ARTICLE IN PRESS

0272-4944/$ - se

doi:10.1016/j.je

�CorrespondE-mail addr

1The results

group MOBIL

Ministry of Ed

program ‘‘Socia

Journal of Environmental Psychology 27 (2007) 277–292

www.elsevier.com/locate/jep

Psychological, sociodemographic, and infrastructural factors asdeterminants of ecological impact caused by mobility behavior1

Marcel Huneckea,�, Sonja Hausteina, Sylvie Grischkatb, Susanne Bohlerc

aRuhr-Universitat Bochum, Faculty of Psychology, Workgroup Environmental and Cognitive Psychology, 44780 Bochum, GermanybLeuphana University of Luneburg, International Research Centre for Environmental and Sustainability Management, Germany

cWuppertal Institute for Climate, Environment and Energy, Germany

Available online 12 August 2007

Abstract

In this study, the relevance of psychological variables as predictors of the ecological impact of mobility behavior was investigated in

relation to infrastructural and sociodemographic variables. The database consisted of a survey of 1991 inhabitants of three large German

cities. In standardized interviews attitudinal factors based on the theory of planned behavior, further mobility-related attitude

dimensions, sociodemographic and infrastructural characteristics as well as mobility behavior were measured. Based on the behavior

measurement the ecological impact of mobility behavior was individually assessed for all participants of the study. In a regression

analysis with ecological impact as dependent variable, sociodemographic and psychological variables were the strongest predictors,

whereas infrastructural variables were of minor relevance. This result puts findings of other environmental studies into question which

indicate that psychological variables only influence intent-oriented behavior, whereas impact-oriented behavior is mainly determined by

sociodemographic and household variables. The design of effective intervention programs to reduce the ecological impact of mobility

behavior requires knowledge about the determinants of mobility-related ecological impact, which are primarily the use of private

motorized modes and the traveled distances. Separate regression analyses for these two variables provided detailed information about

starting points to reduce the ecological impact of mobility behavior.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Environmental behavior; Environmental impact; Attitudes; Mobility behavior; Transportation

1. Introduction

One of the biggest global ecological challenges consistsin the reduction of the ecological impact of individualmobility behavior. According to the Kyoto Protocolindustrialized countries have to reduce their total green-house gas emissions by an average of 5.4% below 1990levels in the first commitment period of 2008–2012(Lenzen, Dey, & Hamilton, 2003). In Germany, withinthe last decades emissions of most pollutants causedby transportation could be reduced, whereas emissionsof greenhouse gases, respectively, CO2, from transport

e front matter r 2007 Elsevier Ltd. All rights reserved.

nvp.2007.08.001

ing author. Tel.: +49 234 32 23030; fax: +49 234 32 14308.

ess: marcel.hunecke@rub.de (M. Hunecke).

are based on research conducted by the junior research

ANZ, which was supported by the German Federal

ucation und Research (BMBF) in the framework of the

l-Ecological Research’’.

increased by about 6.3% between 1990 and 2003 (SRU(German Advisory Council on the Environment), 2005).These tendencies can be found in all western countries(IEA, 2000).Several strategies have been proposed to implement

environmentally sustainable passenger transportation, e.g.an increase of the efficiency of transportation technologies(Lovins & Cramer, 2004), the densification of housing,employment, shopping, and cultural activities (Stead &Marshall, 2001), and regulatory and fiscal measures(ECMT, 2004). In addition, the attractiveness of sustain-able mobility has to be increased by soft policy measuressuch as public awareness campaigns for sustainablemobility and social marketing for public transportation(Brog, Erl, & Mense, 2004). For the design of soft policyinterventions it is necessary to know the motivations of theusers of different transport modes. Stern (2000) introducedthe differentiation between an intent perspective and an

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292278

impact perspective into environmental psychology. Theintent perspective analyzes the motivational basis ofconservation behavior; the impact perspective determinesthe ecological consequences of environmental behavior.

In the present study, we take into account bothperspectives. From an impact perspective we analyze therelation of psychological variables to greenhouse gasemissions resulting from mobility behavior. In order toavoid an overestimation of psychological variables, socio-demographic and infrastructural variables are included inthe analysis. From an intent perspective we analyze themotivational basis of mobility behavior.

1.1. Psychological variables and mobility behavior

In transport science, it is agreed that infrastructuralfactors have a great impact on mobility behavior becausethey determine behavioral options. For example, if nopublic transportation services exist, people have to use thecar, in spite of a high motivation to use a bus or train.Mobility behavior, however, is not solely determined byinfrastructural constraints. There are two types of personalfactors relevant for individual mobility, sociodemographiccharacteristics and attitudinal factors. Sociodemographicaspects include factors such as age or employment status,which determine individual options and necessities formobility activities (e.g. Hanson & Schwab, 1995). Attitu-dinal factors include values, norms, and attitudes, whichaffect preferences for specific activities, destinations,routes, and means of transport (e.g. Anable, 2005; Anable& Gatersleben, 2005; Bamberg & Schmidt, 2001, 2003;Heath & Gifford, 2002; Hunecke, Blobaum, Matthies, &Hoger, 2001; Steg, 2005; Steg, Vlek, & Slotegraaf, 2001).Consequently, the most important task for mobilityresearch is an integrated analysis of the infrastructuraland personal determinants of mobility behavior.

So far, only one interdisciplinary study has testedmultivariate regression models for travel mode choiceand distances traveled by including psychological, socio-demographic as well as infrastructural variables (Van Wee,Holwerda, & Van Baren, 2002). In this study thepsychological influences are operationalized as a preferencefor a certain transport mode. The results indicate anincrease of explanatory power for a model including thepreference variable compared to a model that onlycomprises sociodemographic and infrastructural variables.Furthermore, the analysis shows that the predictive powerof preferences is higher for travel mode choice than fortraveled distances. One crucial restriction of the Van Weestudy is the low reliability of the preference measurementby one item only; here people have to categorize themselvesas preferring a certain mode of transportation.

In social and behavioral research, more sophisticatedtheoretical approaches like the Theory of Planned Beha-vior (TPB; Ajzen, 1991) have been applied to explainmobility behavior by personal factors rather than by simplepreferences for different transport modes. The TPB regards

the constructs attitude, subjective norm (SN), perceivedbehavioral control (PBC), and intention as predictors ofbehavior. Intention is seen as a summary of all the pros andcons a person takes into account when deliberatelyreasoning whether a behavior should be performed ornot. Intention itself is viewed as causally determined byattitude, SN, and PBC. Attitude toward a behavior is thedegree to which the performance of the behavior ispositively or negatively valued. SN is defined as theperceived social pressure to engage or not to engage in abehavior. PBC refers to people’s perceptions of their abilityto perform a behavior. It is assumed to be a direct predictorof both, intention and behavior. The TPB also postulatesthat sociodemographic and contextual factors, values, andgeneral beliefs affect behavior only indirectly via the fourpredictors of the TPB.There are two reasons why the TPB offers an adequate

theoretical framework to explain goal-directed mobilitybehavior: On the one hand, applications of the TPB in thedomain of mobility behavior provide strong empiricalsupport for this model (e.g. Bamberg, Hunecke, &Blobaum, in press; Bamberg & Schmidt, 2001, 2003; Heath& Gifford, 2002). On the other hand, comprising fourpredictors only, the TPB is a comprehensive and economic-al model to explain mobility behavior with the limitedresources of survey studies.In mobility research further mobility-related attitudinal

factors could be identified that affect mobility behavior andare not measured explicitly by the constructs of the TPB.Several studies have demonstrated a positive effect of

personal norm (PN) on the use of environmentally friendlytravel modes (e.g. Harland, Staats, & Wilke, 1999;Hunecke et al., 2001; Nordlund & Garvill, 2003). TheTPB only measures the SN, which is defined as theperceived social pressure to engage or not to engage in abehavior and is determined by normative expectations ofimportant referents. In contrast to SN, PN measures theintrinsic moral obligation to behave morally right(Schwartz, 1977). The relevance of moral norms in travelmode choice is relatively well analyzed. A direct effect ofPN on travel mode choice could not be shown whencontrolling for TPB constructs systematically (Bamberg &Schmidt, 2003; Heath & Gifford, 2002). Instead, therelation between PN and behavior is an indirect one,mediated by intention (Bamberg et al., in press).In addition, the psychological construct perceived

mobility necessities (PMN) extends the TPB providing amore differentiated understanding of the use of environ-mentally friendly transport modes. Haustein and Hunecke(2007) could demonstrate that PMN, defined as people’sperceptions of mobility-related consequences of theirpersonal living circumstances, have an independent effecton travel mode choice in the context of TPB. The factorPMN differentiates the measurement of control beliefs,which were previously only measured implicitly by PBC.Regarding travel mode choice, PBC is defined as people’sperceptions of their ability to use a certain mode of

ARTICLE IN PRESS

2Other methods for analyzing material and energy flows are Life Cycle

Analysis, Product Line Analysis, Process Chain Analysis, Ecological

Footprint, Material Intensity Analysis (Colborn, Meyers, & Dumanoski,

1996; Hofmeister, 1998).

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 279

transportation. In this sense PBC is determined by thetraffic infrastructure as well as by personal living circum-stances. The introduction of PMN separates these twokinds of control beliefs. PBC measures the perception ofthe physical accessibility of transport modes, whereasPMN operationalize the subjective perceptions of mobi-lity-related necessities resulting from social constraints likehaving a job or children, for instance.

Another mobility-related attitude dimension results fromsymbolic-affective evaluations of transport modes. Steget al. (2001) have demonstrated that symbolic-affectivefunctions, like excitement and prestige, as well as instru-mental-reasoned functions, like financial costs and drivingconditions, are important dimensions underlying theattractiveness of car use. In a follow-up study, Steg(2005) has shown that commuter car use is most stronglyrelated to symbolic and affective motives, and not toinstrumental ones. Examining the relative importance ofdifferent instrumental and affective journey attributes,Anable and Gatersleben (2005) found that flexibility andconvenience are the most important instrumental attributesfor car users, whereas freedom is the most importantaffective one. With regard to these aspects the private car isevaluated by far more positively than public transportationor the bike.

Hunecke (2000) has differentiated four basic symbolicdimensions of mobility: autonomy, excitement, status, andprivacy. All symbolic-affective evaluations of transportmodes can finally be reduced to these four dimensions,which are characterized by a functional core of physical orsocioeconomic aspects, but depend strongly on processes ofsocial interpretation. For example, the evaluation ofautonomy of various transport modes is influenced bythe infrastructure and the accessibility of transportsystems. But the extent to which autonomy is evaluatedas necessary or sufficient varies strongly between differentpeople. Theoretically, each transport mode could beevaluated separately with respect to the four basicdimensions. But factor analytical results of two Germanlarge-scale surveys with representative samples have shownthat only public transport is evaluated independently onthe four symbolic dimensions (Hunecke, Schubert, & Zinn,2005; Hunecke & Schweer, 2006). In contrast the car isevaluated homogenously positive or negative in relation toall four dimensions. Against the theoretical background ofan extended norm activation model, which also considersthe constructs of SN and PBC, the autonomy dimensioncould explain additional variance of travel mode choice(Hunecke, 2000).

In the perspective of the TPB the symbolic-affectiveevaluations are behavioral beliefs about how good and bador rather pleasant and unpleasant the transport modes arefor real and potential users. For this reason the symbolic-affective evaluations of excitement, status, and privacy canbe used—in accordance with the TPB—as one possibilityto operationalize the attitude variable in the domain oftravel mode choice (e.g. Haustein & Hunecke, 2007).

In relation to the other three symbolic-affective evalua-tions, the autonomy dimension shows a higher conceptualsimilarity to PBC than to attitude. In spite of thisconceptual similarity, there is still a difference betweenthese constructs. The autonomy dimension measures thegeneral belief that all activities could be carried out by therespective mode of transport, whereas PBC is operationa-lized as the evaluation of the difficulty to use a certainmode of transportation.

1.2. Personal factors as determinants of ecological impact

So far only few studies have been published in whichinterrelations between personal variables and the ecologicalimpact of environmental behavior are analyzed. Gate-rsleben, Steg, and Vlek (2002) have investigated theinfluence of attitudinal variables and sociodemographiccharacteristics on pro-environmental behavior, which wasoperationalized by an index of 13 kinds of behavior, andon the ecological impact of household energy use. Theyfound out that pro-environmental behavior is morestrongly related to attitudinal variables, such as environ-mental awareness and beliefs, whereas the ecologicalimpact of household energy use is primarily related tosocio-economic variables, such as income and householdsize. Their results confirm the assumption of Stern (2000)that it is worthwhile to distinguish two different measuresof environmentally significant behavior: an intent-orientedmeasure and an impact-oriented measure.An additional study by Poortinga, Steg, and Vlek (2004)

has analyzed the role of value dimensions concerningdifferent aspects of quality of life, and general and specificenvironmental concern in the field of home and transportenergy use. So far this study has been the only one focusingon the influence of psychological variables on theecological impact of mobility behavior. Their resultsconfirm the findings of Gatersleben et al. (2002): Homeand transport energy use are especially related to socio-demographic variables, whereas values and environmentalconcern are less important. Only the value dimension‘‘openness to change’’ was found as a relevant predictor ofthe ecological impact of transport energy use.

1.3. Ecological Impact Assessment (EcIA)

The EcIA has its origin in the Material Flow Analysis inbusiness companies. EcIA was developed to identifyecological impact associated with the production and useof goods and services (Baccini & Brunner, 1991).2 Theobjective of EcIA is to make material flows transparent, toimprove environmental-related decisions, and to serve asan instrument for political consulting.

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292280

EcIA has been applied for several consumption areas,also for mobility as an area of need with ecologicalrelevance (INFRAS, 1995). In their application of EcIA tomobility, these studies differ with regard to their choice ofthe examined environmental impact, transportation modes,and life cycle sections of vehicles (production, modes ofusage, infrastructure, disposal). Most surveys investigatetraffic systems as a whole, but do not take into account theecological impact on an individual level, where the usedtransportation mode, the distances traveled, and thecapacity rate of the used transportation mode are thecrucial parameters (Ifeu-Institute (Ifeu-Institute for Energyand Environmental Research, Heidelberg), 2005).

1.4. The present study

The overall objective of the present study was to test anexplanation model for ecological impact of mobilitybehavior that allows specifying the predictive power ofpsychological, sociodemographic, and infrastructural vari-ables. This explanation model should be as exhaustive aspossible with respect to psychological, sociodemographic,and infrastructural factors, so that policy makers can definepriorities when determining the most important startingpoints to promote environmental sustainable transport.

There is only one study that examines the relationshipbetween attitudinal variables and the ecological impactof individual mobility behavior (Poortinga et al., 2004).A central result of this study is that the impact ofenvironmental behavior is more strongly related to socio-demographic and household variables than to values andenvironmental beliefs. However, this result has to beinterpreted with the restriction that behavior specificattitudes have not been included in their analyses. Resultsfrom environmental behavior research indicate that beha-vior specific attitudes and beliefs are better predictors ofbehavior than values or general environmental concern(e.g. Dietz, Stern, & Guagnano, 1998; Oreg & Katz-Gerro,2006). For this reason the current research on psycholo-gical determinants of mobility behavior is based ontheoretically and empirically well-founded models of anattitude–behavior relationship, such as the TPB.

In the present study we aimed at maximizing theexplanatory power of the psychological variables andincluded only mobility-related variables that empiricallyproved to be relevant predictors for travel mode choice.Most of these variables were theoretically derived from theTPB like SN, PBC, and attitude, while the latter wasoperationalized via symbolic-affective evaluations on thefour mobility dimensions autonomy, status, excitement,and privacy. Moreover, the psychological part of the modelwas extended by the constructs PN and PMN as well as byattitude dimensions related to the symbolic dimensions ofmobility, independent of the constructs of the TPB. TheTPB construct ‘‘intention’’ was not integrated into themodel because it does not offer enough information onhow mobility behavior could be changed despite of its high

correlation with behavior. A short value inventory wasincluded in all analyses to test the relevance of values whenmobility-related attitudes were also considered.Another restriction of previous research is the assess-

ment of ecological impact, which has been carried out onlyin an approximate way in psychological studies. In ourinterdisciplinary study we shall overcome this restrictionand realize an ecological assessment of individual mobilitybehavior that satisfies the methodological standards ofcurrent environmental science.

1.5. Hypotheses

Assigning political priorities in planning processes forclimate protection requires information on the relativeimportance of psychological, sociodemographic, and infra-structural factors for the ecological impact of mobilitybehavior. For this reason in a first step we analyzed therelationship between psychological, sociodemographic andinfrastructural variables and greenhouse gases resultingfrom mobility behavior (cf. impact hypothesis).

Impact hypothesis: Psychological variables are significantpredictors of mobility-related greenhouse gas emissionseven if infrastructural and sociodemographic factors arecontrolled for.The design of effective intervention programs to reduce

the ecological impact of mobility behavior demands variousstrategies focusing on different aspects of mobility beha-vior. The two most relevant aspects that have to be changedare the rate of using private motorized modes and thetraveled distances in general. We expect that these twoaspects of mobility behavior are related to differentpsychological, sociodemographic, and infrastructural fac-tors, which require specific interventions in each case. Forthis reason in a second step we analyzed separately therelationship between psychological, sociodemographic, andinfrastructural variables and travel mode choice as well astraveled distances. Regarding psychological variables, weexpect a stronger relation of mobility-related attitudes totravel mode choice than to distances traveled as theseattitudes were operationalized referring directly to thedecision process of travel mode choice. In contrast, forvalues a stronger relation with distances traveled is expectedbecause values as general orientations influence how oftenpeople are mobile, what kind of destinations they choose,and which distances they cover to reach their destinations.

Travel mode hypothesis: Psychological variables aresignificant predictors of the use of private motorizedmodes, even if infrastructural and sociodemographicfactors are controlled for.Additionally, it is expected that mobility-related atti-

tudes are better predictors for travel mode choice thanvalues.

Traveled distances hypothesis: Psychological variablesare significant predictors of traveled distances, even ifinfrastructural and sociodemographic factors are con-trolled for.

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 281

We further hypothesized that the relevance of psycho-logical variables in this analysis is weaker than whenpredicting the use of private motorized modes. Instead,sociodemographic and infrastructural factors are expectedto be the most important predictors. With respect topsychological variables, values are expected to be betterpredictors than mobility-related attitudes.

2. Method

2.1. Sample and procedure

Data for this study was collected from June to December2003, in the German cities of Augsburg, Bielefeld, andMagdeburg. The selection of these three cities was based onseveral indicators. The main indicator was the size ofthe cities and their function in the regional context, whichwas measured by the German BIK classification.3 Accord-ing to the BIK classification the three cities are indicated ascore areas, which is comparable to the classificationStandard Metropolitan Area (SMA) in the USA. TheBIK category core area reflects urban areas that comprise443 German communities and cover 43.3% of the Germanpopulation. A further indicator for the selection of thesurvey areas was the transportation infrastructure, whichcomprised a local tram system as well as a car-sharingservice. Additionally, these cities were not characterized byany topographic specifics that could drastically influencethe choice of transportation modes in any way and theyprovided for a regionally balanced selection. Once Augs-burg, Bielefeld, and Magdeburg were chosen, three types ofcity districts were selected within each city: an inner-citydistrict, a city border district, and a suburban district(cf. Table 1).

For each of the city districts the survey population wasrandomly produced by the cities’ registration offices.11,028 German citizens aged 18–80 received a letterannouncing the survey. The people were personallycontacted by trained interviewers and were asked if theywished to participate. The survey was conducted viastandardized face-to-face interviews that lasted about60min. Altogether, 1991 interviews were carried out, withapproximately 660 interviews per city and about 220interviews per city district. The response rate was 25%after correcting for address errors and for people notcontacted because the number of intended interviews hadalready been achieved. In consideration of the timeinvestment required for a 1-h-interview without incentives,we regard the response rate as satisfactory.

The sample consisted of 1056 women (53%) and 935 men(47%) with a mean age of 46.7. The sample was representa-

3The BIK region types are the most accepted analytical model to

structure space in Germany (Hoffmeyer-Zlotnik, 2000). The model’s main

indicator is the density of population and workplaces (inhabitants+em-

ployees/square kilometer). This indicator reflects the intensity of interac-

tion within regions, which is essential for transportation issues

(Aschpurwis+Behrens GmbH, 2001).

tive for the core areas regarding age and gender, whereaseducation level was above average (43.5% with highereducation). This can be traced back to a higher willingnessto participate of well-educated people and to a high share ofstudents living in the selected inner-city districts.

2.2. Measures

2.2.1. Psychological variables

The TPB constructs PBC, SN, and intention weremeasured with two items each. The statements refer to theuse of environment-friendly means of transport instead ofprivate car use. Attitude was operationalized with statementson the symbolic dimensions autonomy, excitement, status,and privacy, concerning the travel modes private car andpublic transport. For the bicycle only the two dimensionsautonomy and excitement were measured, because previousresearch indicates that the status and privacy dimensions arenot relevant for the use of bicycles in everyday life. Insteadthe weather is an important contextual factor of bicycle use.We took the weather into account by measuring thewillingness to use the bicycle in bad weather conditions.This attitude dimension is called ‘‘weather resistance’’.The PN to use environment-friendly means of transport

was measured with two items, just as perceived mobilitynecessities were. All responses were provided on a 5-point-agreement-scale (1 ¼ do not agree at all, 2 ¼ agree slightly,3 ¼ agree moderately, 4 ¼ agree very much, 5 ¼ agreetotally).4 The items measuring the psychological constructswere presented in random order.Values were assessed by a short version of the Schwartz

Value Inventory (Schwartz & Bilsky, 1990) developed byBamberg (2001). Schwartz distinguishes between 10 moti-vational types of values ordered in a two-dimensionalstructure constituted by four higher order value types,openness to change vs. conservation and self-enhancementvs. self-transcendence (Schwartz, 1992). The pole opennessto change includes stimulation and self-direction, whereasthe opposite pole stresses the preference of tradition,security, and conformity. The pole self-enhancementsubsumes power and achievement, whereas self-transcen-dence includes universalism and benevolence. Each of thefour higher order value types was measured by three itemson a 9-point scale (�1 ¼ opposed to my values, 0 ¼ notimportant, 1–2 [unlabeled], 3 ¼ important, 4–5 [unlabeled],6 ¼ very important, 7 ¼ of supreme importance). Allpsychological items are listed in the Appendix.

2.2.2. Infrastructural variables: spatial characteristics and

accessibility to transport systems

Spatial aspects were operationalized independently fromsubjective judgments through the district types and the citiesthe people belonged to. Each type of district is characterizedby a certain spatial and infrastructural situation, which is

4The equidistance of the German version of this answer scale was

validated by Rohrmann (1978).

ARTICLE IN PRESS

Table 1

General criteria for three types of city districts

Criteria Characteristics

District 1 District 2 District 3

Inner-city district City border district Suburban district

Distance Settlement close to the city center Settlement at the city border Settlement in the suburban area in

interaction with the core city

(labor, leisure)

Density High density of population and

housing

Medium density of population

and housing

Low density of population and

housing

Housing Mostly apartment buildings,

partly historical buildings

Mostly apartment houses but also

single and semi-detached houses

Mostly single and semi-detached

houses

Infrastructure High variety of commercial and

public facilities (for shopping,

cultural, social, leisure purposes)

Medium variety of commercial

and public facilities (mainly for

shopping and social purposes)

Basic needs available (mainly for

shopping and social purposes)

General accessibility of the city

center by foot and bike

High accessibility by both modes Accessibility of the city center by

bike

Limited accessibility

Local Public Transport Easy access to bus and tram Connected by tram or light rail,

bus

Connected by regional train, bus

Long-distance traveling Easy access to main train station Change of bus or tram necessary Change of bus or train necessary

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292282

described in Table 1. By means of the selection of these typesof city districts different levels of accessibility and transporta-tion infrastructure could be analyzed systematically. Inaddition to city districts the cities themselves served ascontrol variables measuring regional specifics that might notbe regarded in the other infrastructural variables.

However, there are also differences in the accessibility todifferent transport modes for residents within the separatedistricts and cities. Thus, the accessibility to transport systemswas also measured for each participant by individual ratingsof the access to public transport and private car. Additionally,the participants were asked about the number of cars perhousehold, possession of a driving license, a season ticket forpublic transport and of a ‘‘Bahncard’’—a discount card forfrequent users of German Rail.

2.2.3. Demographics

Sociodemographic data stating sex, age, education,occupation (full-time/part-time-employment), income,household size, number of children per household, andfamily form (single, living apart together relationship,cohabitant, married couple) were recorded.

2.2.4. Mobility behavior

The mobility behavior was measured by the participants’specifications about their daily purposes and transporta-tion modes. For 14 purposes,5 the participants were askedhow often per week or month they performed these

5The 14 purposes belong to four categories: (1) working (working/

training, trips to second home because of work); (2) shopping (daily

shopping, bulk buying); (3) private errands (trips to administration,

bringing and picking up children, supply of relatives/dependants); (4)

leisure time activities (shopping expedition, meeting the partner, meeting

activities. For each activity, data about the place, thedistance covered and the transportation mode usedwas collected. To obtain realistic information about thechoice of transportation mode and the individualmodal split, the participants could specify the use ofup to three different means of transportation per activity.Two mobility behavior variables were used as dependentvariables. Firstly, the percentage of trips conducted byprivate motorized modes6 and secondly the traveleddistance per year and person. To calculate these variables,mobility data were projected for 1 year, taking factorssuch as average working and holiday days per yearinto account. As a result, traveled distance per yearand person could be determined for all purposes andtransportation modes. Summarizing the distances ofall purposes, the total traveled distance per year andperson was determined. Moreover, for each person thefrequencies of trips by each mode of transportationper year were assessed. On this basis the percentage oftrips conducted by private motorized modes could bedetermined.

2.3. Ecological impact analysis

Assessing the ecological impact only the relevant mattersof the phase of vehicle utilization were considered: pollution,greenhouse gases, and primary energy consumption. In thisstudy we concentrate on the emissions of greenhouse gases

(footnote continued)

friends and relatives, visit of cultural events, sport/association, allotment

garden, day trips).6Private motorized modes include powered two-wheelers, private cars,

car sharing, rental cars and taxis.

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 283

in relation to mobility behavior. The greenhouse gases werecalculated as CO2 equivalents based on the Global WarmingPotential (IPCC (Intergovernmental Panel on ClimateChange), 1996).7

As Fig. 1 illustrates, there are three factors relevant tothe calculation of the ecological impact of individualmobility behavior: distance traveled, the technical data ofthe transportation mode used, and the number ofpassengers per vehicles (capacity rate). These data werelinked with the specific emission factors included in thesoftware-tool and data base TREMOD to calculate theemissions of greenhouse gases for each person caused byhis or her mobility behavior (Ifeu-Institute, 2005).

Apart from the data about the distances traveled and thetransportation modes used (cf. measure of mobilitybehavior), the participants gave detailed informationabout the technical data of private passenger cars andpowered two-wheelers.8 Information about brand andmodel as well as the following data were collected: modeof drive (Diesel, Otto, hybrid (Otto/electric), other),cubic capacity (o1.4 l; 1.4–2.0 l;42.0 l), year of construc-tion and the existence of an air conditioning system.Intra- and extra-urban trips and air conditioningsystems are relevant for fuel consumption. For short trips,which are mostly intra-urban, a cold start factor wastaken into consideration for the specific emission factor.Public transportation data were made available by thetransportation companies and was included in TREMOD(Ifeu-Institute, 2005).

Furthermore, the number of passengers per motorizedtrip is relevant for individual emission calculation. Thecapacity rate of a vehicle makes a big difference forthe individual emission rate. For motorized vehicles likethe personal car, car-sharing, rental car, and motorbike,people indicated the respective number of passengers. Forcollective vehicles, the following capacity rates (percentageof occupied seats) were applied: buses, trams, and subways25.0%; regional trains 23.9%; long distance trains 41.3%;aircraft 60.0%. These values refer to the mean capacityrates of the vehicle types in Germany, 2003.

In the last step, the mobility data were linked with thespecific emission factors. Results of these calculationsrepresent the annual emissions of each participant of thesurvey.

7The Global Warming Potential is defined as the ratio of the time-

integrated radiative forcing from the instantaneous release of 1 kg of a

trace substance relative to that of 1 kg of a reference gas (IPCC, 2001,

p. 385). The considered time horizon is 100 years. CO2 is the reference gas.

The relations of the considered gases are for Methane (CH4): 1 kg CO2/

21 kg CH4; Nitrous oxide (N2O): 1 kg CO2/310 kg N2O (IPCC, 1996).8For alternative transit modes like car-sharing cars, rental cars and

taxis, the following assumptions were made: For the car-sharing cars a

small car, for the rental car a middle class car and for the taxi an upper

class car were assumed. Public transport data were made available by the

transportation companies and are included in the database for motorized

traffic in Germany TREMOD, which was created by the Ifeu-Institute in

Heidelberg, Germany. TREMOD was also used as a basis for specific

emission factors (see Fig. 1).

3. Results

3.1. Descriptive results

3.1.1. Psychological variables

In order to confirm the independence of the fourconstructs SN, PN, PBC7, and PMN, to explore thestructure of the symbolic dimensions of mobility and toreduce the number of psychological variables to theirunderlying dimensions, a principal component analysis(PCA) with varimax rotation was carried out (see Table 2).Retaining only factors with eigenvalues greater than one,we received an eight factors solution, which was wellinterpretable and explained 58.2% of the variance.Against theoretical expectations SN and PN became one

factor, which was called ‘‘ecological norm’’. PBC and thesymbolic dimension with respect to public transportautonomy also formed one common factor. One reversedcar autonomy item similar in content also loaded on thisfactor (cf. Appendix). Because of their conceptual similar-ity it is not surprising that PBC and autonomy load on thesame factor. The common core of both constructs is thesubjective evaluation of objective behavioral options. Thus,they were combined in one construct called ‘‘publictransport control’’.Following the factor solution, perceived mobility neces-

sities became the expected independent factor.In contrast to previous findings, public transport

excitement and status loaded on the same factor. Publictransport privacy, however, could be confirmed as aseparate dimension. In case of the private car andbicycle, several symbolic dimensions were reduced to onefactor, a general car attitude and a general bike attitude.Weather resistance became a separate dimension. Theitems of the eight resulting factors are listed in theAppendix. Based on the eight factor solution mean scaleswere constructed.The four value scales were constructed following the

theoretically specified structure. Table 3 displays mean,standard deviation, and internal consistency (Cronbach’salpha) for the calculated multi-item-scales.

3.1.2. Mobility behavior

With regard to basic mobility behavior data (e.g. numberand availability of the private car), the results of our surveywere quite similar to both of the two representativemobility surveys in Germany MiD (Infas & DIW, 2004)and Mobility Panel (BMVBW (Bundesministerium furVerkehr, Bau- und Wohnungswesen), 2002). With 2.9 tripsper day and person, the number of trips in this study waslower than the result of MiD with 3.3 and of the MobilityPanel with 3.5 trips per day. Working and leisure trips weredominant in people’s mobility (cf. Table 4). The preferredmodes of transportation were private motorized modes,which were used in 48% of all daily trips. Walking wasused in 22%, cycling in 16% and public transport in 14%of all daily trips.

ARTICLE IN PRESS

Annual emissions, differentiated by individual, vehicle and activity [g/Pkm]

Direct emissions

Pollutant and fuel consumption [g]

Electric current consumption [kilowatt hour]

Indirect emissions

Fuel consumption [g emission per kg fuel consumption]

Electric current consumption [g emission per kilowatt

hour electric current consumption]

Annual distances travelled per person for each vehicle and activity1)

Specific emission factor2)

Motorbike [g/vehicle-km]

Specific emission factor2)

Private car [g/vehicle-km]

Specific emission factor2)

aircraft [g/Pkm]

Degree of capacity use [g/vehicle-km / number

of persons in the car per activity]1)

Motorbike1)

Type

Year of constr.

Cylinder capacity

Intra-/extra-urban

Taxi, CarSharing,

Rental car

Assumption of

year of constr.,

cylinder cap.,

fuel, In-/ex-urb

Buses2)

Public service

Vehicle

Coach

Intra-/extra-urban

Trains2)

Fuel

Public Electricity

Electricity of DB

AG

Aircraft2)

Fuel

1) Data Source MOBILANZ

2) Data Source IFEU

linkage

acting on

Specific emission factor2)

Public transport [g/Pkm]

Private car1)

Year of constr.

Cylinder capacity

Fuel

Air condition

Intra-/extra-urban

Fig. 1. Calculation of the individual ecological assessment.

9Interaction terms were not tested in the regression models because no

founded hypothesis about interactions between the independent variables

in their effect on dependent variables existed.

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292284

3.1.3. Ecological impact

For the ecological assessment, only motorized modeshave been taken into consideration. The importance of theprivate car for daily mobility and for ecological impact isindicated by its share of greenhouse gas emissions. Privatemotorized modes had a share of 87% in all transportation-related emissions of the participants, whereas publictransport only had a share of 13%. With regard to activitytype, most greenhouse gases were produced by trips towork followed by leisure trips (cf. Table 4).

3.2. Multivariate analysis

3.2.1. Predicting ecological impact of mobility behavior

The hypothesis that psychological variables are signifi-cant predictors of mobility-related greenhouse gas emis-sions even if infrastructural and sociodemographic factorsare controlled for (impact hypothesis) was tested with ahierarchical regression analysis predicting greenhouse gas

emissions. In this analysis psychological, sociodemo-graphic, and infrastructural variables were entered asindependent variables in two steps.9 In the first step,sociodemographic and infrastructural characteristics wereentered and already explained 40% of the variance ofgreenhouse gas emissions. The strongest predictors werefull-time employment, which was positively related to theecological impact, and age, which was negatively related tothe ecological impact. Car availability was the strongestpredictor of the group of infrastructural variables.In the second step, when psychological variables were

entered, explained variance of the resulting regressionmodel increased from 40% to 48%. Four psychologicalvariables showed significant effects, of which PMN andPBC with beta-weights of .18 and �.11, respectively,

ARTICLE IN PRESS

Table 2

Results of principal component analysis (N ¼ 1991)

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8

PBC 1a,b .67 �.05 .04 .00 .09 .05 �.37 .09

PBC 2 .66 �.01 .09 .02 .19 �.03 �.19 �.06

PT autonomy 1a,b .54 �.25 �.11 .14 .09 .20 �.03 .08

PT autonomy 2 .71 .00 .10 .13 .13 .06 �.09 �.09

Car autonomy 1b .69 �.10 .18 .06 .17 .01 �.21 .09

Car autonomy 2 �.29 .60 .10 �.05 �.01 �.04 �.07 �.28

Car privacy 1 �.07 .43 �.03 �.05 .06 �.39 �.05 �.28

Car privacy 2 �.11 .57 .11 �.10 .04 .02 �.05 �.31

Car excitement 1 �.08 .83 �.01 �.02 �.07 �.04 .06 �.02

Car excitement 2 .14 .59 �.20 .10 �.16 �.14 .05 .24

Car excitement 3 .06 .73 �.07 .06 �.05 �.13 .07 .04

Car excitement 4 �.02 .73 �.01 .06 .02 .02 .05 .03

Bicycle excitement 1 .02 �.10 .79 .07 .08 .04 �.04 .22

Bicycle excitement 2 �.10 �.02 .78 .13 .02 .08 .00 .03

Bicycle autonomy 1 .44 .11 .61 .00 �.01 �.10 .06 .10

Bicycle autonomy 2 .40 .00 .65 .06 .04 �.05 �.03 .16

PT excitement 1 .39 .11 .03 .56 �.04 .12 .19 �.02

PT excitement 2 .30 �.01 .04 .54 �.05 .33 .16 �.10

PT status 1 �.08 .00 .05 .73 .22 �.04 �.06 .09

PT status 2 .22 .03 .09 .47 .16 .01 �.09 �.16

Car status 1b �.09 �.03 .08 .68 .29 �.14 �.14 .05

SN 1 .22 .08 �.10 .08 .56 �.12 .01 .00

SN 2 .09 .00 �.02 .07 .72 .06 .10 .04

PN 1 .06 �.17 .21 .16 .64 .04 .00 .07

PN 2 .22 �.10 .11 .30 .61 .02 .04 .06

PT privacy 1a,b .12 �.11 .02 .02 .01 .83 �.07 .02

PT privacy 2a,b �.02 �.08 .01 �.02 .00 .83 �.07 .05

PMN 1 �.32 .06 .00 �.02 .12 �.06 .81 �.03

PMN 2 �.34 .06 �.01 �.06 .07 �.08 .82 .00

Weather resistance 1a,b �.08 �.09 .25 �.06 .08 .10 �.05 .77

Weather resistance 2 .12 �.04 .43 �.03 .14 .04 �.01 .68

Note: See appendix for an explanation of items.aRecoded.bReversed statement used (high agreement means low parameter value). Mean substitution was used for missing data. Pairwise and listwise deletions led

to comparable results.

Table 3

Description of psychological variables

Factors Constructs (number of items) n M SD Cronbach’s a

Ecological norm SN (2) 1962 2.60 1.00 .67

PN (2)

PT control PBC (2) 1989 3.13 1.06 .80

PT autonomy (2)

Car autonomy (1)

PT status & excitement PT status (2) 1987 2.84 .83 .66

Car status (1)

PT excitement (2)

PT privacy PT privacy (2) 1870 3.55 1.08 .72

Car attitude Car autonomy (1) 1871 3.00 .91 .80

Car privacy (2)

Car excitement (4)

Bicycle attitude Bicycle autonomy (2) 1771 3.54 .99 .77

Bicycle excitement (2)

Weather resistance Weather resistance (2) 1729 2.54 1.23 .70

PMN PMN (2) 1971 3.25 1.37 .84

Openness to change Openness to change (3) 1986 3.11 1.62 .76

Conservation Conservation (3) 1989 4.16 .88 .60

Self-enhancement Self-enhancement (3) 1988 4.44 1.46 .76

Self-transcendence Self-transcendence (3) 1988 5.24 1.32 .80

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 285

ARTICLE IN PRESS

Table 4

Mobility pattern for various mobility purposes

Trips/year and

person

Km traveled per

person and year

Average km

traveled per trip

and person

Greenhouse gas emissions in

CO2 equivalent per person and

year (kg)

Greenhouse gas emissions in

CO2 equivalent per person

and year (%)

(N ¼ 1991) (N ¼ 1991) (N ¼ 1991) (N ¼ 1935) (N ¼ 1935)

Working 286.4 4482.0 15.6 744.3 55.2

Shopping 241.6 486.6 2.0 71.6 5.3

Private errands 137.2 576.6 4.2 86.1 6.4

Leisure time 390.2 4052.6 10.4 446.9 33.1

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292286

proved to be the strongest predictors. Relations of similarquantity were only found between the ecological impactand the sociodemographic variables full-time (b ¼ .25) andpart-time employment (b ¼ .12), age (b ¼ �.15) and theliving apart together relationship (b ¼ .11). Thus, socio-demographic and psychological variables were of similarimportance for greenhouse gas emissions, whereas infra-structural variables were of minor relevance. Within thegroup of infrastructural variables only car availability andthe number of cars per household remained significantpredictors of ecological impact after psychological vari-ables had been entered. The results of the regressionanalysis are summarized in Table 5.

10The significant positive beta-weights of the both cities Augsburg and

Magdeburg, which were entered in the regression analysis, indicate that

the reference city of Bielefeld is negatively related to the ecological impact.

3.2.2. Predicting travel mode choice and traveled distances

The use of private motorized modes and traveleddistances are the most important determinants of theecological impact of individual mobility behavior. Toanalyze the relevance of psychological variables aspredictors of these two variables we conducted hierarchicalregressions, in which psychological, sociodemographic,and infrastructural variables were entered as independentvariables in two steps. Table 6 shows the results of bothregression analyses. In the first step, we entered socio-demographic and infrastructural characteristics. By far thehighest positive relation to the use of motorized privatetransport was shown by car availability and the number ofcars per household, the highest negative relation bypossession of a season ticket, which allows permanentaccess to public transportation. Moreover, people in thecity center less often used private motorized modes than inthe suburbs. This can be explained by a highly accessiblesocial and commercial infrastructure of the city center fornon-motorized modes and public transport. Regardingsociodemographic structure, full-time employment was themost important predictor for the use of private motorizedmodes. When psychological variables were entered in thesecond step, explained variance increased by 14% andreached 60%. Now, the strongest predictor of the use ofprivate motorized modes was public transport control:People with a high perceived ability to use publictransportation used the car less often. Another significantpsychological attitude was weather resistance: The higherthe sensitivity to bad weather conditions the more often

motorized private transport was used. Moreover, theattitude towards the private car positively and the attitudetowards the bike negatively predicted the use of privatemotorized modes.Apart from psychological factors, variables that pertain

to the accessibility of different transport modes remainedstrong predictors of the use of private motorized modes.Here the accessibility of the private car, which comprisescar availability, possession of a driving license, and thenumber of cars in the household as well as the possession ofa season ticket were most important. Compared to thesevariables spatial characteristics showed only little rele-vance. Mainly the city center as a district type still had asignificant effect on the use of private motorized modes.The fact that the city of Augsburg remained a significantpredictor in the regression after psychological variables hadbeen entered can be explained by regional specifics: A posthoc analysis of the traffic situation in Augsburg showed ahigh use of bicycles that decreased the use of privatemotorized modes. Finally, sociodemographic variablesturned out to be of least relevance for the prediction ofthe use of motorized modes. The only significant variablewas full-time employment.In the regression analysis explaining traveled distances,

full-time employment and age were the most importantpredictors in the first step, whereas spatial and infrastruc-tural variables were of minor importance. When psycho-logical variables were entered in the second step, explainedvariance increased by 4% from 33% to 37%. Thus, theincrease of explained variance on account of psychologicalvariables is smaller than in case of travel mode choice. Theonly significant psychological variable in this regression isPMN with a beta-weight of .18. Regarding sociodemo-graphic variables, full-time employment with a beta-weightof .24 and age with a beta-weight of �.18 were thestrongest predictors. Within the infrastructural variablesthere are not any other relations of similar quantity.However, the results show regional specifics concerning thecity of Bielefeld.10 The inhabitants of this city traveled lessdistances in their everyday life—a regional effect for whichwe could not find a valid reason in post hoc analyses.

ARTICLE IN PRESS

Table 5

Summary of hierarchical regression analysis for variables predicting

greenhouse gas emissions (N ¼ 1433)

Variable B SE b

Step 1

City center [1 ¼ yes; 0 ¼ no] �.49 .15 �.08**

Suburban area [1 ¼ yes; 0 ¼ no] .32 .17 .05

Augsburg [1 ¼ yes; 0 ¼ no] .04 .15 .01

Magdeburg [1 ¼ yes; 0 ¼ no] .37 .15 .06

Car availability [1 ¼ yes; 0 ¼ no] 1.29 .20 .19***

Driving license [1 ¼ yes; 0 ¼ no] .58 .21 .07**

Number of cars .53 .10 .15***

Access to PT �.08 .07 �.03

Distance next bus stop .02 .06 .01

Distance next rail station .08 .04 .04

Season ticket [1 ¼ yes; 0 ¼ no] .13 .14 .02

Bahncard [1 ¼ yes; 0 ¼ no] .13 .19 .01

Age �.04 .00 �.21***

Gender �.03 .12 �.01

Higher Education [1 ¼ yes; 0 ¼ no] .33 .13 .06

Number of persons per household �.24 .07 �.11***

Number of childreno18 years �.12 .10 �.03

Living apart together relationship [1 ¼ yes; 0 ¼ no] .98 .19 .11***

Income .07 .04 .05

Full-time employment [1 ¼ yes; 0 ¼ no] 1.81 .15 .31***

Part-time employment [1 ¼ yes; 0 ¼ no] 1.10 .19 .14***

Step 2

City center [1 ¼ yes; 0 ¼ no] �.38 .15 �.06**

Suburban area [1 ¼ yes; 0 ¼ no] .26 .16 .04

Augsburg [1 ¼ yes; 0 ¼ no] .27 .14 .04

Magdeburg [1 ¼ yes; 0 ¼ no] .27 .14 .04

Car availability [1 ¼ yes; 0 ¼ no] 1.14 .19 .16***

Driving license [1 ¼ yes; 0 ¼ no] .41 .20 .05

Number of cars .29 .10 .08**

Access to PT .04 .06 .01

Distance next bus stop .01 .05 .00

Distance next rail station .07 .04 .04

Season ticket [1 ¼ yes; 0 ¼ no] .20 .14 .03

Bahncard [1 ¼ yes; 0 ¼ no] .33 .18 .04

Age �.03 .00 �.15***

Gender �.11 .12 �.02

Higher Education [1 ¼ yes; 0 ¼ no] .31 .13 .05

Number of persons per household �.14 .07 �.07

Number of childreno18 years �.22 .10 �.06

Living apart together relationship [1 ¼ yes; 0 ¼ no] .98 .18 .11***

Income .07 .04 .05

Full-time employment [1 ¼ yes; 0 ¼ no] 1.45 .15 .25***

Part-time employment [1 ¼ yes; 0 ¼ no] .94 .18 .12***

Ecological norm �.13 .07 �.05

Public transport control �.31 .08 �.11***

Public transport status & excitement .21 .08 .06**

Public transport privacy .00 .05 .00

Car attitude .09 .07 .03

Bicycle attitude �.15 .07 �.05

Weather resistance �.19 .05 �.08***

Perceived mobility necessities .38 .05 .18***

Self-enhancement .09 .05 .04

Openness to change .02 .04 .01

Self-transcendence �.05 .05 �.02

Conservation .03 .07 .01

Note: The dependent variable ‘‘greenhouse gas emissions’’ was split in 10

categories of the same size so that the residues follow a normal

distribution. R2¼ .41 for step 1; DR2

¼ .08 for step 2. Adjusted

R2¼ .40 for step 1; D adjusted R2

¼ .08 for step 2. Predictors were

checked for multicollinearity: variance inflation factors (VIF) of all

variables wereo3. **po.01, ***po.001. Pairwise deletion of missing data

was used.

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 287

4. Discussion

The present study examined the relationship betweenpsychological, sociodemographic, and infrastructural vari-ables and the ecological impact of mobility behavior. Withrespect to the impact hypothesis the results confirm ourexpectation that psychological variables show significantrelationships to the ecological impact of mobility behaviorif sociodemographic and infrastructural factors are con-trolled for. Our results challenge the findings of Poortingaet al. (2004), who come to the conclusion that theecological impact of transport behavior rather dependson sociodemographic and household variables than onattitudinal variables. Our results indicate that psychologi-cal variables are not only related to intent-orientedmobility behavior, but also to the ecological impact ofmobility behavior. Two reasons can account for thesecontradictory results: On the one hand, in the present studywe used mobility-related attitudinal variables and notvalues or general beliefs as predictors for the ecologicalimpact. On the other hand, we conducted a reliablemeasurement of all relevant aspects of mobility behaviorand measured the ecological impact with a methodologicalapproach that reflects the state of the art in the assessmentof mobility-related ecological impact.Our results entail important consequences for programs

to support sustainable mobility behavior. Now policymakers can legitimize the application of soft policymeasures, because the ecological impact of mobilitybehavior is not only affected by infrastructural factors orunchangeable sociodemographic characteristics, but alsoby mobility-related attitudinal variables.At this point one could argue that there is an

explanatory gap between psychological variables and theecological impact of mobility behavior. The question hereis: In what way do attitudinal variables affect the emissionof greenhouse gases? We answer that question by ananalysis of the determinants of travel mode choice andtraveled distances as these two aspects of mobility behaviorare the most important behavioral determinants ofindividual caused greenhouse gas emissions.Concerning the travel mode hypothesis the results clearly

confirmed our expectations. Six psychological variables aresignificant predictors for the use of private motorizedmodes if sociodemographic and infrastructural factors arecontrolled for. Psychological variables were responsible foradditional 14% of explained variance compared to a modelbased on sociodemographic and infrastructural predictorsonly. Data also verified our assumption that mobility-related attitudes are better predictors of travel mode choicethan values are.A closer look at the relevance of significant predictors

reveals that public transport control is the strongestpredictor of the use of motorized private transport. Thus,the use of private motorized modes highly depends onpeople’s perception of their ability to use public transpor-tation. Similarly, a relation between the perception of

ARTICLE IN PRESS

Table 6

Summary of hierarchical regression analysis for variables predicting the percentage of trips by private motorized modes and distance traveled (N ¼ 1433)

Variable Percentage of trips by private motorized

modes

Distance traveled

B SE B b B SE B b

Step 1

City center [1 ¼ yes; 0 ¼ no] �11.06 1.82 �.15*** �.15 .16 �.02

Suburban area [1 ¼ yes; 0 ¼ no] 4.32 2.01 .06 .38 .18 .06

Augsburg [1 ¼ yes; 0 ¼ no] �8.15 1.74 �.11*** .39 .15 .06

Magdeburg [1 ¼ yes; 0 ¼ no] �.71 1.76 �.01 .58 .16 .09***

Car availability [1 ¼ yes; 0 ¼ no] 18.60 2.38 .21*** .70 .21 .10**

Driving license [1 ¼ yes; 0 ¼ no] 11.19 2.46 .11*** .55 .22 .07

Number of cars 9.27 1.21 .21*** .30 .11 .08**

Access to PT �2.05 .79 �.06** �.02 .07 �.01

Distance next bus stop .64 .66 .02 �.03 .06 �.01

Distance next rail station 1.30 .51 .05 .07 .05 .04

Season ticket [1 ¼ yes; 0 ¼ no] �14.45 1.69 �.19*** .45 .15 .07**

Bahncard [1 ¼ yes; 0 ¼ no] �10.98 2.23 �.10*** .36 .20 .04

Age �.14 .05 �.07** �.04 .00 �.24***

Gender 3.40 1.48 .05 .07 .13 .01

Higher Education [1 ¼ yes; 0 ¼ no] �.47 1.57 �.01 .39 .14 .07**

Number of persons per household �2.43 .82 �.09** �.18 .07 �.08

Number of childreno18 years 1.96 1.20 .05 �.01 .11 .00

Living apart together relationship [1 ¼ yes; 0 ¼ no] .88 2.29 .01 1.03 .20 .12***

Income .11 .47 .01 .07 .04 .05

Full-time employment [1 ¼ yes; 0 ¼ no] 8.95 1.79 .12*** 1.79 .16 .30***

Part-time employment [1 ¼ yes; 0 ¼ no] 4.68 2.26 .05 1.19 .20 .15***

Step 2

City center [1 ¼ yes; 0 ¼ no] �7.93 1.58 �.10*** �.13 .16 �.02

Suburban area [1 ¼ yes; 0 ¼ no] 4.13 1.73 .05** .32 .17 .05

Augsburg [1 ¼ yes; 0 ¼ no] �4.90 1.52 �.06** .55 .15 .09***

Magdeburg [1 ¼ yes; 0 ¼ no] �2.69 1.54 �.04 .51 .16 .08***

Car availability [1 ¼ yes; 0 ¼ no] 14.49 2.07 .17*** .62 .21 .09**

Driving license [1 ¼ yes; 0 ¼ no] 8.05 2.14 .08*** .45 .22 .06

Number of cars 4.94 1.07 .11*** .13 .11 .04

Access to PT .78 .70 .02 .05 .07 .02

Distance next bus stop .38 .57 .01 �.04 .06 �.01

Distance next rail station 1.02 .44 .04 .07 .04 .04

Season ticket [1 ¼ yes; 0 ¼ no] �11.48 1.51 �.15*** .54 .15 .09***

Bahncard [1 ¼ yes; 0 ¼ no] �5.34 1.94 �.05** .38 .20 .04

Age �.05 .05 �.02 �.03 .01 �.18***

Gender 1.98 1.36 .03 �.01 .14 .00

Higher Education [1 ¼ yes; 0 ¼ no] �.91 1.37 �.01 .34 .14 .06

Number of persons per household �.74 .71 �.03 �.12 .07 �.05

Number of childreno18 years .39 1.04 .01 �.08 .11 �.02

Living apart together relationship [1 ¼ yes; 0 ¼ no] 1.00 1.96 .01 1.05 .20 .12***

Income .43 .41 .02 .08 .04 .05

Full-time employment [1 ¼ yes; 0 ¼ no] 4.83 1.59 .07** 1.43 .16 .24***

Part-time employment [1 ¼ yes; 0 ¼ no] 4.06 1.97 .04 .98 .20 .12***

Ecological norm �2.31 .73 �.06** �.07 .07 �.02

Public transport control �8.71 .84 �.26*** �.18 .08 �.07

Public transport status & excitement 1.06 .85 .02 .16 .09 .05

Public transport privacy �.52 .59 �.02 .02 .06 .01

Car attitude 3.11 .76 .08*** .05 .08 .02

Bicycle attitude �2.72 .74 �.08*** �.09 .07 �.03

Weather resistance �4.38 .58 �.15*** .05 .06 .02

Perceived mobility necessities 2.66 .55 .10*** .39 .05 .18***

Self-enhancement .23 .50 .01 .09 .05 .04

Openness to change �.45 .48 �.02 .05 .05 .03

Self-transcendence .47 .54 .02 �.05 .05 �.02

Conservation 1.12 .74 .03 �.01 .07 .00

Note: Private motorized modes: R2¼ .47 for step 1; DR2

¼ .15 for step 2. Adjusted R2¼ .46 for step 1; D adjusted R2

¼ .14 for step 2; Distance traveled:

R2¼ .34 for step 1; DR2

¼ .04 for step 2. Adjusted R2¼ .33 for step 1; D adjusted R2

¼ .04 for step 2. The dependent variable ‘‘distance traveled’’ was split

in 10 categories of the same size so that the residues follow a normal distribution. Predictors were checked for multicollinearity: variance inflation factors

(VIF) of all variables wereo3. **po.01, ***po.001. Pairwise deletion of missing data was used.

M. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292288

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 289

mobility necessities and the use of private motorized modescould be demonstrated.

From a perspective of behavior change, the six significantpsychological predictors can be divided into two classes ofvariables: On the one hand variables that refer to thesubjective evaluation of the behavioral scope regardingtravel mode choice and on the other hand attitudes thatexpress preferences for the different transport modes. PBCand PMN belong to the first category. Car and bicycleattitude, which relate to the symbolic-affective evaluation ofthe corresponding transport modes, pertain to the secondcategory. Weather resistance is another newly conceptua-lized significant attitude dimension predicting the use ofprivate motorized modes. It belongs to the second categoryas it expresses a preference for cycling. Ecological normforms a preference for environmentally friendly transportmodes and can also be attributed to the second category.

Concerning the traveled distances hypotheses our basicexpectations were also met. Sociodemographic variablesshowed strong relationships to the traveled distances,especially age and full-time employment. Psychologicalvariables turned out to be of minor importance in theprediction of traveled distances. They only explained 4% ofadditional variance that can rather be attributed comple-tely to the only significant psychological predictor PMN.Contrary to our expectations neither values nor infra-structural variables were better predictors than mobility-related attitudes.

The predictive power of the model for traveled distancesis weaker than the predictive power of the model for travelmode choice. Consequently, additional research is neededthat adapts the TPB to traveled distances more specificallyand identifies further attitudinal variables related to thechoice of destinations.

Generally, our results confirm the empirical findings ofother studies that mobility behavior is influenced bysituational and personal factors (Collins & Chambers,2005; Hunecke et al., 2001; Van Wee et al., 2002).Moreover, the conceptualization of personal factorsderived from the theoretical background of an extendedmobility specific TPB offers a good explanatory frameworkfor travel mode choice. However, one limitation of theattitude-based approach used in the present study is that itonly describes a current state of mental representationswithout explaining the dynamics, when mental representa-tions or attitudes are changing. An alternative approachthat has to be applied in explaining travel mode choice isthe behavioral decision theory, which focuses on informa-tion processing (Svenson, 1998). But the models based oninformation processing are far from integrating the wholeset of infrastructural, sociodemographic, and psychologicalvariables that were analyzed in the present study. For thisreason it seems to be adequate to start with a staticattitudinal model that can be operationalized reliably in thecontext of survey research.

Before we finally draw conclusions from our results forprograms to support sustainable mobility, we have to

mention some methodological aspects that restrict theinterpretation of our results. Firstly, we are using correla-tional data, which should be interpreted very cautiouslyin a causal way. Secondly, the results of our regressionanalyses highly depend on the number and operationaliza-tion of the included predictors. Therefore, the beta-weightsof single predictors in our study are not directly comparableto those of other studies. Thirdly, it is still possible toimprove the measurement of mobility behavior. In thecurrent study mobility behavior was measured by aretrospective questioning, which has the disadvantage thatit is susceptible for memory effects. A methodologicalimprovement could be the measurement of mobilitybehavior with a target date method, in which participantsrecord their behavior in a short time distance to itsperformance—usually in the evening of the same day. Butfor an ecological assessment that focuses on the time slot ofa whole year, this method is not practicable because thetarget data method is limited to time slots of a few days as aresult of its high effort. Nevertheless, the results of oursurvey were quite similar to both representative mobilitysurveys in Germany, MiD and the Mobility Panel, whichare based on the target date method. Fourthly, themeasurement of the two symbolic-affective dimensions ofexcitement and status could be improved. These twodimensions should be measured as two separate factorsbecause each of them has a high relevance for the designand promotion of public transport services. Finally,infrastructural aspects could have been controlled for inmore detail on an objective level. Within the three selecteddistrict types, no differentiation was made regardingaccessibility and infrastructural situation on an objectivelevel. Individual differences were only regarded with asubjective measurement of distances to bus stops and railstations. In future studies the individual accessibility andinfrastructural setting as well as timetable data could alsobe measured objectively.In spite of these restrictions the results provide detailed

information for programs to support sustainable mobilitybehavior. The different patterns of predictors betweenthe psychological, sociodemographic, and infrastructuralfactors and the use of private motorized modes andtraveled distances clearly show the necessity for adifferentiated planning of information- and communica-tion-oriented soft policy measures. On the basis ofthe existing results we expect that an attitude-basedstrategy is more promising in achieving a change intravel mode choice than in achieving a reduction oftraveled distances. In case of travel mode choice, providinginformation can help users to realize existing mobilityservices in public transport that offer better or comparableopportunities to travel than by private motorizedmodes. Additionally, a more positive symbolic-affectiveevaluation of public transport can be advanced by theapplication of persuasive communication strategiesin the context of social marketing. However, a precondi-tion of successful social marketing programs is a public

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292290

transport system that offers user-friendly services andworks reliably.

Measures to reduce traveled distances are more difficultto design on the basis of psychological variables. Here thechoice of destinations highly depends on the perception ofpersonal necessities and constraints of spatial structure andinfrastructure, which cannot be changed by soft policyinterventions. Just the appeal for traveling less will not be avery successful intervention strategy to support sustainablemobility behavior.

One important aspect concerning traveled distancespertains to holiday trips. The focus of our study was theecological assessment of daily mobility behavior. Theecological relevance of daily mobility is due to the frequencyof trips conducted, whereas the crucial point of holidaymobility is the long distances that are covered, resulting inhigh emission rates. A comparison of the results of thepresent study with those of a study analyzing holidaymobility (Bohler, Grischkat, Haustein, & Hunecke, 2006)indicates a higher relevance of socio-economic variables—especially income and household size—for the ecologicalimpact of holiday mobility. Consequently, the strategies forchanging holiday behavior also differ from those forchanging daily mobility behavior and it is advisable fortransport policy to consider both aspects separately.

Appendix

Scale

Item

Ecological norm

People who are important to methink that I should use publictransportation instead of myprivate car (SN 1) People who are important to mewould support me in using publictransportation instead of theprivate car (SN 2) For environmental reasons I feelobliged to leave the car unused ineveryday life as often as possible(PN 1) Due to my personal values I feelpersonally obliged to useenvironmental-friendly modes like busor tram for my regular trips (PN 2)

Public transportcontrol

For me, using public transportationinstead of the private car would bedifficult in everyday life (PBC 1,recoded)

Using public transportation insteadof the private car is easy for me if Iwant to (PBC 2) If I used public transport only, Iwould feel restricted in my freedom

of movement (PT autonomy 1,recoded)

Using public transport I can doeverything I want to do (PTautonomy 2) I can deal with my everyday lifewithout a private car. (Carautonomy 1)

Public transportstatus & excitement

I’m impressed by people who covera lot of distances by publictransport (PT status 1)

I think that using public transportis trendy (PT status 2) I look up to people who arrangetheir every day life in a way thatthey do not posses a private car(Car status 1) I like public transportation becausethere are a lot of interesting thingsto see (PT excitement 1) For me using public transportationis relaxing (PT excitement 2)

Public transportprivacy

When using public transport, myprivacy is limited in an unpleasantway (PT privacy 1, recoded)

When I use public transport, otherpersons come close in anunpleasant way (PT privacy 2,recoded)

Car attitude

Driving a car means freedom to me(Car autonomy 2) I like driving a car because I candecide whom to drive with (Carprivacy 1) In my private car I feel safe andsecure (Car privacy 2) Driving a car means fun andpassion to me (Car excitement 1) Sometimes I enjoy driving withouta special destination (Carexcitement 2) Driving a car is sometimes apleasant challenge to me (Carexcitement 3) I enjoy applying my drivingcompetence (Car excitement 4)

Bicycle attitude

I love riding my bike (Bicycleexcitement 1) Riding my bike is relaxing (Bicycleexcitement 2) By bike I can get anywhere (Bicycleautonomy 1) I can reach many of my importantdestinations by bike (Bicycleautonomy 2)

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292 291

Weather resistance

I don’t like riding my bike when theweather is chilly (Weatherresistance 1, recoded) I ride my bike even in bad weatherconditions (Weather resistance 2)

PMN

The organization of my everydaylife requires a high level of mobility(PMN 1) I have to be mobile all the time tomeet my obligations (PMN 2)

Values

How important is y to you as aguiding principle of life?

Openness to change

y having an exciting life y

y having a diversified life y

y being daring y

Conservation

y social order y y national security y

y family safety y

Self-enhancement

y being ambitious y y being competent y y being successful y

Self-transcendence

y unity with nature y

y saving the environment y

y respect for nature y

References

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

and Human Decision Processes, 50, 179–211.

Anable, J. (2005). ‘‘Complacent Car Addicts’’ or ‘‘Aspiring Environmen-

talists’’. Identifying travel behaviour segments using attitude theory.

Transport Policy, 12, 65–78.

Anable, J., & Gatersleben, B. (2005). All work and no play? The role of

instrumental and affective factors in work and leisure journeys by

different travel modes. Transportation Research Part A, 39, 163–181.

Aschpurwis+Behrens GmbH (2001). BIK Regionen: Ballungsraume,

Stadtregionen, Mittel-/Unterzentrengebiete—Methodenbeschreibung

zur Aktualisierung 2000 [Bik regions—Description of method].

Retrieved May 9, 2005, from /http://www.bik-gmbh.de/texte/

BIK-Regionen2000.pdfS.

Baccini, P., & Brunner, P. H. (1991). Metabolism of the anthroposphere.

Berlin, Heidelberg, New York: Springer.

Bamberg, S. (2001). Wertetypen und Umweltverhalten. Empirischer Test

der Theorie integrierter Wertsysteme von Schwartz mit langen sowie

zwei neuentwickelten kurzen Werte-Inventaren [Types of values and

environmentally relevant behavior. An empirical test of Schwartz’s

theory of integrated value systems with the long and two short value

inventaries]. Unpublished manuscript, University of GieXen.

Bamberg, S., Hunecke, M., & Blobaum, A. (in press). Moral norm, social

context and the use of public transportation—Results of two field

studies. Journal of Environmental Psychology.

Bamberg, S., & Schmidt, P. (2001). Theory-driven subgroup-specific

evaluation of an intervention to reduce private car use. Journal of

Applied Social Psychology, 31, 1300–1329.

Bamberg, S., & Schmidt, P. (2003). Incentives, morality, or habit?

Predicting student’s car use for university routes with the models of

Ajzen, Schwartz, and Triandis. Environment and Behavior, 35, 1–22.

BMVBW (Bundesministerium fur Verkehr, Bau- und Wohnungswesen)

(2002). MOP—Mobilitatspanel Deutschland 2002 [MOP—Mobility

panel Germany 2002]. Retrieved November 1, 2005 from /http://

mobilitaetspanel.ifv.uni-karlsruhe.deS.

Bohler, S., Grischkat, S., Haustein, S., & Hunecke, M. (2006).

Encouraging environmentally sustainable holiday travel. Transporta-

tion Research, Part A, 40, 652–670.

Brog, W., Erl, W., & Mense, N. (2004). Individualised marketing:

Changing travel behaviour for a better environment. In: OECD (Ed.),

Communicating environmentally sustainable transport—The role of soft

measures (pp. 83–97). Paris: OECD Publications.

Colborn, T., Meyers, J. P., & Dumanoski, D. (1996). Our stolen future:

Are we threatening our fertility, intelligence and survival? New York:

Dutton.

Collins, C. M., & Chambers, S. M. (2005). Psychological and situational

influences on commuter-transport-mode choice. Environment and

Behavior, 37, 640–661.

Dietz, T., Stern, P. C., & Guagnano, G. A. (1998). Social structural and

social psychological bases of environmental concern. Environment and

Behavior, 30, 450–471.

ECMT. (2004). Transport and spatial policies: The role of regulatory and

fiscal incentives. Paris: OECD Publications.

Gatersleben, B., Steg, L., & Vlek, C. (2002). Measurement and

determinants of environmentally significant consumer behavior.

Environment and Behavior, 34, 335–362.

Hanson, S., & Schwab, M. (1995). Describing disaggregate flows:

Individual and household activity patterns. In S. Hanson (Ed.), The

geography of urban transportation (pp. 166–187). New York: Guilford

Press.

Harland, P., Staats, H., & Wilke, H. A. M. (1999). Explaining

proenvironmental intention and behavior by personal norms and the

theory of planned behavior. Journal of Applied Social Psychology, 29,

2505–2528.

Haustein, S., & Hunecke, M. (2007). Reduced use of environmentally

friendly modes of transport as caused by perceived mobility

necessities—An extension of the theory of planned behavior. Journal

of Applied Social Psychology, 37, 1856–1883.

Heath, Y., & Gifford, R. (2002). Extending the theory of planned

behavior: Predicting the use of public transportation. Journal of

Applied Social Psychology, 32, 2154–2189.

Hoffmeyer-Zlotnik, J. H. P. (2000). Regionalisierung von Umfragedaten.

Eine kleine Handlungsanleitung. Zentrum fur Umfragen, Methoden

und Analysen (ZUMA), How-to-Do-Reihe, No. 4, Manheim.

Hofmeister, S. (1998). Von der Abfallwirtschaft zur okologischen Stoff-

wirtschaft: Wege zu einer Okonomie der Reproduktion [From waste-

management to ecological orientated substance flow management].

Opladen: Westdeutscher Verlag.

Hunecke, M. (2000). Okologische Verantwortung, Lebensstile und Um-

weltverhalten [Ecological responsibility, lifestyles, and ecological

behavior]. Heidelberg: Asanger.

Hunecke, M., Blobaum, A., Matthies, E., & Hoger, R. (2001).

Responsibility and environment—Ecological norm orientation and

external factors in the domain of travel mode choice behavior.

Environment and Behavior, 33, 845–867.

Hunecke, M., Schubert, S., & Zinn, F. (2005). Mobilitatsbedurfnisse und

Verkehrsmittelwahl im Nahverkehr: Ein einstellungsbasierter Ziel-

gruppenansatz [Mobility needs and the choice of transport means in

the public local transport system]. Internationales Verkehrswesen, 57,

26–33.

Hunecke, M., & Schweer, I. (2006). Einflussfaktoren der Alltagsmobili-

tat—Das Zusammenwirken von Raum, Verkehrsinfrastruktur, Le-

bensstil und Mobilitatseinstellungen [Determinants of daily mobility—

The interaction of space, traffic infrastructure, life styles and mobility-

related attitudes]. In K. J. Beckmann, M. Hesse, C. Holz-Rau, & M.

Hunecke (Eds.), StadtLeben—Wohnen, Mobilitat und Lebensstil

(pp. 148–166). Wiesbaden: VS.

IEA (International Energy Agency). (2000). World energy outlook 2000.

Paris: IEA.

Ifeu-Institute (Ifeu-Institute for Energy and Environmental Research,

Heidelberg) (2005). Daten- und Rechenmodell: Schadstoffemissionen

ARTICLE IN PRESSM. Hunecke et al. / Journal of Environmental Psychology 27 (2007) 277–292292

aus dem motorisierten Verkehr in Deutschland 1960–2030. [Software

tool: Emissions resulting from motorized modes in Germany

1960–2030]. Heidelberg.

Infas & DIW (2004). Mobilitat in Deutschland. Ergebnisbericht [Mobility in

Germany. Report on results]. Retrieved November 1, 2005 from /http://

www.kontiv2002.de/pdf/ergebnisbericht_mid_ende_144_punkte.pdfS.INFRAS (Ed.), (1995). Okoinventar Transporte. Grundlagen fur den

okologischen Vergleich von Transportsystemen und den Einbezug von

Transportsystemen in Okobilanzen. Technischer Schlussbericht [Ecolo-

gical inventory of transport. Basics of an ecological comparison of

transport systems and the inclusion of transport systems in life cycle

assessments. Final technical report]. Zurich: INFRAS.

IPCC (Intergovernmental Panel on Climate Change) (1996). Climate

Change 1995: The Science of Climate Change. Contribution of Working

Group I to the Second Assessment Report of the Intergovernmental

Panel on Climate Change. Cambridge: Cambridge University Press.

IPCC (Intergovernmental Panel on Climate Change). (2001). Climate

change 2001: The scientific basis. Cambridge: Cambridge University

Press.

Lenzen, M., Dey, C., & Hamilton, C. (2003). Climate change. In D. A.

Hensher, & K. J. Button (Eds.), Handbook of transport and the

environment (pp. 37–60). Amsterdam: Elsevier.

Lovins, A., & Cramer, D. (2004). Hypercars, hydrogen, and the auto-

motive transition. International Journal of Vehicle Design, 35, 50–85.

Nordlund, A., & Garvill, J. (2003). Effects of values, problem awareness,

and personal norm on willingness to reduce personal car us. Journal of

Environmental Psychology, 23, 339–347.

Oreg, S., & Katz-Gerro, T. (2006). Predicting proenvironmental behavior

cross-nationally: Values, the Theory of Planned Behavior, and Value-

Belief-Norm Theory. Environment and Behavior, 38, 462–483.

Poortinga, W., Steg, L., & Vlek, C. (2004). Values, environmental concern,

and environmental behavior—A study into household energy use.

Environment and Behavior, 36, 70–93.

Rohrmann, B. (1978). Empirische Studien zur Entwicklung von

Antwortskalen fur die sozialwissenschaftliche Forschung [Empirical

studies to develop standardized rating scales for the social sciences].

Zeitschrift fur Sozialpsychologie, 9, 222–245.

Schwartz, S. H. (1977). Normative influence on altruism. In L. Berkowitz

(Ed.), Advances in experimental social psychology, Vol. 10

(pp. 221–279). New York: Academic Press.

Schwartz, S. H. (1992). Universals in the content and structures of values:

Theoretical advances and empirical tests in 20 countries. In M. Zanna

(Ed.), Advances in experimental social psychology, Vol. 25 (pp. 1–65).

Orlando, FL: Academic Press.

Schwartz, S. H., & Bilsky, W. (1990). Toward a theory of universal

content and structures of values. Extensions and cross cultural

replications. Journal of Personality and Social Psychology, 58, 878–891.

SRU (German Advisory Council on the Environment) (2005). Umwelt und

StraX enverkehr. Hohe Mobilitat—Umweltvertraglicher Verkehr. Son-

dergutachten [Environment and road transport. High mobility—

Environmentally sound traffic. Special Report]. Baden-Baden: Nomos.

Stead, D., & Marshall, S. (2001). The relationships between urban form

and travel patterns. An international review and evaluation. European

Journal of Transport and Infrastructure Research,, 1, 113–141.

Steg, L. (2005). Car use: Lust and must. Instrumental, symbolic and

affective motives for car use. Transportation Research, Part A, 39,

147–162.

Steg, L., Vlek, C., & Slotegraaf, G. (2001). Instrumental-reasoned and

symbolic-affective motives for using a motor car. Transportation

Research, Part F, 4, 151–169.

Stern, P. C. (2000). Toward a coherent theory of environmentally

significant behavior. Journal of Social Issues, 56(3), 407–424.

Svenson, O. (1998). The perspective from behavioral decision theory on

modeling travel choice. In T. Garling, T. Laitila, & K. Westin (Eds.),

Theoretical foundations of travel choice modeling (pp. 141–172).

Amsterdam: Elsevier.

Van Wee, B., Holwerda, H., & Van Baren, R. (2002). Preferences for

modes, residential location and travel behaviour: The relevance for

land-use impacts on mobility. European Journal of Transport and

Infrastructure Research, 2, 305–316.