reliability and feasibility measures. J Phys Act Health. 2006;3(s1):S208–S222. PubMed ID: 28834508 doi:
- 66.↑
McKenzie TL, Marshall SJ, Sallis JF, Conway TL. Leisure-time physical activity in school environments: an observational study using SOPLAY. Prev Med. 2000;30(1):70–77. PubMed ID: 10642462 doi:
- 67.↑
Trost SG, Loprinzi PD, Moore R, Pfeiffer KA. Comparison of accelerometer cut points for predicting activity intensity in youth. Med Sci Sports Exerc. 2011;43(7):1360–1368. PubMed ID: 21131873 doi:
- 68.↑
Migueles JH, Cadenas-Sanchez C, Ekelund U, et al. Accelerometer data collection and processing criteria to assess physical activity and other outcomes: a systematic review and practical considerations. Sports Med. 2017;47(9):1821–1845. PubMed ID: 28303543 doi:
- 69.↑
Evenson KR, Catellier DJ, Gill K, Ondrak KS, McMurray RG. Calibration of two objective measures of physical activity for children. J Sports Sci. 2008;26(14):1557–1565. PubMed ID: 18949660 doi:
- 70.↑
Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B. 1995;57(1):289–300. doi:
- 71.↑
Fairclough SJ, Beighle A, Erwin H, Ridgers ND. School day segmented physical activity patterns of high and low active children. BMC Public Health. 2012;12:406. doi:
- 72.↑
Koolman X, van Doorslaer E. On the interpretation of a concentration index of inequality. Health Econ. 2004;13(7):649–656. PubMed ID: 15259044 doi:
- 73.↑
Subramanian SV, Kawachi I. Whose health is affected by income inequality? a multilevel interaction analysis of contemporaneous and lagged effects of state income inequality on individual self-rated health in the United States. Health Place. 2006;12(2):141–156. PubMed ID: 16338630 doi:
- 74.↑
Gold R, Kawachi I, Kennedy BP, Lynch JW, Connell FA. Ecological analysis of teen birth rates: association with community income and income inequality. Matern Child Health J. 2001;5(3):161–167. PubMed ID: 11605721 doi:
- 75.↑
Centers for Disease Control and Prevention. Comprehensive School Physical Activity Programs: A Guide for Schools. U.S. Department of Health and Human Services; 2013.
