Examining neighborhood effects on residents’ daily activities in central Shanghai,China: Integrating “big data” and “thick data”

Abstract

This research combines “big data” and “thick data” approaches to examine the correlation and causation between residential neighborhood features and people’s daily commuting and traveling patterns by integrating two datasets: household survey data and mobile phone data. We focus on “lilong” neighborhoods—a primary form of traditional residential neighborhood in central Shanghai. The characteristics of lilong neighborhoods are assessed using “thick data” from surveys in 105 lilongs, while residents’ daily activities are mapped out using “big data” from two weeks of mobile phone usage. We match these two datasets at neighborhood level based on their geospatial references. Four multinomial logistic regression models are developed to examine neighborhood effects on lilong residents’ daily activities. Our research confirms the major mechanisms of neighborhood effects and unravels their relative importance in shaping the patterns of residents’ daily activities. Conceptually, this study sheds new light on the understanding of how people’s life quality and wellbeing are affected by neighborhood characteristics through highlighting the importance of social interactions and the access to/quality of public facilities. Methodologically, incorporating household survey data (thick data) and mobile phone data (big data) is proven to be a novel and effective approach for examining neighborhood effects at a relatively large scale.

Keywords

neighborhood effects daily activities thick data big data social interactions public facilities

Introduction

Neighborhood effects are a classical topic in social sciences. A vast volume of studies have been dedicated to investigating how certain features of residential neighborhoods restrict people’s socioeconomic opportunities or mitigate adverse impacts from the workplace or social lives (Galster, 2019; Kesteloot et al., 2006; Lewis, 1966). In addition to classical research on socioeconomic outcomes, increasing research attention has been given to neighborhood effects on residents’ health and wellbeing (Ludwig et al., 2012). As suggested by some recent studies, individuals’ activity space, that is, the space within which individuals perform their daily activities, both inside and outside their residential neighborhoods, is of great significance to understanding the inequalities in people’s life quality and wellbeing (Feng et al., 2020; Perchoux et al., 2016). There is also clear evidence of neighborhood effects on people’s daily activities such as travel and commute (Cao, 2010; Gehrke and Wang, 2020) and yet, we still know little about the extent to which neighborhood effects shape the spatial pattern of people’s daily movements and activities, let alone their causal mechanisms and socioeconomic outcomes.

Geographically, most studies on neighborhood effects focus on the context of the US and European cities, characterized by a relatively low degree of social mobility at both individual and societal levels. By comparison, there is a dearth of research explicitly examining neighborhood effects in the Chinese context, except for some relevant studies on impoverished neighborhoods (Liu and He, 2017; Wu et al., 2010), neighborhood attachment (Liu et al., 2017; Li and Wu, 2013), and accessibility to neighborhood-based amenities and related health issues (Su et al., 2017; Xiao et al., 2017; Zhou et al., 2019). For instance, Xiao et al. (2019) reported that vulnerable groups in Shanghai can get equal access to green space, while Zhou et al. (2019) identified significant inequalities in neighborhood visual walkability based on deep learning technologies. Notably, studies in the Chinese context concerning neighborhood effects on daily activities tend to have a narrow focus, for example, neighborhood effects are merely interpreted as one of the causes of physical and psychological health outcomes (Wen et al., 2010). With occasional exceptions (for example, (Cheng and Wang, 2013)’s study on neighborhood effects on the access to jobs), studies of neighborhood effects on daily activities mostly concern older adults. For instance, (Jia et al., 2018) documented that neighborhood characteristics influence middle-aged and older adults’ daily activities and consequently affect their risk of cardiovascular diseases. Also, older adults’ mental health can be improved by physical activities, social interactions within their neighborhood, and volunteering activities (Wang et al., 2019). Other age aspects and neighborhood effects were largely overlooked. This is partly owing to the fast-changing socioeconomic circumstances at all scales that may render neighborhood effects difficult to be detected in relation to individual characteristics and mobilities at micro level and drastic socioeconomic changes at macro level. For instance, a study on impoverished neighborhoods in Chinese cities by Wu et al. (2010) reported a small but significant neighborhood effect on poverty generation in urban China, which was largely linked to the path dependency of social inequalities derived from institutional discrimination under the centrally planned economy. In the post-reform era, Chinese cities have undergone substantial changes in various aspects, during which, market forces have reinforced social inequalities and further widened the rift among neighborhoods inherited from the centrally planned economy. With the deepening of market-oriented reforms, neighborhoods have been endowed with further self-governing power under decentralization and marketization (He, 2015), adding another dimension of socio-spatial inequalities given the varying capacity of neighborhood governance and residents’ varying degrees of dependency to neighborhoods. Considering the rising importance of neighborhood effects on contemporary urban transformation in China, neighborhood effects are insufficiently explored. It is therefore timely and appropriate to expand the discussion on neighborhood effects to a wider population and their daily activities in the Chinese context.

In this research, we aim to fill this void by focusing on the neighborhood effect in relation to residents’ daily traveling patterns in China’s largest city, Shanghai. Renowned for its highly developed market economy and rich history in urban development, Shanghai epitomizes the rapid changes and diverse development paths seen in urban neighborhoods, making it an ideal case to understand neighborhood effects on residents’ daily activities. We are particularly interested in the following questions: What are the neighborhood effects on people’s daily traveling patterns? More specifically, do neighborhood characteristics affect people’s traveling patterns and do people’s perceptions about the physical and social aspects of their neighborhoods affect their daily activities? These questions remain largely unexplored, partially owing to the lack of empirical data. To overcome the problem of data unavailability, we combine conventional household survey data (thick data) with a novel data source: mobile phone data (big data) to delve into these important theoretical and empirical questions.

The remainder of the paper starts with a comprehensive overview of the studies of neighborhood effects to identify the research gap and position our research within the international context. We then introduce the data source and methodologies, followed by the empirical analysis. The paper then concludes with reflections on conceptual and empirical debates on neighborhood effects and discusses the limitations and future directions of this study.

Revisiting neighborhood effects

Theorizing neighborhood effects

The inception of neighborhood effects can be traced to Lewis (1966)’s seminal work on the culture of poverty, in which poor neighborhoods—with a high proportion of disadvantaged groups—were believed to have a negative impact on residents. It is also summarized as “a conceptual model showing the relationship between neighborhood context, individual residents’ attitudes, behaviors and attributes, and opportunities for social advancement” (Galster 2019: 171). There are several sets of theoretical bases for understanding neighborhood effects, for example, social-interactive, environmental, geographical, and institutional mechanisms, which are often employed to interpret individual’s mental health, socioeconomic outcomes, employment opportunities as well as daily mobility (Galster, 2012).

In exploring the social-interactive mechanisms, scholars put forward different conceptualizations of neighborhood effect, for example, focusing on the socialization process and social network, which emphasize the capabilities and personal qualities of residents in the neighborhood. Musterd and Andersson (2006) state that the concentration of “unwanted social ties and role models” can affect the rest of the neighborhood, which might adversely impact on neighborhood opportunities and future development of the society. Given that the extent, type, and quality of social connections may encourage or inhibit the opportunities to discover new ways of life and thus affect the life chances of residents, the benefits of socially diverse neighborhoods, in their more diversified social networks and opportunities for residents are commonly recognized (Andersson, 2001; Ellen & Turner, 1997).

Environmental mechanisms are often employed to study the neighborhood effects on mental and physical health. Relevant studies suggest that exposure to violence, the nature of the physical surroundings, and toxic exposure may trigger strong psychological and physical responses that impair residents’ health and wellbeing (Galster, 2012; Ludwig et al., 2012; Liu et al., 2019). For instance, Ludwig et al. (2012) found that moving to a less-distressed neighborhood environment contributes to long-term improvements in physical and mental health and subjective wellbeing in adults. It was also reported that prolonged exposure to a deteriorated environment weakens residents’ sense of efficacy (CatherineRoss and Pribesh, 2001). So far, however, the mechanisms of how environmental factors affect residents’ behavior, especially daily activities, have not been explicitly explored.

For studies concerning geographical mechanisms, the conceptualization of neighborhood effect is based on the differences in opportunities of access to and the quality of services and public facilities such as the quality of schools, healthcare services and public services. The lack of access to quality public services is argued to produce negative impacts on social opportunity (Kain, 2004). Individuals deprived of sufficient quality services in the neighborhood may experience difficulties in accessing jobs and social opportunities. While institutional mechanisms (e.g., stigmatization, local institutional resources, local market actors) offer important theoretical insights for understanding neighborhood effects (Galster, 2012), they are either less relevant to or lack direct evidence in this research, hence are not further discussed here.

Understanding neighborhood effects on residents’ social lives

Scholars have revealed that neighborhood effects can lead to prolonged poverty at neighborhood scale as well as social exclusion (Kesteloot et al., 2006; Friedrichs et al., 2003). These empirical studies have shown that the concentration of particular socioeconomic groups, for example, the urban poor and ethnic minorities, can generate neighborhood effects and bring behavioral, attitudinal, and psychological impacts. In this regard, scholars have detected specific neighborhood effects on residents’ daily life, affecting education, employment, and health outcomes. These outcomes generated by the neighborhood effect are spatially specific and location-based, which further impacts the population distribution and composition at both neighborhood and city scales. A direct linkage between the area’s reputation and opportunities of employment has been reported (Simpson et al., 2006). A lack of connections and social capital tends to further isolate socially excluded groups and form a vicious cycle of unemployment in that particular neighborhood. By contrast, neighborhood effects are moderated by a high level of social cohesion and attachment (Liu et al., 2017).

Empirical studies on neighborhood effects on daily activities and travel behavior can be decomposed into the effects of physical environments, sense of community, social cohesion and control, relative deprivation, social network, cultural norms and values. Empirical evidence indicates that people’s physical activity and travel behavior are associated with several characteristics of the physical neighborhood such as residential density, street connectivity, land use patterns, local accessibility, and pedestrian-friendly features (Aditjandra et al., 2012; Zhong et al., 2020). While the physical planning dimension has been spotlighted in the literature, other dimensions have received inadequate academic attention. A few exceptions include a study on the role of neighborhood attachment in walking intentions and behavior revealed by a web-based survey conducted in Malmö, Sweden (Ferreira et al., 2016); a positive association between neighborhood social cohesion and the frequency of physical activity in general (Cradock et al., 2009), as well as active travel (Clark and Scott, 2013). While the effects of neighborhood social networks on activity-travel behavior have been well-documented (Kim et al., 2018), the effects of cultural norms and values are inadequately understood. Thus far, our understanding of neighborhood effects on people’s daily activities remains quite limited. With relevant data becoming available, a detailed investigation into neighborhood characteristics and people’s daily activities will advance the studies of neighborhood effects from behavioral and spatial aspects, and extend the explanatory power of existing theories and models of neighborhood effects.

Measuring neighborhood effects

Based on existing theories, continuous efforts have been made to empirically testify and measure neighborhood effects (Andersson, 2001; Friedrichs et al., 2003; Musterd and Andersson, 2006). This research can be further classified into two types: in-depth qualitative studies and large-scale longitudinal research. Both approaches try to explain the existence of neighborhood effects and measure their magnitude, with divergent theoretical and empirical foci. Ideally, both detailed and extensive datasets are required for investigating the existence and magnitude of neighborhood effects. According to Musterd and Andersson (2006), comprehensive research on neighborhood effects necessitates multi-level information at neighborhood and individual scales as well as household attributes. Moreover, the dependent variable should be quantifiable for easy measurement and comparison.

Given the insufficient exploration of neighborhood effects on daily activities, it remains unclear whether existing theories and analytical models can fully explain the patterns of daily commuting and other activities for residents living in different neighborhoods. The main purpose of this research is therefore to combine different data sources and analytical methods in neighborhood research to provide a better understanding of neighborhood effects on residents’ daily activities. To overcome the problems of data unavailability and the constraints of both thick data (e.g. data collected using qualitative methods or self-administrated survey data) and big data (e.g. social media data and mobile phone data), we design an analytical framework that combines both approaches, building upon existing theories of neighborhood effects. The complementary nature of big data and thick data requires deeper integration to interpret the observed behavior, which still remains at the exploratory stage thus far.

Methodologies and data collection

As stated above, this study tries to synergize the approaches of “thick data” and “big data” by integrating two datasets: household survey data and mobile phone data. The housing survey data was conducted in six central Shanghai districts in 2016 (see Figure 1). A total of 1159 individuals from 105 lilong (narrow-alley) neighborhoods were chosen using probability-proportional-to-size sampling. We did not include other types of residential forms such as modern apartment housing or work unit compounds, mainly because the lilong is a very important traditional housing style in Chinese cities (similar to “hutong” in Beijing), and remains a common form of residential neighborhood in Shanghai that can be traced back to the 1870s. Lilongs were the primary residential form in the inner city of Shanghai for more than a century (Zhao, 2004). They are considered to be the first commodity housing type in Chinese dwelling history and also offer a microcosm of the culture of “modern Shanghai” (Arkaraprasertkul, 2009; Zhao, 2004). Lilong can be understood as “neighborhood lanes,” characterized by the efficient “fish-bone” layout to achieve the highest density from minimal urban public space due to their residential form. The small semi-public or collective spaces provided at the neighborhood level enables a “neighborhood watch,” generating a strong sense of security and sense of community (Zhao, 2004; Arkaraprasertkul, 2009). By the end of 2016, this type of housing still took up a total of 14.12 million m² construction area in central Shanghai. Most existing studies on lilong are from the perspectives of architectural design or historical conservation (Arkaraprasertkul, 2009; Zhao, 2004), while seeing lilong through the lens of neighborhood studies remains insufficiently explored. With a specific focus on lilong neighborhoods that are relatively homogeneous in the built environment, this study will be able to unravel the social-interactive mechanism and geographical mechanism of neighborhood effects on daily activities. Both neighborhood- and individual-level data were collected, including location, socioeconomic and demographic characteristics, housing and community attributes at neighborhood level, as well as social support, neighborhood satisfaction, and sense of community at individual level.

Figure 1.

Spatial distribution of the surveyed lilong neighborhoods in Shanghai.

Traditional studies on human movements and activities have mainly relied on two major sources: surveys of individuals in the form of travel diaries (Jones and Pebley, 2014; Zhong et al., 2017), and GPS tracking data (Shen et al., 2013; Marra et al., 2019). It is widely accepted that the mobile phone data may provide new opportunities to move beyond traditional human activity measurements (Xu, 2021), since it offers extensive geo-coded trajectories of population movement across place and time, which may be utilized to understand people’s everyday mobility and activities (Gonzalez et al., 2008; Gong et al., 2020; Xiao et al., 2019; Xu, 2021; Zhou et al., 2018). In contrast to conventional surveys and census data, mobile phone data provides information produced by a large volume of spatiotemporal traces of individual users and is therefore appropriate for assessing trip movements (Gong et al., 2020; Xiao et al., 2019). It can also help overcome the constraints of traditional home surveys, such as sampling bias or a limited sample size, to discover a broader pattern of inhabitants’ daily activities at a bigger scale (Gong et al., 2020; Xu, 2021). The mobile phone data used in this article were gathered from the Shanghai China Mobile Company between March 15th and March 28th, 2014, including the information of mobile station identifier (MSID), signal occurrence time, longitude and latitude coordinates of connecting base station. Since the mobile phone data records the contact between the mobile phone’s signal and the mobile communication base station, the premise for recognizing people’s everyday mobility is that mobile phone signal placement is discretely dispersed in the base station identification region to track the flow and location of human activities (More details are provided in Supp Figure 1). During these two weeks, 6–8 million phone signal records per day were collected.

Assessing neighborhoods characteristics using surveyed data

We first assess the diverse characteristics of 105 surveyed lilong neighborhoods. Neighborhood social and demographic characteristics were measured by household numbers, average monthly individual income, unemployment rate and the proportion of local “hukou” (household registration) holders, ratio of people over 60, ratio of minimum living standard program (MLSP¹) recipients, and proportion of tenants. We also accounted for housing and community physical attributes in terms of lilong types and neighborhood environment, which include access control system, property management and green space. Notably, during the socialist revolution in the 1940s–1950s, most lilong housing has been transformed into shared ownership or tenancy. All variables above are treated as dummy variables. The location of the lilong is measured by several continuous variables including the distance to subway station and community shopping facilities. Besides neighborhood-level information, residents from each lilong were asked about their social support, neighborhood satisfaction, and sense of community. The three variables were measured by a few indicators with 5-point Likert scale based on the degree of satisfaction/agreement. The mean value of these indicators was calculated for assessing these social interaction variables. In order to match the data with mobile phone data, which can only be identified at neighborhood level, we further calculated the average value of each indicator to represent neighborhood level characteristics.

Measuring daily activities using mobile phone data

Following previous frameworks for examining daily activities (Ahas and Mark, 2005; Xiao et al., 2021), we employ the time threshold method, also known as the social position approach, to detect individuals’ activity behavior in this research. We obtained individual users’ position every 60 min based on the records of the base station from which we could identify people’s residential location, working location, and activity location by setting several time thresholds. Specifically, we identified the locations where users stayed between 00:00 and 06:00 for at least five days within the two weeks as their residential locations. The locations where users stayed between 09:00 and 17:30 for at least five workdays within the two weeks were detected as working locations, and activity frequency is defined as number of changes in identified locations. As people’ daily mobility and physical activities have time-of-week differences (Fairclough et al., 2015), we calculated four indices to measure daily activities, namely weekday cumulative distance and frequency of daily activities (excluding commuting behavior), and weekend cumulative distance and frequency of daily activities. The calculation equation is as follows

R_{p} = M a x (\sum_{i = 1}^{\infty} T_{i R 1}, \sum_{i = 1}^{\infty} T_{i R 2}, \sum_{i = 1}^{\infty} T_{i R 3}, \dots \dots, \sum_{i = 1}^{\infty} T_{i R n})

where R_p is the destination locations including residential location, working location and activity location, R₁, R₂⋯⋯, Rn is the mobile phone user’s base station at the specific time period, T_iR1, T_iR2, ⋯⋯, T_iRn is the duration of R₁, R₂⋯⋯, Rn that the mobile phone user stops at the stay location.

Matching mobile phone data with the survey data

Based on the spatial links between neighborhood borders and the position of base stations, we merged these two datasets at the neighborhood level. We identify the spatial patterns of residents’ daily activities using mobile phone data, which are measured by four indicators: weekday activities distance and frequency, weekend activities distance and frequency. All mobile phone data are standardized and categorized using K-means cluster method in STATA 16.0.

With the clusters, calculated by K-means, being the dependent variables, we then run four multinomial logistic models that take into account neighborhoods’ socioeconomic and demographic characteristics, the built environment, and locational variables (Wang & Zhou, 2017). We also explore the impacts of social variables on residents’ daily activities pattern: levels of social support, neighborhood satisfaction, and sense of community. These models examine the causal mechanisms of neighborhood effect and their relative importance under three broad categories: social-interactive mechanisms (social support, neighborhood satisfaction and sense of community), geographical mechanisms (distance to public transportation and commercial center), environmental mechanisms (lilong type, gatedness/ease of public access, green space), alongside demographic and socioeconomic characteristics such as concentration of older adults, concentration of low-income and unemployed population, income, “hukou,” and tenure structure.

Measuring neighborhood effects on daily activities in Shanghai

Table 1 shows the characteristics of four clustered categories of daily activities. The four clustered categories of weekday activities, which represent different daily activity patterns recorded during weekdays, reflect individuals’ uneven access to different socialization opportunities and economic activities. Another four clustered categories reflect the overall (weekend and weekday) activity patterns. The difference between activity patterns from weekdays and weekends indicates the diverse life opportunities and leisure habits among people from various types of neighborhoods, which contributes to a better understanding of neighborhood effects on residents’ daily activities.

Table 1.

Clustering analysis of distance and frequency for daily activities.

Results of clustering analysis for daily activities (weekday)
Variable	Obs.	Mean	Std. Dev.
Frequent long-distance activities
Distance (m)	4,453	49,837.94	19,927.31
Frequency (time)	4453	3.891534	2.045342
Less-frequent medium- to short-distance activities
Distance (m)	43,506	6735.124	6617.271
Frequency (time)	43,506	2.28357	0.4987191
Frequent medium- to long-distance activities
Distance (m)	11,308	11,125.19	7862.369
Frequency (time)	11,308	4.892289	1.373375
Less-frequent short-distance activities
Distance (m)	51,098	26,98.677	3216.585
Frequency (time)	51,098	1	0
Results of clustering for daily activities (weekday and weekend)

Variable	Obs.	Mean	Std. Dev.
More activities on weekends
Distance (weekdays) (m)	20,241	5017.817	5068.394
Frequency (weekdays) (time)	20,241	2.213428	1.256092
Distance (weekend) (m)	20,241	9069.669	7223.717
Frequency (weekend) (time)	20,241	3.60916	1.1623
More activities on weekdays
Distance (weekdays) (m)	10,378	30,213.57	21,119.78
Frequency (weekdays) (time)	10,378	4.56639	1.984588
Distance (weekend) (m)	10,378	8198.432	8805.395
Frequency (weekend) (time)	10,378	2.208229	1.366917
More activities on both weekends and weekdays
Distance (weekdays) (m)	4,377	12,440.38	14,740.05
Frequency (weekdays) (time)	4377	2.418095	1.633801
Distance (weekend) (m)	4,377	49,683.5	19,689.72
Frequency (weekend) (time)	4377	3.543294	1.952945
Fewer activities on both weekends and weekdays
Distance (weekdays) (m)	75,369	4100.806	4719.544
Frequency (weekdays) (time)	75,369	1.596439	0.8070982
Distance (weekend) (m)	75,369	3729.657	4635.541
Frequency (weekend) (time)	75,369	1.336279	0.4724387

With the above clustered activity patterns as dependent variables, two multinomial regression models are developed to examine the neighborhood effects on residents’ employment and socialization opportunities. Table 2 presents the summary statistics of all variables included in the models, including socioeconomic and demographic characteristics, neighborhood built environment, locational variables, and neighborhood sentiment. Lilong neighborhoods appear to have a high concentration of elderly (34% aged above 60) and mid-to low-income population, as evidenced by the relatively high proportion of tenants (43%), MLSP recipients (4%), unemployment (3%), and a small number of neighborhoods with average monthly income above 4500 RMB. In terms of neighborhood built environment, less than one third (27.8%) have green space, while more than half (59.7%) belong to dilapidated old-style “Shikumen lilong,” and only 8.1% are well-decorated detached villas. Although more than 90% are managed by a property management company (PMC), only 22.5% have access control. Since most lilongs are centrally located, the average distances to nearest metro station and commercial center are 1.14 km and 1.49 km, respectively. The average score for social support is 2.18, and the average scores for social satisfaction and sense of community are 3.38 and 3.74, respectively, with 5 being the highest level of satisfaction.

Table 2.

Descriptive statistics of the surveyed neighborhoods.

Variables	Mean	SD
Household size	239.270	390.968
Ratio of local hukou holders (%)	54.0%	0.237
Ratio of residents above 60 (%)	34.0%	0.165
Ratio of MLSP recipients	4.0%	0.058
Unemployment rate	3.0%	0.039
Tenants proportion	43.0%	0.199
Individual monthly income (RMB)
1,500–3,000	34.6%	0.476
3,001–4,500	60.6%	0.489
4,501–6,000	2.5%	0.155
6,001–7,500	2.3%	0.149
Type of lilong
Old-style shikumen lilong	59.7%	0.491
New-style shikumen lilong built before 1949	9.4%	0.292
New-style lilong built before 1949	22.8%	0.419
Detached villas built before 1949	8.1%	0.273
Access control system	22.5%	0.418
PMC service	93.6%	0.245
Green space	27.8%	0.448
Distance to metro station (m)	1.140	0.438
Distance to commercial facilities (m)	1.490	0.937
Social support	2.180	0.417
Neighborhood satisfaction	3.380	0.319
Sense of community	3.740	0.284

Tables 3 and 4 show the results of the multinomial logistic regression models of neighborhood effects on daily activities, which extracts the correlations between neighborhood characteristics and individuals’ daily activities. The first observation is that a neighborhood’s socioeconomic characteristics exhibit a strong impact on residents’ daily activities. Results of the regression models indicate that individuals living in neighborhoods with a higher concentration of MLSP recipients tend to have fewer activities during weekends, indicating a lack of weekend leisure activities for these disadvantageous neighborhoods. The propensity for having frequent long-distance activities on weekdays is relatively high for residents with an average household income within the 4501–6000 RMB per month category. Interestingly, residents living in neighborhoods with an average income of 3001–4500 RMB tend to have fewer activities on both weekdays and weekends compared with the lowest income group. Generally, concentration of poorer population in neighborhoods is associated with negative outcomes (Andersson and Malmberg, 2018)—in this case, this manifests as fewer economic activities on weekdays and fewer leisure activities at the weekend.

Table 3.

Multinomial logistic regression models of neighborhood effects on weekday activities.

	Model 1			Model 2
	Frequent long-distance activities	Less-frequent medium- to short-distance activities	Frequent medium- to long-distance activities	Frequent long-distance activities	Less-frequent medium- to short-distance activities	Frequent medium- to long-distance activities
	Coef.	Coef.	Coef.	Coef.	Coef.	Coef.
Household size	1.47E-05	−9.81E-07	−4.54E-05	1.22E-05	−3.89E-07	−7.65E-05
Ratio of local hukou holders (%)	0.302	0.154*	0.111	0.38	0.146	0.146
Ratio of residents above 60 (%)	−0.175	0.0859	0.107	−0.0529	0.0815	8.58E-05
Ratio of MLSP recipients	−0.432	−0.494*	−0.552	−0.732	−0.536*	−0.63
Unemployment rate	1.398	−0.554	−1.288*	0.936	−0.678	−1.437*
Tenants proportion	0.116	0.287***	0.580***	0.285	0.303***	0.551***
Individual monthly income (RMB) (ref = 1,500–3,000)
3,001–4,500	−0.0225	−0.0426	−0.0617	−0.0411	−0.0425	−0.0386
4,501–6,000	0.209	−0.00872	−0.0747	0.362*	0.0286	−0.0345
6,001–7,500	0.0447	0.0665	0.228	−0.113	0.0672	0.303
Type of lilong (ref = old-style shikumen lilong)
New-style shikumen lilong built before 1949	0.0351	0.0181	0.0261	0.0832	0.0349	0.0225
New-style lilong built before 1949	−0.0356	−0.0730*	−0.0421	0.128	−0.0507	−0.0505
Detached villas built before 1949	0.028	0.0285	0.144	0.211	0.0319	0.0613
Access control system	−0.00769	0.0127	0.0716	−0.0503	0.00305	0.0453
PMC service	0.0231	−0.124*	−0.186*	−0.177	−0.146*	−0.173
Green space	−0.000267	−0.0409	−0.137**	−0.0467	−0.0447	−0.139**
Distance to metro station (m)	0.0813	0.0523	0.112**	0.0821	0.0625*	0.126**
Distance to commercial facilities (m)	0.0666*	−0.00447	−0.0165	0.0462	−0.00831	−0.022
Social support				0.0737	−0.0176	−0.0166
Neighborhood satisfaction				−0.524***	−0.0793	0.0903
Sense of community				0.465**	0.113	0.0768
				−3.147	−1.87	−0.819
Constant	−2.788***	−0.228**	−1.602***	−2.858***	−0.331	−2.134***

***p < .001 **p < .01 *p < .05.

Table 4.

Multinomial logistic regression model of neighborhood effects on daily activities.

	Model 3			Model 4
	More activities on weekends	More activities on weekdays	More activities on both weekends and weekdays	More activities on weekends	More activities on weekdays	More activities on both weekends and weekdays
	Coef.	Coef.	Coef.	Coef.	Coef.	Coef.
Household size	−1.40E-05	4.06E-05	4.60E-05	−1.42E-05	1.47E-05	7.81E-05
Ratio of local hukou holders (%)	−0.0259	0.268*	−0.0203	−0.0502	0.358**	−0.0294
Ratio of aging residents (%)	0.210*	−0.0965	0.00286	0.179	−0.058	0.131
Ratio of MLSP recipients	−1.060***	−0.167	0.0472	−1.137***	−0.38	0.0128
Unemployment rate	−1.835***	−0.788	1.189	−2.069***	−1.034	0.989
Tenants proportion	0.261**	0.389**	−0.188	0.282**	0.472***	−0.0669
Individual income (RMB)(ref = 1,500–3,000)
3,001–4,500	−0.0136	−0.0967*	−0.115	−0.011	−0.0970*	−0.140*
4,501–6,000	0.0665	0.111	0.152	0.148	0.187	0.204
6,001–7,500	0.223	0.0641	−0.267	0.236	0.00163	−0.37
Type of lilong (ref = old-style shikumen lilong)
New-style shikumen lilong built before 1949	−0.0439	0.0931	0.0457	−0.0086	0.0953	0.0926
New-style lilong built before 1949	−0.162***	0.0697	0.0771	−0.122**	0.148*	0.171
detached villas built before 1949	0.00748	0.175	0.165	−0.0026	0.246*	0.302*
Access control system	0.0278	−0.0409	−0.197**	0.00747	−0.0735	−0.200*
PMC service	0.0383	−0.170*	0.114	0.00131	−0.277**	0.00345
Green space	0.0189	−0.0704	0.0605	0.0129	−0.101	0.0455
Distance to metro station (m)	0.143***	0.147***	0.216***	0.166***	0.137**	0.212**
Distance to commercial facilities (m)	−0.00715	0.0647***	−0.0247	−0.0148	0.0542**	−0.0269
Social support				−0.0492	0.0767	0.0211
Neighborhood satisfaction				−0.143*	−0.201*	−0.338**
Sense of community				0.240***	0.205*	0.204
Constant	−1.493***	−3.071***	−1.764***	−2.257***	−2.470***	−2.733***

***p < .001 **, p < .01 *p < .05.

As for demographic characteristics, it is not surprising that neighborhoods with a higher ratio of older adults are likely to have more activities during weekends. Also, the results indicate that residents living in neighborhoods with a higher rate of unemployment are less likely to have medium-to long-distance activities on weekdays and fewer activities on weekends. This resembles the findings associated with lower income groups. In addition, the results show that tenants are also likely to have medium-to short-/medium-to long-distance trips on weekdays, they also tend to have more activities on both weekends and weekdays. This suggests that tenants are likely to engage in various kinds of daily activities and tend to endure longer traveling distances on weekdays.

These regression models also take the built environment into account, in terms of neighborhood type, gating facilities, service of PMC and green space. Neighborhood type is found to significantly impact individuals’ daily activities. Specifically, individuals living in new-style lilongs, in comparison with those living in old-style Shikumen lilongs, tend to have fewer medium-to short-distance activities, and are more likely to have fewer activities on weekends, but have more activities on weekdays. This suggests that old-style Shikumen lilong residents are slightly better off, in terms of travel distance and leisure activities. Meanwhile, residents in detached villa lilongs, which are occupied by a relatively high proportion of middle-class households, are actively engaging in both weekday and weekend activities, indicating that the residents have better access to economic activities and socialization opportunities (Andersson & Malmberg, 2018). Moreover, people living in gated lilong neighborhoods are less likely to have frequent daily activities.

Echoing findings that neighborhood effect influences individual behavior both directly or indirectly through public and private facilities (Friedrichs et al., 2003), in this model, public services and facilities are significantly correlated with residents’ daily activities. Specifically, residents in neighborhoods with PMC service have less-frequent weekday activities and fewer shorter-distance weekday activities. Moreover, individuals living in neighborhoods with green spaces tend to have fewer medium-to long-distance activities on weekdays, which echoes existing findings on the positive association between neighborhood-based physical/leisure activities and green space (Liu et al., 2020). Additionally, the distance to public facilities does have salient effects and is significant in shaping the patterns of individuals’ daily activities—residents tend to have more daily activities as the distance to the nearest metro station increases. Not surprisingly, residents living far from the metro station are more likely to present a medium-to long-distance travel pattern. Similarly, individuals tend to have more weekday activities and longer-distance weekday activities as the distance to the nearest commercial center increases. These findings echo existing findings on the association between the distance to public facilities and residents’ employment status, and the negative impact on social opportunities induced by the lack of access to quality public services (Kain, 2004).

In both Table 3 and Table 4, a second set of models (Model 2 and Model 4) is further developed to examine the social-interactive mechanism in the overall neighborhood effect, by adding three independent variables: social support, neighborhood satisfaction, and sense of community. The results show that both neighborhood satisfaction and attachment have significant impact on residents’ daily activities patterns, while the effect of social support on individuals’ daily activities is not statistically significant. Individuals living in neighborhoods with higher satisfaction tend to have fewer long-distance activities on weekdays, and are less likely to have activities outside the neighborhood at all times. This suggests that, physically and emotionally, neighborhoods with higher level of satisfaction and a pleasant environment can retain their residents—they spend more time within the neighborhood instead of going elsewhere. However, individuals with a higher level of perceived sense of community tend to have more long-distance activities on weekdays and they are likely to have more daily activities both on weekdays and weekends, indicating that they are more likely to be economically active and engage in more social activities with better neighborhood mutual assistance. Conforming to existing literature (Andersson, 2001), these findings suggest that the extent, type, and quality of social interactions within the neighborhood may enable or inhibit residents’ life chances.

Conclusions

Fusing mobile phone big data and questionnaire survey thick data, this study taps into an insufficiently explored area of neighborhood effects studies, namely how neighborhood characteristics affect residents’ daily activities patterns. In relation to the major concepts of neighborhoods effects, especially the social-interactive, geographical and environmental mechanisms, our major findings are three-fold. First, social network and social interaction in the neighborhood play an active role in shaping the pattern of individuals’ daily activities. Residents living in neighborhoods with strong perceived attachment are more likely to enhance their employment opportunities through information exchange and mutual help, which is reflected in the increased frequency and distance of their weekday activities. Nevertheless, we are no able to conclude whether role models and social norms in the neighborhood shape individuals’ daily activities owing to the limitations of the available data.

Second, opportunities of access to and the quality of services and public facilities also affect residents’ daily activities. The competition over scarce resources and facilities is significantly associated with locational variables. For instance, the distances traveled to weekday and weekend activities are both significantly associated with the access to metro stations. This also reflects the classic problem of “spatial (job–housing) mismatch,” which typifies cities with concentrated worksites (Gobillon et al., 2007; Zhou et al., 2018). No doubt, job–housing mismatch has become a very common issue in the process of rapid expansion of the Chinese metropolis. Results of the regression models have shown that residents of neighborhoods far from public facilities or commercial centers travel further for their daily activities, which indicates their restricted opportunities nearby for employment centers and leisure activities.

Thirdly, environmental factors such as an over-crowded built environment, the lack of communal space and underdeveloped infrastructure and public facilities also renders certain lilongs undesirable. In this respect, findings from Shanghai’s lilong neighborhoods offer counterevidence to Permentier et al. (2008)’s argument by highlighting the significant role played by physical and functional neighborhood characteristics in affecting residents’ daily activities and life chances. Although this research does not focus on how environmental characteristics affect residents’ health and wellbeing, our findings can be considered a useful attempt to extend the application of environmental mechanisms to understand neighborhood effects on people’s daily activities. A neighborhood’s socioeconomic characteristics are proven to be closely correlated with the life chances of its residents and/or their access to various public/private resources (Galster, 2012). As our research shows, residents living in neighborhoods with poorer socioeconomic status—indicated by higher percentage of MLSP recipients, relatively higher unemployment rate, lower average household monthly income, and relatively undesirable built environment—tend to have less-frequent leisure activities during the weekends.

Our research largely confirms the major mechanisms through which neighborhood effects shape the patterns of residents’ daily activities, particularly the social-interactive and geographical mechanisms, as well as environmental mechanisms to some extent. Social interactions and access to public facilities are proven to be important in understanding the neighborhood effects on residents’ daily activities in Shanghai’s lilong neighborhoods. Besides the well-understood impact of socioeconomic status on residents’ daily activities, we provide new evidence for the causal mechanism of neighborhood effects on residents’ daily activities by highlighting the significant role of physical and functional neighborhood characteristics (environmental factors). This, again, confirms that the causal mechanisms of neighborhood effects vary from context to context and therefore necessitate constant and collective efforts to unveil their as yet not fully understood “mystery” (Bauder, 2002). In this respect, this paper contributes to the literature on neighborhood effects in two ways. Conceptually, this study on residents’ commuting and other daily activity patterns sheds new light on the understanding of how people’s life quality and wellbeing are affected by neighborhood characteristics by highlighting the importance of social interactions and access to/quality of public facilities, as well as physical and functional neighborhood characteristics. Methodologically, this study contributes by incorporating mobile phone data (big data) and household survey data (thick data) to link the general pattern of residents’ daily activities with social-interactive, accessibility and environmental factors of a neighborhood, which together provide a vivid picture of residents’ travel behavior. This attempt is proven to be a novel and effective approach to examining neighborhood effects at a relatively large scale—105 neighborhoods in central Shanghai in this case.

Nevertheless, several limitations in this research should be acknowledged here. First of all, the mobile phone data and survey data are not collected in the same year, thus there may be potential inconsistency between the two datasets, although the demographic data of central Shanghai in 2014 and 2016 are quite similar. Moreover, we were not able to link mobile phone and survey data at individual level due to privacy concerns, which affects the accuracy of the analysis. We only focus on lilong housing; hence the dataset might contain some self-selection bias. Also, it was not possible to detect the contagiousness of behavior within the neighborhood in this research as we did not conduct in-depth interviews in all neighborhoods. This research would benefit from a bigger pool of data from diversified sources across different time periods to fully explore a range of causal mechanisms of neighborhood effects, particularly the institutional mechanisms that take structural/political economic factors into account (Wu et al., 2010) in addition to existing findings on socialization, accessibility of public services, and neighborhoods’ socioeconomic, physical, and functional characteristics. In general, we believe this study makes a worthwhile contribution to an improved understanding of neighborhood effects, both methodologically and conceptually.

Supplemental Material

sj-pdf-1-epb-10.1177_23998083221078307 – Supplemental Material for Examining neighborhood effects on residents’ daily activities in central Shanghai, China: Integrating “big data” and “thick data”

Supplemental Material, sj-pdf-1-epb-10.1177_23998083221078307 for Examining neighborhood effects on residents’ daily activities in central Shanghai, China: Integrating “big data” and “thick data” by Shenjing He, Chenxi Li, Yang Xiao, and Qiyang Liu in Environment and Planning B: Urban Analytics and City Science

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by the National Natural Science Foundation of China No.: [41871165], and The National Social Science Fund of China No. [19BSH035].

ORCID iDs

Yang Xiao

Qiyang Liu

Author Biographies

Professor Shenjing He is Professor of urban studies and head of the Department of Urban Planning and Design at The University of Hong Kong.

Dr. Yang Xiao is an associate professor in the College of Architecture and Urban Planning at Tongji University.

Miss Chenxi Li is a PhD candidate in the Department of Urban Planning and Design at HKU.

Dr. Qiyang Liu is a post-doctoral fellow in the School of Urban Planning and Design at Peking University.

Supplemental material

Supplemental material for this article is available online.

Notes

References

Aditjandra

Cao

Mulley

(2012) Understanding neighbourhood design impact on travel behaviour: an application of structural equations model to a British metropolitan data. Transportation Research Part A: Policy and Practice 46: 22–32.

Ahas

Mark

(2005) Location based services—new challenges for planning and public administration? Futures 37: 547–561.

Andersson

Malmberg

(2018) Segregation and the effects of adolescent residential context on poverty risks and early income career: A study of the Swedish 1980 cohort. Urban Studies 55: 365–383.

Andersson

(2001) Spaces of socialisation and social network competition: a study of neighbourhood effects in stockholm, sweden. In: Andersen

Kempen

(eds), Governing European cities: social fragmentation, social exclusion and urban governance. Aldershot: Ashgate.

Arkaraprasertkul

(2009) Towards modern urban housing: redefining Shanghai’slilong. Journal of Urbanism: International Research on Placemaking and Urban Sustainability 2: 11–29.

Bauder

(2002) Neighbourhood effects and cultural exclusion. Urban Studies 39: 85–93.

Cao

(2010) Exploring causal effects of neighborhood type on walking behavior using stratification on the propensity score. Environment and Planning A: Economy and Space 42: 487–504.

CatherineRoss

EJM

Pribesh

(2001) Powerlessness and the amplification of threat: neighborhood disadvantage, disorder and mistrust. American Sociological Review 66(4): 568–591.

Cheng

Wang

(2013) Do neighbourhoods have effects on wages? A study of migrant workers in urban China. Habitat International 38: 222–231.

10.

Clark

Scott

(2013) Does the social environment influence active travel? An investigation of walking in Hamilton, Canada. Journal of Transport Geography 31: 278–285.

11.

Cradock

Kawachi

Colditz

, et al. (2009) Neighborhood social cohesion and youth participation in physical activity in Chicago. Social Science & Medicine 68: 427–435.

12.

Ellen

Turner

(1997) Does neighborhood matter? Assessing recent evidence. Housing Policy Debate 8: 833–866.

13.

Fairclough

Boddy

Mackintosh

, et al. (2015) Weekday and weekend sedentary time and physical activity in differentially active children. J Sci Med Sport 18: 444–449.

14.

Feng

Wang

Murray

, et al. (2020) Multi-objective trajectory optimization in planning for sequential activities across space and through time. Environment and Planning B: Urban Analytics and City Science 239980832091330.

15.

Ferreira

Johansson

Sternudd

, et al. (2016) Transport walking in urban neighbourhoods—Impact of perceived neighbourhood qualities and emotional relationship. Landscape and Urban Planning 150: 60–69.

16.

Friedrichs

Galster

Musterd

(2003) Neighbourhood effects on social opportunities: the European and American research and policy context. Housing Studies 18: 797–806.

17.

Galster

G. C.

(2012) The mechanism (s) of neighbourhood effects: Theory, evidence, and policy implications. Neighbourhood effects research: New perspectives. Dordrecht: Springer, 23–56. Available at: https://doi.org/10.1007/978-94-007-2309-2_2.

18.

Galster

(2019) Making Our Neighborhoods, Making Our Selves. Chicago and London: The University of Chicago Press.

19.

Gehrke

Wang

(2020) Operationalizing the neighborhood effects of the built environment on travel behavior. Journal of Transport Geography 82: 102561.

20.

Gobillon

Selod

Zenou

(2007) The mechanisms of spatial mismatch. Urban Studies 44: 2401–2427.

21.

Gong

Jin

Liu

, et al. (2020) Identifying urban residents’ activity space at Multiple geographic scales using mobile phone data. ISPRS International Journal of Geo-Information 9.

22.

Gonzalez

Hidalgo

Barabasi

(2008) Understanding individual human mobility patterns. Nature 453: 779–782.

23.

(2015) Homeowner associations and neighborhood governance in Guangzhou, China. Eurasian Geography and Economics 56: 260–284.

24.

Jia

Xia

, et al. (2018) Cardiovascular diseases in middle aged and older adults in China: the joint effects and mediation of different types of physical exercise and neighborhood greenness and walkability. Environmental Research 167: 175–183.

25.

Jones

Pebley

(2014) Redefining neighborhoods using common destinations: social characteristics of activity spaces and home census tracts compared. Demography 51: 727–752.

26.

Kain

(2004) A Pioneer’s Perspective on the Spatial Mismatch Literature. Urban Studies 41: 7–32.

27.

Kesteloot

Murie

Musterd

(2006) European cities: neighbourhood matters. In: Musterd

Murie

Kesteloot

(eds) Neighbourhoods of Poverty. London: Palgrave Macmillan.

28.

Kim

Parent

Vom Hofe

(2018) The role of peer effects and the built environment on individual travel behavior. Environment and Planning B: Urban Analytics and City Science 45: 452–469.

29.

Lewis

(1966) The culture of poverty. Scientific American 215(4): 19–25.

30.

(2013) Residential satisfaction in China’s informal settlements: a case study of beijing, shanghai, and guangzhou. Urban Geography 34: 923–949.

31.

Liu

(2017) Unpacking the heterogeneity of poor neighbourhoods and neighbouring in large Chinese cities. Area Development and Policy 2: 294–313.

32.

Liu

Wang

Grekousis

, et al. (2019) Neighbourhood greenness and mental wellbeing in Guangzhou, China: what are the pathways? Landscape and Urban Planning 190: 103602.

33.

Liu

, et al. (2017) Changing neighbourhood cohesion under the impact of urban redevelopment: a case study of Guangzhou, China. Urban Geography 38: 266–290.

34.

Liu

Zhang

Jin

, et al. (2020) Spatial pattern of leisure activities among residents in Beijing, China: exploring the impacts of urban environment. Sustainable Cities and Society 52: 101806.

35.

Ludwig

Duncan

Gennetian

, et al. (2012) Neighborhood effects on the long-term well-being of low-income adults. Science 337: 1505–1510.

36.

Marra

Becker

Axhausen

, et al. (2019) Developing a passive GPS tracking system to study long-term travel behavior. Transportation Research Part C: Emerging Technologies 104: 348–368.

37.

Musterd

Andersson

(2006) Employment, social mobility and neighbourhood effects: the case of Sweden. International Journal of Urban and Regional Research 30: 120–140.

38.

Perchoux

Chaix

Brondeel

, et al. (2016) Residential buffer, perceived neighborhood, and individual activity space: New refinements in the definition of exposure areas – The RECORD cohort study. Health & Place 40: 116–122.

39.

Permentier

Van Ham

Bolt

(2008) Same neighbourhood different views? A confrontation of internal and external neighbourhood reputations. Housing Studies 23: 833–855.

40.

Shen

Kwan

Chai

(2013) Investigating commuting flexibility with GPS data and 3D geovisualization: a case study of Beijing, China. Journal of Transport Geography 32: 1–11.

41.

Simpson

Purdam

Tajar

, et al. (2006) Ethnic Minority Populations and the Labour Market: An Analysis of the 1991 and 2001 Census. UK: Department for Work and Pensions.

42.

Xie

, et al. (2017) Community deprivation, walkability, and public health: Highlighting the social inequalities in land use planning for health promotion. Land Use Policy 67: 315–326.

43.

Wang

Zhang

, et al. (2019) Neighborhood social reciprocity and mental health among older adults in China: the mediating effects of physical activity, social interaction, and volunteering. BMC Public Health 19.

44.

Wang

Zhou

(2017) The built environment and travel behavior in urban China: a literature review. Transportation Research Part D: Transport and Environment 52: 574–585.

45.

Webster

(2010) Path dependency and the neighbourhood effect: urban poverty in impoverished neighbourhoods in chinese cities. Environment and Planning A: Economy and Space 42: 134–152.

46.

Xiao

Guo

, et al. (2017) Estimating the willingness to pay for green space services in Shanghai: implications for social equity in urban China. Urban Forestry & Urban Greening 26: 95–103.

47.

Xiao

Miao

Zhang

, et al. (2021) Exploring the health effects of neighborhood greenness on Lilong residents in Shanghai. Urban Forestry & Urban Greening 66.

48.

Xiao

Wang

Fang

(2019) Exploring the disparities in park access through mobile phone data: evidence from Shanghai, China. Landscape and Urban Planning 181: 80–91.

49.

(2021) The Contingency of Neighbourhood Diversity: Variation of Social Context Using Mobile Phone Application Data. Urban, UK: Urban Studies.

50.

Zhao

(2004) From shikumen to new-style: a rereading of lilong housing in modern Shanghai. The Journal of Architecture 9: 49–76.

51.

Zhong

Schläpfer

Müller Arisona

, et al. (2017) Revealing centrality in the spatial structure of cities from human activity patterns. Urban Studies 54: 437–455.

52.

Zhong

Boarnet

(2020) A Two-Dimensional Propensity Score Matching Method for Longitudinal Quasi-Experimental Studies: A Focus on Travel Behavior and the Built Environment. Urban, UK: Environment and Planning B: Urban Analytics and City Science, 48, 2110–2122.

53.

Zhou

Cai

, et al. (2019) Social inequalities in neighborhood visual walkability: Using street view imagery and deep learning technologies to facilitate healthy city planning. Sustainable Cities and Society 50: 101605.

54.

Zhou

Yeh

, et al. (2018) A commuting spectrum analysis of the jobs–housing balance and self-containment of employment with mobile phone location big data. Environment and Planning B: Urban Analytics and City Science 45: 434–451.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.42 MB