exploring the built environment factors in the metro that

Research ArticleExploring the Built Environment Factors in the Metro ThatInfluence the Ridership and the Market Share of the Elderlyand Students

Jiang Ning 1 Tao Lyu 2 and Yuanqing Wang 2

1School of Traffic and Transportation Lanzhou Jiaotong University Lanzhou 730070 China2College of Transportation Engineering Department of Traffic Engineering Changrsquoan University Xirsquoan 710064 China

Correspondence should be addressed to Yuanqing Wang wyqingchdeducn

Received 25 March 2021 Accepted 4 October 2021 Published 13 November 2021

Academic Editor Hongtai Yang

Copyright copy 2021 Jiang Ning et al +is is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

+e metro has developed rapidly in the past two decades and has become one of the crucial patterns of transportation for urbanresidents in China Many studies have explored the factors affecting metro ridership but few have focused on the metro usage ofspecific groups such as the elderly and students +is paper uses the negative binomial regression model to explore the rela-tionship between the built environment and the metro ridership of three types of people (adults the elderly and students) by usingthe metro smart card data of Qingdao We also used the fractional response model to discuss the factors that influence theridership share for the elderly and students +e results show that most variables promote the metro usage of the three groups ofpeople but have a significantly different effect on the market share of those groups Specifically the number of schools hospitalssupermarkets squares parks and scenic spots near metro stations significantly increases the proportion of the elderly metrousage +e number of bus stops and schools substantially increases the share of metro ridership by students +e research resultscan provide valuable insights for promoting the metrorsquos overall ridership and minimizing the gap in allocating public transportresources among different groups

1 Introduction

With the rapid development of urbanization many prob-lems such as traffic congestion and environmental pollutionhave become increasingly prominent To solve this series ofsocial and environmental problems an increasing number ofChinese cities have begun to develop or expand rail transitsystems [1] As of the end of 2020 45 cities in China haveopened rail transit with a total operating mileage of79697 km of which the metro mileage is 62808 km [2]

To adapt to the rapid development of the metro systemmany researchers began to predict and explore the factorsaffecting metro ridership to formulate metro developmentand operation plans +e most widely used ridership modelsare the general four-step model [3] and activity-based model[45] Although these models can quantify urban passengerdemand well they have disadvantages such as higher cost

and lower accuracy [6ndash8] In recent years with the con-tinuous development of geographic information systems(GIS) a large number of studies have adopted statisticalmodels such as the ordinary least squares (OLS) regressionmodel [9] structural equation model (SEM) [10] and directridership models (DRMs) [11] combined with urban builtenvironment to explore the factors influencing metro rid-ership Compared with traditional models they are simplestraightforward interpretation of results and cost less

However most studies have taken total ridership as theresearch object but only a few studies have explored the useof the metro by specific groups such as the elderly andstudents Affected by unfavorable factors such as physiologyand economic income specific groups have more urgentpublic transportation demands than mainstream adultsWith the widespread use of smart card data many re-searchers use card types to classify passengers and explore

HindawiJournal of Advanced TransportationVolume 2021 Article ID 9966794 10 pageshttpsdoiorg10115520219966794

the travel characteristics and differences between main-stream and other groups [12 13] However these studies arebased on bus data and do not involve the metro usage ofspecific groups In addition policy recommendation is morevaluable when exploring the travel patterns and preferencesof multiple groups of people using the metro at the sametime +erefore this paper uses metro smart card data toexplore the metro usage by adults and other groups (theelderly and students) to fill the gaps in the research contentof this direction

+e structure of this study is arranged as follows Section2 reviews related literature Section 3 describes the study areaand data Section 4 introduces the research method andSection 5 presents the research results and discussionSection 6 summarizes the conclusion and limitations

2 Literature Review

Numerous previous studies exist on the travel behavior ofthe elderly and the travel mode choices of students to go toschool However few studies have focused on the metrotravel behaviors of the elderly and students +erefore thissection will review this literature as references

21 Elderly People As problems of aging continue to in-tensify it is crucial to maintain the sense of social partici-pation and mobility of the elderly [14ndash17] As with themainstream population the built environment affects thetravel modes of the elderly [18ndash20] Studies have shown thatneighborhood design characteristics are closely related to theactive travel mode of the elderly For instance older adultswill have more active travel in highly walkable neighbor-hoods [21] At the same time uneven sidewalks pot-holesand other harsh walking environments will affect both theaccessibility and mobility of older adults [22] Since mostolder adults have retired they have more time to engage innonwork activities Compared with young people theirmobility may not have been significantly reduced [17]Meanwhile due to the Chinese cultural background andpractical needs the elderly living with their adult childrenalso have other out-of-home activities under family re-sponsibilities [19] For example compared with others olderpeople with many family members may have more bus tripsbecause they often help take children to school [23]

Previous studies have shown that Western older adultshave strong car dependence [24 25] Affected by the eco-nomic development level and social background the ac-cessibility of public transportation is an essentialdeterminant of the travel behavior of the elderly in China[19] For example a study in Nanjing showed that thenumber of bus stops and bike share stations has a positiveeffect on the travel frequency of the elderly [18] As animportant pattern of travel for the elderly public trans-portation also helps to improve the accessibility and fairnessof the elderly to deal with the unbalanced distribution ofurban public service facilities For instance compared withthe elderly in urban areas the elderly living in suburbs have alonger distance to get medical treatment and are moredependent on public transportation [26]

To further protect the well-being of the elderly manyChinese cities have adopted various preferential policies toencourage the elderly to use public transportation dis-counted fares and monthly public transportation subsidiesAlthough these policies have effectively encouraged the useof public transport the effect may be quite limited A studyin Beijing showed that free bus programs for the elderly onlyinduced less than 5 of bus trips among older adults [23]+erefore to further attract more elderly people to usepublic transportation to maintain their necessary mobilityin addition to public transportation preferential policies wealso need to consider how to increase the attractiveness ofpublic transport to the elderly from other perspectives

22 Elementary andMiddle School Students Going to schoolis the leading travel activity for elementary and middleschool students (6 to 18 years old) A large number of studieshave shown that the built environment is closely related tothe studentsrsquo trip pattern of going to school [27ndash30] Firstthe distance between home and school is the main factor+e increase in travel distance will prompt students to switchfrom active travel mode to motorized mode An Irish studyindicated that 2 km is the dividing line between motorizedand nonmotorized modes for students [31] A Chinese studyshowed that as the distance of travel increases studentswould switch from walking to riding a bicycle When thetravel distance exceeds 3 km they will use more publictransportation and private cars to go to school [32] Secondthe number of street intersections is also related to thestudentsrsquo school transportation patterns A small number ofintersections will encourage children with shorter traveldistances to use active travel mode but the increased in-tersection number will affect the comfort of the journey andreduce the propensity of students with longer travel dis-tances to use public transportation [29] +ird studentsrsquotravel patterns are closely related to the local geography andculture Due to Finnish cycling culture and developed cy-cling infrastructure 43 of students cycle to school inHelsinki [27] In Hong Kong due to safety and topo-graphical issues students rarely ride bicycles to school andtheir main travel modes are walking and school buses [33]Finally due to a long commuting distance and complicatedurban transportation system a single traffic vehicle may notmeet the commuting needs of students+e only study aboutstudents commuting by the metro shows that the density ofbus stops and docks at bike share stations positively affectsstudents who use the metro to travel [30] +is means theconnectivity of transportation also affects the travel patternof students

Many studies recommend that students use active modeto school to protect their health and ease urban trafficcongestion [29 34] However some Western and Chinesestudies have shown that the existing urban education systemis unfair Many students start to go to schools farther away toobtain higher-quality educational resources [29 35 36] it isessential for them to use safe and reliable transportationLimited by operational efficiency and inadequate supervi-sion and management the school bus system in China is not

2 Journal of Advanced Transportation

yet mature [30] +erefore besides private cars publictransportation is a significant pattern for students who havelong commutes to school

23 Literature Summary In summary the existing literaturefocuses on the influence of the built environment on thetravel of the elderly and students Still the relationshipbetween these factors and metro ridership of the elderly andstudents has not been explored+erefore the paper uses thesmart card and built environment data to analyze the travelbehavior of adult the elderly and student metro usage by thenegative binomial regression model In addition consid-ering the importance and necessity of public transportationfor these groups we also used the panel data modelframework to explore the factors affecting the market shareof specific groups to improve their transportation equity+is result can evaluate the built environment preferences ofthe elderly and studentsrsquo metro usage and provide valuableinsights into route planning and station design

3 Study Area and Data

31 Study Area Qingdao is a subprovincial city located inShandong Province an important central city and inter-national port city along the coast of China Qingdaorsquospermanent population was 950 million in 2019 of which645 million lived in the urban area including 220 millionolder adults and 099 million elementary and middle schoolstudents [37] Qingdaorsquos built-up area was 75816 km2 andthe urbanization rate was 7412 [37 38] +ere are twomain reasons for choosing Qingdao as the research objectFirst Qingdao is the second Chinese batch of selectedldquotransit metropolisrdquo model cities with a relatively developedmetro system +e city has already built and operates fourmetro lines by the end of 2019 +e total length of the metronetwork is 176 km and includes 82 stations and an annualpassenger capacity of 187 million Second Qingdao providesexcellent preferential policies to encourage the elderly andstudents to use public transportation Elementary andmiddle school students and older adults (60 to 64 years old)can take the bus andmetro at half price+e elderly of 65 andabove can take free public transportation +ereforeQingdao is an excellent case for this research to explore themetro usage of the elderly and students +e study areaincludes the main urban areas of Shinan Shibei LicangLaoshan and the urban areas of Huangdao and Jimo coveredby the metro network +e metro network of Qingdao isshown in Figure 1

32 Data

321 Data Processing To explore the built environmentrsquosinfluence on the metro usage of three groups of people thisstudy used the Qingdao Metro smart card and built envi-ronment data

We use Qingdao Metro smart card data from 1352019(Monday) to 1952019 (Sunday) +e data come fromQingdao Metro Company Due to route planning issues 3

stations were not yet operational at that time +erefore thisstudy includes card transaction data generated by 79 sta-tions After a passenger enters the station and swipes a carda data set will be generated including passenger cardnumber card type transaction date and transaction timeSee Table S1 in Supplementary Material for the metro smartcard data structure Since passengers need to swipe theircards to enter and exit the metro station the two transactionrecords represent one trip

+e Qingdao Metro smart card types mainly includeregular cards old age cards and student cards +e old agecard is only for adults aged 60 and above+e student card isused only by school students under 18 In this study we usedonly regular cards old age cards and student cards and usedthese card types to represent adults (18ndash59 years old) theelderly and students respectively

+e original data of Qingdao Metro in one week in-cluded a total of 604 million transaction records Afterdeleting some records that had errors or were incompletewe extracted the card types corresponding only to the threegroups Our data set included 3314 million metro trans-actions accounting for 549 of the original data In the dataset adults the elderly and students respectively have 26200489 and 0205 million transaction records+e elderly andstudents respectively accounted for 148 and 62 of thedata set +is proportion is much smaller than the actualproportion of the elderly and elementary and middle schoolstudents in Qingdao (148lt 232 62lt 104) +eresult means the elderly and students are far less likely to usethe metro than adults

In this paper we used point of interest (POI) datacollected from the Baidu map to represent the built en-vironment We selected the explanatory variables as fol-lows First we extracted the characteristics of metrostations and used them to measure which type of metrostation people like to use Specifically four factors wereconsidered bus stops road intersections metro stationentrancesexits and whether the station is in the mainurban area Second considering peoplersquos travel purpose wefurther chose the following variables extracted the hos-pitals squares and parks to represent their medical andleisure demands and selected the scenic spots and super-makets to represent their entertainment and shoppingdemands In addition we extracted elementary and middleschools to represent studentsrsquo daily commuting purposes+e descriptive statistics of variables are shown in Table 1All variables related to numbers were calculated within an800m distance from each metro station which is con-sidered the standard distance and is often used to delimitthe buffer of metro stations [1 39]

322 Data Analysis We divided themetro operating period(6 00ndash23 00) into 16-time intervals and calculated thetravel time distribution of adults the elderly and students asshown in Figure 2 +e three groups of people have obviousmorning and evening peak travel characteristics but areslightly different Specifically the adults have more trips inthe morning peak than in the evening peak For the elderly

Journal of Advanced Transportation 3

the most frequent trips are from 7 00 to 11 00 in the wholeday as well as the proportion of trips between 11 00 and 15 00 is higher than that in other groups +is travel charac-teristic is because most older adults have retired and theirschedule is more flexible +e travel time distribution ofstudents is roughly similar to adults but there are still twodifferences +e first is that studentsrsquo morning and eveningpeaks are earlier than adults because they go to schoolleaveschool earlier than adults go to workoff work Second theproportion of their trips in the morning peak is less than thatin the evening peak because some parents drive their stu-dents to school in the morning [29] +ese students need totake other transportation to get back home by themselves inthe evening

Station-level metro ridership composition is shown inFigure 3 In general the composition of passengers in dif-ferent stations has a significant difference Specificallystations with a high proportion of adults traveling are mainlyconcentrated in the main urban area because this area has ahigh job density Adults have a lot of commuting demands atthese stations Quite differently from the stations with a highproportion of adults the stations with a high proportion ofelderly travelers are mainly located in the northeast andsouthwest areas of Qingdao far away from the city center+is can be explained in two ways First a previous studyshows that both adults and the elderly in Qingdao prefer totravel in main urban areas but the total amount of adulttravel is much more than that of older adults [12] +ereforefrom the overall ridership composition the urban elderlyrsquos

share of trips is relatively low Second as there are manytourism resources along with the metro in the southwest andnortheast of Qingdao this may attract many elderly pas-sengers Students have a higher proportion of trips at thestations on the west coast of Qingdao +is may be becausethe educational resources in the Huangdao district are moreconcentrated than that in other regions attracting muchridership by students along metro lines

4 Method

41 Negative Binomial Regression Model +e aim of thispaper is to explore the influence of station characteristicsand city facilities on the metro usage of adults the elderlyand students because there is much more ridership in somemetro stations than others in other words the sample dataare too discrete and random +erefore we estimated thecorresponding negative binomial regression model for threegroups of people which is widely used for data whose samplevariance significantly exceeds the average and is over-dispersed [40] +e negative binomial regression model canbe described as follows

yi exp β0 + 1113944n

j1βjxij + ε⎛⎝ ⎞⎠ (1)

where yi represents the metro ridership of the ith metrostation i (1 2 3 m) β0 represents the intercept xij

represents the jth independent variable of the ith metro

Laixi City

Jimo District

Pingdu City

N

16Miles

128420

Jiaozhou City

Huangdao District

Chengyang District

Metro StationMetro LineMain Urban AreaUrban Area

Suburban Area

Figure 1 +e metro network of Qingdao


station j (1 2 3 n) βj is the regression coefficient ofthe jth independent variable and ε is the error term of themodel and obeys the gamma-distribution

42 Fractional Response Model Our data set shows thatcompared with adults the share of the elderly and studentsin the metro ridership is much smaller than their actualpopulation +erefore we not only used the linear model toexplore the influence of the built environment on the rid-ership of the three groups of people but also used thefractional response model to explore the determinants of theshare of those groups in the metro ridership

+e fractional response model (also known as thefractional logit model) was first proposed by Papke andWooldridge [41] +e model can effectively deal withbounded dependent variables and does not require ad-ditional calculations and is often used to explain models in

which dependent variables such as share pass rate andproportion are between 0 and 1 In recent years this modelhas also gained some applicability in traffic research Forinstance McDonald used the fractional logit model toanalyze the influence of the Safe Routes to School (SRTS)program on the proportion of students walking or cyclingto school [42] Wang used the fractional logit model toexplore the difference in the influence of environmentalfactors on the share of trips made by men and womenusing bike share [43] +ese studies have verified theadvantages of the fractional response model comparedwith the linear model in dealing with bounded dependentvariables +e structure of the fractional response model isas follows [41 44]

E(y|x) G(x β) (2)

where G() is a nonlinear cumulative function the fittedvalue falls within the unit interval [0 1] +e fractionalresponse model in the form of logistic function can beexpressed as follows

E(y|x) G(x β) exp(xβ)

1 + exp(xβ) (3)

In this study x represents independent variables and βis the coefficient of the corresponding independent variabley is the dependent variable and represents the proportion ofelderly and student trips to the total metro ridership y iscalculated as follows

ythe elderly metro ridership for the elderlytotal ridership of themetro

ystudent metro ridership for the studenttotal ridership of themetro

(4)

To visually indicate the influence of selected variables onthe market share of the metro by the elderly and students wecalculate the average elasticity effects (AEEs) for the builtenvironment variables in the fractional response model +eelasticity effects represent the change in the market share ofthe metro ridership for the specific groups for every 1increase in the dependent variable while holding othersconstant

Table 1 Descriptive statistics of variables

Variable Description Mean Std DevStation characteristicsBus stops Number of bus stops in 800m buffer 921 594Road intersections Number of road intersections in 800m buffer 336 253Main urban area 1 if the metro station in main urban area 0 otherwise 065 048Entrancesexits Number of metro station entrances or exits in 800m buffer 290 122City facilitiesSchools Number of elementary and middle schools in 800m buffer 168 207Hospitals Number of tertiary hospitals in 800m buffer 015 039Supermarkets Number of supermarkets in 800m buffer 064 150Squares and parks Number of squares and parks in 800m buffer 144 176Scenic spots Number of scenic spots in 800m buffer 021 040

018

016

Adults

The elderly

Students

014

012

010

008

006

004

002

0006 8 10 12 14

Time of day (hour)

Pass

enge

r rat

e (

)

16 18 20 22

Figure 2 Proportion distribution of metro ridership among adultsthe elderly and students


Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210

StudentsMain Urban AreaUrban Area

(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210

AdultsThe elderlyStudentsUrban AreaSuburban Area

(b)

Figure 3 Station-level metro ridership composition of (a) east area and (station-level metro ridership composition of east area) and (b) westarea (station-level metro ridership composition of west area)


5 Results and Discussion

Before establishing the models we used the varianceinflation factor (VIF) to check the collinearity of thevariables and found that there was no collinearity amongall the variables +en we retained the variables signifi-cant at the 95 level and the model results are reported inTable 2 +e first three models are negative binomialregression models with the metro trips of three groups ofpeople as the dependent variable +e latter two arefractional response models with the proportion of elderlyand student trips to the total ridership as the dependentvariable To encourage more old adults and students to usethe metro and increase their market share we will focuson the factors that influence metro usage by the elderlyand students

51 Station Characteristics In our case the number of busstops will significantly enhance the attractiveness of themetro to adults and students Moreover the factor will havea positive effect on the trip share of students +e result ispredictable because students usually use the bus as a feedermode when they use the metro to go to school [30]+erefore higher bus stop densities help students transfer tothe metro thereby promoting the metro ridership by stu-dents Previous studies generally demonstrate that busstations positively correlate with metro ridership [45 46]However our results showed that this factor is a negativepredictor of the share of the elderly metro trips +is may bebecause compared with others older people do not liketransfer between metro and bus

+e results show that the number of road intersections ispositively correlated with the metro ridership of adultswhile it has a negative influence factor for the elderlyMoreover we found the factor significantly reduces theshare of the elderly +e results may mean that older peopleprefer metro stations with fewer intersections which pe-destrian environment maymake them feel safe However wedo not have more data to verify this hypothesis Futureresearch will continue to explore the influence of the pe-destrian environment in the catchment area of metro sta-tions on the elderly and students such as road networkdensity and road width

In this study the metro stations in the main urban areaand the number of station entrancesexits are statisticallysignificant and positively correlated with all groups +eseresults are interesting and can be explained by previousresearch For the previous variable Qingdao seniors preferto travel in the old city [12] and most of the elementary andmiddle schools are located in the Qingdao main urban area[47] +erefore compared with the urban area the metrostations located in the main urban area have more ridershipof the elderly and students For the latter variable althoughprevious studies have confirmed that the factor positivelyaffects metro ridership [39] it is still unknown whether thisfactor affects the elderly and students +e research resultsverify and further expand the scope of this conclusionMoreover our research also found that these two variables

have no positive effect on increasing the proportion of olderadults and students+is may be because metro stations withthese characteristics also attract a large number of adulttravelers Although the three groups of people have manytrips in those stations the adults have the most significantincrease

52 City Facilities +e number of elementary and middleschools is positively correlated with the metro ridership ofadults and students and the share of trips made by studentsSurprisingly the factor has also significantly increased theelderlyrsquos metro usage and proportion +e result of theformer is easy to understand +is latter finding can beexplained by a previous study that found some older adultsparticipate in some activities such as shopping in super-markets and escorting children to school to support theolder and younger generations in the family [17] +us thenumber of schools is positively correlated with the numberof trips and market share of the elderly and studentsMoreover the estimated margin effect shows that for every1 increase in the number of elementary andmiddle schoolsnear stations the share of students in the metro ridershipwill increase by 031 +e factor is most effective in in-creasing the proportion of students in metro ridership in allvariables

Regarding hospitals and supermarkets we found thatthese two factors significantly increased the overall ridershipbut only positively influenced the proportion of the elderly+e result means that a large proportion of the ridership bythe elderly meets their medical and shopping needs throughthe metro compared with other groups Previous studieshave verified that public transportation facilities positivelyinfluence the elderlyrsquos travel for medical treatment [26] andshopping [19] Our research reconfirms these conclusions interms of elderly metro usage

+e number of parks and squares is positively correlatedwith the ridership of the three groups of people In additionthose factors have significantly increased the proportion ofolder adult metro usage+e result is not surprising Becauseparks and squares are important leisure places for the elderlyin China they can meet various needs here such as exer-cising and chatting with friends [18 19]

Regarding scenic spots we found the factor is posi-tively associated with the elderly metro usage as well asthe share of the elderly At the same time we found thefactor is negatively associated with the share of studentsrsquotrips +is reason may be because many Chinese elderlyhave retired and have more free time to take the metro tovisit the scenic spots In contrast students may not havemuch free time +erefore the variable only can signifi-cantly improve the use of the elderly +e estimatedmarginal effect shows that every 1 increase in thenumber of scenic spots in the metro catchment area leadsto a 547 increase in the proportion of trips made by theelderly Among the variables associated with the char-acteristics of the built environment scenic spots have thegreatest significant influence on increasing the marketshare of elderly metro usage


6 Conclusions

+e metro is an essential infrastructure for alleviating trafficcongestion and prompting the cityrsquos development Exploringthe factors that affect the metro ridership can help policy-makers better formulate operational plans +is paper ex-plores the travel characteristics and built environmentpreferences of specific groups (elderly and students) usingthe metro +e results verify findings from previous studiesand provide new insights for improving the ridership of theelderly and students in station design and route planning

Compared with adults the share of the elderly andstudents in the metro ridership is much smaller than theproportion of the actual population in our data set+erefore we have not only used the negative binomialregression model to explore the influence of the built en-vironment on the metro usage of the three groups but havealso used the fractional response model to explore factorsthat increase the share of the elderly and students in themetro ridership In our results most variables are positivelyattractive to specific groups but have significant differencesin their influence on the market share of the elderly andstudents +is conclusion provides new ideas for increasingthe metro ridership and reducing the gap in allocating publictransportation resources For example the number ofschools in the catchment area of the metro station not onlyenhances the metro ridership but also increases the pro-portion of the elderly and students in the metro ridership

We found that the distribution of travel time of the threegroups of people showed morning and evening peaks but thepeak travel time of students is earlier than that of adults As theelderly have more free time they also have more trips between9 00 and 15 00 +ese conclusions help policymakers un-derstand the travel differences of different groups and for-mulate diversified metro ticket prices and operation plans tomeet the travel demands of specific groups for instance toalleviate the travel conflicts between the elderly and otherpeople during the morning peak periods+e government can

appropriately reduce the ticket price discount for the elderlyduring peak periods and encourage them to take more tripsduring off-peak periods At the same time policymakersshould consider shortening studentsrsquo travel time and distancethrough good urban planning schemes and education man-agement to ensure the healthy growth of children

Among all the station characteristics the number of busstops is the only factor that increases the number of metrotrips and the market share of students +e result means thatin future transportation planning policymakers can plansome bus routes near high-density childrenrsquos residences andschools to facilitate studentsrsquo changes of transportationmode In addition we found that metro stations located inthe main urban area with many entrancesexits can sig-nificantly attract more elderly and student passengers +isdiscovery is crucial and provides a reference for policy-makers to increase metro ridership from the metro linedesign and station characteristics

+e study found that the number of schools aroundmetro stations significantly increased studentsrsquo usage andshare of the metro ridership Surprisingly the factor alsopositively influences the number and proportion of metrotrips by the elderly +erefore policymakers should considerimproving the travel environment of schools near metrostations to facilitate the travel of students and the elderlyMoreover we found that the number of hospitals super-markets squares parks and scenic spots is significantly andpositively associated with the elderly metro usage and themarket share of the elderly +ese results indicate that themetro is essential for maintaining the quality of older adultsrsquodaily lives In response to the increase in the aging societypolicymakers should consider building metro stations nearthe elderly community and elderly care homes to maintainthe necessary mobility of the elderly

+e study has some limitations First although we ex-plored the travel time distribution of adults senior citizensand students using the metro we did not consider the impactof date attributes on the three groups of people In the future

Table 2 Results for negative binomial regression models and fractional response model

Negative binomial regression model Fractional response modelAdults +e elderly Students +e elderly Students

Coef p-value Coef p-value Coef p-value Coef p-value AEEs Coef p-value AEEsStation characteristicsBus stops 0030 lt0001 0026 lt0001 minus0034 lt0001 minus056 0007 0047 004Road intersections 0064 lt0001 minus0055 lt0001 minus0082 lt0001 minus134Main urban area 1240 lt0001 0671 lt0001 0474 lt0001 minus0259 lt0001 minus423 minus0350 lt0001 minus219Entrancesexits 0129 lt0001 0033 0004 0090 lt0001 minus0093 lt0001 minus151City facilitiesSchools 0080 lt0001 0123 lt0001 0142 lt0001 0057 lt0001 094 0049 lt0001 031Hospitals 0157 lt0001 0267 lt0001 0313 lt0001 0064 0014 105Supermarkets 0136 lt0001 0146 lt0001 0121 lt0001 0028 lt0001 046 minus0021 0035 minus013Squares and parks 0083 lt0001 0087 lt0001 0066 lt0001 0019 0009 032Scenic spots minus0228 lt0001 0077 0009 0335 lt0001 547 minus0072 0038 minus045Constant 2444 2221 0813 minus0548 minus2541AIC 10019510 7483933 5699395 931106 454548BIC 10027350 7491773 5707234 938233 461675AEEs refer to the average elasticity effects


we should collect long-period metro smart card data to an-alyze the travel characteristics of various people on weekdaysweekends and holidays Second due to the limitations of thedata set we did not consider the family characteristics of thestudied groups +e next step should be to design a detailedfamily survey questionnaire covering family members familyincome and car ownership for further research

Data Availability

+e metro smart card data used to support the findings ofthis study are available from the corresponding author uponrequest +e point of interest (POI) data taken from theBaidu map are available at httpsmapbaiducom

Conflicts of Interest

+e authors declare that they have no conflicts of interest

Acknowledgments

+is study was financially supported by the NationalNatural Science Foundation of China (Grant no 51878062)and the Higher Education Discipline Innovation Project111 (Grant no B20035)

Supplementary Materials

Table S1 the metro smart card data structure (Supple-mentary Materials)

References

[1] J Zhao W Deng Y Song and Y Zhu ldquoWhat influencesMetro station ridership in China Insights from NanjingrdquoCities vol 35 pp 114ndash124 2013

[2] China Association of Metros Statistics and Analysis Report ofUrban Rail Transit in 2020 China Association of MetrosBeijing China 2021 httpswwwcametorgcntjxx7647

[3] M G McNally ldquo+e four-step modelrdquo in Handbook ofTransport Modelling 8e Four-step Model D A Hensher andK J Button Eds vol 1 pp 35ndash53 Emerald Group PublishingLimited Bingley England 2007

[4] S Rasouli and H Timmermans ldquoActivity-based models oftravel demand promises progress and prospectsrdquo Interna-tional Journal on the Unity of the Sciences vol 18 no 1pp 31ndash60 2014

[5] J L Bowman and M E Ben-Akiva ldquoActivity-based disag-gregate travel demand model system with activity schedulesrdquoTransportation Research Part a Policy and Practice vol 35no 1 pp 1ndash28 2001

[6] J Gutierrez O D Cardozo and J C Garcıa-PalomaresldquoTransit ridership forecasting at station level an approachbased on distance-decay weighted regressionrdquo Journal ofTransport Geography vol 19 no 6 pp 1081ndash1092 2011

[7] O D Cardozo J C Garcıa-Palomares and J GutierrezldquoApplication of geographically weighted regression to thedirect forecasting of transit ridership at station-levelrdquo AppliedGeography vol 34 pp 548ndash558 2012

[8] J Zhao W Deng Y Song and Y Zhu ldquoAnalysis of Metroridership at station level and station-to-station level in

Nanjing an approach based on direct demand modelsrdquoTransportation vol 41 no 1 pp 133ndash155 2014

[9] M Kuby A Barranda and C Upchurch ldquoFactors influencinglight-rail station boardings in the United Statesrdquo Trans-portation Research Part A Policy and Practice vol 38 no 3pp 223ndash247 2004

[10] K Sohn and H Shim ldquoFactors generating boardings at metrostations in the Seoul metropolitan areardquo Cities vol 27 no 5pp 358ndash368 2010

[11] R Guo and Z Huang ldquoMass rapid transit ridership forecastbased on direct ridership models a case study in wuhanChinardquo Journal of Advanced Transportation vol 2020 p 19Article ID 7538508 2020

[12] F Shao Y Sui X Yu and R Sun ldquoSpatio-temporal travelpatterns of elderly people - a comparative study based onbuses usage in Qingdao Chinardquo Journal of Transport Geog-raphy vol 76 pp 178ndash190 2019

[13] S Zhang Y Yang F Zhen T Lobsang and Z Li ldquoUnder-standing the travel behaviors and activity patterns of thevulnerable population using smart card data an activityspace-based approachrdquo Journal of Transport Geographyvol 90 Article ID 102938 2021

[14] R Alsnih and D A Hensher ldquo+e mobility and accessibilityexpectations of seniors in an aging populationrdquo Trans-portation Research Part a Policy and Practice vol 37 no 10pp 903ndash916 2003

[15] D Julien L Richard L Gauvin et al ldquoTransit use and walkingas potential mediators of the association between accessibilityto services and amenities and social participation amongurban-dwelling older adults insights from the VoisiNuAgestudyrdquo Journal of Transport amp Health vol 2 no 1 pp 35ndash432015

[16] W Y Szeto L Yang R C P Wong Y C Li and S C WongldquoSpatio-temporal travel characteristics of the elderly in anageing societyrdquo Travel Behaviour and Society vol 9 pp 10ndash20 2017

[17] S Y He Y H Y Cheung and S Tao ldquoTravel mobility andsocial participation among older people in a transit me-tropolis a socio-spatial-temporal perspectiverdquo TransportationResearch Part A Policy and Practice vol 118 pp 608ndash6262018

[18] L Cheng X Chen S Yang Z Cao J De Vos and F WitloxldquoActive travel for active ageing in China the role of builtenvironmentrdquo Journal of Transport Geography vol 76pp 142ndash152 2019

[19] J Feng ldquo+e influence of built environment on travel be-havior of the elderly in urban Chinardquo Transportation ResearchPart D Transport and Environment vol 52 pp 619ndash6332017

[20] L Cheng J De Vos K Shi M Yang X Chen and F WitloxldquoDo residential location effects on travel behavior differ be-tween the elderly and younger adultsrdquo Transportation Re-search Part D Transport and Environment vol 73pp 367ndash380 2019

[21] M Winters C Voss M C Ashe K Gutteridge H McKayand J Sims-Gould ldquoWhere do they go and how do they getthere Older adultsrsquo travel behaviour in a highly walkableenvironmentrdquo Social Science amp Medicine vol 133pp 304ndash312 2015

[22] A Stahl and M Berntman ldquoFalls in the outdoor environmentamong older persons-a tool to predict accessibilityrdquo in Pro-ceedings of the 11th International Conference on Mobility andTransport for Elderly and Disabled Persons Montreal CanadaJanuary 2007


[23] Y Zhang E Yao R Zhang and H Xu ldquoAnalysis of elderlypeoplersquos travel behaviours during the morning peak hours inthe context of the free bus programme in Beijing ChinardquoJournal of Transport Geography vol 76 pp 191ndash199 2019

[24] S Rosenbloom ldquoSustainability and automobility among theelderly an international assessmentrdquo Transportation vol 28no 4 pp 375ndash408 2001

[25] L T Truong and S V C Somenahalli ldquoExploring frequencyof public transport use among older adults a study inAdelaide Australiardquo Travel Behaviour and Society vol 2no 3 pp 148ndash155 2015

[26] M Du L Cheng X Li and J Yang ldquoFactors affecting thetravel mode choice of the urban elderly in healthcare activitycomparison between core area and suburban areardquo Sus-tainable cities and society vol 52 Article ID 101868 2020

[27] A Broberg and S Sarjala ldquoSchool travel mode choice and thecharacteristics of the urban built environment the case ofHelsinki Finlandrdquo Transport Policy vol 37 pp 1ndash10 2015

[28] R Ewing W Schroeer and W Greene ldquoSchool location andstudent travel analysis of factors affecting mode choicerdquoTransportation Research Record vol 1895 no 1 pp 55ndash632004

[29] R Zhang E Yao and Z Liu ldquoSchool travel mode choice inBeijing Chinardquo Journal of Transport Geography vol 62pp 98ndash110 2017

[30] Y Liu Y Ji Z Shi and L Gao ldquo+e influence of the builtenvironment on school childrenrsquos metro ridership an ex-ploration using geographically weighted Poisson regressionmodelsrdquo Sustainability vol 10 no 12 Article ID 4684 2018

[31] J A Kelly and M Fu ldquoSustainable school commuting -understanding choices and identifying opportunitiesrdquo Jour-nal of Transport Geography vol 34 pp 221ndash230 2014

[32] S Li and P Zhao ldquo+e determinants of commuting modechoice among school children in Beijingrdquo Journal of TransportGeography vol 46 pp 112ndash121 2015

[33] K Y K Leung and B P Y Loo ldquoDeterminants of childrenrsquosactive travel to school a case study in Hong Kongrdquo TravelBehaviour and Society vol 21 pp 79ndash89 2020

[34] A Ermagun and A Samimi ldquoPromoting active transportationmodes in school tripsrdquo Transport Policy vol 37 pp 203ndash2112015

[35] S Y He and G Giuliano ldquoSchool choice understanding thetrade-off between travel distance and school qualityrdquoTransportation vol 45 no 5 pp 1475ndash1498 2018

[36] M Palm and S Farber ldquo+e role of public transit in schoolchoice and after-school activity participation among Torontohigh school studentsrdquo Travel Behaviour and Society vol 19pp 219ndash230 2020

[37] Qingdao Statistics Bureau Qingdao National Economic andSocial Development Statistics Bulletin Qingdao StatisticsBureau Qingdao China 2020 httpsqdtjqingdaogovcnn28356045n32561056n32561070200327102041515838html

[38] Ministry of Construction PRChina China Urban Con-struction Statistical Yearbook Ministry of ConstructionPRChina Beijing China 2020 httpwwwmohurdgovcnxytjtjzljsxytjgbjstjnj

[39] S Li D Lyu X Liu et al ldquo+e varying patterns of rail transitridership and their relationships with fine-scale built envi-ronment factors big data analytics from Guangzhourdquo Citiesvol 99 Article ID 102580 2020

[40] S Washington M Karlaftis F Mannering andP Anastasopoulos Statistical and Econometric Methods forTransportation Data Analysis CRC Press FL USA 2020

[41] L E Papke and J MWooldridge ldquoEconsometric methods forfractional response variables with an application to 401(k)plan participation ratesrdquo Journal of Applied Econometricsvol 11 no 6 pp 619ndash632 1996

[42] N C McDonald Y Yang S M Abbott and A N BullockldquoImpact of the safe routes to school program on walking andbiking eugene Oregon studyrdquo Transport Policy vol 29pp 243ndash248 2013

[43] K Wang and G Akar ldquoGender gap generators for bike shareridership evidence from Citi Bike system in New York CityrdquoJournal of Transport Geography vol 76 pp 1ndash9 2019

[44] L E Papke and J M Wooldridge ldquoPanel data methods forfractional response variables with an application to test passratesrdquo Journal of Econometrics vol 145 no 1-2 pp 121ndash1332008

[45] D An X Tong K Liu and E H W Chan ldquoUnderstandingthe impact of built environment on metro ridership usingopen source in Shanghairdquo Cities vol 93 pp 177ndash187 2019

[46] M-J Jun K Choi J-E Jeong K-H Kwon and H-J KimldquoLand use characteristics of subway catchment areas and theirinfluence on subway ridership in Seoulrdquo Journal of TransportGeography vol 48 pp 03ndash40 2015

[47] G Gao Z Wang X Liu Q Li W Wang and J ZhangldquoTravel behavior analysis using 2016 Qingdaorsquos householdtraffic surveys and Baidu electric map API datardquo Journal ofAdvanced Transportation vol 2019 p 18 Article ID 63830972019


the travel characteristics and differences between main-stream and other groups [12 13] However these studies arebased on bus data and do not involve the metro usage ofspecific groups In addition policy recommendation is morevaluable when exploring the travel patterns and preferencesof multiple groups of people using the metro at the sametime +erefore this paper uses metro smart card data toexplore the metro usage by adults and other groups (theelderly and students) to fill the gaps in the research contentof this direction

+e structure of this study is arranged as follows Section2 reviews related literature Section 3 describes the study areaand data Section 4 introduces the research method andSection 5 presents the research results and discussionSection 6 summarizes the conclusion and limitations

2 Literature Review

Numerous previous studies exist on the travel behavior ofthe elderly and the travel mode choices of students to go toschool However few studies have focused on the metrotravel behaviors of the elderly and students +erefore thissection will review this literature as references

21 Elderly People As problems of aging continue to in-tensify it is crucial to maintain the sense of social partici-pation and mobility of the elderly [14ndash17] As with themainstream population the built environment affects thetravel modes of the elderly [18ndash20] Studies have shown thatneighborhood design characteristics are closely related to theactive travel mode of the elderly For instance older adultswill have more active travel in highly walkable neighbor-hoods [21] At the same time uneven sidewalks pot-holesand other harsh walking environments will affect both theaccessibility and mobility of older adults [22] Since mostolder adults have retired they have more time to engage innonwork activities Compared with young people theirmobility may not have been significantly reduced [17]Meanwhile due to the Chinese cultural background andpractical needs the elderly living with their adult childrenalso have other out-of-home activities under family re-sponsibilities [19] For example compared with others olderpeople with many family members may have more bus tripsbecause they often help take children to school [23]

Previous studies have shown that Western older adultshave strong car dependence [24 25] Affected by the eco-nomic development level and social background the ac-cessibility of public transportation is an essentialdeterminant of the travel behavior of the elderly in China[19] For example a study in Nanjing showed that thenumber of bus stops and bike share stations has a positiveeffect on the travel frequency of the elderly [18] As animportant pattern of travel for the elderly public trans-portation also helps to improve the accessibility and fairnessof the elderly to deal with the unbalanced distribution ofurban public service facilities For instance compared withthe elderly in urban areas the elderly living in suburbs have alonger distance to get medical treatment and are moredependent on public transportation [26]

To further protect the well-being of the elderly manyChinese cities have adopted various preferential policies toencourage the elderly to use public transportation dis-counted fares and monthly public transportation subsidiesAlthough these policies have effectively encouraged the useof public transport the effect may be quite limited A studyin Beijing showed that free bus programs for the elderly onlyinduced less than 5 of bus trips among older adults [23]+erefore to further attract more elderly people to usepublic transportation to maintain their necessary mobilityin addition to public transportation preferential policies wealso need to consider how to increase the attractiveness ofpublic transport to the elderly from other perspectives

22 Elementary andMiddle School Students Going to schoolis the leading travel activity for elementary and middleschool students (6 to 18 years old) A large number of studieshave shown that the built environment is closely related tothe studentsrsquo trip pattern of going to school [27ndash30] Firstthe distance between home and school is the main factor+e increase in travel distance will prompt students to switchfrom active travel mode to motorized mode An Irish studyindicated that 2 km is the dividing line between motorizedand nonmotorized modes for students [31] A Chinese studyshowed that as the distance of travel increases studentswould switch from walking to riding a bicycle When thetravel distance exceeds 3 km they will use more publictransportation and private cars to go to school [32] Secondthe number of street intersections is also related to thestudentsrsquo school transportation patterns A small number ofintersections will encourage children with shorter traveldistances to use active travel mode but the increased in-tersection number will affect the comfort of the journey andreduce the propensity of students with longer travel dis-tances to use public transportation [29] +ird studentsrsquotravel patterns are closely related to the local geography andculture Due to Finnish cycling culture and developed cy-cling infrastructure 43 of students cycle to school inHelsinki [27] In Hong Kong due to safety and topo-graphical issues students rarely ride bicycles to school andtheir main travel modes are walking and school buses [33]Finally due to a long commuting distance and complicatedurban transportation system a single traffic vehicle may notmeet the commuting needs of students+e only study aboutstudents commuting by the metro shows that the density ofbus stops and docks at bike share stations positively affectsstudents who use the metro to travel [30] +is means theconnectivity of transportation also affects the travel patternof students

Many studies recommend that students use active modeto school to protect their health and ease urban trafficcongestion [29 34] However some Western and Chinesestudies have shown that the existing urban education systemis unfair Many students start to go to schools farther away toobtain higher-quality educational resources [29 35 36] it isessential for them to use safe and reliable transportationLimited by operational efficiency and inadequate supervi-sion and management the school bus system in China is not






32 Data












4 Method


yi exp β0 + 1113944n

j1βjxij + ε⎛⎝ ⎞⎠ (1)



Laixi City

Jimo District

Pingdu City

N

16Miles

128420

Jiaozhou City

Huangdao District

Chengyang District


Suburban Area







E(y|x) G(x β) (2)



1 + exp(xβ) (3)




(4)




018

016

Adults

The elderly

Students

014

012

010

008

006

004

002

0006 8 10 12 14

Time of day (hour)

Pass

enge

r rat

e (

)

16 18 20 22



Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210


(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210


(b)














6 Conclusions













Data Availability




Acknowledgments




References























































32 Data












4 Method


yi exp β0 + 1113944n

j1βjxij + ε⎛⎝ ⎞⎠ (1)



Laixi City

Jimo District

Pingdu City

N

16Miles

128420

Jiaozhou City

Huangdao District

Chengyang District


Suburban Area







E(y|x) G(x β) (2)



1 + exp(xβ) (3)




(4)




018

016

Adults

The elderly

Students

014

012

010

008

006

004

002

0006 8 10 12 14

Time of day (hour)

Pass

enge

r rat

e (

)

16 18 20 22



Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210


(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210


(b)














6 Conclusions













Data Availability




Acknowledgments




References






















































4 Method


yi exp β0 + 1113944n

j1βjxij + ε⎛⎝ ⎞⎠ (1)



Laixi City

Jimo District

Pingdu City

N

16Miles

128420

Jiaozhou City

Huangdao District

Chengyang District


Suburban Area







E(y|x) G(x β) (2)



1 + exp(xβ) (3)




(4)




018

016

Adults

The elderly

Students

014

012

010

008

006

004

002

0006 8 10 12 14

Time of day (hour)

Pass

enge

r rat

e (

)

16 18 20 22



Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210


(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210


(b)














6 Conclusions













Data Availability




Acknowledgments




References























































E(y|x) G(x β) (2)



1 + exp(xβ) (3)




(4)




018

016

Adults

The elderly

Students

014

012

010

008

006

004

002

0006 8 10 12 14

Time of day (hour)

Pass

enge

r rat

e (

)

16 18 20 22



Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210


(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210


(b)














6 Conclusions













Data Availability




Acknowledgments




References



















































Jimo District

N

Chengyang District

AdultsThe elderly

8Miles

64210


(a)

Figure 3 Continued

Jiaozhou CityN

Huangdao District

8Miles

64210


(b)














6 Conclusions













Data Availability




Acknowledgments




References






























































6 Conclusions













Data Availability




Acknowledgments




References




















































Data Availability




Acknowledgments




References



















































exploring the built environment factors in the metro that

Documents