Daily Activity Space for Various Generations in the Yogyakarta Metropolitan Area

Abstract

Two indices of activity space measurements using Euclidian distance measurements have been argued to be able to measure specific visited out-of-home activity locations closer to activity space definitions than other methods. However, the Euclidian distance does not consider any barriers or obstacles, such as the existence of public spaces (e.g., army bases, government offices and airports) or natural barriers (e.g., mountains, hills and agricultural fields that have no road infrastructure). Therefore, this study tries to fill the research gap by measuring the two indices using road network distance. Moreover, this study tries to determine whether the activity space of different generations, namely Generations (Gens) X, Y and Z, is significantly different, and whether some socio-demographic and activity pattern variables can help differentiate the activity space measurements. Using the 2019 Yogyakarta Metropolitan Area (YMA) dataset, this study confirms that measuring activity space using road network distance statistically gives different results from activity space measured using Euclidian distance. Moreover, this study confirms that the oldest generation had opposite activity space patterns in comparison to Gens Y and Z. Unlike the younger ones, the oldest generation visited out-of-home activity locations nearer to their home locations on weekdays but expanded to visit farther out-of-home locations on weekends. Trade-off mechanisms were found between weekdays and weekends, by which Gens X and Y significantly visited out-of-home activity locations farther from their home more often on weekends than on weekdays. However, all generations were observed to visit out-of-home activity locations near their out-of-home activity anchors every day, whereas the oldest tended more often to visit the activity locations farther from their out-of-home activity anchors than the younger generations on Fridays and Sundays.

1. Introduction

The decision to participate in a trip or several trips, and to use a specific travel mode, is determined as a result of the interdependency among various trips and activities on a given day. Interactions among constraints, resources and needs are the mechanisms assumed to be behind the interdependency among various trips and activities, as argued in time–space prism theory [1,2,3,4]. People retain various personal and social characteristics as part of their position in society, life stage and personalities. The uniqueness of combinations among various personal and social characteristics shapes the uniqueness of their constraints and needs, which might vary from day to day. Therefore, single socio-demographic and built environment variables are seen not to be enough to explain people’s activity–travel behaviour.

In the time–space prism perspective, the constraints are defined as capability, coupling and authority [5,6]. The constraints limit the involvement of individuals in various trips and activities, due to the allocation of physiological activities and the influence of people’s skills and inability to perform a specific activity and travel (defined as capability constraints), engagements with other people and materials (coupling constraints) and laws, regulations and agreements, either formal or informal (authority constraints). The constraints will interact with people’s needs (e.g., biological and security, social or love and self-actualisation or self-realisation needs), so that someone’s needs might be different from others’ needs due to their attitudes, social norms, personal goals and other psychological antecedents [7,8,9,10]. Individuals’ constraints such as capability, coupling and authority have been found to explain why people never do their intended activities or never take their intended travel mode [11,12,13]. The collection of various activities and travels on given days is one method of capturing the constraints [12,14,15,16].

Land use or built environment is one of the resources that can generate or limit people’s trips and activities [17,18,19,20,21]. A high degree of mixed land use can be interpreted as an effort to make a city compact [22,23,24], which provides opportunities for individuals to fulfil their needs and desires within walking distance or without private vehicle dependency [25]. In contrast, areas with a low degree of land use diversity can limit individuals’ opportunities to reach some out-of-home discretionary activities and increase the dependency on motorised modes of transport [26,27,28,29].

The movement of people from one place to another place to satisfy their needs and desires causes a change in coordinates, which can be captured in two-dimensional pictures called activity space [5,17,18,30,31]. The movements of people over space captured by activity space should be limited by people spending time at two anchors, namely the home and the workplace [17,18,30,32,33]. Around 70–74% of people’s activity time use takes place at the home anchor [34,35,36], whereas 6–8 h working commitments at the workplace might limit people’s exploration over space [17,18]. Activity space confirms the spatial behaviour or orientation, due to the limitations shaped by constraints and the opportunities offered by resources [5,17,18,31,37,38,39].

In conceptualising the activity space, people might have ellipses or lines of activity space when they have a long commitment time in their out-of-home anchor (e.g., workplaces), but might have circles of activity space when they have no working commitment time and more flexible time as homemakers [37]. Working in a place farther from one’s residential location might widen the familiar area or the potential activity space [31,38]. However, the visited activity locations or activity spaces and the familiar area will be around home, workplace and public transport nodes [39].

The standard deviational ellipse (SDE), network-based buffer (NBB), kernel density and minimum convex-hull polygons (MCP) are some examples of the method for measuring activity space [32,40]. However, those methods likely measure the potential activity space rather than the actual activity space, as the methods cannot explain the specific location of the out-of-home activities undertaken [17,18,32]. Moreover, the variations in the visited activity locations or actual activity space on different days, due to visiting non-routine activity locations, cannot be described by those methods. Those methods only show the standard deviation of the potential activity space, based on the visited activity locations that are closer to people’s familiar areas [38] than the activity space. Revealing the daily modifications of activity space can show the trade-off mechanisms of people’s spatial orientation between weekdays and weekends. In addition, the standard deviation or the dispersion measured by previous methods does not measure how far the activity locations are from people’s home locations and out-of-home anchors. Measuring whether activity space is conducted near the home or an out-of-home anchor might reveal another trade-off mechanism: whether people visit activity locations far from home but near out-of-home activity anchors on a specific day, or near home but far from the out-of-home anchors.

Susilo and Kitamura [17] and Dharmowijoyo et al. [5,18] offered two indices to measure activity space. These two indices measure the exact visited activity locations instead of potential activity space using Euclidian distances. However, the two indices do not measure the variance of visited activity locations, and, thus, overlook the frequency information included in kernel density and NBB. In contrast to SDE, kernel densities, NBB and MCP, the two indices can produce different values on different days; thus, day-to-day variation and trade-off mechanisms among different days can be evaluated. Moreover, they measure the diversity of the visited activity locations (either more dispersed or visiting more activity locations on a day), defined as Ic, and how far the distributions of out-of-home activity locations are from home (Ih). Therefore, whether people visit the out-of-home discretionary activity locations either near home or out-of-home anchors can be analysed. However, this method has not been integrated into Geographic Information System (GIS) software as kernel density, NBB and MCP have, which makes the method difficult to display on the GIS platform.

The Euclidian distance in Ic and Ih measurements does not consider the availability of road or transport networks or the existence of natural barriers or obstacles. The natural barriers (e.g., a hill or a mountain), or public facilities (e.g., airports, army or air force bases, parks, etc.) can make the actual distance using road network distance measurement farther than the Euclidian distance. This study will fill this research gap by improving the methodological part by applying road network distance. This study will investigate Ic and Ih measurements using the shortest path measurements of road networks to measure the distance. This study will offer a combination of two indices, Ic and Ih, with minimum spanning trees using network distance (instead of Euclidian distance) for measuring the distance between activity locations and the centre activity locations, and between the centre activity locations and home locations. Filling the research gap will answer the research question regarding whether the activity space measured using road network distance will be different from activity space measurements using Euclidian distance. This study will also include GIS software for measuring the activity location distance in the Ic and Ih analysis. However, this study only visualises activity space in two-dimensional pictures, which overlooks the third dimension, the time allocation at the locations, as demonstrated by Kwan et al. [41] and Neutens et al. [2]. Including the time dimension in the geo-visualisation of the activity space may describe in which area people will allocate more time to stay, either for in-home or out-of-home activities or for working/studying or out-of-home leisure/maintenance. This is the limitation of the study. This study does not aim to visualise the time dimension in a geographical manner, but to transform the activity space to a value/values that indicate whether people visited out-of-home activity locations near or far from their home (Ih), and have more diversely (more dispersed or more in number) visited out-of-home activity locations (Ic). It is argued that the spatial orientation is highly shaped by people’s activity–travel time-use within the time–space prism perspective. The spatial orientation described by the activity space is a product of their activity–travel commitments throughout the space limited by people’s constraints but expanded by their needs and resources.

Investigations of the two indices have been carried out for workers, students and non-workers [17], individuals [42], and young workers riding motorbikes [43]. However, the tests for various generation groups, such as Generations (Gens) X, Y and Z, have not been examined before. This is another research gap to be filled. A generation is a group of individuals who were born and lived at the same time and have common knowledge and experiences that impact their thoughts, attitudes, values, beliefs and behaviours [44], which correlate to different activity–travel participation on a daily basis. Gen Z and Y are groups of individuals born in 1997–2012 and 1981–1996, respectively, whereas Gen X was born in 1965–1980 [45,46]. These three groups are predicted to be actors who will push to maximise economic growth during the peak period of Indonesia’s demographic bonus [47]. Closing the research gap can answer the research question regarding whether different generations, due to different knowledges, experiences, change of environment, structural conditions and life stage significantly correlate with spatial behaviour and activity–travel patterns. Generational differences in travel activities and behaviour certainly lead to differences in the daily variability of travel–activity patterns and the activity spaces formed. Moreover, filling the research gap can also provide understanding whether different generations in developing countries have similar or different activity–travel and spatial behaviour from those in developed countries. In developed countries, Gen Y is found to use effective modes of transportation more often, make fewer trips, reach smaller distances and participate in more out-of-home and virtual activities than Gen X [48]. Gen Z tends to be more familiar with ICT, perform more online activities, choose transportation that allows socialising during travel, visit activity locations that are close to each other and enjoy more walking, running and cycling [49].

To close the research gap mentioned above, the first objective of this study is to examine the generational differences in activity space measurements using road network and Euclidian distances. The second objective of the study is to investigate whether Gens X, Y and Z have different activity space patterns. In addition to generation differences, the different spatial behaviour can be shaped by activity–travel patterns or commitments to various activities and travel. Therefore, to answer the research question, the study was expanded to include a third objective: to unravel the effects of different socio-demographic and time-use activity–travel patterns on activity space differences. The research used the 2019 Yogyakarta Metropolitan Area (YMA) dataset as the representative of activity space patterns in Indonesia and other developing countries. As the author acknowledges, research in developing countries related to spatial behaviour, particularly the different activity spaces of different generations, is very rare. The study hypothesises that activity space measurements using road network distance are significantly different from activity space measurements using Euclidian distance. Gen X is hypothesised to have significantly different activity space patterns than the younger generations. The older generation is hypothesised to have a higher Ih and Ic, whereas the youngest is hypothesised to have more compact activity spaces near their home locations and out-of-home activity anchors. For the third objective, it is hypothesised that the manner in which people schedule their activity patterns for in-home activities, working and studying significantly affects each generation’s activity space. In addition, some socio-demographic variables, such as employment and income, are hypothesised to differentiate each generation’s activity space.

The dataset explanation and the methodology of the activity space measurements will be discussed in Section 2. Results will be examined in Section 3, whereas the discussion and conclusion will be in Section 4.

2. Materials and Methods

2.1. The Studied Area

The location of the study is the Yogyakarta Metropolitan Area (YMA) or Kartamantul area. Yogyakarta city is the capital of the Special Region of Yogyakarta, which consolidates to form a metropolitan area with its surrounding areas as part of the Sleman and Bantul Regencies [50]. In addition to Yogyakarta, six districts in the Sleman Regency (Depok, some areas in Ngaglik, Mlati, Godean, Gamping and Ngemplak) and three districts in the Bantul Regency (Kasihan, Sewon and Banguntapan) are included in the YMA (see Figure 1 below) [51]. The YMA faces urbanisation and congestion simultaneously, as do other agglomeration areas [52]. Its growth and development as an educational centre, and its many renowned attractive tourist attractions are increasingly encouraging people to settle to work, study or develop business. Therefore, during the data collection in 2019, the economic growth of YMA was 6.60%, or above the national economic growth (5.02%) [53].

Urban sprawl problems are arising as a result of the economic growth and rapid development, in suburban areas, of a built environment that is not balanced by the availability of service facilities [54]. Therefore, it is plausible that high economic growth and urban sprawl are creating congestion, a common problem in many cities in Indonesia, and in turn, air pollution [55]. By 2020, private motorcycles and cars contributed to 92.4% of travel mode share compared to 3.6 and 4% of public transport and ride-sourcing market share, respectively [56,57]. Bus networks continue to try to find ways to increase the demand for their services and expand their networks, which are overly concentrated near the city centre. The rapid transit services are still centred in the city centre, whereas the accessibility of the service in suburban and greater areas is low [58].

2.2. The 2019 YMA Dataset

YMA data from 2019 were used for this study. This study collected the activity diaries of 400 individuals from Thursday to Sunday. The study was part of the study of the effects of time–space prism variables on electric motorcycle adoption. In addition to the activity diary survey, choice experiment, socio-demographic and travel satisfaction information were collected. Therefore, to achieve its objectives, this survey was not designed as a household survey, but as an individual survey. Therefore, not all household members were included in the survey. The respondents included in this survey were treated as an individual response. No other household member effects can be revealed from this survey. The survey applied the probability proportional to size (PPS) method, in which areas with higher populations were sampled relatively more than areas with lower populations. The survey represented 0.03% of the population. The sample size is higher than the 2013 Bandung Metropolitan Area dataset, which only represented 0.02% of the population, as described by Dharmowijoyo et al. [36].

The survey implementation was inspired by Dharmowijoyo et al. [36], in which a member of the survey team was present with the respondents when filling out the survey, particularly the activity diaries. Each surveyor had a responsibility to help around 20 respondents. Community leaders in the studied areas helped the surveyors find the 20 respondents, and assisted the surveyors by having a personal engagement with the potential respondents.

Many studies have revealed the trade-off mechanisms between weekdays and weekends. Therefore, the activity diary was collected for two representative weekdays and two weekend days. Thursday and Friday were chosen for the weekdays, whereas Saturday and Sunday were for weekends. In day-to-day studies, Friday tends to be a pre-weekend day, in which the activity–travel patterns tend to be more varied than on other weekdays [4,59,60] and the activity space tends to be higher than other weekdays, but lower than the weekends [5,17,18]. Even though each day is different [4,5,17,61], Thursday is representative of a common weekday. Compared to Friday, Thursday has more similarity, in the activity–travel undertaken, with Monday, Tuesday and Wednesday [4,59,60].

The activity diary survey was a survey which recorded people’s in-home and out-of-home activity patterns for 24 h. As suggested by Dharmowijoyo et al. [36], to simplify the survey, people’s activity time was divided into 96 time slices, with each time slice equal to a 15 min interval. This method is easier to conduct for the respondents and easier to be validated by the surveyors. However, this method overlooks some activities that were undertaken in under 15 min, such as travelling to a nearby store that requires only a 5 min walk. Moreover, multitasking activities were ignored in this method.

In the survey implementation, 25 activity classifications were used for in-home or out-of-home activities, personal activities (e.g., sleeping, taking a shower) or activities that involved other persons, household chores or working and studying, fixed activities (e.g., sleeping, working and studying) or flexible activities, and offline or online activities. Each respondent was requested to enter the type of activity in each time slice. Surveyors came to visit the respondents each day to ensure the correctness of the data input. Using the 96 time slices, it was easier for the survey team to check and validate the respondents’ responses.

In addition to the activity diaries, the location coordinates of the out-of-home activities undertaken were also recorded using a global positioning system. The respondents shared the address of the location of the activities undertaken at out-of-home locations and at their home location. Then, the surveyors assisted the respondents to find the coordinates of the locations. Checking and cross-checking of the location coordinates were performed several times by the respondents and the surveyors to ensure the accuracy of the recorded location coordinates. In total, 369 respondents succeeded in filling out the coordinates for four consecutive days. However, after screening the activity location coordinates, only 343 respondents, with 1372 trips, were used for further analysis. The profile of the 343 respondents is shown in

Table 1. Profile of all respondents.

Characteristic	%	Average
Females *	47.52
Gen Z	12.54
Gen Y	60.93
Worker *	67.05
Student *	14.87
Married *	53.35
Having a diploma, bachelor’s degree or above *	65.31
Low income (<IDR 3 million/month) *	44.90
Middle income (IDR 3–6 million/month) *	40.23
Having access to a car	38.78
Single and couples *	12.54
Having 3–4 household members *	64.43
Percentage of car use		12.44
Percentage of motorcycle use		57.41
Percentage of public transportation use		0.67
Percentage of non-motorised transport use		10.72
Percentage of ride-sourcing use		0.90
The average number of trips		3.79
The average trip chain		1.88
The average travel time per day (minutes)		99.63
The average time spent on in-home activities per day (minutes)		920.05
The average time spent on out-of-home activities per day (minutes)		345.62
The average time spent on in-home online activities per day (minutes)		52.33
The average time spent on out-of-home online activities per day (minutes)		5.05
The average time spent on online shopping per day (minutes)		17.32

* Males are 52.48%, Gen X are 26.53%, non-workers are 18.08%, not married is 46.65%, high income (>IDR 6 million/ month) is 14.87%, not having access to a car is 61.22% and having ≥5 household members is 23.03%.

.

In the analysis, the activity classifications were condensed into four types: mandatory, leisure, maintenance and online activities. Each type of activity was divided into two subsections: activities performed at home or out-of-home. The survey also included different modes of travel, namely, car, motorcycle, public transport and ride-sourcing. Ride-sourcing, here, is defined as on-demand transport using an application, such as Uber, Grab and Gojek, which emerged since 2010 and now serve around 50 cities in Indonesia [62], with revenue of almost USD 4.369 million [63]. Unlike conventional public transport, ride-sourcing combines the flexibility of private vehicles that can be used anytime, through an application installed on people’s devices or mobile phone, with users benefiting from not owning the vehicles, not needing to find parking nor paying the yearly vehicle owner tax [62].

Table 2. Activity classifications.

Activities Used in the Research	Activity Classifications in the Survey
In-home (IH) mandatory	Sleeping, personal activities and eating at home
IH and out-of-home (OH) working and studying	Working and studying either at home or out-of-home
Picking up/dropping off	Picking up and dropping off children and other household members
IH and OH leisure	Watching TV/listening to radio or music without internet connections, reading newspapers/magazines/comics, relaxing or daydreaming in-home and out-of-home OH leisure also includes going to cinemas/parks/playgrounds, going to recreational places, sightseeing and shopping
IH and OH socialising	Talking/texting with household members/relatives/colleagues/friends either using phone/internet connections or not, visiting/receiving friends/relatives, meeting with friends/relatives, including religious gatherings and volunteering/political activities in-home or out-of-home
IH maintenance	Household activities and in-home babysitting
Grocery shopping	Going to the grocery store
Other maintenance	Going to the bank/post office/health centre and out-of-home babysitting
Sports activities	Doing sports activities out of the home. Sport is not found at home
IH and OH online activities	Social media activities, playing games online, watching movies from internet platforms, browsing, reading online news and any online activities related to leisure

Source: Dharmowijoyo et al. [10].

exhibits the activity classifications.

2.3. Two Indices with Distance-Based Road Network

Two indices are used in the activity space measurement. The first index, Ic, is the total squared distance between all out-of-home activity locations visited and the centre of all out-of-home activity locations visited. The c subscript refers to the diversity level of out-of-home activity locations from their centre of out-of-home activity locations (c). The centre of out-of-home activity locations can be defined as out-of-home anchors. The second index, Ih, is the squared distance between the home location or home anchors and the centre of all out-of-home activity locations. The subscript h indicates home, which shows how far the distributions of out-of-home activity locations are from home locations or home anchors. An example is shown in Figure 2. If an individual only visits an out-of-home activity location, the Ic is 0 because the centre of the out-of-home activity location coincides with the out-of-home activity location [5]. Figure 3 illustrates the step-by-step process for estimating the Ih and Ic, including the formula.

Susilo and Kitamura [17] and Dharmowijoyo et al. [18] warned that the Ic cannot be interpreted solely by the spread or the concentration level of out-of-home activity locations. A value of 20 can be a result of five out-of-home activity locations that are each 2 km away from the centre. Another value of 20 can be gained for two locations, each at a 3.16 km distance. This is illustrated in Figure 4. Therefore, the Ic will be interpreted as the diversity level of the out-of-home activity location. The diversity level means the diversity in distances of the visited activity locations, a higher number of out-of-home activity locations or the degree to which the out-of-home activity locations are spread out [17,18].

More detailed illustrations of estimation results of Ic and Ih are shown in Figure 5. In Figure 5, high Ih is shown when the distance between home anchors and out-of-home anchors are very far, as illustrated by working males from Gens Y and Z groups on Friday. On the other hand, low Ih is indicated with a short distance between home and out-of-home anchors, as described by a working male from the Gen Y group on Sunday. Respondents with low Ic are those with high Ih but who visit out-of-home activity locations very near the out-of-home anchors, as shown by a working male from the Gen Y group on Friday. However, those with high Ic are those who have a shorter distance between home and out-of-home anchors, as shown by a working male from Gen X group on Friday and Sunday.

As road network distance will be used in this study instead of the Euclidian distance, GIS is used as a tool. GIS has long been used as an activity-based micro-simulation tool [15]. Using the help of Esri ArcGIS 10.8 software (California, U.S.), which provides a network analysis extension, the coordinates of the house, out-of-home activities, the centre of the locations of out-of-home activities and the spatial data of the road network in the research area were sourced from the Humanitarian OpenStreetMap (OSM) Team for the Special Region of Yogyakarta, Indonesia. OSM was chosen because the main subject of this study focused on road networks [64]. Additionally, OSM is under an open database license, allowing users to utilise, duplicate, create derivatives, compile, reproduce and communicate to other users publicly from the database they own, as long as they maintain the original attributes and openness [65]. The first stage of the process was to create a road network model in GIS. The second stage was to measure the shortest distance, using a new route analysis, between the out-of-home activity locations and the centre of all out-of-home activity locations, and between the centre of all out-of-home activity locations and the home location. The assumptions did not consider peoples’ route choice perceptions. This was a weakness of this research.

3. Results

3.1. Sample Description

The characteristics of the sample are presented in

Table 3. Sample characteristics.

Characteristic	X	Y	Z
	(n = 91)	(n = 209)	(n = 43)
Gender (%)
Woman	39.56	49.76	53.49
Man	60.44	50.24	46.51
Activity Status (%)
Worker	61.54	78.95	20.93
Student	0.00	9.57	72.09
Non-worker	38.46	11.48	6.98
Married (%)
Yes	90.11	48.33	0.00
No	9.89	51.67	100.00
Having a diploma or above (%)
Yes	69.23	75.60	6.98
No	30.77	24.40	93.02
Share of household income (%)
Low (<IDR 3 million/ month)	32.97	49.28	48.84
Middle (IDR 3–6 million/month)	51.65	37.32	30.23
High (>IDR 6 million/ month)	15.38	13.40	20.93
Access to Car (%)
Yes	52.75	33.01	37.21
No	47.25	66.99	62.79
Household (%)
1–2	9.89	15.31	4.65
3–4	65.93	65.55	55.81
≥5	24.18	19.14	39.53
Percentage of car use	21.73	10.27	3.32
Percentage of motorcycle use	43.07	60.20	74.19
Percentage of public transport use	0.74	0.65	0.58
Percentage of non-motorised transport use	19.49	8.46	3.12
Percentage of ride-sourcing use	0.55	0.36	4.26
The average number of trips	4.25	3.56	3.97
The average travel chain	2.00	1.76	2.22
The average travel time per day (minutes)	115.10	93.09	98.72
The average time spent on working and studying (minutes)	150.41	222.15	229.19
The average time spent on out-of-home socialising activities (minutes)	50.89	51.55	74.13
The average time spent on grocery shopping (minutes)	13.85	10.23	7.76
The average time spent on out-of-home other maintenance (minutes)	22.05	27.42	38.55
The average time spent on out-of-home online activities (minutes)	0.99	5.36	12.12
The average time spent on online shopping (minutes)	25.59	16.22	5.15
The average time spent on in-home online activities (minutes)	19.95	59.23	87.30
The average time spent on in-home activities (minutes)	984.15	911.39	826.48

. The distribution of Gen Y dominates, with a total of 209 respondents, followed by Gens X and Z, with 91 and 43 respondents, respectively. Gen X males were over-represented in the sample, with around 60.44% of the total, whereas the distribution between males and females seems similar to the national statistics [47]. Ninety per cent of Gen X were married in this sample, whereas all Gen Z were unmarried. An equal distribution of marital status was found in Gen Y. Most older respondents (Gens X and Y) were workers, as those people are of productive ages. However, around 72.09% of Gen Z were students. Gens X and Y contained 69.23 and 75.60% of respondents with bachelor’s degrees or above, respectively, whereas 93.02% of Gen Z did not have a degree.

Regarding travel characteristics, Gen X tended to have more access to cars—more than half of the respondents from this group reported having access to a car. However, only 33.01 and 37.21% of Gens Y and Z had access to cars, respectively. Motorcycles were more likely to be the dominant travel mode used in the studied area by all travellers. Motorcycles are still considered the most effective travel mode because they have high flexibility when passing through congested and narrow lanes, such as those found in urban areas in Indonesia [66,67,68]. Gens Y and Z tended to use motorcycles more often, on average, whereas Gen X was more likely to use cars more often, on average. More income security and older age might be the reason why, on average, Gen X used cars more often. Gen X also tended to use public transportation more often on average (around 19.49%), whereas Gens Y and Z, on average, tended to take public transportation for around 8.46 and 3.12% of their total travel time in a day, respectively. Interestingly, Gen Z used ride-sourcing more often (around 4.26% of their total travel time). Gen Z is more likely to have more exposure to digital applications, including ride-sourcing. As has been argued by Mokhtarian and Chen [69], Chikaraishi et al. [70] and Ahmed and Stopher [71], the travel time expenditures of people of different ages or life stages and other socio-demographic variables are different. However, unlike in other studies from developed countries [72,73,74], younger travellers (Gens Y and Z) tended to have shorter travel times than older ones.

Regarding activity patterns, Gen Z tended to spend more time working, studying and in out-of-home discretionary activities than the older Gens, including out-of-home online activities. Gen Z also tended to spend less time on in-home activities but more time on in-home online activities than Gens X and Y. A digital lifestyle with more exposure to online games and other digital applications might explain why Gen Z tended to spend more time on online activities.

3.2. Exploring the Difference in Activity Space Size

The characteristics of the Ih and Ic are presented in

Table 4. Characteristics of Ih and Ic.

Distance	Characteristic	X	Y	Z	All
		(n = 91)	(n = 209)	(n = 43)	(n = 343)
Euclidian	Ih weekdays	59.13 ** (163.40)	116.12 *** (281.21)	196.34 (591.31)	111.06 *** (316.59)
	Ih weekends	160.76 (677.27)	131.96 (413.63)	127.37 (308.50)	139.03 (487.17)
	Ih	109.95 (494.59)	124.04 (353.55)	161.86 (471.49)	125.04 (410.92)
	Ic weekdays	57.28 (184.21)	23.77 (68.20)	15.69 (36.14)	31.65 (110.47)
	Ic weekends	103.83 (311.88)	58.81 (343.82)	62.43 (236.37)	71.21 (324.01)
	Ic	80.55 (256.83)	41.29 (248.33)	39.06 (170.20)	51.43 (242.78)
Road network distance	Ih weekdays	103.08) ** (247.65)	228.29 *** (556.21)	331.55 (999.01)	208.02 *** (577.47)
	Ih weekends	315.81 (1402.98)	259.04 (866.77)	241.68 (620.78)	271.93 (1012.92)
	Ih	209.45 (1011.63)	243.67 (727.96)	286.62 (830.47)	239.97 (824.78)
	Ic weekdays	119.95 (420.73)	56.49 (187.43)	33.33 (77.96)	70.42 (264.30)
	Ic weekends	212.68 (673.40)	139.20 (1021.78)	111.07 (442.56)	155.17 (883.72)
	Ic	166.31 (562.61)	97.85 (735.29)	72.20 (319.22)	112.80 (653.37)

** The mean values of weekdays and weekends are statistically different, with p < 0.01. *** The mean values of weekdays and weekends are statistically different, with p < 0.001.

. The Kruskal–Wallis test was used to check the differences in activity space size, because the Ih and Ic were not normally distributed. Overall, the Ih on weekend days was significantly higher than on weekdays, when using both Euclidian and road network distances. The statistical difference in measurement shows significant results. However, the Ic on both weekdays and weekend days was statistically the same. It can be argued that people in the YMA tended to visit with similar diversity on weekdays and weekend days.

In regard to different life stages, Gens X and Y tended to have a lower Ih and a higher Ic than Gen Z. Similar to the overall measurement, Gens X and Y showed a higher Ih on weekend days than on weekdays using both methods. However, Gen Z explored a similar area on weekdays and weekends. Concerning the Ic, all groups visited out-of-home activity locations with a similar degree of spread between weekdays and weekend days.

Table 5. Summary of Mann–Whitney test results.

Activity Space	Group	Mean Rank	Sum of	Z	p
Ih	Euclidian	1255.20	1,722,134.50	−7.757	0.000 ***
	Road network	1489.80	2,044,005.50
Ic	Euclidian	1326.96	1,820,585.00	−3.153	0.002 **
	Road network	1418.04	1,945,555.00

** p < 0.01, *** p < 0.001.

shows that the Ih and Ic measurements using road network distance were found to be different from the Euclidian distance measurements. In the earlier hypothesis, the author assumed that the two indices’ measurements using Euclidian distance might have statistical differences with the road network distance in urban areas. However, this result argues that the network distance should provide better distance measurements even when applied in urban areas.

3.3. Day-to-Day Variability of the Activity Space per Generation

In general, Gen X tended to spend time at out-of-home activity locations near their home on weekdays, then expanded to farther locations on weekends, as also found in Figure 6. On the other hand, Gens Y and Z tended to have a more stable Ih during weekdays and weekends. Concerning the Ic, Gen X tended to perform more spread-out out-of-home activities, whereas Gens Y and Z tended to spend more time at out-of-home activity locations in more compact areas or near their out-of-home anchors. Meanwhile, Ih and Ic fluctuations broken down into socio-demographic and activity patterns are shown in Figure 7.

Females in Gens Z and Y tended to create a higher Ih on weekdays than males, whereas females in Gen X tended to have opposite patterns, or a lower Ih on weekdays, than males. Females in Gen Z tended to explore more spread-out out-of-home locations, particularly on weekdays, than their male counterparts and Gens Y and X. Females in Gen X tended to have the lowest Ih. On the other hand, they tended to visit out-of-home activity locations that were more diverse or not concentrated near their out-of-home anchors. Females in Gen Z tended to spend time at out-of-home locations that were near their out-of-home anchors or less diverse.

Workers and students in Gens Y and Z tended to have a higher Ih than their non-worker counterparts on weekdays and Sundays. On the other hand, non-workers in Gen X tended to visit farther out-of-home activity locations than their worker and student counterparts. As expected, a higher Ih tended to be traded off with a lower Ic. Workers and students in Gen Z who tended to have a higher Ih on weekdays and Sundays tended to have a lower Ic on the same days than non-workers in Gen Z. In contrast, workers and students in Gen X who tended to have a lower Ih tended to visit more diverse out-of-home activity locations which were farther from their out-of-home anchors or more out-of-home activity locations than non-workers in Gen X.

According to the income data, low-income Gen X members tended to visit out-of-home activity locations farther from their homes on weekdays but nearer their homes on Sundays than their high- and middle-income counterparts. On the contrary, low-income members of Gens Y and Z tended to visit out-of-home activity locations near their homes on weekdays, but farther from their homes on weekends. Regarding car ownership, car owners seemed to visit out-of-home activity locations near their homes, particularly on weekdays. However, non-car owners, who may use motorcycles more often, tended to visit out-of-home activity locations farther from their homes, particularly on weekdays. The Ic patterns seemed easily explained by income and car ownership data.

In regard to the activity patterns, for all generations, those who tended to spend less time socialising tended to have a higher Ih, whereas those who spent more time socialising tended to visit out-of-home activity locations near their home. Gen Y and X members who spent more time socialising tended to visit out-of-home locations farther from their home anchors or more out-of-home locations than those who spent a shorter time socialising. No Ic patterns could be drawn from Gen X broken down into different socialising patterns. Online activity patterns seemed to shape activity space patterns that were different from socialising activity time patterns. Gen X and Y members who had a longer online activity commitment time tended to visit out-of-home activity locations farther from their homes in a compact area or near their out-of-home activity anchors. However, Gen Z members who spent less time on online activities tended to have opportunities to explore out-of-home activity locations farther from their homes and out-of-home anchors.

This method likely captures the trade-offs between Ih and Ic and weekdays and weekends. Moreover, there is a tendency to have different activity space patterns between Gen Z, Gen X and Gen Y.

3.4. The Differences in Activity Space per Generation Broken down into Socio-Demographic and Activity-Pattern Variables

After we acknowledge the possibility of having different activity space patterns among generations broken down into several socio-demographic and activity pattern variables,

Table 6. Average activity space according to socio-demographic variables for each generation and the Kruskal–Wallis test between generations.

Variable	Category	Generation	Ih	Kruskal–Wallis H	p-Value	Ic	Kruskal–Wallis H	p-Value
Gender	Woman	X	190.31	18.283	0.000 ***	144.50	3.623	0.163
		Y	248.98			50.33
		Z	406.07			84.64
	Man	X	221.97	4.806	0.090 +	180.59	13.977	0.001 **
		Y	238.40			144.91
		Z	149.25			57.89
Employment status	Worker and student	X	172.57	3.462	0.177	179.40	17.344	0.000 ***
		Y	262.34			107.90
		Z	282.70			72.20
	Non-worker	X	268.45	5.716	0.057 +	145.38	7.421	0.024 *
		Y	99.70			20.33
		Z	338.82			72.19
Household income	Low (<IDR 3 million/month)	X	163.27	3.635	0.162	209.15	14.482	0.001 **
		Y	248.37			115.26
		Z	175.95			56.11
	Middle (IDR 3–6 million/month)	X	171.47	19.598	0.000 ***	113.88	2.248	0.325
		Y	245.44			69.33
		Z	384.82			47.56
	High (>IDR 6 million/month)	X	435.91	1.068	0.586	250.56	10.671	0.005 **
		Y	221.41			113.25
		Z	402.99			145.33
Access to car	Yes	X	277.71	6.910	0.032 *	126.18	4.778	0.092 +
		Y	207.81			142.23
		Z	104.22			38.52
	No	X	133.24	16.342	0.000 ***	211.12	13.246	0.001 **
		Y	261.34			75.97
		Z	394.71			92.15
Day	Thursday	X	93.12	6.654	0.036 *	100.56	3.910	0.142
		Y	214.89			54.35
		Z	325.79			33.78
	Friday	X	113.05	5.270	0.072 +	139.34	6.668	0.036 *
		Y	241.69			58.63
		Z	337.32			32.87
	Saturday	X	178.42	7.768	0.021 *	98.41	2.071	0.355
		Y	270.53			95.78
		Z	208.65			75.89
	Sunday	X	453.20	9.801	0.007 **	326.95	5.113	0.078 +
		Y	247.55			182.62
		Z	274.72			146.25
Working and studying commitments	Working commitments	X	172.57	5.123	0.077 +	179.40	18.716	0.000 ***
		Y	272.96			112.25
		Z	152.29			42.70
	Studying commitments	Y	174.72	7.819	0.005 **	72.06	0.004	0.948
		Z	320.56			80.76
Out-of-home socialising commitments	<1 h	X	204.81	22.851	0.000 ***	133.93	8.594	0.014 *
		Y	291.98			94.16
		Z	311.07			75.71
	≥1 h	X	218.00	1.453	0.483	226.02	5.207	0.074 +
		Y	126.44			106.79
		Z	255.73			67.77
In-home activities	<14 h	X	346.38	6.115	0.047 *	214.65	22.589	0.000 ***
		Y	340.07			118.25
		Z	237.43			88.20
	≥14 h	X	153.94	9.198	0.010 *	146.72	12.945	0.002 **
		Y	162.94			80.76
		Z	351.75			51.01
Online activities	<1 h	X	189.23	6.624	0.036 *	144.07	13.792	0.001 **
		Y	231.67			107.75
		Z	401.89			118.03
	≥1 h	X	256.75	11.100	0.004 **	218.34	3.156	0.206
		Y	256.74			87.05
		Z	218.31			45.04

*** means p-value < 0.001, ** means p-value < 0.01, * means p-value < 0.05 and ⁺ means p-value < 0.1.

investigates whether the resulting variables are different between generations. In general, socio-demographic variables can differentiate either the Ih or the Ic of people from different generations. However, the in-home, out-of-home and online activity patterns of people from different generations were found to be different from each other. Therefore, those activity patterns were found to statistically distinguish people’s Ic and Ih from different generations.

Figure 6 shows whether each generation’s activity space on one day is significantly different or the same as on other days. However, Figure 6 does not show whether each generation’s daily activity space is significantly different or the same as other generations. Table 6 illustrates that each generation’s Ih was significantly different from that of other generations. It can be argued that Gen X visited out-of-home activity locations nearer their home from Thursday to Saturday, whereas Gens Y and Z had an opposite pattern. Gen X traded off to visit out-of-home activity locations farther from their home on Sunday than on weekdays and Saturday, whereas Gens Y and Z traded off to have a lower Ih on Sunday than on weekdays and Saturdays. In regard to the Ic, each generation had significantly similar Ic patterns on Thursday and Saturday. However, Gen X was found to have more diverse activity spaces—either their locations were more spread out, or they visited more activity locations—on Friday and Sunday than other generations. Gen X’s space exploration was the farthest and the most spread out on Sunday.

Females from different generations had significantly different Ih but similar Ic patterns. In contrast, males from different generations had significantly similar Ih patterns but different Ic patterns. Gen X males visited out-of-home activity patterns in a more spread-out manner, whereas Gen Z males performed more compact out-of-home activity location patterns.

Employment status could differentiate the Ic patterns of people from different generations, but not the Ih. Workers and students from all generations had wider and more diverse out-of-home activity location patterns than non-workers. Workers, students and non-workers in Gen X had the widest and most diverse out-of-home activity location patterns. However, workers and students in Gen Y had a higher Ic than workers and students in Gen Z, whereas non-workers in Gen Y showed the lowest Ic, even compared to the non-workers in Gen Z.

The Ic of people from different generations were differentiated by low or high income. On the other hand, middle income could not differentiate people’s Ic but could categorise the Ih of people from different generations. Low- and high-income Gen X members were found to have the highest Ic. However, high-income Gen Y members had commitments in more compact out-of-home activity locations than high-income Gen Z members. Car availability seemed to determine whether each generation reached farther out-of-home activity locations. Gen X members with cars visited the farthest out-of-home activity locations from their homes compared to Gen Y and Z members with cars. However, Gen X members without cars had the lowest Ih. In trading off with visiting out-of-home activity locations near their home, Gen X members with no cars had the highest Ic.

In-home and online activity patterns are found to significantly differentiate the activity spaces of people from different generations. Gen X members who undertook a shorter in-home activity time (below 14 h/day) were found to have a higher Ih and Ic than Gens Y and Z. This pattern was similar when people spent a long time on online activities. Gen X members who spent a lot of time on online activities seemed to have enough time to visit out-of-home activity locations farther from home and more spread out from their out-of-home activity anchors than their Gen Y and Z counterparts who had similar online activity patterns. However, Gen X tried to trade off their out-of-home activities near their home with more diverse locations (either more spread out from their out-of-home activity anchors or more out-of-home activity locations) when they had a long in-home activity commitment time and short online and out-of-home socialising activity times. In contrast to their Gen X counterparts, Gen Y and Z members who had a long in-home activity commitment time and short online and out-of-home socialising activity times spent out-of-home activities at locations farther from their homes but near their out-of-home activity anchors. It seems that those who spent a lot of time on in-home activities might be similar to those who spent a small amount of time on online activities.

4. Discussion and Conclusions

Using the 2019 Yogyakarta Metropolitan Area (YMA) dataset, even though the YMA is an urban area with a highly road-oriented development, this study confirms that activity space with network distance measurements provides different results than activity space measured by Euclidian distance. Activity space here was measured using two indices, Ih and Ic [5,17,18]. As has been hypothesised, for planning purposes, because people in the YMA and Indonesia, in general, take private vehicles more often, the Ih and Ic calculated with network distance measurements offer better and more valid measurements than the Ih and Ic calculated by Euclidian distance. Ih and Ic measurements using road networks offer better activity space estimation because the measurements consider barriers or obstacles, such as the existence of public spaces (e.g., army bases, government offices and airports) or natural barriers (e.g., mountains, hills and agricultural fields) that have no road infrastructures. However, the Euclidian distance ignores the presence of those barriers and obstacles. However, Ih and Ic calculated using Euclidian distance can still be used for activity space orientation purposes.

Moreover, this study confirms that the oldest generation had opposite activity space patterns to Gens Y and Z. Unlike developed countries [48], in a developing country such as Indonesia, Gen Y members were not found to have fewer trips and visit closer out-of-home activity locations than Gen X members. This study highlights that Gen Z was observed to visit out-of-home activity locations near their out-of-home anchors, but not necessarily to visit more compact out-of-home activity locations, as argued by Larkin et al. [48] using developed country data. In contrast to the younger generations, Gen X visited out-of-home activity locations nearer to their home locations on weekdays but expanded to visit farther out-of-home locations on weekends. The oldest generation, or Gen X, explored the space farther on weekends than the younger generations. Due to their life stage [75], the oldest generation might have experiences exploring more space than others who are younger. Moreover, performing more online or virtual activities, either in-home or out-of-home, aligning with results from developed countries [48], may make the younger generation have more limited time to explore the space; thus, they visit activity locations nearer their home than the oldest generation. To support this argument, Table 3 also showed that Gen X tended to travel longer than the younger generations, whereas the younger ones tended to spend more time in the out-of-home activity locations for out-of-home socialising, maintenance and working and studying. Spending longer time at out-of-home activity locations may cause Gen Y and Z to visit out-of-home activity locations near their home.

It is important to emphasise that Gen Y and Z did not always spend more time on online activities than Gen X. Not observed by research in developed countries [45,48,49], Gen X was found to undertake more time for activities related to shopping online, due to their commitments on more household chores than Gen Y and Z. Therefore, more detailed data collection on a different type of virtual activities might help to provide a better understanding of older and younger generations’ activity–travel behaviour. Furthermore, prioritising spending the longest time for online activities (either out-of-home or in-home) and activities at out-of-home destinations, and the fewest time for in-home activities compared to Gen X and Y, may be the reason why spending more time on online activities makes Gen Z visit the activity locations near their home and out-of-home activity anchors. However, spending more time for both out-of-home and in-home online activities was not observed to make Gen X and Y visit the out-of-home activity locations near their home and out-of-home activity anchors. In contrast to Gen Z, due to their life stage [75], Gen X and Y were found to spend a shorter time on online activities and activities at out-of-home destinations, and more time for in-home activities. This might be in line with the results by Delbosc and Nakanishi [45] showing that online activities might provide more opportunities for people to explore more space and more out-of-home discretionary activities.

Trade-off mechanisms were found between weekdays and weekends, particularly in Gen X’s and Gen Y’s Ih, and between Gen X’s Ih and Ic, particularly on Friday. Due to exposure to more time on online activity and activity at out-of-home activity, Gen Z, significantly, had no variability of activity space during weekends and weekdays. Moreover, the life stage [75] may also cause Gen Z to have a limited experience in space, which makes them have no variability of activity space throughout the given days. All generations were observed to have a similar Ic on a daily basis, whereas the oldest had the highest Ic on Friday and Sunday.

In general, socio-demographic variables can differentiate either the Ih or the Ic of people from different generations. However, the in-home, out-of-home and online activity patterns of people from different generations were found to be different from each other. Therefore, those activity patterns were found to statistically distinguish the Ic and Ih of people from different generations.

Further explorations on whether each generation’s activity space correlates with better social and mental health or subjective well-being should be undertaken in the future. Different activity space patterns among different generations might correlate with different health and well-being impacts. Moreover, activity space measurements are improved by using the people’s actual route choices rather than the shortest path. Thus, the analysis can determine whether the daily route choice influences the activity space measurements and significantly correlates with health and well-being impacts. Another future research is to apply geo-visualisation that includes the time dimension or the three-dimension activity space. The three-dimension activity space can describe the activity space during different time slices, such as morning, afternoon and evening time. The visualisation will show in which area people will spend time during a specific time slice.