The structure of joint leisure trips: analyzing two-person leisure trips of Dutch students

Journal of Transport Geography 31 (2013) 216–225

Contents lists available at SciVerse ScienceDirect

Journal of Transport Geography journal homepage: www.elsevier.com/locate/jtrangeo

The structure of joint leisure trips: analyzing two-person leisure trips of Dutch students Dick Ettema ⇑, Danielle Zwartbol Faculty of Geosciences, Utrecht University, PO Box 80115, 3508 TC Utrecht, The Netherlands

a r t i c l e

i n f o

Keywords: Social network links Social-recreational travel Spatial context Students

a b s t r a c t This paper investigates the spatial structure of students’ leisure trips and to what extent locational characteristics of social network partners influence decisions about their joint leisure activities and travel. To this end a survey was held among university students asking them about details of their last leisure trip made with a friend. Cluster analysis suggests that four typical leisure trip patterns can be derived. Three clusters seem to be determined by the residential locations of ego and alter. Depending on the residential distance between ego and alter, leisure trip distances are either short for ego and alter (if residential distance is very short), somewhat longer for both (if residential distance is slightly longer) or long for at least one partner (if residential distance is large). A fourth cluster includes cases with long leisure trips for both partners, independent of residential distance, representing cases where specific destinations are visited. A more detailed analysis of travel distance suggests that travel distance depends on size of the residential municipality, residential distance and objective and perceived quality of leisure facilities such as cafés, bars and restaurants. Overall, our study provides support for the idea that leisure trip decisions should be understood on the level of social network ties (i.e. ego and alter) rather than based on individual characteristics only. Also, it provides support for the idea that a focus on single ties, rather than on the structure of social networks as a whole, increases our insight in leisure trip decision making.Keywords: Social network links, Social-recreational travel, Spatial context, Students Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Over the past decade, the transportation literature has witnessed an increased interest in the relationship between social and recreational (SR) activities and travel. In transportation science, the increased interest stems from the awareness that the increase in car mobility over the past decades, which has led to adverse effects such as congestion, pollution and CO2 emissions, is caused to a considerable extent by an increase in travel for SR purposes (Schad et al., 2009; Axhausen, 2008). In the Netherlands, for instance, the travel related with leisure activities (excluding holidays) made up 44% of the total distance travelled in passenger transport in 2005 (Harms, 2008). Much (48%) of this leisure-related travel is undertaken for social activities, i.e. to meet and spend time with friends, relatives and others. Ohnmacht et al. (2009) report similar figures for Switzerland. In the Netherlands, the distance travelled for leisure purposes has increased by one third in the period 1985–2003 (Harms, 2008). As this growth was similar to the overall increase in travel, leisure’s share in total travel remained rather stable. Stauffacher et al. (2005) describe similar trends for Germany. ⇑ Corresponding author. Tel.: +31 302532918. E-mail address: [email protected] (D. Ettema). 0966-6923/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jtrangeo.2013.05.006

In contrast to their importance to mobility development, social and recreational trips have received less attention in the literature than for instance commute and shopping trips. Four decades ago Wheeler and Stutz (1971, page 376) already noted that ‘‘considering the depth of the research on social interaction, it is surprising that there are few studies of social travel within metropolitan areas’’ and this statement has considerable purchase even today, although there is a small literature on leisure travel which we will review below. Various aspects of such activities and trips have been examined, including time spent on leisure pursuits (e.g. Ren and Kwan, 2009), the associated travel time (e.g. Cools et al., 2010) and the transport mode used to access leisure activities (e.g. Limtanakool et al., 2006). Yet, the aspect that has been investigated most frequently is the frequency of engagement in leisure. Descriptive information about the frequency with which people engage in various leisure activities has been provided for a representative sample of the Swiss population by Stauffacher et al. (2005), for samples from Germany and Sweden by Schlich et al. (2004), for a German sample by Tarigan and Kitamura (2009), and for a Dutch sample by Sharmeen and Ettema (2010). Although the studies cannot be compared easily, they all indicate that meeting friends, going out to restaurants and bars, visiting theatres and cinemas, active sports and club meetings are important types of leisure activities. Further, various

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

scholars have examined the factors influencing the frequency of engaging in leisure activity and travel (Bhat and Lockwood, 2004; Scheiner, 2006; Sener et al., 2008; Farber and Páez, 2009; Tarigan and Kitamura, 2009). All report that sociodemographic and life-cycle factors have a significant impact on individuals’ propensity to engage in leisure activities and travel. Other factors that are found to influence leisure consumption are spatial setting (Bhat and Lockwood, 2004; Sener et al., 2008), seasonal effects (Bhat and Lockwood, 2004; Sener et al., 2008; Kemperman et al., 2000) and vehicle ownership (Scheiner, 2006; Tarigan and Kitamura, 2009; Farber and Páez, 2009). The spatial variability in leisure travel has also attracted considerable attention and is known to be much larger than for commuting and shopping trips (Stauffacher et al., 2005; Schlich et al., 2004; Tarigan and Kitamura, 2009). In a study observing individuals’ travel behaviour for six consecutive weeks, Schlich et al. (2004) show that participants keep visiting new locations for leisure activities throughout the study period. Although for most participants some eight locations accounted for about 80% of the leisure destinations visited, Schlich and colleagues demonstrated that location choice in leisure travel is more irregular and idiosyncratic than for other kinds of travel. In short, the literature has delivered important insights in various aspects of leisure trips. The approach that is typically taken, however, is to analyze leisure trips as made by a single decision maker. Sharmeen and Ettema (2010), however, report that some 80% of SR trips are made in company of others, and that in about half of the cases, this concerns someone from outside the household. It is emphasized that such trips are fundamentally different from trips made alone or with household members, since the activity partners come from different locations and different households, which will influence the decision making process. Thus, the socio-demographic and locational characteristics of the other, as well as their lifestyle preferences and tastes, may influence decisions regarding the leisure trip. Insight into the influence of characteristics and preferences of leisure partners on their joint leisure activities and travel is however very limited, both in terms of descriptive knowledge and in terms of the underlying processes. Yet, this insight is very relevant given that a considerable share of leisure travel and overall travel consists of joint leisure trips made with non-household members. This relevance is even more prominent with the advance of agent-based models that aim at predicting travel based on social networks and the interactions within such networks. Therefore, this paper aims at increasing the insight into the structure of joint leisure trips, based on a data set collected amongst Dutch students in 2011. While it is recognized that joint leisure trips often include three or more parties, our study is limited to two friends making a joint leisure trip. Focusing on this specific case allows us to gain first insights into the decision making mechanisms of joint trips made by members from different households. On a descriptive level, we will describe typical patterns of joint leisure trips in terms of travel distances, residential locations and travel modes. To increase the insight into travel considerations, we use multivariate statistics to investigate the impact of locational characteristics and personal characteristics on location choice decisions. The paper is structured as follows. In Section 2, we summarize existing empirical and theoretical studies related to social interactions and travel and position our work in the context of this literature. Section 3 describes the data collection procedure. Section 4 gives the empirical results, which include descriptive statistics of students’ joint leisure trips, a classification of joint leisure trips and an ordered logit model to understand the effect of spatial context on trip distance. Section 5 draws some general conclusions and addresses avenues for further research.

217

2. Joint decisions about social–recreational travel While studies that address the detailed interaction between partners involved in joint leisure activities is largely lacking, the awareness that one’s relationships with friends/family influence one’s travel has steadily increased over the past decade. This has invoked a series of studies focusing on the relationship between the composition and structure of social networks and frequency of social or recreational trips. The main assumption underlying these studies is that social networks (SNs) play a role in control over and information about resources and influence cooperation and competition which facilitate or constrain certain opportunities and behaviours. Social activities, defined as joint recreational or support activities, stem from the propensity of different parties to provide or consume companionship or support, which may be influenced by person, tie or network characteristics. Although it is recognized (Carrasco and Miller, 2006) that actors in social networks may include various entities, such as nations, organisations, groups and individuals, the social networks in the context of travel and activities, are usually defined as consisting of friends and family members, elicited by name generator techniques. These could be named ‘informal’ SN, as opposed to ‘formal’ SN such as clubs and associations (Kwak and Campbell, 2010). In this stream of research, Carrasco and Miller, (2006) found that engagement in social activities (e.g. hosting visitors and going to a bar/restaurant), controlled for socio-demographic characteristics, was dependent on the composition of a person’s social network such as the number of family members, friends, and network members from social organisations. Also geographical aspects of the social network appeared to matter, to the extent that increased distance to social network members reduced frequency of engagement in social activities. In a later study, Carrasco and Miller (2009) found that apart from characteristics of the network as a whole (such as density, number of isolates and centrality), characteristics of the alter (age, gender, alter’s position in his/her ego-network) and the tie (similarity, distance, frequency of ICTinteraction) also influenced the probability of engaging in social activities. In an early study, Ginsberg (1975) found that if more friends lived in the same neighbourhood, one is more likely to undertake leisure activities with friends rather than with the spouse. Although this relationship is mediated by gender, education and cultural background, it suggests that the social network of friends and family constitutes a resource for undertaking leisure activities which normally require company. Other scholars have focussed in a more general way on the size and composition of the social network and the frequency of face-to-face contact between network members. Tillema and Dijst (2007) found that the frequency of face-to-face contact correlated negatively with distance to the other network members and increased with the number of persons in the social network, the share of relatives in the social network, use of Internet and SMS and car availability. Mok et al. (2007) also found that frequency of face-to-face contact between social network members depends on geographical distance, and also on the type (kin, friends) and intensity (intimate, non-intimate) of the tie. Regarding the travel implications of face-to-face contacts between network members, Silvis et al. (2006) found that the length of social trips is positively related to the number of people in the social network, proportion and number of non-immediate kin, and the average age of the social network ties. The number of social trips correlated positively with social network size, household size and income. In short, the literature on social and recreational travel has addressed various aspects of SR trip making such as trip frequency, travel mode, travel time and location choice. In addition, it is recognized that a large part of SR trips is made together with

218

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

non-household members such as friends and (external) family members, implying that both the taxonomy and the spatial structure of the social network influence SR trip making. An issue that has received, to the authors’ knowledge, very limited attention to date is how the social network influences SR trip making on the level of individual links and individual trips. That is, how do two or more friends decide about location type, duration and travel mode of a joint activity, given their residential (or other base) locations, travel options and leisure preferences. In a theoretical paper, Ettema and Schwanen (2012) advocate a relational approach to analyzing SR trips, implying that decisions regarding joint leisure trips should be understood in the context of the locations (residential) of participants in the trip and the characteristics and preferences of the participants. With respect to trip generation, they emphasize (based on literature on social relationships and friendships) the role of preferences and skills of trip/activity participants for activity generation, but also less tangible aspects such as identity formation and confirmation and affective and emotional bonds between people. From a longitudinal perspective, also the history of the relationship as well as reciprocity considerations may play a role in the generation of joint SR trips. With respect to location choice of joint SR trips, Ettema and Schwanen (2012) note that time-geographic mechanisms (Hägerstrand, 1970) impact location choice in a straightforward way. In particular, given the base locations of people involved in a joint SR activity, time-geography can be used to find feasible locations where the joint activity may take place, given time constraints and travel options of the participants (see Neutens et al., 2010). However, Ettema and Schwanen (2012) argue that location choice is only to a limited extent based on distance minimization considerations. More important are functional characteristics of locations (affordances), such as opportunities offered by places to participate in sports, have a drink, see a play, and do shopping. Such affordances should match the preferences and requirements of the SR activity partners. In addition, they emphasize the relevance of the atmosphere of places, related to the geographical concept of ‘place’. This entails affective responses to very specific constellations of people and artefacts in specific places, but also the symbolic meaning of places (and the objects in these places) that may serve to confirm one’s identity or membership of a social or lifestyle related group. Thus, to properly understand location choice in joint SR trips, it is important not only to focus on accessibility related attributes but also include functional and atmosphere related characteristics of places. To conclude, there is limited insight into the process of trip generation, location choice and mode choice in joint SR trips. While theoretical notions pertaining to this process have been identified, empirical studies are largely lacking. This paper therefore aims to increase the insight into the decision making in joint SR trips by presenting the results of analyses of data of joint SR trips by Dutch students. First, we aim to get insight into the structure of joint trips by applying descriptive techniques. In addition, we aim to explore the implications of personal and place characteristics related to SR activity participants for location choice by investigating the effect of these factors on trip distance.

Dutch population and represent only a part of the total population in their age cohort. On the other hand, in cities like Utrecht and Rotterdam, university and higher vocational students make up 5.4–11.0% of the population and are relatively heavy users of public transport facilities and cycling infrastructure. In addition, our study is exploratory in nature, and aims to explore variations in the spatial structure of trips and mechanisms in location choice. However, one should be careful to transfer these findings directly to other population segments. Since the aim of the paper is to explore joint trip making on the level of individual trips and the involved company, data have been collected about the most recent SR trip that was made with a friend to a leisure location (i.e. not the friend’s home) and the relationship to this friend. Since we are interested in the role of residential locations and their characteristics for joint SR trip decisions, respondents were instructed to report about the most recent SR trip starting from home for both the respondent and his/her friend. The limitation to two person trips was chosen to obtain a maximum degree of uniformity in terms of decision making process (e.g., a trip with a group of five is likely to have much different dynamics in the decision making process than a two person trip). The focus on the most recent trip (instead of the most common trip) was chosen to obtain variation with respect to types of trips (e.g. dining vs. sports) and characteristics of partners (e.g. daily vs. incidental meetings). The questionnaire included questions regarding the following items:

3. Data

Data were collected among students in Utrecht and Rotterdam. These cities were selected since they differ considerably in structure. Compared to Rotterdam, students account for a larger part of the population in Utrecht. In addition, Utrecht is more compact city, which might influence SR trip making decisions. To collect data, students were approached in selected classes of universities and institutions for higher vocational study in Utrecht and Rotterdam. These classes were chosen such as to maximize

3.1. Data collection Data were collected among Dutch students from universities and higher vocational education organisations in the cities of Utrecht (316,000 inhabitants) and Rotterdam (617,000 inhabitants). It is recognized that students are not representative of the general

(a) Characteristics of the respondent (called ego hereafter), such as age, gender, level of the study followed, city of study, residential location, housing situation (living with parents or living independently), driving license holdership and car availability; (b) Characteristics of the company of the last SR trip (called alter hereafter), including the same characteristics as for the ego; (c) Characteristics of the last joint SR trip, such as location, travel mode (of ego and alter), type of SR activity and duration of SR activity. Travel mode was defined as the mode with which the longest distance was travelled, implying that we do not focus on access and egress trips of public transport and may lose detail on occasional complex trips. Activities in the home location of ego or alter were not regarded as joint SR activities, since in that case the ego or alter does not travel, reducing the activity to an individually made social visit, whereas in this study the emphasis is on joint activity and trip coordination. (d) Characteristics of the relationship between ego and alter, such as frequency of meeting each other face-to-face, mode of interaction in general (face-to-face, email, phone). (e) Functional characteristics of the ego’s residential location. To this end, we included questions regarding ego’s perceptions of the presence and quality of facilities in his/her residential municipality. Respondents were asked whether any of the following facilities (café, restaurant, discotheque, cinema, sports facility, swimming pool) was available in his/her residential municipality and asked to rate the quality of this facility (if present) on a 1–10 scale.

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225 Table 1 Sample characteristics for ego and alter. Ego (%)

Alter (%)

Gender Male Female

38.7 61.3

35.0 65.0

Household type With parents Independent

55.8 44.2

51.9 48.1

Size of residential municipality <50,000 inhabitants 50,000–200,000 inhabitants >200,000 inhabitants

40.5 25.1 34.4

34.4 25.7 39.9

Education level Higher vocational Academic

51.5 48.5

33.7 33.7

University city Utrecht Rotterdam % With drivers license

49.1 50.9 66.9

28.8 25.2 71.8

Car accessibility Always access to car Has to coordinate with other household members No access to car

20.2 40.5 39.3

26.7 36.1 37.2

variation in year of study and in type of study (technical, social, language, law, etc.). Students from universities and institutions for higher education and students in Utrecht and Rotterdam were sampled in equal shares. Data were collected in March and April 2011. In total, 162 usable questionnaires were returned. To allow for analyses of travel distance, the residential locations of ego and alter and the leisure location were used to calculate network distances using Google Maps. It is noted that the precision of the location descriptions differed. The ego’s home location was recorded accurately as their 4-digit zipcode. The 4-digit zip-code has an area of between 1.1 and 8.1 km2. However, the home location of the alter was logically recorded less accurately. 14.3% recorded the friends’ zipcode, 21.7% mentioned the street and municipality and 64.0% mentioned the neighbourhood and municipality. The leisure location was described by the zip-code by 2.5%, by the street and municipality by 42.0%, by the name of the facility (e.g. the name of a bar, club or sports facility) by 25.3% and by neighbourhood by 30.2%. In case of street or facility name, the 4-digit zipcode can be derived accurately. However, in case of a neighbourhood description, the neighbourhood center was manually determined to derive the zip-code. The variety in spatial detail did not allow us to calculate an exact travel distance in all cases. Therefore, distance classes (0–2.5 km, 2.5–5 km, 5–10 km, 10–20 km, 20–40 km, >40 km) were used, which made it possible to assign distance classes to pairs of locations with sufficient reliability.

219

municipality, 34.4% of the egos and 39.9% of the alters lives in a large city (>200,000 inh.). 40.5% of the egos and 34.4% of the alters live in smaller towns (<50,000 inh.). Taken together, this suggests that many students do not live in the city where they study. It should be noted, though, that also students who live independently, may not be able to live in their city of study due to scarcity of housing and settle in adjacent municipalities. The spatial distribution of egos and alters is displayed in Figs. 1 and 2. The figures show a concentration in the cities of Utrecht and Rotterdam, but also many living in smaller cities or peripheral areas. The egos show an almost even split between higher vocational studies and university studies (51.5% vs. 48.5%). Among the alters these studies each account for 33.7% of the sample, suggesting that students not necessarily socialize with other university or higher vocational students but also with people following other types of education or working. Also with respect to the city of study, the egos are split evenly between Utrecht and Rotterdam (49.1% vs. 50.9%). For alters, this is 28.8% and 25.2% respectively. This implies that students from Utrecht/Rotterdam do not only socialize with people having other occupations, but also with students from institutions in other cities. The majority of both egos and alters owns a driving license (66.9% and 71.8%), although only a minority (20.2% and 26.7%) always has access to a car. Many (40.5% and 36.1%) have to coordinate use of the car, for instance when living in a household with one car but multiple drivers. Since leisure trips (as studied in this paper) are made in couples, it makes sense not only to analyze ego and alter characteristics, but also how these are distributed within couples. In this respect, the vast majority of leisure trips are made with a person of the same sex (80.9%). 41.9% of the couples both live with their parents and 34.6% both live independently, implying that 76.5% have the same household situation. Also in terms of residential municipality size, the majority of couples shares similar characteristics. 28.5% both live in a small town, 15.8% both live in a medium size town and 28.5% both live in a large city. Taken together, 72.8% of the couples live in municipalities of about the same size. In addition, 51.5% live in the same municipality and 62.6% live less than 10 km away from each other. Obviously, these outcomes are related to the relationship between spatial location and social network formation, implying that people living in a shorter distance have more chance to meet and establish a social relationship (Tillema and Dijst, 2007). Less congruency is observed in terms of educational level. About half of the couples follow the same type of education (25.6% both academic, 27.2% both higher vocational). Finally, it is noted that the couples in our sample interact with a high frequency. 90.2% has contact more than once per week. This reflects the important role of peers in early adulthood, before other types of social relations (partner, colleagues) become more prominent.

4. Results 3.2. Sample characteristics 4.1. The structure of students joint SR trips Table 1 summarizes characteristics of egos and alters. The results suggest that the age of egos ranges from 16 to 30 years and from 16 to 35 years for alters. The majority however, is between 18 and 24 years old (90.8% of egos and 85.1% of alters). The majority of the egos and alters is female (61.3% and 65.0% respectively). Since the share of female students is 59.0% in Utrecht and 51.5% in Rotterdam this indicates a biased representation. With respect to household situation, 55.8% of the egos and 51.9% of the alters live with their parents. This implies that about half also live independently. In the Netherlands, this often implies living in or near to the city where one studies and is thought to be associated with a different lifestyle (more outgoing and fun seeking) which might affect leisure travel choices. With respect to the size of the

Having discussed the characteristics of egos, alters and couples, this section describes the characteristics of leisure trips, related to (selected) ego, alter and couple characteristics. With respect to the more specific purpose of the leisure trip, it is found that visiting restaurants (20.4%), bars (40.1%), cinemas (19.1%) and discotheques (12.3%) are the dominant activities during a leisure trip. These activities were precoded in the questionnaire. These precoded activities covered 95.7% of the joint SR activities. 7 respondents indicated other activities, such as outside activities (3), club activities (2), shopping (1) or a visit to a third party (1). Although no minimum duration of the joint SR activity was required, the duration of the joint activity is mostly rather long. 91.1% of the

220

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

Fig. 1. Spatial distribution of egos’ residential locations.

Fig. 2. Spatial distribution of alters’ residential locations.

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

221

Fig. 3. Spatial distribution of joint social–recreational activity locations.

Table 2 The relationship between trip distance of ego and alter and residential distance. Trip distance alter

Trip distance ego <5 km 5–20 km >20 km

<5 km

5–20 km

>20 km

69 (80.2%) 5 (5.8%) 12 (14.0%)

2 (10.0%) 12 (60.0%) 6 (30.0%)

15 (31.2%) 2 (4.2%) 31 (64.6%)

Distance between partners <5 km

5–20 km

>20 km

72 (70.6%) 10 (9.8%) 20 (19.6%)

3 (25.0%) 7 (58.3%) 2 (16.7%)

11 (27.5%) 2 (5.0%) 27 (67.5%)

Shortest trip distance <5 km 71 (69.6%) 5–20 km 10 (9.8%) >20 km 21 (20.6%)

6 (50.0%) 4 (33.3%) 2 (16.7%)

9 (22.5%) 6 (15.0%) 25 (62.5%)

Trip distance ego <5 km 5–20 km >20 km

activities take more than 2 h, 31.6% takes even more than 4 h. Only 1.6% takes less than 1 h. Activity duration is not related to travel duration. Chi-square tests of the relation between travel distance class (either of ego or alter) and activity duration class appeared not to be significant (p = 0.99 and p = 0.67 respectively). With respect to travel modes used, it appears that cycling (37.4% of egos and 38.0% of alters), bus/tram (20.2% and 19.6%) and car (22.1% and 22.7%) are the dominant travel modes. It is also noted that in 82.8% of the cases, ego and alter use the same travel mode. This is logical given that the partners usually have similar travel distance (see later). In addition, they usually share the same household type (with implications for car accessibility) and municipality size (with implications for access to public transport). Also, using the same travel mode may be related to actually travelling

together for whole or part of the trip. However, this was not recorded in the current data set. The majority of joint leisure trips are made over a short distance. 55.8% of the trips by egos and by alters is shorter than 5 km. However, 31.8% of ego trips and 31.2% of alter trips are more than 20 km, suggesting that there is considerable heterogeneity with respect to trip lengths. These figures are mirrored in the spatial distribution of SR activity locations (Fig. 3). SR activities are heavily concentrated in the cities of Utrecht and Rotterdam. Combining this with the more scattered distributions of residential locations (Figs. 1 and 2) results in the travel distance distributed outlined before. Table 2 also suggests a positive relationship between egos’ and alters’ travel distance, which is highly significant according to a Chi-square test (p < 0.001). The fact that many (51.5%) live in the same municipality and live close to each other is a logical explanation for the distribution displayed in the table. To further explore the relationship, Table 2 displays two additional relationships: 1. between the residential distance of ego and alter and the travel distance of the ego; 2. between residential distance of ego and alter and the shortest distance travelled by either ego or alter (thus, if the ego travels 5 km and the alter travels 10 km, the shortest distance is 5 km). This table suggests that with increasing residential distance, the travel distance for the ego increases. A potential explanation would be that at least one partner has to travel farther if the leisure activity takes place close to the residence of the other party, which may apply to the ego. However, also the shortest distance travelled (by ego or alter) increases with increasing residential distance (p < 0.001, according to a Chi-square test), suggesting that overall, a larger residential distance leads to a geographically wider range of destinations considered. To appreciate this finding, it is noted

222

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

Table 3 Clusters of student leisure trips. 1 35

2 48

3 41

4 30

Short travel distances, partners live close

Slightly longer travel distances and residential distance

Long residential one partner travels longer distance

Both partners travel a long distance, residential distance varies

Distance (short) 0.5–2.5 km 2.5–5 km 5–10 km 10–20 km 20–40 km >40 km

35 (100%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%)

2 (4.2%) 28 (80.0%) 11 (22.9%) 7 (14.6%) 0 (0.0%) 0 (0.0%)

5 (12.2%) 11 (26.8%) 4 (9.8%) 4 (9.8%) 5 (12.2%) 12 (29.3%)

3 (10.0%) 3 (10.0%) 2 (6.7%) 4 (13.3%) 18 (60.0%) 0 (0.0%)

Distance (long) 0.5–2.5 km 2.5–5 km 5–10 km 10–20 km 20–40 km >40 km

30 (85.7%) 5 (14.3%%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%)

0 (0.0%) 14 (29.2%) 18 (37.5%) 16 (33.3%) 0 (0.0%) 0 (0.0%)

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (7.3%) 38 (92.3%)

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 30 (100%) 0 (0.0%)

Distance (between) 0.5–2.5 km 25 (71.4%) 2.5–5 km 9 (25.7%) 5–10 km 1 (2.9%) 10–20 km 0 (0.0%) 20–40 km 0 (0.0%) >40 km 0 (0.0%)

28 (58.3%) 5 (10.4%) 9 (18.8%) 6 (12.5%) 0 (0.0%) 0 (0.0%)

4 (9.8%) 1 (2.4%) 1 (2.4%) 0 (0.0%) 2 (4.9%) 33 (80.5%)

12 (40.0%) 2 (6.7%) 2 (6.7%) 1 (3.3%) 13 (43.3%) 0 (0.0%)

n Interpretation

Table 4 Behavioural and residential characteristics of clusters.

Description

Cluster 1

Cluster 2

Cluster 3

Cluster 4

Short travel distances, partners live close

Slightly longer travel distances and residential distance

Long residential distance, one partner travels longer distance

Both partners travel a long distance, residential distance varies

10.4% 35.4% 12.5%

41.5% 26.8% 14.6%

20.0% 26.7% 36.7%

41.7% 0.0%

17.1% 0.0%

13.3% 3.3%

2.1% 31.2% 16.7%

24.4% 24.4% 26.8%

20.0% 26.7% 36.7%

47.9% 0.0% 2.1%

22.0% 0.0% 2.4%

13.3% 3.3% 0.0%

54.2%

41.5%

16.7%

45.8%

34.1%

66.7%

16.7%

39.0%

33.3%

37.5%

26.8%

Travel mode longest trip – Train 0.0% – Car 2.9% – Bus/tram/ 8.6% metro – Bicycle 71.4% – Walking 17.1% Travel mode shortest trip – Train 0.0% – Car 0.0% – Bus/tram/ 2.9% metro – Bicycle 80.0% – Walking 17.1% – Other 0.0% Share living independent

65.7%

Size of residential municipality <50,000 22.9% inhabitants 50,000–200,000 11.4% inhabitants >200,000 65.7% inhabitants

that the residential distance is also correlated with size of municipality (p < 0.001, according to a Chi-square test): those living in larger cities have shorter residential distances. Thus, the mechanism seems to be that those living in larger cities not only live closer to each other (as a result of higher population densities), but also have leisure facilities (mainly bars, restaurants, discotheques, cinemas in this sample) available on within shorter distances. In short, the outcomes suggests that joint leisure trips of Dutch students are in most cases symmetric, in the sense that travel distance and mode choice of ego and alter are similar. In addition, it is found that if ego and/or alter live in smaller towns, this leads to

0.00%

longer residential distances but also to longer leisure distances. Activity durations mostly take more than 2 h, and visiting restaurants, bars, discotheques and cinemas are the prominent type of leisure. 4.2. Classification of joint leisure patterns Based on the travel distance of ego, alter and the residential distance, a classification was developed of typical leisure trip patterns in relation to the residential locations. To this end a two stage cluster analysis was carried out on three variables:

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225 Table 5 Estimation results ordered logit model. Variable

Coefficient

p

Threshold parameters <2.5 km 2.5 to <5 km 5 to <10 km 10 to <20 km 20 to <40 km

0.582 0.809 1.423 2.100 3.726

0.456 0.304 0.073 0.009 0.000

Locational characteristics # Jobs in hotels, restaurants, cafes within15 min cycling # Jobs in services within15 min cycling Quality of cafes is insufficient Study in Utrecht

0.002 0.002 1.279 0.905

0.037 0.040 0.003 0.023

Lives in town <50,000 inh. Lives in town 50,000–200,000 inh. Lives in town >200,000 inh. (base)

2.278 3.446

0.000 0.000

Friend lives in town <50,000 inh. Friends lives in town 50,000–200,000 inh. Friend lives in town >200,000 inh. (base)

0.580 1.174

0.264 0.045

Residential Residential Residential Residential Residential Residential

2.461 2.700 1.447 2.328 0.803

0.000 0.000 0.049 0.024 0.221

0.066 0.556

0.851 0.253

Studies language, law Studies maths, physics, chemistry, technology Studies social sciences (base)

2.111 0.542

0.010 0.395

Higher vocational (university = base) Has driving license

0.126 0.656

0.741 0.193

Always access to car Sometimes access to car No access to car (base)

0.697 0.375

0.242 0.420

Leisure partner of other sex

1.016

0.033

Meets friend daily Meets friend Ponce per week Meets friend
0.798 0.011

0.239 0.986

Nagelkerke rho-square

0.513

distance distance distance distance distance distance

<2.5 km 2.5–5 km 5–10 km 10–20 km 20–40 km >40 km (base)

Sociodemographics and tie characteristics Male Lives with parents

1. the shortest travel distance of ego and alter; 2. the longest travel distance of ego and alter; 3. the residential distance. Given that we use categorical variables, distance between clusters is based on the loglikelihood criterion. A maximum number of branches of 8 and a maximum tree depth of 3 was used. The number of clusters was determined based on Schwartz’s Bayesian Criterion, resulting in 4 clusters. The silhouette measure of 0.4 suggests that the solution is acceptable. The outcomes suggest the existence of four distinct classes (Table 3). A first cluster consists of trips between partners living close to each other (<5 km) and making short leisure trips (<5 km). A second category consists of partners living close (<5 km) or somewhat farther (<20) away from each other, and making trips between 2.5 and 20 km. A third cluster consists of partners living far away from each other (>40 km) of which one partner travels a long distance, and the other mostly much less (<10 km). The final cluster consists of trips in which ego and alter both travel longer distances, with their residential distance either being very large or very small. Given the definition of clusters we tested whether selected trip characteristics differed significantly between clusters (Table 4), using Chi-square tests. With respect to the main travel mode for the shortest trip we find significant differences (p < 0.001). In

223

particular, in the ‘micro’ cluster 1, bicycle is the dominant mode, whereas in the local cluster 2 also car and bus/tram are important modes. In case both parties travel long distance (cluster 3), train is a relatively important travel mode, along with car and bus/metro. If only one party travels a long distance (cluster 4), the other party has varying trip distances and uses a variety of modes. Also the travel mode of the longest trip differs significantly between clusters (p < 0.001). Again, if both partners travel a very short distance (cluster 1) bicycle is the dominant mode, and if both travel less than 10 km, bicycle, car and bus/tram are the dominant modes. In case the longest trip is over 20 km (clusters 3 and 4) the train becomes relatively more important, although also car and bus/metro are used. Surprisingly, several students report to use their bicycle for distances over 20 or even 40 km. In addition, differences between the clusters were found in terms of socio-demographic and locational characteristics. The share of living independently differed significantly between clusters (p = 0.001). The ‘micro’ cluster 1, where students live close to each other and travel short distances occurs mostly among those living independently. Most likely, this is related to living in dense urban areas with high concentrations of fellow students. Also cluster 2 (‘local’ trips) occurs slightly more often among independent living students. Those living with their parents are more likely to be involved in cluster 3 and 4 trips, characterized by longer travel distances and longer residential distances. In addition there is a significant relationship between trip cluster and municipality size (p < 0.001). The micro cluster 1 occurs most often in large cities, whereas cluster 4 occurs more often in small towns, with longer distances between ego and alter and longer distances to urban facilities. However, cluster 2, with relatively short travel distances, is also occurring often in smaller towns. This may reflect leisure trips with local friends visiting local facilities, which requires a longer distance given the lower densities in these towns. We did not find relationships between leisure trip type (cluster) and the city where ones studies (Utrecht or Rotterdam) (p = 0.984), frequency of contact with the alter (p = 0.669), specific trip purpose (p = 0.325) or duration of the activity (p = 0.754). Taken together, it appears that the spatial configuration of the leisure trip (locations of ego, alter and activity and the distances travelled) corresponds strongly with the urban setting of ego and alter. Trips between students who live close together and travel a short distance are typically observed if they both live in a large city, whereas trips in which ego and alter live 5–10 km and both travel less than 10 km are observed often if both live in a smaller town. If ego and alter live in different cities, cluster three is more dominant, in which one of the partners travels a longer distance. The cluster where both travel a longer distance (cluster 4) takes place for a variety of locational characteristics of ego and alter, but not if both live in a large city.

4.3. Travel distance Given the importance of locational characteristics of ego and alter for leisure travel outcomes, a remaining issue is their importance relative to more qualitative locational factors and sociodemographic characteristics of alter and ego. As noted before, we hypothesize that functional characteristics of leisure locations, which do not necessarily coincide with municipality size, as well as atmosphere related qualities, which determine whether locations match students’ lifestyles, influence location choice and therefore distance travelled. To investigate these relationships, an ordered logit model was estimated, in which travel distance of the ego (expressed in categories) is the dependent variable. Explanatory variables include the variables used to interpret the clusters (size of the residential municipality of ego and alter),

224

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

socio-demographics of the ego, social tie relationships and perceived and objective indicators of the quality of leisure places. Qualitative characteristics (see Ettema and Schwanen, 2012) may refer to the perception of availability and quality of facilities, as well as objective, more detailed measures of availability of services. With respect to the perceived quality of facilities, we asked respondents to grade the presence and quality of a series of facilities (café, restaurant, discotheque, cinema, sports facility, swimming pool) on a 1–10 scale, which were tested for explanatory power in the model. In addition, two objective measures were used: the number of jobs in hotels/restaurants/cafes and the number of jobs in services that can be accessed within 15 min by bicycle. We use the latter as a proxy for the availability of leisure facilities. The estimation results (Table 5) confirm the role of residential location for leisure trip decisions: the trip distance of those living in medium or small size towns is larger than of those living in large cities. In addition, trip distance increases with residential distance, as discussed before. Thus, a larger residential distance leads to the consideration of more remote locations, including locations close to the alter’s place. In addition, it is found that travel distance of the ego diminishes if the alter lives in a smaller (medium size) town. Most likely, if the alter lives in a smaller town, this is associated with lower availability of options for joint leisure (bars, restaurants, cinemas), making the alter’s town a less favourable destination to visit by the ego. However, apart from municipality size and residential distance, also more qualitative locational characteristics play a role. First, students studying in Utrecht travel shorter distances than students studying in Rotterdam. Since we control for other locational characteristics, this is likely an effect of spatial structure. The city of Utrecht has a higher population density than Rotterdam (3279 inh./km2 vs. 2850 inh./km2) which could lead to shorter distances in general. However, given that the same share of students studying in Utrecht and Rotterdam live outside their university town, this may also be related to the location of residential areas relative to Utrecht and Rotterdam (the cities used most frequently for leisure activities). In addition, the facilities of the residential municipality play a role in destination choice and the resulting travel distance. Distance decreases if more employment in cafes/bars/restaurants (a proxy for presence of bars) is present within 15 min of travel. Also, if quality of the cafes/bars/restaurants is rated as satisfactory, travel distance is less. This suggests that both presence and quality of these facilities in the municipality lead to more local leisure activities and less travel. However, presence of jobs in services leads to longer travel distances. Apparently, services represented in this indicator (banks, post office, library) do not play a role in students’ leisure participation and are located in other places than the leisure facilities used by students. In general, if a leisure trip is made to visit a bar/restaurant/café, travel distance will be less, since these facilities are more local, whereas facilities such as discotheques, cinemas, swimming pools or sports facilities require more travel. Finally, it is found that, controlled for locational characteristics, students living independently make longer leisure trips as compared to those living with their parents. Thus, students living independently may be more outgoing and inclined to spend more time travelling to leisure locations. Apart from locational characteristics, several socio-demographic and tie relationships influence travel distance. We find that students in law, literature, etc. make longer leisure trips and that same-sex leisure trips tend to take place over shorter distances. No difference was found between males and females. In addition, some variables that might theoretically affect travel distance are not found to do so for this sample. Owning a driving license and having access to a car do not influence trip distance, although about 20% of the trips is made by car. Thus apparently, car is not

used to make longer trips than the combined options of walking, cycling and public transport. Also, frequency of contact is not related to travel distance, although one might expect that if one interacts less frequently with someone, one is prepared to invest more travel in the joint activity.

5. Conclusion This paper has investigated the spatial structure of students’ leisure trips and to what extent locational characteristics of social network partners influence decisions about their joint leisure activities and travel. Descriptive analyses of our sample reveal that students’ joint leisure trips are mostly made to visit bars, restaurants, cinemas or discotheques. These activities usually last for more than 2 h and often involve short travel distances. Main travel modes used include cycling, car and bus/tram. In many cases ego and alter use the same travel mode and have about the same travel distance. Cluster analysis suggests that four typical leisure trip patterns can be derived. Three clusters seem to be determined by the residential locations of ego and alter. Depending on the residential distance between ego and alter, leisure trip distances are either short for ego and alter (if residential distance is very short), somewhat longer for both (if residential distance is slightly longer) or long for at least one partner (if residential distance is large). A fourth cluster includes cases with long leisure trips for both partners, independent of residential distance, representing cases where specific destinations are visited. A more detailed analysis of travel distance suggests that travel distance depends on size of the residential municipality, residential distance and objective and perceived quality of leisure facilities such as cafés, bars and restaurants. Overall, our study provides support for the idea that leisure trip decisions should be understood on the level of social network ties (i.e. ego and alter) rather than based on individual characteristics only. Also, it provides support for the idea that a focus on single ties, rather than on the structure of social networks as a whole, increases our insight in leisure trip decision making. We feel that the outcome of this and follow up studies could feed efforts in developing forecasting models of leisure travel based on social networks. In particular, reliable simulations of the spatial structure of social networks, combined with the mechanisms found in this paper could lead to more realistic prediction of joint leisure trips. Although we feel that our results contribute to insight in joint leisure trip making, we recognize that this study can be extended in a number of ways. First, it is noted that we used a very specific sample (students in higher education). It is recognized that this group differs from other population segments in important ways. First, their time use patterns and corresponding time constraints with respect to leisure differ from more time constrained groups, such as the working population, with potential implications for leisure trip decisions. On the other hand, their access to transportation by car will be more limited. More importantly, the social network structure of students is likely to differ significantly from that of other groups. Social networks may be more local and based in the residential area as compared to e.g. professionals who have moved to other places for work and still maintain links with former fellow students. On the other hand, students living independently may maintain networks in both their study town and home town. Further study is needed into the social network structures of different population segments in order to increase the understanding in joint leisure trip making. Another limitation of this study is that we have focused on leisure trips made by two persons only, whereas travel parties may also consist of more individuals. Although Neutens et al. (2010) have looked into options for multi party leisure trips in a norma-

D. Ettema, D. Zwartbol / Journal of Transport Geography 31 (2013) 216–225

tive time-geographic framework, descriptive and modeling efforts of such trips are still lacking and require additional research.

References Axhausen, K.W., 2008. Social networks, mobility biographies, and travel: survey challenges. Environment and Planning B: Planning and Design 35, 981–996. Bhat, C.R., Lockwood, A., 2004. On distinguishing between physically active and physically passive episodes and between travel and activity episodes: an analysis of weekend recreational participation in the San Francisco Bay area. Transportation Research Part A: Policy and Practice 38, 573–592. Carrasco, J.-A., Miller, E.J., 2006. Exploring the propensity to perform social activities: a social network approach. Transportation 33, 463–480. Carrasco, J.-A., Miller, E.J., 2009. The social dimension in action: a multilevel, personal networks model of social activity frequency between individuals. Transportation Research Part A: Policy and Practice 43, 90–104. Cools, M., Moons, E., Wets, G., 2010. Assessing the impact of public holidays on travel time expenditure: differentiation by trip motive. Transportation Research Record: Journal of the Transportation Research Board 2157, 29–37. Ettema, D., Schwanen, T., 2012. A relational approach to analysing leisure travel. Journal of Transport Geography 24, 173–181. Farber, S., Páez, A., 2009. My car, my friends, and me: a preliminary analysis of automobility and social activity participation. Journal of Transport Geography 17, 216–225. Ginsberg, Y., 1975. Joint leisure activities and social networks in two neighborhoods in Tel Aviv. Journal of Marriage and the Family 37, 668–676. Hägerstrand, T., 1970. What about people in regional science? Papers in Regional Science 24, 7–21. Harms, L., 2008. Overwegend Onderweg [Mostly Mobile]. SCP, The Hague. Kemperman, A.D.A.M., Borgers, A.W.J., Oppewal, H., Timmermans, H.J.P., 2000. Consumer choice of theme parks: a conjoint choice model of seasonality effects and variety seeking behaviour. Leisure Sciences 22, 1–18. Kwak, N., Campbell, S., 2010. Mobile communication and social capital: an analysis of geographically differentiated usage patterns. New Media and Society 12, 435–451. Limtanakool, N., Dijst, M., Schwanen, T., 2006. The influence of socioeconomic characteristics, land use and travel time considerations on mode choice for

225

medium- and longer-distance trips. Journal of Transport Geography 14, 327– 341. Mok, D., Wellman, B., Basu, R., 2007. Did distance matter before the Internet? Interpersonal contact and support in the 1970s. Social Networks 29, 430–461. Neutens, T., Schwanen, T., Miller, H., 2010. Dealing with preferences, privileges and timing in opportunities for joint activity participation. Geographical Analysis 42, 245–266. Ohnmacht, T., Götz, K., Schad, H., 2009. Leisure mobility styles in Swiss conurbations: construction and empirical analysis. Transportation 36, 243–265. Ren, F., Kwan, M.-P., 2009. The impact of the Internet on human activity–travel patterns: analysis of gender differences using multi-group structural equation models. Journal of Transport Geography 17, 440–450. Schad, H., Ohnmacht, T., Götz, K., 2009. Leisure mobility styles in Swiss conurbations: construction and empirical analysis. Transportation 36, 243–265. Scheiner, J., 2006. Does the car make elderly people happy and mobile? Settlement structures, car availability and leisure mobility of the elderly. European Journal of Transport and Infrastructure Research 6, 151–172. Schlich, R., Schönfelder, S., Hanson, S., Axhausen, K.W., 2004. Structures of leisure travel: temporal and spatial variability. Transport Reviews 34, 219–237. Sener, I.N., Copperman, R.B., Pendyala, R.M., Bhat, C.R., 2008. An analysis of children’s leisure activity engagement: examining the day of week, location, physical activity level, and fixity dimensions. Transportation 35, 673–696. Sharmeen, F., Ettema, D., 2010. Whom to Hang Out With and Where? Analysis of the Influence of Spatial Setting on the Choice of Activity Company. Paper Presented at the 12th WCTR Conference, 11–15 July, 2010, Lisbon, Portugal. Silvis, J., Niemeier, D., D’Souza, R., 2006. Social Networks and Travel Behavior: Report from an Integrated Travel Diary. Paper Presented in the 11th International Conference on Travel Behaviour Research, Kyoto, Japan. Stauffacher, M., Schlich, R., Axhausen, K.W., Scholz, R.W., 2005. The Diversity of Travel Behaviour: Motives and Social Interactions in Leisure Time Activities. (accessed 07.10.10). Tarigan, A.M., Kitamura, R., 2009. Week-to-week leisure trip frequency and its variability. Transportation Research Record 2135, 43–51. Tillema, T., Dijst, M., 2007. Electronic and Face-to-Face Communication in Maintaining Social Relationships. Paper Presented in Workshop Frontiers in Transportation 2007, Amsterdam. Wheeler, J.O., Stutz, F.P., 1971. Spatial dimensions of urban social travel. Annals of the Association of American Geographers 61, 371–386.