Skip to main content
NCBI home page
Elsevier - PMC COVID-19 Collection logoLink to Elsevier - PMC COVID-19 Collection
. 2023 Apr 4;88:102008. doi: 10.1016/j.jenvp.2023.102008

Developing and testing a protocol to systematically assess social interaction with urban outdoor environment

Shuolei Chen a,∗, Ole Sleipness b, Keith Christensen b, Bo Yang c, Hao Wang a
PMCID: PMC10082984  PMID: 37065612

Abstract

Social interaction is significant for individuals’ mental and psychological well-being in communities. With the growing demand for outdoor activities in urban settings under the COVID‑19 pandemic, urban parks have become important public resources for human social activities. Researchers have developed numerous instruments to measure park use behaviors, but most are designed for assessing physical activity while ignoring social interactive behaviors. Despite the importance, no single protocol objectively assesses the range of social interactions in urban outdoor environments. To bridge the research gap, we have developed a social interaction scale (SIS) based on Parten's scheme. The innovative protocol, named Systematically Observing Social Interaction in Parks (SOSIP) was developed based on the SIS, allowing systematic evaluation of human’s interactive behaviors in outdoor environment both from their levels of social interaction and group size. The psychometric properties of SOSIP were established through the verification of content validity and reliability tests. Additionally, we applied SOSIP to explore relationships between park features and social interaction via hierarchical linear models (HLMs). Statistical comparisons between SOSIP and other forms of social interaction were discussed and indicated strong reliability of applying SOSIP. The results indicated SOSIP is a valid and reliable protocol for objectively assessing social interactive behaviors within urban outdoor environments and informing better understanding of individuals' mental and psychological health benefits.

Keywords: Social interaction, Urban open space, Mental health benefit, Outdoor, Physical activity, Psychological well-being

1. Introduction

1.1. Outdoor activities and the health benefits

Urban dwellers often face physical and mental health issues ranging from social isolation to lack of physical exercise (Zhou & Parves Rana, 2012). Urban open space such as parks can mitigate these issues by providing opportunities for outdoor activities (Xu, 2021), including physical exercise and social interaction (Veitch et al., 2021). In particular, lockdowns and restrictions since the COVID-19 pandemic have resulted in increased demand for social interaction in outdoor spaces, and parks became an important public resource for people's outdoor social activities (Geng et al., 2021; Venter et al., 2021). Social interaction in parks is a significant activity for urban residents to achieve the mental health benefits in addition to opportunities for physical activities (PA).

Social interaction is significant for individuals’ mental health and social cohesion of communities (James et al., 2015; Jennings & Bamkole, 2019). By reducing social isolation for urban dwellers since the pandemic, outdoor social interaction contributed to the positive relationships in the communities and improved mental health (Schiller et al., 2021). Growing numbers of scholars have noted the significance of social interaction in urban open space and recommended its inclusion in evaluations of outdoor activity (Dadvand et al., 2019; Maas et al., 2009; Veitch et al., 2021). However, most tools that are used to assess park use were developed from the perspective of PA while ignoring social interaction as a vital outdoor activity within urban open space (Chen et al., 2020).

Multiple disciplines have measured PA, regarded for decades as the significant park use activity for achieving physical and mental health benefits. Tools to measure PA, such as the System for Observing Play and Recreation in Communities (SOPARC) (McKenzie et al., 2006) and Physical Activity in the Park Setting (PA-PS) (Walker et al., 2009), often assess park-based PA according to their intensity and number of participants. As the most prevalent tool for assessing park-based PA, SOPARC has become popular with researchers (Joseph & Maddock, 2016). It also has been adapted for a range of purposes beyond PA assessment (Whiting et al., 2012). SOPARC allows observers to use momentary time sampling to record the number of park users and degree of PA intensity via systematic scans of visitor behavior (McKenzie et al., 2006). The application of protocols such as SOPARC ensured that various case studies assessed the activity in a standardized manner (Poortinga et al., 2017).

1.2. Methods for assessing outdoor social interaction

Social interaction refers to the degree of bonds or interactivities between two or more individuals (Rasidi et al., 2012). It can be reflected by both the sense of connectedness to others and one's degree of interactive behaviors. Social interactions occurring in urban outdoor environments were assessed through documenting the number of socializing individuals as well as the patterns and degrees of observed social activities (Kaźmierczak, 2013; Veen et al., 2016), which has been summarized as reflections of effective mechanisms of mental health benefits and well-being in open spaces (Wan et al., 2021). Examples capturing social interaction happening in urban outdoor environments including parks are provided in Table 1 .

Table 1.

Literature measuring social interaction/contacts in urban outdoor environment.

Literature Employed a Social interaction protocol/tool? Methods Measures related to social interaction/
Skjœveland (2001) No Questionnaire-based survey and field observation Survey measures: supportive act in neighbors, social ties, neighborhood attachment, neighbor annoyance; Observation measures: number of individuals presented in the parks
Maas et al. (2009) No Survey Loneliness; social support; contacts with neighbors and friends in the community
Seeland et al. (2009) No Questionnaire-based survey Residents' peer groups; ways of making friends
Peters et al. (2010) No Observation Presented social groups (whether they interact, and if so, their peer groups); gender, age, origin; number of their social activities; the sub-areas where the social activities happened
Yamada and Knapp (2010) No Questionnaire-based survey Preferred types of social interaction in parks
Rasidi et al. (2012) No Observation (representative of weekday, weekend, and public holiday) Social division: individual/group; social preference: activity/program; microclimate; personal info: gender, age group, race/ethnic
Kemperman and Timmermans (2014) No Survey Social contacts with neighbors: six statements about the respondents' relation with neighbors in the direct living environment
Campbell et al. (2016) No Combines observation with structured interviews of park users Human activities and signs of prior human use: functionally grouped human activities (sitting, socializing, bicycling, exercise, nature recreation); observed age (youth, adults, senior), race
Hillier et al. (2016) No SOPARC Total number of people present, whether at least one male and one female were present, only males were present, or only females were present, and whether any children, teens, adults, or seniors were present
Moulay et al. (2017) No Questionnaire-based survey Intensity of contact; engagement with parks; types of interaction
Salih and Ismail (2018) No Questionnaire-based survey Social interaction types and forms
Dadvand et al. (2019) No Survey Number of close friends, number of days per week spending with friends
Schmidt et al. (2019) No Observation (modified SOPARC) Number of social interaction (if two or more persons are talking, walking, running, biking, and sitting together), their age and gender
Tao et al. (2020) No Survey Reported frequency of communication with neighbors
Open in a new tab

SOPARC: System for Observing Play and Recreation in Communities (McKenzie et al., 2006).

In the literature, the significance of measuring outdoor social interaction has been documented. Numbers of quantitative studies have explored the relationships between social interaction and the features or characteristics of built environments, such as parks, to inform the planning and management of these places (Campbell et al., 2016; Rasidi et al., 2012; Veitch et al., 2021). Many survey-based studies have measured social interaction from respondents' self-reported sense of contact with neighbors or friends (Dadvand et al., 2019; Kemperman & Timmermans, 2014; Maas et al., 2009; Skjœveland, 2001; Tao et al., 2020), but have not focused on social behaviors and activities occurring in urban open spaces. Survey methods can inquire about people's subjective sense of contact and attachment with their community. Self-reported methods like surveys assess social interaction indirectly through description and interpretation, which has not been established to directly provide objective measures for statistical analysis (Hill et al., 2014).

Additionally, some scholars observed human social interactions in parks and open spaces (Campbell et al., 2016; Peters et al., 2010; Rasidi et al., 2012). Observational studies often record human behaviors and activities, such as participation in sports, walking, chatting, and resting. Compared to surveys, systematic observation allows for simultaneous generation of information and ensures the collected information has strong internal validity, especially for quantitative research (McKenzie & Van Der Mars, 2015). These studies entailed counting the number of individuals engaged in different categories of human activity but did not classify the individuals’ level of social interaction. For example, Peters et al. (2010) recorded the number of people present in the park, and whether they interacted with others. In another example, Campbell et al. (2016) grouped human activities in parks according to functional categories such as sitting, exercising, socializing, and nature-based recreation. Other observational studies noted the importance of following a protocol but did not identify one designed specifically for assessing social interaction. Consequently, they used SOPARC, which was not designed for measuring social interaction but rather for PA (Hillier et al., 2016; Schmidt et al., 2019). Although Schmidt et al. (2019) modified SOPARC to fit their research goal of measuring social interaction, SOPARC is still tailored for PA; the validity and reliability of using the protocol for measuring social interaction have not been established.

1.3. Levels of social interaction

Most observational methods measured social interaction in urban outdoor environment by recording the number of other people with whom an individual socialized (Campbell et al., 2016; Dadvand et al., 2019; Peters et al., 2010; Rasidi et al., 2012). Limited study considered the degree of socialization within groups. According to the definition of social interaction, both the level of social interaction and the number of socializing people should be considered as measurements of social interaction (Kaźmierczak, 2013; Veen et al., 2016).

Generalizing the degree of social activities into patterns allowed people's narrative descriptions of social behaviors to be characterized as coded data for scientific research (Bakeman & Gottman, 1997; Bode et al., 2017). To reflected the mental health benefits achieved from the degree of involvement and interaction with others, Parten (1932) generalized the social activities into six levels and established an influential coding scheme for assessing social interaction among children through systematic observations. It has been psychometrically well-established and commonly

used to assess levels of social interaction (Bakeman & Brownlee, 1980). The predefined activities according to Parten included 1) Unoccupied Play: Child is observing, not playing; 2) Solitary Play: Child plays alone and is uninterested or unaware of others; 3) Onlooker Play: Child observes other children playing but does not take part; 4) Parallel Play: Child plays next to another child and is more interested in the activity than the partner; 5) Associative Play: Child interacts with other children but in an unorganized and uncoordinated manner; and 6) Cooperative Play: Child engages with other children in an organized activity.

Parten's characterization of social interactions has been widely applied to other age groups in different settings (Bakeman & Gottman, 1997; Charko et al., 2016; Coplan et al., 2015; Donohue, 2015; Loebach et al., 2020). Subsequent scholars have discussed whether subtle distinctions between the codes were necessary (Bakeman & Gottman, 1997). For instance, Smith (1978) reduced Parten's six levels of social interaction to three, which are: 1) Alone (combining the first three categories of Unoccupied, Onlooker, and Solitary), 2) Parallel (using the same definition as the Parallel as Parten's scheme), and 3) Group (combining Parten's last two categories of Associative and Cooperative). Parten's coding scheme established a proper theoretical foundation for the development of a social interaction scale to examine levels of social interaction in outdoor spaces, and multiple versions of this scale should be tested.

1.4. Research objectives

With the growing demand for outdoor social interaction, researchers have quantified social interaction variables and studied the relationships between social interactive behaviors and environmental features. However, these studies measured social interaction without following an established instrument, and a reliable and valid protocol had not been developed. Compared to other types of urban outdoor environments, urban parks are equipped with more facilities, amenities, and nature environments to support residents' different outdoor activities. While some scholars considered the degree of social interaction and the number of social participants in outdoor environments like neighborhoods, park use research still primarily focused on PA or highlighted some social activities, such as chatting and meeting, but without defining the social park use activities according to the degree of social interaction. Most notably, no single protocol has assessed social interaction comprehensively by simultaneously considering both the degree of contact and the number of individuals in the group. The current study strives to bridge the research gaps through the development and testing of a protocol based on Parten's coding theory, which was designed to measure social interaction in urban outdoor environment, especially for urban park settings.

2. Methods

2.1. Development of SOSIP

We developed Systematically Observing Social Interaction in Parks (SOSIP). It provides objective data on both the number of participants and their corresponding degree of social interaction through systematic observation in urban outdoor environments. The Social Interaction Scale (SIS), which is the theoretical framework of SOSIP, interprets the degrees of people's social behavior.

2.1.1. Social interaction scale (SIS)

Based on Parten's coding scheme, SIS defined the level of social interaction from low to high using six levels: 1) Solitary, 2) Solitary Onlooker, 3) Onlookers, 4) Parallel, 5) Associative, and 6) Cooperative. Point values from 1 to 6 were assigned to each level to quantitatively assess the social interaction of the general population in urban outdoor environment (Table 2 ).

Table 2.

Interpretation of SIS (Social Interaction Scale) for applying SOSIP (Systematically Observing Social Interaction in Parks) in park settings.

SIS Point Value Descriptions Examples
Alone
 Solitary 1 An individual who is alone and uninterested or unaware of others. An individual is working/reading/writing in a park without noticing anyone besides themselves.
Solitary Onlooker
 2
 An individual who is alone but is interested in or observing others.
 An individual sitting on a lawn by themself but watching others' activities.
Group Onlookers 3 Individuals in a group setting who are observing others' movement but not taking part in or communicating with each other. A group of people sitting next to each other watching a ball game but not talking with each other.
Parallel 4 People are in a group activity, but they are more interested in the activity than the partner beside them. A group of boys may skateboard together in a park, but they are focused on skateboarding without communicating with each other. People go fishing together but remain in their worlds and do not talk.
Associative 5 Individuals in a group are interacting with others but in an unorganized or informal manner. A group of people is gathering informally for a birthday party in a park.
Cooperative 6 A group of people engaged with others in an organized or formal activity. A group of people played a basketball game in a park, in which each one of them may have a distinct role in the game.
Open in a new tab

2.1.2. Observation techniques of SOSIP

SOSIP is innovatively designed to evaluate social interaction, according to SIS. According to the first step of scanning in Fig. 1 , the observers with SOSIP are required to walk across the target park to locate all present park users, and scan the whole area from left to right for momentary congregation of park users in discrete groups. In the occasions of large size or high user density, observers are suggested to subdivided the park into sub-areas and scan these areas for an accurate counting of park users' data. Data from the sub-areas are summed to provide an overall measurement for the whole park. For example, a picnic area on a park's periphery would be recognized as a sub-area for discrete observation separately from a athletic field in middle of the same large park due to the observers can't scan them at the same time.

Fig. 1.

Fig. 1

Open in a new tab

Summary of steps in SOSIP for training of the observers.

To ensure that the systematic observation captured spontaneous behaviors, observers using SOSIP were required to conduct unobtrusive observations. Based on the momentary time sampling for systematic observation suggested by SOPARC (McKenzie et al., 2006), observers with SOSIP collected momentary data while walking across the park and visually scanning the area to capture the activities of visitors on the first scan, and determine the corresponding levels of social interaction according to SIS. Observers can take pictures of the area to record the level of social interaction and the group size, to characterize their observations as snapshots of social interaction.

However, mischaracterizations of social interaction assessment can occur on momentary time sampling occasionally if the observers fail to capture the accurate information on the first scan, especially for beginners using SOSIP without the assistant of a camera. To avoid mischaracterizations, the training of SOSIP allowed the observers to stay on site if they need enough time to determine the levels of social interaction and/or the group size of the first scan, but the changes of social interaction and group size cannot be accounted during the rest of the observation time. For instance, an observer with SOSIP found a group of boys playing basketball in a park, but the group stopped playing and turned to a water break when the observer counted the number of participants. For this circumstance, the observer only recorded the level of social interaction on the first scan, which is basketball playing rather than a water break. SOSIP observers can continually watch the park users' activities, but only for determining the social interaction and/or group size if they didn't catch the information on the first scan and ignore if there is changes of social interaction level of the user(s) or their group size.

2.1.3. Identification and data collection

Based on SIS, observers following SOSIP protocol can identify the social interaction for each individual/group (SISUP) (the step of record in Fig. 1). SISUP were assigned points from 1 to 6 (Table 2). First, observers must note whether park visitors are alone (by themselves) or in a group (at least two people). Individuals who gathered and undertaking the same social activities are defined as a group with SOSIP. For instance, if observers found four people stayed close in a park, with two people chatting while two others played a card game, there were two distinct groups of two undertaking different social activities. If a park visitor was alone, he or she needed to be further classified as Solitary (1 point) or Solitary Onlooker (2 points). If there was a group, SISUP needed to be categorized as Onlookers (3 points), Parallel (4 points), Associative (5 points), or Cooperative (6 points).

Observers following SOSIP also needed to record how many people gathered and joined the activity (group size) and their apparent demographic information, including gender—male (M), female (F), both (within a group [B], or cannot be ascertained [O]; and race and ethnicity—classified into: White (W), Hispanic (H), Black (B), Asian (A), Mixed (mixed-race groups gathered (M), and Others (previously unmentioned races (O). Apparent age groups were also needed to be recorded as follows: youth under 18 years old (Y), adults aged 18–65 (A), seniors over the age of 65 (E), or people from different age groups gathered in a group, such as a family (M). The weather and temperature were also noted.

2.1.4. Repeat and calculation of the social interaction score

The protocol aimed to capture the different use patterns at different times on a single day and different days of a week (the step of repeat in Fig. 1). The target parks/spaces were required to be observed in the morning and afternoon for at least three days, including a weekday, a Saturday, and a Sunday, which had been indicated as a proper frequency of systematic observations (Cohen et al., 2011). In this case, each park was observed six times—two times (morning and afternoon) on three different days (a weekday, a Saturday, and a Sunday), which was the step of repeat in Fig. 1. Each observation created a social interaction score, and each park had a total of six social interaction scores.

After each observation, the observers collected the data including SISUP of all park users across the park and their group size (the last step in Fig. 1). The social interaction score for each observation is calculated by the sum of multiplying the SISUP and the corresponding group size for each group showing in the park during the observation, as illustrated in Equation (1) below (i equals the counting of presented groups/individuals on the first scan for each observation). In this case, people's social interaction behaviors can be measured through both their levels of social interaction and the number of the individuals in that group. For example, observer found 10 boys playing basketball and a woman reading alone in a park for an observation. An SISUP of 6 and group size of 10 were recorded for the group of boys, whereas an SISUP of 1 and group size of 1 were recorded for the woman. The social interaction score for this observation is 6 * 10 + 1 * 1 = 61.

Social Interaction Score = ⅀ (SISUPi * Group Sizei) (1)

2.2. Testing of SOSIP

2.2.1. Validity of SOSIP

The content validity of SOSIP was established through construction of the SIS. The SIS is developed according to Parten's (1932) scheme, which was tested for the validity to determine the levels of social interactions and established the psychometric properties (Bakeman & Brownlee, 1980; Bakeman & Gottman, 1997). Even though Parten's scheme was originally applied to children, it has been evaluated as a valid coding scheme for assessing social interaction for different age groups (Charko et al., 2016; Coplan et al., 2015) and in different settings including urban open space (Donohue, 2015; Loebach et al., 2020). To verify the psychometric utility of SOSIP, a parallel-forms of reliability test was conducted to demonstrate the internal consistency of the protocol.

2.2.2. Interrater reliability of SOSIP

Following approval of the institutional review board (IRB) from Utah State University in fall 2019, a preliminary observation with SOSIP was conducted in Logan, Utah, to test the protocol's reliability. Four observers were trained and assigned in pairs to assess people's social interaction with SOSIP in 30 selected parks (Fig. 2 ). Every time, two of the observers joined the systematic observation simultaneously and independently, to ensure the results were not influenced by others. A total of 180 observations across the 30 parks were implemented with two observers' records.

Fig. 2.

Fig. 2

Open in a new tab

Distribution of the selected urban parks in Logan and North Logan.

A one-way random intraclass correlation coefficient analysis (ICC) was calculated to estimate the interrater reliability of social interaction scores in IBM SPSS Statistics 24. Several ICC tests were conducted to estimate the interrater reliability of social interaction scores categorized by demographics, including age, race, and gender.

2.2.3. Applicability and parallel forms reliability of SOSIP

As an increasing number of studies focused on the relationships between social interaction and the features of built environments, we demonstrated the application of SOSIP in exploring associations between park features and social interaction as an example. Here, we used some park features as example variables in relation to social interaction. Additionally, the parallel form reliability of SOSIP and multiple versions of SIS were tested.

The example independent variables (IV) included quality, size, and type of parks. We employed a direct observational instrument, the Parks, Activity, and Recreation among Kids (PARK) tool, which was evaluated the validity and reliability and constructed the psychometric properties (Bird et al., 2015), to measure park quality of the 30 selected parks in Logan (Chen et al., 2019). Park quality consisted of several park features which have been summarized as contributory to mental health by facilitating social activities in parks, such as landscaping, safety, amenities (Wan et al., 2021). The psychometric properties of these park features were established through multiple statistical explorations in various tools (Besenyi et al., 2018; Kaczynski et al., 2012; Malek & Nashar, 2018). The dependent variable (DV), social interaction, was collected during the preliminary test.

To further establish the reliability of the protocol, the comparative dependent variables (CDV) transformed from the original DV were used to test the reliability of the parallel form. The CDV was created in response to some argument of unnecessary distinctions of the six levels of social interaction. We tested whether it is necessary to assess social interaction according to Parten's coding scheme. To achieve this, we reclassified the collected SIS into the three-level social interaction scale, according to Smith (1978), which introduced a reduced version of Parten's coding scheme (Table 3 ). Also, the corresponding sample sizes based on the six-level scale were regrouped into Smith's consolidated three-level scale. The CDV was the sum of multiplying the three-level SIS by the reclassified group size.

Table 3.

Reclassification of SIS (Social Interaction Scale) according to a reduced version of Parten's coding scheme.

Six-level SIS Point value Description Three-level SIS (Smith, 1978) New point value
Solitary 1 an individual who is alone and uninterested or unaware of others Alone
 1
Solitary Onlooker 2 an individual who is alone but is interested in or observing others
Onlookers
 3
 individuals in a group setting who are observing others movement but not taking part in or communicating with each other
Parallel
 4
 people are in a group activity, but they are more interested in the activity than the partner beside them
 Parallel
 2
Associative 5 individuals in a group, interacting with others, but in an unorganized and uncoordinated manner Group 3
Cooperative 6 a group of people engaged with others in an organized activity
Open in a new tab

In our example, the associations between IV (park feature) and DV (social interaction) and the CDV (social interaction with the three levels of SIS) were explored using the hierarchical linear model (HLM), because the variables demonstrated a hierarchical data structure. Because each park was observed six times and six social interaction scores were created, the IV—park feature—was the higher-level unit (Level 2) in the hierarchical structure; it allowed the intercept and slope to vary randomly across a higher level. Social interaction scores (DVs) were nested under each park and became the Level 1 unit. All random slope HLM analyses were implemented with the use of the “lme4” package in R programming language (Bates et al., 2013).

The other control variables (nominal) were coded as park type (community park, C; greenway, G; neighborhood park, N; special use park/facility, S; pocket park, P), weather (sunny, S; rainy/cloudy, R), weekday/weekend (weekday, A; Saturday, B; Sunday, C), and time (morning, A; afternoon, B). To keep all the variables in a consistent unit, we standardized them into a 0–1 scale. The first HLM analysis (HLM1) explored the associations between IV and DV. Keeping the IV consistent, another HLM (HLM2) studied the relationships between park features and social interaction on a reduced scale (CDV).

3. Results

3.1. Sample characteristics

The descriptive statistics of the IV, DV, and CDV are shown in Table 4 . The histograms with standard curves indicate that the skewness of IV, DV, and CDVs is between −1 and 1, whereas park size and temperature are normally distributed.

Table 4.

Descriptive statistics of the IV, DV, CDV, and Continuous Control Variables.

Mean SD Range
IV
Park Feature (Score) 53.3 17.15 (0, 78)
(standardizeda) 0 1 (-2.51, 1.60)



DV
Social interaction (Score) 78.60 112.30 (0, 873)
(standardizeda) 0 1 (-.58, 6.40)



CDV (Reclassified Social Interaction Score) 46.73 80.53 (0, 529)
(standardizeda) 0 1 (-.52, 5.81)



Control Variables (Continuous)
Temperature (°C) 16.62 5.53 (7, 29)
(standardizeda) 0 1 (-1.30, 2.25)
Park Size (Acres) 7.7 7.3 (.46, 25.18)
(standardizeda) 0 1 (-.96, 2.39)
Open in a new tab

IV: Independent Variables, DV: Dependent Variables, CDV: Comparative Dependent Variables.

a

Standardized to 0–1 scale.

3.2. Reliability of SOSIP

3.2.1. Interrater reliability

Table 5 shows the summary of percent agreement by domain indicated by the ICC scores, as collected by the paired observers. When ICCs >0.75, it is considered to reflect strong reliability; when the ICC is between 0.50 and 0.75, it means moderate reliability; and when ICCs <0.5, it illustrates low reliability (Portney & Watkins, 2009). Accordingly, the ICCs indicate strong reliability in the sample sizes in each level of SIS (1–6). It also indicated strong interrater reliability of the summed domain social interaction scores. Strong to moderate reliabilities are shown for the social interaction scores classified by park users’ gender, race and ethnicity, and age (Table 6 ).

Table 5.

ICCs of social interaction scores for all park users.

ICC scores by domain
Social Interaction Scale (SIS)
 Score
1 2 3 4 5 6
ICC Scores .92* .81* .83* .84* .91* .92* .78*
Open in a new tab

*p < 0.001.

ICC: intraclass correlation coefficient analysis.

Table 6.

ICCs of social interaction scores classified by different demographics.

Park users' classification
Gender Race and Ethnicity Age Group
ICC Scores M: .81* W: .77* Y: .81*
F: .77* H: .72* A: .68*
B: .68* B: .76* E: .72*
O: .62* A: .73* M: .72*
M: .68*
O: .52*
Open in a new tab

*p < 0.001.

ICC: intraclass correlation coefficient analysis.

Gender: M (Male), F (Female), B (Both male and females), O (Others); Race and ethnicity: W (White), H (Hispanic), B (Black), A (Asian), M (Mixed group), O (Others); Age group: Y (Young), A (Adults), E (Elderly), M (Mixed group).

3.2.2. Parallel forms reliability

In HLM1, when the social interaction score was the DV, the estimate for the park variance was 413.7, and the residual was 7132.1. Thus, the total variance was 413.7 + 7132.1 = 7545.8. The variance partition coefficient (VPC) was 413.7/7545.8 = 0.055, which indicated that 5.5% of the variance of DV could be attributed to the park level variance.

For the HLM2 (social interaction score reclassified on a reduced scale as the CDV1), the estimate for the park variance was 117.7, and the residual is 2159.6. Thus, the total variance was 117.7 + 2159.6 = 2277.3. The VPC was 117.7/2277.3 = 0.052, which indicated that 5.2% of the variance of CDV1 could be attributed to the park level variance.

In Table 7 , the results of the two HLM models displayed very similar patterns. Park features (IV) were significantly correlated with the social interaction indicators (DV and CDV). The effect sizes in the comparative group were smaller than in the HLM1. The correlations between the DVs and the control variables in the regressions both show similar patterns. Also, the effect sizes of the intercepts in HLM1 were often larger than those of HLM2 (Table 7).

Table 7.

Results of the HLM analyses exploring association between IVs and the different versions of social interaction in the HLMs.

 HLMs
Intercept The original HLM1: with DV Comparative: HLM2 with CDV
Level 2
Park quality 76.98** 43.31**
Park size 27.15** 17.32**
Park type P 130.80** 72.23**



Level 1
Week B 98.31** 74.38**
Week C 46.88** 27.41**
Temperature −4.05 −.55
Weather S 48.31* 45.91**
Time B 18.11 9.46
Open in a new tab

**p < 0.001; *p < 0.05.

Park type P: Pocket Park, Week B: Saturday, Week C: Sunday, Weather S: Sunny, Time B: afternoon.

HLMs: Hierarchical Linear Models; DV: Dependent Variables, CDV: comparative dependent variables.

4. Discussion

While most existing protocols measuring PA focuses on people's physical health in communities, this paper describes the development and functionality of SOSIP for systematically assessing social interaction with urban outdoor environment to inform mental and psychological health benefits in urban outdoor environment. Affected by the COVID-pandemic, outdoor social activities have become increasingly valuable for helping urban residents find relief from the pressures from work and daily routines, leading to improved physical and psychological health, thus reducing mortality risks (Corley et al., 2021; Mintz et al., 2021; Zhou et al., 2022). Compared to the other types of urban outdoor environment, the combination of nearby nature benefits found from visiting park and green space has been indicated crucial in promoting social relations (Jennings & Bamkole, 2019; Shanahan et al., 2016) and support people's physical and psychological well-being by providing various ecosystem services (Pinto et al., 2022; Zhou & Parves Rana, 2012). Potential users of the protocol include researchers, practitioners, managers, policymakers, city planners, park designers, and anyone interested in measuring social interaction and managing urban spaces, such as urban parks, that benefit people's mental health. Park designers and managers can use SOSIP to plan for improving the park users' social needs through improved design and management of urban parks to support a range of spaces and activities that support social interaction.

The development of SOSIP, especially the direct observational techniques, was inspired by the SOPARC designed for measuring PA. Both instruments allow observers to objectively assess behaviors in outdoor space through a reliable and feasible process. While most existing studies examine social interaction via survey and questionnaire, the systematic observational method provides an objective measure of social interaction that is currently missing from the literature. Using a protocol for objectively evaluating social interaction makes it easier for users to understand the same measure in different cases and provides quantitative data for scholars to explore relationships between social interaction and the environmental features, and ultimately contribute to a healthy community (Chen et al., 2022). Because of the protocol's low equipment and technology requirements, it is accessible for a broad range of users. By design, users do not need to predict social interaction from various types of activities without an instrument. Instead, users of SOSIP can accurately measure levels of social interaction according to the SIS and comprehensively assess social interaction, in addition to the number of socializing individuals.

As the SOSIP protocol was developed through the theoretical framework of SIS which modified from Parten's coding scheme, the established psychometric properties of Parten's scheme provided SIS validity for assessing levels of social interaction in terms of mental health. The consistency between the two versions of SIS in the parallel forms of reliability tests proved good internal consistency and psychometrics of SOSIP. This study indicates that social interaction measured through SOSIP is an effective method for quantitively studying associations between social interaction and the identified park features which bring positive impacts for human mental and psychological health through encouraging outdoor social activities. SOSIP can be used by scholars interested in exploring people's social interaction behaviors or relationships with environmental indicators, such as park features, size, and other environmental characteristics.

With construction of the content validity of SIS and multiple tests for reliability, SOSIP has been evaluated as a valid and reliable protocol for evaluating social interaction, with the establishment of psychometric properties. This study found strong interrater reliability of SOSIP and at least moderate interrater reliabilities for using SOSIP to assess the social interaction of park users classified by apparent gender, race, and age group. This allows SOSIP to be used as a reliable protocol for assessing social interaction of specific groups, such as children, elderly citizens, and minority groups. Social interaction between races is an important research topic but previous efforts to measure social interaction in parks used SOPARC, which was not designed for that purpose (Hillier et al., 2016). The development of SOSIP addressed the issue and allowed for assessing social interaction of disadvantaged groups for studies where environmental and social justice is a particular consideration.

The application of SOSIP established positive parallel form reliability of SOSIP. The results in the comparative statistics demonstrated a strong consistency between the social interaction scores and the reduced-level social interaction scores. This indicated the consistency between the six levels of SIS with the three-level SIS. In addition, the original HLM and the comparative model both successfully illustrated the significant correlations between park features and social interaction, which also proved SOSIP was reliable in parallel forms and internal consistency.

Comparing SOSIP with the CDV in the regressions proved the protocol to be superior to the reduced forms of social interaction measurement in application. Even though both HLM models detected significant correlations, compared to the comparative measurement of social interaction, SOSIP illustrated a more pronounced correlation between park features and social interaction in the statistical results by providing a larger effect size. That is because the six-level SIS established a more elaborate measure of social interaction than the three-level SIS. Regarding the statement of unnecessary six-level classification of social interaction, the statistical comparisons between HLM1 and HLM2 explained that the model using a finer scale can display a more considerable correlation than the reduced scale by providing larger effect sizes. As the reduced scale also illuminated the significant correlation and showed consistent results with the six-level scale, users of SOSIP may find the three-level scale an acceptable alternative.

Some literature suggests social interaction can be estimated via the amount of time people spend in a place, thereby indicating their engagement in a public open space and the intensity of contact (Carmona et al., 2010; Gehl, 2011). However, SOSIP was designed to gather social interaction through the momentary time-sampling approach. For researchers who intend to collect data on the persistence of activities with SOSIP, we have some suggestions. Because the level of social interaction (1–6) is determined at the beginning of the observation period in SOSIP, any subtle changes in social interaction behavior—while uncommon during the observation period but still inevitable over time—could lead to inaccuracy in the dataset. Regarding this circumstance, we suggest that the researchers either note changes in social interaction behaviors within the 15-min observation period as a limitation or conduct a math equation to weight different levels of interaction based on the distribution of time for each level. For instance, an SOSIP observer finds a group of park users sitting and chatting under the tree (Level 5) for the first 10 min (10/15 * 100% = 67%), and then the group moves to the soccer field and played soccer (Level 6) for the remaining 5 min (5/15 * 100% = 33%). The level for that group would then be calculated as 5 * 67% + 6 * 33% = 5.33. This weighting process is difficult to be implemented in a crowded park, but the momentary time sampling is adaptive for various degree of crowding of a park.

4.1. Limitation and recommendation

This protocol was appropriate for the example in Logan, Utah, because the density of park users and park size were limited because of the local prevalence of neighborhood and community parks. For crowded park settings in larger cities, there may be some larger social activities, such as attending large events. While it is challenging to apply SOSIP in these parks and the social interaction score may not be as accurate as in the parks of Logan, which should be acknowledged as a limitation of the protocol, the protocol can be scaled to larger contexts with technological assistance.

For potential users of SOSIP who are interested in measuring social interaction in crowded settings, we have some recommendations for adaptations both in the format of observation and the SIS that would improve the utility and scalability of the protocol. Instead of systematic observation in person, some video techniques, such as unmanned aerial vehicle observations, may provide more accurate data on park users' activities in very large and crowded parks. The data can be either processed following SOSIP by the researcher watching the videos or coded by using Deep Learning of Computer Version techniques to recognize park users’ social behaviors. Compared to the six-levels of SIS, observers with the protocol are suggested to employ the 3-levels SIS to determine the degree of social interaction, which is adapted for an efficient and accurate judgment in busy settings.

5. Conclusion

This study built on an innovative protocol designed to measure social interaction with urban outdoor environment through SIS and group size. Establishment of the protocol's theoretical validity and multiple reliabilities has indicated that SOSIP is valid and reliable for scholarly studies. The social interaction variables captured through SOSIP can be further analyzed and explored in relation to environmental features and inform researchers, planners, and policy makers to facilitate urban residents' mental health and social cohesion, especially during the pandemic. By enabling researchers to assess levels of social interaction in parks and the factors that influence those interactions, protocols like SOSIP could address larger concerns regarding psychological health, community well-being, and environmental justice.

Author statement

Shuolei Chen: Conceptualization, Methodology, Software, Formal analysis, Investigation, Writing - Original Draft, Visualization. Ole Sleipness: Validation, Resources, Supervision, Writing - Review & Editing, Project administration. Keith Christensen: Validation, Resources, Supervision, Project administration. Bo Yang: Validation, Supervision, Project administration. Hao Wang: Resources, Funding acquisition.

Funding information

The authors would like to acknowledge the funding that they had received from:

  • (1)

    Organization: National College Students Innovation and Entrepreneurship Training Program, Grant numbers: 202210298045Z, 202210298046Z.

  • (2)

    Organization: Jiangsu Office of Philosophy and Social Science, Grant number: 2022SJYB0127.

  • (3)

    Organization: National Key Laboratory Foundation of China, Grant number: 2019YFD1100404.

Handling editor: Giulia D'Aurizio

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jenvp.2023.102008.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Multimedia component 1
mmc1.pdf (101.5KB, pdf)

References

  1. Bakeman R., Brownlee J.R. The strategic use of parallel play: A sequential analysis. Child Development. 1980:873–878. [Google Scholar]
  2. Bakeman R., Gottman J.M. Cambridge University Press; 1997. Observing interaction: An introduction to sequential analysis. [Google Scholar]
  3. Bates D., Maechler M., Bolker B., Walker S. 2013. lme4: Linear mixed-effects models using Eigen and S4. CRAN. [Google Scholar]
  4. Besenyi G.M., Schooley B., Turner-McGrievy G.M., Wilcox S., Wilhelm Stanis S.A., Kaczynski A.T. The Electronic Community Park Audit Tool (eCPAT): Exploring the use of mobile technology for youth empowerment and advocacy for healthy community policy, systems, and environmental change. Frontiers in Public Health. 2018;6:332. doi: 10.3389/fpubh.2018.00332. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bird M.E., Datta G.D., Van Hulst A., Kestens Y., Barnett T.A. A reliability assessment of a direct-observation park evaluation tool: The Parks, activity and recreation among kids (PARK) tool. BMC Public Health. 2015;15(1):906. doi: 10.1186/s12889-015-2209-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bode N.W., Sutton A., Lacey L., Fennell J.G., Leonards U. A method for detecting characteristic patterns in social interactions with an application to handover interactions. Royal Society Open Science. 2017;4(1):160694. doi: 10.1098/rsos.160694. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Campbell L.K., Svendsen E.S., Sonti N.F., Johnson M.L. A social assessment of urban parkland: Analyzing park use and meaning to inform management and resilience planning. Environmental Science & Policy. 2016;62:34–44. [Google Scholar]
  8. Carmona M., Tiesdell S., Heath T., Oc T. 2010. Public places- urban spaces: The dimensions of urban design. MA architectural press. [Google Scholar]
  9. Charko L., Fraser C., Jones D., Umangay U.K. Faculty, candidates, and children at play: Perceptions and dissonances. InPlay: A theory of learning and change. 2016:55–70. (Springer, Cham) [Google Scholar]
  10. Chen S., Christensen K.M., Li S. A comparison of park access with park need for children: A case study in cache county, Utah. Landscape and Urban Planning. 2019;187:119–128. [Google Scholar]
  11. Chen S., Sleipness O., Xu Y., Park K., Christensen K. A systematic review of alternative protocols for evaluating non-spatial dimensions of urban parks. Urban Forestry and Urban Greening. 2020;126718 [Google Scholar]
  12. Chen S., Wu Z., Sleipness O.R., Wang H. Benefits and conflicts: A systematic review of dog park design and management strategies. Animals. 2022;12(17):2251. doi: 10.3390/ani12172251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cohen D.A., Setodji C., Evenson K.R., Ward P., Lapham S., Hillier A., McKenzie T.L. How much observation is enough? Refining the administration of SOPARC. Journal of Physical Activity and Health. 2011;8(8):1117–1123. doi: 10.1123/jpah.8.8.1117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Coplan R.J., Ooi L.L., Rose-Krasnor L. Naturalistic observations of schoolyard social participation: Marker variables for socio-emotional functioning in early adolescence. The Journal of Early Adolescence. 2015;35(5–6):628–650. [Google Scholar]
  15. Corley J., Okely J.A., Taylor A.M., Page D., Welstead M., Skarabela B., Russ T.C. Home garden use during COVID-19: Associations with physical and mental wellbeing in older adults. Journal of Environmental Psychology. 2021;73:101545. doi: 10.1016/j.jenvp.2020.101545. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dadvand P., Hariri S., Abbasi B., Heshmat R., Qorbani M., Motlagh M.E., Kelishadi R. Use of open spaces, self-satisfaction, and social contacts in adolescents: A population-based caspian-V study. Environmental Research. 2019;168:171–177. doi: 10.1016/j.envres.2018.09.033. [DOI] [PubMed] [Google Scholar]
  17. Donohue M.V. Social behavioral goals and reimbursable developmental activity strategies. Occupational Therapy in Mental Health. 2015;31(4):311–327. [Google Scholar]
  18. Gehl J. Island Press; 2011. Life between buildings: Using public space. [Google Scholar]
  19. Geng D.C., Innes J., Wu W., Wang G. Impacts of COVID-19 pandemic on urban park visitation: A global analysis. Journal of Forestry Research. 2021;32(2):553–567. doi: 10.1007/s11676-020-01249-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hillier A., Han B., Eisenman T.S., Evenson K.R., McKenzie T.L., Cohen D.A. Using systematic observations to understand conditions that promote inter-racial experiences in neighborhood parks. Urban Planning. 2016;1(4):51. doi: 10.17645/up.v1i4.756. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hill J.M., Jobling R., Pollet T.V., Nettle D. Social capital across urban neighborhoods: A comparison of self-report and observational data. Evolutionary Behavioral Sciences. 2014;8(2):59. [Google Scholar]
  22. James P., Banay R.F., Hart J.E., Laden F. A review of the health benefits of greenness. Current Epidemiology Reports. 2015;2(2):131–142. doi: 10.1007/s40471-015-0043-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Jennings V., Bamkole O. The relationship between social cohesion and urban open space: An avenue for health promotion. International Journal of Environmental Research and Public Health. 2019;16(3):452. doi: 10.3390/ijerph16030452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Joseph R.P., Maddock J.E. Observational park-based physical activity studies: A systematic review of the literature. Preventive Medicine. 2016;89:257–277. doi: 10.1016/j.ypmed.2016.06.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kaczynski A.T., Stanis S.A.W., Besenyi G.M. Development and testing of a community stakeholder park audit tool. American Journal of Preventive Medicine. 2012;42(3):242–249. doi: 10.1016/j.amepre.2011.10.018. [DOI] [PubMed] [Google Scholar]
  26. Kaźmierczak A. The contribution of local parks to neighbourhood social ties. Landscape and Urban Planning. 2013;109(1):31–44. [Google Scholar]
  27. Kemperman A., Timmermans H. Open spaces in the direct living environment and social contacts of the aging population. Landscape and Urban Planning. 2014;129:44–54. [Google Scholar]
  28. Loebach J., Cox A., Little S. 2020. Behavior mapping to support the development of youth-friendly public outdoor Spaces.The routledge handbook of designing public spaces for young people: Processes, practices and policies for youth inclusion. [Google Scholar]
  29. Maas J., Van Dillen S.M., Verheij R.A., Groenewegen P.P. Social contacts as a possible mechanism behind the relation between open space and health. Health & Place. 2009;15(2):586–595. doi: 10.1016/j.healthplace.2008.09.006. [DOI] [PubMed] [Google Scholar]
  30. Malek N.A., Nashar A. Measuring successfulness of malaysian green open spaces: An assessment tool. Theoretical and Empirical Researches in Urban Management. 2018;13(2):21–37. [Google Scholar]
  31. McKenzie T.L., Cohen D.A., Sehgal A., Williamson S., Golinelli D. System for observing play and recreation in communities (SOPARC): Reliability and feasibility measures. Journal of Physical Activity and Health. 2006;3(s1):S208–S222. doi: 10.1123/jpah.3.s1.s208. [DOI] [PubMed] [Google Scholar]
  32. McKenzie T.L., Van Der Mars H. Top 10 research questions related to assessing physical activity and its contexts using systematic observation. Research Quarterly for Exercise & Sport. 2015;86(1):13–29. doi: 10.1080/02701367.2015.991264. [DOI] [PubMed] [Google Scholar]
  33. Mintz K.K., Ayalon O., Nathan O., Eshet T. See or be? Contact with nature and well-being during COVID-19 lockdown. Journal of Environmental Psychology. 2021;78:101714. doi: 10.1016/j.jenvp.2021.101714. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moulay A., Ujang N., Said I. Legibility of neighborhood parks as a predicator for enhanced social interaction towards social sustainability. Cities. 2017;61:58–64. [Google Scholar]
  35. Parten M.B. Social participation among preschool children. Journal of Abnormal and Social Psychology. 1932;27(pa3):243. [Google Scholar]
  36. Peters K., Elands B., Buijs A. Social interactions in urban parks: Stimulating social cohesion? Urban Forestry and Urban Greening. 2010;9(2):93–100. [Google Scholar]
  37. Pinto L.V., Inácio M., Ferreira C.S.S., Ferreira A.D., Pereira P. Ecosystem services and well-being dimensions related to urban green spaces–A systematic review. Sustainable Cities and Society. 2022;104072 [Google Scholar]
  38. Poortinga W., Calve T., Jones N., Lannon S., Rees T., Rodgers S.E., Johnson R. Neighborhood quality and attachment: Validation of the revised residential environment assessment tool. Environment and Behavior. 2017;49(3):255–282. doi: 10.1177/0013916516634403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Portney L.G., Watkins M.P. Vol. 892. 2009. (Foundations of clinical research: Applications to practice). (Pearson/Prentice Hall) [Google Scholar]
  40. Rasidi M.H., Jamirsah N., Said I. Urban open space design affects urban residents' social interaction. Procedia-Social and Behavioral Sciences. 2012;68:464–480. [Google Scholar]
  41. Salih S., Ismail S. Determining the factors affecting social interaction in the parks of Baghdad city. Iraq.ArchNet-IJAR. 2018;12(3) [Google Scholar]
  42. Schiller B., Tönsing D., Kleinert T., Boehm R., Heinrichs M. Effects of the COVID-19 pandemic nationwide lockdown on mental health, environmental concern, and prejudice against other social groups. Environment and Behavior. 2021;139165211036991 [Google Scholar]
  43. Schmidt T., Kerr J., Schipperijn J. Associations between neighborhood open space features and walking and social interaction in older adults—a mixed methods study. Geriatrics. 2019;4(3):41. doi: 10.3390/geriatrics4030041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Seeland K., Dübendorfer S., Hansmann R. Making friends in Zurich's urban forests and parks: The role of public open space for social inclusion of youths from different cultures. Forest Policy and Economics. 2009;11(1):10–17. [Google Scholar]
  45. Shanahan D.F., Bush R., Gaston K.J., Lin B.B., Dean J., Barber E., et al. Health benefits from nature experiences depend on dose. Scientific Reports. 2016;6:28551. doi: 10.1038/srep28551. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Skjœveland O. Effects of street parks on social interactions among neighbors: A place perspective. Journal of Architectural and Planning Research. 2001:131–147. [Google Scholar]
  47. Smith P.K. A longitudinal study of social participation in preschool children: Solitary and parallel play reexamined. Developmental Psychology. 1978;14(5):517. [Google Scholar]
  48. Tao Y., Yang J., Chai Y. The anatomy of health-supportive neighborhoods: A multilevel analysis of built environment, perceived disorder, social interaction and mental health in beijing. International Journal of Environmental Research and Public Health. 2020;17(1):13. doi: 10.3390/ijerph17010013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Veen E.J., Bock B.B., Van den Berg W., Visser A.J., Wiskerke J.S. Community gardening and social cohesion: Different designs, different motivations. Local Environment. 2016;21(10):1271–1287. [Google Scholar]
  50. Veitch J., Ball K., Flowers E., Deforche B., Timperio A. Children's ratings of park features that encourage park visitation, physical activity and social interaction. Urban Forestry and Urban Greening. 2021;58:126963. [Google Scholar]
  51. Venter Z.S., Barton D.N., Gundersen V., Figari H., Nowell M.S. Back to nature: Norwegians sustain increased recreational use of urban open space months after the COVID-19 outbreak. Landscape and Urban Planning. 2021;214:104175. [Google Scholar]
  52. Walker J.T., Mowen A.J., Hendricks W.W., Kruger J., Morrow J.R., Jr., Bricker K. Physical activity in the park setting (PA-PS) questionnaire: Reliability in a California statewide sample. Journal of Physical Activity and Health. 2009;6(s1):S97–S104. doi: 10.1123/jpah.6.s1.s97. [DOI] [PubMed] [Google Scholar]
  53. Wan C., Shen G.Q., Choi S. Underlying relationships between public urban green spaces and social cohesion: A systematic literature review. City, Culture and Society. 2021;24:100383. [Google Scholar]
  54. Whiting J.W., Larson L.R., Green G.T. Monitoring visitation in Georgia state parks using the System for observing play and recreation in communities (SOPARC) Journal of Park and Recreation Administration. 2012;30(4):21–37. [Google Scholar]
  55. Xu Z. The urban parks in nanjing, 1900s–2000s: A brief introduction. Planning Perspectives. 2021;36(6):1269–1292. [Google Scholar]
  56. Yamada N., Knapp D. Participants' preferences for interpretive programs and social interactions at a Japanese natural park. Visitor Studies. 2010;13(2):206–221. [Google Scholar]
  57. Zhou S., Chen F., Xu Z. Evaluating the accessibility of urban parks and waterfronts through online map services: A case study of shaoxing, China. Urban Forestry and Urban Greening. 2022;77:127731. [Google Scholar]
  58. Zhou X., Parves Rana M. Social benefits of urban open space: A conceptual framework of valuation and accessibility measurements. Management of Environmental. 2012 [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.pdf (101.5KB, pdf)

Articles from Journal of Environmental Psychology are provided here courtesy of Elsevier

RESOURCES