How do users choose their routes in public transport? The effect of individual profile and contextual factors

Transportation Research Part F 51 (2017) 24–37

Contents lists available at ScienceDirect

Transportation Research Part F journal homepage: www.elsevier.com/locate/trf

How do users choose their routes in public transport? The effect of individual profile and contextual factors Elise Grison a,b,⇑, Jean-Marie Burkhardt a, Valérie Gyselinck a a b

Psychology of Behavior and Mobility Lab, IFSTTAR, France Memory and Cognition Lab, UMR S894, Paris Descartes University and INSERM, France

a r t i c l e

i n f o

Article history: Received 30 September 2016 Received in revised form 9 June 2017 Accepted 29 August 2017

Keywords: Decision-making Route-planning Transportation User profiles

a b s t r a c t The aim of the present study was to better understand how public transport users make their choice of route, in order to favor the use of public transport (henceforth PT) in large cities. Based on decision-making theories, a classical choice paradigm (Slovic, 1975), and recent findings in psychology (Chowdhury & Ceder, 2013; Grison et al., 2016) we developed a new method to investigate the effect of contextual and individual factors on PT route choices. We proposed to sixty PT users realistic forced choices between two PT routes that differed in affective (level of physical comfort) and instrumental (number of transport modes) attributes. We also varied the context of the decision (long or short route; route to go to work, to a leisure activity, etc.), and we recorded various individual characteristics (age, sex, attitude towards PT, habits, etc.). Our results highlight that: the comfort of the route is preferred to the number of transport modes, especially for long trips; the choice of the comfortable alternative or the one with only one transport mode depends on user characteristics; the length of the trip and habits are the most important variables in the decision, but attitudes also seem to play a major role. Our study furthers current knowledge about the psychological process of PT route choice in light with multi-attribute choices theories, and provides new insights that can contribute to improving current large city route-planning aids. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction The promotion of public transport (henceforth PT) use can be a promising way to reduce the current increase in air pollution and traffic congestion in large cities. A first avenue of research has been dedicated to understanding the behavioral switch from the use of the private car to the use of public transport (see for example Litman, 2008; Popuri, Proussaloglu, Ayvalik, Koppelman & Lee, 2011; Rubens, Gosling, & Moch, 2011). These studies have been essential to show the influence of various structural, contextual, and individual factors on transport mode choice. Recently, a complementary promising way to study this issue has been promulgated. Chowdhury and Ceder (2013) argued that focusing on the route choice behavior of current public transport users should enable to extract information that could then be used to enhance the use of public transport (see also for example Grison, Gyselinck, & Burkhardt, 2016). This proposal is particularly suitable for large cities, where various transport modes and routes are available. For example in Paris (France), taking only the public transport network into consideration, travelers can take the subway, the bus, the tramway, or ⇑ Corresponding author at: Psychology of Behavior and Mobility Lab, IFSTTAR, France. E-mail address: [email protected] (E. Grison). http://dx.doi.org/10.1016/j.trf.2017.08.011 1369-8478/Ó 2017 Elsevier Ltd. All rights reserved.

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

25

they can combine several of these transport modes. Moreover, the density of the transportation network allows people to choose between many alternatives resulting from different transport modes or combinations. Consequently, users have to compare the alternatives, and select the alternative that suits them best depending on their own preferences (Tversky, 1972). In the current literature, how people make such route decisions between several PT routes remains under-investigated. As a consequence, to date no model explains this particular behavior from a user perspective. Thus, the aim of the present study is to better understand how PT users plan their route, how they proceed, what the factors involved in their decision are, and how these factors influence their choice. From an applied viewpoint, clarifying the factors and the decision processes involved in PT situations can provide fruitful insights to develop innovative solutions for supporting and promoting the use of PT, such as the improvement of route planning aids. The paper is structured as follows. First, a review of the literature concerning experimental data acquired theses last years on PT route choice is presented. Then a theoretical framework to explain PT route choice is discussed followed by the research question. We follow by the presentation of the methodology of the study. Finally, the results are presented and discussed in light of the current literature and of their implications in improving route-planning aids. 2. Literature review 2.1. Routes choices in public transport system 2.1.1. Attributes and criteria considered in PT route choice When making a choice of transport to travel, people consider various aspects of the route to make their decision. In this paper, we will use the term ‘‘attribute” to refer to the characteristics of the route; and we will refer to the term criteria when speaking of the user’s preference when making a choice, such as favoring one attribute rather another, or having a cut-off on an attribute. In global transport literature on mode choice, three categories of attributes have been identified: instrumental (i.e., cost, travel time), symbolic (i.e., norms, social representation), and affective (i.e., emotional dimension linked to the travel/mode) (Steg, 2005). To date, the attributes highlighted in literature on PT route choice can be classified in the instrumental and affective categories (Ben-Elia, Di Pace, Bifulco, & Shiftan, 2013; Bovy & Hoogendoorn-Lanser, 2005; Chiu, Lee, Leung, AU, & Wong, 2005; Guo & Wilson, 2011; Raveau, Guo, Munoz & Wilson, 2014; Raveau, Munoz & de Grange, 2011). The main findings of studies investigating PT route choices are that beyond some well-established instrumental attributes such as the travel time and cost, other attributes like the transfers’ characteristics (Guo & Wilson, 2011), and the network topology (Raveau, Guo, Muñoz, & Wilson, 2014; Raveau, Muñoz, & de Grange, 2011) can be involved in the decision process. The results also highlight that trips are composed of different slices of time (i.e., waiting time, walking time, in-vehicle time) that are evaluated differently by users when making their choice. Indeed, people assess the waiting time and walking time as being more important for their decision than the in-vehicle time (Raveau et al., 2014). The finding that slices of the travel time can have different values depending on their nature is important since one of the most important attribute of PT routes is that they frequently imply transfers between lines or transport modes. In most doorto-door trips using PT, users have to make transfers, which are negatively perceived because they involve waiting time, walking, uncertainty, and loss of control over the trip (Friman, 2010; Hine & Scott, 2000). Moreover, Heye and Timpf (2003) observed that transfers could lead to a difficulty in accomplishing the trip, especially due to the physical and cognitive load on wayfinding induced by transfers. In their attempt to modeling wayfinding in public transport, Rüetschi and Timpf (2005) proposed that finding our way in unfamiliar public transport is a difficult task because numerous signs are present, and the information is not always well displayed. It has been indeed shown that understanding signs in the public transport system is essential for the wayfinding task, especially at transfer or egress (Fontaine & Denis, 1999; Timpf, 2002). Moreover, Hannes, Janssens, and Wets (2009) demonstrated from interview data that having a good spatial representation of the environment help in finding the best transport mode to make the trip, and facilitated the accessibility of the transport mode. Recent work also shows that the more transfers the route has, the more difficult the route planning task is (Grison, Gyselinck, Burkhardt, & Wiener, 2016). For all the above-mentioned reasons, users tend to avoid transfers or to plan ahead when they expect to have transfers in the PT route (Hine & Scott, 2000). As a result, the literature focusing on PT route choice has mainly investigated the factors that can facilitate the use of PT routes including transfers. Chowdhury and Ceder (2013), for example, observed in their survey-based study that if some instrumental aspects of transfer – such as the reliability between modes, the walking time, and traffic information or waiting time - were improved, users would be more prone to use PT routes with transfers. All these studies show that various instrumental attributes can be considered when making a PT route choice. In addition to this kind of attributes, a few studies have reported that users can also consider affective dimension of the route to make a decision. For example, Hölscher, Tenbrink, and Wiener (2011) asked participants about the reasons of their choice of route (i.e., walking routes) and found out that some affective attributes such as the attractive characteristics (esthetics) of the route can be considered at the same level as instrumental ones. In a recent study on PT route choice, Grison et al. (2016), asked participants to describe real PT route events. The descriptions included the reasons of their choice of route. The authors point out the fact that affective attributes such as the comfort in the transport (e.g., be seated, quietness), and the fact to travel with a friend can be strong reasons to choose a particular route. To date however, while these recent studies report that affective attributes are considered by users to make their choice, these attributes have been understudied compared to instrumental ones.

26

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

2.1.2. Influence of contextual and individual factors on PT route choice Understanding the factors associated to the selection and hierarchy of the attributes considered by users is another recent issue emerging in the literature. Studies in this field (Chowdhury, Ceder, & Sachdeva, 2014; Grison et al., 2016) have been strongly influenced by previous work on transport mode choice, showing that three kinds of factors influence the attribute considered and the chosen criteria when making a transport mode choice. Structural factors are the first kind, for example the need for mobility or the accessibility to transportation (Brisbois, 2010; Gandit, 2007). The second kind is individual factors, such as socio-demographic characteristics (e.g., age, sex, and number of children), attitude towards transport modes or ecology, and habits (Brisbois, 2010; Hunecke, Blöbaum, Matthies, & Höger, 2001; Klöckner & Friedrichsmeier, 2011; Vredin Johansson, Heldt, & Johansson, 2006). These individual factors are referred to the ‘‘user profile” in recent studies (Grison et al., 2016; Klöckner & Friedrichsmeier, 2011), as we will do in the following. Finally, the third kind of factor is contextual, such as the aim of the specific trip, the weather, the time of day, or the emotional state of the user when taking a decision (Dieleman, Dijdt & Burghouwt, 2002; Klöckner & Friedrichsmeier, 2011; Mann & Abraham, 2006). Interestingly, Klöckner and Friedrichsmeier (2011) recently highlighted that when a user chooses a transport mode, the criteria taken into account result from interactions between the three kinds of factors. They asked participants to fill in a questionnaire to assess their transport mode choice, their user profile (attitude towards alternative transport modes to the car, habits, sociodemographic characteristics), as well as the context (aim of the trip, length of the trip, moment of the week). Among the various results obtained by the authors, they showed that a strong car use habit together with a negative attitude towards alternative transport modes to the car was associated with a higher impact of the travel time criterion on the intention to use the car. They also showed that users with a positive attitude towards the use of alternative modes to the car used their car less for the purpose of shopping. Following this line of research, but in the particular field of PT route choice, Chowdhury et al. (2014) reported the effect of user profile - including sociodemographic characteristics, and frequency of PT use - on the use of PT routes with transfers. In their study, the authors proposed several PT route choices to participants. At each trial, participants were first described verbally a PT route without transfer that took 40 min. Then, three alternatives were presented, each with one transfer. The characteristics of the transfer were varied: information given at the transfer point or not, and a shelter to wait or not. The alternatives allowed participants to gain from 10 to 20 min. Results showed that participants who frequently used PT and women were 6 times more prone to choose the PT route with one transfer than participants who did not use PT frequently, or than men. This study has been a first conclusive attempt to demonstrate that individual characteristics play a role in the PT route choice. Recently, Grison et al. (2016) have deepened this line of research. They analyzed real users’ previous experience, and showed the influence of the user profile as well as of the context on route choice. They focused on the choice of using a route with multiple transport modes (i.e., multimodal route, including transfers) or not. In their study, participants (PT users) were asked to report on previous satisfying or dissatisfying PT route experiences. For each route, participants had to describe the route, and the context of the route in detail. They also had to describe if the route was satisfying or not. The authors recorded participants’ sociodemographic characteristics, their attitude towards PT, and their frequency of use of multimodal routes. The results highlighted different patterns of routes associated to specific user profiles. For example, the users the most inclined to choose multimodal routes were those who frequently used multimodal routes in their daily life and who had a positive attitude towards PT; they were preferentially men aged 35–50. These users were more prone to take multimodal routes when going back home in the evening, based on affective criteria. Another user profile was preferentially associated to women aged 25–34 used to unimodal routes (i.e., one transport mode), and who did not have a positive attitude towards PT. Users with this profile based their choice on instrumental criteria. This study shows that depending on user profile and context, users will consider more or less instrumental or affective attributes of the route to make their choice. Like Chowdhury et al. (2014), the study by Grison et al. (2016) highlights that behavior is strongly influenced by familiarity with a specific behavior, in that users familiar with routes including transfers or changes of mode are more inclined to use this kind of route than others. This result also shows that when making a decision, the past experience – satisfying or not - and habits play an important role in the future decision. As we previously shown, transfers are important in the decision to use one route or another (Hine & Scott, 2000). Researches seem to show that an important factor in the decision to use transfers is related to users sense of direction ability (Hannes et al., 2009). The sense of direction can be defined as the general ability to find our way, and it involves various spatial abilities such as spatial visualization, and spatial manipulation (Hegarty, Montello, Richardson, Ishikawa, & Lovelace, 2006). The sense of direction is an ability also involved in the construction of mental spatial representation, and in routeplanning (Hegarty, Richardson, Montello, Lovelace, & Subbiah, 2002). It has been shown in literature important individual differences in the sense of direction ability (Hegarty et al., 2002). As this ability is involved in route-planning activity, we could hypothesize that depending on the level of users in the sense of direction ability, the route choice will differ. To date, however no studies have investigated this question in the context of PT route choice. Altogether, the recent results obtained suggest that when choosing a PT route, users consciously compare the alternatives on various attributes, and choose the alternative that better fit their criteria. Moreover, the chosen criteria seem to depend on user profile, and context. To our knowledge, however, no models or theories explain the PT route choice from the user perspective, and by integrating those kinds of factors. Accordingly, the extent to which these factors influence the decision is not well known.

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

27

2.2. Multi-attributes choice: a theoretical framework for PT route choices 2.2.1. Definition of multi-attributes choice When making a PT route choice, people are faced with several route alternatives that are characterized by different attributes. This particular task has been commonly called a multi-attribute decision-making task (Abelson & Levi, 1985). For example, it can be a choice between one route taking 30 min with one transfer, and one route taking 35 min without transfer. In this example, the travel time and the number of transfers are the attributes of the routes. Multi-attribute decision-making was first studied in the domain of economics and management, in order to support the process of choosing the best mathematical option (see for example: Huber, 1974; Zanakis, Solomon, Wishart, & Dublish, 1998). A major assumption of this approach is that the decision-maker will compare all the existing options to decide which one is the best. An underlying assumption is that s/he has access to all the useful information on every option to enable their comparison. However, this assumption rises two main problems when applied to human behavior: (1) people do not systematically analyze all the alternatives and attributes, but rather rely on heuristics to make their choice (Bettman, Johnson, & Payne, 1990; Svenson, 1979); and (2) people do not necessarily choose the best mathematical option, but an option that best fits their preferences and needs (Golledge, 1995; Grison & al., 2016; Hölscher, & al., 2011). 2.2.2. The ‘‘prominence hypothesis” to explain PT route choice Other theories have therefore been proposed in the literature to better explain the cognitive processes involved when users make choices. A very large number of ‘‘rules” have been proposed and empirically tested (Montgomery & Svenson, 1976; Payne, Bettman, & Johnson, 1988; Russo & Dosher, 1983; Svenson, 1979). Amongst the various rules proposed, the Elimination-By-Aspect (EBA, Tversky, 1972) may be particularly suited to explain the choice between different routes in public transport. This rule requires that the user has previously defined criteria for each attribute. Tversky (1972) proposes a stage-by-stage process. At each stage, an attribute is selected and evaluated for each alternative of the choice. Alternatives that do not meet the criterion on the specific attribute are rejected. Tversky (1972) suggested that the first attribute examined is the most important one for the decision-maker. For example, in a choice between several alternative PT routes, if for the user, the most important attribute is the travel time, the user will first review each alternative on the basis of its travel time. This rule proposes that the multi-attribute choice process is a conscious choice, as the decision maker has to establish cutoffs on criteria, and compare alternatives based on those cut-offs. This assumption seems in accordance with the results obtained in PT route choice literature. Indeed, after a choice, users can easily list the important attributes, and the criteria they used for their decision (Raveau et al., 2014), or explain their reasoning (Grison et al., 2016). The EBA rule also proposes that the comparison process between alternatives begin by the most important attribute for the user. This aspect of the rule involves that users have preferences towards specific attributes, and do not have all the same criteria, as it has been shown in PT route literature (Chowdhury et al., 2014; Grison et al., 2016). For these reasons, the EBA rule appears to be a valid theoretical framework to explain PT routes choices. Considering the EBA rule, an hypothesis named the ‘‘prominence hypothesis” has been put forward (Slovic, 1975) which can be formulated as follows: decision-makers should choose the alternative with the highest value on the most important attribute. This hypothesis has been tested in various scenarios. For example, in one of his studies, Slovic (1975, experiment 2) presented participants with various choices between two driving route alternatives, each composed of two attributes (distance, and time). The participants had previously equally valued the alternatives. After making their choice, participants were asked to report what was the most important attribute for them. Results showed that participants tended to choose the alternative for which the value of the most important attribute was higher and thus confirmed the underlying hypothesis. This hypothesis has been validated in a wide variety of contexts, such as choosing between two baseball players, television commercials, secretarial applicants (Slovic, 1975; Slovic, 1995; Tversky, Sattath, & Slovic, 1988), but to our knowledge no other studies have investigated this hypothesis in the PT route choice field. To sum up, this body of research shows that the decision making process is not only based on a rational choice consisting in a systematic and detailed comparison of all alternatives -, but rather this process seems subjective and based on the decision maker preferences. Bringing together this theoretical framework and previous works on PT route choice, we expect that preferences of PT users – determined by user profile and context - should influence significantly the kind of attribute favored, and consequently the chosen route. 3. Research question While a better understanding of human PT route choice can be promising to improve current route planning aids, and consequently favor the use of PT, we still do not know much about how users solve such a task. We investigated this question by studying PT route choice in light of the ‘‘prominence hypothesis”, and considering the influence of individual as well as contextual factors. Participants who were all PT users, were presented with a 2 alternatives forced choice paradigm, such as those proposed in the multi-attribute decision-making field (e.g., Slovic, 1975). The choices had to be made between two PT routes described on two major attributes. One of the attributes was one of the most important instrumental attributes identified when considering PT routes (i.e., number of transport modes); the second was one of the affective attributes previously evoked in PT route choice literature (i.e., comfort). Participants had to make a choice in different contexts (i.e., various aims and durations of the trip), and certain individual characteristics (e.g., sociodemographic, habits, attitude towards PT) were

28

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

recorded to assess participants’ user profiles. Based on the literature reviewed above, we hypothesized that the PT route choice should be explained by the ‘‘prominence hypothesis” (Slovic, 1995; Tversky et al., 1988): participants would choose the alternative consistent with their preferences – i.e., the one that had the highest value on the most important attribute (i.e., instrumental or affective). Considering the recent results obtained in literature on PT routes choice, we also hypothesized that the user profile and the context would influence the preferences of users, and consequently their criterion (i.e., favoring instrumental or affective attributes). We thus should observe that the user profile characteristics and the context have an effect on the chosen route (Grison et al., 2016). 4. Method 4.1. Participants Sixty participants (16 males, 44 females), mostly students at Paris Descartes University, took part in the study in exchange for course credits. They were aged from 18 to 41 years old (M = 21.88, SD = 4.28). All participants were frequent users of public transport (i.e., PT) in the Île-de-France region (France). They all had a PT transit pass (i.e., pass Navigo) except for two of them, who bought tickets or traveled free of charge. Their average commuting time was 59.7 min (SD = 24.02). Amongst participants, 27 used a unimodal PT route (only one transport mode with or without transfer) to commute, and 33 used a multimodal PT route (at least two different transport modes). 4.2. Material In order to fulfill our objective - i.e., to better understand PT users’ route choice and the influence of context and user profile - we created numerous forced-choice situations as classically used in the multi-attribute decision-making literature (Slovic, 1975, 1995; Tversky et al., 1988). In order to observe whether some criteria were more or less favored in the decision, we contrasted each route choice on two attributes based on the two major kinds highlighted in the literature (Grison et al., 2016; Guo & Wilson, 2011): comfort (high or low) for the affective attribute, and the number of transport modes (one or three) for the instrumental attribute. To assess the effect of some contextual variables, we varied the overall length of the route, either long or short, and we created different realistic trip situations in which the decision had to be made in accordance with previous findings (Grison et al., 2016; going to work, to a leisure activity, or returning home). Finally, to assess user profile characteristics, we used different questionnaires and scales. 4.2.1. Route choices Each route choice was composed of two PT routes (see Fig. 1) described with respect to their level of comfort (i.e., affective attribute) and their number of transport modes (i.e., instrumental attribute). The two routes of each choice were systematically opposed on these two attributes, i.e., one route had only one transport mode (i.e., direct unimodal) but was uncomfortable, whereas the other one was comfortable but involved three transport modes (i.e., multimodal).

Fig. 1. An example of a choice.

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

29

We developed the route choice by creating eight different kinds of comfort information associated to different combinations of three transport modes; and eight different kinds of discomfort information associated to different single transport modes. The information about comfort concerned the presence or absence of factors such as seat availability, overcrowding, noise, and the possibility of reading or listening to music. Concerning the combinations of transport modes, we rotated the order of transport modes so that each transport mode occurred at least once at the beginning of the route, in the middle, and at the end (see Table 1 for the different combinations). Finally, the eight resulting choices were duplicated in two travel time conditions. Eight route choices were short (less than 30 min, in comparison to the Île-de-France average; Caenen, Courel, Paulo, & Schmitt, 2011) and eight were long (above 45 min). This gave 16 choices between two PT routes (see Fig. 1 for an example), eight in both travel time conditions, and for each one different kinds of information on comfort were provided and different combinations of transport modes were proposed. 4.2.2. Trip situations used as support to investigate route choice In order to propose trip situations close to the real situations that users experience in their daily life, we first created 27 different situations, they were based on the contexts highlighted by real users’ experience in the study by Grison et al. (2016). Therefore, the situations were created based on the following elements: – – – –

The The The The

aim of the trip: commuting to work or place of study, returning home, or going to a leisure place; time of day: morning, afternoon, or evening; weather: sunny, cloudy, or rainy; mood of the person: good, neutral, or bad.

The situations were proposed in the form of a small text and pictures representing the important elements of the sentence (see example in Fig. 2). A pre-test was conducted to select three situations among the 27 initial ones, one for each aim (commuting, returning home, leisure). Thirty-one participants who did not take part to the main experiment (12 males, M = 22.3 years old, SD = 4.8) were asked to judge the valence and the realism of the 27 situations on a 7-point Likert scale (1 = negative or not agree; 4 = neutral; and 7 positive or totally agree).

Table 1 Combinations of transport modes in route choices. Multimodal routes (three modes)

Unimodal routes (one mode)

1st mode

2nd mode

3rd mode

Subway Bus Train Tramway

Bus Subway Tramway Train

Tramway Train Subway Bus

Bus or Tramway Bus or Subway Subway or Train Tramway or Train

Fig. 2. Example of the presentation of one situation.

30

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

We chose scenarios with a neutral valence and a high level of realism in order to avoid any effect due to the valence of the scenario (notably because scenario deals with participant’s feelings, e.g., ‘‘you feel as usual”), or to its realism. The following three situations were thus selected. – Commuting: ‘‘You are commuting in the morning around 9 a.m. Today, you feel as usual, you’ve slept well and the weather is cloudy.” (Valence: M = 4.4/7, SD = 1.19; Realism: M = 5.9/7, SD = 1.07); – Returning home: ‘‘You are returning home at the end of the afternoon around 7 p.m. Nothing special happened during the day and the weather is cloudy.” (Valence: M = 4.5/7, SD = 0.87; Realism: M = 6.3/7, SD = 0.97); – Leisure: ‘‘You are going to your leisure activity in the afternoon. The weather is cloudy and you feel well.” (Valence: M = 5.3/7, SD = 0.69; Realism: M = 5.6/7, SD = 1.42). 4.2.3. Tests to assess user profile characteristics 4.2.3.1. Sense of direction scale. We used the Santa Barbara Sense-of-Direction scale (SBSOD; Hegarty et al., 2002) to measure participants’ global sense of direction (i.e., spatial preferences, spatial experiences, spatial and navigation abilities). It has been chosen because of the strong links observed between the results obtained at this scale and the real navigation behavior (Hegarty et al., 2006). This scale is composed of 15 items using a 7-point Likert scale (1 = not agree, 7 = totally agree). 4.2.3.2. Attitude towards public transport scale. This scale, developed to measure the perception that participants have of public transport, is an adaptation of the one used by Gandit (2007). This adaptation has already been previously used to assess users’ attitude towards public transport (Grison et al., 2016). It measures the global perception of public transport using 19 statements (5-point Likert scale, 1 = not agree, 5 = totally agree) focusing on advantages (12 statements) as well as disadvantages (7 statements) of public transport. It thus provides two scores: perceived advantages and perceived disadvantages. 4.2.3.3. Questionnaire on use of public transport. This questionnaire has been used in a previous study (Grison et al., 2016) in order to collect information about the frequency of use of PT, and about the PT routes that participants use to commute or to go to a leisure place. It also allowed us to assess the criteria that participants used when choosing PT routes. 4.3. Procedure Each participant was welcomed individually in a quiet room and the total duration of the experiment was around one hour. Once participants had given their informed consent, the experimenter described the choice task. They were told that their task was to choose between two PT routes presented on a laptop screen (15 min). We explained that for each trial they would see the two propositions and they had to choose the one they preferred. We then explained that they would have to make their choices in different situations. At the beginning of each session, a situation was presented on the screen; participants were asked to project themselves into the situation and to perform the following choices by imagining themselves in that situation. They were also informed that they would have all the time they needed to read the situation carefully. Once the participants had understood the instructions, they were shown an example. The training situation was first presented. Participants took all the time they needed to imagine themselves in the situation. When they were ready, they pressed the SPACEBAR and the first choice was displayed. If they chose the PT route on the left they had to press the ‘‘S” key on the keyboard (AZERTY); if they chose the route on the right they had to press the ‘‘L” key. In the training phase, 4 choices were proposed. After the training phase, the test phase began. Each participant performed the 16 choices in all three situations (commuting, returning home and leisure), making a total of 48 choices. The order of the situations and choices was randomized. The position of the PT routes in the choice (either left or right) was counterbalanced across participants. When participants had performed all the trials in the choice task, they were asked to fill in the various scales and questionnaires (SBSOD, attitude towards PT and questionnaire about the use of PT).

5. Data analysis 5.1. Scales and questionnaire 5.1.1. Sense of direction scale For each item, the participants rated a score from 1 to 7. The scores of the 15 statements were summed to obtain a global score out of 105. The higher the score, the better the participant’s sense of direction. 5.1.2. Attitude towards PT scale In this 19 items scale, 12 items (statements) refer to PT advantages, and 7 to disadvantages. For each statement, the participant responds on a 5 points Likert scale and obtains a score from 1 to 5. Two scores are calculated from this scale: one

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

31

score describes the perceived advantages (12 items, score out of 60), and one the perceived disadvantages (7 items, score out of 35). The higher the score, the more advantages/disadvantages with PT the participant perceives. 5.1.3. Public transport use questionnaire We assessed the frequency of public transport use by summing three scores, one describing the use of public transport to go to work, one to go to a leisure place during the week, and one to go to a leisure place during the week-end. Each one was measured with a 5-point Likert scale (total score out of 15). We called this variable PT_Frequency. We also recorded the usual type of route taken to commute: unimodal (only one transport mode) or multimodal (multiple transport modes). We called this variable Usual_Mode. Finally, we asked participants in the questionnaire if they usually took into consideration the number of transfers or the comfort of the transport mode to make their route choice. We called these variables Transfers_Criteria and Comfort_Criteria. 5.2. Route choice analysis We used a binary code for the choice task responses. If participants chose the comfortable PT route involving three transport modes it was coded ‘‘1”, whereas if they chose the uncomfortable PT route with only one mode it was coded ‘‘0”. We used the Chi2 test to analyze whether the proportion of choices towards one or another alternative changed according to the length of the route or the context of the decision. In order to explore whether our contextual and individual variables could explain the route choice, we ran a logistic binÒ Ò ary regression (IBM SPSS Statistics 21). This analysis is widely used in psychological research to explain binary dependent variables (Adler & Kuskowski, 2003; Bryden, Charlton, Oxley, & Lowndes, 2013; Preux, Odermatt, Perna, Marin, & Vergnenègre, 2005). Moreover, it was recently used to explain user route choice by individual variables (Chowdhury et al., 2014). The current literature does not provide evidence to support an hypothesis about the order of entry of predictive variables. Consequently, we have adopted an explorative perspective and run an exploratory stepwise logistic binary regression based on the likelihood ratio to explore the relative importance of our predictive variables. The predictive variables used in the analysis were as follows: – – – – – – – – – – –

Sex: female (0), male (1); Age Length of PT route: short (0), long (1); Context: commuting (0), returning back home (1), leisure (2); SBSOD (score out of 105); Perceived PT advantages (score out of 60); Perceived PT disadvantages (score out of 35); PT_Frequency (score out of 15); Comfort_Criteria: not important (0) or important (1) Transfers_Criteria: not important (0) or important (1) Usual_Mode: unimodal (0) or multimodal (1). Ò

The Context variable was transformed into three binary variables by the SPSS software. They were: Context_0 (Work = 1; Others = 0), Context_1 (Return = 1; Others = 0) and Context_2 (Leisure = 1; Others = 0). 6. Hypotheses According to the ‘‘prominence hypothesis” (Slovic, 1995; Tversky et al., 1988), we should observe that participants who usually choose their route according to the comfort criterion choose the comfortable alternative more frequently. Conversely, participants using the criterion of the minimum number of transfers should prefer the alternative with only one transport mode although it has a low level of comfort. Concerning the effect of context, we hypothesized that participants would choose the comfortable alternative more often when going to a leisure activity or returning home. The reverse effect should be observed for commuting. We also hypothesized that the longer the route is, the more participants will choose the comfortable alternative. Based on previous studies (Chowdhury et al., 2014; Grison et al., 2016; Hannes et al., 2009; Rüetschi & Timpf, 2005), our hypotheses concerning the effect of the various dimensions of user profile are as follows: the comfortable alternative involving three transport modes should be chosen by participants who find a lot of advantages with public transport, and few disadvantages, by participants who frequently use PT, who are used to multimodal routes, and who have a good sense of direction. The older the participants are, the more they should choose this alternative too. We also hypothesized that men would choose this alternative more than women. Finally, concerning the relative importance of our variables of interest to explain the route choice, as travel time has been shown to be a highly important variable in transport mode choice (Litman, 2008), current literature only allows us to hypothesize that the length of the route would be one of the first variables integrated in our model.

32

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

7. Results 7.1. Changes in route choices according to contextual variables Among the 2880 choices recorded, 1082 (37.6%) selected the uncomfortable alternative without transfer, and 1798 (62.4%) the comfortable alternative involving two transfers. This result shows that in general participants prefer to choose the comfortable alternative even if it involves two transfers. However, as presented in Table 2, the choice of one or the other alternative differed depending on the length of the route (Chi2(1) = 185.52, p < 0.001). While the proportion of choices between the two alternatives was more or less the same when the route was short (49.9% and 50.1%), participants showed a clear preference for the comfortable alternative with two transfers for long routes (25.3% vs. 74.17%). The context i.e. the purpose of the trip like going to a leisure activity, to work or home, does not affect significantly the proportion of choices between one or the other alternative (see Table 3, Chi2(2) = 2.1, p = 0.35). 7.2. Description of the user profile variables Table 4 presents the descriptive statistics of the numerical variables. Two of our variables do not follow the normality: age and frequency of PT use. Concerning age, this is due to the fact that most of our participants were young, from 18 to 25 years old (N = 54). Regarding the frequency of PT use, the result indicates that the majority of our participants frequently used PT to commute. Table 5 presents the frequency of use of the comfort or transfer criteria by our sixty participants. There were no significant differences between the use of the two criteria (v2(1) = 0.096, p = 0.76). 7.3. Route choice explained by the context and the user profile 7.3.1. Importance of variables in the decision In order to explore the relative importance of variables in the choice, we analyzed the order of entry of variables in the regression model. The stepwise logistic binary regression shows up eight steps. Each step significantly improves the model (see Table 6). Consequently, eight of the eleven variables significantly predict the choice (see Table 7): Length, Usual_Mode, Transfers_Criteria, PT Advantages, SBSOD, PT Disadvantages, sex, and Age. The significance and likelihood (omnibus tests of model coefficients) of the eight steps of the analysis and the corresponding models are presented in Table 6.

Table 2 Proportion of choices in percentage of one or the other alternative, depending on the length of the route. Length of the route

Choice

Transfer Comfort

Short

Long

49.9 50.1

25.3 74.7

Table 3 Proportion of choices in percentage between one or the other alternative, depending on the context of the route. Aim of the route

Choice

Transfer Comfort

Commute

Return

Leisure

38.3 61.7

35.7 64.3

38.6 61.4

Table 4 Descriptive statistics on the variables Age, SBSOD, PT advantages, PT disadvantages, and PT frequency for our 60 participants.

Age SBSOD PT advantages PT disadvantages PT Frequency

Mean (SD)

Min

Max

Median

Mode

Frequency of mode

Lower quartile

Upper quartile

Normality (Shapiro-Wilk)

21.9 63.2 36.5 19.5 12.5

18 27 21 12 6

41 95 54 29 15

21 67.5 37 19.5 13

19 73 38 Multiple 15

14 6 5 7 19

19.5 49.5 31 16 11

22 74.5 42 22 15

W = 0.06 W = 0.96 W = 0.98 W = 0.97 W = 0.88

(4.28) (17.9) (7.72) (4.32) (2.41)

(p < 0.001) (p = 0.08) (p = 0.69) (p = 0.20) (p < 0.001)

33

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37 Table 5 Frequency of comfort and transfer criteria reported by participants for their usual choice of PT routes (N = 60).

Yes No

Comfort criteria

Transfer criteria

25 35

35 25

Table 6 The eight steps and associated models (omnibus test of model coefficients), as well as their significance and their likelihood. Chi-square

df

p-value

Step 1

Step Model

188.134 188.134

1 1

0.000 0.000

3624.508

2 Log likelihood

Nagelkerke R square 0.086

Step 2

Step Model

184.405 372.539

1 2

0.000 0.000

3440.103

0.165

Step 3

Step Model

51.932 424.47

1 3

0.000 0.000

3388.171

0.187

Step 4

Step Model

23.271 447.741

1 4

0.000 0.000

3364.901

0.196

Step 5

Step Model

15.029 462.769

1 5

0.000 0.000

3349.872

0.202

Step 6

Step Model

16.168 478.938

1 6

0.000 0.000

3333.704

0.209

Step 7

Step Model

5.378 484.316

1 7

0.002 0.000

3328.325

0.211

Step 8

Step Model

5.902 490.218

1 8

0.015 0.000

3322.424

0.213

Table 7 OR (odds-ratio), confidence interval and P-Value of variables in the equation at step 8.

Length Usual_Mode Criteria_Transfers PT Advantages SBSOD PT Disadvantages Sex Age

OR

95% confidence interval

P-Value

Entry step

3.338 2.960 0.506 1.016 0.992 0.946 0.774 1.029

2.820–3.950 2.385–3.673 0.414–0.617 1.004–1.029 0.987–0.997 0.923–0.970 0.627–0.955 1.005–1.053

0.000*** 0.000*** 0.000*** 0.012* 0.001*** 0.000*** 0.017* 0.018*

1 2 3 4 5 6 7 8

At the first step, the analysis included the Length variable (OR = 2.940, CI = 2.511–3.441, p < 0.001). This result indicates that the Length variable is the most important in predicting the choice of route, and confirms our hypothesis. Concerning the order of entry of other variables, for which we did not set up predictions, the Usual_Mode variable was added after the Length (OR = 3.018, CI = 2.564–3.552, p < 0.001), followed by the Transfers_Criteria variable included at step three (OR = 0.497, CI = 0.410–0.601, p < 0.001). At the fourth step, the analysis included the PT Advantages variable (OR = 1.028, CI = 1.017–1.040, p < 0.001). The variable SBSOD was then included at step five (OR = 0.991, CI = 0.986–0.995, p < 0.001), followed by the PT disadvantages variable at step six (OR = 0.954, CI = 0.932–0.976, p < 0.001). The variable Sex was added at step seven (OR = 0.779, CI = 0.631–0.962, p = 0.020), and finally the Age variable at step eight (OR = 1.029, CI = 1.005–1.053, p = 0.018). Five variables are out of the equation at this final step: Context_0 (p = 0.286), Context_1 (p = 0.115), Context_2 (p = 0.357), PT_Frequency (p = 0.548), and Comfort_Criteria (p = 0.105). 7.3.2. Influence of context and of user profile on the decision In order to verify our specific hypothesis on each variable, we analyzed the effect of each significant variable. Table 7 presents the significant variables in the model at the final step in the order of inclusion. The results of the variables Length, attitude towards PT, Transfers_Criteria, Usual-Mode and age are in line with our prediction. Indeed, the longer the route is, the more participants choose the comfortable alternative (OR = 3.338, CI = 2.820– 3.950, p < 0.001), and the more accustomed participants are to multimodal routes, the more they choose the comfortable

34

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

alternative (OR = 2.960, CI = 2.385–3.673, p < 0.001). Moreover, the more participants take into consideration the number of transfers in their daily life, the less they choose the comfortable alternative (OR = 0.506, CI = 0.414–0.617, p < 0.001). Concerning the effect of attitudes towards PT, the more the participants find advantages with PT, the more they choose the comfortable alternative (OR = 1.016, CI = 1.004–1.029, p = 0.012), and the more the participants find disadvantages with PT, the less they choose the comfortable route with three transport modes (OR = 0.946, CI = 0.923–0.970, p < 0.001). As expected, the more accustomed the participants are to taking multi-modal routes, the more they choose the comfortable route involving three transport modes (OR = 2.96, CI = 2.385–3.673, p < 0.001). Finally, the older the participants are, the more they choose the comfortable alternative that involves two transfers (OR = 1.029, CI = 1.005–1.053, p = 0.018). However, the effects of the sense of direction and of the sex contradict our hypotheses. In fact, the higher their SBSOD score is, the less participants choose the comfortable alternative (OR = 0.992, CI = 0.987–0.997, p = 0.001). Results also show that men are significantly less inclined to choose the comfortable PT route than women (OR = 0.774, CI = 0.627–0.955, p = 0.017).

8. Discussion The aim of the present study was to better understand how users choose their route in a public transport network. More particularly, we have tested if the ‘‘prominence hypothesis” (Slovic, 1975) can explain the PT route choice with a focus on the effects of context as well as of the user profiles on the decision. To do so, we asked participants to choose between two PT routes contrasted on two kinds of attribute: an instrumental attribute (i.e., the number of transport modes), and an affective attribute related to comfort (e.g., be seated or not). We varied the context of the decision (length, and aim of the route), and we recorded various individual characteristics (attitudes towards PT, use of PT, age, sex, etc.). One main issue of the study was to test the ‘‘prominence hypothesis” (Slovic, 1975; Tversky et al., 1988) in the context of PT route choice. Interestingly, our results validate part of this hypothesis: participants who reported on the questionnaire that the instrumental attribute (i.e., number of transfers) was usually the most important attribute they considered when choosing a route, preferred to choose the route with the highest value on this attribute. This result provides, for the first time, arguments to consider the ‘‘prominence hypothesis”, and consequently the Elimination-by-Aspects rule (Slovic, 1975; Tversky, 1972) as a valuable theoretical framework to explain PT route choices. However, this effect rule does not hold when applied to the affective attribute: people who usually consider the comfort as the most important attribute did not prefer the route with the highest level of comfort. This can be explained by the way comfort was defined in our scenarios. Indeed, based on the results of our previous research (Grison et al., 2016), we defined the comfort as the availability of seats or the quietness in transport. They both refer to the physical comfort that makes the travel enjoyable. In the literature focusing on user experience, however, comfort has mainly been defined in a global way, as for example in the definition of emotional comfort (Allinc, Cahour, & Burkhardt, 2015; Cahour, 2008). Emotional comfort is defined as a global feeling, which is dynamically constructed through the affective states (e.g., confidence, surprise, fear) experienced during a specific activity that involves the body and the mind. In this respect, routes with transfers can be perceived as uncomfortable because of their difficulty, and because they can produce a feeling of loss of control for users (Hine & Scott, 2000). When we asked participants if they based their choices of route on the comfort criterion, the nature of the comfort was not specified. Our result can thus be explained by the fact that participants considered emotional comfort in their final evaluation, and not the physical comfort in transport mode as designed in our choice task. Future work should better specify the type of comfort assessed to better understand its effect, as it might be an important affective attribute of PT routes choice that we have underestimated here (Allinc et al., 2015; Grison et al., 2016). A second major hypothesis of our study was that the context of the decision should influence the preference of users. Results show that our participants preferred the affective rather than the instrumental attribute when making their choices. The length of the route tended to strengthen this result: participants prefer the comfortable alternative for longer routes. These results reveal a strong effect of one contextual factor when people choose their route. However, the aim of the trip did not influence the choice between instrumental or affective attributes, while previous work has shown a strong effect of this factor on route choice (Grison et al., 2016). Our study also aimed to highlight the effect of user profile on PT route choice. Our results showing that habits, attitudes towards PT, sense of direction, age, and sex influences the decision of our participants, confirm our hypothesis. More interestingly, our results analysis allows us to explore the relative importance of each of these user profile characteristics. Thus, the first user profile variable included in the model was the participants’ travel mode habit (unimodal vs multimodal route). Results show that participants who were used to multimodal routes were more prone to choose the route with three transport modes that was comfortable. The fact that this variable was the first user profile characteristic included in the model reveals the strong impact of habits in PT route choice. It extends, in the particular field of PT route choice, years of research in psychology showing the importance of habits on choice (see for a review, Ajzen, 2011 or Armitage & Conner, 2001). The next two dimensions of the user profile characteristics included in the model were the perceived advantages and disadvantages of PT. Results demonstrate the importance of the user’s attitude towards PT during route choice. While the influence of attitude on transport mode choice has been largely demonstrated, until now little attention had been paid

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

35

to its effect on PT route choice. Our results confirm those obtained by Grison et al. (2016), and show that having a positive attitude towards PT makes participants considering the affective attribute, and not only the instrumental one. The sense of direction was then included in the model. Including this characteristic in the user profile was an original feature of our study. Indeed, it has been shown in the literature that transfers are one of the most negative aspects of PT, especially because they involve navigational skills to find the way at transfers (Timpf, 2002). Following this line of results, we suggested that the better the sense of direction that participants had, the more they would favor the comfort of the route even if it required two transfers. Our results, however, do not validate this hypothesis, and even show the opposite effect. An explanation comes from the literature that has shown interaction effects between several variables when choosing a transport mode or a route (Grison et al., 2016; Klöckner & Friedrichsmeier, 2011). In the present study the choices were contrasted on two attributes: the level of comfort, and the number of transport modes. Moreover, if one of the two attributes was negative, the other one was positive. Thus, when making their choice, participants may have favored the affective aspect of the route rather than the instrumental one, and consequently they chose the comfortable route involving two transfers. It is also possible that other variables that have a stronger impact on the decision, such as habit or attitude, oriented the choice more than the sense of direction. This assumption is in accordance with our results showing that habits and perceived PT advantages were included before the sense of direction in the model. The last two individual variables added to the model were sociodemographic – i.e., age and sex. Whereas the effect of age confirms the one obtained by Grison et al. (2016) – i.e., the older the participants, the more they choose the affective attribute -, results on sex are however in contradiction with our hypothesis, since men chose the comfortable alternative requiring two transfers less than women. This result is however in line with the one obtained by Chowdhury et al. (2014) showing that women tend to choose routes with transfers more than men. Taking these results all together, it appears that women facing fictitious choices, even if these choices are realistic, tend to choose routes with transfers more than men. In studies on users’ real experience, however, men tend to choose routes with transfers more than women (Grison et al., 2016). Another explanation lies in the fact that the alternative with two transfers was also the more comfortable one. It has been largely shown that women rely more on emotion than men (see Brody & Hall, 2010 for a review); consequently they may have paid more attention to the comfortable aspect of the route than to the number of transport modes. We did not observe however any effects of the PT use frequency on choice. Chowdhury et al. (2014) showed that the more participants are frequent users of PT, the more they are inclined to use transfers. The absence of effect in our study can be explained by the fact that the majority of our participants were frequent users of PT for multiple purposes. Further investigation will be needed to better target the effect of this variable on PT route choice. We acknowledge that the fact to use hypothetical scenarios might be a shortcoming. This may explain some of our results in contradiction with our hypotheses, especially the effect of the aim of the travel. Indeed, even if we took care to propose realistic trip situations, the participants were aware of the laboratory-testing context, and our methodology does not guarantee that participants were fully immersed in the situation. This procedure ensures the control of many variables that are not considered in some other studies, and it allows highlighting some results that are sometimes difficult to observe in real situations due to the various factors involved. We might however question its validity as regards the choices made in real situations. Immersive environments, such as field studies, could be a way to better assess the effects of context. Another point is that we chose to focus on the information about the comfort in transport mode, because it has been proven a relevant factor, but to date this kind of information is not present in navigational aid applications, so this factor is most of the time unpredictable in reality. Nevertheless, despite differences between our scenarios and reality, our results are coherent with results obtained from real user experience (Grison et al., 2016). It is thus important in future to conduct both kinds of research, and to always compare the results from real and fictitious situations. Nevertheless, all our results taken together bring new evidences that solving a multi-attributes choice is not only based on rational, and on an analytical comparison between the various options, but that preferences, past experiences, habits, gender, etc. modulate the choice and tend to interfere with this choice process. Thus, our results extend some previous results of the literature concerning transport mode choices, showing that habits interfere with the choice process (Aarts, Verplanken, & van Knippenberg, 1997; Verplanken, Aarts, & Van Knippenberg, 1997). When users have strong habits they do not go through the entire comparison process, but choose according to their habits. Our results suggest that other variables such as past experiences or preferences can lead to the same kind of mechanism. This last assumption is supported by the ‘‘prominence hypothesis” suggesting that the choice process is strongly influenced by user preferences (Slovic, 1975; Tversky et al., 1988). Questions that arise from our results concern the relative importance and the relation between habits, past experiences, and preferences in the particular context of PT route choices. We can hypothesize that through repeated experiences users will develop habits for routes taken frequently, but also develop global preferences regardless of some of the routes attributes. Then, depending on the context of the decision (usual route or new one), users will rely on their specific habits, or their general preferences. 9. Conclusion To sum up, our study provides new data on the under-investigated question of route planning in PT. On the theoretical side, our results confirm, in the context of route choice, that the decision is not always rational, but based on preferences. This result allows a first step to the theorization of PT route choice in regards to actual multi-attributes choices theories.

36

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

Our results also highlight that the choice between two PT routes contrasted in two major kinds of attribute is determined by both contextual factors and user profile. We also show that for some people and in some contexts, a route with a high affective value is favored over instrumental considerations. This result is new in the field of PT route choice that has, so far, mostly focused on instrumental aspects. We have also explored the order of importance of the factors which account for a PT route choice. This analysis reveals that some contextual factors such as the length of the route are the ones most often taken into consideration when making a decision. The type of usual route also appears to be highly important, and reveals that habits are strong predictors of choice. Other individual factors such as attitudes towards PT or the age and sex of people play a role in the choice. This last finding suggests that, even if these factors are less important in the choice, user profile should be taken into consideration to explain PT route choices. Our results also provide new insights for the body of literature that aims to establish models of PT route planning in order to improve current route planning aids. In most cases, current route planning applications describe routes by their instrumental attributes (e.g., the shortest path, the minimum number of transfers, and the walking time). They do not, however, consider affective aspects or factors related to the context or to individual characteristics. The integration of factors such as the context, habits, attitudes or of attributes such as overcrowding in the transport mode in route planning applications may help to better target users’ expectations, and consequently help to propose routes that are better adapted to their needs.

References Aarts, H., Verplanken, B., & Knippenberg, A. Van (1997). Habits and information use in travel mode choice. Acta Psychologica, 96(3), 1–14. http://dx.doi.org/ 10.1016/s0001-6918(97)00008-5. Abelson, R. P., & Levi, A. (1985). Decision making and decision theory. Handbook of Social Psychology, 1, 231–309. Adler, G., & Kuskowski, M. (2003). Driving cessation in older men with dementia. Alzheimer Disease & Associated Disorders, 17(2), 68–71. Ajzen, I. (2011). The theory of planned behaviour: Reactions and reflections. Psychology & Health, 26(9), 1113–1127. Allinc, A., Cahour, B., & Burkhardt, J.-M. (2015). Psychological comfort and discomfort in transport modes. Proceedings of the European Congress of Cognitive Ergonomics, 8–10. http://dx.doi.org/10.1145/2788412.2788435. Armitage, C. J., & Conner, M. (2001). Efficacy of the theory of planned behaviour: A meta-analytic review. British Journal of Social Psychology, 40(4), 471–499. Ben-Elia, E., Di Pace, R., Bifulco, G. N., & Shiftan, Y. (2013). The impact of travel information’s accuracy on route-choice. Transportation Research Part C: Emerging Technologies, 26, 146–159. Bettman, J. R., Johnson, E. J., & Payne, J. W. (1990). A componential analysis of cognitive effort in choice. Organizational Behavior and Human Decision Processes, 45(1), 111–139. Bovy, P. H., & Hoogendoorn-Lanser, S. (2005). Modelling route choice behaviour in multi-modal transport networks. Transportation, 32(4), 341–368. Brisbois, X. (2010). Le processus de décision dans le choix modal: Importance des déterminants individuels, symboliques et cognitifs. France: Université de Grenoble [Thèse de doctorat]. Brody, L. R., & Hall, J. A. (2010). Gender, emotion, and socialization. In J. C. Christer & D. R. McCreary (Eds.), Handbook of gender research in psychology. Gender research in general and experimental psychology (vol. 1, pp. 429–454). New York, NY: Springer. Bryden, K. J., Charlton, J. L., Oxley, J. A., & Lowndes, G. J. (2013). Self-reported wayfinding ability of older drivers. Accident Analysis & Prevention, 59, 277–282. Caenen, Y., Courel, J., Paulo, C., & Schmitt, D. (2011). Les Franciliens utilisent autant les transports en commun que la voiture pour se rendre au travail. Île-deFrance à la, 353. Cahour, B. (2008). Discomfort, affects and coping strategies in driving activity. In Proceedings of the 15th European conference on cognitive ergonomics the ergonomics of cool interaction. http://dx.doi.org/10.1145/1473018.1473046. Chiu, D. K., Lee, O. K., Leung, H. F., Au, E. W., & Wong, M. C. (2005). A multi-modal agent based mobile route advisory system for public transport network. In Proceedings of the 38th Annual Hawaii International Conference on System Sciences, 2005. HICSS’05 (pp. 92b–92b). Chowdhury, S., & Ceder, A. (2013). A Psychological investigation on public-transport users’ intention to use routes with transfers. International Journal of Transportation, 1(1), 1–20. http://dx.doi.org/10.11648/j.pbs.20130201.12. Chowdhury, S., Ceder, A., & Sachdeva, R. (2014). The effects of planned and unplanned transfers on public transport users’ perception of transfer routes. Transportation Planning and Technology, 37(2), 154–168. Dieleman, F., Dijst, M., & Burghouwt, G. (2002). Urban form and travel behaviour: Mirco-level household attributed and residential context. Urban Studies, 39 (3), 507–527. http://dx.doi.org/10.1080/0042098022011280. Fontaine, S., & Denis, M. (1999). The production of route instructions in underground and urban environments. Spatial information theory. In Cognitive and computational foundations of geographic information science (pp. 83–94). Berlin Heidelberg: Springer. Friman, M. (2010). Affective dimensions of the waiting experience. Transportation Research Part F: Traffic Psychology and Behaviour, 13(3), 197–205. http://dx. doi.org/10.1016/j.trf.2010.04.006. Gandit, M. (2007). Déterminants psychosociaux du changement de comportement dans le choix du mode de transport: Le cas de l’intermodalité. France: Université Pierre Mendès-France-Grenoble II. Thèse de doctorat. Golledge, R. G. (1995). Path selection and route preference in human navigation: A progress report (pp. 207–222). Berlin Heidelberg: Springer. Grison, E., Gyselinck, V., & Burkhardt, J. M. (2016). Exploring factors related to users’ experience of public transport route choice: Influence of context and users profiles. Cognition, Technology and Work, 18(2), 287–301. Grison, E., Gyselinck, V., Burkhardt, J. M., & Wiener, J. M. (2016). Route planning with transportation network maps: An eye-tracking study. Psychological Research Psychologische Forschung, 1–15. Guo, Z., & Wilson, N. H. M. (2011). Assessing the cost of transfer inconvenience in public transport systems: A case study of the London Underground. Transportation Research Part A: Policy and Practice, 45(2), 91–104. http://dx.doi.org/10.1016/j.tra.2010.11.002. Hannes, E., Janssens, D., & Wets, G. (2009). Does space matter? Travel mode scripts in daily activity travel. Environment and Behavior, 41, 75–100. Hegarty, M., Montello, D. R., Richardson, A. E., Ishikawa, T., & Lovelace, K. (2006). Spatial abilities at different scales: Individual differences in aptitude-test performance and spatial-layout learning. Intelligence, 34(2), 151–176. Hegarty, M., Richardson, A. E., Montello, D. R., Lovelace, K., & Subbiah, I. (2002). Development of a self-report measure of environmental spatial ability. Intelligence, 30(5), 425–447. Heye, C., & Timpf, S. (2003). Factors influencing the physical complexity of routes in public transportation networks. In Electronic proceedings of the 10th international conference on travel behaviour research, Lucerne. Hine, J., & Scott, J. (2000). Seamless, accessible travel: users’ views of the public transport journey and interchange. Transport Policy, 7(3), 217–226. Hölscher, C., Tenbrink, T., & Wiener, J. M. (2011). Would you follow your own route description? Cognitive strategies in urban route planning. Cognition, 121 (2), 228–247. Huber, G. P. (1974). Multi-attribute utility models: A review of field and field-like studies. Management Science, 20(10), 1393–1402. http://dx.doi.org/ 10.1287/mnsc.20.10.1393.

E. Grison et al. / Transportation Research Part F 51 (2017) 24–37

37

Hunecke, M., Blöbaum, A., Matthies, E., & Höger, R. (2001). Responsibility and environment ecological norm orientation and external factors in the domain of travel mode choice behavior. Environment and Behavior, 33(6), 830–852. http://dx.doi.org/10.1177/00139160121973269. Klöckner, C. A., & Friedrichsmeier, T. (2011). A multi-level approach to travel mode choice–how person characteristics and situation specific aspects determine car use in a student sample. Transportation Research Part F: Traffic Psychology and Behaviour, 14(4), 261–277. http://dx.doi.org/10.1016/j. trf.2011.01.006. Litman, T. (2008). Valuing transit service quality improvements. Journal of Public Transportation, 11(2), 43–63. Mann, E., & Abraham, C. (2006). The role of affect in UK commuters’ travel mode choices: An interpretative phenomenological analysis. British Journal of Psychology, 97(2), 155–176. http://dx.doi.org/10.1348/000712605X61723. Montgomery, H., & Svenson, O. (1976). On decision rules and information processing strategies for choices among multiattribute alternatives. Scandinavian Journal of Psychology, 17(1), 283–291. Payne, J. W., Bettman, J. R., & Johnson, E. J. (1988). Adaptive strategy selection in decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 14(3), 534. Popuri, Y., Proussaloglou, K., Ayvalik, C., Koppelman, F., & Lee, A. (2011). Importance of traveler attitudes in the choice of public transportation to work: Findings from the regional transportation authority attitudinal survey. Transportation, 38(4), 643–661. http://dx.doi.org/10.1007/s11116-011-9336-y. Preux, P. M., Odermatt, P., Perna, A., Marin, B., & Vergnenègre, A. (2005). Qu’est-ce qu’une régression logistique? Revue des maladies respiratoires, 22(1), 159–162. Raveau, S., Guo, Z., Muñoz, J. C., & Wilson, N. H. M. (2014). A behavioural comparison of route choice on metro networks: Time, transfers, crowding, topology and socio-demographics. Transportation Research Part A: Policy and Practice, 66, 185–195. http://dx.doi.org/10.1016/j.tra.2014.05.010. Raveau, S., Muñoz, J. C., & de Grange, L. (2011). A topological route choice model for metro. Transportation Research Part A: Policy and Practice, 45(2), 138–147. http://dx.doi.org/10.1016/j.tra.2010.12.004. Rubens, L., Gosling, P., & Moch, A. (2011). Favoriser le report modal: Connaître les raisons liées au choix d’un mode de déplacement pour le changer. Pratiques Psychologiques, 17(1), 19–29. http://dx.doi.org/10.1016/j.prps.2009.12.002. Rüetschi, U. J., & Timpf, S. (2005). Modelling wayfinding in public transport: Network space and scene space. In Spatial cognition IV. Reasoning, action, interaction (pp. 24–41). Berlin Heidelberg: Springer. Russo, J. E., & Dosher, B. A. (1983). Strategies for multiattribute binary choice. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9(4), 676. Slovic, P. (1975). Choice between equally valued alternatives. Journal of Experimental Psychology: Human Perception and Performance, 1(3), 280. Slovic, P. (1995). The construction of preference. American Psychologist, 50(5), 364. Steg, L. (2005). Car use: lust and must. Instrumental, symbolic and affective motives for car use. Transportation Research Part A: Policy and Practice, 39(2), 147–162. Svenson, O. (1979). Process descriptions of decision making. Organizational Behavior and Human Performance, 23(1), 86–112. Timpf, S. (2002). Ontologies of wayfinding: A traveler’s perspective. Networks and Spatial Economics, 2(1), 9–33. http://dx.doi.org/10.1023/ A:1014563113112. Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79(4), 281. Tversky, A., Sattath, S., & Slovic, P. (1988). Contingent weighting in judgment and choice. Psychological Review, 95(3), 371. Verplanken, B., Aarts, H., & Van Knippenberg, A. (1997). Habit, information acquisition, and the process of making travel mode choices. European Journal of Social Psychology, 27(5), 539–561. Vredin Johansson, M., Heldt, T., & Johansson, P. (2006). The effects of attitudes and personality traits on mode choice. Transportation Research Part A: Policy and Practice, 40(6), 507–525. http://dx.doi.org/10.1016/j.tra.2005.09.001. Zanakis, S. H., Solomon, A., Wishart, N., & Dublish, S. (1998). Multi-attribute decision making: A simulation comparison of select methods. European Journal of Operational Research, 107(3), 507–529.