Developing a user typology considering unimodal and intermodal mobility behavior: a cluster analysis approach using survey data

The mobility behavior of people living in cities varies greatly. Cities offer a wide range of different mobility options and people are confronted with multi-optionality in their everyday lives. In large cities in particular, users can choose from a wide variety of different transport modes, including a dense public transport network and good conditions for walking and cycling [4, 8, 34]. To manage the urban transport system for the benefit of people living in cities, it is important to gain a comprehensive understanding of users’ mobility behavior and their characteristics. Using different forms of grouping procedures, a number of different mobility user typologies have already been developed to make the mobility behavior of highly diverse individuals more applicable and suitable for subsequent analyses [2, 17, 18, 22, 33, 45, 46].

Mobility and mobility behavior in cities is very complex. The wide variety of transport alternatives in addition to the density and mix of uses in urban areas provide a good basis for people to use and combine different modes of transport in their everyday mobility in a flexible, individual and situational way [15, 32]. Using different modes over the course of a week (multimodality) [1, 39] or within one trip (intermodality) accounts for a considerable percentage of the total number of trips undertaken, especially in cities [10, 23]. Regarding trips, we use the term unimodal trip for the use of one transport mode on a single trip for one trip purpose. In contrast, an intermodal trip is characterized by the combination of different transport modes on a single trip for one trip purpose [28]. Intermodal mobility behavior has also been analyzed empirically [23, 41‐43, 51]. These studies have shown that the share of intermodal mode choice at the modal split is quite low by comparison with unimodal usage. Nevertheless, the results demonstrate that intermodal combinations are a relevant option for many people in their everyday mobility portfolio. Furthermore, combining intermodal and unimodal modes in everyday life, according to different situations and different trip purposes, varies widely. Intermodal mobility behavior and intermodal users therefore merit a differentiated view. There is not merely one universal intermodal user because mobility behavior varies. A typology of users can be helpful for addressing the diversity of intermodal mobility behavior.

Both intermodal and multimodal mobility behavior are discussed as being crucial for a more efficient and sustainable urban transport system [6, 10, 29, 38]. In this context, it is important to understand the characteristics, background and logic behind this varied mobility behavior from the user’s perspective. We have to look at intermodal and unimodal use within the overall context of the individual’s mobility behavior and we must also take both intermodal and unimodal mode use into account in user typologies to enable us to understand the mobility behavior exhibited by different types of users. Since intermodal trips are less frequent than unimodal trips, typologies based on travel surveys using fixed reference dates often fail to provide information about intermodal mobility behavior. The few existing studies on intermodal users focus mainly on a specific means of transport (e.g. [37] for bike & car-sharing). At present, there is no systematic segmentation of the full range of different intermodal users. This paper aims to provide a user typology that combines intermodal and unimodal mobility behavior in an effort to obtain an overview of intermodal users within the context of their overall mobility behavior.

The common feature of many segmentation approaches established in transport research is that they usually categorize individuals with a certain travel behavior which can then be used to develop user-specific measures. For this purpose, it is necessary to identify segmented typologies with which the behavior of different groups can be understood [7]. Today, segmentation approaches [36, 47] are an established means of analyzing daily travel determinants [31, 44, 45] and are used by different disciplines such as psychology (e.g. [22]), sociology (e.g. [25]) and also, increasingly, transport sciences (e.g. [17]). Transport providers and municipalities use market segmentation as a basis for targeted interventions to increase the use of sustainable transport modes [13].There are two methodological arguments which suggest the superiority of typologizing. The epistemological argument is based on the lack of sensitivity in linear analysis concepts to significant cause-effect relationships that are only detectable in subgroups of the total population. The pragmatic argument relates to improving the possibilities for communication between scientists and practitioners by identifying homogeneous groups so as to reduce the complexity of heterogeneous populations ([12, 22];).

The segmentation approaches and the methods applied (factor, cluster correspondence analysis or qualitative typology) differ depending on the research question and the object of investigation. In the field of transport research, segmentation studies have identified groups of people with similar conditions and travel behavior [17, 33] or attitudes [2, 25, 26, 45]. The work of Kutter [33], which introduced the concept of behavioral homogeneous groups, provided significant impetus for working with types that differ significantly from each other due to their socio-demographic characteristics, combined with their practiced mobility behavior. More recent approaches focus more on attitudinal characteristics (e.g. [2, 25, 45]). Although these psychographic segmentation approaches reveal an added value for the explanation of behavior, several studies have reinforced the focus on behavior-related characteristics since there are obvious differences in needs and orientations between users with different usage intensities [5, 11, 21]. Empirical evidence increasingly indicates the existence of higher-level mobility orientations that influence all dimensions of an individual’s mobility behavior [26, 48]. Vij et al. [48, 49] emphasize the existence of modality styles, or “behavioral predispositions, characterized by a certain mode or set of modes that an individual habitually uses” ([48]: 1). Modality styles such as the innermost component of the concept proposed by Vij et al. [48] are embedded in the larger concept of an individual’s mobility style and, ultimately, lifestyle ([41, 48, 49]).

Existing user typologies rarely consider intermodal mobility behavior. As a consequence, intermodality is not usually represented in common mobility types. So far, there has been no user typology that differentiates between intermodal users (e.g. bike + public transport, car + public transport) and also considers both intermodal and unimodal behavior. Reflecting the work of Vij et al. [48, 49], we use the construct of modality styles and operationalize this concept in our aim of identifying mobility types that incorporate both unimodal and intermodal mode use. Against this background, this study is in line with segmentation studies of mobility behavior that do not focus on one means of transport alone or only on the amount of use (e.g. [41, 48, 49]). We argue that analyzing the use of different travel modes in conjunction with the purpose of travel is extremely important for detecting differences in travel patterns [41]. Our goal is to identify a user typology from the sum of unimodal and intermodal travel behavior. The objective of this paper is to identify different mobility types in a first step and to describe the mobility types identified in more detail according to socio-demographic characteristics in a second step. In this way, it is possible to formulate highly illustrative enhanced mobility types (EMT) (unimodal and intermodal behavior, socio-demographic, resources, etc.). This provides a better understanding of intermodal mobility behavior from the user’s perspective and can help planners and practitioners to consider the requirements of different users and to formulate target-group-specific measures.

We address this issue in our paper and identify a user typology that includes unimodal and intermodal travel behavior. The user typology draws on a cluster analysis with empirical data from a survey which we conducted in Berlin in 2016. Against this background, the aims of the paper are:

Section 2 below provides an overview of the study design, including the empirical survey data and the methodological procedure using principal component analysis (PCA) and cluster analysis. Section 3 presents the results from the PCA and cluster analysis and the resulting mobility types. In section 4, we discuss the results in addition to the pros and cons of the methodological procedure. Finally, in section 5 the conclusion sums up the main findings and answers our research questions.

The user typology that is presented in this paper is based on empirical survey data. In this section we present the overall methodological procedure and general information about the survey we conducted, we describe the variables considered and explain the procedure of the cluster analyses.

This section illustrates the results of the applied methodology. This includes the outcomes of the clustering approaches (the intermodal and unimodal styles) and the subsequent combination of the resulting clusters (mobility types) as well as a detailed description of the enhanced mobility types (EMT) derived.

The procedure presented uses a two-step clustering approach to identify a wide range of user groups in respect of everyday mobility behavior with a special focus on intermodality. The results demonstrate that this methodology is suitable for including important behavioral aspects with low usage frequency in user types without losing an overall picture of travel behavior. This facilitates the ability to focus on forms of transportation that are not yet fully accepted. This was demonstrated for the topic of intermodality but can also be used for other forms of upcoming transport provision or changes in travel behavior, such as car-sharing or ride-sharing.

Within the resulting typology of user groups, each user is matched to one unimodal and one intermodal style which can be refined into a comprehensive enhanced mobility type (EMT) including all aspects of travel behavior (unimodal and intermodal), available mobility resources and socio-demographic characteristics. At the same time, the approach enables us to take a differentiated look at selected user types according to a specific research question and therefore forms a valuable basis for developing target-group-specific actions in urban and transport planning. Alternatively, certain user types can be combined to generate more aggregated target classes (e.g. the initial intermodal users and unimodal users or all users with a certain amount of car use), if needed. As a result, the mobility types are based on the user’s actual mobility behavior rather than on one means of transport alone or on socio-demographic characteristics. In contrast, existing mobility typologies often mix mobility behavior with socio-demographic characteristics or values and opinions in the classification process, in order to create predefined target-specific user groups [12, 22]. This leads to homogeneous population groups but does not allow for statements on specific, less frequent behavioral aspects. In our case, socio-demographic characteristics were explicitly not considered in the first step of the classification (step B in Fig. 2) to avoid the creation of merely socio-demographic groups (pupils, working people, the elderly, etc.). The aim was to place special emphasis on mobility behavior and avoid superimposing other attributes when creating the groups. Instead, socio-demographic characteristics and information about individual mobility resources were used later to describe enhanced mobility users (EMT) (step D in Fig. 2). This enabled a detailed view of individuals with a combination of specific intermodal and unimodal mobility behavior. At the same time, the high level of detail of each mobility type and the resulting large number of different mobility types may also be a limitation as it involves a more complex handling and makes application more challenging.

As a consequence of this methodological procedure, the most suitable mobility types can be selected and further investigated depending on a specific question from science or practice. Since intermodal usage of different modes of transport can be seen as a promising alternative to unimodal car usage (I1U5) [10], the three groups with a high proportion of intermodal behavior (I3U1, I5U3 and I6U5) are selected as an example and are further discussed in this section to show the practical value of this work. By understanding which kind of individuals are using intermodal combinations in which situations, these groups can be specifically addressed for further usages. For instance, we are able to determine that people who combine car and public transport (I3U1) are mostly retired, living in couple households or alone, in decentralized neighborhoods and are mostly part of higher age groups. This is very valuable information when planning and designing infrastructure for intermodal interchanges [14, 16] as it is possible to provide features that are adapted to the specific needs of the corresponding user group. In contrast, the group of people who combine bike and public transport (I5U3) are younger (mostly between 20 and 45), working full-time, living in urban environments and are often part of family households. In addition to combining bike and public transport, this group has below average intermodal car usage and often uses the bike unimodally for non-work purposes, which means they are already using intermodal and unimodal provision. This information can help to identify areas in which many people of this group live. This enables the possibility of customizing the public transport infrastructure and its provision to the needs of this specific group. For example, mobility stations at public transport stations are currently being tested in many places to promote intermodal mobility. By having information about the predominant characteristics and type of residential area of mobility types with intermodal behavior, location and facilities of mobility stations can easier be determined. Furthermore, these groups may be explicitly involved in the planning process of mobility stations. As a result, their requirements for such new offers can be better taken into account in planning and implementation, which leads to greater acceptance of these offers in practice. At the same time, groups of people who are not yet intermodal could also be particularly considered and involved in the planning process to meet their requirements and needs. The same applies to multimodal users (I6U5), as they have similar characteristics to the intermodal bike and public transport user (I5U3) but live in more decentralized neighborhoods.

Subsequent qualitative analyses (interviews, workshops) show that the EMT match up very well, not only with the users’ mobility behavior but also the type of person. Specific user types can therefore be asked explicitly about their preferences with regard to the provision of public transport, infrastructure or even innovative vehicle concepts. The latter has been successfully demonstrated in interdisciplinary collaboration for the creation of user-oriented vehicle concepts [30].

In addition, integrating the purposes of trips in the methodological procedure, as with Vij et al. [49], constitutes a distinctive characteristic in the user typology presented. The mobility behavior of an individual can vary and be specific to certain situations, especially in an urban context where there are many different options. Allowance is made for this by differentiating the mobility behavior according to trip purposes. Looking at trip purposes may also help to transfer the EMT identified to specific real-life situations so that planning can address these needs. People who are intermodal in certain situations have a good chance of combining several modes of transport also for other trip purposes since they already have the necessary resources available. For example, individuals who regularly combine bike and public transport on their trip to work are likely to have both a public transport pass and a bicycle at their disposal. Thus, they already have the basic requirement to combine these means of transport also for other trip purposes. Campaigns that aim to motivate intermodal mode use for different trip purposes are likely to be particularly successful in these groups and should therefore be tailored to these.

The typology has another significant advantage in that it is empirically based on more than 1000 cases, enabling us to pursue this differentiated approach and also identify user groups that are less well represented. The empirical database needs to have a relatively large sample size in order to achieve a robust sample for the single user types. However, this is at the same time a limitation of the methodological procedure presented and its transferability, since the use of other datasets for this approach requires a certain number of cases and the differentiated query of mobility behavior. As a result, not every dataset can be processed with this approach to gain this kind of mobility types. Generating user types applicable to questions on intermodal travel behavior required an elaborate methodological procedure comprising two different cluster analyses with variables on intermodal and unimodal mode use and different trip purposes. At the same time, the resulting EMT presented in Table 2 have clearly defined and convincing characters that incorporate the complex findings in a comprehensive and straightforward manner.

This study provides a user typology that facilitates user-specific analyses of comprehensive mobility behavior with a special focus on integrating intermodal behavior. It combines intermodal and unimodal travel behavior with personal characteristics (socio-demographic characteristics and available mobility resources) and allows us to work on user specific research questions. In addition to the traditionally focused unimodal travel behavior, the proposed enhanced mobility types (EMT) provide an overview of the spectrum of intermodal user behavior which has not previously been part of any existing user typology. In this paper, we have outlined five different EMT as an example.

These EMT are the starting point for a number of further research questions. For example, information about selected user types will be further summarized and presented illustratively by creating short profiles and idealized example users. These are useful for presenting a clear yet simplified image and for translating the complex results into practice and using them in interdisciplinary collaborations. Further analyses of the conducted qualitative interviews with representatives of the selected user types will help to further specify the reasons and requirements for mode choice. This will enable us not only to describe but also understand intermodal mobility behavior within the context of the user’s overall mobility behavior. This will involve analyzing different user perspectives in respect of interchanges, interchange behavior and preferences for intermodal mode use, and will also involve transferring the findings into practice.

Another important issue, not dealt with as yet, is the transferability of the mobility types and the question as to whether the mobility types identified can also be found in other cities. We see the possibility of assigning individuals to a mobility type by adding up their individual socio-demographic characteristics. This also links up with discussions about the possibility of integrating the user types identified and their intermodal mobility behavior into travel demand modeling. In conclusion, it can be stated that there are wide-ranging options for employing and further developing the mobility types presented in research and for applying them in practice.