Environmentally friendly consumer choices: Cultural differences in the self-regulatory function of anticipated pride and guilt

Please download to get full document.

View again

of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Environmentally friendly consumer choices: Cultural differences in the self-regulatory function of anticipated pride and guilt
Document Share
Document Transcript
  Environmentally friendly consumer choices: Cultural differences inthe self-regulatory function of anticipated pride and guilt Marleen C. Onwezen  a ,  b ,  * , Jos Bartels  c , Gerrit Antonides  b a LEI, Wageningen University and Research Centre, The Netherlands b Department of Social Sciences, Wageningen University, The Netherlands c Department of Communication Science, VU University Amsterdam, The Netherlands a r t i c l e i n f o  Article history: Available online 30 July 2014 Keywords: Anticipated emotionsSelf-conscious emotionsSelf-regulationPro-environmental choicesCollectivisticSelf-construal a b s t r a c t Anticipated self-conscious emotions, such as pride and guilt, help individuals to behave in line with theirpersonal and social standards regarding the environment. We seek to explore whether this self-regulatory role of anticipated pride and guilt functions similarly across individuals from different cul-tures ( N  ¼ 3854). We show that there are no differences across countries in the self-regulatory functionof anticipated pride and guilt  within  collectivistic and individualistic cultures but that there are differ-ences  between  collectivistic and individualistic cultures. For example, for individuals from individualisticcountries, anticipated emotions are more stronglyaffected byattitudes than theyare for individuals fromcollectivistic countries. The results provide a  󿬁 rst indication that the function of emotions is more socialin nature for individuals from collectivistic than individualistic cultures. These  󿬁 ndings imply that cul-tural differences in the function of emotions are associated with cultural differences in self-construal. ©  2014 Elsevier Ltd. All rights reserved. 1. Introduction The world is confronted with environmental issues such asclimate change, air and water pollution, and loss of biodiversity.Many of these problems are caused by human behaviour (DuNannWinter & Koger,2004;Gardner & Stern,2002)andthereforecanbemanaged by guiding consumer behaviour in a way that reducesenvironmental impact (Steg  &  Vlek, 2009).Recently is shown that anticipated pride and guilt guide in-dividuals to behave themselves in accordance with pro-environmental social norms and attitudes thus pointing to a self-regulatory function of these emotions (Onwezen, Bartels,  & Antonides, 2013). We aim to explore whether the self-regulatoryfunction of pride and guilt can be validated across a range of countries. We conduct a  󿬁 rst test of whether this self-regulatoryfunction differs between individuals from collectivistic and indi-vidualistic cultures. Based on previous studies (Mesquita, 2001;Tracy  &  Robins, 2007), we suggest that cultural differences in thefunction of emotions are primarily caused by differences in theconstrual of the self, i.e. regarding how consumers perceivethemselves in relation to others.Accordingly, we offer theoretical insights into the nature andfunctioning of self-conscious emotions, respond to the call forresearch on emotions in the context of environmentally friendlybehaviour (Kals  &  Maes; Vining  &  Ebreo, 2002), and providepractical insights into ways of stimulating pro-environmentalbehaviour cross-culturally. We focus on purchases of organicproducts because they are among others purchased out of concernfor the environment (Lockie, Lyons, Lawrence,  &  Mummery, 2002;Magnusson, Arvola, Hursti, Åberg,  &  Sj € od  en, 2003).We  󿬁 rst provide an overview of the literature on the self-regulatory function of anticipated pride and guilt and on the pro-posed differences between individualistic and collectivistic cul-tures. Next, we report on our study procedures and results, anddiscuss their implications. 2. Theoretical framework   2.1. Self-regulatory function of anticipated pride and guilt  Self-conscious emotions are evoked when people evaluate theirbehaviour with respect to a set of personal or social standards(Lewis, 1993; Tracy  &  Robins, 2004a). The current study focusesspeci 󿬁 cally on two self-conscious emotions, pride and guilt,because these emotions seem to be especially relevant in thecontextofpro-environmentalbehaviour. Pride isapositiveemotion *  Corresponding author. LEI, Wageningen University and Research Centre, TheNetherlands. E-mail address:  marleen.onwezen@wur.nl (M.C. Onwezen). Contents lists available at ScienceDirect  Journal of Environmental Psychology journal homepage: www.elsevier.com/locate/jep http://dx.doi.org/10.1016/j.jenvp.2014.07.0030272-4944/ ©  2014 Elsevier Ltd. All rights reserved.  Journal of Environmental Psychology 40 (2014) 239 e 248  that is experienced as a pleasant feeling and that often accom-panies feelings of self-worth (Rodriguez Mosquera, Manstead,  & Fischer, 2000; Tracy  &  Robins, 2007).  Guilt   is a negative emotionthat leads to feeling tense, remorseful, and worried (Baumeister,Stillwell,  &  Heatherton, 1994; Ferguson, Stegge,  &  Damhuis,1991). Pride (Lewis, 1993; Mascolo  &  Fischer, 1995; Tangney,1999; Tracy  &  Robins, 2004b) and guilt (Kugler  &  Jones, 1992;Tangney, Miller, Flicker,  &  Barlow, 1996) have some commoncharacteristics. Both arise when one feels responsible for an indi-vidual act and evaluates this act with respect to personal or socialstandards. Individuals tend to strive towards and maintain pride,and to avoid and get rid of guilt. Subsequently, these emotionsguide behaviour in accordance with personal and social standards.A limited body of recent research shows evidence for a self-regulatory function of self-conscious emotions (e.g., Hynie,MacDonald,  &  Marques, 2006; Su, Lu,  &  Lin, 2011), such that theyhelp to monitor and adapt one's own behaviour to re 󿬂 ect one'sstandards or goals (Carver  &  Scheier, 1998). Anticipated negativeself-conscious emotions mediate the relationship between atti-tudes and social norms on intentions in the context of condom use(Hynie et al., 2006) and in the context of textbook piracy (Su et al., 2011).In the context of environmentally friendly behaviour, previousstudies show that pride (Harth, Leach,  &  Kessler, 2013) and guilt(Arvolaetal.,2008;Bamberg,Hunecke, & Bl € obaum,2007;Bamberg &  M € oser, 2007; Carrus, Passafaro,  &  Bonnes, 2008; Ferguson  & Branscombe, 2010; Harth et al., 2013; Kaiser, Schultz, Berenguer,Corral-Verdugo,  &  Tankha, 2008; Verhoef, 2005) are associatedwith pro-environmental behaviour. In addition, Onwezen, Bartels,et al. (2013) recently explored the function of pride and guilt inthe context of the environment. They show evidence for a self-regulatory function of anticipated pride and guilt, such that theymediate the effects of personal norms, attitudes, and social normson environmentally friendly intentions. We therefore hypothesisethe following: Hypothesis 1 .  Anticipated pride and guilt regarding the environ-ment mediate the effects of attitudes and social norms on environ-mentally friendly purchase intentions .Fig.1 shows a graphical representation of the conceptual model.Next, we describe the proposed differences between individualisticand collectivistic cultures.  2.2. Similarities within individualistic and collectivistic cultures inthe self-regulatory function of anticipated pride and guilt  Studiesshowthattheproposedbasicmechanism(Hypothesis1)exists across a range of cultures such that the self-regulatoryfunction of guilt is found in both individualistic (i.e. Canada;Hynie et al., 2006) and collectivistic (i.e. Taiwan; Su et al., 2011) countries. Mesquita (2001), furthermore, states that althoughmultiple differences in emotions (e.g., concerns, appraisals andaction readiness) exist between individualistic and collectivisticcultures, individuals within individualistic and collectivistic cul-tures have comparable emotions and functions of emotions. Wetherefore hypothesize that: Hypothesis2 .  Themediatingeffectsofanticipatedprideandguiltonthe effects of attitudes and social norms on organic purchase in-tentions do not differ within individualistic or within collectivistic cultures .  2.3. Differences between individualistic and collectivistic cultures inthe self-regulatory function of anticipated pride and guilt  Although we hypothesized that anticipated pride and guilt havea self-regulatory-function in both individualistic and collectivisticcultures (Hypothesis 2), we propose that the strength of the asso-ciations differs  across  individualistic and collectivistic cultures.Several researchers suggest that a different construal of the self    how individuals see themselves in relation to others (Markus  & Kitayama, 1991) leads to different self-conscious emotions, asthese emotions rely on self-awareness and self-evaluations (Tracy &  Robins, 2004a; Tracy, Robins,  &  Tangney, 2007). Because in-dividuals from different cultures have a different construal of theself, self-conscious emotions are sensitive to cross-cultural differ-ences (Eid & Diener, 2001; Tangney & Fischer,1995).In cross-cultural psychology, two prototypical types of self-construal have been distinguished by several authors: the inde-pendent self and the interdependent self (e.g., Markus & Kitayama,1991). The independent self is emphasised more in individualisticcultures,whiletheinterdependentselfisrepresentedmoreoftenincollectivistic cultures (Cross, Hardin,  &  Gercek-Swing, 2011). Anindependent self is associated with individual goals, attributes,abilities, and preferences, independent from others. An interde-pendent self encompasses larger social groups (e.g., family, neigh-bourhood, or a sports team) and is associated with a self that isregulated by the emotions, thoughts, and actions of other people(Markus & Kitayama, 1991).Previous studies show that individuals from collectivistic andindividualisticculturesdifferintheirsensitivitytosocialnormsandattitudes. Collectivistic cultures are shown to be more inclined tofollow social norms and less inclined to follow attitudes in theirgreen purchasing behaviour compared to individualistic cultures(Chan & Lau, 2002). We believe that these differences occur due tocultural differences in the underlying emotional mechanism, suchthat collectivistic and individualistic cultures differ in the medi-ating effects of self-conscious emotions between the norm-intention and attitude-intention association. Additionally, webelieve that these cultural differences result from to differences inself-construal, as that we expect to  󿬁 nd differences betweencollectivistic and individualistic cultures, and not within thesecultures. Below we formulate speci 󿬁 c propositions based on pre-vious research  󿬁 ndings.Previous  󿬁 ndings have shown that self-construal plays animportantroleinpredicting thoughtsandbehavioursrelatedtotheenvironment (Arnocky, Stroink,  &  DeCicco, 2007; McCarty  & Shrum, 2001). However, research on cross-cultural differences intheself-regulatoryfunctionofanticipatedprideandguiltislacking.Previous studies do compare cultural differences in the effects of personal and social standards in evoking pride and guilt (i.e.  󿬁 rstpart of mediation effect) and in the effects of pride and guilt onintentions or behaviour (i.e. second part of mediation effect).Regarding the 󿬁 rst partof the mediation effect two studies indicatethat social norms have a larger effect on emotions in collectivisticcompared to individualistic cultures. Norm violations related to Fig. 1.  Proposed conceptual model. M.C. Onwezen et al. / Journal of Environmental Psychology 40 (2014) 239 e  248 240  social acceptance had stronger effects on guilt (and shame) forcollectivistic cultures than for individualistic cultures (Bierbrauer,1992; Ersoy, Born, Derous,  &  van der Molen, 2012). We proposethat individuals from collectivistic cultures, who are more likely todevelop an interdependent self which encompasses importantgroups and includes norms and goals of these groups, are moresensitive to social standards. On the other hand, individuals fromindividualistic cultures, who are more likely to develop an inde-pendent self which relates to uniqueness and individual goals, aremoresensitivetopersonalstandards. Wethereforehypothesisethefollowing: Hypothesis 3 .  Attitudes have a stronger, and social norms a weaker,impact on anticipated pride and guilt regarding the environment for individuals from individualistic cultures than for individuals fromcollectivistic cultures .Regarding the second part of the mediation effect  e  the in- 󿬂 uence of anticipated emotions on behavioural intentions  e several studies indicate that pride has a stronger effect onbehaviour in individualistic cultures, whereas guilt has a strongereffect in collectivistic cultures (Lee, Aaker,  &  Gardner, 2000;Markus  &  Kitayama, 1991). We distinguish two arguments sup-porting this proposition. First, ego-focused emotions such as pridetend to be associated with internal attributions, uniqueness andindividual awareness and therefore have a stronger impact onbehaviour within individualistic cultures. Other-focused emotionssuch as guilt tend to be associated with a social context or withrelying on others and therefore have a stronger impact onbehaviour within collectivistic cultures (Markus  &  Kitayama,1991).Second,peoplewithanindependentself-construalaregenerallymore promotion focused (reinforced by individual aspirations andwishes) and therefore more sensitive to positive emotions such aspride, while people with an interdependent self-construal aregenerally more prevention focused (reinforced by preventing theviolation of social norms) and therefore more sensitive to negativeemotions such as guilt (Lee et al., 2000). Kim and Johnson (2013) underscore this reasoning. They show that the in 󿬂 uence of prideon purchase intentions was stronger for US than for Korean par-ticipants, whereas the in 󿬂 uence of guilt was weaker for USthan forKorean participants.In the context of the environment these cultural differences inthe effects of pride and guilt are not yet found. Kaiser et al. (2008)show no cultural differences between collectivistic and individu-alistic countries in the effects of guilt (and embarrassment) on in-tentions. As such, it remains unclear whether previous  󿬁 ndings oncultural differences in the effects of pride and guilt on intentionsare applicable to the context of environmental behaviour. Wetherefore aim to test the following hypothesis: Hypothesis 4 .  Anticipated pride regarding the environment has astronger and anticipated guilt has a weaker impact on environmen-tally friendly purchase intentions for individuals from individualistic cultures than individuals from collectivistic cultures . 3. Study overview  We  󿬁 rst aim to validate the self-regulatory function of antici-pated pride and guilt (e.g., Onwezen, Bartels, et al., 2013) across arange of individualistic and collectivistic countries. We expect thattheeffectsofattitudesanddescriptivesocialnormsonintentionstobuy organic products are mediated by both anticipated pride andguilt (Hypothesis 1). Then we explore whether these effects arecomparable  within  individualistic countries, and  within  collectiv-istic countries (Hypothesis 2). Finally, we explore whether the self-regulatory function of pride and guilt, as expected, differs  across individualistic and collectivistic countries (Hypotheses 3 & 4). 4. Method 4.1. Participants Participants were selected online by a research agency inDecember 2010. The research agency selected representativesamples of respondents in terms of age, gender and educationallevel. In total, the sample consisted of 3854 respondents. We usedthe individualism dimension of Hofstede's framework (Hofstedecentre; Hofstede  &  Bond, 1984) to select individualistic andcollectivistic countries. For individualistic countries we selected:Australia (score  ¼  90;  n  ¼  507), Canada (score  ¼  80;  n  ¼  510),Germany (score  ¼  67;  n  ¼  514), the Netherlands (score  ¼  80; n ¼ 507),theUnitedKingdom(score ¼ 89; n ¼ 503),andtheUnitedStates (score ¼ 91;  n ¼ 507). The sample included 49.6% males and50.4% females, and the mean participant age was 46.6 years( SD  ¼  15.6). For collectivistic countries we selected: Malaysia(score  ¼  26;  n  ¼  403) and Singapore (score  ¼  20;  n  ¼  403). Thesample included 54.2% males and 45.8% females, and the mean agewas 37.4 years ( SD ¼ 11.9). 4.2. Measures All items were translated by the research agency into the nativelanguages of the selected countries: Dutch, German, and English(English is one of the most commonly spoken languages inMalaysia and Singapore). These translations were checked bynative speakers. 4.2.1. Attitudes Attitudes were measured following the theory of plannedbehaviour (Ajzen, 1991). Respondents answered four items indi-cating on a 5-point scale to what degree they associated thefollowing attributes with organic products fromtheirowncountry:Nice, Friendly, Reliable, and Successful (1  ¼  “ not at all ”  to5  ¼  “ completely ” ). Cronbach's alpha for these items was .94 forindividualistic and .91 for collectivistic countries. 4.2.2. Descriptive social norms An existing scale from the work of  Nolan, Schultz, Cialdini,Goldstein, and Griskevicius (2008) was used to measureperceived descriptive social norms around buying organic prod-ucts. The respondents rated three items (ranging from 1 ¼ “ never ” to 5  ¼  “ very often ” ) regarding the frequency with which theybelieve that their relatives/colleagues/average people from theircountry buy organic products. Cronbach's alpha for this scale was.72 for individualistic .69 for collectivistic countries. 4.2.3. Anticipated pride and guilt  Anticipated pride and guilt were each measured with threeitems selected from Holbrook  &  Batra's inventory (1987). To mea-sure anticipated pride, the respondents were asked to rate thefollowing items:  “ If I would behave in an environmentally friendlyway, then I feel: proud/worthy/extremely good. ”  To measureanticipated guilt, the respondents were asked to rate the followingitems:  “ If I would behave in an environmentally unfriendly way,then I: feel guilty/feel remorseful/have a bad conscience. ”  All itemswere rated on 5-point scales (ranging from 1  ¼  “ not at all ”  to5  ¼  “ very much ” ). Cronbach's alphas for the scales were high forboth individualistic ( a pride ¼  .86 and  a guilt ¼  .96) and collectivistic( a pride ¼ .88 and  a guilt ¼ .96) countries. M.C. Onwezen et al. / Journal of Environmental Psychology 40 (2014) 239 e  248  241  4.2.4. Intention to buy organic products The intention to buy organic products was measured followingthe theory of planned behaviour (Ajzen, 1991). The respondentsrated on a 5-point scale the likelihood of their buying organicproducts in the next two weeks (ranging from 1  ¼  “ extremelyunlikely ”  to 5 ¼ “ extremely likely ” ).The overall measurement model showed an adequate  󿬁 t forindividualistic (relative  c 2 ¼  441.0/59  ¼  7.5;  p  <  .001;RMSEA  ¼  .046; SRMR   ¼  .024; CFI  ¼  .987; TLI  ¼  .983) and collec-tivistic countries (relative  c 2 ¼  242.6/59  ¼  4.11;  p  <  .001;RMSEA ¼ .063; SRMR  ¼ .025; CFI ¼ .975; TLI ¼ .967). The 󿬁 t indicesare described in detail in the analysis section. 4.3. Analysis Mplus (version 6.11) was used to estimate structural regressionmodels with latent variables. The conceptual model was tested(Hypothesis 1) and used to explore differences within individual-istic, within collectivistic countries (Hypothesis 2), and acrossindividualistic and collectivistic countries (Hypotheses 3  &  4).Comparison across countries consisted of two main steps: prepar-ing the data by establishing measurement invariance across coun-tries and testing the structural regression model across countries. 4.3.1. Testing for measurement invariance across countries Differences in structural relations between groups only havemeaning when the constructs used are similar for all includedgroups. Establishing the invariance of the measurement model wastherefore an essential step in our analysis. Different forms of invariancehavebeenidenti 󿬁 edbypreviousresearch.Inthecurrentstudy, we only tested for con 󿬁 gural and metric invariance (and, forexample,notforscalarinvariance),becausethesearenecessaryandsuf  󿬁 cient conditions for testing differences in structural relations(Muth  en & Asparouhov, 2013; Vandenberg & Lance, 2000). Con  󿬁  gural invariance  is observed when a similar factor structureholds across the identi 󿬁 ed groups. We therefore tested the pro-posed measurement model for each of the groups separately. If themeasurement models show a good  󿬁 t for all identi 󿬁 ed groups,con 󿬁 gural invariance holds (Van de Schoot, Lugtig,  &  Hox, 2012).We used the following  󿬁 t indices to explore model  󿬁 t: relative  c 2 ,RMSEA, SRMR, CFI, and TLI. Metric invariance referstowhetherthemeaningof theconstructis equal across groups. Following previous studies (Van de Schootet al., 2012; Vandenberg  &  Lance, 2000), we estimated a baselinemodel (all parameters varied freely across groups) and a model inwhich factor loadings were assumed equal across groups. Tocompare these models, we discussed the critical values of  Cheungand Rensvold (1999; 2002). They concluded that changes in CFI( D  ¼  .01), Gamma Hat ( D  ¼  .001), and McDonald's NoncentralityIndex (NCI;  D  ¼  .02) provided the best performance in modelcomparison, in that these values are not in 󿬂 uenced by sample size,model complexity, and overall  󿬁 t measures of the baseline model.To make a  󿬁 nal decision, we used chi-square difference test, asrecommended by Vandenberg and Lance (2000) because the othertests only provide critical values and not a test of signi 󿬁 cant dif-ference across models. 4.3.2. Testing structural regression models across countries If measurement invariance is shown, wecan exploredifferencesin structural relations between countries. Based on the conceptualmodel two multi-group structural regression models with latentvariables wereestimated. First, all parameters in the structural partof the model were constrained to be equal across groups (universalmodel). Second, all parameters in the structural part were esti-mated freely (country-speci 󿬁 c model). Model comparisons withchi-squaredifference testswere used totestwhetherthe structuralrelationships between the latent variables differ across countries. 5. Results Table 1 shows means, standard deviations, and correlation co-ef  󿬁 cients for all of the included variables. All correlations aremoderate to strong and in the expected directions. 5.1. Part 1: comparing differences in the self-regulatory functionwithin individualistic and within collectivistic countries5.1.1. Testing for con  󿬁  gural invariance across individualistic countries 1 Con 󿬁 gural invariance was observed based on the acceptable  󿬁 tof the measurement models of all the included countries: Australia(relative  c 2 ¼  144.042/94  ¼  1.47;  p  <  .001; RMSEA  ¼  .048;SRMR  ¼ .039;CFI ¼ .980;TLI ¼ .975),Canada(relative c 2 ¼ 186.592/94  ¼  2.05;  p  <  .001; RMSEA  ¼  .063; SRMR   ¼  .052; CFI  ¼  .961;TLI  ¼  .950), Germany (relative  c 2 ¼  231.887/94  ¼  2.30;  p  <  .001;RMSEA  ¼  .061; SRMR   ¼  .037; CFI  ¼  .970; TLI  ¼  .962), theNetherlands (relative  c 2 ¼  147.198/94  ¼  1.59;  p  <  .001;RMSEA  ¼  .040; SRMR   ¼  .030; CFI  ¼  .987; TLI  ¼  .984), the UnitedKingdom(relative c 2 ¼ 215.262/94 ¼ 2.59;  p < .001;RMSEA ¼ .059;SRMR  ¼ .044;CFI ¼ .970;TLI ¼ .961),andtheUnitedStates(relative c 2 ¼  173.797/94  ¼  1.83;  p  <  .001; RMSEA  ¼  .050; SRMR   ¼  .041;CFI ¼ .978; TLI ¼ .972). 5.1.2. Testing for metric invariance across individualistic countries Chi-square difference tests (Table 2) showed that the measure-ment model was not fully invariant across countries, such thatModel 1 had a signi 󿬁 cantly lower  󿬁 t compared to the baselinemodel. A detailed analysis, however, revealed that the factorloadings were clearly signi 󿬁 cant for all countries (all in the ex-pected direction) and that the magnitude of the differences inparameter estimates between the countries was modest. The re-sults thus show neither perfect invariance nor evidence of com-plete inequality. This situation is termed partial measurementinvariance (Byrne, Shavelson, & Muth  en,1989). No explicit guidingrules decide which level of partial invariance is acceptable,although it is important that at least two latent variables are  󿬁 xedand the observed degree of invariance is reported with the results(e.g., Byrne et al., 1989).Toassesspartial metricinvariance,wechecked themodi 󿬁 cationindicesandstep-by-stepidenti 󿬁 editems thatdiffered substantiallyacross the sixcountries. Eventually, we estimated a model inwhichwe freed 5 out of 17 items (Table 2). This partially invariant modelhad an excellent 󿬁 t and did notdiffer signi 󿬁 cantlyin model 󿬁 t fromthe baseline model. The chi-square difference test seemed in thiscontextmorerigorousthanthecriticalvaluesofCFIandNCI,inthatthese values indicated that full metric invariance is in place.Gamma Hat on the other hand seemed more rigorous than chi-square difference test, as that this value rejects full and partialmetric invariance. Adopting the chi-square-difference test(Vandenberg  &  Lance, 2000), the partially invariant model wasused for further analyses. 1 Note that we assessed measurement invariance across individualistic countriesfor a measurement model which included a measurement of behaviour. Weincluded behaviour to ascertain that our conceptual model follows the reasoning of vested theories, such as the Theory of Planned Behaviour (Ajzen, 1991), positingthat intentions are the main predictors of behaviour. The results of the modelincluding behaviour are reported later in this paper. M.C. Onwezen et al. / Journal of Environmental Psychology 40 (2014) 239 e  248 242  5.1.3. Testing structural regression models across individualistic countries To assess whether the proposed structural relations differedacross individualistic countries, we estimated both a universalmodel and a country-speci 󿬁 c model. Chi-square difference testsshow that the country-speci 󿬁 c model (relative  c 2 ¼  934.543/458  ¼  2.04;  p  <  .001; RMSEA  ¼  .057; SRMR   ¼  .082; CFI  ¼  .976;TLI  ¼  .976) and the universal model (relative  c 2 ¼  964.484/498  ¼  1.94;  p  <  .001; RMSEA  ¼  .054; SRMR   ¼  .084; CFI  ¼  .977;TLI  ¼  .974) were not signi 󿬁 cantly different in model  󿬁 t ( D c 2 (40)  ¼  29.941;  p  ¼  n.s.). These results con 󿬁 rm Hypothesis 2, indi-cating that the structural relations do not differ signi 󿬁 cantly acrossindividualistic countries. 5.1.4. Testing for con  󿬁  gural invariance across collectivistic countries The results revealed a model with an acceptable  󿬁 t for in-dividuals from Singapore (relative  c 2 ¼  2.78;  p  <  .001;RMSEA  ¼  .067; SRMR   ¼  .024; CFI  ¼  .973; TLI  ¼  .965) and fromMalaysia(relative c 2 ¼ 2.33;  p < .001;RMSEA ¼ .058;SRMR  ¼ .033;CFI ¼ .977; TLI ¼ .970). These results indicate con 󿬁 gural invariance,such that the measurement model was considered valid in bothgroups. 5.1.5. Testing for metric invariance across collectivistic countries Table 3 shows that the measurement model was not fullyinvariant across the groups: Models 1, 2, and 3 had a signi 󿬁 cantlylower  󿬁 t compared to the baseline model. Subsequently, weaimed to assess partial metric invariance. We used the samestep-by-step procedure as for individualistic countries andeventually estimated a model in which we freed 2 out of 17items. This model had an excellent  󿬁 t and did not differ signif-icantly in model  󿬁 t from the baseline model, which indicatedpartial metric invariance. The results show again that the criticalvalues of NCI and CFI were less rigorous, in that these indicatedfull metric invariance, whereas Gamma Hat was more rigorousthan the chi-square difference test in that this test indicated nofull or partial metric invariance. We adopted the chi-squaredifference tests and used the partially invariant model forfurther analyses. 5.1.6. Testing structural regression models across collectivistic countries To explore whether there were differences in the structural re-lations between individuals from the two collectivistic countries,we estimated both a country-speci 󿬁 c and a universal model. Thegroup-speci 󿬁 c model (relative  c 2 ¼ 2.43;  p  <  .001; RMSEA ¼ .060;SRMR   ¼  .050; CFI  ¼  .973; TLI  ¼  .966) and the universal model(relative  c 2 ¼  2.33;  p  <  .001; RMSEA  ¼  .058; SRMR   ¼  .048;CFI ¼ .973; TLI ¼ .969) showed no signi 󿬁 cant differences in model 󿬁 t ( D c 2 (8) ¼  4.471;  p  ¼  n.s.). These results con 󿬁 rm Hypothesis 2:the self-regulatory function of anticipated pride and guilt did notdiffer across the two collectivistic countries. 5.2. Part 2: comparing differences in the self-regulatory functionbetween individualistic and collectivistic countries In this phase, the self-regulatory function of anticipated prideand guilt was compared across individualistic (Australia, Canada,Germany, the Netherlands, UK, and US) and collectivistic countries(Malaysia and Singapore).  Table 1 Means, standard deviations and correlations for all included variables within individualistic ( N  ¼ 3048; lower triangular matrix) and collectivistic ( N  ¼ 806; upper triangularmatrix) countries.Individualistic countries Collectivistic countries 1. 2. 3. 4. 5. M SD M SD 1 Anticipated pride 3.06 1.05 3.81 0.89  e  .431** .398** .305** .458**2 Anticipated guilt 2.85 1.18 3.40 1.09 .509** a e  .288** .237** .345**3 Attitude 3.20 0.92 3.23 0.82 .506** .370**  e  .455** .479**4 Descriptive social norm 2.43 0.67 2.73 0.71 .286** .222** .386**  e  .435**5 Intention 2.62 1.35 3.20 1.18 .433** .371** .558** .416**  e 6 Purchase of products 2.42 1.00  e e  .429** .320** .469** .429** .639** a **  p  <  .001;  M  ¼ mean;  SD ¼ standard deviation.  Table 2 Fit indices for measurement invariance across six individualistic countries. c 2 df   RMSEA; SRMR; CFI; TLI; GH; NCIBaseline model 1098.779 564 RMSEA ¼ .054; SRMR  ¼ .043;CFI ¼ .975; TLI ¼ .968;GH ¼ .829; NCI ¼ .626Model 1: metricinvariance1279.581 644 RMSEA ¼ .055; SRMR  ¼ .081;CFI ¼ .970; TLI ¼ .967;GH ¼ .804; NCI ¼ .627 D c 2 ¼ 180.802;  p  <  .00180  D GH ¼ .025; D NCI ¼ .001;  D CFI ¼ .005Model 1.1 a 1186.205 619 RMSEA ¼ .054;SRMR  ¼ .080; CFI ¼ .970;TLI ¼ .966; GH ¼ .821;NCI ¼ .626 D c 2 ¼ 87.426;  p ¼ n.s.55  D GH ¼ .007;  D NCI ¼ .000; D CFI ¼ .005 a This is a partially invariant model in which the item loadings for  “ I feel supe-rior ” ;  “ How often did you buy organic food/organic cotton the last two weeks ” ;  “ Iassociate organic products with being successful ” ; and  “ How often do you think theaverage person from your country buys organic? ”  were free to vary;  df  ¼ degrees of freedom.  Table 3 Fit indices for measurement invariance across two collectivistic countries. c 2 df   RMSEA; SRMR; CFI; TLI; GH; NCI; D GH;  D NCI;  D CFIBaseline model 301.255 118 RMSEA ¼ .063; SRMR  ¼ .031;CFI ¼ .975; TLI ¼ .967;GH  ¼ .934; NCI ¼ .623.Model 1: metricinvariance347.856 131 RMSEA ¼ .065; SRMR  ¼ .056;CFI ¼ .971; TLI ¼ .965;GH  ¼ .922; NCI ¼ .623 . D c 2 ¼ 46.601;  p  <  .00117  D GH ¼ .012;  D NCI ¼ .000; D CFI ¼ .004Model 1.1 a 329.749 129 RMSEA ¼ .063; SRMR  ¼ .050;CFI ¼ .973; TLI ¼ .967;GH  ¼ .927; NCI ¼ .623 . D c 2 ¼ 28.494;  p ¼ n.s.11  D GH ¼ .007;  D NCI ¼ .000; D CFI ¼ .002 a This is a partially invariant model in which the item loadings for  “ I feel worthy ” and  “ How often do you think the average person from your country buys organic? ” were free to vary;  df  ¼ degrees of freedom. M.C. Onwezen et al. / Journal of Environmental Psychology 40 (2014) 239 e  248  243
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks