to reduce and not to reduce resource consumption? that is two questions

11

Click here to load reader

Upload: juliette-richetin

Post on 05-Sep-2016

219 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: To reduce and not to reduce resource consumption? That is two questions

at SciVerse ScienceDirect

Journal of Environmental Psychology 32 (2012) 112e122

Contents lists available

Journal of Environmental Psychology

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

To reduce and not to reduce resource consumption? That is two questions

Juliette Richetin a,*, Marco Perugini a, Mark Conner b, Iqbal Adjali c, Robert Hurling d, Abhijit Sengupta d,Danica Greetham e

a Faculty of Psychology, University of Milan-Bicocca, Piazza dell’Ateneo Nuovo 1 (U6), 20126 Milan, Italyb Institute of Psychological Sciences, University of Leeds, Leeds LS2 9JT, UKcModelling & Analytics Lloyds Banking Group, 5th Floor, 48 Chiswell Street, London EC1Y 4XX, UKdUnilever Discover, Colworth Science Park Sharnbrook, Bedford MK44 1LQ, UKeDepartment of Mathematics and Statistics, University of Reading, Reading RG6 6AX, UK

a r t i c l e i n f o

Article history:Available online 8 January 2012

Keywords:Attitudinal modelsDoingNot doingGoalsPro-environmental behavioural prediction

* Corresponding author. Tel.: þ39 026448 3863; faE-mail addresses: [email protected]

lloydsbanking.com (I. Adjali).

0272-4944/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.jenvp.2012.01.003

a b s t r a c t

Recent research shows that because they rely on separate goals, cognitions about not performinga behaviour are not simple opposites of cognitions about performing the same behaviour. Using thisperspective, two studies (N ¼ 758 & N ¼ 104) examined the psycho-social determinants ofreduction in resource consumption. Results showed that goals associated with reducing versus notreducing resource consumption were not simple opposites (Study 1). Additionally, the discriminantvalidity of the Theory of Planned Behaviour constructs associated with reducing versus notreducing resource consumption was demonstrated (Study 1 & 2). Moreover, results revealed theincremental validity of both Intentions (to reduce and to not reduce resource consumption) forpredicting a series of behaviours (Study 1 & 2). Finally, results indicated a mediation role for theimportance of ecological dimensions on the effect of both Intentions on a mock TV choice anda mediation role for the importance of non ecological dimensions on the effect of Intention of notreducing on the same TV choice. Discussion is organized around the consequences, at both theo-retical and applied levels, of considering separate motivational systems for reducing and notreducing resource consumption.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

In the general domain of attitudinal models of behaviouralprediction such as the Theory of Planned Behaviour (TPB, Ajzen &Madden, 1986; Ajzen, 1991, 2002a), very few studies haveconsidered constructs related to both performing and not per-forming an action. Although some research has examined choicesbetween two or more alternatives (e.g., Jaccard, 1981), sinceFishbein’s (1980) original study in which he considered both theintention to quit smoking and the intention to smoke for pre-dicting smoking quit attempts, little research has pursued this lineof investigation. More specific to environmental behaviour, someempirical work has been devoted to showing the simultaneousvalidity of constructs somehow related to performing and notperforming an action (e.g., Letirand & Delhomme, 2005) or toconsidering constructs related to performing an ecological action

x: þ39 026448 3788.(J. Richetin), iqbal.adjali@

All rights reserved.

and to performing a non ecological alternative. However, a mainissue remains on choosing the best alternative in order to betterpredict behaviour. We argue that not performing a pro-environmental behaviour is the simplest alternative to perform-ing that behaviour. As the main theoretical reason for consideringboth alternatives when predicting behaviour, Richetin, Conner,and Perugini (2011) argue that cognitions concerning not doingare not the simple opposite of those for doing because performingor not performing an action for an individual can be based ondifferent goals and involve different self-regulation strategies.Considering that those goals could be separate rather than oppo-sites, they provide psychological distinctiveness to performing andnot performing an action. Applying this argument in the TPBframework (Ajzen, 1991, 2002a; Ajzen & Madden, 1986) theyempirically demonstrated in different domains that individuals’action can be influenced by both intentions to act and intentionsnot to act. Consistent with previous research that has shown thatan ecological behaviour and a non ecological alternative are linkedto different motivations (e., de Boer, Boersema, & Aiking, 2009) butincluding all alternatives, the present contribution explores

Page 2: To reduce and not to reduce resource consumption? That is two questions

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122 113

whether pro-environmental intentions and behaviours can also besimultaneously predicted by cognitions related to performing andnot performing. We first briefly review results from studies thathave used the TPB to investigate psycho-social determinants ofpro-environmental behaviours, as well as studies that includeadditional constructs to enhance TPB predictive power. We thendiscuss the reasons why it would be important to consider bothcognitions around reducing and not reducing resource consump-tion and whether it would further improve TPB predictions, andpresent two empirical studies to illustrate the benefit of thisapproach for pro-environmental behaviours.

1.1. Using and expanding the Theory of Planned Behaviour topredict pro-environmental behaviour

Since their development, the Theory of Reasoned Action (TRA)(Ajzen& Fishbein,1969; Fishbein&Ajzen,1975) and its extension, theTPB (Ajzen, 1991, 2002a; Ajzen & Madden, 1986) have been exten-sively applied to many domains in order to better understand somepsycho-social determinants of a broad range of behaviours. Since theattention towards pro-environmental behaviours has increasedsteadily in the last years as a consequence of the greater awareness oftheir relevance, the TPB has also been used to determine some keyfactors leading to pro-environmental intentions and behaviours. Infact, in contrast to many “anti-environmental” behaviours that arebased on habits and therefore mostly driven by automatic processesdifficult to disrupt (e.g., Kolmuss & Agyeman, 2002), pro-environmental ones are often not regular enough to becomehabitual and are therefore more likely to be guided by a deliberateprocess in which different alternatives are considered. As a conse-quence, theymay be best predicted by consciously formed intentionstowards the behaviour (Aarts, Verplanken, & vanKnippenberg,1998).Hence, the TRA and the TPB have been successfully applied to variousdifferent pro-environmental behaviours such as environmentalactivism (e.g., Fielding,McDonald, & Louis, 2008), water conservation(e.g., Corral-Verdugo & Frías-Armenta, 2006), sustainable agriculturepractices (e.g., Fielding, Terry, Masser, & Hogg, 2008), recycling (e.g.,Cheung, Chan, & Wong,1999; Goldenhar & Connell, 1993), compost-ing (e.g., Taylor & Todd, 1997), energy use (e.g., Harland, Staats, &Wilke, 1999), and purchase of sustainable food products (e.g.,Vermeir & Verbeke, 2006). In their meta-analysis of studies applyingthe TPB in this area, Bamberg and Möser (2007) report on averagethat intentions explain 27% variance in self-reported pro-environ-mental behaviours.

However, the extent to which the TPB explains intentions andbehaviours has often been questioned (e.g., Conner & Armitage,1998) and this has also been in the case in the domain of pro-environmental behaviour (e.g., Stets & Biga, 2003). In fact,a considerable amount of research has demonstrated that theinclusion of personal, descriptive or moral norms (e.g., Rivis &Sheeran, 2004), personality traits (e.g., Courneya, Bobick, &Schinke, 1999), anticipated regret (e.g., Sheeran & Orbell, 1999),past behaviour (Bagozzi, 1981), behavioural and goal desire (e.g.,Perugini & Bagozzi, 2001; Perugini & Conner, 2000) improves theprediction of intention or behaviour. For example, in relation to theprediction of pro-environmental intentions and behaviours, studieshave shown that the inclusion of self-identity (Sparks & Shepherd,1992), personal norms (e.g., Harland et al., 1999), guilt (e.g., Kaiser,2006), or group membership (Fielding, McDonald, & Louis, 2008)improved the prediction mainly of intention but also of behaviour.In particular, Fielding et al. (2008) demonstrated that environ-mental group membership and self-identity predicted intention toengage in environmental activism over and above the TPBconstructs; Bamberg and Möser (2007) showed that the combi-nation of the Norm Activation Model (Schwartz, 1977) with the TPB

lead to an average of 52% variance explained in Intentions. More-over, research has considered further improvements of TPB throughthe integration of goal-oriented aspects of action (Perugini &Bagozzi, 2001, 2004a, 2004b; Perugini & Conner, 2000).

1.2. Separate goal systems for acting and not acting

Previous research shows the important role that self-regulationmechanisms and goals in particular play in pro-environmentalbehaviours (e.g., Aarts & Dijksterhuis; 2000; Pelletier, Tuson,Green-Demers, Noëls, & Beaton, 1998). Few research have alsoshowed that different motivations are attached to performinga pro-environmental action and to performing a non pro-environmental alternative (e.g., de Boer et al., 2009). With this inmind we hypothesized that goals also play a role in performing allnon pro-environmental alternatives that are included in not per-forming pro-environmental behaviours. Empirical and theoreticalwork has posited the existence of two different motivationalsystems such as approach and avoidance (e.g., Cacioppo & Berntson,1994) or of two different motivational orientations such aspromotion and prevention (e.g., Higgins, 2000). However in eachtheory and for both orientations, action and inaction are consideredconceptually orthogonal dimensions. For example, avoidance/prevention can result in action (e.g., running away from a negativestimulus) or inaction (e.g., not moving towards the negative stim-ulus/freezing) and approach/promotion could also result in bothoptions (e.g., going towards a positive stimulus versus not runningaway from a positive stimulus). From the perspective of the actionversus inaction in the environmental domain, one could not reduceone’s own resource consumption because of a prevention focus (toavoid disrupting a comfortable life) or because of a promotion focus(to get an even more comfortable life); one could reduce one’s ownresource use because of a prevention focus (to not destroy theplanet) or because of a promotion focus (to preserve wildlife).Therefore these theories do not allow us to disentangle theprocesses underlying doing and not doing. Instead, the best theo-retical background for our claim refers to the theory of goal system(Kruglanski et al., 2002). This theory proposes the existence ofnetworks in which goals are associated to means that serve theirachievement. Goals can be connected to other goals and meanswith other means with facilitative links, preferably between goalsand their corresponding means, or inhibitory links, betweencompeting goals or competing means. Following this principle,Richetin et al. (2011) argue that the intention to perform an actionand the intention to not perform the same action belong to tworelatively separate goal systems, providing separate rather thanopposite reasons for each alternative. As a consequence, pathwaysto action and to inaction should not be judged asmutually exclusiveand should be considered together in order to better understandand predict behaviour.

1.3. Not doing is not the opposite of doing: application to pro-environmental behaviour

Applying this logic to the environmental domain, we argue thatwhile performing a pro-environmental action and not performingthe same action can be considered as logical opposites, they arepsychologically different because each is associated with separate(and not necessarily opposite) goals. For instance, whereas per-forming a pro-environmental behaviour might be triggered by thegoal to preserve the planet, it seems less likely that those notdoing a pro-environmental behaviour have a goal of destroyingthe planet. A more likely goal might rather be to maintaina convenient way of life. Although the goals underlying doing andnot doing can, in some cases, be perceived as conflicting they are

Page 3: To reduce and not to reduce resource consumption? That is two questions

1 There were only two main effects of Order of questionnaire on the TPBconstructs whereas there was no main effect on behavioural measures (allps > .238). Norms related to reducing and to not reducing resource consumptionwere higher when the measures related to not reducing were presented first (M¼ 4.52, SD ¼ .98 and M ¼ 3.65, SD ¼ .93, respectively) than when the measures relatedto reducing were presented first (M ¼ 4.35, SD ¼ .96 and M ¼ 3.40, SD ¼ 1.00,respectively), t(756) ¼ �2.37, p ¼ .018 and t(756) ¼ �3.57, p < .001, respectively.There were only one moderation effect of Order of questionnaire on the validity ofthe Intention to reduce resource consumption in predicting SRB (R2 ¼ 4.3%). Orderwas not a significant predictor (b ¼ .05, p ¼ .305) whereas the Intention of notreducing (b ¼ �.28, p < .001) and the interaction term was (b ¼ �.14, p ¼ .039). Allother interactions were not significant (all ps > .405).

2 Participants were missing at random for two main reasons. The main reasonwas that for some high-schools the data collection for the second session was notpossible for logistic reasons. In addition, some students were also randomly absentfor the second session. However, we tested whether the students who took part tothe two sessions differed from the students who took part to only the first sessionin terms of means in TPB constructs related to reducing and not reducing resourceconsumption. No significant differences were detected across all measures.

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122114

not necessarily opposites. In fact previous research has demon-strated that the motives underlying why people do things to helpprotect the environment are different from the motives underlyingwhy people do not do things to help protect the environment (e.g.,Pelletier, Dion, Tuson, & Green-Demers, 1999). Thus, the reasoningand the influences underlying the decision to act or not act pro-environmentally can be very different as can the pathways bywhich such decision making processes influence action. Thiswould suggest that both intentions to act and intention to not actpro-environmentally can co-exist and that behaviour can be pre-dicted by both. From the perspective of the TPB this reasoningwould suggest that behaviour can be better predicted by consid-ering all components of the TPB (e.g., intentions, attitudes, norms,and perceived behavioural control) in relation to both performingand not performing the behaviour. From a statistical point of view,this would imply that the constructs related to the two processes(i.e., doing versus not doing) can show evidence of incrementalvalidity; that is they can uniquely contribute to the prediction ofintention and/or behaviour.

1.4. Aims of this contribution

In line with previous work showing that performing anecological behaviour and performing a non ecological alternativeare linked to different motivations, we assume that theconstructs related to performing and to not performing anecological behaviour are not opposites because each option relieson separate goals. We examine the cognitions for reducing andnot reducing resource consumption in the context of the TPB intwo studies set within a high-school (Study 1) and a university(Study 2). We chose reducing and not reducing resourceconsumption as the target behaviour for several reasons. Thisphrasing includes a large range of behaviours from reducingspeed when driving to not letting the water run while brushingteeth. This allowed us to examine performance of a range ofdifferent behaviours and also to compute a behaviouralcomposite with increased reliability. These behaviours are alsorelatively low-cost compared to other pro-environmentalbehaviour such as installing a solar panel on the roof. Thiscontribution has four main aims. First, we examine the goalsunderlying both reducing and not reducing resource consump-tion, hypothesizing that they would be different rather thanopposite (H1). Second, we test if TPB constructs related to per-forming and to not performing a behaviour have discriminantvalidity (H2). Third, we hypothesize that one’s cognitions aboutdoing something and about not doing something can bothcontribute to the prediction of the respective intentions (H3).Fourth, we argue that one’s intentions of doing something and ofnot doing something can both predict behaviour (H4).

2. Study 1

2.1. Method

2.1.1. Participants and procedureEight hundred and fifteen high-school students (439 women,

376 men, Age range: 16e22; M Age ¼ 18.01, SD ¼ .85) wererecruited from six different high-schools of Italy to take part ina two-session study. Participants were tested in groups of atmaximum 25. In a first session, they were asked to completea series of questionnaires on goals related to reducing and notreducing resource consumption, along with measures of the TPB(Ajzen, 1991) constructs (i.e., Attitude, Norms, Perceived Behav-ioural Control, and Intentions) for both options. Half of theparticipants completed the measures related to doing (versus not

doing) first whereas the other half completed the measuresrelated to not doing first.1 The study included additional measuresnot considered here. In a second session, 5e9 days later,participants were asked to complete the behavioural measures.In total, 441 completed the second session.2 From the firstsession, the data from 57 participants were discarded. Addition-ally, the data from 32 participants in the second session werediscarded, leaving a total of 758 for the first session and a total of409 for the second session. All excluded participants did notcomplete all the measures.

2.1.2. MeasuresGiven the similarity between the measures concerning reducing

resource consumption and not reducing resource consumption,only the measures applied to reducing resource consumption willbe presented.

2.1.2.1. Goals. Participants wrote down the main goal related toreducing resource consumption and the main goal related to notreducing resource consumption.

2.1.2.2. Attitude. Participants were presented with the stem“Reducing my resource consumption for me is” followed by sevenbipolar scales (i.e., uselesseuseful, harmfulebeneficial,stupideintelligent; not enjoyableeenjoyable, unpleasantepleasant,boringeamusing; negativeepositive) on a 7-step answer scaleranging from 1 to 7.

2.1.2.3. Norms. Norms were assessed by six items on 7-point scalesranging from 1 to 7: 1) “People who are important to me think Ishould reduce my resource consumption” (extremelyunlikelyeextremely likely); 2) “People who are important to mewould approve my reducing of my resource consumption” (not atall-very much); 3) “People who are important to me would behappy if I would reduce my resource consumption” (falseetrue); 4)“Most people who are important to me reduce their resourceconsumption” (falseetrue); 5) “The extent with which most peoplewho are important to me reduce their resource consumption is”(lowehigh); 6) “How many people do you know who reduce theirresource consumption?” (noneemany).

2.1.2.4. Perceived Behavioural Control. Perceived BehaviouralControlwas assessedwith four items on 7-point scales ranging from1 to 7: 1) “I have control over reducing my resource consumption”(completely disagreeecompletely agree); 2) “Howmuch control doyou have over reducing your resource consumption?” (nocontrolecomplete control); 3) “I am confident that I can reduce my

Page 4: To reduce and not to reduce resource consumption? That is two questions

Table 1Goals associated with reducing versus not reducing resource consumption (Study 1;N ¼ 758).

Reducing resource consumption Not reducing resource consumption

Save, protect and respect nature240 (31.66%)

Maintain actual lifestyle, easy life413 (54.49%)

Better future, well-being of futuregenerations 229 (30.21%)

Seek for immediate pleasure92 (12.14%)

Improve quality of life 160 (21.10%) Trust in resources 78 (10.29%)Resources are limited 51 (6.73%) Lack of time, time for doing other

things 70 (9.23%)General well-being (man, nature)

29 (3.83%)Lack of interest 43 (5.67%)

Avoid environmental disasters14 (1.85%)

Personal well-being and interest27 (3.56%)

Better balance man/nature14 (1.85%)

Essential need for resources14 (1.85%)

Excessive consumption11 (1.45%)

Laziness 10 (1.32%)

Save money 10 (1.32%) No other alternatives 6 (.79%)Trust in human kind for findingother resources 5 (.66%)

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122 115

resource consumption” (completely disagreeecompletely agree);4) “If I wanted to, it would be easy for me to reduce my re sourceconsumption?” (falseetrue).

2.1.2.5. Intention. Intention was assessed by three items on 7-pointscales ranging from 1 to 7: 1) “I plan to reduce my resourceconsumption (completelydisagreeecompletelyagree); 2) “I intend toreduce my resource consumption (completely disagreeecompletelyagree); 3) “How likely is it that you will reduce your resourceconsumption” (very unlikelyevery likely).

2.1.2.6. Behavioural measures. Behaviour was measured in twoways. First, there was a hypothetical behavioural measure: Partic-ipants were asked to read sequentially two hypothetical scenariosin which they are the main character (Scenario). Each scenariodescribed a situation in which two actions were possible, one morein favour of resource consumption with a down side and the othernot (i.e., “You are in your bathroom cleaning your teeth. You knowthat by closing the tap you would consume less water but you findit very inconvenient to have to close it and then to open it again.”;“You are in class, you have paper you need to throwaway. You knowthat recycling paper reduces deforestation but the recycling box isin the corridor while in the classroom there is a rubbish bin”).3

After reading each scenario, participants rated the likelihood ofacting in the direction of reducing resource consumption on 7-point scales from 1 (very unlikely) to 7 (very likely). Second,participants indicated the frequency with which they performeda series of nine behaviours related to reducing resourceconsumption (e.g., switching off lights when leaving the room, notletting thewater runwhile washing hands, unplugging a cell phonecharger when charging is completed, reducing speed while drivinga car or motorcycle) during the previous week (SRB) on 6-pointscales from 0 (never) to 5 (very frequent). The list of behaviourshas been developed with special attention to the age of theparticipants considering only behaviours that were highly likely tobe performed (e.g., unplugging a cell phone charger when chargingis completed) and excluding those which were more likely to beperformed by older adults in the household (e.g., waiting for a fullload before starting the dishwasher).

2.2. Results

2.2.1. Goals associated with reducing versus not reducing resourceconsumption

As shown in Table 1, the most frequently reported main goalassociated with reducing resource consumption was environ-mentally related (i.e., to protect nature) whereas the mostfrequently reported main goal associated with not reducingresource consumption was personal (i.e., to maintain one’s easyway of life). The two most frequently reported goals underlyingreducing resource consumption (61.87%) were other- or generalinterest-oriented whereas the two most frequently reported goalsunderlying not reducing (66.63%) were self-oriented. Theseresults confirm that the goals associated with not reducingresource consumption are not the opposites of the ones associ-ated with reducing resource consumption (i.e., ‘to protect theplanet’ was not necessarily mirrored by ‘to not protect theplanet’), supporting H1.

3 We have included the mention of deforestation in the scenario about recyclingpaper because it was possible that not all high-school students may be aware thatrecycling reduces deforestation. This allowed us to place all students at the samelevel of knowledge thus reducing the influence of other inter-individual differencesdifferent from the one we wanted to investigate.

2.2.2. Construct and discriminant validityTo examine the construct and discriminant validity in addition

to the reliabilities and the correlations between constructs, we ranconfirmatory analyses for each construct separately with modelsthat always included two factors (i.e., one factor for reducing andone for not reducing): One model (M1) in which the correlationbetween the two factors is fixed to e 1, implying that the twofactors fully overlap and so was conceptually equivalent to a modelwith one single general factor. The other model (M2) in which thecorrelation between the two factors was left to be estimated. Thisstrategy was chosen because these two models are nested andtherefore their difference in fit can be formally tested by consid-ering their difference in c2. If the best solution appears to be M1, itwould imply that the measures related to reducing and notreducing resource consumption assess the same construct andtherefore the constructs related to doing and not doing would notachieve construct and discriminant validity. On the contrary, if M2would appear to be the best solution, it would imply that themeasures related to reducing and not reducing resourceconsumption do not assess the same construct and therefore theconstructs related to doing and not doing would achieve satisfac-tory construct and discriminant validity.

As shown in Table 2, the correlations between the constructswere all significant for reducing resource consumption and for notreducing resource consumption. More central to our concerns, thecorrelations between reducing and not reducing resourceconsumption (underlined values) were all negative, significant andranged from�.44 (Intentions) to�.01 (PBCs). However, correlationswere sufficiently different from�1 to reject the conclusion that theconstructs were simply opposites. Indeed the degree of correlationbetween factors indicated at most a moderate degree of overlap.The results therefore indicate a substantial independence in theevaluation of reducing versus not reducing resource consumption.

Table 3 reports the more formal test of discriminant validity,that is the c2 difference between models (c2 M1 � c2 M2) with itsassociated significance level as well as the correlation between thetwo latent factors of reducing and not reducing resourceconsumption. The results for Study 1 clearly indicated that thesolution with two correlated but distinct factors organized aroundDoing and Not doing (M1) is the best solution. These resultsconverge to the conclusion that all constructs (Attitude, Norms,Intention) achieve construct and discriminant validity (4 from�.55to .11) for reducing versus not reducing resource consumption,providing strong support to H2.

Page 5: To reduce and not to reduce resource consumption? That is two questions

Table 2Correlations among constructs (Study 1).

1 2 3 4 5 6 7 8 9 10

1. Attitudereducing (.88)2. Normsreducing .37*** (.81)3. Perceived Behavioural Controlreducing .39*** .40*** (.82)4. Intentionreducing .54*** .46*** .66*** (.91)5. Attitudenot reducing �.33*** �.17** �.16** �.24*** (.90)6. Normsnot reducing �.22*** �.38*** �.18*** �.20*** .23*** (.75)7. Perceived Behavioural Controlnot reducing �.15** �.07 �.01 �.13** .25*** .25*** (.63)8. Intentionnot reducing �.43** �.28*** �.29*** �.44*** .42*** .41*** .25*** (.87)9. Scenario .26** �.14** .26*** .32*** �.23** �.15** �.10* �.27*** (.28a)10. Self-Reported Behaviour .20** .21*** .33*** .37*** �.14* �.17** �.13** �.26*** .39*** (.68)

Note. Reliabilities are indicated into brackets when applicable.The correlations between reducing and not reducing resource consumption constructs are indicated with underlining.*p < .05. **p < .01. N ¼ 758; for Behaviours, N ¼ 409.

a Only two items r, p < .001.

Table 4Incremental validity of constructs for the prediction of intention of reducingresource consumption.

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122116

2.2.3. Incremental validity2.2.3.1. Intentions. The incremental validity for predicting Inten-tion to reduce resource consumption was tested with a series ofregression analyses in which as a first step Attitudereducing,Normsreducing and PBCreducing were entered, and as a second stepAttitudenot reducing, Normsnot reducing and PBCnot reducing were added.The incremental validity for predicting Intention to not reduceresource consumption was tested by entering these sameconstructs but in a reverse order, namely variables related to notreducing resource consumption first followed by the ones related toreducing resource consumption.

When considering the TPB constructs for the prediction of theIntention of reducing resource consumption (Table 4, Study 1), bothAttitudes show incremental validities and Norms and PBC relatedto reducing were significant predictors. When considering theprediction of the Intention of not reducing resource consumption(Table 5, Study 1), again both Attitudes showed incremental val-idity. Moreover, both PBCs showed incremental validities andNorms related to reducing was a significant predictor. Therefore forpredicting both Intentions to reduce and not to reduce resourceconsumption, both Attitudes play a significant role, supporting H3.

2.2.3.2. Behaviours. To test the incremental validity of doing andnot doing constructs in predicting each of the two behaviours (i.e.,Scenario; SRB), first a series of regression analyses were run inwhich both Intentions were entered to examine the unique vari-ance explained by each. A second series of analyses examined theadditional role of PBCs. For each behaviour, in the first step Inten-tionreducing and PBCreducing were entered, as a second step Inten-tionnot reducing and PBCnot reducing were added (Table 6).

When both Intentions were entered in the regression to predictthe hypothetical behaviour (Scenario, R2 ¼ .13), both Intentions ofreducing and not reducing were significant (b ¼ .26, p < .001 andb ¼ �.17, p ¼ .001, respectively). When the PBCs (Step 2) wereentered into the equation, both Intentions still significantly pre-dicted, although the PBCsdidnot. For thepredictionof SRB (R2¼ .17),

Table 3Chi-square difference between Model 1 and Model 2 and correlation between thelatent factors.

Study 1 Study 2

c2(1) p F c2(1) p 4

Attitude 3617.14 < .001 �.35 82.93 < .001 �.66Norms 603.20 < .001 �.55 79.08 < .001 �.77Perceived

Behavioural Control490.64 < .001 .11 54.89 < .001 .19

Intention 960.98 < .001 �.50 100.88 <.001 �.57

both Intentions were significant (b ¼ .34, p < .001 and b ¼ �.13,p ¼ .012, respectively). When considering PBCs (Step 2), bothIntentions andPBC related to reducing resourcewere still significant.To sum up, both Intentions of reducing and not reducing resourceconsumption are important in the emergence of hypothetical andself-reported pro-environmental behaviours, supporting H4.

2.3. Discussion

Constructs related to performing and to not performing a behav-iour are associatedwith two different sets of reasons or goal systemsand hence should be considered not as opposite but as psychologi-cally distinct. Therefore, in the context of the TPB applied to pro-environmental ecological behaviour, constructs related to perform-ing and to not performing should be means to different but notnecessarily opposite goals, should show construct and discriminantvalidities and can simultaneously contribute to both Intentions ofreducing and of not reducing resource consumption and to behav-iours. The results of thisfirst studymainly supported thehypotheses.

First, not reducing resource consumption and reducing resourceconsumption are associated with different but not opposite sets ofgoals, reducing resource consumption mainly triggered by reasonslinked to the protection of the planet whereas not reducingresource consumption was mainly triggered by goals linked tomaintaining a comfortable life. Previous research has shown andargued that goals associated to pro-environmental behaviour havemainly to do with concern of others and with self-interest (forconcern of other, Schwartz, 1977; for self-interest, Ajzen, 1991; fora discussion see Bamberg & Möser, 2007). Our results, althoughshowing that self-interest is also a reason for reducing resourceconsumption (i.e., saving money), largely demonstrated that the

Study 1 Study 2

Step 1 Step 2 Step 1 Step 2 Step 3

R2 .51 .53 .44 .48 .63Attitudereducing .29 .25 .40 .22 .16Normsreducing .14 .14 .09 .17 .15Perceived Behavioural

Controlreducing.46 .47 .35 .32 .07

Attitudenot reducing L.07 L.29 L.25Normsnot reducing .04 .08 .11Perceived Behavioural

Controlnot reducing

�.09 �.01 �.04

Past Behaviourreducing .51

Note. Significant factors are indicated in bold.

Page 6: To reduce and not to reduce resource consumption? That is two questions

Table 5Incremental validity of constructs for the prediction of intention of not reducingresource consumption.

Study 1 Study 2

Step 1 Step 2 Step 1 Step 2 Step 3

R2 .31 .39 .35 .38 .39Attitudenot reducing .30 .23 .46 .30 .28Normsnot reducing .32 .27 .08 .08 .07Perceived Behavioural

Controlnot reducing

.15 .14 .18 .25 .24

Attitudereducing L.23 L.21 L.18Normsreducing .04 .06 .02Perceived Behavioural

ControlreducingL.13 .13 .14

Past Behaviourreducing L.18

Note. Significant factors are indicated in bold.

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122 117

main goals associated with reducing resource consumption areother-oriented. Second, all TPB constructs related to reducing andnot reducing resource consumption achieve good construct anddiscriminant validities with correlations between cognitions aboutreducing and not reducing ranging from weak to moderate. Third,both Attitudes simultaneously predicted both Intentions to reduceand not to reduce resource consumption. Finally, the resultsshowed the incremental validity of both Intentions for predictinghypothetical and self-reported pro-environmental behaviours.

Taken together, these results support the empirical and theo-retical value of considering constructs related to both doing andnot doing a behaviour (Richetin et al., 2011) and extend theirrelevance to the pro-environmental domain. In this first study, inaddition to self-reported pro-environmental behaviours, weexamined hypothetical behaviours in which two actions werepossible, one more in favour of resource consumption with a downside and the other not. However with this procedure it is possiblethat the desirability of one behavioural option compared to theother might have been too salient and therefore could have biasedthe answers. The direction of the question itself (i.e., participantsrated the likelihood of acting in the direction of reducing resourceconsumption) might also have encouraged responses towards thereducing resource consumption option. It would be interesting totest the incremental validity of both Intentions in predictinga behavioural measure in which elements of choice are presentedless dichotomously. Furthermore, the role of Past Behaviour hasnot been investigated. Past Behaviour might partially account forthe impact of constructs related to not reducing resourceconsumption and hence could diminish or eliminate their incre-mental validity in predicting Intentions and behavioural criteria.Finally, it would be interesting to understand how each Intentionaffects behavioural measures. The second study aimed to addressthese issues through replicating the first study’s results with theinclusion of a measure of past behaviour and additional behav-ioural criteria.

Table 6Incremental validity of constructs for the prediction of behaviours (Study 1).

Scenario Self-ReportedBehaviour

Step 1 Step 2 Step 1 Step 2

R2 .11 .13 .16 .17Intentionreducing .27 .20 .31 .26Perceived Behavioural Controlreducing .08 .08 .12 .13Intentionnot reducing L.16 L.11Perceived Behavioural Controlnot reducing .03 �.07

Note. Significant factors are indicated bold.N ¼ 409.

3. Study 2

3.1. Method

3.1.1. Participants and procedureA hundred and four (76 women, 28 men, M Age ¼ 22.65,

SD ¼ 2.08) students from an Italian university were recruited toparticipate in a one-session study. They were asked to completea series of questionnaires with the same measures used in Study 1except for those related to goals for reducing versus not reducingresource consumption. Therefore participants completed themeasures of the TPB (Ajzen, 1991) constructs (i.e., Attitude, Norms,Perceived Behavioural Control, and Intentions) and Past Behaviourrelated to reducing versus not reducing one’s own resourceconsumption. Half of the participants completed the measuresrelated to doing first whereas the other half completed themeasures related to not doing first.4 Then, participants completedbehavioural measures. One month later, participants were con-tacted to complete an additional online behavioural measure(N ¼ 73).5 The study included additional measures not consideredhere.

3.1.2. Measures3.1.2.1. Attitude, Norms, Perceived Behavioural Control, Inten-tions. For each construct, participants completed the samemeasures previously described in Study 1.

3.1.2.2. Past Behaviour. Past Behaviour (PB) was assessed by twoitems: “I usually reduce my resource consumption”, on a 7-pointscale from 1 (strongly disagree) to 7 (strongly agree) and “In thelast month, how frequently did you do reduce your resourceconsumption?”, on a 5-point scale from 1 (never) to 5 (very often).Factor scores from a Principal Component Analysis (with one factorexplaining 90.34% of variance) were used in the analyses.

3.1.2.3. Behavioural measures. The first behavioural criterion wasthe multi-attribute consumer choice task, developed fromVerplanken and Holland’s (2002) procedure and consisted ofchoosing between 20 television sets. The television sets weredescribed by seven attributes (e.g., screen dimensions) of whichonly one was related to environmental aspects. Each attribute wasbriefly detailed. Environmental aspects were said to refer toelectricity use, the presence of environmentally hostile materials,and the presence of materials that can be recycled. Moreover,indications were given on how favourable or unfavourablea particular television set was on a particular attribute (seeAppendix). Participants were asked to indicate which televisionset they would choose if they were to buy one, assuming thatprices were approximately equal. Each television was codedaccording to their favourability towards environmental aspects.

4 There were only two main effects of Order of questionnaire on the TPBconstructs. PBC and Norms related to not reducing resource consumption werehigher when the measures related to not reducing were presented first (M ¼ 4.35,SD ¼ 1.01 and M ¼ 3.39, SD ¼ 1.03, respectively) than when the measures related toreducing were asked first (M ¼ 3.79, SD ¼ .98 and M ¼ 2.99, SD ¼ 1.00, respec-tively), t(102) ¼ 2.83, p ¼ .006 and t(102) ¼ 2.04, p ¼ .044. There was no main effecton behaviours (all ps > .256). And there were no significant moderation effects ofOrder on the validity of both Intentions in predicting behaviours (all ps > .130).

5 Like in Study 1, we tested whether the participants completed the two sessionsdiffered from the participants who completed only the first session in terms ofmeans in TPB constructs related to reducing and not reducing resource consump-tion. There was only one main significant effect for the Attitude towards reducingresource consumption with participants who did the two sessions showing higherAttitude (M ¼ 5.90, SD ¼ .73) than participants who did only the first session(M ¼ 5.47, SD ¼ .83), t(102) ¼ �2.55, p ¼ .014 (for all other constructs, ps > .104).

Page 7: To reduce and not to reduce resource consumption? That is two questions

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122118

After the choice task, participants indicated how important each ofthe seven attributes had been for their choice on 7-point scalesranging from 1 (not at all important) to 7 (very important). Twoscores were considered, one concerning the importance ofecological dimensions (1 attribute) and another one concerningthe importance of non ecological dimensions (6 attributes) in theTV choice. As a second behavioural criterion participants werecontacted after one month and asked to indicate the frequencywith which they performed a series of 20 behaviours during theprevious month.6

3.2. Results

3.2.1. Construct and discriminant validityAs shown in Table 7, the correlations between the constructs

related to reducing resource consumption and not reducingresource consumption respectively were all significant. Morecentral to our concerns, the correlations between reducing and notreducing resource consumption (underlined values) were allnegative; ranging from �.64 (Attitudes) to �.05 (PBCs). The resultsoverall confirm the previous results, indicating a substantial inde-pendence in the cognitions about reducing versus not reducingresource consumption.

For a more formal test, the results from Table 3 for Study 2showed that the solution with two correlated but distinct factorsabout reducing and not reducing resource consumption was thebest solution. As in Study 1, all constructs achieve construct anddiscriminant validity (4 from �.77 to �.13) for reducing versus notreducing resource consumption, again supporting H2.

3.2.2. Incremental validity3.2.2.1. Intentions. As for Study 1, a series of regression analyseswere performed in order to examine the incremental validity of TPBconstructs in predicting both Intentions of reducing and notreducing resource consumption. When considering the predictionof the Intention of reducing resource consumption (Table 4, Study2), both Attitudes show incremental validity, that remained whenentering PB, which in turn was a significant predictor. In addition,PBC for reducing resource consumption predicted Intention but itdid not when considering PB. For the prediction of the Intention ofnot reducing resource consumption (Table 5, Study 2), both Atti-tudes and PBC related to not reducing were significant predictors.This pattern remained also when entering the significant predictorPB (Step 3). Considering the simultaneous validity of both Attitudesfor predicting both Intentions, results confirmed those obtained inStudy 1, supporting H3.

3.2.2.2. Behaviours. Both Intentions of reducing and not reducingresource consumption predicted TV Choice (b ¼ .23, p ¼ .032and b ¼ �.27, p ¼ .013, R2 ¼ .19). When PBCs were entered intothe equation (Step 2) both Intentions were significant predictors,but when PB was also entered only the Intention of not reducingremained significant (Step 3). For predicting the importance ofecological dimensions in the TV Choice (R2 ¼ .22), the Intentionto reduce was a significant predictor (b ¼ .36, p ¼ .001) whereasthe Intention to not reduce was not (b ¼ �.17, p ¼ .102). Thispattern remained unchanged when considering PBCs (Step 2)

6 The difference in terms of number of behaviours compared to Study 1 has twomain reasons. First, we wanted to increase the range of behaviours related toreducing resource consumption so to increase the range and hence validity of themeasure. Second, university students are likely to be potentially engaged ina greater range of ecological behaviours than high-school students, given theirgreater independence in everyday life.

and PB (Step 3). On the contrary, for predicting the importanceof non ecological dimensions (R2 ¼ .11) the Intention to reducewas not a significant predictor (b ¼ �.07, p ¼ .521) and theIntention to not reduce was (b ¼ .29, p ¼ .009). This validityremained when considering PBCs and PB. Finally, PBC related tonot reducing was significant in predicting importance of nonecological dimensions in the TV choice. For SRB (R2 ¼ .24), bothIntentions were significant predictors (b ¼ .26, p ¼ .029 andb ¼ �.30, p ¼ .012, respectively). The two Intentions were againsignificant when entering PBCs (Step 2). When considering alsoPB in the equation, only the Intention a related to not reducingremained significant predictors. Globally, the inclusion of PBweakened the conclusion from the results towards a supportto H4.

3.2.3. Mediation of effect of both Intentions on TV Choice byImportance of Ecological versus non-ecological dimensions

As shown above (see also Table 8), the Intention of reducingand the Intention of not reducing resource consumption predictedthe importance of ecological and non ecological dimensions,respectively. Moreover, the importance of ecological and nonecological dimensions and the TV choice were significantly inter-correlated (see Table 7). Note that, mirroring the original proce-dure of the task (Verplanken & Holland, 2002, Study 1 & 2), theimportance of ecological and non ecological dimensions wererated after the TV choice because choice was the primarydependent variable. We hypothesized that the importance of theecological dimensions would be a stronger mediator of the effectof the Intention to reduce resource consumption on TV choicethan of the effect of the Intention to not reduce resourceconsumption on TV choice (H5). Inversely, the importance of nonecological dimensions would mediate the effect of the Intention ofnot reducing resource consumption on TV choice to a greaterextent than the effect of the Intention of reducing on TV choice(H6). We used bootstrap estimation to determine the significanceof indirect effects (Preacher & Hayes, 2004) by formally testing themediation hypotheses utilizing the SPSS macro developed byPreacher and Hayes (2004). Bootstrap methods involve samplingwith continuous replacement from a particular sample to drawa large number of unique samples and consequently provideincreased statistical power (e.g., MacKinnon, Lockwood, Hoffman,West, & Sheets, 2002; MacKinnon, Lockwood, & Williams, 2004).For each test, we estimated 5000 bootstrap samples and wecomputed 95% confidence intervals to conclude whether theindirect effect was significantly different from zero at the signifi-cance level of.05 (two-tailed).

First, the relation between Intention of reducing resourceconsumption and TV choice (cf. Fig. 1) was fully mediated by theimportance of ecological dimensions (M effect ¼ .36, SE ¼ .08, 95%CI ¼ .20, .52) whereas it was partially mediated by the importanceof non ecological dimensions (M effect ¼ .09, SE ¼ .05, 95%CI ¼ .008, .19), supporting H5. Second, the relation betweenIntention of not reducing resource consumption and TV choicewas partially mediated by the importance of non-ecologicaldimensions (M effect ¼ �.12, SE ¼ .04, 95% CI ¼ �.22, �.04) and itwas also partially mediated by the importance of ecologicaldimensions (M effect ¼ �.27, SE ¼ .07, 95% CI ¼ �.42, �.15), notsupporting H6. Therefore, whereas the importance of ecologicaldimensions was a stronger mediator of the effect of the Intentionto reduce resource consumption on TV choice than it was of theeffect of the Intention to not reduce resource consumption, theimportance of non ecological dimensions was not a strongermediator of the effect of the Intention of not reducing resourceconsumption on TV choice than it was of the effect of the Intentionof reducing it.

Page 8: To reduce and not to reduce resource consumption? That is two questions

Table 7Correlations among constructs (Study 2).

1 2 3 4 5 6 7 8 9 10 11 12 13

1. Attitudereducing (.85)2. Normsreducing .29* (.83)3. Perceived Behavioural Controlreducing .38* .37** (.84)4. Intentionreducing .56** .33* .54** (.91)5. Past Behaviour .41** .31* .59** .72** (.81a)6. Attitudenot reducing �.64** �.12 �.33* �.53** �.38** (.82)7. Normsnot reducing �.23* �.62** �.31* �.31* �.33* .41** (.83)8. Perceived Behavioural Controlnot reducing �.17 �.16 �.05 �.14 �.17 .29* .39** (.61)9. Intentionnot reducing �.46** �.15 �.27* �.53** �.39** .56** .40** .28* (.86)10. TV Choice .36** �.01 .27* .37** .37** �.36** �.08 .12 �.39** e

11. Importance Ecological Dimensions .28** �.03 .35** .45** .43** �.29** �.04 .13 �.36** .69** e

12. Importance Non Ecological Dimensions �.21* .08 �.10 �.23* �.20* .25** .18 .36* .33** �.39** �.40** (.63)13. Self-Reported Behaviour .35** �.03 .19 .40** .45** �.41** �.07 �.20 �.42** .35** .43** �.27* (.81)

Note. Reliabilities are indicated into brackets when applicable.The correlations between reducing and not reducing resource consumption constructs are indicated with underlining.*p < .05. **p < .01.N ¼ 104; for SRB, N ¼ 73.

a Only two items r, p < .001.

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122 119

3.3. Discussion

Study 2 replicated the main findings of Study 1; all conceptsrelated to reducing and to not reducing resource consumptionshowed construct and discriminant validity. Both Attitudes pre-dicted both Intentions. As expected, both Intention to reduce andIntention to not reduce predicted the TV choice and SRB. Therefore,Study 2 provides additional evidence that the cognitions related toreducing resource consumption are distinct from those about notreducing resource consumption. Moreover, in this second study,results revealed that the importance of ecological dimensions inthe TV choice was predicted by the Intention to reduce whereas theimportance of non ecological dimensions was predicted by theIntention to not reduce. Finally, mediation analyses demonstratedthe predominant mediator role of the importance of ecologicaldimensions for the effect of both the Intention to reduce and theIntention to not reduce resource consumption on TV choice.

4. General discussion

Considering that the pathways to action and to inaction can bedifferent and not mutually exclusive, the main aim of this contri-bution was to show that cognitions underlying performing and notperforming pro-environmental behaviours were different and notopposite and hence should be simultaneously considered in theprediction of pro-environmental intention and behaviour.Although in line with research on improving TPB prediction ofintention and behaviour (e.g., Sparks & Shepherd, 1992), this workdiffers through its focus on behavioural alternatives reflected in notperforming a behaviour, rather than adding new constructs. We cansummarize the key results from the two studies in four points: 1)

Table 8Incremental validity of constructs for the prediction of behaviours (Study 2).

TV Choice ImportancDimension

Step 1 Step 2 Step 3 Step 1

R2 .15 .20 .21 .22Intentionreducing .32 .18 .08 .37Perceived Behavioural Controlreducing .09 .10 .05 .14Intentionnot reducing L.26 L.27Perceived Behavioural Controlnot reducing �.02 .00Past Behaviourreducing .18

Note. Significant factors are indicated in bold.N ¼ 104, for SRB, N ¼ 73.

goals underlying reducing and not reducing resource consumptionare different and not necessarily opposites, 2) TPB constructsrelated to reducing and not reducing resource consumption showdiscriminant validity, 3) some constructs related to both reducingand not reducing resource consumption simultaneously predictboth Intentions of reducing and not reducing resource consump-tion, and 4) the Intention of not reducing resource consumptionshows incremental validity for predicting behaviours, suggestingthat a strong intention of not reducing resource consumption is notthe same as a weak intention of reducing resource consumption.The latter is a critical point, as it implies both that reducing and notreducing resource consumption follow two relatively differentmental processes and that both can contribute to explain andpredict ecologically relevant behaviours.

Results from the first study indicate that goals linked to theprotection of the planet primarily underlie reducing resourceconsumption whereas goals linked to maintaining a comfortableway of life mainly underlie not reducing resource consumption.Previous research has investigated the goals, motivations and, ata more general level, the values associated with pro-environmentalbehaviour and their role on intentions or behaviour (e.g., de Groot &Steg, 2007, 2010; Pelletier et al., 1998; Stern, 2000). Althoughresearch has been devoted to demonstrate that different motiva-tions are attached to an ecological behaviour and to one of its nonecological alternative (e.g., de Boer et al., 2009), very few researchstudies have investigated the goals or motivations associated withthe simplest alternative that is not behaving pro-environmentallyand their role as determinants of intentions and behaviour (i.e.,de Groot & Steg, 2010; Pelletier et al., 1999). Moreover, previouswork, while distinguishing between behaving and not behavingpro-environmentally, have assumed that not behaving pro-

e Ecologicals

Importance Non EcologicalDimensions

Self-Reported Behaviour

Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3

.24 .25 .05 .19 .19 .16 .24 .30

.28 .18 e.24 �.09 �.06 .41 .25 .06

.15 .10 .03 .01 .02 �.01 .03 �.10�.16 �.16 .21 .21 L.28 L.28�.04 .03 .29 .28 �.09 �.06

.17 �.04 .37

Page 9: To reduce and not to reduce resource consumption? That is two questions

Fig. 1. Mediation analyses of the Intention of reducing versus of not reducing resource consumption on TV Choice. The value in parentheses indicates the effect of the Intention ofreducing versus not reducing resource consumption on TV Choice after controlling for the Importance of Ecological versus non Ecological dimensions in the TV choice. *p < .05.*p < .01. ***p < .001.

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122120

environmentally is mostly a lack of motivation in behaving pro-environmentally. Pelletier et al. (1999) argue,“.the factors at thesource of people’s motivation for engaging in environmentalbehaviours may be quite different than the factors underlyingenvironmental amotivation. Thus, a better understanding of theamotivation phenomenon may contribute to clarify the complexdynamic involved in environmental action” (p. 2484), definingamotivation as “a state in which individuals are not able to perceivea contingency between their behavior and the subsequent outcomes oftheir behavior” as Deci and Ryan (1985) originally did. We agreewith the importance of focussing also onwhy people do not behavepro-environmentally but our results suggest that the motivationlinked to not reducing resource consumption is not necessarilyequivalent to the amotivation linked to reducing resourceconsumption. We wish to stress that this is not a question ofsemantics but of substance. The results of these studies show thatpeople do not engage in pro-environmental behaviours because ofa lack of motivation or amotivation but because they have a moti-vation to not do so. We believe that to better understand pro-environmental behaviours this distinction is very important. Con-firming previous results in different domains (Richetin et al., 2011)this research has shown that reducing and not reducing resourceconsumption are means to different goals and belong to twodifferent goal systems. These different goals provide psychologicaldistinctiveness between reducing and not reducing resourceconsumption that in turn lead to the importance of considering thedecision making processes underlying both doing (i.e., reducingresource consumption) and not doing (i.e., not reducing resourceconsumption). We also found that the use of established models ofattitude such as the TPB considering the relevant cognitiveconstructs (Intentions, Attitudes, Norms, PBC) for both doing andnot doing helps to better understand and predict pro-environmental behaviour. We acknowledge that the TPB format

implies a specific framing of the behaviour and therefore leadsparticipant to think only to one alternative at a time. This could beperceived as forcing a separation between cognitions that ineveryday life situations may be activated in a more mixed fashion.However, we believe that this may also be a positive feature for atleast two reasons. First, it allows to separate cognitions anddistinguish motivations underlying both options to do and to notdo, hence increasing the theoretical understanding of their specificrole. Second, it allows to significantly increase the overall predictivevalidity because it allows to consider both cognitions simulta-neously. Note that this increase in predictive validity should nothappen if the not doing cognitions weremerely the opposites of thedoing cognitions (for a similar argument, see Ajzen, 2011, p. 1117).

In this work both Attitudes, to doing and to not doing, (Study 1 &2) and both PBCs (Study 1, Intention of not doing) explained anadditional portion of variance in Intentions. As well as confirmingprevious research showing the importance of Attitude andPerceived Behavioural Control in the Intention of behaving pro-environmentally (see Bamberg & Möser, 2007 for a review), theseresults demonstrate the importance of also considering theconstructs related to the specific alternative that consists in notbehaving pro-environmentally. However, the inclusion of all TPBconstructs for both alternatives might not be always the bestsolution in order to better predict pro-environmental Intentionsand behaviours. Norms, as well as PBCs do not systematically showincremental validity. For example, although the overall resultssupport the assumption that norms predict intentions to behavepro-environmentally (e.g., Goldenhar & Connell, 1993; Taylor &Todd, 1997), only Norms related to reducing played a role forIntention of reducing and Norms related to not reducing forIntention of not reducing and only in Study 1. It should also beacknowledged that the introduction of past behaviour has led to thereduction of the additional predictive contribution of the two

Page 10: To reduce and not to reduce resource consumption? That is two questions

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122 121

intention measures for SRB in Study 2. However, besides thecontroversial theoretical status of past behaviour (cf. Ajzen, 2002b),its inclusion led to the predictive validity only of the Intention ofnot reducing resource consumption. Therefore, although consid-ering past behaviour might reduce the contribution of somecognitions related to performing or not performing pro-environmental behaviour, our results suggest that the beststrategy is to first consider them all together so to establishempirically which cognitions of which option (doing versus notdoing) may play a more important role in a given study.

With the second study, this contribution also sheds some lighton potential mechanisms underlying the differential effects of theIntentions of reducing and not reducing resource consumption onpro-environmental behaviour. More precisely, the results show thatthe importance of ecological dimensions in the TV choice task ispredicted by the Intention to reduce whereas the importance ofnon ecological dimensions is predicted by the Intention to notreduce resource consumption. Mediation analyses provide supportonly to the hypothesis that the importance of ecological dimen-sions is a stronger mediator of the effect of the Intention to reduceresource consumption on TV choice than of the effect of theIntention to not reduce resource consumption on TV choice. In fact,contrary to our expectation, the importance of non-ecologicaldimensions does not mediate the effect of the Intention of notreducing resource consumption on TV choice to a greater extentthan the effect of the Intention of reducing on TV choice. However,caution is required when drawing conclusions from these results.The ratings of importance were taken after the TV choice, becausechoice was the primary dependent variable, which may haveconferred a particular status to these hypothesizedmediators in theparticipants’ decision making process. In fact, one might wonder ifthe ratings could be considered as a post-choice rationalization.However, recent research has provided empirical evidence thatpost-choice ratings can be valid retrospective indicators of theimportance given to key features when choosing because they canreflect an important step inherent to the choice based on a coher-ence-generating mechanism that operates to make decisions (seeSimon, Krawczyk, & Holyoak, 2004; Simon, Pham, Le, & Holyoak,2001). Future studies may be needed to confirm the resultsemploying a procedure that allows online monitoring of informa-tion acquisition behaviour (cf., the information-display-boardparadigm in Verplanken & Holland, 2002, Study 3). Such orsimilar procedures would allow a further look at some of themechanisms involved in the decision making process throughwhich both Intention of reducing and Intention of not reducingaffect an environmentally relevant behaviour.

It should be acknowledged that the variance explained inbehaviours is relatively modest (from 11 to 30%) and the incrementgiven by considering cognitions related to both doing and not doingis relatively small (from 1 to 14%). However, the gain in theoreticalunderstanding is substantial, as our results have clarified that thedecision making processes related to not reducing resourceconsumption play a role of their own rather than being just anopposite (or a lack of) those related to reduce resource consump-tion. Furthermore, besides the non-negligible role of automaticprocesses in “anti-environmental” behaviour (e.g., Kolmuss &Agyeman, 2002) that might explain a portion of the intention-behaviour gap, probably one should also consider as quiteremarkable that we have been able to predict specific behaviourswith general stems such as reducing versus not reducing resourceconsumption, leading to poor correspondence between measure-ment of constructs and measurement of behaviour (Eagly &Chaiken, 1993; Fishbein & Ajzen, 1975). This is particularly truefor the more specific behaviours such as the TV choice or thehypothetical scenarios which have been predicted by measures of

the TPB constructs are at a very general level (reducing versus notreducing resource consumption). On the other side, using veryspecific construct measures (e.g., about buying a TV based on itsenvironmental qualities) would have been awkward and probablycould have even inflated the results due to some potentially blatantdemand effects.

The common motivation to all the research in the pro-environmental domain that examine processes triggering inten-tions and behaviours is to find ways to change them so to increasethem. By showing that the processes underlying not doing are notthe opposite of the processes underlying doing, this contributionalso has some implications for research into behaviour change. Infact, pro-environmental behaviour can mean doing something suchas recycling glass but can also mean not doing something else suchas littering or leaving the water running while brushing teeth. Inthe perspective of behavioural change, by showing that thecognitions underlying each doing and not doing option of the twobehaviours are not the opposites, one might want to start consid-ering different approaches for increasing the performance of onebehaviour (e.g., recycling) or for decreasing another one (e.g.,leaving the water running while brushing teeth). Moreover, byknowing that both Intentions might simultaneously predictbehaviour, one might want to simultaneously work on factors thatwould increase the Intention of acting pro-environmentally and onfactors that would decrease the Intention of not acting pro-environmentally. Finally, de Groot and Steg (2007) suggest thatfuture studies should focus on examining how values can best bechanged or how to motivate people to act upon their values. Westrongly concur and would add that research could fruitfully focuson the goals or values that are linked to behaving pro-environmentally as well as on the goals or values that are linkedto not behaving pro-environmentally. Based on the fact that thetwo options (doing and not doing) belong to two different systems,future research should be devoted to further investigate thedifferences between these two systems and the conditions underwhich one system dominates. Those conditions could be inducedby situational factors such as the accessibility of one goal more thanthe other or by individual differences such as self-control or even bythe type of behaviour such as tempting versus virtuous behaviour.

To conclude, by considering separately the triggers for not per-forming a pro-environmental action from those for performing thesame action such as reducing and not reducing resourceconsumption we believe that this contribution prolongs previousresearch that considers alternative behaviours together withecological ones. The focus on the two simplest alternatives that aredoing and not doing might allow to better understand the under-lying mechanisms of engagement with pro-environmental behav-iours that would go beyond improving the predictive power ofmodels and to provide new insights on how to more successfullyintervene.

Acknowledgements

Study 1 and 2 data were collected as part of a Unilever grant(SPHERE CH-2009-1108) to Marco Perugini and Juliette Richetin.

We would like to thank Federico Selvaggio and Denny Mondinifor their help in data collection in Study 1 and Simona Zec for herhelp in data collection in Study 2.

Appendix. Supplementary material

Supplementary material associated with this article can befound, in the online version, at doi:10.1016/j.jenvp.2012.01.003.

Page 11: To reduce and not to reduce resource consumption? That is two questions

J. Richetin et al. / Journal of Environmental Psychology 32 (2012) 112e122122

References

Aarts, H., & Dijksterhuis, A. (2000). The automatic activation of goal-directedbehaviour: The case of travel habit. Journal of Environmental Psychology, 20,75e82.

Aarts, H., Verplanken, B., & van Knippenberg, A. (1998). Predicting behaviour fromactions in the past: Repeated decision making or a matter of habit? Journal ofApplied Social Psychology, 28, 1356e1375.

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

Ajzen, I. (2002a). Perceived behavioral control, self-efficacy, locus of control, and thetheory of planned behavior. Journal of Applied Social Psychology, 32, 665e683.

Ajzen, I. (2002b). Residual effects of past on later behavior: Habituation andreasoned action perspectives. Personality and Social Psychology Review, 6,107e122.

Ajzen, I. (2011). The theory of planned behavior: Reactions and reflections.Psychology & Health, 26, 1113e1127.

Ajzen, I., & Fishbein, M. (1969). The prediction of behavioral intentions in a choicesituation. Journal of Experimental Social Psychology, 5, 400e416.

Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes,intentions, and perceived behavioral control. Journal of Experimental SocialPsychology, 22, 453e474.

Bagozzi, R. P. (1981). Attitudes, intentions and behavior: A test of some keyhypothesis. Journal of Personality and Social Psychology, 41, 607e627.

Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera:A new meta-analysis of psycho-social determinants of pro-environmentalbehaviour. Journal of Environmental Psychology, 27, 14e25.

de Boer, J., Boersema, J., & Aiking, H. (2009). Consumers’ motivational associationsfavoring free-range meat or less meat. Ecological Economics, 86, 850e860.

Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and eval-uative space: A critical review, with emphasis on the separability of positiveand negative substrates. Psychological Bulletin, 115, 401e423.

Cheung, S. F., Chan, D. K. S., & Wong, Z. S. Y. (1999). Reexamining the theory ofplanned behavior in understanding waste paper recycling. Environment andBehavior, 31, 587e612.

Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behavior: Areview and avenues for further research. Journal of Applied Social Psychology, 28,1429e1464.

Corral-Verdugo, V., & Frías-Armenta, M. (2006). Personal normative beliefs, anti-social behavior, and residential water conservation. Environment and Behavior,38, 406e421.

Courneya, K. S., Bobick, T. M., & Schinke, R. J. (1999). Does the theory of plannedbehavior mediate the relationship between personality and exercise behavior?Basic and Applied Social Psychology, 21, 317e324.

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in humanbehavior. New York: Plenum.

Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. New York: HarcourtBrace.

Fielding, K. S., McDonald, R., & Louis, W. R. (2008). Theory of planned behaviour,identity, and intentions to engage in environmental activism. Journal of Envi-ronmental Psychology, 28, 318e326.

Fielding, K. S., Terry, D. J., Masser, B., & Hogg, M. A. (2008). Integrating social identitytheory and the theory of planned behaviour to explain decisions to engage insustainable agricultural practices. British Journal of Social Psychology, 47, 23e48.

Fishbein, M. (1980). A theory of reasoned action: Some implications. In H. Howe, &M. Page (Eds.), 1979 Nebraska symposium on motivation (pp. 65e116). Lincoln,NE: University of Nebraska Press.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior. Reading, MA:Addison-Wesley.

de Groot, J. I. M., & Steg, L. (2007). General beliefs and the theory of plannedbehavior: The role of environmental concerns in the TPB. Journal of AppliedSocial Psychology, 37, 1817e1836.

de Groot, J. I. M., & Steg, L. (2010). Relationships between value orientations, self-determined motivational types, and pro-environmental behavioural inten-tions. Journal of Environmental Psychology, 30, 368e378.

Goldenhar, L. M., & Connell, C. M. (1993). Understanding and predicting recyclingbehavior: An application of the theory of reasoned action. Journal of Environ-mental Systems, 22, 91e103.

Harland, P., Staats, H., & Wilke, H. A. M. (1999). Explaining pro-environmentalintention and behaviour by personal norms and the theory of planned behav-iour. Journal of Applied Social Psychology, 29, 2505e2528.

Higgins, E. T. (2000). Making a good decision: Value from fit. American Psychologist,55, 1217e1230.

Jaccard, J. (1981). Attitudes and behavior: Implications of attitudes toward behav-ioral alternatives. Journal of Experimental Social Psychology, 17, 286e307.

Kaiser, F. G. (2006). A moral extension of the theory of planned behaviour: Normsand anticipated feelings of regret in conservationism. Personality and IndividualDifferences, 41, 71e81.

Kolmuss, A., & Agyeman, J. (2002). Mind the gap: Why do people act environ-mentally and what are the barriers to pro-environmental behavior? Environ-mental Education Research, 8, 239e260.

Kruglanski, A. W., Shah, J. Y., Fishbach, A., Friedman, R., Chun, W. Y., & Sleeth-Keppler, D. (2002). A theory of goal systems. In M. P. Zanna (Ed.), Advances inexperimental social psychology (pp. 331e378). San Diego: Academic Press.

Letirand, F., & Delhomme, P. (2005). Speed behaviour as a choice between observingand exceeding the speed limit. Transportation Research Part F: Traffic Psychologyand Behaviour, 8, 481e492.

MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002).A comparison of methods to test mediation and other intervening variableeffects. Psychological Methods, 7, 83e104.

MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for theindirect effect: Distribution of the product and resampling methods. Multivar-iate Behavioral Research, 39, 99e128.

Pelletier, L. G., Dion, S., Tuson, K., & Green-Demers, I. (1999). Why do people fail toadopt environmental protective behaviors? Toward a taxonomy of environ-mental amotivation. Journal of Applied Social Psychology, 29, 2481e2505.

Pelletier, L. G., Tuson, K. M., Green-Demers, I., Noëls, K., & Beaton, A. M. (1998). Whyare you doing things for the environment?: The motivation toward the envi-ronment scale. Journal of Applied Social Psychology, 28, 437e468.

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

Perugini, M., & Bagozzi, R. P. (2004a). An alternative view of pre-volitional processesin decision making: Conceptual issues and empirical evidence. In G. Haddock, &G. R. Maio (Eds.), Contemporary perspectives on the psychology of attitudes: TheCardiff symposium (pp. 169e201). Hove, UK: Psychology Press.

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

Perugini, M., & Conner, M. (2000). Predicting and understanding behavioral voli-tions: The interplay between goals and behaviors. European Journal of SocialPsychology, 30, 705e731.

Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirecteffects in simple mediation models. Behavior Research Methods, Instruments, &Computers, 36, 717e731.

Richetin, J., Conner, M., & Perugini, M. (2011). Not doing is not the opposite of doing:Implications for attitudinal models of behavioral prediction. Personality andSocial Psychology Bulletin, 37, 40e54.

Rivis, A. J., & Sheeran, P. (2004). Descriptive norms as an additional predictor in thetheory of planned behaviour: A meta-analysis. Current Psychology, 22,264e280.

Schwartz, S. H. (1977). Normative influence on altruism. In L. Berkowitz (Ed.),Advances in experimental social psychology, Vol. 10 (pp. 221e279). New York:Academic Press.

Sheeran, P., & Orbell, S. (1999). Augmenting the theory of planned behavior: Rolesfor anticipated regret and descriptive norms. Journal of Applied SocialPsychology, 29, 2107e2142.

Simon, D., Krawczyk, D. C., & Holyoak, K. J. (2004). Construction of preferences byconstraint satisfaction. Psychological Science, 15, 331e336.

Simon, D., Pham, L. B., Le, Q. A., & Holyoak, K. J. (2001). The emergence of coherenceover the course of decision making. Journal of Experimental Psychology: Learning,Memory, and Cognition, 27, 1250e1260.

Sparks, P., & Shepherd, R. (1992). Self-identity and the theory of planned behaviour:Assessing the role of identification with “green consumerism”. Social PsychologyQuarterly, 55, 388e399.

Stern, P. C. (2000). Toward a coherent theory of environmentally significantbehavior. Journal of Social Issues, 56, 407e424.

Stets, J. E., & Biga, C. F. (2003). Bringing identity theory into environmental soci-ology. Sociological Theory, 21, 398e423.

Taylor, S., & Todd, P. (1997). Understanding the determinants of consumer com-posting behavior. Journal of Applied Social Psychology, 27, 602e628.

Vermeir, I., & Verbeke, W. (2006). Sustainable food consumption: Exploring theconsumer “attitude-behavioral intention” gap. Journal of Agricultural and Envi-ronmental Ethics, 19, 169e194.

Verplanken, B., & Holland, R. (2002). Motivated decision making: Effects of acti-vation and self-centrality of values on choices and behavior. Journal of Person-ality and Social Psychology, 82, 434e447.