lable at ScienceDirect

Journal of Environmental Psychology 38 (2014) 76e83

Contents lists avai

Journal of Environmental Psychology

journal homepage: www.elsevier .com/locate/ jep

The bicycle and the city: Desires and emotions versus attitudes, habitsand norms

Paola Passafaro a,*, Alessandra Rimano b, Maria Paola Piccini c, Renata Metastasio a,Valeria Gambardella a, Gisella Gullace a, Consolata Lettieri a

aDepartment of Developmental and Social Psychology, University of Rome “La Sapienza”, ItalybDepartment of Social Sciences, University of Rome “La Sapienza”, Italyc Salesian Pontifical University of Rome, Italy

a r t i c l e i n f o

Article history:Available online 11 January 2014

Keywords:Sustainable mobilityBicycleAttitudesEmotionsPast behaviour

* Corresponding author. Department of DevelopmUniversity of Rome “La Sapienza”, Via dei Marsi, 78,0649917721.

E-mail address: paola.passafaro@uniroma1.it (P. Pa

0272-4944/$ e see front matter � 2014 Elsevier Ltd.http://dx.doi.org/10.1016/j.jenvp.2013.12.011

a b s t r a c t

The paper explores the motivational basis of the use of the bicycle for daily travel necessities in a big city.Different from previous research on cycling, this study used a social psychological model (Perugini andBagozzi’s model of goal-directed behaviour) as a theoretical framework and focussed on prediction ofdesires. A total of 387 residents of Rome (Italy) were administered a structured questionnaire measuringattitudes, subjective norms, perceived control, positive and negative anticipated emotions, desire, andpast behaviour concerning their use of the bike in the city. Results showed how positive emotionsmediated the effects of attitudes, perceived social norms, and negative emotions on desire. This suggeststhat the affective factors could play a more direct role in orienting people’s desire in relation to both thebehaviour considered and the particular context: a large city where mobility is based on motorisedsystems and support facilities for bicycles are scarce.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

One of the main goals of the recent transport policies proposedby the European Union is enhancing mobility within and amongEuropean cities, while at the same time reducing local congestion,accidents, and pollution (European Commission, 2012). In otherwords, people should feel free to get around in urban areas, buthave to minimise the social and ecological impact of their travelchoices. As private cars and motorbikes are still responsible for aconsiderable amount of pollution and a significant number of roadaccidents and health problems in many urban contexts, thedevelopment of alternative transport methods, such as publictransportation and bicycles, has often been recommended(European Commission, 2012). The European Commission, forexample, has suggested how most of the state members of theEuropean Union could develop bicycle-based sustainable mobilityfor their cities (Dekoster & Schollaert, 1999).

However, policies directed at such implementation have barelytaken off in most European countries, and the same can be said forthe social psychological literature which should inform such

ental and Social Psychology,00185 Rome, Italy. Tel.: þ39

ssafaro).

All rights reserved.

policies. As a matter of fact, although several studies have beenconducted with the aim of detecting the main social psychologicalfactors that affect the diffusion of sustainable mobility in urbancontexts, most of these studies have focussed on public trans-portation. Moreover, as Gatersleben and Uzzel (2007) recentlystated, most of the research on travel behaviour has been limited tostudies focussed on utilitarian appraisal, while other aspects, suchas the affective dimensions, have often been neglected or studiedmainly from the sole perspective of commuters’ stress.

More generally, little is known about how emotional factors cancombine with other social psychological factors such as attitudes,norms, and habits in explaining people’s travel mode choices. Thisis particularly true for the studies concerning the use of bikes in thecity: although many of them have investigated the relationshipexisting among selected factors (such as attitudes and everydayactual or self-reported commuting habits), to our knowledge noattempt has been made to analyse the relationship among thesepredictors using a comprehensive social psychological model as atheoretical framework. Furthermore, most studies have focussed onpeople’s intentions or self-reported behaviours (see, e.g.,Gatersleben & Haddad, 2010), but have neglected desires which, forexample, might be particularly appropriate to take into accountwhen dealing with contexts where cycling is not a common prac-tice and structural facilities are lacking. Previous research hasshown how, even in such contexts, many people can hold positive

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e83 77

attitudes toward cycling, although they might not actually developan intention to cycle (see for e.g., Gatersleben & Appleton, 2007;Oja, Vuori, & Paronen, 1998; Rimano, Piccini, Metastasio,Chiarolanza, & Passafaro, 2012; Rimano et al., submitted forpublication). This could be due to the fact that they might nothave developed the corresponding desire. Thus, analyzing desiremight be particularly useful in these cases because it might giveinsights into how cycling uptake could increase if, for example,other conditions which are known to affect intentions are positive.However, no previous study has attempted to study the de-terminants of the desire to use a bike in the city.

This paper aims to address these shortcomings by applying thetenets of a theoretical framework that enables simultaneousconsideration of both the attitudinal or cognitive (pros and cons)and the affective (emotional) implications of travel mode choices,with particular reference to the desire to use a bike for daily ne-cessities in a big city.

2. The social psychological determinants of the use of a bikein the city

Commuting by bike in the city can be “healthy” and “dangerous”at the same time; and people seem to be aware of the controversialcharacter of this means of transportation. These are some of theaspects that emerge from the international scientific literaturepublished on the topic of the use of bikes in the city. For example,studies of sports medicine and kinesiology have suggested thesubstantial health-enhancing potential of physical activity when itis of moderate intensity and is incorporated into people’s daily liferoutines, as in the case of commuting by bike to work or to school(see Matheson et al., 2011; for a recent review; see also Haskellet al., 2007; Oja et al., 1998). Considerations regarding the bene-fits for health (and the environment) are those reported more oftenby bicyclists when they are asked to explain why they commute bybike (e.g., Gatersleben & Appleton, 2007; Hopkinson & Wardman,1996). However, riding a bike in the city has its drawbacks, one ofwhich is the risk of road accidents. This is one of the main concernsof cyclists and can affect an individual’s decisions about whether touse a bike in the city or not. In this sense, its role is thought to begreater than that of other already ascertained factors, such as thegradient (or “hilliness”) of the trail, the distance from the destina-tion, the time of the ride, the climate of the area, the traffic con-ditions, and the quality of the cycling facilities (see, e.g.,Gatersleben & Appleton, 2007; Hopkinson & Wardman, 1996).These considerations are usually referred to as to the practical,instrumental, or utilitarian reasons that may motivate the use ofbicycles in an urban context. They are typically linked to the atti-tude toward this specific behaviour and to the perceived controlover it. Gatersleben and Appleton (2007) showed that attitudestoward cycling and the perceptions of various personal andexternal barriers can affect an individual’s decision to use a bike forhis or her daily commute to work or to school. Yet more generalattitudes also seem helpful in understanding the determinants ofthis behaviour. Bopp et al., assessed the role of eco-friendly atti-tudes (EFA) and found that individuals higher in EFA “were morelikely to actively commute (walk or bike to work) and less likely todrive, and reported more self-efficacy, fewer barriers, and moremotivators for AC [active commuting]” (Bopp, Kaczynski, &Wittman, 2011, p. E9). Hence, the authors concluded that “publichealthebased interventions to promote AC may use ecology-themed messages for greater reach and impact” (Bopp et al., 2011,p. E9).

However, more recent contributions have directed attentionalso to other social psychological factors able to orient people’sdecisions in favour of or against the use of bikes in the city. For

example, Gatersleben and Uzzel (2007) investigated the affectiveappraisals of the daily commute, and compared four transport al-ternatives (driving, using public transportation, walking, andcycling). Findings indicated that “walking and cycling journeys arethe most relaxing and exciting and therefore seem the most opti-mum form of travel from an affective perspective” (p.416). Resultsof qualitative studies (see, e.g., Rimano, et al., 2012) confirmed thatpeople tend to mention several emotional aspects (both positiveand negative) when asked to indicate the positive and negativeaspects that they would associate with commuting by bike in thecity. Feelings of joy and happiness (linked to the sensation offreedom, for example) are among the positive feelings recorded,but several fears are typically mentioned as well, particularly linkedto the risk of accidents (always recurrent) and health problems (forexample, due to breathing polluted air). Other important aspectswhich emerged in the literature concern the role of habits andsocial norms. Many people reported that they did not intend tocommute by bike because they were not willing to change theirhabit of using other means of transportation (typically private cars;Rimano et al., 2012). Moreover, when people were asked to indicatewho they thought would approve or disapprove of them if theydecided to commute by bike, different relevant categories tended tobe listed, ranging from relatives and peers to specific social groups,such as environmentalists and bike producers (Rimano et al., 2012).

All in all, these studies suggest that practical (utilitarian) con-siderations and emotional reactions combine to determine people’swillingness to use a bike in the city. Moreover, there is a possibilitythat normative considerations and habits might play a role as well.However, none of these studies have assessed the relative roleplayed by all of these possible determinants combined and, moregenerally, to our knowledge no attempt has beenmade to test themusing a comprehensive social psychological model. This is ashortcoming because the use of theoretical models is of crucialimportance for a full understanding of socially relevant behavioursand for planning adequate promotional strategies.

3. Predicting desires to use a bicycle in the city

A further aspect that emerges from the literature is that manypeople tend to declare that they would actually like to use a bike fortheir daily commute in the city (for example, to work or to school),but that they are discouraged by several practical obstacles, such asthe absence of cycling facilities and/or the bad quality of existingones (Gatersleben & Appleton, 2007; Rimano et al., 2012). In otherwords, there seem to be peoplewho desire to use a bikemore often,but do not for various contingent reasons. Perugini and Bagozzi(2004) have defined desire as “a state of mind whereby an agenthas a personal motivation to perform an action or to achieve a goal.Such motivation is based on an integration of different sources ofappraisals (e.g., emotional, evaluative, social, etc.) and representsthe first step towards a decision to act, typically followed by anintention to do so” (p.71). Perugini and Bagozzi (2004) furtherspecified the differences existing between desires and intentions inrelation to three criteria: the perception that the subject canperform the action (perceived performability of the action), theperceived connection of the action with a goal that the subjectwants to attain (action connectedness), and the definite versusindefinite time within which the action is framed (temporalframing). The authors have empirically demonstrated that desirescompared to intentions tend to be associated with lower perceivedperformability, lower connection to action, and a more indefinitetemporal frame (Perugini & Bagozzi, 2004). For example, a personcan have the desire to use a bike in the city even when lacking theperception that he/she will be able to do so in his/her city, and inthe absence of a specific commitment to act. Moreover, this person

Fig. 1. The model of Goal Directed Behavior (Perugini & Bagozzi, 2001). NotePNORMS ¼ prescriptive norms; DNORMS ¼ descriptive norms; PBC ¼ perceivedbehavioural control; NEGEM ¼ negative anticipated emotions; POSEM ¼ positiveanticipated emotions; PASTB ¼ past behavior.

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e8378

may desire to use a bike, but has not necessarily defined where andwhen to cycle. Intentions, on the other hand, are typically definedat a lower level of abstraction, which includes at least a basic idea ofhow, where, and when the action will take place. For this reason, itcould be said that intentions are a general measure of commitment,while desires are a general measure of motivation (Perugini &Bagozzi, 2004). The authors have proposed a specific theoreticalmodel in order to explain desires: the Model of Goal-DirectedBehaviour (MGB; Perugini & Bagozzi, 2001). As the model wasintroduced with the explicit aim of “broadening and deepening”Ajzen’s (1991) Theory of Planned Behaviour (TPB), it includes all itsconstituting factors (attitudes, subjective norms, perceived behav-ioural control, behavioural intentions, and behaviours). Moreover,it also considers relevant factors such as anticipated emotions andpast behaviour. However, in the MGB the direct predictors of in-tentions are the desire to perform a certain behaviour and thefrequency with which that behaviour has been performed in thepast, while in Ajzen’s TPB intentions were predicted directly byattitudes, subjective norms, and perceived control. Indeed, in theMGB desire (not intention) is directly predicted by attitudes, sub-jective norms, and perceived behavioural control.

Various studies have shown the efficacy of the model in theprediction of socially relevant desires, intentions, and behaviours(Chen, 2013; Kim, Lee, Lee, & Song, 2012; Leone, Perugini, &Ercolani, 2004; Perugini & Bagozzi, 2001; Song, Lee, Kang, & Boo,2012; Song, Lee, Norman, & Han, 2012), among which there isalso the use of public transportation (Carrus, Passafaro, & Bonnes,2008). In particular, these studies showed not only that the MGBmodel tends to explain more variance than the TPB, but that it also“provides an account for how reasons for acting become integratedand transformed into intentions to act, which the TPB does notaddress” (Taylor, Bagozzi, & Gaither, 2005, p.199). There are reasonsto believe that the model could be of help also in the study of theuse of bikes in the city. First, although the ultimate goal of themodel is to predict intentions and behaviour, it is the only socialpsychological model that provides a theoretical framework for thestudy of desires. Previous studies conducted on the use of bicycleshave used apparently similar constructs, such as the willingness touse a bike, but have omitted to frame it within a clear theoreticalframework. This is needed because the same studies suggest theimportance of distinguishing between having developed a desire(willingness) to use a bike, and an intention (commitment) to do so.Moreover, it has been suggested that the motivational basis (andthe promotional strategies) that would prompt people who “wouldlike” to use a bike (but do not) might differ considerably from themotivations (and strategies) that could encourage people whoalready cycle (though not frequently) to do it more regularly(Gatersleben & Appleton, 2007). Furthermore, these two groups ofpeople might differ even substantially (in terms of motivations)from those who do not consider cycling as an option at all. Morespecifically, Gatersleben and Appleton (2007) found that some ofthe peoplewho have contemplated cycling, but never tried to do so,tended to have strong positive attitudes and perceptions concern-ing the use of bikes, while they lamented the existence of too manybarriers. Hence, the authors suggested that these people could bemotivated to start using bikes by increasing their perceived controlover the behaviour. This could be done, for example, by increasingthe endowment of cycling lanes, helping these people make spe-cific action plans, and organising social events in support of the useof the bike. People who already cycle but not on a regular basistended to perceive fewer barriers, and thus could be persuaded tocycle more often by enhancing positive feedback (about theirfitness improvements, for example, their environmental awareness,or their financial savings). The authors based their analysis, and thesubsequent practical suggestions, on the work of Prochanska and

colleagues (Prochanska, 1994; Prochanska & Di Clemente, 1984)who have elaborated a model of Behavioural Change for the healthdomain. This model considers behavioural change as a processencompassing different stages, starting with precontemplation(unawareness of problems and unwillingness to take action), andproceeding through contemplation (awareness of problems andthinking about change), action planning (developing an intentionto change), Action (the change is enacted), and actionmaintenance.Although the authors do not discuss it in these terms, it is evidentthat the passage from the precontemplation to the contemplationstage requires that the individual develop some kind of desire toperform the behaviour, which can be translated into an intentionunder certain conditions. While the model provides practical sug-gestions on how to intervene in order to prompt the passage fromone stage to the other, it neither provides a clear theoretical justi-fication for the reasons why this passage should occur, nor, it seemsto encompass a sufficiently wide range of psychological de-terminants possibly involved at each stage. For this reason, whilebehavioural models such as Prochanska’s do have an evidentpractical utility, it has been shown how their efficacy can beenhanced if they are integrated with social psychological modelsthat enable researchers to identify the relevant factors involved ateach stage/level, and to explain/predict the patterns of relationshipamong these factors, while also providing a theoretical justificationfor them (Bamberg & Schmidt, 2000). Unfortunately, as alreadymentioned, cycling seems to be one of those social domains inwhich no social psychological model has been tested before.Considering that the MGB model provides an adequate theoreticaljustification for studying the determinants of desires, and seeingthat it includes a rather wide range of determinants, most of whichhave already been shown to play a role in this specific domain, itwould be worthwhile to start such tests from this specific model.

4. Aims of the study and hypotheses

The general aim of the study reported herein is to assess therelative roles of social psychological factors of a cognitive, affective,

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e83 79

normative, and behavioural nature in the prediction of the desire touse a bike for daily travel necessities in a big city. The focus is on theuse of the bike as a means to increase sustainable urban mobility ingeneral, and thus the attention was directed to all daily transportnecessities, which include commuting (to school/to work) but arenot limited to it. The Model of Goal-Directed Behaviour (Perugini &Bagozzi, 2001) was chosen as the major theoretical anchor becauseit encompasses many of the relevant factors already observed inliterature on the use of bikes in the city, and because it allows forthe prediction of desires. The major hypothesis is that, coherentlywith the model specifications (see Fig. 1), attitudes, subjectivenorms, perceived behavioural control, anticipated emotions (posi-tive and negative), and frequency of past behaviour will signifi-cantly (and independently) predict desires to use a bicycle for dailytravel necessities in a big city.

5. Method

5.1. Context and participants

The study was carried out in Rome (the city with the largestextension and population in Italy). Although a certain number ofcycling lanes are available, the city is among the big urban areaswith the lowest endowment of cycling facilities in Italy and Europe(ISFORT, 2008e2011; The Gallup Organization, 2007). The transportsystem is thus still fully based on private cars and public trans-portation, and the bicycle is an option adopted by quite a lowproportion of inhabitants (less than 5%; Agency for the control andquality of the public services of Rome, 2010; ISFORT, 2008e2011).Nevertheless, according to recent surveys, many people in Romedeclared a desire to use a bike more often for their daily transportnecessities (Agency for the control and quality of the public servicesof Rome, 2010; Rimano et al., 2012). The city therefore seems anadequate context in which to study the determinants of the indi-vidual desire to use a bike. A convenience sample of 387 residentsof Rome participated in the study. People were contacted by trainedinterviewers in public places of the city (shops, city streets, malls,squares, work places, etc.) and some of them were personal ac-quaintances of the interviewers. The sampling procedure wasaimed at balancing participants by sex and age so that the finalcomposition of participants resulted in 196 females (50.6%) and 191males (49.4%) with an age range of 18e83 (M ¼ 41,06; DS ¼ 14,19).As cycling is not a common practice in Rome, it has been difficult tofind people who regularly use this means of transportation, so thesample could not be balanced according to this aspect. Neverthe-less, participants who claimed to use a bike two times per week ormore reached 10% of the sample and this ensured at least anacceptable variability of responses for factors such as past behav-iour. Education level was “primary school” for seven participants(1.8%), “junior high school” for 26 (6.7%), “senior high school” for188 (48.6%), and “college/university degree” for 166 (42.9%). Par-ticipants’ professional occupation was heterogeneous and distrib-uted as follows: 166 (42.9%) employees in the public and privatesectors, 71 (18.3%) students, 25 (6.5%) unemployed, with theremaining 125 (42%) composed of teachers, retirees, “blue collars,”housewives/husbands, shop owners, and other working conditions.Data collection was highly accurate, so no questionnaire had to bedropped for missing responses.

5.2. Measurement instrument

A paper-and-pencil questionnaire was created, measuring atti-tudes, subjective norms, perceived behavioural control, positiveanticipated emotions, negative anticipated emotions, frequency of

past behaviour, and desire to use a bicycle for daily travel in the city.In particular:

Attitudes were measured through items of a semantic differen-tial type introduced by the following statement: “For me, using abicycle in Rome for my daily travel necessities is..” Responseswere recorded on a 6-point bipolar scale featuring the followingadjectives: “good/bad; appropriate/inappropriate; right/wrong;useful/useless.”

Prescriptive norms were measured through the following twoitems: “Most people important to me think that I should use a bi-cycle for my daily travel necessities in Rome”; “Most peopleimportant to me would agree if I used a bicycle for my daily travelnecessities in Rome.” Responses to these items ranged from 1(completely disagree) to 6 (completely agree).

Descriptive norms were measured through the following twoitems: “How many of the people who are important to you use abicycle for their daily travel necessities in Rome?”; “Most of thepeople important to me use a bicycle for their daily travel neces-sities in Rome.” Responses to these items ranged from 1 (few) to 6(many), and from 1 (completely agree) to 6 (completely disagree),respectively.

Perceived behavioural control was measured through two itemsasking respondents to rate (on a 6-point scale) how difficult/easyand complicated/simple it was for them to use a bicycle for theirdaily travel necessities in Rome.

Frequency of past behaviourwasmeasured through the followingitem: “How often do you use a bicycle for your daily travel neces-sities in Rome?” Responses were recorded on a 6-point scaleheaded as follows: 1 (never), 2 (1e2 times per year), 3 (1e2 timesper month), 4 (1e2 times per week), 5 (more than two times perweek), and 6 (I always use this means).

Positive and negative anticipated emotions were measured bymeans of a set of 16 items (eight for positive emotions and eight fornegative emotions) introduced by a question expressed in theconditional form: “If you used a bicycle for your daily travel ne-cessities in Rome, how.do you think you would feel?” The itemsfeatured the following positive emotions: “happy, excited, glad,gratified, proud of yourself, satisfied, self-assured, relaxed.” Nega-tive emotions proposed were: “angry, strained, disappointed,frustrated, fearful, unsatisfied, depressed, discontented.” A 6-pointresponse scale was used in this case ranging from 1 (not at all) to 6(completely). The range of emotions proposed was selected on thebasis of the results of the study by Rimano et al. (2012), in whichspecific emotional reactions emerged from the interviews con-ducted on a sample of Rome residents.

Desirewas measured through the following two items: “I desireto use a bicycle for my daily travel necessities in Rome”; “My desireto use a bicycle for getting around in Rome is..” Responses to theseitems were recorded on a 5-point bipolar scale, ranging from 1(extremely false/weak) to 6 (extremely true/strong).

The questionnaire also included measures concerning otheraspects that are not taken into account in this discussion.

6. Results

6.1. Descriptive statistics

The number of observed variables was reduced in order to avoidcomputational problems and obtain smaller standard errors in thesubsequent statistical analysis. Cronbach’s a was computed tocheck for the internal consistency of all measures, with theexception of past behaviour (measured through a single item).Table 1 shows the pattern of correlations among the latent con-structs and the reliability check (Cronbach’s a) for the corre-sponding measures. As can be seen, age showed no statistically

Table 1Bivariate correlations, reliability check and descriptive statistics.

1 2 3 4 5 6 7 8 Mean scores

1. Attitudes .77 4,29 (1.11)2. Prescriptive norms .43* .67 2.86 (1.31)3. Descriptive norms .21* .43* .59 1.93 (1.01)4. Perceived behavioral control .30* .35* .27* .79 2.83 (1.39)5. Positive anticipated emotions .50* .36* .18* .31* .89 3.52 (1.12)6. Negative anticipated emotions �,46* �.35* �.08 �.24* �.56* .87 2.38 (1.07)7. Past behavior .30* .35* .24* .37* .35* �.19* e 1.77 (1.10)8. Desire .40* .34* .21* .29* .64* �.38* .47* .95 3.41 (1.64)9. Age �.09 �.11* .00 .05 .00 �.06 �.06 �.02 41.06 (14.19)

Note: N ¼ 387; * Correlations are significant at the .05 level or lower. Cronbach’s alpha are reported in italics in the diagonal; standard deviations are reported in parenthesis;Past behaviour and age were measured through a single item.

Fig. 2. Modified model predicting desire to use the bicycle for the daily travel necessities in the city. Note: N ¼ 387; PNORMS ¼ prescriptive norms; DNORMS ¼ descriptive norms;NEGEM ¼ negative anticipated emotions; POSEM ¼ positive anticipated emotions; PASTB ¼ past behavior.

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e8380

significant correlations with the constructs considered in the study(with the sole exception of prescriptive norms), so it was not usedin the subsequent analysis.

6.2. Test of the MGB

Structural equation modelling (SEM) with latent variables wasapplied to the data in order to test the original MGB model (Bollen,1989). To assess the fit of the model, we relied on various indexesbesides the classic c2 test. The high sensitivity of c2 to sample sizemay in fact result in an overestimation of even small differencesbetween the observed and the predicted matrix. This can lead toerroneous rejection of acceptable solutions (Bollen, 1989). There-fore, the RMSEA (Root Mean Square Error of Approximation), NNFI(Non-Normed Fit Index) and CFI (Comparative Fit Index) were alsoconsidered. Overall fit indexes showed that the hypothesisedmodel did not fit the data well: not only was c2 statistically sig-nificant, but the alternative indexes also led to the rejection of thissolution (RMSEA¼ .15; CFI¼ .76; NNFI¼ .70). In particular, contraryto our expectations, only positive anticipated emotion and pastbehaviour were shown to be statistically significant predictors ofdesires. Apparently, these results are not due to multicollinearityproblems, as in no case did tolerance values lower than .59 emerge.However, inspection of the modification indexes indicated severalpaths among the predictors, which were added in a modifiedmodel. Moreover, a check on the patterns of correlations among the

predictors also confirmed that a different pattern of relationshipsexisted among the constructs considered.

6.3. Modified model

In the modified model, positive anticipated emotions and pastbehaviour were considered the only direct predictors of desires.Moreover, positive anticipated emotions were expected to mediatethe effects of negative anticipated emotions, attitudes, andperceived behavioural control on desire. Negative anticipatedemotions were expected to mediate the effect of prescriptive anddescriptive norms on positive anticipated emotions. A direct effectof attitudes on negative anticipated emotions was included in themodel, while perceived behavioural control was excluded from it.The final model tested is reported in Fig. 2. The modified modelcontains 15 observed variables and eight latent factors. It explains57% of variance, and, although the c2 was statistically significant,the alternative indexes indicated a satisfactory fit for this model(RMSEA ¼ .046; CFI ¼ .99; NNFI ¼ .98).

7. Conclusions and discussion

Previous studies have suggested that factors such as emotionsand desires, as well as attitudes, habits, and social normsmight playa crucial role in the understanding of the use of bikes in an urbancontext. However, no attempt has previously been made to deter-mine the pattern of relationships existing among these factors with

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e83 81

regard to this specific behaviour. As Perugini and Bagozzi’s (2001;2004) Model of Goal-Directed Behaviour encompasses all of thesefactors, it was used as the theoretical framework for this study inorder to verify whether the pattern of relationships among theconsidered predictors was coherent with the one expected on thebasis of the model. Results confirmed that emotions, attitudes,habits, and social norms all play a role in directly or indirectlypredicting people’s desire to use bikes in the city. However, thepattern of relationships that emerged among the hypothesisedpredictors appeared to be different from that expected on the basisof the model. In particular, our results indicated that positiveanticipated emotions and past behaviour are in this case the mostdirect predictors of desire, while social norms and attitudes seemsto predict desire indirectly through anticipated positive and nega-tive emotions. Such findings are unusual in the literature on theMGB and need to be taken cautiously until they will be supportedby further studies on different samples of people. Nonetheless theymay yield some explanations. For example, Taylor (2007) assertedthat a certain variability in results should be expected fromempirical investigations using models of judgement and decisionmaking such as the MGB, “particularly related to differences inrespondent’s level of affective versus cognitive involvement” withthe stimulus (Taylor, 2007, p.740). Indeed, there are cases in whichthe affective responses to a stimulus may “overwhelm analyticcomputations” (Taylor, 2007, p. 741; see also Wilson & Arvai, 2006).Thus, it is possible that emotions play a more prominent role thanexpected in determining people’s desire to use a bike in the city.This in turn could be due to the fact that the conceptual relationshipbetween emotions and desires is particularly strong in this case.However, no study has determined which pattern is more likely toemerge when a stimulus tends to elicit more affective rather thancognitive responses (or vice versa), so the plausibility of such anexplanation has to be ascertained by further investigations.

Particularly curious is the mediation effect of positive antici-pated emotions on the relationship between negative anticipatedemotions and desires. This could be due to a redundancy due to thefact that these concepts are conceptually related: the twoemotional response measures might account for the same variancein other response variables. In other words, this seems to be a casein which positive and negative emotions do not provide indepen-dent explanations of desires. Other possible explanations for theseresults concern measurement issues regarding the number of in-dicators used to build the corresponding measures. Indeed, some ofthe constructs of the model (namely descriptive norms, subjectivenorms, perceived behavioural control, and desires) were measuredusing only two indicators. Although, Cronbach’s alpha valuesindicated acceptable levels of reliability for these constructs, infuture studies it would be sensible to use more than two indicatorsfor measuring the corresponding variables. However, besides thenumber of indicators, other measurement issues might also haveaffected the results obtained. For example, it was shown how,within models such as the MGB, the pattern of relationships amongthe considered factors may vary in the function of the specificcomponents of the constructs which are measured. In this sense,Taylor (2007) showed how measuring attitudes in relation to their“utilitarian” rather than their “hedonic” component may lead todifferent results. In this specific case, the author found that desireswere more positively related to the “utilitarian” than to the “he-donic” component of attitudes. In the current study this distinctionwas not made, as the two components were measured univocally,and this might have weakened the predictive capacity of thisconstruct in relation to desire. Similarly, Taylor, Ishida, and Wallace(2009) found that desires correlated positively with PBC when thiswas measured as “control,” but did not correlate with it when thiswas measured as “difficulty.” This might explain why, in the study

here reported, perceived behavioural control did not predict desireto use a bike in the city over and above the other factors considered.In other words, the weakness of the PBC as an additional predictorof desires could be ascribed to the fact that it was measured as“difficulty” rather than as “control.” In this sense, another differ-ence in the measures used in the current study concerns subjectivenorms. Most of the studies that adopted the MGB model as atheoretical framework have tended to measure subjective norms as“prescriptive norms”: that is, as the normative beliefs about theexpectations of others related to the considered behaviour. Variousstudies have demonstrated that “descriptive norms” (concerningindividual beliefs about whether relevant others would or wouldnot perform the target behaviour) are also important aspects toconsider, depending on the kind of behaviour studied. Carrus et al.(2008) have shown the plausibility of distinguishing the two kindsof measures in relation to the use of public transportation whenadopting the MGB framework. Following Carrus et al. (2008), thedistinction in the current study between prescriptive and descrip-tive subjective norms was maintained, but this might have affectedthe pattern of relationships which finally emerged.

Nonetheless, considering the results of previous qualitative andquantitative studies on the use of bikes in the city, our results areonly partially surprising. These studies provided indications thatemotions do have an important role in explaining people’s desiresto use a bike in the city. For example, while most of the literature ontravel mode choice has focussed on the cognitive/utilitarian factors,Gatersleben and Uzzel (2007) found that the role of affective re-sponses had been overlooked in this field. The authors showed howevery travel mode elicits a different affective responsewhich can beinterpreted along the two main affective dimensions developed byRussell and Snodgrass (1987): pleasantness and arousal. The resultsof their study showed howcycling tends to be perceived as pleasantand exciting by bicyclists, and, although these authors did not takeinto consideration the emotional response of those who do notcycle, other studies did, obtaining similar results. For example,emotions were found to be a crucial component of the personalexperiences reported by cyclists and non-cyclists interviewed in astudy by Rimano et al. (2012). The emotions expressed were ratherarticulated and in many cases mixed feelings were recorded.Indeed, the authors reported that the idea of using a bike in the cityas a primary means of transportation seemed able to elicit bothhopes and fears, enthusiasms and reservations, appreciation andworries. Such emotions were explained by participants in relationto the various positive and negative outcomes associated with theuse of bikes in an urban context. Fears were thus mainly linked tobeliefs regarding the risk of accidents and several other well-known inconveniences (distance/time of the ride, weather condi-tions, hilliness of the trail, etc.). However, a number of positiveimplications emerged as well, some of which were linked to utili-tarian aspects such as economic savings and physical wellness,while others were more connected to symbolic motivations(freedom, beauty, psychological well-being, etc.). Hence, emotionsare an important component of people’s experience of the use ofbikes in an urban context, and their role should be further inves-tigated. The data here discussed seems, for example, to indicatethat positive and negative anticipated emotions might be based onpartially different antecedents. While both positive and negativeanticipated emotions are directly predicted by attitudes, the latterare also directly predicted by social norms, while the former arenot. Moreover, positive anticipated emotions are predicted by pastbehaviour, while negative anticipated emotions are not. Thus, itmight be that, at least in this specific case, negative anticipatedemotions could be linked to the existence of shared beliefs con-cerning the riskiness of riding a bicycle in the city, while positiveemotions might be more linked to the personal experience of

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e8382

individuals who have ascertained the existence of many beneficialeffects. Although the existence of negative beliefs shared by thepopulation from which the sample was extracted is pure specula-tion here, there are cues that this might not be a hazardous hy-pothesis. Indeed the study of Rimano et al. (2012) and Rimano et al.(submitted for publication) focused on the use of bikes in an urbancontext and found evidence in support of the Agenda SettingTheory (e.g., McCombs & Shaw, 1972; Shaw, 1979; see also Atkin,1980). According to this theory, a correspondence tends to existbetween the kind of arguments people express in relation to so-cially relevant objects and the way in which these objects areaddressed by the mass media. In their study the authors found thatalthough the arguments reported by the people interviewed weremore articulated, most of the themes which emerged were alsopresent in the discourse of a set of national online newspapersanalysed in that occasion which dealt with the theme of bikes. Inparticular, while the risk of accidents was one of the fears mostcited by the subjects interviewed, report of accidents in which cy-clists were involved represented the most recurrent topicaddressed by the online newspapers in the period considered.

An aspect worthy of assessment in future studies is whether theapparent prominent role of emotions in determining the individualdesire to use a bike is a peculiarity of this kind of domain, or couldalso concern other kinds of behaviours, and if so, why. For example,it could be speculated that the bike is a non-motorised means oftransportation and that this could have some kind of implicationsin terms of what their relevant predictors are and the pattern ofrelationships that exist among them. Previous studies have foundhow substantial differences exist between cycling or walking (non-motorised ways of travelling) and driving or using public trans-portation (motorised ways of travelling) in terms of the emotionaland attitudinal responses elicited (Gatersleben & Appleton, 2007;Gatersleben & Uzzel, 2007). However, the implications of thesedifferences for the prediction of desires have not been systemati-cally ascertained so far. Future studies could thus compare thedesire to use non-motorised means of transportation (such as thebike) with the desire to use motorised ones in order to ascertainwhether differences in the role of emotions could also be trackedback to this feature.

The study reported herein also has several practical implica-tions. First, if our results can be confirmed by other studies, theymight indicate that in order to increase people’s desire to use bikesin the city, local administrators should plan strategies to have botha cognitive and an emotional impact on urban residents (i.e.,changing attitudes and norms, and turning negative emotions intopositive ones). A first step in this sense might be that of assessingpeople’s behavioural and normative beliefs as suggested by Ajzen(1991) in order to understand which are the positive and nega-tive beliefs that support social norms and personal attitudes. It alsoseems important to measure people’s intentions and actualbehaviour. Indeed, a limit of our study is that we did not measuresuch constructs, so wewere unable to assess the relationship of thepredictors with them or to compare our participants’ desires withtheir actual intentions to use a bike. Such comparisons might berelevant for clarifying the extent of the actual demand for cyclingfacilities in an urban context, distinguishing, for example, thosewho have the desire to use a bike and actually intend to do it fromthose who have the same desire but do not actually intend to do it.This could be particularly meaningful for those urban contexts (likethe one investigated here) where cycling facilities are lacking, andmobility is still based on private cars and public transportation.While 56% of the participants in this study declared that they neverused a bike for their daily travel in Rome, 40% of them stillexpressed some desire to do so. It is thus possible to imagine thatany attempt to increase the use of bikes in a context similar to the

one considered here should be directed first at identifying thesekinds of people and trying to understand how to help themtransform their desire into intentions and then into behaviour byremoving possible barriers and/or adopting possible supportingmeasures (see, for example, Geller, 2002; Vandenbergh, Stern,Gardner, Dietz, & Gilligan, 2010). This could have a circular effect,because as the number of people using bikes in the city increases,perceptions of social norms in favour of the use of bikes shouldincrease as well: these in turn could affect positive beliefs andsubsequently influence the desire of those who were not in favourof using such means of transportation. An issue that remains openis whether different results would have been obtained if the studyhad been undertaken in a different urban context and/or with adifferent sample of participants. Such an issue acquires substantialrelevance if one considers the discrepancies recorded here betweenthe expected (MGB) model and the data driven one. However,addressing such an issue will entail a deeper understanding of thecontextual aspects able to affect the individual desire to ride a bikeover and above the psychological factors here considered (atti-tudes, social norms, perceived control, anticipated emotions, andpast behaviour). In this sense, meaningful factors to take into ac-count could be the overall population size of the city because, forexample, surveys conducted in Italy and in Europe found that bikeusage reaches the lowest rates in urban areas with high populationsize and the highest rates in urban areas with low population size(e.g., ISFOR, 2008; The Gallup Organization, 2007). Yet these samestudies found that, in some cases, cities with comparable popula-tion size have substantially different rates of bike usage, thus sug-gesting the existence of an interaction with other contextualaspects as well. These could be, for instance, the territorialconformation of the urban area and its climate, the overallendowment of cycling facilities, and/or the perceived quality ofthese (Hopkinson & Wardman, 1996). Also, the economic structureof the city might play a role because, for instance, industrial citiescould have different transport necessities than tourist cities.However, more studies are needed in order to ascertain whetherany of these (or other) contextual factors could affect the re-lationships among the psychological factors here investigated, andwhy. In this sense, future studies could also use more heteroge-neous samples of participants in terms of, for example, cyclingexperience. It is reasonable to think that desires, goals, and ex-pectations of people who habitually use a bike in the city might bedifferent from those who never do so, and thus distinctions couldbe made on the basis of the various possible reasons people do ordo not cycle.

References

Agency for the control and quality of the public services in Rome. (2010). Indaginesulle piste ciclabili a Roma [A survey on cycling lanes in Rome]. Retrieved June,18th 2013 http://www.agenzia.roma.it/documenti/monitoraggi/430.pdf.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and HumanDecision Processes, 50, 179e211.

Atkin, C. K. (1980). Effects of the mass-media. NewYork: McGraw-Hill.Bamberg, S., & Schmidt, P. (2000). Regulating transport: Behavioural changes in the

field. Journal of Consumer Policy, 22, 479e509.Bollen, K. (1989). Structural equations with latent variables. New York: Wiley.Bopp, M., Kaczynski, A. T., & Wittman, P. (2011). The relationship of eco-friendly

attitudes with walking and biking to work. Journal of Public Health Manage-ment Practice, 17, E9eE17.

Carrus, G., Passafaro, P., & Bonnes, M. (2008). Emotions, habits and rational choicesin ecological behavior: The case of recycling and use of public transportation.Journal of Environmental Psychology, 28, 51e62.

Chen, F.-Y. (2013). The intention and determining factors for airline passengers’participation in carbon offset schemes. Journal of Air Transport Management, 29,17e22.

Dekoster, J., & Schollaert, U. (1999). Cycling: The way ahead for towns and cities.Belgium: European Commission DG IX, Environment, Nuclear, Safety and CivilProtection.

P. Passafaro et al. / Journal of Environmental Psychology 38 (2014) 76e83 83

European Commission. (2012). Directorate general for mobility and transport. In To-ward low Carbon transport in Europe. European Union.

Gatersleben, B., & Appleton, K. (2007). Contemplating cycling to work: Attitudesand perceptions in different stages of change. Transportation Research Part A:Policy and Practice, 41, 302e312.

Gatersleben, B., & Haddad, H. (2010). Who is the typical bicyclist? TransportationResearch Part F, 13, 41e48.

Gatersleben, B., & Uzzel, D. (2007). Affective appraisal of the daily commute:Comparing perceptions of drivers, cyclists, walkers and users of public trans-port. Environment and Behavior, 39, 416e431.

Geller, E. S. (2002). The challenge of increasing pro-environmental behaviour. InR. B. Bechtel, & A. Churchman (Eds.), Handbook of environmental psychology (pp.28e54). New York: Wiley.

Haskell, W. L., Lee, I. M., Pate, R. R., Powell, K. E., Blair, S. N., Franklin, B. A., et al.(2007). Physical activity and public health: Updated recommendation for adultsfrom the American college of sports medicine and the American heart associ-ation. Medicine and Science in Sports and Exercise, 39, 1423e1434.

Hopkinson, P., & Wardman, M. (1996). Evaluating the demand for new cycle facil-ities. Transport Policy, 3, 241e249.

ISFORT. (2008e2011). Audimob. Osservatorio sui comportamenti di mobilità degliitaliani [Audimob. Observatory on the mobility behaviors of Italians]. In Ladomanda di mobilità degli Italiani. http://www.isfort.it/sito/statistiche/Congiunturali/Annuali/RA_2011.pdf. Retrieved June 18th, 2013.

Kim, M.-J., Lee, M. J., Lee, C. K., & Song, H. J. (2012). Does gender affect Koreantourist’s overseas travel? Applying the Model of Goal-Directed behavior. AsianPacific Journal of Tourism Research, 17, 509e533.

Leone, L., Perugini, M., & Ercolani, A. P. (2004). Studying, practicing, and mastering:A test of the Model of Goal-Directed behavior (MGB) in the software learningdomain. Journal of Applied Social Psychology, 3, 1945e1973.

Matheson, G. O., Klügl, M., Dvorak, J., Engebretsen, L., Meeuwisse, W. H.,Schwellnus, M., et al. (2011). Responsibility of sport and exercise medicine inpreventing and managing chronic disease: Applying our knowledge and skill isoverdue. British Journal of Sports Medicine, 45, 1272e1282.

McCombs, M. E., & Shaw, D. L. (1972). The Agenda-setting function of the Press.Public Opinion Quarterly, 36, 176e187.

Oja, P., Vuori, I., & Paronen, O. (1998). Daily walking and cycling to work: Theirutility as health-enhancing physical activity. Patient Education and Counseling,33, s87es94.

Perugini, M., & Bagozzi, R. P. (2001). The role of desires and anticipated emotions ingoal-directed behaviors: Broadening and deepening the theory of plannedbehavior. British Journal of Social Psychology, 40, 79e98.

Perugini, M., & Bagozzi, R. P. (2004). The distinction between desires and intentions.European Journal of Social Psychology, 34, 69e84.

Prochanska, J. O. (1994). Strong and weak principles for progressing from pre-contemplation to action on the basis of twelve problem behaviors. HealthPsychology, 13, 47e51.

Prochanska, J. O., & Di Clemente, C. C. (1984). The transtheoretical approach: Crossingtraditional boundaries of change. Homewood, IL: Dow Jones/Irwin.

Rimano, A., Piccini, M. P., Metastasio, R., Chiarolanza, C., & Passafaro, P. (2012).Determinants of the use of bike in the city: Between practical motivations andsocial cultural values. In Poster presented at the planet under pressure e Newknowledge towards solutions e Conference, London, March 26e29th, 2012.

Rimano, A., Piccini, M. P., Passafaro, P., Metastasio, R., Chiarolanza, C., Boison, A.,et al. (2014). The bicycle and the dream of a sustainable city: An explorativecomparison of the image of bicycle in the mass-media and the common public.submitted for publication Urban Studies.

Russell, J. A., & Snodgrass, J. (1987). Emotions and the environment. In D. Stokols, &J. Altman (Eds.), Handbook of environmental psychology (Vol. 1); (pp. 245e280).New York: John Wiley.

Shaw, E. F. (1979). Agenda setting andmass communication theory.Gazette, 2, 96e105.Song, H. J., Lee, C.-K., Kang, S. K., & Boo, S.-J. (2012). The effect of environmentally

friendly perceptions on festival visitors’ decision-making process using anextended model of goal-directed behavior. Tourism Management, 33, 1417e1428.

Song, H. J., Lee, C.-K., Norman, W. C., & Han, H. (2012). The role of responsiblegambling strategy in forming behavioral intention: An application of a Model ofGoal-Directed behavior. Journal of Travel Research, 51, 512e523.

Taylor, S. A. (2007). The addition of anticipated regret to attitudinally based, goal-directed models of information search behaviours under conditions of uncer-tainty and risk. British Journal of Social Psychology, 46, 739e768.

Taylor, S. D., Bagozzi, R. P., & Gaither, C. A. (2005). Decision making and effort in theself-regulation of hypertension: Two competing theories. British Journal ofHealth Psychology, 10, 505e530.

Taylor, S. A., Ishida, C., & Wallace, D. W. (2009). Intention to engage in digitalpiracy: A conceptual model and empirical test. Journal of Service Research, 11,246e262.

The Gallup Organization. (2007). Attitudes on issues related to EU transport policies.In Flash Eurobarometer, 206b. Retrieved June 18th, 2013 http://ec.europa.eu/public_opinion/flash/fl_206b_en.pdf.

Vandenbergh, M. P., Stern, P. C., Gardner, G. T., Dietz, T., & Gilligan, M. (2010).Implementing the behavioral wedge: Designing and adopting effective car-bon emission reduction programs. Environmental Law Reporter, 40, 10547e10554.

Wilson, R. S., & Arvai, J. L. (2006). When less is more: How affect influences pref-erences when comparing low and high-risk options. Journal of Risk Research,9(2), 165e178.