Park Use and Park-Based Physical Activity in Low-Income Neighborhoods

Abstract

Routine physical activity is important for everyone, and most urban areas have an infrastructure of neighborhood parks that are intended to serve as a setting for recreation and leisure. However, parks are not used proportionally by all age groups, genders, and socioeconomic groups. This paper explores factors associated with park use by different age and gender groups in low-income neighborhoods in Los Angeles, CA. We found that women’s visits to parks generally centered around children, whereas men’s visits were more likely to be associated with their own physical activity. Barriers for seniors are associated with limited facilities and programming that meet their needs. Park managers should consider park renovations that include social meeting places, comfortable sitting areas, and walking paths to better serve women and seniors.

National surveys of health behaviors consistently show that physical activity declines as people age (Troiano et al., 2008) and that physical activity is lower in females than in males across the life span. This is particularly problematic because reductions in physical activity compound the increased risks of chronic diseases associated with aging, including heart disease, diabetes, osteoporosis, and cancer (U.S. Department of Health and Human Services, 2008; Office of the Surgeon General (US), 2010). Our society has created an infrastructure of parks and most urban residents live within 1 mile of a park (The Trust for Public Land, 2015). However, there are still many communities that lack equitable access to high-quality parks. (Babey, Wolstein, Krumholz, Robertson, & Diamant, 2013; Gordon-Larsen, Nelson, Page, & Popkin, 2006; Rigolon, 2016; Wen, Zhang, Harris, Holt, & Croft, 2013) Along with disparities in access, there are also disparities in park use: Parks are used more by younger than by older groups and more by males than females (Cohen et al., 2016; Evenson, Jones, Holliday, Cohen, & McKenzie, 2016).

Another barrier to physical activity as people age is the increasing prevalence of infirmities and chronic diseases (Mora & Valencia, 2018). It is likely that physical limitations play a role in how people decide to spend their leisure time. Nevertheless, people who maintain routine physical activity have a lower incidence of chronic diseases compared with individuals who fail to meet physical activity guidelines (Blair & Brodney, 1999; Haskell, Blair, & Hill, 2009).

Leisure-time activities are also patterned by gender roles. This has been manifested by men being overrepresented as park visitors (Evenson et al., 2016) as well as being more involved in organized sports than women (Cohen et al., 2016). Although hormonal changes contribute to high rates of osteoporosis in women, gender disparities in engagement in vigorous physical activity exacerbate its prevalence (Chang & Do, 2015).

Beyond the associations among age, gender, and physical activity and park use, several studies have found that parks in low-income neighborhoods are used less than parks in higher-income neighborhoods (Cohen et al., 2012, 2016), even though time-use studies indicate that low-income groups have more leisure time (Aguiar & Hurst, 2007). Some of the factors associated with the disparity include differences in park conditions (Weiss et al., 2011), park size (Kaczynski, Potwarka, & Saelens, 2008), the amount of programming, and concerns about safety (Echeverria, Luan Kang, Isasi, Johnson-Dias, & Pacquiao, 2014; Lapham et al., 2015; Ou et al., 2016) and crime (Han, Cohen, Derose, Li, & Williamson, 2018). However, one study found that concerns about safety had a very small association with park visitation, once individual-level factors like health were accounted for (Tucker-Seeley, Subramanian, Li, & Sorensen, 2009). In the United States, low-income neighborhoods in urban areas have large proportions of ethnic-racial minorities. An issue recently coming to light, but not well covered in the scientific literature, is concern about racial profiling and police harassment, which may lead to people avoiding public spaces. Beyond this, personal preferences likely play a role in that relaxation, TV watching, and other passive sedentary behaviors may be prioritized over outdoor leisure activities.

Although previous research has uncovered patterns of and barriers to park-based physical activities based on income, gender, and age separately (Carlson, Brooks, Brown, & Buchner, 2010; Echeverria et al., 2014; Loukaitou-Sideris, Levy-Storms, Chen, & Brozen, 2016; Mowen, Orsega-Smith, Payne, Ainsworth, & Godbey, 2007; Yen, Scherzer, Cubbin, Gonzalez, & Winkleby, 2007), we know considerably less about how those barriers intersect. Thus, in this study, we investigated whether women and men of different age groups in low-income neighborhoods of Los Angeles use local parks differently, even after controlling for a variety of individual, park-level, and neighborhood-level factors. Using the socioecologic model of health behavior as a theoretical foundation (Sallis et al., 2006) to help identify components of the social, physical, and policy environment relevant to park use, this study examines factors associated with park use among a representative sample of adult residents of low-income neighborhoods who live within a 1-mile radius of neighborhood parks. We also draw upon the writings of Jane Jacobs, whose keen observation of urban life identified the importance of structure and location, as well as “demand goods”—the activities, events, and facilities that are needed to draw people to parks (Jacobs, 1961). While ½ mile (about a 10-min walk) has been considered optimal access to parks (The Trust for Public Land, 2017), surveys of neighborhood park users indicate that 34% come from more than a mile a way, and 18% from between 1/2 and 1 mile away (Cohen, 2017). Studying a population that lives near parks removes proximity as a barrier and, thus, can shed light on how other facilitators and barriers to park use and park-based physical activity vary by age and gender.

It is likely that the physical structural characteristics of parks, the social structures of society, and individual choices all contribute to disparities in park use and park-based physical activity. Specifically, we investigated whether women and men of different age groups use neighborhood parks. Ultimately, this information can be used to inform the design and management of parks, so parks can more equitably serve all local residents.

Methods

The data for this study were taken from surveys that addressed park use and physical activity from a larger randomized trial in the City of Los Angeles intended to promote physical activity in low-income neighborhoods (Cohen et al., 2017). We defined low-income as the percentage of households in poverty above the city median of 19%. After eliminating six ineligible parks, we randomly selected the 48 (60%) of the 86 parks with recreation centers in low-income neighborhoods in the City of Los Angeles, optimizing geographical dispersion to avoid contamination that could occur if parks were too close. Parks were considered ineligible if they only provided specialized services or were in isolated housing projects, and use by the general public was prohibited.

The primary data for our analyses come from intercept surveys of 2,973 residents living within 1 mile of the selected parks, half at baseline, and half during the intervention period. Using ArcGIS, the household sample was identified within the buffers so that ⅓ of the sample lived within ¼ mile of the park, ⅓ within ¼-½ mile, and the remainder with ½ to 1 mile of the park’s mailing address. Table 1 shows the characteristics of the neighborhoods within a 1-mile radius of each park. We also separately interviewed a sample of 3,143 park users, who were asked to note whether they lived within ¼ mile, ½ mile, 1 mile, or further from the park. Although the City of Los Angeles Department of Recreation and Parks planned these parks to serve a population within a 2-mile radius, we limited our surveys to within 1 mile as we found that among visitors to the low-income area parks, the majority lived within 1 mile of the park. Only 16% of park visitors came from more than 1 mile away. A statistical analysis system procedure called PROC SURVEYSELECT was used to randomly select addresses within each buffer. Field staff enrolled participants using verbal consent and asked respondents about their park use, physical activity behavior using two items on frequency and duration from the Minnesota Hearth Health Program (Jacobs, Ainsworth, Hartman, & Leon, 1993), and awareness of and participation in park-sponsored classes and activities. We asked respondents how much time outside of work they spent watching television, DVDs, or videos; playing video games; or using a computer on an average day to measure screen time.

Table 1.

Park Neighborhood Characteristics (1-Mile Radius)

Park characteristics (N = 48)	Mean	Range
Acres	8.4	1.5–25.8
Number of facilities	8.1	4.0–14.0
1-Mile population estimate (average per park)	52,310	21,530–133,123
%Households in poverty	27.2	17.1–41.0
Population race or ethnicity (%)
Hispanic	68.7	15.7–94.9
White	9.0	0.3–71.0
African American	11.7	0.7–67.2
Asian	9.0	0.2–31.0
other/multiple	1.6	0.2–4.5
Violent crime rate (crimes per 1,000 population within 1-mile radius of parks)	72.8	11.8–164.5

Baseline survey participation rates were 82.5% for household respondents. The data collectors were Latino and the survey was available in English and Spanish. The RAND Human Subjects Protection Committee approved the study and an oral consent procedure for the surveys.

Annual crime data from 2012 to 2015 were obtained through the Los Angeles Open Data website (https://data.lacity.org). Data files contained geocoded crime records provided by the Los Angeles Police Department. Using ArcGIS, crime data were mapped using the provided latitude and longitude. A 1-mile buffer was created around each park using the geocoded park addresses. All reported crimes within the 1-mile buffers were selected and defined as frequency of crimes per 10,000 people in the 6-month period, with the population in the 1-mile radius as the denominator. To obtain the park-level poverty estimates (percentage of households whose income in the past 12 months is below the poverty level), we used the 2007–2011 American Community Survey 5-Year Estimates for the estimates at the Census Tract and ZIP code tabulation area level (U.S. Census, 2012). Estimates were based on a 1-mile buffer around each park using the geocoded park address as the centrum.

Data Analysis

We first stratified the data by gender and age group, for 18–25, 26–39, 40–59, and 60 and above, and examined responses to questions on individual sociodemographic characteristics, park use, and personal health. Because the participants were largely Latino, interaction analyses between race-ethnicity and other variables were not conducted.

We fitted a mixed-effect model to estimate relationships between age, gender, and park use, adjusting for individual, park-level, and neighborhood covariates, as well as within-park clustering. To explore the joint effects of age, gender, and child status, we divided respondents into three groups by using variables indicating gender and child status; one is for male, another one for female without a child under 18 years old, and the other one for female with a child under 18 years old. We did not stratify male by child status because there was no notable difference in the age effect among males with and without children. We included this three-level variable as well as its interaction with age, to estimate the group-specific associations with age. We controlled for other individual factors such as race/ethnicity, education, current employment status, health status, obesity, distance between the park and residence, screen time, and perception of safety. We also controlled for park-level factors such as park size, the number of people in the park, the proportion of males, and the number of organized park activity sessions, and for neighborhood-level factors including the proportion of households below the poverty line, population density within 1 mile, and the crime rate. We also controlled for the year of survey administration to control for any time trends. Random effects at the park level are included to consider park-level clustering. These mixed-effect models were all implemented in SAS 9.4 (SAS Institute Inc., Cary, NC).

Results

Population Characteristics of Residents

The majority of participants were Latino (67% of men and 78% of women) and roughly 10% White, 10% African American, and 2% Asian comprised the remainder. The largest age group of respondents was 40–59 years (42% of men and 45% of women). Other characteristics including employment and education are listed in Table 2. Except for asthma, where men aged 18–25 had higher rates than men aged 26–59, all medical problems increased with age. Except for heart disease, rates of medical problems among senior women were higher than senior men.

Table 2.

Descriptive: Population Characteristics of Resident Sample

	All men (N = 1,209)	Men aged 18–25 (n = 136)	Men 26–39 (n = 353)	Men 40–59 (n = 509)	Men 60+ (n = 211)	p value	All women (N = 1,741)	Female 18–25 (n = 196)	Female 26–39 (n = 578)	Female 40–59 (n = 781)	Female 60+ (n = 186)	p value
Race/ethnicity (%)
Latino	67.4	69.6	51.7	74.9	74.4	<.0001	78.0	79.1	69.7	83.4	80.1	<.0001
White	12.2	11.1	18.8	10.3	6.6	<.0001	8.6	9.7	10.8	6.2	11.3	.01
Black	12.7	14.1	15.6	11.1	10.9	.19	7.8	6.6	9.2	7.7	4.8	.24
Asian/Pacific Islanders	2.0	3.0	2.3	1.6	1.9	.74	2.0	2.0	3.5	0.9	2.2	.01
American Indian/Alaska Native	0.2	0.0	0.3	0.2	0.0	.83	0.0	0.0	0.0	0.0	0.0
other	5.5	2.2	11.4	2.0	6.2	<.0001	3.7	2.6	7.1	1.9	1.6	<.0001
Have child(ren) under age 18 (%)	36.2	9.6	48.6	49.7	0.0	<.0001	52.7	22.2	82.4	50.8	1.1	<.0001
Employment (%)
full time	57.2	24.4	75.4	73.9	7.6	<.0001	33.7	26.7	43.2	35.8	3.2	<.0001
part time	12.5	46.7	12.5	4.9	8.5	<.0001	16.5	27.7	16.3	16.3	6.5	<.0001
self-employed	7.8	3.7	6.0	10.7	6.6	.01	5.9	0.5	3.5	9.5	3.8	<.0001
unemployed	5.6	11.1	5.4	5.5	2.4	.01	3.9	4.6	3.8	4.3	2.2	.57
retired	11.8	0.0	0.0	1.8	63.0	<.0001	5.9	0.0	0.0	0.5	52.7	<.0001
disabled	6.2	0.0	0.0	4.6	24.6	<.0001	1.3	0.0	0.0	1.0	8.1	<.0001
student	8.6	46.7	11.1	0.4	0.0	<.0001	10.6	59.0	9.7	1.7	0.0	<.0001
home/parent	1.2	0.7	0.9	1.8	1.0	.56	45.7	10.8	48.0	54.8	37.1	<.0001
other	0.0	0.0	0.0	0.0	0.0		0.3	0.5	0.0	0.3	1.1	.11
Education (%)
high school	36.9	17.0	15.7	45.0	66.2	<.0001	30.1	4.6	14.7	38.7	68.9	<.0001
high school	25.9	40.7	27.4	27.4	10.5	<.0001	37.8	31.3	46.3	39.3	11.5	<.0001
some college	13.2	32.6	14.8	9.1	7.6	<.0001	18.0	52.3	18.3	11.3	8.7	<.0001
associate degree	12.2	7.4	22.2	7.1	10.5	<.0001	8.2	8.2	10.8	6.7	6.0	.03
bachelor’s	9.7	0.7	17.7	8.9	3.8	<.0001	5.6	3.6	9.4	3.3	4.9	<.0001
graduate school/degree	2.2	1.5	2.3	2.6	1.4	.72	0.5	0.0	0.5	0.6	0.0	.50
Medical diagnoses (%)
asthma	5.5	5.1	1.8	2.2	19.8	<.0001	4.3	1.6	2.4	2.1	22.4	<.0001
diabetes	14.4	0.0	0.9	16.5	39.6	<.0001	12.3	0.0	1.8	13.5	53.0	<.0001
high cholesterol	13.1	0.0	2.7	17.5	27.1	<.0001	11.0	0.0	2.8	13.5	37.4	<.0001
heart disease	5.5	0.0	0.0	4.4	20.3	<.0001	2.0	0.0	0.0	1.3	13.7	<.0001
hypertension	18.5	0.0	1.5	21.3	50.2	<.0001	15.9	0.0	1.2	20.9	57.9	<.0001
weight problem	8.5	1.7	4.7	10.8	13.0	<.0001	10.1	2.60	6.5	13.8	14.2	<.0001

Among men and women, there was a bell-shaped distribution in terms of ages and frequency of park use, with the peak use between ages 26 and 39 and the lowest use among seniors. In all age groups, except ages 40–59, men reported using the neighborhood parks more frequently than women. The duration of park use tended to decrease with age, although among women the differences by age were not significant. For women, the decline between ages 18 and 25 was only 6 min (p = .06), whereas in men, it was 41 min (p < .0001). For all age groups, except among seniors, men stayed longer in the park than women. Differences in use and duration of stay by age group are in Table 3.

Table 3.

Park Use, Neighborhood Residence Duration, and Physical Activity by Gender and Age of Resident Sample

	All men (N = 1,209)	Men aged 18–25 (n = 136)	Men 26–39 (n = 353)	Men 40–59 (n = 509)	Men 60+ (n = 211)	p value	All women (N = 1,741)	Female 18–25 (n = 196)	Female 26–39 (n = 578)	Female 40–59 (n = 781)	Female 60+ (n = 186)	p value
Frequency of use ≥l×/week	31.3	39.3	41.9	22.4	29.9	<.0001	25.9	26.2	31.3	24.1	16.8	.0004
Never visits park	52.9	42.2	34.0	55.1	53.1	<.0001	45.3	46.2	36.0	47.6	63.8	<.0001
Duration of stay (%)
0–60 min	27.0	9.0	18.5	32.3	48.5	<.0001	29.7	27.4	23.0	34.7	39.7	.0009
1–2 hr	43.1	39.7	48.7	44.1	30.3	.02	55.7	55.6	63.5	51.8	36.8	<.0001
≥2 hr	29.9	51.3	32.8	23.6	21.2	<.0001	14.6	17.0	13.5	13.5	23.5	.13
Average duration minutes	95	119	101	89	78	<.0001	84	88	87	81	81	.06
Transport mode
walk	43.6	43.6	33.9	38.6	47.5	.10	49.0	54.7	47.3	47.8	55.9	.34
bike	7.8	25.6	4.3	4.4	10.1	<.0001	0.5	0.9	0.5	0.5	0.0	.87
car	51.9	25.6	60.9	57.0	39.4	<.0001	49.7	42.5	51.9	51.2	39.7	.11
bus/public transportation	0.6	0.0	0.4	0.0	3.0	.01	0.6	0.9	0.3	0.5	2.9	.08
other	0.8	5.1	0.4	0.0	0.0	.004	0.2	0.9	0.0	0.0	1.5	.02
Frequency exercise/week (days)	2.5	3.00	3.0	2.2	1.9	<.0001	2.1	2.2	2.3	1.9	1.8	.01
≥150 min/week	30.8	39.3	47.4	23.9	14.7	<.0001	23.3	27.6	27.4	21.4	14.1	.0005
Run into people they know	72.9	92.3	76.4	67.1	62.6	<.0001	62.7	77.4	64.2	58.3	57.4	.003
Length of time in neighborhood
up to 5 years	47.2	56.0	69.9	40.6	19.4	<.0001	51.4	63.1	66.9	42.7	28.0	<.0001
5 years or more	52.8	44.0	30.1	59.4	80.6	<.0001	48.6	36.9	33.1	57.3	72.0	<.0001
Do not know staff	64.9	52.5	53.11	73.25	72.93	<.0001	65.47	63.19	57.86	69.61	76.19	<.0001
Park is safe	77.4	85.2	85.86	71.78	69.54	<.0001	73.39	78.38	76.51	69.14	76.32	.0156
Does not usually exercise	36.8	23.7	25.3	43.5	48.3	<.0001	47.6	46.2	41.5	50.13	57.30	.0005
Place of exercise
park	30.9	35.0	30.4	29.4	32.4	.74	33.5	25.7	37.3	35.0	20.3	.009
home	29.9	37.9	33.5	25.9	24.1	.03	33.5	38.1	33.4	31.6	36.7	.58
private club	14.9	14.6	23.6	11.2	3.7	<.0001	10.5	17.1	14.0	6.2	7.6	.0005
street/sidewalk	14.2	10.7	8.4	13.6	33.3	<.0001	19.1	15.2	14.0	21.3	34.2	.0002
other	10.1	1.9	4.2	19.9	6.5	<.0001	3.5	3.8	1.2	5.9	1.3	.005
Ever participated in park program	5.9	13.4	5.8	4.0	5.9	.0008	8.0	6.3	11.0	6.3	7.7	.01
Screen time
0–60 min	10.7	11.1	8.8	13.0	8.1	.13	7.9	8.7	7.8	7.6	8.6	.93
1–2 hr	26.0	16.3	21.3	34.0	21.0	<.0001	24.2	21.0	25.0	26.9	14.0	.002
2–4 hr	33.8	40.0	41.2	32.8	22.4	<.0001	40.2	41.0	43.0	43.2	18.3	<.0001
≥4 hr	20.7	26.7	23.6	13.8	28.6	<.0001	19.7	25.6	18.7	17.2	26.9	.0030
Do not know	8.7	5.9	5.1	7.3	20.0	<.0001	8.0	3.6	5.6	5.1	32.3	<.0001
Average minutes	159	175	169	141	178	<.0001	165	173	163	160	184	.006

The percentage of residents who walked to the park as their usual mode of transport did not differ by age group or gender. The youngest men and women were more likely than other age groups to report running into people they knew at the park. By contrast, the oldest age groups of both genders, despite being more likely to live in the neighborhood the longest (5 years or more), were the least likely to say they knew the park staff, inversely paralleling the frequency with which they visited the park.

Respondents named the park and their homes as the two top places they usually exercised, with home being more common among the younger men and women, aged 18–25, and park being more common among adults aged 26–59. The percentage of respondents who said they did not usually exercise increased with age for both men and women; except that women aged 18–25 reported “not usually exercising” more than women aged 26–39 (46% vs. 42%). Persons aged >60 years were the least likely to exercise (48% of men and 57% of women). Perception of safety declined with age for men, but did not uniformly decline among women. After age 40, the prevalence of perceptions that the park was safe was similar among men and women (Table 3).

Figure 1a–1c shows the most common park activities by age and gender. For both women and men, sitting in the park was the most cited activity and increased with age for men. For men, basketball was the second, whereas for women, using the playground area was the second. Men cited sports as a reason to go to the park more than women, but we did not distinguish between participating in the sport or spectating. Meeting friends at the park was among the top activities but declined with age for both genders and was highest among men aged 26–39 (49%) and women aged 40–59 (41%).

Figure 1 —

Top activities in parks reported by neighborhood residents: (a) men by age; (b) women by age; (c) by gender.

Predictors of Park Use Among Residents

Controlling for numerous other individual, neighborhood, and park-related factors, age has significant relationships with frequency of park visitation and duration which differ among men, women with children, and women without children (Table 4). For the number of visits to parks in the last 7 days, age had a significantly negative relationship among men (β = −0.014, p < .01), but was insignificant for women. In younger ages, women had significantly fewer park visits than men on average. Due to the differential age association, the difference diminishes among older groups. For example, controlling for all covariates, at age 25, women without children had 0.7 fewer weekly visits (p < .001) and women with children had 0.4 fewer weekly visits (p = .02) compared with men of the same age. However, at the age of 50, women without children still had 0.5 fewer visits per week (p < .001), but women with children visited parks at the same rate as men of the same age.

Table 4.

Multivariate Associations Between Age and Park Use and PA Among Neighborhood Residents

	No. of visits to parks in last 7 days (n = 1,950)	Typical duration of park visit (min; n = 1,374)
Characteristics	Mean [95% CI]	Mean [95% CI]
Individual
intercept	0.78 [−0.97, 2.54]	93.0 [33.7, 152.3]**
females having children <18 years old (vs. male)	−0.83 [−1.52, −0.13]*	−33.0 [−52.0, −14.0]***
females not having children <18 years old (vs. male)	−0.96 [−1.52, −0.41]**	−27.5 [−44.1, −11.0]**
age × males	−0.014 [−0.023, −0.005]**	−0.8 [−1.0, −0.5]***
age × females not having children <18 years old (vs. age × males)	0.009 [−0.003, 0.021]	0.4 [0.0, 0.7]*
age × females having children <18 years old (vs. age × males)	0.019 [0.002, 0.036]*	0.6 [0.1, 1.1]*
Black vs. Whites/others	0.08 [−0.25, 0.40]	8.8 [−0.7, 18.3]
Latino vs. Whites/others	−0.30 [−0.56, −0.05]*	−4.7 [−11.9, 2.6]
High School graduate (vs. <HS)	0.04 [−0.16, 0.23]	1.1 [−4.7, 6.9]
some college (vs. <HS)	0.31 [0.07, 0.56]*	−1.3 [−8.3, 5.7]
college graduate (vs. <HS)	0.35 [0.08, 0.62]*	−7.3 [−15.0, 0.4]
obese (BMI > 30)	0.07 [−0.13, 0.26]	3.9 [−1.8, 9.5]
fair or poor health status	0.36 [0.15, 0.57]***	2.3 [−4.0, 8.6]
lives 0–¼ mile from park (vs. >½ mile)	0.74 [0.57, 0.92]***	−3.8 [−9.0, 1.4]
lives ¼–½ mile from park (vs. >½ mile)	0.29 [0.11, 0.47]**	2.2 [−3.3, 7.6]
perceives park as safe	0.29 [0.11, 0.48]**	9.8 [3.6, 15.9]**
screen time (12 hr) vs. 0–60 min	0.17 [−0.12, 0.45]	−1.8 [−10.0, 6.4]
screen time (2–4 hr) vs. 0–60 min	0.06 [−0.21, 0.33]	1.8 [−6.2, 9.8]
screen time (4+ hr) vs. 0–60 min	0.21 [−0.08, 0.50]	3.8 [−4.7, 12.4]
currently works full time	−0.13 [−0.30, 0.04]	1.4 [−3.6, 6.3]
currently works part time	0.42 [0.20, 0.64]***	−0.7 [−6.8, 5.4]
Park
acres	0.03 [0.00, 0.05]*	0.0 [−0.8, 0.9]
observed number of park users	0.00 [0.00, 0.01]	0.3 [0.1, 0.5]**
%male park users	0.00 [−0.02, 0.02]	0.2 [−0.6, 1.0]
number of organized activity sessions	0.00 [−0.02, 0.01]	−0.2 [−0.7, 0.2]
Neighborhood
number of total crimes 6 months after start of baseline observation	0.00 [0.00, 0.00]	0.0 [−0.1, 0.0]
%households in poverty within 1 mile	0.00 [−0.02, 0.02]	−0.2 [−0.7, 0.4]
pop. within 1 mile	0.01 [−0.07, 0.10]	−0.3 [−3.4, 2.8]

Note. Boldface indicates statistical significance. BMI = body mass index; CI = confidence interval; HS = high school.

^*p < .05.

^**p < .01.

^***p < .001.

For the typical duration of a park visit, age had a consistently negative relationship in all three age-gender-child status groups, but the strength of the association differed. The relationship was the strongest in men (β = −0.75, p < .001), followed by women without children (β = −0.37, p = .02), and then nonexistent in women with children (β = −0.16, p = .44). Controlling for all covariates, an increase of 20 years in age was related to an average decline of 15, 7.4, and 3.2 min in a typical park visit in men, women without children, and women with children, respectively. A 65-year-old man would stay in a park nearly 30 min less per visit than a 25-year-old man on average, controlling for other covariates. For the same comparison, a woman without children would stay 15 fewer minutes and a woman with children would stay 6 fewer minutes.

There were a few other significant covariates in the models including education, race/ethnicity, employment status, proximity to the parks, safety perceptions, and self-rated health (Table 4). Park size was associated with frequency of park visits, but the impact was relatively small such that every additional acre was associated with an increase of 0.03 visits per week. This would translate to a person visiting a 12-acre park 0.3 times more per week than a 2-acre park. The number of observed park users was associated with longer duration of visits to the park, such that seeing an additional 50 people in the park would be associated with a 13.5-min longer stay. The number of park visits and duration of stay were not associated with the rate of violent crime, but were associated with individual perception of safety.

Discussion

Our findings are consistent with the general age-related trajectory of physical activity decline and confirm gender-related disparities in park use among low-income residents (Floyd et al., 2011; Joseph & Maddock, 2016). The results suggest that family related responsibilities may explain a great deal of the discrepancy of park use and park-based physical activity among men and women, and that for women, the age-related decline associated with park use is mitigated by having children. Women’ s visit to parks in low-income neighborhoods appears to be largely related to their being the caretakers who bring children to parks. Their activities while in the park tend to be more sedentary (sitting, playground, celebrations) and rarely involve sports-related activities. In contrast, men’s visits to parks are more likely to involve their own participation in sports and other recreation.

Surprisingly, leisure time appears not to be closely related to park use. Individuals with part-time jobs visited parks more than those who were unemployed. Seniors, who are most likely to be retired, visit parks the least. A study by Loukaitou-Sideris et al. (2016) indicates that an important barrier to seniors’ use of parks was the presence of younger age groups, by whom they felt threatened with respect to potentially being inadvertently knocked over by a skater or runner, hit with a ball, or even the victim of assault.

Across both genders, the percentage of people who said they do not usually exercise is relatively high, 36.8% for men and 47.6% for women. This frequency increases substantially with age so that more than half of all seniors do not usually exercise. These self-reported rates among low-income area residents are higher than the national average of 25% of adults who do not engage in routine leisure-time physical activity (Centers for Disease Control and Prevention, 2014).

There is a high level of screen time among low-income populations, which may be related to lower-income groups experiencing higher levels of stress, which can lead to cognitive overload, higher participation in passive activity, and lower capacity to make more thoughtful decisions that yield long-term benefits (Mullainathan & Shafir, 2013). A higher stress level may play a role in fewer people choosing to be involved in outdoor activities, given that it requires more effort than passively engaging with electronic media.

Yet screen time was not associated with park visits or duration of stay, possibly because the time of the days when screens are viewed may not coincide with times that parks can be visited (e.g., after dark). However, concerns about safety were strongly associated with fewer and shorter park visits. In our model, total crimes were not associated with park use, similar to several other studies that indicate that perception of crime is more strongly related to individual factors than objective measures of crime (McGinn, Evenson, Herring, Huston, & Rodriguez, 2008; Nehme, Oluyomi, Calise, & Kohl, 2016; Rader et al., 2014). In particular, in a previous study, only crimes related to homicide and shooting were associated with park use, but not perceptions of safety (Han et al., 2018). Furthermore, our analysis is limited to low-income areas, and the data may not have sufficient variability in crime, especially if a threshold has already been reached across the 48 neighborhoods.

Given the amount of attention paid to screens or technology, electronic media may be the best venue to promote more park use and outdoor activity to a low-income population. Media should consider devoting a portion of their broadcasts to launch messages to remind people of the importance of physical activity and the availability of local recreational facilities and programming.

Sitting was the top park-based activity mentioned by women of all ages and by most men. Walking was the third most common activity for both men and women. Yet not all parks had attractive sitting areas, and only 31% had walking paths. Indeed, what people do in the park is constrained by what facilities and programming are available. The reason that meeting friends in the park declined with age among women may be correlated with children growing up and for men, with lower participation in sports. The low number of seniors using public parks may be, in part, related to health conditions that make it difficult to go to the park. Paradoxically, however, those who rated their health as poorer were more likely to visit the park. On the other hand, those who are in poor health may be more motivated to increase their level of exercise to improve it.

Limitations

The study participants were selected from low-income neighborhoods, so the findings cannot be generalized beyond this population. In addition, because we used a household sample, we recruited more females than males, and possibly more unemployed persons than what might be the true population prevalence. In addition, the results rely on the self-report of physical activity and park use. Self-report of physical activity is known to be positively biased (Craig et al., 2003), although reports of park use have been validated as reliable (Evenson, Wen, Golinelli, Rodriguez, & Cohen, 2013).

Conclusion

Given that physical activity is important across the life span, and especially important to mitigate and prevent chronic diseases associated with aging, parks need to be developed to meet the needs of an aging population. Their current emphasis on sports fields and courts favors young males, who are more likely to take advantage of these facilities. Playgrounds should be improved to include more opportunities for active participation by caretakers (who are primarily female), so they could increase their own levels of physical activity while supervising children. Future park renovations should consider the creation of social meeting places in parks, comfortable sitting areas and walking paths. Moreover, in low-income communities, efforts will need to be even greater than in wealthier communities to overcome barriers related to safety, daily stress, and the proclivity toward passive entertainment. Special efforts to reduce the perception that people of color may risk police harassment in public parks may be necessary to address fears among local residents. Consideration should also be given to routinely implementing activities, more programming, and classes targeting women and seniors, community events, including festivals, evening programs like Los Angeles County’s Parks After Dark program (Fischer, Welsing, Aragon, & Simon, 2014) that encourage a greater proportion of the population to become physically active.