4.1. Formulation of QOL indicator
The proposed framework integrates both subjective and objective aspects into an urban environment-related QOL. Specifically, the urban environment-related QOL is determined by each QOL-component indicator in each district, weighted by the individuals’ subjective values of that indicator. The QOL is formulated as
$${QOL}_{i}^{p}=\sum _{k}{w}_{k}^{p}\bullet ({x}_{k,i}-{\stackrel{-}{x}}_{k})$$
1
where, \({QOL}_{i}^{p}\)is the QOL value of individual attribute \(p\) in district \(i\), \({w}_{k}^{p}\)is the subjective value of individual attribute \(p\) for service \(k\), and \({x}_{k,i}\) is the QOL component indicator, which determines the accessible value of service \(k\)obtained in district \(i.\) \({\stackrel{-}{x}}_{k}\) is the mean of \({x}_{k,i}\) within the urbanized area in the entire subject region. Thus, the QOL value describes the extent to which the urbanized area is better or worse than the standard living environment. It also determines the attribute-dependent variations in this difference.
To determine the spatial internal structure within the city, the resolution was set to 500-m mesh. The QOL component indicators in each district were calculated from statistical data and GIS applications. The subjective values were estimated from the results of a questionnaire survey. This evaluation considered the individual attributes (such as gender, age, income, and region) of the participants. The QOL values were then obtained in each 500-m mesh and for each individual attribute, including the subjective aspects.
4.2. QOL components and indicators related to the urban environment
The QOL components related to the urban environment in previous researches (Supplementary Table 2) were reorganized into five dimensions of outcomes attributable to citizen-oriented urban policies. Health, which has been the norm in many previous studies, was incorporated into the five dimensions as health care access and traffic safety. The QOL dimensions are described below.
1) Economic opportunity
This dimension evaluates the employment and economic opportunities within a living area, including number of job opportunities and accessibility of commuting 46. It also evaluates the affordability of an area in terms of house rents and housing affordability 47.
2) Living and Cultural opportunity
This dimension evaluates the accessibility of foundational daily-life services in a living place. The importance of accessibility to all aspects of life and culture, including shopping, healthcare, and cultural activities, has been the focus of many studies 45,50.
3) Amenity
This dimension assesses the comfort of housing and its neighborhood. Housing comfort is related to the size and performance of the living space57. In a residential neighborhood, it is related to the availability of open space and environmental problems such as noise and vibration58,59.
4) Safety and security
This dimension evaluates the safety and security of the living area, including vulnerability to natural disasters, traffic safety and public security 60, and health hazards such as air pollution41,61.
5) Environmental friendliness
This dimension assesses the environmental impact of living in an area. The uses of public transportation and energy within a housing are influenced not only by lifestyle, but also by the characteristics of the living area. The environmental impact is multifaceted and includes energy consumption, impact on global warming, resource recycling, and biodiversity.
Three indicators were set for each of the five dimensions (Table 1). Ideally, we should include all urban-environment indicators that are assumed to vary from one individual or region to another, but such a comprehensive inclusion is impractical. Here 15 indicators were selected based on a) ease of obtaining their data and b) the presence of regional differences. Access time to other cultural and lifestyle facilities (LC4) was added as another service access that reflects the diversity of facility locations in different areas.
Table 1
List of QOL component indicators
Dimension
|
Sub-Dimension
|
QOL Component Indicator
|
Economic opportunity
|
Job opportunity
|
EC1: Active job openings-to-applicants ratio
|
Access to housing
|
EC2: House rent (JPY/month)
|
Commuting level
|
EC3: Commuting time(min)
|
Living and Cultural
opportunity
|
Commercial access
|
LC1: Access time to nearest shopping (min)
|
Health care access
|
LC2: Access time to nearest clinic (min)
|
Public service access
|
LC3: Access time to nearest train station (min)
|
Other service access
|
LC4: Access time to other cultural and lifestyle facilities (min)
(Convenience stores, Restaurants, Amusement facilities, Large-scale retail stores, City center, General hospital, Town hall, Banks/Post offices, Daycare, Elementary school, Libraries/Community centers, Cultural facilities, Sports facilities, Parks, Shared offices)
|
Amenity
|
Housing environment
|
AM1: Living space (m2/capita)
|
Natural environment
|
AM2: Access time to nearby playground/square (min)
|
Sound environment
|
AM3: Noise level (dB)
|
Safety and Security
|
Natural safety
|
SS1: Flood depth (m)
|
Traffic safety
|
SS2: Traffic fatalities per 100,000 population
|
Public safety
|
SS3: Number of criminal offenses per 1,000 population
|
Environmental Friendliness
|
Low-carbon
|
EF1: Energy saving performance (%)
|
Biodiversity
|
EF2: Green coverage rate (%)
|
Cleanliness
|
EF3: Time to garbage collection point (min)
|
The QOL component indicators can be broadly classified into two categories: indicators related to transport (e.g., access time to a shopping facility) and indicators not related to transport (e.g., living space). The transport-related indicators were calculated by Eq. (2), which considers that a diversity of available choices might improve accessibility:
$${x}_{k,i,m}=-\frac{1}{\mu }{ln}{\sum }_{j}{{AT}}_{k,j}\cdot {exp}(-\mu \cdot {t}_{ij,m})$$
2
where \({x}_{k,i,m}\) is the access time from district \(i\) to the facility providing service \(k\)with diverse choices of travel mode \(m\), \({AT}_{k,j}\)は is the number of facilities providing service \(k\) in district \(j\), \({t}_{ij,m}\) is the access time by travel mode \(m\) between districts \(i \text{a}\text{n}\text{d} j\), and \(\mu\) is a distance attenuation parameter. The value of \(\mu\) depends on the travel purpose and is here set as \(\mu =1\).
4.3. Spatial data collection
Table 2 shows the calculation method and data for determining the QOL component indicators. The data in each 500-m mesh were extracted from geospatial information and GIS tools. The data distributions of each indicator in each metropolitan area are shown in Supplementary Method 2.
Table 2
Indicator
|
Data source and calculation method
|
EC1: Active job openings-to-applicants ratio
|
Source: Active job openings-to-applicants ratio of published by Public employment security office
|
EC2: House rent (JPY/month)
|
Source: Housing and Land Survey of Japan and real estate information
|
EC3: Commuting time (min)
|
Source: Commuting time based on the Housing and Land Survey of Japan
|
LC1: Access time to nearest shopping (min)
|
Calculate the access time from the center of the 500m mesh to the facility using ArcGIS network analyst.
Source of facilities: Digital national land information and open street map
|
LC2: Access time to nearest clinic (min)
|
LC3: Access time to nearest train station (min)
|
LC4: Access time to other cultural and lifestyle facilities (min)
|
AM1: Living space (m2/capita)
|
Source: Population and Housing Census of Japan
|
AM2: Access time to nearby playground/square (min)
|
Calculate the access time from the center of the 500m mesh to the facility using ArcGIS network analyst.
Source of facilities: Digital national land information and open street map
|
AM3: Noise level (dB)
|
Noise regulation standard values for each city's zoning by 500m mesh
|
SS1: Flood depth (m)
|
Calculate area-weighted average of assumed maximum inundation depth of hazard maps by 500m mesh.
Source of hazard maps: Ministry of Land, Infrastructure, Transport and Tourism of Japan
|
SS2: Traffic fatalities per 100,000 population
|
Source: Institute for Traffic Accident Research and Data Analysis
|
SS3: Number of criminal offenses per 1,000 population
|
Source: National Police Agency
|
EF1: Energy saving performance (%)
|
Spatial differences in this indicator are not considered in this study
|
EF2: Green coverage rate (%)
|
Percentage of green space within 500m mesh using GIS
|
EF3: Time to garbage collection point (min)
|
Spatial differences in this indicator are not considered in this study
|
4.4. Subjective-value estimation methods
(Basic 15 indicators)
The weights of multiple indicators can be determined by various methods such as the analytic hierarchy process, the contingent valuation method, and conjoint analysis. In the present study, the weight parameters of the 15 basic indicators were estimated through pairwise-comparison conjoint analysis. This method was chosen for two reasons: 1) it calculates weights through monetary conversions and other units, and 2) it sets trade-offs and constraints to control overestimation. For pairwise-comparison conjoint analysis, the survey respondents were asked to repeatedly select their preferred profile from two residential district profiles. The profiles were described by several indicators and their weight parameters were estimated with a binary logit model using the maximum-likelihood method. The formulations are given as
$${P}^{p}\left(i\right)={exp}({\beta }_{k}^{p}\cdot {x}_{k,i})/{\sum }_{j}{exp}({\beta }_{k}^{p}\cdot {x}_{k,j}),$$
3
$${\beta }_{k}^{p}={\beta }_{k}^{0}+{\sum }_{n}{\beta }_{k}^{n}$$
4
where \({P}^{p}\left(i\right)\) is the probability of individual attribute \(p\) selecting residential district \(i\), \({\beta }_{k}^{p}\) is the weight parameter of indicator \(k\), \({x}_{k,i}\) is the value of indicator k in residential district \(i\), \(n\) is an attribute (such as gender, age, income) of the individual, \({\beta }_{k}^{0}\) is the weight parameter of the reference attribute, and \({\beta }_{k}^{n}\) is the weight-parameter difference between the reference and individual attributes. The various individual attributes are combined into the weight parameter \({\beta }_{k}^{p}\).
The estimated weight parameters of the different indicators have different units. In this study, the QOL measure was given in monetary units. The monetary equivalent \({w}_{k}^{p}\) of the weight parameter of each indicator was calculated in terms of the house rent parameter \({\beta }_{EC2}^{p}\) as follows:
$${w}_{k}^{p}=-{\beta }_{k}^{p}/{\beta }_{EC2}^{p}$$
5
When designing the questionnaire for the conjoint analysis, the good and bad levels of each indicator were set as shown in Table 3. Along each of the five dimensions, profiles consisting of four indicators (three EC indicators plus house rent as a conversion indicator) were created as orthogonal arrays.
Table 3
Attribute levels for the conjoint analysis
Dimension
|
Indicator
|
Level 1
|
Level 2
|
Economic opportunity
|
EC1: Active job openings-to-applicants ratio
|
2.0
|
1.0
|
EC2: House rent (10k JPY/month)
|
10k JPY cheaper
|
10k JPY more expensive
|
EC3: Commuting time (min)
|
15 min
|
45 min
|
Living and Cultural
opportunity
|
LC1: Access time to nearest shop (min)
|
10 min
|
30 min
|
LC2: Access time to nearest clinic (min)
|
10 min
|
30 min
|
LC3: Access time to nearest train station (min)
|
10 min
|
30 min
|
Amenity
|
AM1: Living space (m2/capita)
|
40 m2
|
20 m2
|
AM2: Access time to nearby playground/square (min)
|
5 min
|
15 min
|
AM3: Noise level (dB)
|
40 dB
|
60 dB
|
Safety and Security
|
SS1: Flood depth (m)
|
Less than 0.5 m
|
1.5 m
|
SS2: Traffic fatalities per 100,000 population
|
2 people
|
6 people
|
SS3: Number of criminal offenses per 1,000 population
|
3 people
|
10 people
|
Environmental Friendliness
|
EF1: Energy saving performance (%)
|
20%
|
0%
|
EF2: Green coverage rate (%)
|
40%
|
5%
|
EF3: Time to garbage collection point (min)
|
5 min
|
15 min
|
(Other cultural and lifestyle facilities)
The other life-service opportunities (15 facilities) were weighted based on a pairwise comparison questionnaire with two attributes related to burden on the survey respondents. Specifically, the relative importance of a) reduced access time to nearby shopping destinations and b) equally reduced access time to another certain facility was weighted on a seven-point scale from 0 to 2.
$${w}_{k}^{p}={w}_{LC1}^{p}\cdot {\beta }_{k}^{p}$$
6
(Transport mode preferences)
The time of reaching a facility depends on the availability of transport modes. In addition, the transport preferences and associated residential preferences vary among people62,63. This study evaluated the participants’ preferences for the following transport modes: walking, bicycle, car, and public transportation. In the questionnaire, the participants were presented with two transport modes with different access times to the destination, and were repeatedly asked to state their preference.
The weight parameters were estimated using a logit model based on the maximum-likelihood method shown in Equations (3) and (4) and the 15 basic indicators. The probability of selecting transport mode m depended on the individual attributes, \({S}_{k,i,m}^{p}\) and was calculated using Eq. (7). The accessibility value was based on the transportation preferences and was calculated using Eq. (8):
$${S}_{k,i,m}^{p}={exp}({\beta }_{m}^{p}\cdot {x}_{k,i,m})/{\sum }_{r}{exp}\left({\beta }_{r}^{p}\cdot {x}_{k,i,r}\right)$$
7
$$\left({x}_{k,i}-{\stackrel{-}{x}}_{k}\right)={\sum }_{m}{S}_{k,i,m}^{p}\bullet \left({x}_{k,i,m}-{\stackrel{-}{x}}_{k,m}\right)$$
8
4.5. Questionnaire survey for estimating subjective values
The weights of the QOL component indicators were based on the individual attributes and the results of the online questionnaire survey, conducted in March of 2021. The target respondents were citizens aged 20 years or older living in five major metropolitan areas in Japan. The sample size was equally allocated among the metropolitan areas, genders, and age groups, and responses were obtained from 3,090 respondents.
The subjective values of the participants were gained from three major categories of responses: 1) values of the 15 basic indicators, 2) importance of other life-service opportunities (15 facilities), and 3) transport mode preferences. Each participant gave 1) 36 responses to different combinations of profiles, 2) 15 responses to different facilities (one choice per facility), and 3) 15 responses to different combinations of access time and transport mode.
The following individual attributes were also obtained: gender, age, household composition, occupation, income, housing type, and usual car usage (Supplementary Table 5).
The weight parameters were estimated as shown in Supplementary Method 3.