Park-Based Physical Activity in Diverse Communities of Two U.S. Cities: An Observational Study

Introduction 

Lack of physical activity among U.S. residents is a major health concern, particularly among low-income and minority populations. National studies show that adults and children from racial and ethnic minority groups get less physical activity than their white counterparts.1, 2 Racial and ethnic minorities and low-income populations also bear a disproportionate risk of experiencing chronic diseases3 among which obesity and overweight, stroke, diabetes, depression and anxiety, colon cancer, and cardiovascular diseases are linked to physical inactivity.4 Efforts to increase physical activity in diverse communities could have positive health outcomes.

The ecologic model of health behavior examines how the modification of built environment features can positively affect behaviors such as physical activity.5, 6 Public parks can play a substantial role in increasing leisure-time physical activity because they offer a wide range of free or low-cost activities close to where people live and because their existence, design, and quality are influenced through public policy.7, 8 Access to parks and recreation areas has been identified as an important predictor of physical activity,9, 10, 11 and a national study estimates that 70% of U.S. residents live within walking distance of a public park.12 Moreover, 80% of U.S. residents report using public parks, and nearly one person in four uses them “frequently.” If neighborhood parks are to help increase physical activity in diverse and disadvantaged communities, research is needed to describe how parks are used and identify which settings support physical activity. Studies of ethnically diverse and underserved populations are also needed to address disparities in physical activity through the provision of parks and recreation facilities.

Few studies have examined how the capacity of parks and the activity spaces within them contribute to physical activity. McKenzie et al.13 observed users of eight large parks in Los Angeles and found that 62% of male park users and 71% of female park users were sedentary. Vigorous-intensity activity in specific activity areas ranged from 2% to 34% and was significantly lower in picnic (13%) and open-space areas (28%) than in all sport facilities (32%–34%) except for baseball/softball fields (23% vigorous activity). Another study found that girls living near parks with playgrounds, basketball courts, walking trails, swimming areas, and tracks accumulated more moderate-to-vigorous physical activity (MVPA) than girls living farther away.11 However, residential areas with predominantly African-American and Hispanic populations appear to have lower access to park and recreation facilities.14, 15, 16

These study findings are helpful but provide an incomplete understanding of the association between public parks and physical activity in diverse communities. While parks have the potential to support physical activity, a substantial amount of use can be sedentary. Considerable variation exists in the conduciveness of specific activity settings within parks to support physical activity. Most studies of parks and physical activity have focused on proximity of parks and have not linked physical activity to specific modifiable park attributes. Because public parks can be influenced through public policy, identifying features most likely to support MVPA in diverse communities could suggest how park settings can be managed to increase physical activity among residents. The objectives of the present study were to (1) assess levels of physical activity in selected neighborhood parks, (2) compare levels of physical activity observed in parks located in neighborhoods of different racial/ethnic and income composition, and (3) examine whether levels of physical activity associated with specific activity areas vary by the racial/ethnic and income composition of neighborhoods.

Methods 

Measures 

Physical activity 

Physical activity was measured using a modified version of the System for Observing Play and Leisure Activity in Youth (SOPLAY)17 similar to the method developed by McKenzie et al.13 Observation codes accounted for age group (children/adult), gender, and activity levels (sedentary, walking/moderate, and vigorous). Construct validity of these physical activity codes has been established in previous studies.18, 19 Trained observers recorded observations of physical activity in the parks between 10 am and 6 pm from Friday through Sunday during the spring (Tampa, March–April) and early summer (Chicago, May–June) of 2005. Following an established protocol, separate scans were made for girls, boys, women, and men. Park activity areas were scanned visually from left to right and the codes representing park users’ activity levels were recorded on a standardized form. Four scanning periods were conducted for each activity zone (two for am, two for pm hours). Cohen’s kappa coefficients for inter-observer agreement between paired observers ranged from 0.79 to 0.97, which is well within the acceptable range.20

Physical activity codes were converted to energy expenditure (kcal/kg/min), providing a second measure of physical activity using previously validated codes.13 Energy expenditure was estimated by summing the number of individuals in sedentary, walking, and vigorous categories and then multiplying by their respective constants, 0.051kcal/kg/min, 0.096kcal/kg/min, and 0.144kcal/kg/min.17 Calculation of energy expenditure per person enables transformation of these data into a linear format to compare relative activity levels across parks, activity spaces, and neighborhoods. This also allows for a comparison of means through ANOVA, using energy expenditure per person as the dependent variable and activity areas and neighborhood composition as independent variables.

Age group and gender 

Observers categorized individuals in parks into two age groups, children and adults. Children were coded as anyone who appeared to be 12 and under following previously validated protocols.13 High inter-rater reliability indicated a sufficient agreement between observers on categorizations. Similarly, categorization as male or female was based on apparent gender.

Park activity zones 

Activity zones for all parks and their boundaries were mapped by two members of the research team prior to observations. In most cases, activity zones coincided with established recreation use areas such as playgrounds, courts (e.g., tennis, basketball), picnic areas, sports fields, and open spaces (see Table 1 for complete listing). Each member of the observation team was instructed on zone boundaries and carried a map of activity zones into the field.

Table 1. Primary activity zones in study parks used for physical activity observations by neighborhood type

		Neighborhood type
		African-American, high-income	African-American, low-income	White, high-income	White, low-income	Hispanic, high-income	Hispanic, low-income
	Courtsa
	Tampa	0	79	0	28	15	100
	Chicago	105	113	50	21	85	51
	Sports fieldsb
	Tampa	0	72	0	487	0	0
	Chicago	109	157	229	212	152	184
	Shelters
	Tampa	0	785	480	0	1487	719
	Chicago	9	0	0	0	0	0
	Playgrounds
	Tampa	0	216	130	40	1001	164
	Chicago	100	52	239	60	99	127
	Open spaces
	Tampa	0	180	143	78	643	98
	Chicago	0	0	55	62	0	385

a Courts include tennis, racquetball, volleyball, and basketball courts. b Sports fields include soccer, football, and baseball/softball fields.

Neighborhood composition 

Neighborhood racial/ethnic composition was a categoric variable with three attributes (white, African American, and Hispanic) taken from topologically integrated geographic encoding and referencing (TIGER) census files. Neighborhood income was a dichotomous variable (low and middle/upper) from the same data source. A combined racial/ethnic and income variable (e.g., low-income white, high-income white) was created to examine physical activity by neighborhood composition.

Results 

Variation in Physical Activity by Neighborhood Type 

Mean energy-expenditure-per-person values for parks in different neighborhood types are shown in Table 2. Overall, significant differences in mean energy expenditure were observed in Tampa and Chicago parks. In Tampa, parks in neighborhoods (census tracts) with large concentrations of Hispanic Americans showed the highest mean energy expenditure per person (mean=0.069), followed by parks in predominantly white areas (mean=0.068) and parks in predominantly African-American areas (mean=0.067) (F=3.06, p=0.047). Post-hoc tests revealed significant differences in energy expenditure between parks in Hispanic and African-American neighborhoods. In Chicago, users of parks in neighborhoods identified as African American showed the highest energy expenditure (mean=0.087), followed by parks in Hispanic (mean=0.082) and white (mean=0.082) neighborhoods (F=6.75, p=0.001). Scheffe’s post-hoc tests showed that mean energy expenditure of park users in African-American neighborhoods was significantly greater than mean energy expenditure of park users in Hispanic and white neighborhoods. Significant differences in energy expenditure were also observed according to neighborhood income (Table 3). In both cities, greater mean energy expenditure was observed in parks in higher-income neighborhoods, with the association stronger in Chicago (F=10.17, p=0.001) than Tampa (F=6.44, p=0.011).

Table 2. Mean energy expenditure per person by racial and ethnic neighborhood composition

	Racial/ethnic composition	n	Mean	SD	F	Eta2
	Tampa
	White	1389	0.068a,c	0.030	3.06⁎	0.001
	African-American	1332	0.067a	0.028
	Hispanic	4322	0.069c	0.029
	Chicago
	White	980	0.082a	0.037	6.75⁎⁎	0.006
	African-American	664	0.087b	0.037
	Hispanic	767	0.082a	0.037

a–c Means with different superscript are significantly different at p<0.05 (Scheffe’s post-hoc test). ⁎ p<0.05; ⁎⁎ p<0.01.

Table 3. Mean differences in energy expenditure per person by neighborhood income composition

	Income composition	n	Mean	SD	t
	Tampa
	Low-income	3046	0.068	0.029	−2.54⁎
	High-income	3985	0.069	0.030
	Chicago
	Low-income	1146	0.081	0.036	−3.19⁎⁎
	High-income	1265	0.086	0.038

⁎ p<0.05; ⁎⁎ p<0.01.

Analysis of variance was also used to examine variation in energy expenditure in neighborhood parks defined jointly by racial/ethnic and income composition (Table 4). In Tampa parks, differences in energy expenditure in parks of different racial/ethnic and income composition were statistically significant (F=8.96, p<0.001). Scheffe’s post-hoc tests indicate that energy expenditure was greatest in parks in neighborhoods identified as high-income Hispanic and low-income white, and lowest in high-income white and low-income Hispanic neighborhoods. Different results were obtained from Chicago parks. Although energy expenditure in parks of different racial/ethnic and income composition was statistically significant (F=10.16, p<0.001), parks in neighborhoods identified as high-income African-American had higher energy expenditure than all of the remaining racial/ethnic–income neighborhood types.

Table 4. Mean energy expenditure per person in parks from neighborhoods categorized by income and racial/ethnic composition

	Income and racial/ethnic composition	n	Mean	SD	F	Eta2
	Tampa
	Low-income white	633	0.072a	0.031	8.96⁎	0.005
	High-income white	756	0.066b	0.028
	Low-income African American	1332	0.067b	0.028
	High-income African Americanc	––	––	––
	Low-income Hispanic	1085	0.066b	0.028
	High-income Hispanic	3237	0.070a	0.030
	Chicago
	Low-income white	385	0.083a	0.036	10.16⁎	0.021
	High-income white	595	0.080a	0.037
	Low-income African American	330	0.080a	0.035
	High-income African American	334	0.096a	0.037
	Low-income Hispanic	431	0.079b	0.037
	High-income Hispanic	336	0.086a	0.037

a–b Means with different superscripts are significantly different at p<0.05 (Scheffe’s post-hoc test). c A park for this designation (higher-income and predominantly African-American) could not be found using our criteria for neighborhood selection. ⁎ p<0.001.

It is important to note that these patterns are influenced by the activity zones where observations of physical activity occurred. For example, in Tampa parks in low-income white neighborhoods, 76.8% and 12.4% of observations were in baseball/softball and open-space activity areas, respectively. In high-income and Hispanic neighborhoods, 47.5% and 32% of observations occurred near picnic shelters and playgrounds, respectively. For the remaining subgroups, the percentage of observations conducted near picnic shelters ranged between 59% and 69.5%. As detailed below, picnic shelters were associated with low energy expenditure relative to other activity zones. In Chicago parks, baseball/softball fields and open-space areas generated the lowest energy expenditure values. In low-income white neighborhoods, 64% of observations were in baseball/softball fields (26%) and open-space areas (38%). In high-income white neighborhoods, 51% of observations occurred in baseball/softball fields (40.8%) and open-space areas (10.1%).

Physical Activity by Activity Zones 

How activity zones contributed to physical activity can be reported in greater detail. The percentages of people observed walking or engaged in vigorous physical activity in primary activity spaces in Tampa parks were as follows: tennis/racquetball courts (75%); basketball courts (74.4%); open-space areas (45.5%); playgrounds (44.6%); baseball/softball fields (32.5%); fishing areas (19.3%); and picnic shelters (16.9%). In Chicago, the percentages of people observed in walking or vigorous physical activity in primary activity zones were as follows: tennis/racquetball courts (54.5%); basketball courts (58.3%); open-space areas (46.8%); playgrounds (50.9%); and baseball/softball fields (37%).

The ANOVA results of energy expenditure by activity areas are reported in Table 5. For Tampa parks, the greatest energy expenditure (0.098) was associated with racquet sports (tennis and outdoor racquetball) and basketball courts. Dog play areas (0.057), picnic shelters (0.059), and fishing piers (0.060) were associated with the lowest energy expenditure. Scheffe’s post-hoc tests revealed significant separation among the activity areas. Tennis/racquetball and basketball courts (0.096) had greater energy expenditure than all other areas. Interestingly, energy expenditure for baseball/softball fields was not significantly different from energy expenditure documented in picnic, fishing, and dog play areas. In Chicago parks, less separation in terms of mean differences was observed among activity zones. Mean energy expenditure per person on basketball courts (0.088), playgrounds (0.088), and soccer fields (0.094) was significantly higher than that observed on baseball/softball fields (0.074).

Table 5. Mean energy expenditure per person in parks by activity zones

	Activity zones	n	Mean	SD	F	Eta2
	Tampa
	Picnic shelters	3471	0.059a,b	0.020	144.13⁎	0.127
	Dog play areas	36	0.057a,b	0.016
	Fishing piers	83	0.060a,b	0.020
	Baseball/softball fields	549	0.071b	0.031
	Open spaces	1010	0.078c	0.033
	Playgrounds	1551	0.080c	0.035
	Basketball courts	90	0.098d	0.033
	Tennis/racquetball courts	132	0.098d	0.034
	Chicago
	Dog play arease	8	0.057a	0.016	12.83⁎	0.039
	Baseball/softball fields	781	0.074a	0.033
	Open spaces	173	0.083a,b	0.036
	Volleyball	52	0.084a,b	0.034
	Tennis/racquetball	198	0.086a,b	0.037
	Playgrounds	574	0.088b,c	0.039
	Basketball court	175	0.088b,c	0.036
	Soccer fields	249	0.094b,c	0.039

a–d Means with different superscript are significantly different at p<0.05 (Scheffe’s post-hoc test). e This result should be interpreted with caution given the small N. ⁎ p<0.0001.

Discussion 

Seventy percent of Tampa park users and 51% of Chicago park users were observed in sedentary activities. Among Tampa park users, 21% and 8% were observed in walking and vigorous activity, respectively. Among Chicago park users, 28% and 22% were observed in walking and vigorous activity, respectively. These findings are similar to reports from past studies using observational methods.18, 21 While many types of park use, both active and passive, combine to provide an array of social, economic, and psychological benefits sought through leisure experience,22 the prevalence of sedentary activity in park settings suggests that there may be further opportunities to encourage physically active park use. Public parks are widely available, subject to public policy influence,8 and can promote population level changes in physical activity,1 so studies like the present one can provide information that leads to more health-promoting park management policies.

The association between age group and physical activity was statistically significant in both cities. In Tampa parks, 44% of children were observed walking or engaged in vigorous activity compared to 23.2% of adults. In Chicago parks, 52% of children were observed walking or engaged in vigorous activity versus 47% for adults. This evidence that many children are getting physical activity by using parks is encouraging. These results can inform policymakers on the importance of neighborhood parks as critical community spaces where children can be physically active. Additionally, the study provides evidence for improving facilities conducive to physical activity in existing parks and creating new parks as places where children can engage in MVPA.

The ANOVA procedures showed that physical activity, energy expenditure in particular, varied by neighborhood racial/ethnic and income composition. For example, in Tampa the highest levels of energy expenditure were generated in parks from high-income Hispanic neighborhoods and low-income white neighborhoods. The lowest energy expenditure was associated with high-income white neighborhoods and low-income Hispanic neighborhoods. These findings reflect the similar composition of Hispanic and white neighborhoods. In Tampa, block groups with a population greater than 50% Hispanic were also nearly 50% white. In Chicago, the greatest energy expenditure was recorded from parks in high-income African-American neighborhoods. In both cities, the association between activity zones and physical activity appears to underlie differences by racial/ethnic and income composition. These physical activity patterns suggest that ethnic and racial groups vary in their use of parks. It was not possible to establish those differences in the present study. Further investigation is needed to identify resources and configurations of parks that would most effectively encourage people to be active in each kind of neighborhood.

To put energy expenditure into perspective, a hypothetical 150-pound man who lived in the least-active neighborhood (African-American neighborhoods in Tampa), who visited the parks three times a week at 30 minutes per time, and who had the average energy expenditure would burn 21,379 kcal over the course of a year. A person living in the most-active neighborhood type (African-American neighborhoods in Chicago) and visiting the parks for the same amount of time would expend 27,760 kcal in a hypothetical year. This is a difference of 6381 kcal a year, or almost 2 pounds.

Sedentary behavior and lower levels of energy expenditure were associated with dog play areas, picnic shelters, baseball/softball fields, and open-space areas. MVPA and higher energy expenditure were generated by the use of soccer fields and playgrounds and by basketball, tennis/racquetball, and volleyball courts. Relative differences in physical activity by activity zones were consistent across cities. While these patterns would be expected, the present study provides quantitative evidence of how various activity areas within parks facilitate and constrain physical activity. It also highlights the need to consider how activities and facilities now available in parks located in communities at greater risk of inactivity and its health consequences might be redesigned or better programmed to stimulate physical activity and reduce racial/ethnic and income inequalities in physical activity. Future research using more rigorous designs, such as quasi-experimental evaluations of park renovations, can build on these results to better understand how specific configurations of facilities enhance moderate and vigorous physical activity in parks. Future studies should also identify the activity and program preferences of neighborhood residents. Perhaps moderate and vigorous physical activity in public parks can be increased in ethnically diverse communities if programs and facilities and other interventions are “culturally salient and appropriate.”23

The study has several limitations. First, the SOPLAY observations consist of momentary time sampling, meaning that each park user’s activity level was assessed only at one moment and each physical activity category encompassed a range of intensities.13, 24 The energy expenditure measures were based on the SOPLAY physical activity categories, so they are not precise. Second, the observations did not represent early morning, weekday, and seasonal park use. Different patterns of physical activity could result if broader coverage was achieved. Third, the neighborhood types in Tampa can be better described as “mixed” areas where there were high concentrations of both Hispanic and white residents. In contrast, Chicago neighborhoods exhibited greater residential segregation. The study did, however, present data on how parks located in ethnically diverse residential areas contribute to physical activity. An additional strength is that the data were obtained by established protocols from 28 parks in two large cities.13, 17

The present results underscore the need to better understand how public parks contribute to physical activity in diverse communities. Evidence of the extent of sedentary behavior in parks demonstrates the need to consider how parks can be designed and managed to encourage physically active park visits. Although parks are frequently touted as critical resources for physical activity, clearly more research is needed to guide managerial decisions and policy. Future studies should investigate how park infrastructure, amenities, and programs in activity areas affect physical activity in diverse communities. Another research priority would be to evaluate interventions specifically designed to increase physical activity in activity areas dominated by sedentary behavior.