On the one hand, travelers’ preferences for travel information (the “I” in ICT) and teleworking and telecommunication facilities (the “C” in ICT) are expected to be interrelated, and, on the other hand, empirical evidence is lacking to prove this expectation. This chapter presents a preliminary empirical study that aims to provide inputs for the following model development in this thesis and also aims to provide related insights that are expected to be of practical relevance. The study is based on a web-survey conducted among 261 Dutch commuters about their daily life behavior in terms of using travel information and teleworking, and their potential behavior in terms of using travel information and teleworking under hypothetical conditions. This study develops a structural model and uses structural equation modeling (SEM) to analyze the collected data, and explores a) whether (and to what extent) a correlation exists between travelers’ preferences for teleworking and travel information as expected, and b) what are the relations between travelers’ preferences for teleworking and
travel information and other factors that could be of influence, such as travelers’ perception of the reliability of commuting time, the availability and reliability of travel information, the availability of teleworking and the quality of the teleworking environment and facilities. The final model provides good model fit and construct validity, and the following findings are derived from the SEM analysis of the 261 sample respondents:
y The perceived extent to which travelers are allowed to telework is positively related to travelers’ preference for teleworking;
y The quality of the teleworking environment and facilities is positively related to travelers’ preference for teleworking;
y The perceived availability of travel information to travelers is positively related to travelers’ preference for travel information;
y The hypothesized relation of “the perceived reliability of travel information” on travelers’ “preference for travel information” is found insignificant in this study, suggesting that, among the sample respondents, the reliability of travel information does not affect their preference for travel information. However, this inference should be treated with caution; y The perceived reliability of commuting time is found to be negatively related to both
travelers’ preference for teleworking and travelers’ preference for travel information – the less reliable travelers’ commuting time is, the more travelers prefer to telework or acquire travel information;
y It appears that the reliability of commuting time perceived by travelers has a larger impact on their preference for travel information than on their preference for teleworking, ceteris paribus;
y A significant correlation is found between travelers’ preference for teleworking and preference for travel information, suggesting that other common underlying factors or personal traits than what are considered in the study are relevant to travelers’ preferences for both teleworking and travel information. This also implies that travelers who prefer to use travel information seem to also prefer teleworking and vice versa (as both are driven by the common underlying factors or personality traits);
y Other factors than the ones that are considered in the model exist of effects on travelers’ preference for teleworking and preference for travel information;
y It appears that for the sample respondents in the study, the perceived availability of teleworking and the perceived quality of teleworking environment and facilities do not affect their preference for travel information, while the perceived availability and reliability of travel information do not affect their preference for teleworking.
These aforementioned results are expected to be of relevance to this research and are also expected to be of practical relevance. In particular, this study is of relevance to the following model development in this thesis. The study provides empirical insights, and also empirical evidence to support the rationale, for the model development to consider the interrelations between the “I” and the “C” in ICT. The details of the model will be elaborated in the following chapters of this thesis (Chapters 4 and 5). The derived insights will be also used for derivation of practical and policy implications in Chapter 6.
However, it should be noted that, as what has been addressed in detail in the previous sections in this chapter, aspects including the aim and scope of the study, the way in which the model is operationalized, the resulted unexpected relations subject to the given sample data, etc. should be taken into consideration in the interpretation of these findings. Further study is also advised for some of these issues (e.g., insignificant relations of certain factors on travelers’ preferences for teleworking and travel information) before making general conclusions.
Appendix
Results of other tested structural models
The full-fledged structural model with all hypothesized paths included
Table 3-7: Model fit of the full-fledged structural model with all hypothesized paths included
N Chi-square d.f. P-value CFI TLI RMSEA SRMR
261 145.355 115 0.0293 0.975 0.967 0.032* 0.045
*: Estimate: 0.032; 90 Percent C.I.: 0.011 ~0.0473UREDELOLW\506($.05: 0.980
Figure 3-3 presents the full-fledged estimated model, wherein all the paths defined in the conceptual model are estimated.
Perceived quality of teleworking environment and facilities (QUTE)
Preference for travel information (PEIN) Preference for teleworking (PETE) Perceived reliability of commuting time (RECO) Perceived availability of teleworking (AVTE) Perceived reliability of travel information (REIN) -0.007 (0.936) Perceived availability of travel information (AVIN) 0.216 (0.027) 0.853 AVTE1 AVTE2 0.671 AVTE3 0.805 AVTE4 0.570 QUTE2 QUTE1 QUTE3 PETE1 0.749 PETE2 0.736 RECO1 RECO2 RECO3 PEIN2 0.858 PEIN3 0.646 0.558 0.958 AVIN1 AVIN2 REIN1 0.866 0.789 PEIN1 0.526 (0.000) 0.304 (0.000) 0.311 (0.000) 0.217 (0.001) -0.074 (0.307) -0.017 (0.799) 0.027 (0.726) 0.048 (0.525) -0.084 (0.360) 0.032 (0.660) 0.050 (0.484) Legend: : Latent construct : Measurement variables : Effect (significant) : Effect (insignificant) : Correlation (significant) : Correlation (insignificant)
: Measurement variable for construct (all significant)
0.627
E
0.845
E
E : Error/unexplained variance of latent construct
0.556 0.949 0.567 0.597 0.796 0.562 0.021 (0.816) 0.020 (0.822) -0.040 (0.592 0.448 (0.000) -0.189 (0.008) 0.040 (0.643) -0.279 (0.000) 0.255 (0.002)
Figure 3-3: The estimated full-fledged structural model with all hypothesized paths included
The structural model with the path between “perceived reliability of travel information” and “preference for travel information” deleted
Table 3-8: Model fit of the structural model with the path between “perceived reliability of travel information” and “preference for travel information” deleted
N Chi-square d.f. P-value CFI TLI RMSEA SRMR
261 136.378 107 0.0291 0.975 0.968 0.032* 0.047
Perceived quality of teleworking environment and facilities (QUTE)
Preference for travel information (PEIN) Preference for teleworking (PETE) Perceived reliability of commuting time (RECO) 0.440 (0.000) Perceived availability of teleworking (AVTE) -0.170 (0.010) -0.261 (0.000) Perceived availability of travel information (AVIN) 0.268 (0.000) 0.215 (0.018) 0.854 AVTE1 AVTE2 0.673 AVTE3 0.802 AVTE4 0.570 0.563 0.606 QUTE2 QUTE1 QUTE3 0.787 PETE1 0.746 PETE2 0.735 RECO1 0.538 RECO2 0.984 RECO3 0.548 PEIN2 0.858 PEIN3 0.647 0.584 0.915 AVIN1 AVIN2 0.790 PEIN1 0.526 (0.000) 0.315 (0.000) -0.091 (0.216) -0.022 (0.754) 0.025 (0.755) 0.024 (0.723) 0.053 (0.444) Legend: : Latent construct : Measurement variables : Effect (significant) : Correlation (significant) : Correlation (insignificant) 0.639 E 0.847 E
E : Error/unexplained variance of latent construct : Measurement variable for construct (all significant)
Figure 3-4: The estimated structural model with the path between “perceived reliability of travel information” and “preference for travel information” deleted
127
4 The effects of different forms of ICT on
accessibility – A behavioral model and numerical
examples
Lu, R., Chorus, C., & Van Wee, G.P., (2014). The effects of different forms of ICT on accessibility - a behavioural model and numerical examples. Transportmetrica A: Transport
Science, 10(3), 233-254. 95
Abstract
This paper develops a utilitarian model that captures the effects of different forms of ICT (information and communication technologies) on accessibility. The role of constraints that an individual may face when making decisions is explicitly considered, as well as the presence of uncertainty. Two major forms of ICT are considered: travel information provided by ICT and teleactivities that may be performed using the ICT. The model is developed to measure the combined effects of both travel information and teleactivities as well as their potential interaction effects on accessibility. Numerical examples show the plausibility of the model in capturing the effects of different forms of ICT on accessibility and reflect the combined effects of travel information and teleactivities on accessibility in different situations.
95 Note that in this thesis, as is common practice, Chapter 4 is presented in a way, without any
4.1 Introduction
Accessibility is a key concept for both policy makers and researchers and it is studied by scholars of several disciplines, such as transport engineering, economics, geography, and sociology. Although defined and operationalised in various ways, accessibility is widely used to describe the correlation between land use patterns and transportation systems (Dong et al., 2006) and one definition of accessibility proposed by the U.S. Department of Environment (1996) is the ease and convenience of access to spatially distributed opportunity with a choice of travel. It is important at this point to note that the notion of accessibility over the past decades has been operationalised in numerous ways by transportation researchers and geographers (see Geurs and van Wee (2004) for an overview), and that these different operationalisations are not necessarily all mutually consistent. As will become clear below, in this paper we operationalize accessibility as being the utility of a choice set, which in this paper is operationalised as the utility of the most preferred alternative in that set. This approach, in our view, is in line with the LogSum-approach to accessibility as developed by Ben-Akiva and Lerman (1985) – but note that we forego the use of error terms in our model, hence we arrive at a deterministic equivalent of the well-known LogSum-formulation. However, we anticipate that there will be readers who find this operationalization of accessibility less appropriate or useful, and we of course respect this. Those readers may to rather interpret our measure of accessibility as being a measure of user benefits (which is a more general and neutral concept); these readers are invited to substitute the term “accessibility” for the term “user benefits” throughout the remainder of this paper.
A paramount goal of transport policy is to improve accessibility. Whereas in past decades, most attempts to increase accessibility have focused on the development of transportation infrastructure (roads, railways), it is to be expected that Information and Communication Technologies (ICTs96) may also have a large impact on accessibility. For example, travel information may help travellers reduce travel times, and the option to telework may make someone’s office very accessible also when roads are heavily congested. In light of this intuitive relation between ICT and accessibility, it is surprising that most ICT-related studies in the field of transportation and geography have focused on the effects of ICT on travel behaviour97, while there is few effort in measuring the effects of ICT on accessibility – the ease and convenience of access to spatially distributed opportunities with a choice of travel. Two main streams of literature concerning the effects of ICT on travel behaviour can be distinguished. One stream focuses on the effects of travel information on travel behaviour in the context of uncertain travel times (see e.g., Chorus et al., 2006b), and another stream deals with teleactivities, such as teleworking and teleshopping, that are enabled by ICT (see e.g., Andreev et al., 2010). Hardly any literature exists that combines both streams, whereas in real life most ICTs (such as laptops, mobile phones) combine both functions. Furthermore, we know of no study that explicitly deals with possible interactions between the effects of travel
96 In our research, ICT refers to all the technologies that provide access to information or teleactivities.
We consider ICT broadly, including “old” ICT such as radio and television and “new” ICT such as the Internet and mobile phones. We in particular distinguish the two major functions of ICTs: information technologies and communication technologies (section 4.3.1).
97
In our research, travel behaviour refers to personal activities and related travel, examples of which include commuting to work and travel to shopping.
information and teleactivities (e.g., a commuter might assess travel times for his or her route to work, before deciding whether or not to work from home for a few hours or the entire day). As mentioned above, there is hardly any literature on the effects of ICT on accessibility (Van Wee et al., 2013), and, to the best of our knowledge, no literature exists that discusses the combined effects of on the one hand travel information and on the other hand teleactivities on accessibility. Choices with respect to travel information and teleactivities could be mutually dependent, as do related accessibility benefits. For example, in the above described situation where a commuter decides whether or not to telework based on received travel information, the value of travel information depends on the availability of the option to telework, and vice versa. In other words, it is likely that in many situations, interactions between travel information and teleactivities (in terms of accessibility benefits) exist. Ignoring such interactions could lead to an under- or over-estimation of the effects of combined ICT on accessibility.
This paper aims to fill the identified gaps in the literature by presenting a model that is able to capture the combined effects of different forms of ICT on accessibility, including possible interactions. The model adopts the utility paradigm (see, Ben-Akiva and Lerman, 1979), where the maximum utility that an individual can derive from a set of activity-travel pattern options is considered to be a measure of accessibility. We apply the recently developed constrained multinomial logit model (see, Martínez et al., 2009) to capture the impact of constraints on travel behaviour and accessibility. Our model aims at allowing analysts to quantitatively measure and evaluate the impact of different forms of ICT accessibility when travellers are faced with uncertainty and constraints.
The remaining part of this paper is structured as follows. Section 4.2 gives an overview of literature on ICT’s (potential) effects on accessibility. Section 4.3 firstly introduces at a conceptual level the behavioural model of ICT’s effects on accessibility, followed by a formal derivation of the model. Section 4.4 presents numerical examples to show the model’s plausibility and investigate the combined effects of different forms of ICT on accessibility. Finally, Section 4.5 summarizes the main conclusions and discusses the model’s applicability and paths for further research.