Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

10
Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans Paula Castro * , Margarida Garrido, Elizabeth Reis, Joa ˜o Menezes Departamento de Psicologia Social e das Organizaço ˜es, Instituto Superior de Cie ˆncias do Trabalho e da Empresa, Av. das Forças Armadas, 1649-026 Lisboa, Portugal article info Article history: Available online 30 November 2008 Keywords: Conservation behaviours Ambivalence Social representations Environmental attitudes Contradiction abstract As resource conservation is an imperative for sustainable development, it is crucial to achieve a deeper understanding of the factors involved in people’s decisions to recycle. This is even more so because the level of environmental concern is usually higher than the level of ecological behaviours. In this paper we take this fact as indicating that decision-making regarding conservation behaviours happens in the context of an internal debate where contradictory ideas are weighed up and the possibility of ambivalence arises. The main aim of this paper is therefore to explore how contra- diction and ambivalence impact upon the attitudes, intentions and pro-ecological behaviours of the private sphere. The paper focuses specifically on the separation and deposition of metal cans, and compares the predictive capacity of beliefs, attitudes and intentions for two groups of respondents – one with a high and another with a low level of ambivalence, as assessed with a direct measure. The role of personal identity and the influence of structural constraints are also explored. Results demonstrate a clear moderating effect of experienced ambivalence, and show how beliefs, particularly negative ones, present a higher predictive capacity of the attitude in the high-ambivalence group, and personal identity plays a relevant role in predicting behaviour in both groups. We discuss the importance of pursuing the study of ambivalence when analysing decision-making in the conserva- tion area. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction As concerns with sustainability become more central in our societies, the conditions under which individuals make decisions to behave in ways that affect levels of resource consumption are increasingly researched. The environmental psychology of ‘‘sustainability’’ has consequently grown in the past years (Bonnes & Bonaiuto, 2002) and one recurrent finding of research in this field regards the lack of correspondence, or gap, between ideas and behaviour (Se ´guin, Pelletier, & Hunsley, 1998; Van Liere & Dunlap, 1981; Vining & Ebreo, 2002). It has often been shown that in spite of the widespread high levels of environmental concern in the pop- ulation ‘‘people greatly differ in the amount of time and energy that they are willing to invest’’ in behaviours aimed at preserving the quality of the environment (Se ´guin et al., 1998, p. 628), like conservation behaviours. One consequence of recent efforts to understand this gap was an accrued awareness of the multidi- mensionality, variety and often contradictory nature of environmental views 1 (Corral-Verdugo, Bechtel, & Fraijo, 2003; Kurz, Donaghue, Rapley, & Walker, 2005; Lima & Castro, 2005; Steg & Sievers, 2000). Another consequence was the realization that clearer typologies of ecological behaviours and their specific predictors were needed (Stern, 2000). The present article examines conservation behaviour, specifi- cally addressing the recycling of metal cans 2 , the material with the lowest deposition rates in Portugal in 2004 (see Sociedade Ponto Verde). Drawing from research highlighting the variety and * Corresponding author. Tel.: þ351 21 790 3001; fax: þ351 21 790 3002. E-mail address: [email protected] (P. Castro). 1 In the environmental literature the label ‘‘environmental attitudes’’ is commonly used to refer to several aspects such as values, beliefs, evaluations, etc, regarding the natural environment. Here we will use the following notion of ‘‘attitude’’: the explicit ‘‘evaluation of an object, concept or behavior along a dimension of favor or disfavor’’ (Ajzen & Fishbein, 2000, p. 3). This definition assumes ‘‘attitudes’’ to be conceptually distinct from other notions such as values or beliefs (Zanna & Rempel, 1988) which may also refer to the same object – here, the natural environment - and which, together with attitudes, or evaluations, comprise what is here termed ‘‘environmental views’’. 2 All types of cans, made of steel and aluminum, are being considered here, for in Portugal they are all to be deposited together in the same drop-off curbside container. Contents lists available at ScienceDirect Journal of Environmental Psychology journal homepage: www.elsevier.com/locate/jep 0272-4944/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jenvp.2008.11.003 Journal of Environmental Psychology 29 (2009) 24–33

Transcript of Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

Page 1: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

lable at ScienceDirect

Journal of Environmental Psychology 29 (2009) 24–33

Contents lists avai

Journal of Environmental Psychology

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

Ambivalence and conservation behaviour: An exploratory study onthe recycling of metal cans

Paula Castro*, Margarida Garrido, Elizabeth Reis, Joao MenezesDepartamento de Psicologia Social e das Organizaçoes, Instituto Superior de Ciencias do Trabalho e da Empresa, Av. das Forças Armadas,1649-026 Lisboa, Portugal

a r t i c l e i n f o

Article history:Available online 30 November 2008

Keywords:Conservation behavioursAmbivalenceSocial representationsEnvironmental attitudesContradiction

* Corresponding author. Tel.: þ351 21 790 3001; faE-mail address: [email protected] (P. Castro).

0272-4944/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.jenvp.2008.11.003

a b s t r a c t

As resource conservation is an imperative for sustainable development, it is crucial to achievea deeper understanding of the factors involved in people’s decisions to recycle. This is even more sobecause the level of environmental concern is usually higher than the level of ecological behaviours.In this paper we take this fact as indicating that decision-making regarding conservation behaviourshappens in the context of an internal debate where contradictory ideas are weighed up and thepossibility of ambivalence arises. The main aim of this paper is therefore to explore how contra-diction and ambivalence impact upon the attitudes, intentions and pro-ecological behaviours of theprivate sphere. The paper focuses specifically on the separation and deposition of metal cans, andcompares the predictive capacity of beliefs, attitudes and intentions for two groups of respondents –one with a high and another with a low level of ambivalence, as assessed with a direct measure. Therole of personal identity and the influence of structural constraints are also explored. Resultsdemonstrate a clear moderating effect of experienced ambivalence, and show how beliefs, particularlynegative ones, present a higher predictive capacity of the attitude in the high-ambivalence group, andpersonal identity plays a relevant role in predicting behaviour in both groups. We discuss theimportance of pursuing the study of ambivalence when analysing decision-making in the conserva-tion area.

� 2008 Elsevier Ltd. All rights reserved.

1 In the environmental literature the label ‘‘environmental attitudes’’ iscommonly used to refer to several aspects such as values, beliefs, evaluations, etc,regarding the natural environment. Here we will use the following notion of‘‘attitude’’: the explicit ‘‘evaluation of an object, concept or behavior alonga dimension of favor or disfavor’’ (Ajzen & Fishbein, 2000, p. 3). This definitionassumes ‘‘attitudes’’ to be conceptually distinct from other notions such as values orbeliefs (Zanna & Rempel, 1988) which may also refer to the same object – here, thenatural environment - and which, together with attitudes, or evaluations, comprise

1. Introduction

As concerns with sustainability become more central in oursocieties, the conditions under which individuals make decisions tobehave in ways that affect levels of resource consumption areincreasingly researched. The environmental psychology of‘‘sustainability’’ has consequently grown in the past years (Bonnes& Bonaiuto, 2002) and one recurrent finding of research in this fieldregards the lack of correspondence, or gap, between ideas andbehaviour (Seguin, Pelletier, & Hunsley, 1998; Van Liere & Dunlap,1981; Vining & Ebreo, 2002). It has often been shown that in spite ofthe widespread high levels of environmental concern in the pop-ulation ‘‘people greatly differ in the amount of time and energy thatthey are willing to invest’’ in behaviours aimed at preserving thequality of the environment (Seguin et al., 1998, p. 628), likeconservation behaviours. One consequence of recent efforts tounderstand this gap was an accrued awareness of the multidi-mensionality, variety and often contradictory nature of

x: þ351 21 790 3002.

All rights reserved.

environmental views1 (Corral-Verdugo, Bechtel, & Fraijo, 2003;Kurz, Donaghue, Rapley, & Walker, 2005; Lima & Castro, 2005; Steg& Sievers, 2000). Another consequence was the realization thatclearer typologies of ecological behaviours and their specificpredictors were needed (Stern, 2000).

The present article examines conservation behaviour, specifi-cally addressing the recycling of metal cans2, the material with thelowest deposition rates in Portugal in 2004 (see Sociedade PontoVerde). Drawing from research highlighting the variety and

what is here termed ‘‘environmental views’’.2 All types of cans, made of steel and aluminum, are being considered here, for in

Portugal they are all to be deposited together in the same drop-off curbsidecontainer.

Page 2: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–33 25

multidimensionality of environmental views, the paper exploresthe notion that this variety may result in people experiencingambivalence, i.e., evaluating attitude objects as simultaneouslypositive and negative (Thompson, Zanna, & Griffin, 1995), andpresents an exploratory study of the moderating role of experi-enced ambivalence regarding recycling, in the context of modelslike the Theory of Reasoned Action. The following sections discussthe main theoretical assumptions.

2. Thinking about the environment – complexity, diversity,contradiction

As mentioned, a number of recent studies concur that envi-ronmental views are multidimensional and varied (Castro, 2006;Diekmann & Franzen, 1999; Stern, Dietz, & Guagnano, 1995; Uzzell,2000). In the past, research on environmental views monitoredwhether social support was raising towards a so-called ecologicalworldview (Catton & Dunlap, 1980), over a number of years andwithin countries (see Arcury & Christianson, 1990; Dunlap, VanLiere, Mertig, & Jones, 2000). Or it compared different countriesregarding their levels of support of either ecological or anthropo-centric views, often with the New Environmental Paradigm (NEP)scale (Furman, 1998; Gooch, 1995). It was assumed that only one bi-polar factor organized these views and people were distributed ona continuum, rejecting one pole and accepting the other. However,research has come to highlight the multidimensionality of envi-ronmental views (Bonaiuto, Carrus, Martorella, & Bonnes, 2002;Ellis & Thompson, 1997; Lima & Castro, 2005; Milfont & Duckitt,2004; Uzzell, 2000). Recent studies show how old anthropocentricviews (or the previously Dominant Social Paradigm, as they werealso-called (Catton & Dunlap, 1980)) can co-exist with newecological ideas, not only in society, but also in the same individual(Castro & Lima, 2001). This notion is also present in the concept of‘‘sustainable development’’ (Bruntland, 1987), which assumes thepossibility of conciliating and balancing the use and preservation ofnature (Bonnes & Bonaiuto, 2002; Milfont & Duckitt, 2004). The useand preservation of resources are now viewed as complementarygoals and not opposite ones (Milfont & Duckitt, 2004). This iscorroborated by positive correlations found between the anthro-pocentric and ecological sub-scales of the NEP scale for somecountries (Bechtel, Verdugo, & Pinheiro, 1999; Castro & Lima, 2001).

The diversity and conflictive content of environmental viewsamong all publics has in fact been claimed from the beginning byCultural Theory (Douglas & Wildavsky, 1982). For this approach, thehigh levels of agreement with ecological ideas revealed in surveysmay indicate only a surface consensus (Ellis & Thompson, 1997),enabled by the possibility of maintaining ideas and behavioursuncoordinated. Behind this surface consensus, heterogeneity andcontroversy would be the reality (Steg & Sievers, 2000). Thisheterogeneity and complexity of meaning are however not visibleto the same extent in all countries researched (Milfont & Duckitt,2004). One possible reason for this may be the fact that ecologicalideas quickly became socially desirable in some publics (Uzzell,2003). This may be linked to the fact that some countries haveinstitutionalized sustainability goals through legal instrumentsfaster than other countries. This institutionalization originates notjust new laws but also environmental campaigns and educationprojects and therefore makes the expression of views opposinglegal norms and goals more difficult.

To summarize, these recent studies seem to suggest that whennovel ideas enter a society’s common sense – as ecological ideasdid, from the late 1960s on in most western countries (Hayward,1994) – the existing representational system does not change ina sudden and simple manner, discarding the old ideas andaccepting the new ones. On the contrary, change seems to resultin the co-existence of competing representations, as Social

Representations Theory has stressed (Wagner & Hayes, 2005).Moreover, by stimulating the social debate, change originates newcounter-arguments (Moscovici, 1988). As a consequence, varied andopposing ideas regarding the same object become available forindividual decision-making.

In this sense, the literature on sustainability suggests theimportance of considering that social change progresses in stages.The research findings and reflections of the last decades regardingchange in the ecological direction indicate that Stage 1–the Emer-gence of change stage – was characterized by the emergence ofenvironmental concern as a new social trend, already clear amongthe general public of most world countries in the 1970s (Dunlap,Gallup, & Gallup, 1993; Dunlap & Mertig, 1995; Hayward, 1994; VanLiere & Dunlap, 1981). Stage 2–the Institutionalization one – hasfollowed, with national legislation changing in many countries inthe sequence of international treaties, and new formal goals forsustainability were established. Several governments, like theEuropean Union ones, signed commitments bounding them toattain, for instance, certain levels of recycling of domestic wastematerials. This took various countries to another phase: theGeneralization one, or Stage 3. In this stage, making new ideas andnew everyday practices generalize to a population becomesa priority for governments and pressure groups, like environmen-talists. Individuals are encouraged to change by media campaigns,environmental projects, and other means, which emphasise thepositive aspects of the new pro-ecological behaviours.

The societal pressures for change of Stage 3 have severalconsequences at the psycho-social level. As mentioned, researchshows how people come to align their ideas with the new envi-ronmental laws and norms more rapidly than their practices(Bonaiuto et al., 2002; Castro & Lima 2001; Costarelli & Colloca,2004). This suggests that when we are dealing with norm guidedbehaviour we need to consider not just the imperfect capacity ofattitudes and beliefs for predicting behaviour but also that attitudespresent lower means than behaviour, i.e., we need to consider thatwe are dealing with a specific type of inconsistency. This is relevantbecause inconsistency can also be expressed in the other wayaround, as is the case with drive-motivated behaviour, like eating orsmoking (Bamberg & Schmidt, 1998), where low means for theattitudes can be found together with high means for the behaviour.The specific direction the inconsistency assumes is important inpsycho-social terms, since it informs about people’s awareness ofthe norms. Regarding conservation behaviours, specifically, theinconsistency pattern found may suggest that people are indeedaware that conservation ideas are socially desirable. A similarawareness of the norm may also be acting when, in response to thepressures for change, hybrid views and discourses, which accom-modate conflictive meaning, emerge (Castro, 2006; Uzzell, 2003)and sometimes even create barriers for changing practices (Kurzet al., 2005).

In summary, the idea that many societies, and EU ones inparticular, are in the Stage of Generalization of ecological change ishere taken as a hypothetical descriptor of a moment when peoplecan be expected: (1) to be aware of the new formal and informalnorms regulating discourse and action regarding ecological issues,(2) to avoid blatant opposition to them, (3) while still circum-venting them in various ways, namely by maintaining ideas andactions uncoordinated. This is congruent with a view of socialchange as shaped by the debate between heterogeneous ideas. Onepsycho-social consequence of this is that people take decisions ina context where contradictory ideas exist for supporting divergentpaths of action. A methodological consequence is that researchdesigns need to assess contradictory ideas. This may help us tobetter understand the nature of decision-making when contradic-tory views co-exist for individuals to evaluate the new practicestheir societies require them to initiate.

Page 3: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–3326

3. Predicting environmentally relevant actions

As the interest in both understanding and changing individualdecisions remained central in sustainability psychology, severalreasons were along the years advanced to account for the difficultyof beliefs or attitudes for predicting behaviours: (1) the lack ofa consistent theoretical framework regarding the levels of analysisof the variables used (Stern et al., 1995); (2) the lack of a unitaryconcept of attitude (Kaiser, Wolfing, & Fuhrer, 1999), with differentconcepts named ‘‘attitudes’’ (Castro, 2006); (3) the lack ofa typology of ecological behaviours (Stern, 2000); (4) the failure toconsider structural constraints and the costs of actions (Corraliza &Berenguer, 2000; Kaiser, Oerke, & Bogner, 2007; Kaiser et al., 1999).

Regarding the first problem, Stern et al. (1995) proposed anintegrative model which assumes that distal ecological views, suchas those tapped by the NEP scale, help to predict proximal beliefs,which, in turn, predict specific attitudes, which are the bestpredictors of behaviours. This model uses a definition of ‘‘attitude’’which sees these as evaluations (Fishbein & Ajzen, 1975), and takesbeliefs, as antecedents of attitudes, to regard the content of what isthought and said about a certain social object (Zanna & Rempel,1988). It both conceptually distinguishes between beliefs andattitudes and between distal (generic beliefs) and proximal(specific beliefs), assuming also that proximal variables carrya stronger potential for predicting both attitudes and behaviours.Nevertheless, several studies in sustainability psychology havecontinued to focus on the role of more distal variables for behaviourprediction (Nordlund & Garvill, 2002), and the respective role ofdistal and proximal variables still needs clarification.

For overcoming the second problem above identified – the needof a unified concept of attitude – the Theory of Planned Behaviour(TPB) has been considered central for the ecological domain (Kaiseret al., 1999). This theory, like the Theory of Reasoned Action (TRA)(Ajzen & Fishbein, 2000; Fishbein & Ajzen, 1975),3 defines attitudesas evaluations and establishes them as the best predictors ofintentions (Ajzen & Fishbein, 2000). In this tradition attitudes areseen as determined by cognitive, affective and behavioural ante-cedents, and as having cognitive, affective and behavioural conse-quents, but as consisting only of evaluations measured in an explicitform (Ajzen & Fishbein, 2000; Zanna & Rempel, 1988). Theseoptions enable researchers to examine whether or not the sameattitude can be predicted by different beliefs (van der Pligt, Eiser, &Spears, 1986), and therefore allow them to examine the impact ofthe multidimensional nature of environmental beliefs upon atti-tudes. This is an important goal, for models of the type of TPB, dueto the fact that they conceptually differentiate beliefs and attitudesalso ‘‘more readily invite the consideration of the notion ofambivalent attitudes’’ (Thompson et al., 1995, p. 362). And if theexistence of conflicting environmental views is now clear, theimpact of contradiction and controversy on attitudes needs to beexamined, which means looking at attitudinal ambivalence (seeCostarelli & Colloca, 2004 for a pioneer study). This may beparticularly important when behavioural prediction is not the solepurpose of the research, and a better understanding of the factorsthat may be relevant for fostering future change in norm guidedactions is also sought, as is the case of the present study.

These conceptual clarifications are even more relevant now thatclearer typologies of pro-ecological behaviours are available (Stern,2000), meeting the third shortcoming above identified. Thebehaviour here studied – domestic waste recycling – has now beencharacterized as an ecological behaviour of the private sphere,a type of behaviour with a potentially positive environmental

3 TPB and TRA are similar models, except for the fact that consideration of theimpact of perceived behavioral control was added to TRA and transformed it in TPB.

impact which only becomes significant when many people inde-pendently choose to perform it (Stern, 2000). This transforms thesituation under which decisions concerning recycling are madeinto a doubly dilemmatic one. On the one hand, people have to dealwith the dilemmas opened by the existence of contradictory beliefsregarding recycling. On the other, they face the social dilemma ofknowing that their acts count only if others do as they do, whilethey cannot control the behaviour of others.

4. The role of ambivalence and self-identity

The first dilemma above mentioned makes it important toexamine the role of ambivalence regarding recycling; the secondbrings to the fore the role of personal involvement and self-identityin this domain. These variables were investigated in the context ofattitude models such as the TRA and TPB, which were along theyears subjected to two types of expansions: ‘‘theory broadening’’attempts and ‘‘theory deepening’’ efforts (Perugini & Bagozzi,2001). The former add to the standard model new predictors, andthe latter try to find moderators or mediators. Regarding modera-tors, studies point to the role of ambivalence (Jonas, Broemer, &Diehl, 2000). Self-identity is among the predictors recently shownto be relevant (Conner & McMillan, 1999; Mannetti, Pierro, & Livi,2004). They are now approached in turn.

Ambivalence can be defined as the degree to which an attitudeobject is evaluated positively and negatively at the same time(Thompson et al., 1995; Wegener, Downing, Krosnick, & Petty,1995). The literature identifies two types of ambivalence – expe-rienced or subjective ambivalence and objective or indirectambivalence. The later is deduced by the researcher through themeasurement of both positive and negative evaluations combinedthrough some formula. Experienced measures imply offeringpeople the opportunity to express their subjective feelings ofambivalence by having them describing how much they feeldivided regarding the evaluation of some object or experiencecontradictory ideas about it. Comparisons have shown correlationsbetween both measures (e.g., Priester & Petty, 1996). For the envi-ronmental domain, a reasonably high correlation of .43 betweenobjective and subjective ambivalence was found (Costarelli & Col-loca, 2004). This indicates that the two measures, although related,do not tap exactly into the same reality, and the use of one or theother is to be determined by the goals of the research. Experiencedambivalence is more coherent with the empirical aims of thepresent article. This is so because, as suggested by low Portugueserecycling rates, the new separation and deposition practices are notyet installed habits, but instead decisions of a deliberate nature.What in this domain constitute habits, or automatic responsesexecuted without much deliberation (Verplanken, Myrbakk, &Rudi, 2005), are practices of non-separation. Therefore, it isreasonable to suppose that to separate and to deposit are norm-related actions demanding deliberation and vigilance. They arehence reasoned in the sense of the TRA: derived from accessiblebeliefs about the attitude object (Ajzen & Fishbein, 2000), whichcan, however, be contradictory, as mentioned. This means thata central question to be explored is whether individuals are awareof how they react to the co-existence of contradictory ideas, i.e.,whether they indicate experiencing ambivalence.

Although some authors contend that ambivalence shouldweaken the links between attitude and intention, and betweenintention and behaviour (Sparks, Hedderley, & Shepherd, 1992)others argue that ambivalence may be accompanied by consider-able elaboration, consequently becoming a catalyst for the inte-grative resolution of contradiction (Jonas, Diehl, & Broemer, 1997).This reasoning is consistent with the existence of stronger attitude–behaviour relations for high levels of ambivalence, and with weakerlinks for low levels of ambivalence, for these may reflect the

Page 4: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–33 27

absence of efforts towards coping. In this paper, consistently withthis, we expect stronger links when there is awareness of one’s ownambivalence and more intense elaboration can be expected. In sum,regarding ‘‘theory deepening’’ variables, the paper will consider themoderating role of experienced ambivalence.

Regarding the ‘‘theory broadening’’ variables, it will explore therole of self-identity (Mannetti et al., 2004; Terry, Hogg, & White,1999). As mentioned, self-identity, defined as ‘‘the salient part of anactor’s self which relates to a particular behaviour’’ (Conner &Armitage, 1998, p. 1444), was shown to increase the intention ofrecycling for those who regard recycling as an important compo-nent of their self-identity (Terry et al., 1999). It was also shown toindependently predict the intention of performing ecologicalbehaviours (Mannetti et al., 2004; Sparks & Shepard, 1992), and willbe explored here.

Finally, regarding the prediction of behaviour, some studiessuggest the importance of considering also how structural featuresmay either constrain or facilitate pro-ecological behaviour,addressing the problem of the costs of behaviour (Kaiser et al.,2007). Variables regarding structural features were also added tothe attitude models and proved relevant (see Corraliza & Berenguer,2000), which makes important the consideration of their role. Inlight of this, we opted for studying the recycling of metal cans byexploring the moderating role of experienced ambivalence and byadding to the attitude models questions tapping the role ofpersonal identity and structural constraints.

5. General aim and specific objectives

This study assumes that in Portugal we are currently in Stage 3of ecological change, the Generalization one. The assumption istaken as a hypothetical descriptor of a moment in most societies,and in European Union ones in particular, when ecological goalshave been made mandatory through new laws and policycommitments and ecological ideas are socially desirable (Uzzell,2003), but, on the other hand, contradictory ideas about these goalscirculate in society, available for decision-making. Based on these,ambivalence may emerge when individuals are evaluating certainactions now socially required from them, like recycling. Portugal isa society where governmentally defined recycling goals exist, and,in this context, it seems reasonable to expect that expressed beliefs,attitudes and intentions – the level of ideas – will be in accordancewith the norm, showing high consensus across individuals. It is alsoreasonable to expect that non-normative beliefs, like ideas againstrecycling, will receive more support from those who experienceambivalence. Because our society now expects us to maintainrecycling, it is reasonable to anticipate that, when aware of theirambivalent leanings, individuals will struggle to maintain it. Forthis reason, stronger belief–attitude, attitude–intention andintention–behaviour links are expected for individuals experi-encing more ambivalence. Low experienced ambivalence would, inturn, propitiate less deliberation and, therefore, less clear links.These hypotheses, focused only on recycling practices regardingdomestic waste, have not yet, to our knowledge, been explored. Itwas demonstrated that objective ambivalence mediates the role ofsubjective ambivalence for predicting the intention for compositepro-environmental behaviours (Costarelli & Colloca, 2004).However, the ambivalence felt towards pro-ecological behavioursin general might not be the same as the one experienced regardinga specific norm guided behaviour of the private sphere (Stern,2000), which is the object of societal commitments and efforts. Thismakes it relevant to examine whether the responses of those whodo not experience ambivalence differ from the responses of thosewho do. Moreover, using a measure of subjective ambivalenceavoids the biases identified by Ullrich, Schermelleh-Engel, andBottcher (2008) when ambivalence is measured by collapsing

separate measures of positive and negative attitudes into a singleindex. The comparison between high- and low-ambivalencerespondents is then the general purpose of the present exploratorystudy, which has the following specific objectives and hypotheses:

1. Comparative objectives considering ambivalence asmoderator:

Hyp1: when high- and low-ambivalence respondents arecompared, high and similar means (consensus) for pro-ecological beliefs (i.e., distal beliefs), for pro-recycling beliefs(i.e., proximal beliefs) and for the attitude and intentionmeasures are expected, due to the normative status of thesevariables.Hyp2: when high- and low-ambivalence respondents arecompared, differentiation on the variables not affected bysocial desirability is expected: for anti-ecological beliefs andbeliefs against recycling, higher means for high-ambivalencerespondents, as compared to low-ambivalence ones areanticipated. Regarding self-reported behaviour and ecologicalself-identity differentiation is also expected: lower means forthose who experience high-ambivalence are anticipated.There are no expectations for perceived structural features.

2. Predictive objectives considering ambivalence as a moderator.2.1. High- and low-ambivalence respondents will be compared in

order to examine whether the traditional predictors of TPBfollow different paths for the two groups, and to examinewhether the links are stronger in the high-ambivalence group.

2.2. High- and low-ambivalence respondents will be compared inorder to examine whether the new predictors – self-identityand structural features – follow different paths for the twogroups, and to examine whether the links are stronger in thehigh-ambivalence group.

Hyp3: For the above mentioned analyses, the predictivecapacity of variables is expected to be higher for the high-ambivalence group.

2.3. For both ambivalent and non-ambivalent groups, the role ofboth distal and proximal beliefs in the prediction of attitudestowards the recycling of metal cans will be explored. Nospecific predictions are made regarding the predictive powerof distal and proximal beliefs because the literature presentsconflicting findings.

6. Method

6.1. Procedure and participants

Data for the present study were collected in the context ofa broader project regarding metal and plastic repositioning,commissioned to us by the organization in charge of the manage-ment of domestic waste in Portugal (Sociedade Ponto Verde), ina context where the recycling levels of these materials needed to beincreased to meet governmental commitments. The applied goalsof the project included achieving a better understanding of theseparation and deposition decisions in order to implement adver-tising campaigns, environmental education and other relevantprograms; they also included an appraisal of the recycling program,and of other aspects connected with the Sociedade Ponto Verde.The variety of applied goals created a number of constraintsregarding the specific purposes of the present paper. The instru-ment could not be long, and the complexity of the measures used toapproach a diversified population could not be high. In an idealsituation, the attitude and intention measures would be morecomplex than the ones used. The same is true regarding theambivalence and the constraint measures. This is why the presentresearch represents only an exploratory study.

Page 5: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

Table 1Items used to assess views about the environment.

Pro-ecological beliefs Factor loads

Ideally, ecological behaviours should become automatic habits .751It is possible to control environmental problems if the government

produces laws indicating clearly what we can and cannot do.656

Environmental problems can only be controlled if people areforced to radically change their behaviour

.631

Anti-ecological beliefsThe so-called ‘‘ecological crisis’’ facing humankind has been

greatly exaggerated.782

There are more important things in life than protecting theenvironment

.682

Nowadays, one already feels like resisting toenvironmentalist excesses

.455

We do not need to worry about environmental problemsbecause science and technology will be able to solve them

.450

Table 2Items used to assess positive and negative beliefs about metal recycling.

Pro-recycling of metal beliefs Factor loadings

The separation and deposition of domestic waste in metal isa way of preserving our natural heritage

.769

The separation and deposition of domestic metal waste helpsprotecting future generations

.757

The separation and deposition of domestic metal wastecontributes to reducing pollution

.728

The separation and deposition of domestic metal residualscontributes to reduce landfills

.726

The separation and deposition of domestic metal waste savesnatural resources

.712

The separation and deposition of domestic metal waste isimportant for the protection of people’s health

.666

It is important to separate and deposit domestic metal wastebecause in this way the material is not wasted

.651

The separation and deposition of domestic metal waste allowsfor the material to be useful to other people

.624

The separation and deposition of domestic metal waste isimportant because the material has value in itself

.537

Anti-recycling of metal beliefsI do not have the patience to separate and deposit my

metal waste.691

I find it uncomfortable to separate my metal waste fromthe remaining waste

.683

I find it annoying to keep hearing about separation anddeposition of domestic waste

.625

The separation and deposition of domestic metal wasteis a business that only benefits some people

.595

The separation and deposition of domestic metal wasteis an habit hard to acquire

.495

To recycle metal waste is more expensive than not to recycle it .429In terms of environmental protection, the separation

and deposition of domestic metal waste is useless.410

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–3328

The first step of the study was the conduction of two focusgroups, following Fishbein and Ajzen’s (1975) advice to try toensure that the heterogeneous pool of beliefs circulating in societyis represented in questionnaires. Participants in the groups wereresidents in the area of great Lisbon (Ngroup1¼7, age: Mgroup1¼22;Ngroup2¼ 6; Mgroup2¼ 29), which for 2 h discussed environmentalprotection, recycling and the use and separation of metal cans. Thequestionnaire was designed taking into account the literature andthe content analysis of the group discussions, particularly helpfulfor collecting negative beliefs about recycling.

The sample was selected using quotes to ensure that sex,education and age groups were represented. On the whole, 394participants were surveyed, of which 59.5% were women. The agerange varied from 14 to 63 years (M¼ 29.44; SD¼ 12.31). Forty fivepercent of the participants had completed high school, 34.4% helda university degree, and around 20% had completed elementary orjunior high education. Nearly half (42%) were still living with theirparents, 38% lived with a partner and about 36% had children. Theparticipants who lived in the two major metropolitan areas of Lis-bon (78.6%) and Porto (21.4%), were approached in large shopping-malls and completed the questionnaires in a face-to-face situation.

6.2. Variables

6.2.1. General ecological viewsSome items used to assess general distal ecological views were

adapted from previous studies (Dunlap et al., 2000; Lima & Castro,2005; Steg & Sievers, 2000) and others were created from thediscussions from the focus groups. Items were responded on a 5-point Likert scale, ranging from 1¼ totally disagree to 5¼ totallyagree. An Exploratory Factor Analysis with varimax rotation yieldedtwo factors (with eigenvalues greater than one; variance explain-ed¼ 47%; Factor 1 – pro-ecological beliefs; a¼ .53, mean inter-itemcorrelation¼ .27; Factor 2 – non-ecological beliefs; a¼ .51, meaninter-item correlation¼ .214). Two scores were computed aver-aging the items, which are listed in Table 1.

6.2.2. Beliefs about metal recyclingTo measure beliefs about metal recycling, items were adapted

from previous studies and from the discussions in the focus groups.

4 For examining the reliability of each factor two indicators were used: internalconsistency (Cronbachs alpha) and homogeneity (inter-item correlations). Thisdouble procedure follows the recommendations of Briggs and Cheek (1986) forovercoming Cronbach’s alpha coefficient sensitivity to the length of the scale. Infact, these authors affirm that ‘‘the mean inter-item correlation is not influenced byscale length and it is therefore a clearer measure of items homogeneity’’ (p. 115),and they consider the optimal value of this indicator to be between .2 and .4 (p.115),a criteria fulfilled by the factors used in this study.

An Exploratory Factor Analysis forced to extract two factors5 withvarimax rotation explained¼ 42% of the variance (Factor 1 – posi-tive beliefs about metal recycling; a¼ .86, mean inter-itemcorrelation¼ .41; Factor 2 – negative beliefs about metal recycling;a¼ .65, mean inter-item correlation¼ .21). Two scores werecomputed, averaging the items, listed in Table 2.

6.2.3. Attitude, intention and behaviourAttitude – to assess participants’ attitude towards separation and

deposition of metal waste, we used the following item: ‘‘I considerthe separation and deposition of metal waste as something’’ (from1¼ very negative to 5¼ very positive).

Intention – to tap participants’ intention regarding separationand deposition of metal waste, we used the following item: ‘‘Iintend to start/continue separating and depositing my domestic metalwaste’’ (from 1¼ totally untrue to 5¼ totally true).

Behaviour – to assess self-reported behaviour, participants wereinvited to state the ‘‘frequency with which I separate and deposit mymetal waste in the Ecoponto’’6 (from 1¼ never to 5¼ always).

6.2.4. Attitudinal ambivalenceFor the measurement of subjective, or experienced, attitudinal

ambivalence two questions were employed: ‘‘Regarding the sepa-ration and deposition of my metal waste I feel that my attitude is’’(from 1¼ not at all contradictory to 7¼ very contradictory);‘‘Regarding the separation and deposition of my domestic waste (ingeneral) I feel that my attitude is’’ (from 1¼ not at all contradictory to

5 A four factor solution (the one yielded by the criterion of eigenvalues greaterthan one) would increase the percentage of explained variance up to 55% but wouldlead to all factors presenting lower reliabilities.

6 Ecoponto is the Portuguese word that designates the street containers used toseparately deposit the recyclable domestic waste in our drop-of system.

Page 6: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

Table 3Description and comparison of the main variables for the two ambivalence groups.

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–33 29

7¼ very contradictory). The two items (r¼ .66) were averaged intoa single variable.

6.2.5. Self-identityFor assessing self-identity we used an index with the following

items: ‘‘I like to think of myself as someone with ecological concerns’’;‘‘I recycle my metal waste because I want to feel that I personallycontribute to the protection of the environment’’; ‘‘I think of myself assomeone actively committed to the environmental cause’’ (answeredin a 5-point Likert scale, from 1¼ totally disagree to 5¼ totallyagree; a¼ .68; mean inter-item correlation¼ .41). This measureprovides an assessment of self-identity which is both adjusted tothe specific case (I recycle my metal waste because.), and moreencompassing, connecting the specific and the more general levelsof the ecologically relevant self-concept.

6.2.6. Structural featuresTo tap the most central structural constraint to recycling

behaviours in our drop-of system, we used the item ‘‘I have anEcoponto near my house’’, answered in a 5-point Likert scale (from1¼ totally untrue to 5¼ totally true).

6.2.7. Preliminary data analysis: establishing ambivalence asa moderator

In order to examine the role of ambivalence as a moderator, wefirst needed to asses whether ambivalence altered the indepen-dent–dependent variables relation in a step way, following Baronand Kenny (1986, p. 1176), who established that if this is the case,one can then dichotomize the moderator and form contrastinggroups. A study of the correlations among these variables for threegroups of low-, medium- and high-ambivalence, which werecreated by partitioning the ‘‘ambivalence’’ index, showed that thecorrelations between attitude and intention (rlow-ambivalence

group¼ .15; rmedium-ambivalence group¼ .27; rhigh-ambivalence group¼ .10)and between intention and self-reported behaviour (rlow-ambivalence

group¼ .27; rmedium-ambivalence group¼ .19; rhigh-ambivalence group¼ .43)support the conclusion that the independent and dependent vari-ables are in a step relation with the moderator (Baron & Kenny,1986). It is also the case that the variances are not different acrosslevels of the moderator (the variance of the attitude measure isthe same across the three levels of low, medium- and high-ambivalence [L(2,333)¼ 2,3, n.s.], as is the variance of the intention[L(2,343)¼ 0,22, n.s.] and the variance of self-reported behaviour[L(2, 344)¼ 1,8, n.s.]), another characteristic that supports studyingthe moderator as a dichotomy (Baron & Kenny, 1986). Based onthese results, we dichotomized the ambivalence measure, con-structing, as Baron and Kenny (1986) advise, two contrastinggroups and eliminating the medium ambivalence group, i.e., thosewithout a clear position regarding their own ambivalence7. In thefollowing sections we therefore compare two contrasting groups:a low-ambivalence group (n¼ 171, Mambivalence¼ 2.15) and a high-ambivalence group (n¼ 73, Mambivalence¼ 5.27).

Low-ambivalence High-ambivalence

GeneralM

N M SD N M SD t p

Pro-ecological beliefs 4.0 169 3.99 .70 72 4.07 .68 �.880 n.s.Anti-ecological beliefs 2.4 168 2.31 .66 72 2.53 .77 �2.285 .02Pro-recycling of

metal beliefs4.1 167 4.16 .62 72 4.02 .63 1.573 n.s.

Anti-recycling ofmetal beliefs

2.5 164 2.37 .69 69 2.82 .61 �4.767 .000

7. Results

7.1. Comparative aims considering ambivalence as moderator

In order to compare high- and low-ambivalence respondentsand test the comparative hypotheses Hyp1 and Hyp2, means and

7 Original observations, measured in a 1–7 scale, were split into three groups: thefirst included observations with values ranging from 1 to 3; the second groupincluded observations with value 4; the remaining observations (5–7) wereincluded in group 3. Group 2 was then excluded from the analysis because wewanted well contrasted and separated groups.

standard deviations for each of the considered variables werecomputed and the respective t tests performed. Results are pre-sented in Table 3. The pattern obtained is consistent with the firstprediction (Hyp1). High and similar means (consensus) wereexpected and were found regarding pro-ecological beliefs(Mlow¼ 3.99; Mhigh¼ 4.07; p¼ n.s.) and pro-recycling beliefs(Mlow¼ 4.16; Mhigh¼ 4.02; p¼ n.s.). Consensus was also observedfor attitudes (Mlow¼ 4.36; Mhigh¼ 4.11; p¼ n.s.) and intentions(Mlow¼ 3.81; Mhigh¼ 3.55; p¼ n.s.) towards recycling. This meansthat, as predicted, the two groups do not show significant differ-ences on variables with a normative component.

The predictions (Hyp2) regarding negative beliefs, behaviour,and self-identity were also supported. The results show how thelow-ambivalence group presents lower levels of support for anti-ecological (Mlow¼ 2.31; Mhigh¼ 2.53; p< .02) and anti-recyclingbeliefs (Mlow¼ 2.37; Mhigh¼ 2.82; p< .00) and reports a higherperformance of the behaviour (Mlow¼ 2.85; Mhigh¼ 1.90; p< .00).In turn, high-ambivalence respondents show higher support foranti-recycling beliefs and perform the behaviour less frequently.For self-identity, a variable with a less obvious normative status,differentiation was also observed in the expected direction, withthe low-ambivalence group showing a higher mean (Mlow¼ 3.71;Mhigh¼ 3.25; p< .00). These were also the ones perceiving thestructural recycling features as more positive (Mlow¼ 3.53;Mhigh¼ 2.92; p¼ .02).

These findings suggest that when in a society change isproposed by new legislation and norms, more agreement is seenwhen the variables assessed are similar to the norm (the case ofpro-ecological or pro-recycling beliefs). This seems consistent withthe notion that the awareness of what is the ‘‘right thing’’ to think iswhat first generalizes.

7.2. Predictive aims in the context of ambivalence as amoderator – proximal and distal beliefs as attitude predictors

Regarding the role of distal ecological beliefs and proximalecological beliefs in the prediction of attitudes towards recycling ofmetal cans, findings (see Table 4) support the idea that it is worthintegrating both, given that both seem to impact upon the forma-tion of attitudes. While general environmental views makea significant contribution for the explanation of the attitudetowards the recycling of metal cans, especially in high-ambivalencegroups (in Model 1, Adj. R2¼ .22), the combination of both generaland proximal beliefs increases the overall variance of the attituderesponses accounted for in the cases of both the low- and the high-ambivalence groups, but in particular for the latter (see Model 2,Adj. R2¼ .34).

Attitude 4.3 163 4.36 1.05 71 4.11 1.18 1.566 n.s.Intention 3.7 165 3.81 1.18 73 3.55 1.11 1.587 n.s.Behaviour metal waste 2.6 169 2.85 1.33 73 1.90 1.23 5.169 .000Structural features 3.3 170 3.53 1.63 50 2.92 1.74 2.283 .02Self-identity 3.6 157 3.71 .80 69 3.25 .83 3.938 .000

Note: Levine’s Test for Equality of Variances supported the equal variancesassumption for all the variables considered.

Page 7: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

Table 4Regressions of positive and negative distal and proximal beliefs to attitude.

Low-ambivalence High-ambivalence

Model 1 Model 1F(2, 157)¼ 9.4,p< 0.000, Adj. R2¼ .10

F(2, 66)¼ 10.3,p< 0.000, Adj. R2¼ .22

b Sig. b Sig.

Pro-ecological beliefs .14 .08 .04 n.s.Anti-ecological beliefs �.25 .00 �.47 .00

Model 2 Model 2F(4, 146)¼ 6.5,p< 0.000, Adj. R2¼ .13

F(4, 60)¼ 9.1,p< 0.000, Adj. R2¼ .34

b Sig. b Sig.

Pro-ecological beliefs .12 n.s. .01 n.s.Anti-ecological beliefs �.26 .00 �.35 .01Pro-recycling of metal beliefs .21 .00 .02 n.s.Anti-recycling of metal beliefs .10 n.s. �.40 .00

Table 6Regressions of intention, self-identity and structural features to behaviour.

Low�ambivalence High�ambivalence

Model 1 Model 1F(1, 161)¼ 13.1,p< 0.00, Adj. R2¼ .07

F(1, 71)¼ 15.9,p< 0.00, Adj. R2¼ .17

b Sig. b Sig.

Intention .27 .00 .43 .00

Model 2 Model 2F(3, 150)¼ 14.6,p< 0.00, Adj. R2¼ .22

F(3, 46)¼ 7.81,p< 0.00, Adj. R2¼ .31

b Sig. b Sig.

Intention .09 n.s. .28 .03Self-identity .39 .00 .43 .00Structural features .09 n.s. .19 n.s.

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–3330

Also relevant, although unforeseen by our predictions, is the factthat negative beliefs are the best overall predictors of the attitude.This suggests that the refusal to agree with ideas that denigrateecological protection and recycling plays a relevant role insustaining a positive attitude towards recycling. Taken together,these two facts – that beliefs play a more important role for thehigh-ambivalence group and that negative beliefs are betterpredictors of the attitude – may imply that when people are awareof their ambivalence they more carefully consider the variouscounter-arguments associated with recycling. This may show thatinner debate about recycling is at this point more focused onnegative beliefs than on positive ones.

7.3. Predictive aims – predicting intentions and behaviours

To test the strength of the links between attitude and intention,and intention and behaviour, in the context of experiencedambivalence as a moderator, the correlations for both groups werecalculated and are displayed in Table 5. Findings show that theattitude–intention link is weak (contrarily to TPA predictions), andnot very different between the two groups, whereas the intention–behaviour link is clearer in both groups, and particularly clear forthe high-ambivalence group. Together, these results may indicatethat when people make decisions in a situation where they weighpros and cons, and are aware of it, and then finally decide that theyintend to perform the behaviour, then they do perform it. In otherwords, in the high-ambivalence group, performance of the behav-iour is less frequent, but seems more consistent with the intention.

7.4. Predictive aims – attitude ambivalence and the new variablesin the TRA

In order to explore the role of self-identity and structuralconstraints in the prediction of behaviour and to compare the rela-tive strength they acquire in ambivalent and non-ambivalent groups,a series of regression analysis were performed. Intention was firstused as the sole predictor of behaviour and then, in a second step, thenew predictors were added. Results are displayed in Table 6. As

Table 5Correlations between attitude and intention and between intention and behaviour.

Low�ambivalence High�ambivalence

r Sig. r Sig.

Attitude-intention .15 .06 .11 n.s.Intention-behaviour .27 .00 .43 .00

already shown in Table 5, and in line with the classic theory, inten-tion alone predicts behaviour. Also as stressed, results show itacquires a higher predictive capacity when decisions are taken ina context of high-ambivalence. Particularly relevant is the fact thatthe overall predicting capacity of self-identity is important for bothgroups. As expected, this capacity is particularly evident for thosepeople who are ambivalent and know it. These seem to both rejectnegative beliefs and to draw on personal identity for their decision-making. The structural feature assessed seems less relevant for them.

7.5. Self-identity as a mediator of intention in behaviour

The high relevance that self-identity assumes as a directpredictor of behaviour required further testing, and the possiblemediating role of this variable regarding the intention–behaviourlink was determined for the high- and low-ambivalence groups,following Baron and Kenny’s (1986) recommendations. Findingsare reported in Fig. 1 (low-ambivalence group) and Fig. 2 (high-ambivalence group). As shown in Fig. 1, self-identity in fact medi-ates the intention–behaviour relationship: results indicate that forlow-ambivalence individuals, self-identity is the sole predictor ofbehaviour (b¼ .41, B¼ .67), since the effect of intention becomesnon significant (b¼ .11, B¼ .12) when it is considered together withidentity, although it was significant when independently consid-ered (b¼ .27, B¼ .31), and the relation between intention andidentity is strong (b¼ .45, B¼ .31). The Sobel test (z¼ 3.9, p< .000)indicates full mediation for this case. In the high-ambivalencegroup (Fig. 2) the adjusted model explains 29% of the variation ofthe self-reported behaviour, the association of identity to behaviouris strong ((b¼ .40, B¼ .70) and there is a mitigated mediating rolefor identity: the association of intention to behaviour (b¼ .28,B¼ .36) is attenuated when identity is considered, but the Sobeltest (z¼ 1.3, p< .20), indicates that there is no full mediation, evenif identity attenuates the direct relation (b¼ .43, B¼ .47).

8. Discussion

This paper departed from the assumption that social changeproceeds in a complex manner that can be organized into three

SELF-IDENTITY

INTENTION BEHAVIOUR

0.45 (0.47) 0.41

(0.27) 0.11 n.s.

Adj. R2=.21

Fig. 1. Mediation effect of self-identity for the low-ambivalence group.

Page 8: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

SELF-IDENTITY

INTENTION BEHAVIOUR

0.34(0.50) 0.40

(0.43) 0.28

Adj. R2=.29

Fig. 2. Mediation effect of self-identity for the high-ambivalence group.

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–33 31

main stages (i.e., emergence, institutionalization, and generali-zation). During the generalization stage, which comes after newideas are institutionalized through new legislation and policycommitments, certain goals demanding behavioural change aremade normative. Guided by this hypothetical proposal, the paperexplored the notion that during this Stage people may experienceattitudinal ambivalence when making decisions about the newnormative behaviours, due to the existence of opposing ideas aboutthem.

Findings show a moderating effect of experienced ambivalence.The comparison between low- and high-ambivalence respondentsshows a similar agreement with all variables that tap ideas thatresemble the norms and/or can be directly deduced from them:pro-ecological beliefs, pro-recycling beliefs, attitude and intentiontowards recycling. Instead, variables measuring less clearlynormative aspects – identity and the structural feature assessed –receive more agreement from low-ambivalence respondents. Theseparation and deposition of metal cans, in turn, is not frequent ineither group, but is even less frequent among those who experienceambivalence. These results can be discussed taking into account therole of habit in decision-making. If it so happened that for thoseindividuals low on ambivalence the recycling of cans was a habitualbehaviour, then ‘‘habit’’ could be a variable obscuring the role ofambivalence. However, the results show that the behaviour is nota habitual one for either group. In fact, although low-ambivalencerespondents report higher levels of behaviour (M¼ 2.85) thanhigh-ambivalence respondents (M¼ 1.9), both groups presentmeans below the theoretical mean of the scale (3), indicatinga reduced performance of the behaviour. This seems to suggest thatchanging towards recycling is under reasoned control.

The study also examined the role of both distal and proximalbeliefs, since their compared capacities as possible predictors ofattitudes have been discussed (Stern et al., 1995). Models such as theTPA or TRA assume specific beliefs about the behaviour to be the bestpredictors of attitudes, but many studies in sustainability psychologyhave attempted to show the capacity of general beliefs and values forpredicting behaviour (Nordlund & Garvill, 2002; Steg & Sievers,2000). It was shown that both types of beliefs were capable of pre-dicting independent parts of the variance in attitudes, and that thereare, therefore, advantages in continuing to consider both.

Another distinction, besides the distal-proximal one, emergedas relevant in these findings: the positive–negative beliefsdistinction. Fishbein and Ajzen (1975) advise the researcher toelicit heterogeneous beliefs for constructing the questions, but notall studies do this. There seem to be reasons, however, to be carefulin this regard, since the present results have shown that it is therejection of negative beliefs, and not so much the acceptance ofpositive ones, that seems to be forming the attitudes at this point.Here we need to recall that these issues – ecology and sustain-ability – have been the object of enlarged social discussion andmedia campaigns during the last 20 years in European Unioncountries, and people seem to be aware of both the ‘‘right thing’’ tothink and the possible counter-arguments. These results suggestthat the rejection of anti-recycling and anti-ecological argumentsseems to play a larger role, at this point of the sustainability debate,than the acceptance of the ‘‘right’’ ones. Although exploratory, andrequiring confirmation in more controlled conditions, these

findings seem corroborated by others which show also clearer linksbetween negative beliefs regarding water conservation and waterconservation behaviour (Corral-Verdugo et al., 2003). They alsopoint to a new direction to be explored: they allow the hypothesisthat education campaigns for the sustainable use of resourcesmight be better succeed if orchestrated for refuting negativebeliefs. The investment into building an awareness of the positiveconsequences of environmental protection and recycling seems tohave succeeded. People agree with the benefits. These results maynow indicate that the negative beliefs have to be brought out in theopen and refuted, because they seem to play a relevant role indecision-making in the high-ambivalence group of respondents.

It is also in the high-ambivalence group that the rejection ofnegative ideas is more important for attitude formation. The studyhere presented is an exploratory one, and no strong conclusions canbe drawn, but this result seems to suggest that decision-making isprobably a more deliberative process for those who are aware oftheir ambivalence (Jonas et al., 2000), and may go through theprocess of refuting negative arguments in a more reasoned way.The same reason may also explain why the intention–behaviourlink is also stronger for the high-ambivalence group: their morethorough consideration of beliefs may strengthen it. This maysuggest, in other words, that those who deliberately consider andthen dismiss negative beliefs and decide they will perform thebehaviour, actually perform it, and those who do not dismiss themand who do not decide as such, do not perform it. In contrast, forthose deciding without ambivalence or awareness of it, the link isnot only less strong, but also attenuated when we consider the roleof self-identity.

In this regard, these results suggest, in line with previous research(Terry et al., 1999), that viewing recycling as an important compo-nent of self-identity is predictive of behaviour. This variable mediatesthe role of intention for low-ambivalence respondents, explainingbehaviour over and above it. This suggest that self-identity is animportant variable to predict behaviour in the domain of recycling ofdomestic waste, which may have consequences for campaignsdesigned to promote it, as it may imply that they need to focus on theconnection of the self with the behaviour. As mentioned, recyclingdecisions are experience in a doubly dilemmatic context: (1)opposing arguments exist regarding resource conservation; (2)individual decisions are under social dilemma conditions, i.e., indi-viduals know that they are effective only if others also make them.For these reasons, the decision of maintaining the practice while it isnot habitual may depend on personal commitment and self-defini-tions compatible with the behaviour. These decisions cannot bemaintained by ‘‘looking around’’ to see what others do, but byviewing these goals as part of the self, regardless of what others do.For incentive campaigns, this means attempting to re-enforcepersonal definitions in the ecological direction.

At this point, a comment may be made concerning the fact thatwe found the relationship between intention and behaviour to bestronger for high-ambivalence respondents, while Conner, Povey,Sparks, James, and Shepherd (2003) found the opposite. In thisregard, the distinction between drive-motivated behaviours (likeeating or smoking) and norm-related behaviours (see Bamberg &Schmidt, 1998) must not be forgotten. May be for the first type ofbehavious ambivalence weakens the links, while for the latter itstrengthens them. And we have to consider, moreover, that thenorm-related behaviour here studied is both formally linked toexplicit governmental goals and is not habitual, demanding delib-erative effort, which may be one of the reasons strenghtening thelinks.

A comment on the obtained adjusted R2 which were not veryhigh, is needed. The type of models here used present betterpredictive capacities for aggregated behaviours (Kaiser et al., 2007).Taking into account that only one type of behaviour and only one

Page 9: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–3332

material were considered, the adjusted R2 are not particularly low,and on the whole suggest that one of the dimensions added to thestandard model, self-identity, may be an important target forcampaings aiming at changing recycling behaviour. The dimensionsadded to the model will never be able to predict all the variance inbehaviour, but when the goal is intervention and the modificationof behaviour, they indicate which are the aspects worth focusingon and those that are irrelevant. In view of this, we need to regretthe fact that the structural constraints were assessed in a ratherpoor way in the present study. This prevented us from achievinga better knowledge of how specific costs may predict specificbehaviours.

Other limitations of the present study regard its reliance onsingle items for the measurement of attitude and intention and ona reduced number of indicators to assess experienced ambivalence.This prevents a confident generalization of the results here pre-sented, because we do not know how stable and replicable they are.They need to be improved for a more reliable test of the role ofexperienced ambivalence regarding recycling decisions. However,the present findings suggest interesting paths for pursuing theresearch, namely indicating the importance of addressing in futurestudies the inter-relations between self-identity, subjectiveambivalence and objective ambivalence, using stronger indicatorsthan ours and possibly using also the multivariate approach to dataanalysis suggested by Ullrich et al. (2008). This would permitexploring whether mediation effects found in previous studies(Costarelli & Colloca, 2004) can be replicated for specific conser-vation behaviours and when the mediating role of self-identity isalso considered. It would also help clarifying what may be thedistinctive roles played by different types of ambivalence, becausewith our data it is not possible to know whether those respondentsreporting low-ambivalence would also have low means on anindirect measure of ambivalence. Nevertheless, our direct measuredoes suggest that some individuals are aware of experiencingambivalence, and in face of the present findings it seems worthexamining in closer detail what this awareness is associated with.Another shortcoming regards the fact that some of the factors thatwe worked with presented barely acceptable indicators. This canbe, nevertheless, counter-balanced by the fact that they are theresult of previous careful consultation through focus groups. A finalremark regards the non-inclusion of measures tapping normativebeliefs and subjective norms. These were not included because theyseem to be, in general, a weaker component of these models (Ajzen& Fishbein, 2000), a fact corroborated by a previous study in ourcountry (Castro, Reis, & Menezes, 2003) on the recycling of plastic,where their predictive role was small.

As a general conclusion, we concur with the notion that thestudy of ambivalence can open new and useful research lines in thestudy of pro-environmental behaviour (Costarelli & Colloca, 2004),in particular if ambivalence is seen as linked to the relationsbetween the three levels of the social, the contextual, and thepersonal (Bonaiuto et al., 2002; Castro, 2006; Uzzell, 2000),something not attempted in this paper but important for futureones. If the three levels are researched in conjugation, the notion ofambivalence may help understanding how normative elementsdefined at the societal level (e.g., the importance of a sustainableuse of resources) are welcome by all, but often conflict with func-tional beliefs regarding the Self (e.g., recycling is time consuming),and are resisted in everyday practices, and therefore sometimesprogress slowly through society.

References

Ajzen, I., & Fishbein, M. (2000). Attitudes and the attitude–behaviour relation:reasoned and automatic processes. In: European review of social psychology, Vol.11. Chichester, GB: Wiley. pp. 1–34.

Arcury, T. A., & Christianson, E. H. (1990). Environmental worldview in response toenvironmental problems: Kentucky 1984 and 1988 compared. Environment &Behavior, 22, 387–407.

Bamberg, S., & Schmidt, P. (1998). Changing travel-mode choice as rational choice:results from a longitudinal study. Rationality and Society, 10, 223–252.

Baron, R., & Kenny, D. (1986). The moderator–mediator variable distinction in socialpsychological research: conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychology, 51, 1173–1182.

Bechtel, R. B., Verdugo, V. C., & Pinheiro, J. Q. (1999). Environmental belief systems:United States, Brasil, and Mexico. Journal of Cross-Cultural Psychology, 30(1),122–128.

Bonaiuto, M., Carrus, G., Martorella, H., & Bonnes, M. (2002). Local identityprocesses and environmental attitudes in land use changes: the case of naturalprotected areas. Journal of Economic Psychology, 23, 631–653.

Bonnes, M., & Bonaiuto, M. (2002). Environmental Psychology: from spatial–phys-ical environment to sustainable development. In R. B. Bechtel, & A. Churchman(Eds.), Handbook of environmental psychology (2nd ed.). (pp. 28–54) New York:John Wiley & Sons.

Briggs, S. R., & Cheek, J. M. (1986). The role of factor analysis in the development andevaluation of personality scales. Journal of Personality, 54, 106–148.

Castro, P. (2006). Applying social psychology to the study of environmental concernand environmental worldviews: some contributions from social representa-tions approach. Journal of Community and Applied Social Psychology, 16, 247–266.

Castro, P., & Lima, L. (2001). Old and new ideas about the environment and science:an exploratory study. Environment and Behavior, 33, 400–423.

Castro, P., Reis, E., & Menezes, J. (2003). O reposicionamento do plastico [Plasticrepositioning]. Research report presented to Sociedade Ponto Verde.

Catton, W. R., & Dunlap, R. E. (1980). A new ecological paradigm for post-exuberantsociology. American Behavioral Scientist, 24, 15–47.

Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behavior:a review and avenues for further research. Journal of Applied Social Psychology,28, 1429–1464.

Conner, M., & McMillan, B. (1999). Interaction effects in the theory of plannedbehaviour: studying cannabis use. British Journal of Social Psychology, 38, 195–222.

Conner, M., Povey, R., Sparks, P., James, R., & Shepherd, R. (2003). Moderating role ofattitudinal ambivalence within the theory of planned behaviour. British Journalof Social Psychology, 42, 75–94.

Corral-Verdugo, V., Bechtel, R., & Fraijo, S. B. (2003). Environmental beliefs andwater conservation: an empirical study. Journal of Environmental Psychology, 23,247–257.

Corraliza, J. A., & Berenguer, J. (2000). Environmental values, beliefs and actions –a situational approach. Environment and Behavior, 32, 832–848.

Costarelli, S., & Colloca, P. (2004). The effects of attitudinal ambivalence on pro-environmental behavioural intentions. Journal of Environmental Psychology, 24,279–288.

Diekmann, A., & Franzen, A. (1999). The wealth of nations and environmentalconcern. Environment & Behavior, 31, 540–549.

Douglas, M., & Wildavsky, A. (1982). Risk and culture: An essay on the selection oftechnological and environmental dangers. Berkeley: University of California Press.

Dunlap, R. E., Gallup, G. H., Jr., & Gallup, A. M. (1993). Of global concern: results ofthe health of the planet survey. Environment, 35, 33–39.

Dunlap, R. E., & Mertig, A. G. (1995). Global concern for the environment: is afflu-ence a prerequisite? Journal of Social Issues, 51, 121–137.

Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). Measuringendorsement of the New Ecological Paradigm: a revised NEP scale. Journal ofSocial Issues, 56, 425–442.

Ellis, R. J., & Thompson, F. (1997). Culture and environment in the Pacific Northwest.American Political Science Review, 91, 885–897.

Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behaviour: An intro-duction to theory and research. Reading, MA: Addison-Wesley.

Furman, A. (1998). A note on environmental concern in a developing country:results from an Istambul survey. Environment and Behavior, 30, 520–534.

Gooch, G. D. (1995). Environmental beliefs and attitudes in Sweden and the BalticStates. Environment & Behavior, 27, 513–539.

Hayward, T. (1994). Ecological thought. Cambridge: Polity Press.Jonas, K., Bromer, P., & Diehl, M. (2000). Attitudinal ambivalence. In W. Stroebe, &

M. . Hewstone (Eds.), European Review of Social Psychology, Vol. 11 (pp. 35–74).Chichester, GB: Wiley.

Jonas, K., Diehl, M., & Broemer, P. (1997). Effects of attitudinal ambivalence oninformation processing and attitude–intention consistency. Journal of Experi-mental Social Psychology, 33, 190–210.

Kaiser, F. G., Oerke, B., & Bogner, F. X. (2007). Behaviour-based environmentalattitudes: development of an instrument for adolescents. Journal of Environ-mental Psychology, 27, 242–251.

Kaiser, F. G., Wolfing, S., & Fuhrer, U. (1999). Environmental attitude and ecologicalbehaviour. Journal of Environmental Psychology, 19, 1–19.

Kurz, T., Donaghue, N., Rapley, M., & Walker, I. (2005). The ways people talk aboutnatural resources: discursive strategies as barriers to environmentallysustainable practices. British Journal of Social Psychology, 44, 603–620.

Lima, M. L., & Castro, P. (2005). Cultural theory meets the community: worldviewsand local issues. Journal of Environmental Psychology, 25, 23–25.

Mannetti, L., Pierro, A., & Livi, S. (2004). Recycling: planned and self-expressivebehaviour. Journal of Environmental Psychology, 24, 227–236.

Milfont, T. L., & Duckitt, J. (2004). The structure of environmental attitudes: a first-and second-order confirmatory analysis. Journal of Environmental Psychology,24, 289–303.

Page 10: Ambivalence and conservation behaviour: An exploratory study on the recycling of metal cans

P. Castro et al. / Journal of Environmental Psychology 29 (2009) 24–33 33

Moscovici, S. (1988). Notes towards a description of social representations. Euro-pean Journal of Social Psychology, 18, 211–250.

Nordlund, A. M., & Garvill, J. (2002). Value structures behind proenvironmentalbehaviour. Environment and Behavior, 34, 740–756.

Perugini, M., & Bagozzi, R. P. (2001). The role of desires and anticipatedemotions in goal-directed behaviours: broadening and deepening thetheory of planned behaviour. British Journal of Social Psychology, 40, 79–98.

Priester, J. R., & Petty, R. E. (1996). The gradual threshold model of ambivalence:Relating the positive and negative bases of attitudes to subjective ambivalence.Journal of Personality and Social Psychology, 71, 431–449.

van der Pligt, J., Eiser, J. R., & Spears, R. (1986). Attitudes towards nuclear energy:familiarity and salience. Environment and Behavior, 18, 75–93.

Seguin, C., Pelletier, L. G., & Hunsley, J. (1998). Toward a model of environmentalactivism. Environment & Behavior, 30, 628–652.

Sociedade Ponto Verde. Quantidades Recicladas. [Recycled quantities]. http://www.pontoverde.pt/artigo_objecto.aspx?sid%3D58b81a33&ndash;74cb-4ca7-a967-3e8a3e1d6942&cntx%3DrMcP3cPOkJBl1ENxXTxCT0PaPFv1giWxMHIsJBDI8ok%3DRetrieved March, 2008.

Sparks, Hedderley, & Shepherd, R. (1992). An investigation into the relationshipbetween perceived control, attitude variability and the consumption of twocommon foods. European Journal of Social Psychology, 22, 55–71.

Sparks, P., & Shepherd, R. (1992). Self-identity and the theory of planned behavior:assessing the role of identification with ‘‘green consumerism’’. Social PsychologyQuarterly, 55, 388–399.

Steg, L., & Sievers, I. (2000). Cultural theory and individual perceptions of envi-ronmental risks. Environment and Behavior, 32, 250–269.

Stern, P. C. (2000). Toward a coherent theory of environmentally significantbehavior. Journal of Social Issues, 56, 407–424.

Stern, P. C., Dietz, T., & Guagnano, G. A. (1995). The new ecological paradigm insocial–psychological context. Environment & Behavior, 27, 723–743.

Terry, D. J., Hogg, M. A., & White, K. M. (1999). The theory of planned behaviour:self-identity, social identity and group norms. British Journal of SocialPsychology, 38, 225–244.

Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Let’s not be indifferent about(attitudinal) ambivalence. In R. E. Petty, & J. A. Krosnick (Eds.), Attitude strength:Antecedents and consequences (pp. 361–386). NJ: Lawrence Erlbaum.

Ullrich, J., Schermelleh-Engel, K., & Bottcher, B. (2008). The moderator effect thatwasn’t there: statistical problems in ambivalence research. Journal of Personalityand Social Psychology, 95, 774–794.

Uzzell, D. (2000). The psycho-spatial dimension of global environmental problems.Journal of Environmental Psychology, 20, 307–318.

Uzzell, D. (2003). Our uncommon future. In R. G. Mira, J. M. S. Cameselle, &J. R. Martınez (Eds.), Culture, environmental action and sustainability (pp. 21–39).Gottinger: Hogrefe & Huber.

Van Liere, K. D., & Dunlap, R. E. (1981). Environmental concern: does it makea difference how it’s measured? Environment & Behavior, 13, 651–676.

Verplanken, B., Myrbakk, V., & Rudi, E. (2005). The measurement of habit. InT. Betsch, & S. Haberstroh (Eds.), The routines of decision making (pp. 231–247).Mahwah, NJ: Lawrence Erlbaum.

Vining, J., & Ebreo, A. (2002). Emerging theoretical and methodological perspectiveson conservation behavior. In R. B. Bechtel, & A. Churchman (Eds.), Handbook ofenvironmental psychology (pp. 541–558). NY: John Wiley and Sons.

Wagner, W., & Hayes, N. (2005). Everyday discourse and common-sense – The theoryof social representation. Basingstoke: Palgrave-Macmillan Publishers.

Wegener, D. T., Downing, J., Krosnick, J. A., & Petty, R. E. (1995). Measures andmanipulations of strength related properties of attitudes – current practice andfuture directions. In R. E. Petty, & J. A. Krosnick (Eds.), Attitude strength: Ante-cedents and consequences (pp. 455–488). NJ: Lawrence Erlbaum.

Zanna, M. P., & Rempel, J. K. (1988). Attitudes: a new look at an old concept. InD. Bar-Tal, & A. W. Kruglanski (Eds.), The social psychology of knowledge (pp.315–334). Cambridge: Cambridge University Press.