• No se han encontrado resultados

Objetivos y Plan de Trabajo

1.3. RESULTADOS Y DISCUSIÓN

The analysis explained in this subsection looks at modelling the choice of using a combination of travel information sources.5 The sources are categorised by media type: traditional and technology oriented. Descriptive statistics, modelling results and conclusions are presented.

5 The acquisition of travel information sources by media was investigated and results were presented

at the European Association for Research in Transportation (hEART) conference in 2013; the study was extended to investigate interactions of explanatory variables for the use of different media for travel information. The work in this subsection was presented at the Transport Research Arena (TRA) conference in May 2014: “Understanding the revolution in travel information: A model of information source acquisition and use”. It is included in the 2014 TRA proceedings and the paper corresponding to the presentation is available at:

Chapter 3. Exploratory study on travel information acquisition and use

3.2.2.1 Acquisition of TI by media channel: traditional and technology oriented

alternatives

The data used for the subsequent analyses consists of responses from 3,239 individuals from the 2007/2008 SHS dataset. The SHS questionnaire asked respondents how they sought travel information before setting out on a journey during the last month. The respondents could answer any combination of 14 responses depicting ways they acquired (or not) information before travelling, which would bring a power set of 16,384 possible outcomes. Keeping all different combinations derived from those 14 information sources as alternatives implies complex computations and results that would have been difficult to interpret. The alternative set was simplified down to four alternatives, in Table 3.3, categorised by the media channel(s) used to access information. The last alternative joining information sources from both media channels is considered uncorrelated with individuals choosing one or the other media. We assume that, by selecting a mix of options, those individuals manage the way they check information differently. Another analysis including correlation amongst alternatives could be investigated in future work. Table 3.3 shows the choice set for each respondent and describes the type of media sources included in each alternative.

Table 3.3 Choice of travel information media: descriptive statistics by alternative Alternative Information sources per alternative Number of times

chosen

Percentage of times chosen 1 no information

consulted (reference) Never go to unfamiliar places Know the route 1231 38%

2 traditional way to consult information

Road map/schedule Asked a friend

Someone else plans the route for me Phoned the AA or RAC

Phoned Traveline

Contacted the venue/attraction Never planned, relied on road signs

797 25%

3 technological way to consult information

Used journey planner on internet Used Transport Direct internet portal Checked Teletext for roadworks/congestion Used Satellite Navigation such as Tom Tom Used Traveline web-site

396 12%

4 both traditional and

technological ways All sources from alternative 2 and 3 815 25% In this study, individual demographics characterise the choice of travel information per media category. In the SHS survey, familiar and unfamiliar trips were not differentiated

Chapter 3. Exploratory study on travel information acquisition and use

because individuals did not answer for a particular trip but for their way-of-life in general. The question therefore encompassed all types of trips and the proportion of unfamiliar trips was negligible. Ideally, and for future investigation, a dataset including travel information associated with a specific trip would allow trips to be distinguished by purpose.

3.2.2.2 Acquisition of TI by media channel: modelling results

The logistic regression was estimated6 for the following categorical information source choices: traditional travel information (alternative 2), technology-related information (alternative 3) and a mix of traditional and technology (alternative 4) with respect to no information sought (alternative 1). The results are presented in Table 3.4. Although the overall fit of the model, as measured by rho-squared, is modest, indicating the presence of significant unobserved influences, a number of significant effects associated with the included explanatory variables are identified.

Table 3.4 Acquisition of travel information by media channel: modelling estimates Traditional information

source

Technology information source

Both traditional and technological

Variable Β t-test β t-test β t-test

Constants -0.681 -2.46 -2.340 -5.61 -2.270 -6.75 Age 0.003 0.08 -0.212 -4.17 -0.150 -3.68 Female 0.004 0.04 -0.184 -1.49 0.002 0.02 Income 0.044 1.34 0.141 4.47 0.125 4.25 Education: category 1 0.252 1.92 0.217 1.02 0.544 3.03 Education: category 2 0.477 3.53 0.653 3.12 1.200 6.82 Education: category 3 0.315 1.65 0.600 1.74 0.530 1.78 Internet use 0.063 0.57 1.730 7.40 1.500 9.02 Sedentary lifestyle -0.005 -0.12 -0.117 -1.72 -0.164 -3.06 Car use frequency -0.918 -2.10 0.573 0.88 1.260 2.40

PT use frequency 3.570 3.69 1.700 1.17 3.840 3.47

Congestion experience

frequency -0.619 -1.37 1.490 2.85 1.330 3.07

Bolded values are significant at the 95% confidence level. Reference alternative: No information consulted

Observations 3,239 Initial Log-Likelihood -4490.2 Final Log-Likelihood -3850.5 Log-Likelihood ratio 1279.3 Rho-square 0.142 Adjusted rho-square 0.134

6 Using BIOGEME v1.8, Bierlaire, M. (2003). BIOGEME: a free package for the estimation of discrete

Chapter 3. Exploratory study on travel information acquisition and use Since this model is a logistic regression, the constants or intercepts represent the mean response value of the outcome if all factors were set to zero. However, not all factors may be able to be set to zero, in which case they cannot be sensibly interpreted.

The female variable does not show any significant results for any of the alternatives. The internet dummy parameter unsurprisingly infers that using the internet strongly increases the likelihood of respondents seeking information from at least one technological source: alternative 3, technology-related information or alternative 4, a mix of traditional and technology. This is expected, as they would already be familiar with these devices. With respect to no education, respondents with some education level are slightly more likely to use a mix of information channels. Individuals with higher education are significantly more likely to use at least a traditional channel, or a technological way or, even more, a mix of ways to acquire travel information. For category 3, no significant results were found as this category probably includes a group with heterogeneous levels of education. A higher income is positively linked with a higher probability of choosing information sources using technology or a mix of sources (alternative 3 and 4), probably due to a better affordability and familiarity of these devices. The age parameter confirms that senior respondents may in general be more reticent to use new technology devices. Respondents who have lived in one place for many years are less likely to use a mix of information sources, probably due to the habit of following the same source for a long time. Car and public transport (PT) use frequencies are the number of times that respondents use, respectively, their car and public transport in a month. Frequent car users are less likely to use traditional channels than not look for travel information, and more likely to access information via a mix of traditional and technological means. As they get familiar and experienced on the road, frequent car or public transport users may need to use a combination of different sources to reduce information uncertainty, as they may encounter more congestion. Frequent PT users are more likely to use at least one traditional media such as asking relatives, or getting information from stations/stops. They may also look for additional information using technology, favouring this way over a mix of sources. Finally, individuals were asked how often they would experience congestion using their main mode of transport and it was found that those experiencing frequent congestion were more likely to use technology- related sources to seek travel information.

In order to compare the relative magnitudes of the effect of variables on the choice of travel information sources, elasticities for the probabilities of the alternatives with respect to attributes were calculated and graphically represented in Figure 3.1.

Chapter 3. Exploratory study on travel information acquisition and use

Figure 3.1 Elasticity values of alternative probabilities with respect to attributes

Few parameter values of the variables influencing respondents to acquire information using traditional sources (alternative 2) are statistically significant. Car use has the least weak effect, followed by the use of public transport and having a higher education but all of their magnitudes are low relative to elasticities across all alternatives. For alternative 3, the percentage change in the probability of using traditional ways to acquire information is greater when associated with percent change in variables such as internet, age and income as compare to others. For alternative 4, all parameter values are statistically significant and with a large magnitude. The percentage increase in probability of using a mix of ways to look for travel information is strongly related with a percentage increase in variables for internet use indicator, income and car use and a percentage decrease in age. In general, the internet use indicator, the age and the income of respondents are the variables that have higher effect on the choice of media category for travel information sources.

3.2.2.3 Acquisition of TI by media channel: conclusions

The preliminary results shown in these analyses are intuitively coherent. Internet use, age and income are shown to be the factors with the largest effect on the type of media that respondents use to acquire travel information. The internet use dummy coded variable is highly significant with a strong effect. An issue could be that if respondents have the internet available, they would use it by default and this may bring a problem of self- selection to the model. Another issue could be the aspect of causality between information

Chapter 3. Exploratory study on travel information acquisition and use and travel. In particular, the choice set of transport modes is strongly determined by availability and therefore dictates specific sources of information, but the final choice amongst different modes could be strongly dependent on information. This simple analysis still provides detailed and valuable user profiles per source. These conclusions are important for transport demand modellers who need to be aware of those factors that play a role in travel information and change behaviour that do not usually appear in those models. One of the next steps towards improving the model would be to add alternative attributes related to information sources and consider model structures that allow the representation of the simultaneous acquisition of a bundle of sources and their frequency of use.

3.2.3 Use of travel information, its usefulness and its effect on travel