schüssler & axhausen psychometric...onmentalism and

29
Research Collection Conference Paper Psychometric scales for risk propensity, environmentalism and and variety seeking Author(s): Schüssler, Nadine; Axhausen, Kay W. Publication Date: 2011 Permanent Link: https://doi.org/10.3929/ethz-a-006689653 Rights / License: In Copyright - Non-Commercial Use Permitted This page was generated automatically upon download from the ETH Zurich Research Collection . For more information please consult the Terms of use . ETH Library

Upload: vanmien

Post on 03-Jan-2017

222 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Research Collection

Conference Paper

Psychometric scales for risk propensity, environmentalism andand variety seeking

Author(s): Schüssler, Nadine; Axhausen, Kay W.

Publication Date: 2011

Permanent Link: https://doi.org/10.3929/ethz-a-006689653

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

Page 2: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 1

PSYCHOMETRIC SCALES FOR RISKPROPENSITY, ENVIRONMENTALISM ANDAND VARIETY SEEKING

9TH INTERNATIONAL CONFERENCE ON SURVEY METHODS IN TRANSPORT:TERMAS DE PUYEHUE, CHILE, NOVEMBER 14-18, 2011

Nadine Schüssler

(Corresponding Author), IVT, ETH Zurich, CH-8093 Zurich

Kay W. Axhausen, IVT, ETH Zurich, CH-8093 Zurich

ABSTRACT

There is a growing understanding in the transport modelling community that socio-demographic attributes are not sufficient to characterise travellers and to model their travelbehaviour. Thus, an increasing number of studies integrate latent variables representing atti-tudes and perceptions in their choice models. Since it is impossible to measure latent variablesdirectly psychometric scales are used as indicators.

In this paper psychometric scales for risk propensity, environmentalism and variety seeking arepresented that were developed with the aim to use them in latent variable models for mode andpublic transport connection choice. To achieve this goal, a two-step study was designed. Thefirst step of the survey is a paper-and-pen mail-out survey in which the psychometric scalesare complemented by a one-day travel diary and a household and person characteristics ques-tionnaire. In the second step, the scales are employed in a web-based survey and combinedwith a one-week GPS-based travel diary. This paper focusses on the first part of the survey anddiscusses the sample structure, the overall response patterns and the usability of the psychome-tric scales for characterising the respondents’ attitudes and integrating them into latent variablemodels.

Page 3: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

1 INTRODUCTION AND RELATED WORK

In recent years, there is a growing understanding in the transport modelling community thatsocio-demographic attributes are not sufficient to characterise travellers and to model theirtravel behaviour. Thus, an increasing number of studies integrate latent variables represent-ing attitudes and perceptions in their choice models. Since it is impossible to measure latentvariables directly, psychometric scales are used as indicators.

The traditional way of accounting for attitudes and perceptions in transport models is to mea-sure them employing psychometric scales, reduce them with factor analysis and use the factorscore in a regression or discrete choice model. However, recent methodological progress has ledto a more sophisticated way of integrating attitudes and perceptions into choice models. Sinceattitudes and perceptions cannot be observed directly, they are modelled as latent variables forwhich the psychometric indicators serve as manifestations. The modelling framework for thiswas presented by Ben-Akiva et al. (2002) and consists of two components. The measurementmodel describes the relationship between the indicators and the psychological factors whereasthe structural model explains the psychological factors with the help of person characteristicsand, thus, allows the analyst to distinguish between the influence of person characteristics onthe latent variable and their effects on other aspects of the decision.

The applications of psychometric indicators in transport behaviour models cover a variety ofdifferent topics. On the one hand, there are the short term decisions such as route choice, modechoice or the number of trips per day. On the other hand, there are long-term decisions likeresidential location choice or the choice of the number and type of vehicles. While some topicshave been addressed from very different angels, the attitudes and perceptions accounted for inother choice contexts are rather limited.

The prime example for a choice context approached from varying directions is mode choice.The attitudes and perceptions investigated so far range from the perception of mode relatedcharacteristics – such as reliability (Prashker, 1979) and convenience of usage (Atasoy et al.,2011) – over habits (Tudela et al., 2011; Vij et al., 2011) and social norms (Galdames et al.,2011) to lifestyle in general and environmentalism in particular (Kitamura et al., 1997; Schwa-nen and Mokhtarian, 2005; Johansson et al., 2006; Atasoy et al., 2010). Then again, the in-fluences of attitudes on car route choice has just recently been gained attention. First attemptshave examined how the willingness to take risks (Tsirimpa et al., 2010) or the tendency to be acar-lover (Abou-Zeid et al., 2011) interact with route choice decisions.

Page 4: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 3

The main determinant explored with respect to residential location choice (Gärling et al., 1998;Mokhtarian and Salomon, 2001; Cao and Mokhtarian, 2005; La Paix et al., 2011), vehicleownership (Choo and Mokhtarian, 2007) and the propensity to travel (Bagley and Mokhtarian,1999) is the – actual or aspired – lifestyle of a person. Since these three choice dimensionsare also inherently linked with each other, as shown by Kitamura et al. (1997) and La Paixet al. (2011), it is, however, an open research issue whether this linkage can be exhaustivelyexplained by the lifestyle or if there are still other factors accountable.

A completely different direction – that is, however, equally important for the travel surveymethods community – was explored in the recently published study by Hess and Stathopoulos(2011) who try to explore how the engagement of a survey participant in the survey influencestheir behaviour in a stated preference mode choice experiment by modelling the engagementas a latent variable.

This paper continues the line of research on measuring attitudes and perceptions that can beused as latent variables in choice models. The aim is to identify scales that are most effective inmeasuring the respondents’ attitudes and to establish how these attitudes can best be integratedinto latent variable choice models. Under investigation are risk propensity, environmentalismand variety seeking and their influence on mode and public transport connection choice. Severalof the studies mentioned above have already shown that a positive attitude towards protectingthe environment leads to a higher probability of choosing public transport. The assumptionregarding risk propensity is that risk averse persons are more prone to use the car and to choosepublic transport connections with fewer number of transfers. With respect to variety seeking,it is assumed that persons who search for variety in their life are more eager to experiment andhave therefore bigger choice sets.

In order to investigate these effects, a two-step study was designed. The first step is a paper-and-pen mail-out survey in which the psychometric scales are complemented by a one-day traveldiary and a household and person characteristics questionnaire. The second step combines thepsychometric scales with a GPS-based travel diary. The participants are recruited via telephone,asked to carry a person-based GPS receiver for a week and to fill out the psychometric scalesand the household and person questionnaire online. Moreover, they take part in a web-basedprompted recall survey in which they confirm and correct the results generated by the automaticpost-processing of their GPS tracks.

The design of the second part of the survey is described in the parallel paper by Schüssler et al.

(2011). This paper focusses on the first part, the paper-and-pen survey. The discussion includes

Page 5: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

an analysis of the response patterns and the sample structure. The answers to the psychome-tric scales are evaluated in terms of their usability for the characterisation of travellers, theirattitudes and the relationship between attitudes and travel behaviour.

The remainder of this paper is structured as follows. In the next section, the design of thepaper-and-pen study and particularly of the psychometric scales is discussed. Then, the re-sponse patterns are analysed with respect to the overall response rate and the socio-economiccharacteristics as well as the response to individual scale items. Subsequently, the answersto the psychometric scales are analysed with the aim to identify the scale items that are notsuitable for differentiating between the respondents’ attitudes and to obtain first indications onhow the attitudes can be integrated into latent variable choice models. The paper ends withconclusions and an outlook on further work.

2 SURVEY DESIGN

The paper-and-pen survey evaluated in this paper consists of three components: a questionnaireasking for person and household characteristics, three psychometric scales covering the topicsrisk propensity, variety seeking and environmentalism and a one-day travel diary. The person-and household questionnaire and the diary are slightly adapted versions of the ones used byWeis et al. (2010). The household and person questionnaire asks for a large variety of socio-economic variables whereas the trip diary included, beside the standards set of trip and activitycharacteristics, questions concerning the number of accompanying persons and the planninghorizon of the trip. The major part of the survey, however, were the three psychometric scales.They were designed following an extensive literature review that is briefly summarised below.

In total, 2000 questionnaires were sent out to respondents living in Canton Zurich in 2 wavesduring summer and fall 2010. After the first wave, a problem with the age distribution ofthe addresses bought from an address vender was discovered. About 45% of the respondentswere 65 years or older. Thus, the second wave was restricted to persons younger than 65years of age. For the diary, the respondents were assigned a survey day that was a Tuesday,Wednesday or Thursday to capture only ordinary weekdays. The potential respondents wereasked to send back the questionnaires within a week after the assigned survey day in a postpaidreturn envelope. In case they did not do this, they recieved a reminder via mail including a newassigned survey date.

Page 6: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 5

2.1 Design of the psychometric questionnaires

Corresponding to the three attitude domains that are investigated in this study, three separatescales have been developed: one measuring the risk propensity of the respondents, one address-ing their attitude towards the environment and environmental protection and one quantifyingthe level of variety the persons seek in their life. Each scale is presented to the respondentswith a 5-point agree-disagree scale. To minimise effects resulting from the order of the scaleitems, their order is determined randomly with three different random orders for each scale.

Risk propensity

There is a growing understanding in risk propensity research that a person’s degree of risktaking does not only depend on individual, group and cultural factors but also on the domainin which the risk occurs. While it is still an open research issue whether this is caused byvariations in the attitude towards risk over different domains or by varying perceptions of risks,Weber et al. (2002) argue that for the modelling and prediction of risk behaviour this distinctionis irrelevant and that it is sufficient to observe the person’s risk behaviour in the domain ofinterest. They derived an extensive 101 item risk propensity scale that covers five risk domains– namely financial, health and safety, recreational, ethical and social risks.

Nicholson et al. (2005) added the domain of career related risks to this set to derive their "RiskTaking Index" which they tested with the relationship between risk taking and performanceamong financial traders in investment banks. They found that risk propensity in any domainis strongly rooted in one’s personality and that personality traits associated with risk takingare extraversion and openness whereas neuroticism, agreeableness, conscientiousness usuallyare accompanied by risk averseness. Lepp and Gibson (2003) and Hatfield and Fernandes(2009) used the risk propensity of their respondents to explain certain aspects of their transportbehaviour. Lepp and Gibson (2003) analysed the influence of novelty seeking and risk percep-tion on the choice of international tourism destinations. They learned that familiarity seekingtravellers perceive higher degrees of risk compared to novelty seeking travellers and that thiscombination is – besides past experience with travelling abroad – the major influence on thetype of journey and the destination choices. Hatfield and Fernandes (2009) studied the drivingbehaviour of young drivers and how it is influenced by their risk propensity. They found thatyoung drivers show higher levels of risky driving despite their higher perception of risks butthat there are several motives for risky driving, such as sensation seeking or peer pressure, thatare strongly present for young drivers, especially for males.

The risk propensity scale used in this study is shown in Table 1. It combines a reduced version

Page 7: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Table 1: Scale items measuring the attitude towards risk

Code QuestionR1 I admit if my taste differs from that of my friends *R2 I argue with a friend if we have different opinions *R3 I ask my boss for a raise when I think that I earned it *R4 I would date a coworker *R5 I would openly disagree with my boss in front of my coworkers *R6 I speak my mind about unpopular issues at social occasions *R7 I wear unconventional clothes *R8 I would cheat a fair amount on my income tax *R9 I still drive home after I had three drinks in the last two hours *R10 I would forge somebody’s signature *R11 I have used cable TV without paying for it *R12 I use office materials provided by my employer for private purposes *R13 I would shoplift a small item (e.g. a lipstick or a pen) *R14 I have at least once used illegally copied software *R15 I go camping in the wild *R16 I ski down slopes that are too difficult for me *R17 I would like to do a safari in Kenya *R18 I would go whitewater rafting at high water in spring *R19 I would go on a 2 week vacation in a foreign country without booking ahead *R20 I engage in dangerous sports, e.g. paragliding *R21 I tried out bungee jumping at least once *R22 I eat food that is beyond its expiration date if is still looks good *R23 I ignore pain as long as possible before consulting a doctor *R24 I rarely use sunscreen before sunbathing *R25 I rarely wear a seat-belt *R26 I would engage in unprotected sex outside a relationship *R27 I usually ride my bike without wearing a helmet *R28 I smoke at least one packet of cigarettes per day *R29 I would co-sign a loan for a new car for a friend *R30 I would invest 10% of my annual income in a blue chip stock *R31 I would invest 10% of my annual income in speculative stocks *R32 I would invest 10% of my annual income in government bonds *R33 I would lend my best friend an amount of money equivalent to one month’s income *R34 I would bet a day’s income in a casino *R35 I would accept a job that is paid solely based on commission *R36 I always take the latest possible public transport connection to the train stationR37 I start earlier if I assume that there will be congestion on my routeR38 I prefer public transport connections with very short transfer timesR39 If I don’t know the way I just start into the general direction and search my way step by stepR40 I avoid streets that are occasionally congestedR41 I start earlier if I have to drive an unfamiliar routeR42 I try to be at the airport at the latest possible time

(*) Source: Weber et al. (2002)

Page 8: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 7

of the domain specific risk propensity scale by Weber et al. (2002) with seven additional itemsfor transport related risks. Overall, the scale contains 42 items covering the domains social,ethical, recreational, financial, health/safety and transport-related risks.

Environmentalism

Due to the increasing awareness of environmental issues, a lot of work regarding the mea-surement of environmentalism has been published in recent years. One of the earliest and mostwell-known studies is the land-use and transport behaviour study by Kitamura et al. (1997) whomeasured environmentalism using a 10 item scale. Subsequently, Schultz (2001) argued thatenvironmental concern has to be differentiated between concern for oneself, other people andthe biosphere because different values and awarenesses of harmful consequences are attachedto them. Gatersleben et al. (2002) investigated the relationship between environmental atti-tudes and believes, socio-economics, social science indicators of pro-environmental behaviourand measurements of direct and indirect energy consumption. They found that self-reportedpro-environmental behaviour is indeed related to environmental attitudes but only marginallyto household energy consumption which is more strongly linked to household size and income.They concluded that environmental attitudes are especially relevant when behavioural changesdo not cost too much effort or change in comfort. Still, as soon as people have the financialability they are tempted to consume more energy. Following the theory of planned behaviour(Ajzen, 1991), Anable (2005) developed a 105 item scale to examine the influence of habits,moral norms, environmental attitudes, felt efficacy and perception of other persons’ behaviouron mode choice and showed that the mode choice behaviour of different attitudinal populationsegments is indeed very different.

After reviewing, amongst others, the scales of these authors, the scales used by Gaterslebenet al. (2002) and Kitamura et al. (1997) were judged to be most appropriate for the study athand. To use the advantages of both scales, they were combined into the 25 item scale pre-sented in Table 2 that takes into account general concern for the environment, awareness ofconsequences for oneself, others and the biosphere and the evaluation of measures for environ-mental protection.

Variety seeking

Compared to the variety of studies employing measures for environmentalism relatively littleresearch has so far been directed towards the quantification of variety seeking and its incorpora-tion in models for daily transport behaviour. The few studies aiming in this direction investigatethe phenomenon of travel for its own sake, i.e. undirected travel or travel with unnecessary de-

Page 9: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Table 2: Scale items regarding environmentalism

Code QuestionE1 I worry about environmental problems *E2 Too much attention is paid to environmental problems *E3 Environmental problems are exaggerated *E4 The attention for the greenhouse effect is exaggerated *E5 I am optimistic regarding the state and future of our environment *E6 Environmental pollution affects my health *E7 Environmental problems have consequences for my life *E8 I can see with my own eyes that the environment is deteriorating *E9 Environmental problems are a risk for the future of our children *E10 Saving threatened species is unnecessary luxury *E11 We should be careful with our environment because we depend on it *E12 Vehicle emissions increase the expenses for health care **E13 Environmental protection starts with myself *E14 People who do not care about environmental protection avoid their responsibilities *E15 Behavioural change requires more environmental friendly products *E16 Behavioural change requires a right example by the government *E17 Pro-env. beh. is only useful if everybody cooperates and I don’t think this will happen *E18 Environmental protection costs too much **E19 Environmental protection is good for the economy **E20 Jobs are more important than the environment **E21 Stricter vehicle smog control should be enforced **E22 The price of gas should be raised to reduce pollution **E23 Using tax dollars to pay for public transport is a good investment **E24 There should be incentives for using electric vehicles **E25 Who causes environmental damage should pay to repair it **

(*) Source: Gatersleben et al. (2002)(**) Source: Kitamura et al. (1997)

tours (e.g. Mokhtarian and Salomon, 2001). A wider recognition of the influence of varietyseeking on travel behaviour can be found in the tourism literature. Examples are the studies byBello and Etzel (1985) and Niininen et al. (2004) who investigated the role of variety seekingand consumer loyalty on holiday destination choice. Bello and Etzel (1985) derived their own50 item scale and asked respondents to evaluate their most recent trip for more than 2 nightswith respect to – amongst others – novelty of the experience. They concluded that novelty seek-ers felt the trip to be more deserved, gave more importance to the pleasure aspect of the holiday

Page 10: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 9

Table 3: Scale items evaluating the variety seeking tendency

Code QuestionV1 I like to experience novelty and change in my daily life *V2 I sometimes look for ways to change my daily routine *V3 I like to have lots of activity around me *V4 I prefer a clearly structured, repetitive daily scheduleV5 Reoccurring rituals give me a feeling of control and securityV6 I love surprisesV7 A week in which all my evenings are similar bores meV8 Shops with exotic herbs and fragrances fascinate me *V9 When eating out I like to try unusual items *V10 The content of my shopping cart looks pretty much the same all the timeV11 I buy only trendy clothesV12 I prefer seasonal fruits and vegetablesV13 I actively search for bands whose music I do not yet knowV14 I always shop at the same supermarketV15 I like to explore unknown towns or parts of my townV16 I prefer to spend my holidays always at the same locationV17 I prefer having drinks always at my regular pubV18 I like to try new types of sportsV19 Cultures completely different from my own fascinate meV20 I prefer to organise my holidays spontaneouslyV21 I always keep an open door for surprise visitorsV22 I like to meet new peopleV23 I like to explore new places in my town or new towns **V24 I like to try new routes to familiar destinationsV25 I sometimes take a longer route to see something newV26 I like to drive around just for the fun of itV27 When commuting I always take the same routeV28 I like to meet new people while travelling by train

(*) Source: Mehrabian and Russell (1973)(**) Source: Weber et al. (2002)

and spent more time and money on it. Niininen et al. (2004), who used the Arousal SeekingTendency scale by Mehrabian and Russell (1973), found that there is a link between the needfor variety and the patterns of holiday destination choice but that this link is only modest.

Since none of the scales reported in the literature was completely satisfactory and the author of

Page 11: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Table 4: Response rates in the paper and pen survey

Sample 1 Sample 2 Total [%]Sent out 1000 1000 100.0Invalid addresses 26 28 2.7Reminder – 856 85.6Valid addresses 974 972 100.0Returned questionnaires 117 157 14.1Returned questionnaires 117 157 100.0Invalid hh & person quest. 4 5 3.3Invalid psychom. scales 4 13 6.2Invalid diary 12 22 12.4Valid questionnaires 100 122 81.0

the Arousal Seeking Tendency scale would make it only available for face-to-face interview-ing, we constructed our own scale including some of the questions reported by Mokhtarian andSalomon (2001). The variety seeking scale reported in Table 3 contains 28 questions measur-ing the desire for variety in the daily routine in general and in shopping, eating, recreationalactivities and transport behaviour in particular.

3 RESPONSE PATTERNS

Table 4 shows the overall response statistics for the paper-and-pen survey. In total, 2000 let-ters were sent out in two waves of which 2.7% were returned due to invalid addresses. Thereminder letter was sent out to 856 potential respondents of the second wave with the aim toincrease the response rate. Of the 1946 potential respondents with valid addresses 274 returnedat least one questionnaire resulting in a response rate of 14.1%. However, not all of the returnedquestionnaires were usable for later analysis. Some respondents answered only one or two ofthe three questionnaires or left out several questions. The trip diary, however, seems to havecaused the most difficulties leading to 34 questionnaires that were unusable even after carefulmanual correction by the analysts. With respect to the usability of the psychometric scales, aquestionnaire was sorted out if more than one third of the psychometric scale items were leftblank. Altogether, the questionnaires of 222 persons remained after cleaning for subsequentanalysis.

To evaluate the representativeness of the study, the household and person characteristics ofthe respondents were compared to the weighted persons and households in the Microcensus

Page 12: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 11

Table 5: Socio-economic attributes of the respondents compared to the Microcensus 2005

Attribute Study participants [%] MZ 2005 [%]

Gendermale 48.6 48.8female 51.4 51.2

Age

< 25 0.0 20.325 - 34 10.4 15.535 - 44 19.4 18.345 - 54 25.2 15.155 - 64 27.5 13.5>= 65 16.7 17.2

Education

None 1.4 2.6Obligatory school 3.2 12.9Matur 5.4 7.0Apprentice 38.7 49.1Prof. diploma 10.8 9.7Univ. of appl. sc. 24.3 7.0University 16.2 11.7

Employment status

In training 1.4 16.8Full time employed 46.4 37.6Part time employed 26.1 16.6Unemployed 2.3 2.9Housewife/-husband 5.4 6.2Retired 18.0 18.8

Household size

1 25.7 32.92 37.4 37.13 14.0 12.14 16.2 13.2>= 5 6.3 4.7

Monthly HH income

< 4’000 6.3 20.64’000 - 8’0000 38.7 46.88’000 - 12’000 23.9 21.512’000 - 16’000 17.6 7.0> 16’000 10.8 4.1

Driving licenseyes 90.5 79.7no 9.5 20.3

Car availabilityAlways 65.3 72.7Sometime 19.4 20.8Never 13.1 6.5

PT subscriptions

Nationwide season ticket 14.9 8.6Halbtax 59.5 26.3Other PT sub. 50.9 17.3None 13.1 38.0

Page 13: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Figure 1: Share of missing values for the risk propensity scale

2005 - the Swiss national travel survey - living in the study area. The comparison is presentedin Table 5. As expected, there is an under-representation of young respondents and smallhouseholds and an over-representation of persons with high education, high income, a partor full-time employment and households with four or more members. Moreover, there is aconsiderably higher share of driving license and public transport subscription holders and alower share of persons that have always a car available, implying that our sample is probablymore mobile and more public transport oriented than the average population of Canton Zurich.

As discussed above, not all respondents indicated their level of agreement to each of the scaleitems. While some respondents only filled out very few items and were consequently excludedfrom further analysis, others left only some questions blank. Thereby, certain scale items wereleft out more often than others. In order to better understand the developed scales, a closer lookis taken at these scale items in the following.

Figure 1 compares the number of missing answers per risk propensity scale item for the re-sponses usable for model estimation with those that were sorted out. As expected, the overallnumber of missing answers is higher for the persons that are sorted out and certain scale itemsclearly posed bigger problems for the respondents than others with one scale item clearly out-standing. This is scale item R3 "I ask my boss for a raise when I think that I earned it". Other

Page 14: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 13

Figure 2: Share of missing values for the environmentalism scale

scale items that were often left out were R5 "I would openly disagree with my boss in front ofmy coworkers", R15 "I go camping in the wild", R35 "I would accept a job that is paid solelybased on commission" and R12 "I use office materials provided by my employer for privatepurposes". Interestingly, nearly all of these scale items are related to work topics and about onethird of the participants who left these scale items out were at least part time employed. Thus,the high number of missing answers cannot only be attributed to persons not being in this kindof situation. Something else makes these items hard to answer. Perhaps it reflects the persons’conflict of truthfully stating something that is socially less acceptable. For instance, it is widelyaccepted or even demanded that one should ask for a raise if one earned it. However, a lot ofpeople shy away from it because it bears conflict potential.

The patterns of the missing answer distributions for the environmentalism scale presented inFigure 2 differ substantially from those for the risk propensity scale. The overall number ofmissing answers is considerably lower for the persons usable for model estimations and higherfor those that were sorted out. Moreover, there is no scale item that stands out the way thatscale item R3 did. Nevertheless, there are some scale items for which answers are missingmore often than for others. These are E7 "Environmental problems have consequences for mylife", E24 "There should be incentives for using electric vehicles" and E12 "Vehicle emissionsincrease the expenses for health care". Since there is no apparent reason why these questions

Page 15: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Figure 3: Share of missing values for the variety seeking scale

should be more difficult than others, it will be interesting to see if they will also stand out inthe web-based survey or if this is just a coincidence here.

For the variety seeking scale, there is again one scale item that clearly stands out in terms ofthe number of missing values: V27 "When commuting I always take the same route". Butthis time there is a simple explanation for it. All the persons who did not answer this scaleitem where retired or working as housewife or househusband, so they do not commute and thequestion does not apply to them. The other scale items with the most missing values were V17"I prefer having drinks always at my regular pub" and V7 "A week in which all my eveningsare similar bores me". Since, however, these scale items were only left out by four or threepersons, respectively, in the group of persons usable for model estimation, this might as wellbe just a coincidence.

The internal consistency of the scales was tested by calculating Cronbach’s alpha for eachscale and the answers given by the responses usable for model estimation. All scales performsatisfactory with a Cronbach’s alpha of 0.82 for the risk scale, 0.88 for the environmentalismscale and 0.83 for the variety seeking scale. This finding is also inline with the consistencyvalues between 0.7 and 0.89 reported by Weber et al. (2002) and Gatersleben et al. (2002) fortheir respective scales.

Page 16: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 15

4 ANALYSIS OF THE RESULTS

The analyses presented in this section pursue two goals. On the one hand, they should help toidentify the scale items that are not well suited for measuring the respondents attitudes. On theother hand, some first indications should be derived on how the attitudes can be integrated intolatent variable choice models. As discussed in the introduction, a latent variable model consistsof two components. The measurement model describes the relationship between the psychome-tric scale items and the underlying psychological factors whereas the structural model explainsthe psychological factors with the help of person characteristics. Therefore, the results of threeanalyses are presented in the following. First, an examination of the standard deviation andmean of the answers gives an indication which scale items are not usable to distinguish attitudeprofiles between respondents. Second, a factor analysis is conducted on each psychometricscale to identify the most dominant underlying psychological constructs before the relationshipbetween these factors and certain socio-economic variables is examined in the third subsection.

4.1 Standard deviation of the answers to scale items

Figure 4 depicts the standard deviation of the answers to each scale item relative to the meanvalue of the these answers. For this figures some of the scale items were recoded in a way thata low mean value corresponds to risk aversion, low attitude towards environmental protectionand low desire for variety. As expected, the mean values for risk propensity are systematicallylower than those for environmentalism and variety seeking and those for environmentalismare the highest. The respondents are in general risk averse and environmentally friendly – orat least they like to present themselves that way. The risk propensity scale shows by far thelargest range of standard deviations with scale items that have a very low standard deviationand others with a very high one. In contrast, the standard deviation for the variety seeking scaleitems does not vary much.

Since the aim of this analysis is to identify which scale items help the least to distinguishdifferent attitude profiles, the scale items with lowest standard deviation were identified in thenext step. For the risk propensity scale these are

• R26 I would engage in unprotected sex outside a relationship• R13 I would shoplift a small item (e.g. a lipstick or a pen)• R10 I would forge somebody’s signature• R25 I rarely wear a seat-belt

Page 17: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Figure 4: Mean value versus standard deviation of answers to scale items

These questions are mainly ethical questions with a strong social norm that defines them asbad. Accordingly, nearly all respondents stated total disagreement leading to a very low meanvalue.

Regarding the environmentalism scale the scale items with the lowest standard deviation are

• E11 We should care for our environment because we depend on it• E6 Environmental pollution affects my health• E13 Environmental protection starts with myself• E25 Who causes environmental damage should pay to repair it• E9 Environmental problems are a risk for the future of our children

Here, most respondents stated a strong agreement with these statements, reflecting their aware-ness of environmental consequences and that the ongoing education of the population withrespect to environmental protection had at least the effect of creating a strong social norm.

Page 18: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 17

Table 6: Results of the factor analysis for risk propensity

Question Factor 1 Factor 2 Factor 3R1 Admit if my taste differs from that of my friends – – 0.599R2 Argue with a friend we have different opinions – – 0.663R5 Openly disagree with my boss in front of my coworkers – – 0.453R6 Speak my mind about unpopular issues at social occasions – – 0.611R8 Cheat a fair amount on my income tax 0.684 – –R9 Drive home after I had three drinks in the last two hours 0.491 – –R10 Forge somebody’s signature 0.508 – –R13 Shoplift a small item (e.g. a lipstick or a pen) 0.617 – –R14 Use illegally copied software 0.516 – –R26 Engage in unprotected sex outside a relationship 0.471 – –R31 Invest 10% of my annual income in speculative stocks 0.436 – –R35 Accept a job that is paid solely based on commission – 0.489 –R41 Start earlier if I have to drive an unfamiliar route – -0.496 –R42 Try to be at the airport at the latest possible time – 0.566 –

For the variety seeking scale, the scale items with the lowest standard deviation were V12 "Iprefer seasonal fruits and vegetables" and V23 "I like to explore new places" but their standarddeviation does not differ much from the scale items with the highest standard deviation, i.e. V8"Shops with exotic herbs and fragrances fascinate me" and V20 "I prefer to organise my holi-days spontaneously" and their mean value is close to three indicating more balanced answersof the respondents.

4.2 Results of the factor analysis

The results of the factor analyses are presented in Tables 6, 7 and 8. In order to improvereadability, only the factor scores with an absolute value of at least 0.4 are shown. The fac-tor analysis was conducted with SPSS and the best results were achieved using a principalcomponent analysis and a Varimax rotation with three factors for each scale.

Since the risk propensity scale covers such a variety of domains, the three main factors iden-tified in the factor analysis explain only about 26% of the variance. They do, however, followthe domains specified beforehand. The first factor mainly covers ethical risks such as illegalbehaviour or cheating someone. The second factor entails mainly scale items that are con-cerned with transport related risks while the third factor summarises some of the social risks

Page 19: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Table 7: Results of the factor analysis for environmentalism

Question Factor 1 Factor 2 Factor 3E2 Too much attention is paid to environmental problems – 0.765 –E3 Environmental problems are exaggerated – 0.718 –E4 The attention for the greenhouse effect is exaggerated – 0.695 –E6 Environmental pollution affects my health 0.648 – –E7 Environmental problems have consequences for my life 0.517 – –E9 Env. prob. are a risk for the future of our children 0.440 – –E10 Saving threatened species is unnecessary luxury – – -0.403E11 We should care for our env. because we depend on it 0.499 – –E12 Vehicle emissions increase the need for health care 0.536 – –E13 A better environment starts with myself 0.483 – –E15 Behav. change requires more env. friendly products – – 0.687E16 Behav. change requires a right example by the government – – 0.428E18 Environmental protection costs too much – 0.464 –E19 Environmental protection is good for the economy 0.418 – –E20 Jobs are more important than the environment -0.425 – –E21 Stricter veh. smog control should be enforced 0.543 – –E22 The price of gas should be raised to reduce pollution 0.506 – –E24 There should be incentives for using electric vehicles – – 0.536E25 Who causes environmental damage should pay to repair it 0.513 – –

addressed by the scale items.

Weber et al. (2002) presented a factor analysis with 5 factors with the intention of having onefactor for each of their risk categories financial, health/saftey, recreational, ethical and socialrisks. However, they found only a distinct factor assignment for ethical, social and recreationalrisk to one factor each. All other risk areas were distributed over several factors. This findingwas confirmed by us in a factor analysis with 6 factors for our model reflecting the six riskareas in our scale. There, only the ethical and social risks were assigned to one unique factoreach while all other risk areas where distributed over several factors.

The three main factors for environmentalism shown in Table 7 explain about 43% of the vari-ance in the data. The first factor describes the respondents awareness of the negative conse-

quences of environmental pollution and our responsibility to restrict behaviour that is harmfulto the environment. The second factor for environmentalism characterises a denial of environ-

mental problems and the attitude that environmental problems are exaggerated and too much

Page 20: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 19

Table 8: Results of the factor analysis for variety seeking

Question Factor 1 Factor 2 Factor 3V1 I like to experience novelty and change in my daily life – – 0.443V2 I sometimes look for ways to change my daily routine – 0.647 –V3 I like to have lots of activity around me 0.406 – –V4 I prefer a clearly structured, repetitive daily schedule – -0.490 –V5 Rituals give me a feeling of control and security – -0.442 –V6 I love surprises – – 0.628V8 Shops with exotic herbs and fragrances fascinate me 0.660 – –V9 When eating out I like to try the most unusual items 0.411 – –V15 I like to explore unknown towns or parts of my town 0.665 – –V19 Cultures completely different from my own fascinate me 0.527 – –V21 I always keep an open door for surprise visitors – – 0.406V23 I like to explore new places 0.725 – –V24 I like to try new routes to familiar destinations – 0.624 –V25 I sometimes take a longer route to see something new – 0.481 –V26 I like to drive around just for the fun of it – 0.493 –V27 When commuting I always take the same route – -0.460 –V28 I like to meet new people while travelling by train – – 0.408

attention is paid to them. The third factor summarises the attitude towards measures for envi-

ronmental protection.

For variety seeking, only 33% of the variance is explained by the three main factors foundin the factor analysis, but they are more diverse in their interpretation than the factors forenvironmentalism or risk propensity as can be seen in Table 8. The first factor describes a desire

for making new experiences by trying new food, visiting new places and getting in contact withcultures different from those of the respondents. The second factor characterises a completelydifferent type of variety seeking. Persons who score high on this factor search for (small)variations from their daily routine. Therefore, this factor is called interest in varying one’s

daily routine. Finally, the third factor captures a liking for surprises and unexpected changes.

4.3 Socio-economic profiles of the factors

The subsequent Figures 5, 6 and 7 show the distribution of factor scores for different populationsegments and the first two factors of each scale. The population segments were chosen basedon the strongest correlation between the factor scores and the socio-economic characteristics.

Page 21: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Figure 5: Relationship between socio-economic variables and risk propensity

(a) Ethical risks and education (b) Ethical risks and employment status

(c) Transport related risks and gender (d) Transport related risks and income

If these socio-economic characteristics are indeed the most important influence factors has tobe tested in the latent variable model estimation.

The most distinct relationship between ethical risks and population segments was found for thelevel of education and the employment status. The more educated respondents tend to indicatea higher propensity to take ethical risks as do persons who are at least part time employedcompared to persons who are retired, unemployed or housewives/-husbands. The sample sizefor persons in training is too small to draw any reasonable conclusions for them.

The propensity to take transport related risks strongly depends on the gender and the income

Page 22: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 21

Figure 6: Relationship between socio-economic variables and environmentalism

(a) Environmental awareness and number of cars inhousehold (b) Environmental awareness and bike ownership

(c) Environmental denial and number of cars in house-hold (d) Environmental denial and education

of the respondents. Male respondents are more likely to take this kind of risks as are personswith higher household incomes. Both effects were expected based on the assumption thatpeople with a higher value of travel times savings are likely to be more risk prone in transportdecisions and males and persons with higher incomes tend to have higher values of travel timesavings.

For the awareness of environmental problems, the most distinct relationship between factorscore distribution and socio-economic variables can be seen for the number of cars in the house-hold and bike ownership. The higher the number of cars in the household the lower the factor

Page 23: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Figure 7: Relationship between socio-economic variables and desire for new experiences

(a) Desire for new experiences and number of trips re-ported (b) Desire for new experiences and gender

(c) Interest in varying one’s daily routine and numberof household members

(d) Interest in varying one’s daily routine and numberof cars in household

score tends to be, whereas the availability of at least one bike in the household leads by trend toa higher factor score. This is inline with the expectation that environmentally aware householdsuse more environmentally friendly modes and are correspondingly equipped with the matchingmobility tools.

Analogously, a higher number of cars in the household leads to an increased tendency to denyenvironmental issues. More interesting, however, is the influence of education on the denialfactor. High factor scores for this variable are only found by less educated respondents. Thismight indicate that less educated persons are not well reached by education programs address-

Page 24: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 23

ing environmental problems and protection.

The desire for new experiences seems to be related to the number of trips the respondentsmade on the survey day with persons scoring high on the new experiences factor also makingmore trips than others. This is reasonable since one would expect that persons who like newexperiences are also more active than others. Another possible segmentation is by genderbecause the females in our study have a higher desire to make new experiences than males.

The interest in varying one’s daily routine depends most heavily on the number of householdmembers. The bigger the household the smaller is the factor score maybe because a largehousehold itself provides enough variation in the daily schedule. Another socio-economic vari-able that influences the interest in varying one’s daily routing is the number of cars in thehouseholds. This is counterintuitive but to some extent the number of cars is correlated withthe household size.

5 CONCLUSION AND OUTLOOK

This paper investigates the suitability of three psychometric scales – one for risk propensity, onefor environmentalism and one for variety seeking – for measuring the attitudes of the respon-dents and incorporate these attitudes into latent variable models. Therefore, several analysesare conducted. First, the response patterns for this relatively new type of survey are analysed.Second, the scale items that were least suitable for differentiating respondents are searched.Third, first insights into how the attitudes can be integrated into choice models are gained.

The overall response rate was in the expected range taking into account the high burden on therespondents. However, 20% of questionnaires had to be removed, mainly because the diariescould not be interpreted. The socio-economic characteristics of the sample were in-line withprevious studies with an over-representation of persons that are highly educated, have a highincome, live in large households and are more mobile and public transport oriented than theaverage population in the sampling area. Regarding the response patterns of the individualscale items, some scale items appeared to be more difficult to answer than others, probablybecause they reflected the persons’ conflict of stating something that is socially less acceptable.

When analysing the standard deviation of the individual scale items, it can be seen that the threescales cover substantially different ranges of standard deviations. While the standard deviationsof the answers to the scale items of the variety seeking scale are very similar to each other, theenvironmentalism and especially the risk propensity scale show distinctively different answer

Page 25: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

patterns depending on the scale item. In particular, there were a few scale items for whichpretty much all the respondents provided similar answers. These were typically scale itemsconnected to strong social norms. Thus, it is questionable if these scale items are suitable todifferentiate respondents.

In the subsequently conducted factor analysis, several factors are established that can be testedin transport behaviour models. Ethical and transport related risks are the most interesting fac-tors regarding risk propensity while the awareness of negative consequences and the denial ofthe environmental problems should be used as indicators for environmentalism and for varietyseeking the desire for making new experiences and the interest in varying one’s daily routineshould be tested. For each of the these factors different socio-economic variables seem tobe relevant and should be checked in the structural component of the latent variable models.For risk propensity gender, education, income and employment status were most promisingwhereas for environmentalism these were education, number of cars and bike ownership andfor variety seeking gender, household size, number of trips on survey days and number of cars.

As a next step, these findings will be used to estimate latent variable models for modechoice based on the one-day diaries that were collected together with the psychometric scales.Thereby, the attitudinal factors established above will be tested individually and combined andin association with different socio-economic characteristics. Afterwards, the more extensivedataset derived from the web-based survey with the one-week GPS diary will be exploited.First, it will be checked whether there is a systematic difference in the answering patternscompared to those described in this paper taking into account the potentially different socio-economic composition of the samples. Then, a set of choice models will be estimated to in-vestigate the influence of the attitudes on mode and public transport connection choice and thenumber of trips made per day.

Page 26: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 25

6 ACKNOWLEDGEMENTS

The authors would like to thank the Swiss State Secretariat for Education and Research forfunding this research that is part of the research project "Route choice in urban public transportsystems" within the COST Action "TU0603 - Buses with high levels of service". Special thanksgoes to Eveline Beutler for her support with the field work and data entry.

Page 27: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

REFERENCES

Abou-Zeid, M., M. E. Ben-Akiva, M. Bierlaire, C. F. Choudhury and S. Hess (2011) Attitudesand value of time heterogeneity, paper presented at the 90th Annual Meeting of the Trans-

portation Research Board, Washington, D.C., January 2011.

Ajzen, I. (1991) The theory of planned behavior, Organizational Behavior and Human Decision

Processes, 50 (2) 179–211.

Anable, J. (2005) “Complacent Car Addicts” or “Aspiring Environmentalists”? Identifyingtravel behaviour segments using attitude theory, Transport Policy, 12 (1) 65–78.

Atasoy, B., A. Glerum and M. Bierlaire (2011) Mode choice with attitudinal latent class: ASwiss case-study, paper presented at the 2nd International Choice Modelling Conference,Leeds, July 2011.

Atasoy, B., A. Glerum, R. Hurtubia and M. Bierlaire (2010) Demand for public transport ser-vices: Integrating qualitative and quantitative methods, paper presented at the 10th Swiss

Transport Research Conference, Ascona, September 2010.

Bagley, M. N. and P. L. Mokhtarian (1999) The role of lifestyle and attitudinal characteris-tics in residential neighborhood choice, in A. Ceder (ed.) Transportation and Traffic Theory

- Proceedings of the 14th International Symposium on Transportation and Traffic Theory,Elsevier, Oxford.

Bello, D. C. and M. J. Etzel (1985) The role of novelty in the pleasure travel experience, Journal

of Travel Research, 24 (1) 20–26.

Ben-Akiva, M. E., J. L. Walker, A. T. Bernardino, D. Gopinath, T. Morikawa and A. Poly-doropoulou (2002) Integration of choice and latent variable models, in H. S. Mahmassani(ed.) In Perpetual Motion: Travel Behavior Research Opportunities and Application Chal-

lenges, 431–470, Elsevier, Oxford.

Cao, X. and G. Mokhtarian (2005) How do individuals adapt their personal travel? Objectiveand subjective influences on the consideration of travel-related strategies for San FranciscoBay Area commuters, Transport Policy, 12 (4) 291–302.

Choo, S. and P. L. Mokhtarian (2007) What type of vehicle do people drive? the role of attitudeand lifestyle in influencing vehicle type choice, Transportation Research Part A: Policy and

Practice, 38 (3) 201–222.

Page 28: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Psychometric scales for risk propensity, environmentalism and and variety seeking 27

Galdames, C., A. Tudela and J. A. Carrasco (2011) Exploring the role of psychological factorson mode choice models using a latent variables approach, paper presented at the 90th Annual

Meeting of the Transportation Research Board, Washington, D.C., January 2011.

Gatersleben, B., L. Steg and C. Vlek (2002) Measurement and determinants of environmentallysignificant consumer behavior, Environment and Behavior, 34 (3) 335–362.

Gärling, T., R. Gillholm and A. Gärling (1998) Reintroducing attitude theory in travel behaviorresearch: The validity of an interactive interview procedure to predict car use, Transporta-

tion, 25 (2) 129–146.

Hatfield, J. and R. Fernandes (2009) The role of risk-propensity in the risky driving of youngerdrivers, Accident Analysis & Prevention, 41 (1) 25–35.

Hess, S. and A. Stathopoulos (2011) Linking response quality to survey engagement: A com-bined random scale and latent variable approach, paper presented at the 2nd International

Choice Modelling Conference, Leeds, July 2011.

Johansson, M. V., T. Heldt and P. Johansson (2006) The effects of attitudes and personalitytraits on mode choice, Transportation Research Part A, 40 (6) 507–525.

Kitamura, R., P. L. Mokhtarian and L. Laidet (1997) A micro-analysis of land use and travel infive neighbourhoods in the San Francisco Bay Area, Transportation, 24 (2) 125–158.

La Paix, L., M. Bierlaire, E. Cherchi and A. Monzón (2011) How urban environment affectstravel behaviour? Integrated choice and latent variable model for travel schedules, paperpresented at the 2nd International Choice Modelling Conference, Leeds, July 2011.

Lepp, A. and H. Gibson (2003) Tourist roles, perceived risk and international tourism, Annals

of Tourism Research, 30 (3) 606–624.

Mehrabian, A. and J. A. Russell (1973) A measure of arousal seeking tendency, Environment

and Behavior, 5 (3) 315–333.

Mokhtarian, P. L. and I. Salomon (2001) How derived is the demand for travel? Some concep-tual and measurement considerations, Transportation Research Part A: Policy and Practice,35 (8) 695–719.

Nicholson, N., E. Soane, M. Fenton-O’Creevy and P. Willman (2005) Personality and domain-specific risk taking, Journal of Risk Research, 8 (2) 157–176.

Page 29: Schüssler & Axhausen PSYCHOMETRIC...ONMENTALISM AND

Niininen, O., E. Szivas and M. Riley (2004) Destination loyalty and repeat behaviour: An ap-plication of optimum stimulation measurement, International Journal of Tourism Research,6 (6) 439–447.

Prashker, J. N. (1979) Scaling perceptions of reliability of urban travel modes using indscal andfactor analysis methods, Transportation Research Part A: General, 13 (3) 203–212.

Schüssler, N., L. Montini, C. Dobler and K. W. Axhausen (2011) Improving automatic post-processing routines for GPS observations using prompted-recall data, paper presented at the9th International Conference on Survey Methods in Transport, Termas de Puyehue, Novem-ber 2011.

Schultz, P. W. (2001) The structure of environmental concern: Concern for self, other people,and the biosphere, Journal of Environmental Psychology, 21 (4) 327–339.

Schwanen, T. and P. L. Mokhtarian (2005) What affects commute mode choice: Neighborhoodphysical structure or preferences toward neighborhoods?, Journal of Transport Geography,13 (1) 83–99.

Tsirimpa, A., A. Polydoropoulou and C. Antoniou (2010) Development of a latent variablemodel to capture the impact of risk aversion on travelers’ switching behavior, Journal of

Choice Modelling, 3 (1) 127–148.

Tudela, A., K. M. N. Habib, J. A. Carrasco and A. O. Osman (2011) Incorporating the explicitrole of psychological factors on mode choice: A hybrid mode choice model by using datafrom an innovative psychometric survey, paper presented at the 2nd International Choice

Modelling Conference, Leeds, July 2011.

Vij, A., A. Carrel and J. L. Walker (2011) Capturing modality styles using behavioral mixturemodels and longitudinal data, paper presented at the 2nd International Choice Modelling

Conference, Leeds, July 2011.

Weber, E. U., A.-R. Blais and N. E. Betz (2002) A domain-specific risk-attitude scale: Mea-suring risk perceptions and risk behaviors, Journal of Behavioral Decision Making, 15 (4)263–290.

Weis, C., C. Dobler and K. W. Axhausen (2010) Stated adaptation survey of activity schedulingreactions to changing travel conditions: Field work and preliminary results, paper presentedat the 12th World Conference on Transportation Research, Lisbon, July 2010.