Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/23540033

AComparativeStudyofAttitudeTheoryandOtherTheoreticalModelsforUnderstandingTravelBehaviour

ArticleinEnvironmentandPlanningA·February2008

DOI:10.1068/a39151·Source:RePEc

CITATIONS

35

READS

83

3authors,including:

Someoftheauthorsofthispublicationarealsoworkingonthese

relatedprojects:

DistributivejusticeandequityintransportationViewproject

MartinDijst

UtrechtUniversity

233PUBLICATIONS3,516CITATIONS

SEEPROFILE

TimSchwanen

UniversityofOxford

155PUBLICATIONS3,486CITATIONS

SEEPROFILE

AllcontentfollowingthispagewasuploadedbyMartinDijston02December2016.

Theuserhasrequestedenhancementofthedownloadedfile.

1 IntroductionAnalytical studies of travel behaviour in which statistical methods are applied toempirical data are often carried out within a utility-maximization framework. Theyassume that people's preferences or utility levels are determined by their sociodemo-graphic characteristics, ownership of technical means, skills, and other attributes ofpeople's features and land-use features. They also assume that people choose the optionor activity-travel pattern that provides the highest level of utility (Cervero, 2002;Louviere et al, 2000, for example). Such an approach can be criticized for decontextual-izing choice behaviour. Social psychologists, for instance, have argued that preferencesfor choice alternatives are not stable, which is assumed in stated choice approaches.Instead, such preferences are partly contingent on situations that are not taken intoconsideration directly in microeconomic utility functions (Fujii and Ga« rling, 2003).Moreover, geographers and sociologists would argue that, even though microeconomicstudies sometimes account for the attributes of time periods and spatial environments,they tend not to consider fully the sociomaterial contexts with which people andobjects are associated, nor how these contexts influence behavioural decisions. Suchdecisions cannot be properly understood by accounting for personal or land-useattributes, because these obtain their meanings in their interaction with other people

A comparative study of attitude theory and other theoreticalmodels for understanding travel behaviour

Martin DijstUrban and Regional Research Centre Utrecht (URU), Faculty of Geosciences, Utrecht University,PO Box 80.115, 3508 TC Utrecht, The Netherlands; e-mail: m.dijst@geo.uu.nl

Sendy FaragôCentre for Transport and Society (CTS), Faculty of the Built Environment, University of theWest of England, Frenchay Campus, Coldharbour Lane, Bristol BS16 1QY, England;e-mail: Sendy.Farag@uwe.ac.uk

Tim SchwanenUrban and Regional Research Centre Utrecht (URU), Faculty of Geosciences, Utrecht University,PO Box 80.115, 3508 TC Utrecht, The Netherlands; e-mail: t.schwanen@geo.uu.nlReceived 27 April 2006; in revised form 12 July 2006; published online 5 October 2007

Environment and Planning A 2008, volume 40, pages 831 ^ 847

Abstract. In this study we investigate the extent to which the integration of psychological mechanismsfrom attitude theory into conventional analytical approaches can advance our understanding of travelbehaviour. Three models, which explain volitions (intentions) in different ways, are specified anddiscussed: a customary model (CM) that directly links attitude and external variables (eg socio-demographics) to volition; a simplified version of the extended model of goal-directed behaviour(EMGB) that links attitude to volition via behavioural desire; and a hybrid model (HM) that integratesthe two former models. Using survey data about the volition to buy media products (books, music,and DVDs, for example) online and in-store collected in four locations in the Utrecht region, theNetherlands, we find that shopping behaviour is reasonably well explained by the simplified EMGB.Past behaviour, perceived behavioural control, and subjective norms all have a statistically significantimpact on the volition to shop online, while goal desire and perceived behavioural control signif-icantly affect the volition to shop in-store. The results of the HM indicate that it is important totake external variables, such as access to physical stores and gender, into account when explainingshopping behaviour.

doi:10.1068/a39151

ô Corresponding author.

and material objects in time ^ space (Latour, 2005; Thrift, 1996, for instance). Processesof experiencing, intending, attuning, and acting that occur in people's ecologicalsettings undeniably influence their behavioural choices.

There are various theoretical approaches that have pointed to the relevance ofstudying specific contexts for understanding daily practices. Time geography, forinstance, has long since investigated behaviour within interrelated space ^ time situa-tions as subject to constraints (Dijst and Kwan, 2005; Ha« gerstrand, 1970). Morerecently, nonrepresentational theory (Thrift, 1996), actor-network theory (Latour,2005), and the new mobilities paradigm (Sheller and Urry, 2006) have addressed behav-ioural choices within associations of relations. Although efforts inspired by theseapproaches to contextualize quantitative empirical studies on travel behaviour shouldbe stimulated, we will explore the merits of another contextual approach in this paperöattitude theory. In this theory, behaviour is investigated in sociopsychological contextsin which attitudes, perceptions of opinions of other people, and other psychologicalfactors dominate.

Various studies in travel-demand analysis have investigated the relationship betweenattitudes and revealed behaviour. Most of these revealed-preference studies have inves-tigated how attitudes affect behaviour (Koppelman and Pas, 1980; Mokhtarian andSalomon, 1997; Schwanen and Mokhtarian, 2005), although some have consideredthe dynamic interplay and two-way causality of attitudes and revealed behaviour(Golob, 2001; Golob et al, 1979; Reibstein et al, 1980). However, it is long since knownthat the correspondence between attitudes and realized behaviour tends to be ratherlow, which implies that attitudes are often poor predictors of behaviour (Kraus, 1995;Wicker, 1969, for example). To address such shortcomings, social psychologists havedeveloped theoretical models in which the relationships between attitudes and behaviourare studied, while other behavioural factors are taken into account (Armitage andConner, 2001). Examples include the theory of planned behaviour (TPB) (Ajzen, 1991)and the extended model of goal-directed behaviour (EMGB) (Perugini and Conner,2000). In these models the relationship between attitude and behaviour is mediatedby intention, the most proximal cognitive antecedent to actual behaviour (Eagley andChaiken, 1993).

In this paper we intend to investigate the extent to which the integration of psycho-logical mechanisms from attitude theory into more conventional, quantitative approachesin travel-demand analysis will advance our understanding of travel behaviour. We havetherefore constructed three types of models for understanding travel-behaviour choicesthat differ in terms of the mechanisms governing choices. The first model, the customarymodel (CM), is akin to models for explaining realized behaviour in many transportationstudies. In this model, various sociodemographic and land-use variables, skills, andpersonality traits are employed to explain shopping choices. A modified version of theEMGB developed by Perugini and Conner (2000) is the second model that we willapply. This model represents one of the main streams of attitude theory developed bysocial psychologists. Finally, a hybrid model (HM) is developed which is based ona combination of the CM and simplified EMGB. In investigating the insights thateach of these models provides, we hope to advance the understanding of variousmechanisms of behavioural choices in transportation.

To assess the empirical validity of the three models, we have applied these in aninvestigation of the adoption of in-store shopping and online shopping. Shoppingmobility has been chosen because the total frequency of shopping trips undertakenexceeds that of commuting trips which have traditionally attracted the bulk of attentionin transportation studies (Schwanen et al, 2001). In addition, as activity-based trans-port studies have suggested, the choice between in-store and online shopping has

832 M Dijst, S Farag, T Schwanen

important implications for activity pattern formation in general (Bhat et al, 2003;Salomon and Koppelman, 1988). Online shopping may, for instance, reduce individ-uals' need to chain activities into complex, multistop journeys. Furthermore, the spatialimplications of e-commerce are increasingly attracting attention from geographers(Aoyama, 2003; Wrigley and Currah, 2006, for instance).

Structural equation modelling is applied to data of online and in-store shoppingcollected in the Utrecht region, the Netherlands (Farag et al, 2007). Coefficients for thethree models introduced above will be estimated and compared. The comparison willconcentrate on behavioural insights from each model, because formal statistical com-parisons of model types are not feasible as a result of differences in the estimatedcovariance matrices (see section 3.2).

2 Theoretical framework2.1 Attitude theoriesIn attitude theory, behavioural choices are dependent on the association of a person'sgoals, attitudes, and perceived abilities with the perceptions and opinions of otherpeople with whom he or she is connected. The contextualization of behaviour offeredby attitude theory is thus sociopsychological in character. In this theory an attitude isdefined as a subjective evaluation of a behaviour, which disposes a person to behave ina certain way towards it (Eagly and Chaiken, 1993; Ga« rling et al, 1998). A crucial stepin the history of attitude theory has been the development of the theory of reasonedaction (TRA) and its successor, the TPB (Armitage and Conner, 2001). Both theories arebased on the idea that behaviours depend jointly on motivation (intention) and ability(behavioural control) (Ajzen, 1987; 1991). However, empirically, the TRA operationaliza-tion (Fishbein and Ajzen, 1975) states that intention intermediates between behaviourand the attitude towards the behaviour. Intention depends not only on attitudes, butalso on subjective norms or the perceived social pressure exerted by important others,such as parents and good friends, to perform, or not to perform, a behaviour.

The TPB (Ajzen, 1991) extended the TRA by including two behavioural controlvariables. Perceived behavioural control is the confidence an individual has to undertakea particular behaviour in a particular situation. This concept is different from locus ofcontrol as applied, for example, by Mokhtarian and Salomon (1997) and Stern (1998),which is a generalizedöand thus more decontextualizedöexpectancy that remainsstable across situations and forms of action (Ajzen, 1991). Perceived behavioural con-trol is assumed to have a direct impact on intention and behaviour (Ajzen, 1991). Actualbehavioural control, which refers to the availability of requisite opportunities andresources such as time, money, and the cooperation of others, has a direct impact onbehaviour. Because this construct is often difficult to measure, perceived behaviouralcontrol is, in attitude theory studies, usually used as a proxy for actual behavioural control(Eagly and Chaiken, 1993). It is often assumed that perceived and actual behaviouralcontrol are correlated (Ga« rling et al, 1998). However, little attention has been givenin empirical research to the operationalization of external factors that could act asconstraints on behaviour. We return to this issue later.

In the course of time, several researchers have sought to extend the TPB by intro-ducing new predictors of intentions and, in turn, behaviour (Bagozzi, 1981; Triandis,1977; Verplanken et al, 1997, for example). One extension is the incorporation of habit.A habit is a learned act that becomes an automatic response to a situation and can befunctional in obtaining certain goals or end states (Verplanken et al, 1997). In otherwords, through their direct impact on behaviour, habits limit the implementation ofan intention in behaviour, or prevent a different intention from being consideredor formed (Bagozzi, 1981; Triandis, 1977; Verplanken et al, 1997). Through habits,

Models for understanding travel behaviour 833

the appreciation of a situation in which a behaviour has to take place and the searchfor information about alternative choices are minimized (Verplanken et al, 1997).

Another important extension to the TPB is the addition of goals. These can varyfrom concrete goals, such as buying clothes, to higher level ends, such as a feeling ofwell-being acquired by wearing new clothes. Goals are considered in the EMGB(Perugini and Conner, 2000), an advanced alternative to the TPB. Habits are, in thismodel, also taken into account through the concept of past behaviour. The EMBG ispremised on the notion that the choice of behaviours is based on utilitarian considera-tions (Perugini and Conner, 2000). Behaviours are assumed to be selected because oftheir usefulness in achieving a goal. If the behaviour is important in itself (that is to say,if it is an end in itself and not a means to an end), considerations related to pleasant-ness or ease could be more important in selecting a certain behaviour than utilitarianconsiderations. The EMGB extends the model of goal-directed behaviour (MGB) byadding goal desirability (the strength of an action's end state) and goal feasibility(the ease or difficulty of reaching the end state) (Perugini and Conner, 2000).

A crucial construct in the EMGB is behavioural desire, defined by Perugini and Conner(2000, page 706) as ` the motivational state of mind wherein appraisals and reasons toact are transformed into a motivation to do so.'' Behavioural desire is treated as themost proximal antecedent of intention. This concept deals with motivation, which isbroader than the transportation concept of `preference' that implies comparing andselecting alternatives (eg car versus public transport). Anticipated emotions representpositive or negative emotions in achieving or not achieving, respectively, a certaingoal for which the behaviours are instrumental. Anticipated emotions are believed toaffect behavioural desire, together with perceived goal feasibility, goal desire, attitude,perceived behavioural control, subjective norms, and past behaviour.

Finally, the EMGB uses a wider definition of intention, namely volition, whichconstitutes a further extension of the TPB. It encompasses a direct assertion of one'sintention, a statement of engagement in plans to perform instrumental acts in theservice of goal pursuit, and an expression of commitment and of effort needed to enactthe behaviour (Perugini and Conner, 2000).

2.2 Characteristics of the proposed modelsWe investigate three models in this paper, through which we hope to advance theunderstanding of the mechanisms underlying people's behavioural choices. We applythese models on in-store and online shopping. The following models will be specifiedfor both shopping modes:1. The customary model (CM): various sociodemographics, land-use features, owner-ship of technical means (eg cars, credit cards), skills (eg Internet experience), andpersonality traits (eg feeling time pressured) are applied to explain the choice for ashopping mode. These variables can mould the preference for a shopping mode butalso constrain the choice of shopping mode (Desbarats, 1983; Ha« gerstrand, 1970;Mokhtarian and Salomon, 1994). For example, monetary and time constraints couldimpact shopping behaviour negatively. The same factor may be either a constraint ora facilitator, depending on the negative or positive sense of its presence (Mokhtarianand Salomon, 1994). Credit card ownership would be a facilitator for online buying,whereas not owning a credit card would be a constraint. Similarly, Internet experiencecould be either a constraint or a facilitator for e-shopping, while shop accessibility andcar ownership could act as such for in-store shopping (Farag et al, 2007; Forsytheand Shi, 2003; Swinyard and Smith, 2003; Van and Senior, 2000). A limitation of theCM is that there is no underlying theory to guide the inclusion of variables in the model,therefore leading to ad hoc explanations of behaviour. Moreover, attitude is directly

834 M Dijst, S Farag, T Schwanen

linked to intention or behaviour, without recognition of the importance of intermediatemotivational processes for forming intentions.2. The extended model of goal-directed behaviour (EMGB): the short overview of atti-tude theories brings us to a simplified version of the EMGB applied in this paper forchoices in shopping modes (see figure 1). The EMGB has been tested for bodyweightregulation and study effort (Perugini and Conner, 2000). Compared with these goals,purchasing a product is a comparatively easy activityöone which would evoke rela-tively weak emotions if a purchase were (un)successful. Considering this, we haveomitted goal feasibility and anticipated emotions from this study. Restrictions imposedby the acceptable length of our questionnaire prevented us from measuring all theconstructs. A selection was therefore made of those we considered most importantwith respect to shopping purposes.3. The hybrid model (HM): Attitudinal models, such as that shown in figure 1, are oftenseen as a complete theory of behavioural choices in the sense that no other variablesinfluence behaviour (Eagly and Chaiken, 1993). As mentioned above, actual behav-ioural control has been considered theoretically, but has not often been operationalizedin empirical research. Furthermore, it is unclear what causes the variation in psycho-logical constructs such as perceived behavioural control and subjective norms. There isan implicit assumption that the psychological mechanisms work in the same way forevery individual across different geographical and social contexts. It has, however, beenrecognized that external and internal constraints could affect these mechanisms inseveral ways: for example, by constricting the opportunity set and moulding the forma-tion of preferences (Desbarats, 1983; Ha« gerstrand, 1970; Mokhtarian and Salomon,1994). Thus, intentions are only partly explained, since a conceptualization of theconstraints within which intentions and behaviour occur is largely omitted in attitudetheory. Constraints could, however, play an important part in an individual's choiceprocess. These constraints are part of the CM. By integrating the EMGB and the CMwe hypothesize that external variables such as sociodemographics, land-use features,ownership of technical means, skills, and personality traits have an impact on theantecedents of behavioural desire and volition, such as subjective norms, perceivedbehavioural control, and past behaviour. Thus, the HM can be viewed as an expansionof the simplified EMGB.

Product characteristics play an important part in the decision to buy a productonline or in store (Chang et al, 2005). Search goods (books and music, for example)are more suited to buying online than experience goods (clothing and furniture, forexample). Mixing product categories in e-shopping behaviour research is likely to giveinconsistent results (Cao and Mokhtarian, 2005). We therefore focus on one type ofproduct, namely search goods (books, CDs, videos, DVDs, and computer software).

"

VolitionGoal desire

Past behaviour

Subjective norms

Behaviouraldesire

Perceivedbehavioural control

" "

!3

"

Attitude

!

Figure 1. A simplified version of the extended model of goal-directed behaviour (EMGB) (afterPerugini and Conner, 2000).

Models for understanding travel behaviour 835

3 Methodology3.1 Research design and method of analysisA questionnaire consisting of questions about daily and nondaily in-store shoppingbehaviour and attitudes, e-shopping behaviour and attitudes, Internet use, and socio-demographic data was designed. The research area consists of four municipalities (oneurban and three suburban) located in the centre of the Netherlands; they were selectedon the basis of their urbanization and shop-availability levels. See Farag et al (2007) formore details.

With the help of the municipalities' population administration, 8000 householdswere randomly drawn. The households were sent a selection questionnaire askingwhether they would like to participate in the main questionnaire and if so, how theywould like to do soöonline or with paper and pencil. Some 24% of the householdsreturned the selection questionnaire; of these, 80% were willing to participate in themain questionnaire (1566 respondents) This relatively low response rate might implythat people who have an affinity with in-store shopping, e-shopping, or both, areoverrepresented in our sample. Of the respondents willing to participate, 77% wereInternet users and therefore belonged to our research population and received the mainquestionnaire (1210 respondents). An Internet user was defined as a person using theInternet for work or private reasons. In total, 826 people completed the shoppingquestionnaire, of which 44% did so online.

Comparing our sample with a nationwide sample of Internet users (defined aspeople who have ever used the Internet), we found that our sample is characterizedby an overrepresentation of highly educated people, females, and older people.(1) In thenationwide sample 28% of the Internet users had completed an academic degree, whilein our sample 57% of the respondents had done so. This might be attributed to the factthat residents in (strongly) urbanized areas often are highly educated (Farag, 2006).Moreover, Utrecht contains two major higher educational institutions of which thosealumni who still live in Utrecht might have responded to the survey. Of the DutchInternet users in 2003, the majority was male (54%) and mostly aged between 25 and 44years (44%). In our sample, women form the majority (61%), and 54% of the respond-ents are aged between 25 and 44 years. A possible explanation of the high percentageof females in our study is that shopping appeals more to women than to men. Hence,women would be more willing to fill out a questionnaire about shopping than men.When interpreting the empirical findings of our research it has to be kept in mind thatwomen might shop more often in-store than men, and that highly educated peoplemight shop more often online than less-educated people.

Half of the sample received a questionnaire with a section that asked about thedesire to buy a media product (books, CDs, videos, DVDs, or computer software)within the forthcoming month, while the other half received a version with a sectionon buying clothes and shoes. These sections contained questions about buying a productonline and in-store. Half of the 826 respondents reported having a desire to purchaseeither media products or clothes and shoes in the forthcoming month. Of these respon-dents 245 had filled in the section about media products. Our analysis is restricted to themedia products, because the majority of products bought online are media products;comparison with buying these products in-store is therefore easier (TNS, 2002).

We have applied structural equation modelling (SEM), which is a common methodof analysis in applications of attitude theory (Hansen et al, 2004; Perugini and Conner,2000; Shim et al, 2001, for example). In SEM a variable can simultaneously be dependent

(1) Information on Internet use by the Dutch population in 2003, received upon request 11 November2005 from Statistics Netherlands, Heerlen.

836 M Dijst, S Farag, T Schwanen

(that is, an outcome variable) and independent (that is, an explanatory variable). Moreover,SEM distinguishes between direct, indirect, and total effects (Jo« reskog and So« rbom,2001). A total effect consists of a direct and one or more indirect effects via third-partyvariables. Maximum likelihood was used as the method of estimation. In addition to acovariance matrix, an asymptotic covariance matrix was calculated as input for theanalysis. In this way, standard errors and chi-squares were corrected for nonnormality(Jo« reskog, 2001). A disadvantage of constructing an asymptotic covariance matrixis that a list-wise deletion procedure is applied, and this resulted in many missingcases (19%). Therefore, we imputed values for missing items using the technique ofexpectation maximization (EM), which substitutes values for missing data through amaximum-likelihood estimation procedure (Olinsky et al, 2003). This approach performswell with small sample sizes (Olinsky et al, 2003). A recursive structural equation modelwith latent variables was estimated using LISREL software version 8.54 (Jo« reskog andSo« rbom, 2001).

There are several goodness-of-fit measures that can be used to assess the outcomeof an SEM analysis. Frequently used measures include (Golob, 2003): the root meansquare error of approximation (RMSEA), which is based on chi-square values andmeasures the discrepancy between observed and predicted values per degree of free-dom (a good model has an RMSEA value of less than 0.05); the comparative fit index(CFI), which compares the proposed model with a baseline model with no restrictions(a good model should exhibit a value greater than 0.90); and goodness-of-fit measuresthat compare the sample and model-implied variance ^ covariance matrices, such asthe standardized root mean square residual (SRMR) (a value less than 0.05 is con-sidered a good fit) and the adjusted goodness-of-fit index (AGFI) (the greater the value,the better the fit). Another goodness-of-fit measure is the Satorra ^Bentler chi-square,which takes nonnormality into account by using an asymptotic covariance matrix(Jo« reskog, 2001). Within each of the three models it is possible to compare the good-ness-of-fit indicators (see Golob, 2003) for e-shopping and in-store shopping and toassess the extent to which a specified model fits the empirical data. However, a directcomparison of goodness-of-fit indicators between the three models is very difficult dueto the varying sizes of the covariance matrices that are estimated. Hence, it is notpossible to assess formally which model explains shopping behaviour best in statisticalterms. It is, nonetheless, possible to compare the insights that each model gives in termsof the explanations offered for individuals' intentions to shop online and in-store.

3.2 Operationalization of variablesThe operationalization of the variables and their frequency distribution prior to theimputation of values for missing items are shown in table 1. In operationalizingthe various psychological constructs from attitude theory, we have strictly followedPerugini and Conner (2000), who have tested the constructs' validity. We have trans-lated their questions literally into Dutch and used these in our questionnaire. We havealso produced correlation matrices and carried out reliability tests by estimatingCronbach's alpha statistics for each construct (available from the authors uponrequest). These analyses showed that, for `subjective norms' only, the correlations ofthe original items were too low for the development of a single construct. All the otherconstructs were reliable (Cronbach's alpha being greater than 0.50).

Further, when specifying our structural equation models, we have correlated theerror terms of the various constructs in the EMGB to assess their independence(available from the authors upon request). As the correlation scores were all below0.3, we concluded that the various constructs could all be included in the models.

Models for understanding travel behaviour 837

Table 1. Operationalization and frequency distribution of variables (OL � online shopping; IS � in-store shopping).

Variables N % Mean SD

OL IS OL IS OL IS

EMGB constructsGoal desireMy desire to buy media products within the forthcoming month is:1 � very weak, 7 � very strong 240 240 5.01 5.01 1.39 1.39

AttitudeI think that to buy media products onlinea within the forthcoming month would be for me:1 � unpleasant, 7 � pleasant 240 239 3.76 4.44 1.81 2.081 � not enjoyable, 7 � enjoyable 240 236 3.67 4.35 1.71 2.041 � unattractive, 7 � attractive 239 237 2.98 4.11 1.93 2.21

Subjective normsPeople who are important to me think I should buy media products online within theforthcoming month:1 � strongly disagree, 7 � strongly agree 243 240 2.03 3.01 1.54 1.91

Perceived behavioural controlFor me buying media products online within the forthcoming month would be:1 � very difficult, 7 � very easy 244 239 5.07 6.09 1.92 1.36I am confident that I could buy media products online within the forthcoming monthif I wanted to:1 � strongly disagree, 7 � strongly agree 244 240 5.58 5.93 2.01 1.88

Behavioural desireMy desire to buy media products online within the forthcoming month is:1 � very weak, 7 � very strong 244 239 2.86 4.93 1.84 1.62I want to buy media products online within the forthcoming month:1 � false, 7 � true 243 239 2.72 4.89 1.96 1.78

VolitionI will invest time in order to buy media products online within the forthcoming month:1 � very unlikely, 7 � very likely 241 237 2.72 4.92 1.98 1.80I intend to buy media products online within the forthcoming month:1 � strongly disagree, 7 � strongly agree 243 237 2.78 5.03 1.95 1.73

838M

Dijst,

SFarag,T

Schwanen

N:/psfiles/epa4004w

/

Models

forunderstanding

travelbehaviour

839Table 1 (continued).

Variables N % Mean SD

OL IS OL IS OL IS

EMGB constructs (continued)Past behaviourHow many times did you buy media products online during last year?Number of times during last year (continuous). 238 238 2.05 11.38 2.96 7.94

External variablesFrequency of Internet use for work and/or private reasons

less than once a week 26 11at least once a week 63 25at least once a day 157 64

Number of years using Internet for work/private reasons (continuous) 242 5.33 2.71Frequency online buying for private use in the past year (continuous) 241 2.78 3.95Time pressured (` I do not have enough time for myself '')1 � completely disagree, 7 � completely agree 245 4.37 1.76Shop accessibility (number of media stores one can reach from home)

on foot within 15 minutes 242 3.76 9.94by bicycle within 20 minutes 242 27.01 24.30

Gender% female 246 61

Age (continuous) 246 39.67 11.75Income (net household income per month)

low (<1700 Euros) 69 29medium (1700 ± 3100 Euros) 89 37high (>3100 Euros) 82 34

Household car ownershipno car 56 231 car 132 5452 cars 56 23

Credit card ownership% credit card owner 246 70

a `online' was replaced by `in-store' for the questions about in-store shopping.

Attitudes were operationalized as affective attitudesöthat is, attitudes which expressan individual's (dis)liking (Mokhtarian and Salomon, 1994). Perugini and Conner(2000) also included cognitive attitudes (`facts' as viewed by the individual as usefulor effective, for example) in the EMGB. Initially, attitudes were operationalized con-taining both affective and cognitive items. However, the e-shopping model performedconsiderably better when the cognitive attitudes were left out, while the model fit ofthe in-store shopping model did not change appreciably. We therefore consider onlyaffective attitudes in the current paperöit seems that affective attitudes are moreimportant than cognitive attitudes for e-shopping.

The reader of this paper should be aware that the external variable Internetexperience is more general than the EMGB construct of past behaviour. The formerrefers to the number of years that someone has been using the Internet and thefrequency of Internet use, while the latter refers to the number of times that someonehas bought a media product online (or in-store) during the past year. Further, theshop-accessibility measures described in table 1 were developed with the Flowmap 7software (Van der Zwan et al, 2003). Regular proximity counts were used that measurethe number of stores selling books, CDs/videos/DVDs, and computer software that arespondent could reach from the place of residence (defined in four-digit postal codezones) either on foot or by bicycle within fifteen or twenty minutes, respectively.

4 ResultsIn this section the estimation results for the CM, the EMGB, and the HM arepresented both for e-shopping and in-store shopping (table 2). In the CM, attitudehas a direct effect on volition, which is similar to the way attitudes are often incorpo-rated in conventional modelling approaches in transportation studies, assuming thatattitudes directly affect revealed behaviour. The model fit is reasonably good for boththe e-shopping and in-store shopping CMs. No significant effect from attitude onvolition occurs, which is as expected, since its effect should be mediated via behaviouraldesire according to the EMGB (figure 1). There are, however, statistically significantimpacts of various external variables on the volition to shop online or in a physicalstore, indicating that it is important to account for constraints in models that explainshopping behaviour (table 2). In the e-shopping model the strongest relationship is thatpeople with more Internet experience have a stronger volition to buy a media productonline within the next month. Having a credit card also positively affects the volition tobuy a media product online. These findings show the importance of having certainresources for online buying. Further, we see that the more media stores one can reachfrom home by bicycle within twenty minutes, the weaker the volition to buy a mediaproduct online.

With respect to in-store shopping, people who own one or two cars have a rela-tively weak volition to buy a media product. Earlier research has found that owners ofone or two cars make fewer shopping trips than people who do not own a car (Faraget al, 2007; Van and Senior, 2000). A possible explanation for the findings here may bethat car owners tend to live in the suburbs and would have to make more effort toreach a shopping centre, which, in the Netherlands, are often located in the centralbusiness districts (CBDs) of towns and cities. They may be more comfort oriented andmay attach more value to efficiency compared with people without a car, which couldlead to fewer shopping trips. Furthermore, frequent online buyers have a relativelyweak volition to buy a media product in-store. It seems that the more online-buyingexperience that people have, the less inclined they are to buy media products in-store.The advantages of buying these types of product online probably outweigh theadvantages of buying them in-store.

840 M Dijst, S Farag, T Schwanen

In our simplified version of the EMGB for e-shopping and in-store shopping,attitude indirectly affects behaviour (or volition, as in our case) via behavioural desire.Overall, the EMGB performs reasonably well when applied to shopping behaviour,although the model fit for the volition to shop online is somewhat better explainedthan for the volition to shop in-store. This is, for instance, indicated by the Satorra ^Bentler chi-square statistic, which is statistically significantly larger than zero forin-store shopping. In both models behavioural desire has a statistically significantpositive impact on volition, which is also the strongest effect of all the specifiedrelationships. This finding indicates that the people who have a greater wish to buymedia products in the forthcoming month are more willing to do so either online orin-store. However, contrary to our expectations, in neither of the models does attitudehave a statistically significant effect on behavioural desire. It seems that other psycho-logical constructs are more capable of explaining the volition to shop within the nextmonth. One may have, in general, positive attitudes towards shopping, but planning togo shopping within the next month may connote a particular need that might not fitthe circumstanceöeven for those who have positive attitudes.

In the e-shopping model, past behaviour does not have a significant direct effect onvolition, but it does have the strongest (positive) effect on behavioural desire (table 2).It seems that past experience in buying media products online influences the wish to doso again in the forthcoming month. As expected, statistically significant positive effectsof perceived behavioural control and subjective norms on behavioural desire were alsofound. Concerning the total effects, we can conclude that past behaviour, perceivedbehavioural control, and subjective norms have a significant total effect on the volitionto buy a media product online, via behavioural desire (table 2). In the in-store shoppingmodel, goal desire for buying a media product generally has a significant positive effecton behavioural desire: the stronger the wish to buy a media product, the stronger thewish to buy it in-store. Perceived behavioural control has the strongest (positive) effecton the volition to shop in-store.

The HM illustrates the importance of adding external variables to the EMGB(table 2). Although the e-shopping and in-store shopping models have SRMR valuesgreater than 0.05, they perform well when the RMSEA and Satorra ^ Bentler chi-squarestatistics are considered. Nonetheless, the AGFI and CFI indicators suggest that thee-shopping model performs slightly better than the in-store shopping model. Here, incontrast with the simplified EMGB, the in-store shopping model shows significantpositive direct and total effects from past behaviour on behavioural desire and volition.This suggests that people who have bought media products in-store in the past have astronger volition to buy such products in-store in the forthcoming month. Once theimpact of external variables is taken into account, both shopping modes are thusaffected by previous shopping behaviour. The e-shopping model is similar to the oneestimated in the simplified EMGB. However, past behaviour has now a statisticallysignificant negative direct effect on volition: the more media products one has boughtonline in the past, the weaker the volition to do so again in the near future. Thiscounterintuitive result might indicate that media products bought online belong to theless-frequently purchased type of consumer goods. Hence, some time needs to passbefore one forms the volition to buy such products online again. Alternatively, thisresult could also mean that media products bought online are often impulse purchases.

In the e-shopping model several external variables affect subjective norms, perceivedbehavioural control, and past behaviour. All of the significant direct effects reportedbelow are also significant total effects on behavioural desire and volition. The subjectivenorms are affected negatively by shop accessibility: the more media stores one can reachwithin twenty minutes from home by bicycle, the less encouraging significant others

Models for understanding travel behaviour 841

Table 2. Standardized coefficients of direct and total effects (in italic) for the customary model,simplified extended model of goal-directed behaviour (EMGB), and hybrid model (significanceat least p < 0:05, unless indicated otherwise).

Customary model Simplified EMGB

online in-store online in-store

VO VO BD VO BD VO

Measurement modelAttitudePleasent to buy media product 0.989 0.973 1.002 0.974Enjoyable to buy media product 0.824 0.973 0.814 0.971Attractive to buy media product 0.487 0.882 0.480 0.883Perceived behavioural controlEasy buying media product 0.993 0.732Confident buying media product 0.417 0.622Behavioural desireDesire to buy media product 0.923 0.892Want to buy media product 0.934 0.884VolitionInvest time to buy media product 0.834 1.142 0.904 0.880Intend to buy media product 0.944 0.656 0.871 0.851Internet experienceYears using Internet 0.579Frequency Internet use 0.775

Structural modelEMGB constructsGoal desire 0.092# 0.463

0.092# 0.101# 0.463 0.448Attitude 0.060# ÿ0.043# 0.023# 0.054#

0.060# ÿ0.043# 0.023# 0.025# 0.054# 0.052#

Subjective norms 0.214 ÿ0.008#0.214 0.236 ÿ0.008# ÿ0.008#

Perceived behaviour control 0.234 0.5220.234 0.259 0.522 0.505

Behavioural desire 1.104 0.9681.104 0.968

Past behaviour 0.611 ÿ0.104# ÿ0.009# 0.1070.611 0.571 ÿ0.009# 0.009#

External variablesInternet experience 0.466

0.466Frequency online buying ÿ0.237

ÿ0.237Time pressure

Shop accessible within 15 minuteswalking

Shop accessible within 20 minutes ÿ0.218cycling ÿ0.218

Female

AgeIncome

Car ownership ÿ0.220*ÿ0.220*

Credit card ownership 0.2380.238

Goodness-of-fit indicatorsR 2 (reduced form) 0.385 0.115 0.725 0.710 0.575 0.632Degrees of freedom 19 10 40 40Satorra ±Bentler w 2 10.912 8.988 43.797 63.247p-value Satorra ±Bentler w 2 0.927 0.533 0.314 0.011Standardized root mean square residual 0.029 0.059 0.046 0.050(SRMR)

Root mean square error of approximation 0.000 0.000 0.020 0.049(RMSEA)

Comparative fit index (CFI) 0.993 0.969 0.976 0.940Adjusted goodness-of-fit index (AGFI) 0.950 0.895 0.876 0.791

* Significant at p < 0:10, # not significant.Note: VO � volition, BD � behavioural desire, AT � attitude, SN � subjective norms,PBC � perceived behavioural control, PB � past behaviour.

Table 2 (continued).

Hybrid model

online in-store

BD VO SN PBC PB BD VO AT SN PBC PB

0.972 0.9730.839 0.9720.495 0.883

0.916 0.4970.440 0.577

0.918 0.8850.929 0.866

0.897 0.8610.862 0.828

0.527 0.5150.714 0.889

0.093 0.5690.093# 0.103# 0.569 0.567ÿ0.012# ÿ0.003#ÿ0.012# ÿ0.013# ÿ0.003# ÿ0.003#0.229 0.023#

0.229 0.253 0.023# 0.023#

0.295 0.707*0.295 0.326 0.707* 0.704

1.106 0.9971.106 0.997

0.604 ÿ0.109* 0.093* 0.1140.604 0.559 0.093* 0.207

1.072 0.587 ÿ0.3840.670 0.677 1.072 0.587 0.001# 0.001#

ÿ0.436*ÿ0.308 ÿ0.307 ÿ0.436*

ÿ0.282ÿ0.083* ÿ0.092 ÿ0.282

ÿ0.127 0.135*0.010# 0.025# ÿ0.127 0.135*

ÿ0.137 ÿ0.162* ÿ0.210ÿ0.206 ÿ0.205 ÿ0.137 ÿ0.162* ÿ0.210

0.327 0.152 0.411*0.096 0.107 0.327 0.294 0.293 0.152 0.411*

0.1840.004# 0.004# 0.184

ÿ0.392 ÿ0.257ÿ0.271 ÿ0.271 ÿ0.392 ÿ0.257

0.275 ÿ0.1890.081 0.090 0.275 ÿ0.018# ÿ0.039* ÿ0.189

0.2900.175 0.162* 0.290

0.448 0.445 0.019 0.739 0.476 0.459 0.466 0.148 0.076 0.425 0.066125 12452.370 86.1761.000 0.9960.063 0.095

0.000 0.000

0.924 0.8850.794 0.752

are with respect to buying media products online. The people who perceive that theyhave a lot of control over buying media products online tend to be women, experiencedInternet users, or car owners. People who perceive that they have little control overbuying media products online tend to be people who feel time pressured, have a highincome, and have several media stores within twenty minutes from their home bybicycle.With respect to past behaviour, it was found that people who have often boughtmedia products online in the past tend to be experienced Internet users, credit cardowners, people who have relatively few media stores within twenty minutes by bicyclefrom home, or people with a relatively low income. Perhaps the latter group of peopleuse the Internet to find cheap products and so buy online more often than higherincome groups. In conclusion, Internet experience has the strongest (positive) effecton the volition to buy online via perceived behavioural control and past behaviour.

Perceived behavioural control and past behaviour are also affected by externalvariables in the in-store shopping model.With regard to perceived behavioural control,we found that frequent online buyers perceive that they have less control over buyingmedia products in-store. Their positive experience with online buying appears to makethem reluctant to undertake a shopping trip for media products. Hence, they couldperceive in-store shopping to be more difficult than e-shopping. With respect to pastbehaviour, in order to purchase media products, car owners have made fewer shoppingtrips than people who do not own a car. Thus, the negative effect of online buying andcar ownership on the volition to shop in-store in the CM reflects the direct effects ofthese variables on perceived behavioural control and past behaviour. The numberof trips made in the past to purchase media products was also affected by shopaccessibility: the more media stores one can reach from home within fifteen minuteson foot, the more shopping trips one makes. Finally, women not only perceive onlinebuying to be easy, but also perceive in-store shopping to be equally so. Online-buyingexperience has the strongest (negative) effect on the volition to buy in-store via per-ceived behavioural control. Together with the finding that shop accessibility negativelyaffects the volition to buy online, these findings might suggest substitution betweene-shopping and in-store shopping trips for media products. However, more inves-tigations are required to substantiate this conjecture and to better understand theinteractions between shopping modes, given that the relations between physical andvirtual shopping tend to be very complex and contingent upon their sociospatial context(Aoyama, 2003; Farag, 2006).

In summary, we believe that the simplified EMGB provides more insight than theCM, in which attitude and external variables (such as sociodemographics) are directlyrelated to volition. The HM, which expands the simplified EMGB with external varia-bles, provides further insights into the determinants of the constructs in the EMGB.It can thus be concluded that an individual's context and resources are importantin explaining shopping volition and, ultimately, shopping behaviour. It has been shownthat, in such a modified version of the EMGB, Internet experience has the strongest(positive) effect on the volition to buy online, while online-buying experience has thestrongest (negative) effect on the volition to buy in-store. These external variables affectthe volition to shop online and in-store via perceived behavioural control and pastbehaviour.

5 Conclusions and discussionIn this paper we have addressed the theoretical issue of whether a contextualisationconventional, quantitative approaches in travel demand analysis, through the integra-tion of psychological mechanisms from attitude theory, will increase our understandingof travel behaviour. For that purpose we constructed three different types of models

844 M Dijst, S Farag, T Schwanen

and applied these on choices in shopping for search goods. These models were (1) a CM,which includes various sociodemographics and land-use features, ownership of tech-nical means, skills, and personality traits, (2) a simplified version of the EMGB(Perugini and Conner, 2000), and (3) an HM, which is an integration of the EMGB andthe CM.

The findings suggest that the simplified EMGB (in which the effect of attitude onvolition is mediated by behavioural desire) provides more insight than the CM, inwhich attitude is directly related to volition. The EMGB performs reasonably wellwhen applied to shopping behaviour. Although no significant effects of attitude onbehavioural desire were found, other psychological constructs had statistically sig-nificant impacts. The effects of past behaviour, perceived behavioural control, andsubjective norms on volition to shop online were statistically significant, while thevolition to shop in-store was influenced by goal desire and perceived behaviouralcontrol. Estimating the HM, in which external variables were added to the simplifiedEMGB, proved to be rewarding both for the e-shopping and for the in-store shoppingmodels.

The results clearly show that application of concepts from attitude theory increasethe understanding of shopping behaviour. They nonetheless suggest that this contextualapproach is not the best choice for all types of behavioural choices. For example, thefact that the simplified EMGB was better at explaining e-shopping than in-store shop-ping choices might be explained by the degree to which the behaviour is nonhabitual.In contrast to in-store shopping, online shopping is a rather new type of behaviour,which is not yet well established in society. This probably means that this behaviour ismuch more subject to conscious weighing and the influence of opinions of others thanthe routines associated with in-store shopping.

Although additional research is necessary, we believe that the integration of theo-retically informed attitudinal concepts in conventional quantitative approaches couldalso provide novel insights into other travel-related choices. For example, revealedbehavioural or stated choice studies about the impact of new public transport lines orparking policies on mode choice for commuting, of congestion pricing on departure timechoices, of telecommuting schemes on working at home, and of real-time informationprovided via mobile technologies on route and destination choices. The usefulness of suchan approach may extend to other fields of investigation: for instance, to study the impactof information and communication technologies on social networks and the impact ofthe supply of affordable nurseries on women's participation in the labour market.

In this paper we have limited the analysis to an application of attitude theory,arguing that this type of theory only partially explains intentions because it does notadequately conceptualize the relations between intentions and behaviour. Of course,this omission could be resolved, as we have done in this paper, by using externalvariables such as shop accessibility and Internet experience as proxy for these con-straints. However, it might be more fruitful to expand this contextual approach withconceptual and methodological ideas from time geography. For example, the applica-tion of the `prism' and `action space' concepts (Dijst and Kwan, 2005; Ha« gerstrand,1970) could offer a more concrete operationalization of a person's time ^ space context.The same is true for concepts such as `relative space', `associations of actants', and`translation' (Latour, 2005; Murdoch, 2006) taken from actor-network theory. Theseconcepts may be employed to better understand the interactions between interrelatedpeople and material items at various physical distances, and to understand the trans-formation of attitudes, goals, thoughts, and emotions that may take place during theseinteractions.

Models for understanding travel behaviour 845

Acknowledgements. The authors wish to express their gratitude for the helpful comments fromDr Marco Perugini (Department of Psychology, University of Essex), Dr Jan Faber (Department ofInnovation and Environmental Sciences, Utrecht University), and three anonymous reviewersof this paper.

ReferencesAjzen I, 1987, `Attitudes, traits and actions: dispositional prediction of behavior in personality

and social psychology''Advances in Experimental Social Psychology 20 1 ^ 63Ajzen I, 1991, ` The theory of planned behavior''Organizational Behavior and Human Decision

Processes 50 179 ^ 211AoyamaY, 2003, ``Sociospatial dimensions of technology adoption: recent M-commerce and

E-commerce developments'' Environment and Planning A 35 1201 ^ 1221Armitage J C, Conner M, 2001, ` Efficacy of the theory of planned behavior: a meta-analytic

review'' British Journal of Social Psychology 40 471 ^ 499Bagozzi R P, 1981, `Attitudes, intentions, and behavior: a test of some key hypotheses'' Journal

of Personality and Social Psychology 41 607 ^ 627Bhat C R, Sivakumar A, Axhausen KW, 2003, `An analysis of the impact of information and

communication technologies on non-maintenance shopping activities'' TransportationResearch B 37 857 ^ 881

Cao X, Mokhtarian P L, 2005, ` The intended and actual adoption of online purchasing: a briefreview of recent literature'', Institute of Transportation Studies (ITS), Department of Civiland Environmental Engineering, University of California, Davis, http://www.its.ucdavis.edu/publications/2005/UCD-ITS-RR-05-26.pdf

Cervero R, 2002, ` Built environment and mode choice: toward a normative framework''Transportation Research D 7 265 ^ 284

Chang M K, CheungW, Lai V S, 2005, ` Literature derived reference models for the adoptionof online shopping'' Information and Management 42 543 ^ 559

Desbarats J, 1983, ` Spatial choice and constraints on behavior''Annals of the Association ofAmerican Geographers 73 340 ^ 357

Dijst M, Kwan M-P, 2005, `Accessibility and quality of life: time-geographic perspectives'', inSocial Dimensions of Sustainable Transport Eds K Donaghy, S Poppelreuter, G Rudinger(Ashgate, Aldershot, Hants) pp 109 ^ 126

Eagly A H, Chaiken S, 1993 the Psychology of Attitudes (Harcourt Brace Jovanovich, Orlando, FL)Farag S, 2006 E-shopping and Its Interactions with In-store Shopping (Utrecht University, Utrecht)Farag S, Schwanen T, Dijst M, 2007, ``Shopping online and/or in-store? A structural equation

model of the relationships between e-shopping and in-store shopping'' TransportationResearch A 41 125 ^ 141

Fishbein M, Ajzen I, 1975 Belief, Attitude, Intention and Behavior: An Introduction to Theory andResearch (Addison-Wesley, Reading, MA)

Forsythe S, Shi B, 2003, ` Consumer patronage and risk perceptions in Internet shopping'' Journalof Business Research 56 867 ^ 875

Fujii S, Ga« rling T, 2003, `Application of attitude theory for improved predictive accuracy of statedpreference methods in travel demand analysis'' Transportation Research A 37 389 ^ 402

Ga« rlingT, Gillholm R, Ga« rling A, 1998, ` Reintroducing attitude theory in travel behavior research''Transportation 25 129 ^ 146

Golob T F, 2001, ` Joint models of attitudes and behavior in evaluation of the San Diego I-15congestion pricing project'' Transportation Research A 35 495 ^ 514

Golob T F, 2003, ``Structural equation modeling for travel behaviour research'' TransportationResearch B 37 1 ^ 25

Golob T F, Horowitz A D,Wachs M, 1979, `Attitude-behaviour relationships in travel-demandmodelling'', in Behavioural Travel Modelling Eds D A Hensher, P R Stopher (Croom Helm,London) pp 739 ^ 757

Ha« gerstrand T, 1970, ` What about people in regional science?''Papers in Regional Science 24 7 ^ 21Hansen T, Jensen J M, Solgaard H S, 2004, ` Predicting online grocery buying intention:

a comparison of the theory of reasoned action and the theory of planned behaviour''International Journal of Information Management 24 539 ^ 550

Jo« reskog KG, 2001, `Analysis of ordinal variables 2: Cross-sectional data'',http://www.ssicentral.com/lisrel/ord2.pdf

Jo« reskogKG, So« rbomD, 2001LISREL8:User'sReferenceGuide (Scientific Software International,Lincolnwood, IL)

846 M Dijst, S Farag, T Schwanen

Koppelman F S, Pas E I, 1980, ` Travel-choice behavior: models of perceptions, feelings, preference,and choice'' Transportation Research Record 765 26 ^ 33

Kraus S J, 1995, `Attitudes and the prediction of behaviour: a meta-analysis of the empiricalliterature'' Personality and Social Psychology Bulletin 21 58 ^ 75

Latour B, 2005Reassembling the Social: An Introduction toActor-network-theory (OxfordUniversityPress, NewYork)

Louviere J J, Hensher D A, Swait J D, 2000 Stated Choice Methods: Analysis and Application(Cambridge University Press, Cambridge)

Mokhtarian P L, Salomon I, 1994, ` Modeling the choice of telecommuting: setting the context''Environment and Planning A 26 749 ^ 766

Mokhtarian P L, Salomon I, 1997, ` Modeling the desire to telecommute: the importance ofattitudinal factors in behavioral models'' Transportation Research A 31 35 ^ 50

Murdoch J, 2006 Post-structuralist Geography: A Guide to Relational Space (Sage, London)Olinsky A, Chen S, Harlow L, 2003, ` The comparative efficacy of imputation methods for missing

data in structural equation modeling'' European Journal of Operational Research 151 53 ^ 79Perugini M, Conner M, 2000, ``Predicting and understanding behavioral volitions: the interplay

between goals and behaviors'' European Journal of Social Psychology 30 705 ^ 731Reibstein D J, Lovelock C H, Dobson R P, 1980, ` The direction of causality between perceptions,

affect, and behavior: an application to travel behavior''Journal ofConsumerResearch 6 370 ^ 376Salomon I, Koppelman F, 1988, `A framework for studying teleshopping versus store shopping''

Transportation Research A 22 247 ^ 255Schwanen T, Mokhtarian P L, 2005, ` What if you live in the wrong neighbourhood? The impact

of residential neighbourhood dissonance on distance travelled'' Transportation Research D10 127 ^ 151

Schwanen T, Dieleman F M, Dijst M, 2001, ` Travel behaviour in Dutch monocentric andpolycentric urban systems'' Journal of Transport Geography 9 173 ^ 186

Sheller M, Urry J, 2006, ` The new mobilities paradigm'' Environment and Planning A 38 207 ^ 226Shim S, Eastlick M A, 1998, ` The hierarchical influence of personal values on mall shopping

attitude and behavior'' Journal of Retailing 74 139 ^ 160Statistics Netherlands, 2005, Internet use of the Dutch population in 2003, (Received upon request

11 November 2005 from Statistics Netherlands, Heerlen)Stern E, 1998, ` Travel choice in congestions: modeling and research needs'', in Theoretical

Foundations of Travel ChoiceModelingEds T Ga« rling,T Laitila, K Westin (Pergamon, Oxford)pp 173 ^ 199

Swinyard W R, Smith S M, 2003, ` Why people (don't) shop online: A lifestyle study of theInternet consumer'' Psychology and Marketing 20 567 ^ 597

Thrift N, 1996 Spatial Formations (Sage, London)TNS Interactive, 2002, ` Global e-commerce report 2002'' http://www.tnsofres.com/ger2002/Triandis H C, 1977 Interpersonal Behavior (Brooks/Cole, Monterey, CA)Van U-P, Senior M, 2000, ` The contribution of mixed land uses to sustainable travel in cities'',

in Achieving Sustainable Urban Form Eds K Williams, E Burton, M Jenks (E and F N Spon,London) pp 139 ^ 148

Van der Zwan J,Van der Wel R, Ritsema van Eck J, De Jong T, Floor H, 2003 Manual Flowmap,Version 7 Faculty of Geosciences, Utrecht University

Verplanken B, Aarts H,Van Knippenberg A, 1997, ` Habit, information acquisition, and theprocess of making travel mode choices'' European Journal of Social Psychology 27 539 ^ 560

Wicker AW, 1969, `Attitudes versus actions: the relationship of verbal and overt behaviouralresponses to attitude objects'' Journal of Social Issues 24 41 ^ 77

Wrigley N, Currah A, 2006, ` Globalizing retail and the `new e-conomy': the organizationalchallenge of e-commerce for the retail TNCs'' Geoforum 37 340 ^ 351

ß 2007 a Pion publication printed in Great Britain

Models for understanding travel behaviour 847

Conditions of use. This article may be downloaded from the E&P website for personal researchby members of subscribing organisations. This PDF may not be placed on any website (or otheronline distribution system) without permission of the publisher.