The Impact of Individual and Group Feedback on Environmental Intentions and Self-Beliefs

Method

Results

Through funneled debriefing, 18 participants expressed varying degrees of suspicion about the verity of the feedback or the purpose of the study, although most did so only after being informed about its purpose. Chi-square tests showed that the frequency of suspicion did not differ as a function of condition, all p values > .05. MANOVA revealed that these 18 participants did not significantly differ in any way from the remaining participants, all p values > .05, and hence all 90 participants were included in subsequent analyses.

On the modified version of the Inclusion of Other in the Self Scale (Aron et al., 1992), the mean response was 4.41 (SD = 1.21), indicating that the average participant thought that the relationship between himself or herself and students at the university was best described as moderately connected. There was no correlation between scores on the Inclusion of Other in the Self Scale and any of the dependent measures, all p values > .05.

Environmental Attitudes and Intentions

The first analysis tested whether the patterns of individual and group feedback directly affected participants’ attitudes toward the environment. Participants’ ratings of the degree to which they agreed with the 13 environmental attitude statements were averaged, after reverse scoring negatively worded items; Cronbach’s alpha coefficient was .86. (See Table 1 for means and correlations of all dependent measures). This average attitude score was analyzed with a 2 × 2 ANCOVA, with responses to the EIQ (which assessed participants’ pretest environmental behaviors) as a covariate. No significant effects emerged, all ps > .1. Thus, the manipulations did not affect participants’ environmental attitudes. But did they influence participants’ intentions of engaging in proenvironmental behavior in the future?

OPEN IN VIEWER

Table 1.

Means, Standard Deviations, and Intercorrelations of Dependent Measures.

	M	SD	1	2	3	4
1. Environmental attitudes	7.50	1.56	—	.67**	.29**	−.18*
2. Environmental behaviors	8.07	1.87		—	.39**	−.32**
3. Emotions	5.08	2.84			—	−.58**
4. Self-ratings	5.37	1.61				—

*

p < .05. **p < .01.

Participants’ scores on the 10 items that measured the likelihood of performing environmental behaviors were averaged, after reverse scoring negatively worded items; Cronbach’s alpha coefficient was .88. A 2 × 2 ANCOVA (with EIQ scores as a covariate) revealed a significant two-way interaction, F(1, 85) = 4.01, p < .05, η² = .03. Neither the main effect of individual feedback nor group feedback was significant, all ps >.05. Tests of simple effects showed a significant simple main effect of group feedback when individual feedback was highly negative such that participants who received moderately negative group feedback (n = 22, M = 8.58, SD = 0.32) were more likely to intend to perform environmental behaviors than those who received highly negative group feedback (n = 23, M = 7.70, SD = 0.31), F(1, 85) = 3.89, p = .05. No significant simple effects emerged when individual feedback was moderately negative between moderately negative group feedback (n = 22, M = 7.81, SD = 0.32) and highly negative group feedback (n = 23, M = 8.19, SD = 0.31), nor were there any simple effects within the group feedback levels.

We next explored whether certain behavioral intentions were more affected by the feedback manipulation than others. A principal-axis factor analysis was conducted on the 10 items that measured the likelihood of performing environmental behaviors. (Two participants had missing environmental intention data and so their responses were excluded from the factor analysis.) Based on an examination of the scree plot of eigenvalues, two factors were retained and rotated to an oblique (direct oblimin) solution. The two factors reflected two levels of engagement in environmental behaviors—low and highly committed environmentalism (eigenvalues = 4.96 and 1.16, respectively). The low-commitment factor involved conserving energy and water, recycling, and limiting driving when possible. The high-commitment factor involved more deeply engaged behaviors such as dietary restrictions, active research into businesses’ ecological practices, and dissemination of environmental knowledge. (See Table 2 for unstandardized means of items found in each factor.)

OPEN IN VIEWER

Table 2.

Means and Standard Deviations of Low- and High-Commitment Environmental Behaviors by Condition.

Group feedback
	Moderately negative	Highly negative
Low-commitment behaviors
Individual feedback
Moderately negative	8.98 (.34)	9.18 (.33)
Highly negative	9.33 (.34)	8.59 (.33)
High-commitment behaviors
Individual feedback
Moderately negative	6.06 (.41)	6.70 (.40)
Highly negative	7.45 (.41)	6.37 (.40)

Note: Means are the averages for items that loaded highly on the low- and high-commitment factors, adjusted for mean scores on the Environmental Impact Questionnaire.

The two factor scores were analyzed with a 2 × 2 MANOVA, with responses to the EIQ as a covariate. The analysis revealed a significant multivariate interaction, mF(2, 82) = 3.56, p < .05, η² = .08. At the univariate level, a significant interaction was obtained for high-commitment behavior intentions, F(1, 83) = 7.19, p < .01, η² = .05, but not for low-commitment behavior intentions, F(1, 83) = 3.16, p = .08, η² = .03. These interactions for the low- and high-commitment behaviors are shown in Figure 1. Neither the main effect of individual feedback nor the main effect of group feedback reached significance at either the multivariate or univariate level, all ps > .05.

OPEN IN VIEWER

Figure 1.

Intentions to perform low- and high-commitment environmental behavior as a function of individual and group feedback.

Tests of simple effects revealed two significant pairwise differences between means on the standardized factor scores for the high-commitment factor. First, participants who received moderately negative group feedback but highly negative individual feedback (n = 22, M = 0.32, SD = 0.15) were more likely to intend to perform high-commitment environmental behaviors than those who received both individual and group feedback that was moderately negative (n = 22, M = −0.23, SD = 0.15), F(1, 83) = 6.40, p = .05. Second, among participants who received highly negative individual feedback, receiving moderately negative group feedback (n = 22, M = 0.32, SD = 0.15) resulted in greater high-commitment intentions than receiving highly negative group feedback (n = 22, M = −0.18, SD = 0.15), F(1, 83) = 5.25, p < .05. No significant simple effects were found for those receiving moderately negative individual feedback between the moderately negative group feedback condition (n = 22, M = −0.23, SD = 0.15) and the highly negative group feedback condition (n = 22, M = 0.90, SD = .15). Overall, these patterns show that participants expressed the highest intentions to behave in environmentally friendly ways when their personal feedback was worse than the group feedback.

Discussion

The primary goal of this study was to examine the effects of feedback about one’s own and one’s group’s environmental impact on people’s environmental intentions and self-ratings. Participants reported on their current environmental behaviors and received false feedback regarding their personal impact on the environment that indicated that their actions were either moderately negative or highly negative, as well as information about the impact of their group.

Participants’ self-ratings and intentions to behave in proenvironmental ways were influenced by the feedback that they received, although not always in the predicted pattern. Individual and group feedback interacted to influence environmental intentions as anticipated, but the direction of the influence on feedback did not fully replicate the boomerang effect found in past research. Furthermore, the effect of feedback on behavior was not mediated by emotions or self-evaluations, as had been anticipated.

Overall, participants expressed the greatest proenvironment intentions when their individual feedback was worse than that of the group average, although the effects of feedback on behavioral intentions varied as a function of the commitment needed for the behaviors; the effect of feedback was significant only for intentions to perform behaviors that required a great deal of personal commitment. The nonsignificant findings for low-commitment behaviors suggested that participants were least likely to endorse proenvironmental intentions when both their individual and group feedback were highly negative. However, when asked about their intentions to enact high-commitment behaviors, participants were least likely to endorse proenvironmental intentions when they and their group received moderately negative feedback. Despite this difference, both low- and high-commitment behaviors showed the same general pattern (see Figure 1). Furthermore, for both low- and high-commitment behaviors, the lowest behavioral intentions occurred when participants thought that their carbon impact matched that of their group.

These findings complement that of research on social identity and social norms in the environmental domain that have demonstrated the power of reference groups. Research based on social identity and social categorization theories has highlighted the differential effects of social comparisons to in-groups versus out-groups (Smith & Louis, 2008, 2009). When comparing themselves with out-group members, people tend to contrast their environmental attitudes and behaviors with those of the out-group. For example, Rabinovich, Morton, Postmes, and Verplanken (in press) found that when people were asked to compare their in-group with out-groups that were more or less environmentally friendly, participants’ environmental attitudes and behaviors moved in the opposite direction to those of the out-group. When presented with a more eco-friendly out-group, participants were less environmentally friendly themselves, and vice versa.

Our findings extend the out-group contrast effects obtained by Rabinovich et al. (in press) by showing that participants conformed to the ostensible environmental behavior of their in-group, which is also consistent with social identity research. Although people contrast their behaviors from out-group members, they tend to conform their behaviors to in-group members. As Haslam, Ellemers, Reicher, Reynolds, and Schmitt (2010) explained,

it is precisely through individuals’ identification of, and conformity to, norms that are perceived to be shared with others in a particular context that their potentially idiosyncratic views become socially organized and consensual. Through this process, individual views are coordinated and transformed into shared values, beliefs, and behaviors. (p. 350)

The individual and group feedback given to participants in our study encouraged participants to compare themselves with members of an important in-group. The effect of this approach closely mirrored that found in social norm research on in-group comparisons.

In-group membership status was made salient in this study by explicitly mentioning the average student at the participants’ university. Research has shown that the salience of the in-group is an important aspect of in-group comparisons, and successful social comparison campaigns have often made this comparison group clear (Smith & Louis, 2009; Wellen, Hogg, & Terry, 1998). For example, Goldstein et al. (2008) found that telling people about the positive actions of community members (e.g., reusing hotel-room towels) made hotel guests more likely to conserve. Although those studies did not provide individual feedback, presumably hotel guests were familiar with their own towel-reuse habits. For those who did not normally conserve—roughly 65% based on the results for the control group—learning that most other hotel guests reuse their towels would have essentially resulted in feedback that was worse than the group average (Goldstein et al., 2008).

Yet, Cialdini and his colleagues also found ironic increases in littering and destruction of national parks when participants were led to believe that other people acted in environmentally damaging ways (Cialdini, 2003; Cialdini, Reno, & Kallgren, 1990). Similarly, our participants showed a lower intention to behave environmentally when they believed that their group’s environmental impact was as negative as their own. Thus, participants appeared motivated to change their behavior only when they believed they were less environmentally conscious than their peers. We suspect that most of our participants had not previously given much thought to how their environmental actions might compare with other students’, so the feedback offered them a new perspective on how environmentally friendly they were compared with other people. Presumably, participants whose individual feedback mirrored the group feedback had little comparative reason to change their behavior.

Interestingly, this study did not show the undesirable boomerang effect found in past social norm studies, in which people who receive better-than-group feedback decrease their proenvironmental behaviors to match the descriptive social norm (Schultz et al., 2007). Participants whose individual feedback was better than the group average were no less likely to endorse proenvironmental behavioral intentions than those whose feedback matched the group. This effect could be due to two things. First, participants in this study indicated behavior intentions rather than actually performing the behavior. Thus, the detriment to proenvironmental behavior may not occur until after intentions have been made and actual behavior begins. Alternatively, participants may have implicitly perceived an injunctive proenvironmental norm indicating that they should be environmentally friendly. Schultz and colleagues (2007) found that providing an injunctive proenvironmental norm buffered against the detrimental effects of receiving better-than-group feedback. Although we did not provide smiley faces or other forms of explicit injunctive norms, participants may have been primed by the study materials to think that they should behave more proenvironmentally.

In addition to providing information about the environmental behavior of participants’ in-group, our study included data about the environmental behavior of the wider world’s population. Although most people know that those in developed countries (including their in-group) disproportionally contribute to environmental damage, they differ in their assumptions about the extent of this difference. By controlling for participants’ beliefs about global environmental norms in this way, we could examine the effects of in-group behavior within the constraints of knowledge about the world’s actual behavior.

Other studies have noted the role of negative affect, particularly guilt and embarrassment, in engendering a sense of responsibility for the environment that can lead to proenvironmental behaviors (Carrus et al., 2008; Kaiser & Shimoda, 1999; Kaiser et al., 2008). Although participants in the present study rated themselves more negatively when provided with feedback indicating that they had a highly rather than moderately negative impact on the environment, their emotions (guilt, shame, embarrassment, and remorse) were not influenced by the feedback. This finding supports the notion that social norms do not require a great deal of internal reflection or self-judgment to be effective if they are used as informational guides or heuristics (Cialdini et al., 1990). Participants in the present study who were told that their behavior went against that of their group may have used that new information as a behavioral guide rather than an indictment. Thus, their behavior change can be seen as a rational decision based on new information rather than on feelings of guilt or shame.

However, despite the fact that the feedback did not affect moral emotions, participants did incorporate it into their self-ratings. Not surprisingly, participants felt most positively about themselves when they received feedback that was better than that of their group. However, the participants who felt the most immoral, irresponsible, worthless, bad, and unethical were also the least likely to intend to change their behaviors. This pattern was obtained when both group and individual feedback was negative, suggesting that although participants internalized the negative feedback to some degree, their negative self-images did not motivate them to improve their behaviors.

This finding adds credence to research showing social motivations for prosocial actions (Bamberg & Möser, 2007). Participants were seemingly not motivated by internal feelings of guilt or shame, and the experimental manipulations did not change their environmental attitudes. Rather, their intentions to behave differently were affected by comparisons between themselves and an important reference group, suggesting that the desire to fit in with their group played a role in spurring environmental action (Cialdini, 2003; Cialdini et al., 1990; Goldstein et al., 2008). This lack of emotional reaction to social comparison feedback also helps to explain how normative social influence often goes underdetected (Nolan et al., 2008). If social comparison feedback induced stronger emotional responses, then people might more easily detect the influence of this feedback. In the absence of strong emotions, they are left to infer that their decisions are influenced by other factors (Bem, 1967).

On the surface, the finding that people who reported stronger intentions to perform high-commitment behaviors did not report feeling more moral than others might seem at odds with research on moral licensing. Moral licensing occurs when people use their previous positive actions to buffer against the guilt that may arise from subsequent undesirable behavior (Mazar & Zhong, 2010). In this study, participants’ self-ratings of morality, goodness, ethicality, responsibility, and value were in line with their perceived environmental harm—just as with moral licensing—yet their behavioral intentions were not. One possible explanation for this difference is that participants completed self-ratings and reported their intentions to perform good behaviors following negative feedback, whereas in moral licensing, self-ratings and undesirable behaviors follow positive feedback. Given that indulging in bad behavior is often less effortful and requires less self-control than enacting new good behavior (Fishbach & Trope, 2008), participants in the present study may have faced a steeper challenge behaving in accordance with self-ratings than do participants in moral licensing studies.

Our finding that high-commitment behaviors were affected more strongly by feedback than low-commitment behaviors suggests that behaviors that require strong personal commitment are particularly amenable to social comparisons. Perhaps high-commitment actions (such as restricted diets, research into corporate ecological practices, and discussion of environmental knowledge) are less common and familiar to student participants than low-commitment behaviors such as turning off lights, driving less, and recycling. Participants may have thus relied on social comparison with the group even more when making decisions about their intentions regarding these high-commitment behaviors. Supporting this explanation, the means for low-commitment behavior intentions were higher than those for high-commitment behaviors in all conditions. Thus, although people are more likely to perform easy and common behaviors in general, they are more influenced to adjust rare behaviors according to social norms.

Although the moderately negative and highly negative feedback were designed to correspond to the environmental impact of the average American and more negative impact than the average American, respectively, participants probably perceived moderately negative feedback fairly negatively. Given that the average American uses resources at a rate that is more than three times greater than what is needed to sustain the earth, moderately negative feedback is only comparatively less negative than blatantly highly negative feedback. Therefore, it is important to determine how negatively people perceive feedback that indicates that they are damaging the earth at about the same rate as most other people in their reference group.

The primary conclusion from this study is that people may be more motivated to behave in environmentally friendly ways when they believe that their current environmental impact differs from that of an important reference group. As noted, other research has shown the power of group norms in affecting environmental behavior (see Cialdini, 2003), but the current study extends that finding beyond particular environmental behaviors to beliefs regarding the general impact that one has on the environment compared with the impact of one’s reference group. The current research also shows that attitudes, emotions, and self-evaluations do not mediate these effects, something that could not be tested in previous field studies.

Although these findings are promising, they have some limitations. First, the study measured behavioral intentions rather than actual behaviors. This was done purposefully to capture a wide range of behavioral domains, but it comes at a possible cost of ecological validity. Behavioral intentions have only a moderate relationship with actual behavior, and so we cannot be sure that people’s intentions in these studies translated into behavior change (Hornsey et al., 2007; Sheeran, 2002). We opted to focus on intentions in the current study but recommend that researchers test these effects while measuring actual behaviors.

Second, our results may have been underpowered given our sample size. Nevertheless, significant interactions were obtained, leading us to believe that a larger sample size would not change the overall results of this study. The mediation analyses did not even approach statistical significance, so it seems unlikely that a larger sample would have found mediation effects.

These results suggest that carbon footprint calculators could be designed to provide feedback in ways that highlight differences between people’s personal impact and the impact of important reference groups. Apparently, simply knowing that one uses five times more of the earth’s resources than one equitably should use is not always sufficient to change intentions. Feedback that compares the user’s carbon footprint to particular reference groups might set processes in motion that increase people’s intentions to behave proenvironmentally.