<<

Pride, , and Pro-Environmental Behavior: The Role of Experienced Self-Conscious

Emotions in an Individual’s Response to Carbon Footprint Feedback

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in

the Graduate School of The Ohio State University

By

Ian J. Adams

Graduate Program in Environment & Natural Resources

The Ohio State University

2019

Thesis Committee

Dr. Nicole Sintov, Advisor

Dr. Alia Dietsch

Dr. Robyn Wilson

1

Copyrighted by

Ian Jae Adams

2019

2

Abstract

Feedback on one’s consumption, for instance, via a carbon footprint calculator, is a common strategy used in attempts to promote pro-environmental action. Although evidence suggests that feedback can be effective in reducing consumption, little research examines the psychological processes that occur between feedback reception and subsequent behavior. To illuminate this gap, the present study uses attribution and identity theories to examine the cognitive and emotional processes that occur upon receipt of such feedback. Three hundred ninety-seven adults received a positively-, neutrally-, or negatively-framed bogus feedback message from a source high or low in credibility. Feedback frame did not impact pro-environmental behavior directly, but instead impacted the extent to which participants made internal attributions for their feedback and their pride and guilt responses. Guilt, but not pride, mediated the relationship between feedback and pro-environmental behavior. Further, this research found that the indirect effect of feedback frame on pro-environmental behavior through guilt is stronger for participants with higher levels of environmentalist identity. Our findings provide a better understanding of the circumstances by which pride and guilt lead to pro-environmental action, and recommendations for feedback designers and practitioners are detailed.

i

Acknowledgements

My sincere thanks to my advisor, Dr. Nicole Sintov, for her thoughtful discussion, encouragement, and advice that made this research possible. Additionally, I would like to express my gratitude to my partner, John, for his continued support and keeping me going throughout this process.

ii

Vita

2013…………………………………………Rocky River High School

2017…………………………………………B.S. Marketing, The Ohio State University

2017 – 2018…………………………………Graduate Fellow, School of Environment and Natural Resources, The Ohio State University

2018 – Present………………………………Graduate Research Associate, School of Environment and Natural Resources, The Ohio State University

Fields of Study

Major Field: Environment and Natural Resources

iii

Table of Contents

Abstract ...... i

Acknowledgements ...... ii

List of Tables ...... vii

List of Figures ...... viii

Chapter 1: Introduction ...... 1

Chapter 2. The Influence of Carbon Footprint Feedback Framing on Cognitive and

Emotional Pathways to Pro-environmental Behavior ...... 8

Abstract ...... 8

Introduction ...... 10

Background ...... 10

Attribution Theory and Self-Conscious Emotions ...... 11

Self-Conscious Emotions and Pro-Environmental Behavior ...... 14

Contributions ...... 16

Overview of Research and Hypotheses ...... 18

Methods ...... 19 iv

Participants ...... 19

Procedures ...... 22

Measures ...... 26

Manipulation Check ...... 33

Dropout Analyses ...... 34

Results ...... 36

Discussion and Conclusions ...... 45

Limitations ...... 49

Future Directions ...... 51

Chapter 3. The Influence of Environmentalist Identity on Emotional Response to Carbon

Footprint Feedback ...... 53

Abstract ...... 53

Introduction ...... 55

Background ...... 55

Process Model of Self-Conscious Emotions ...... 56

Identity Theory ...... 57

Research Context ...... 62

Contributions ...... 62

Hypotheses ...... 63

v

Methods ...... 65

Results ...... 66

Discussion and Conclusions ...... 76

Limitations ...... 80

Future Directions ...... 82

Chapter 4. Overview and Conclusions ...... 83

References ...... 88

Appendix A: Experimental Instrument ...... 102

Appendix B: Chapter 2 Supplemental Information ...... 121

Appendix C: Chapter 3 Supplemental Information ...... 126

vi

List of Tables

Table 1. Sample vs. U.S. Demographic Characteristics...... 21

Table 2. Promax Rotated Factor Loadings for Causal Attributions Scale Items...... 28

Table 3. Descriptive Statistics for Key Chapter 2 Outcome Variables...... 31

Table 4. Manipulation Check Results...... 34

Table 5. Summary of Chapter 2 Hypothesis Testing...... 36

Table 6. Mean Internal Attribution and Standard Deviations by Experimental Condition.

...... 39

Table 7. Summary of Path Coefficients, Indirect Effects, Direct Effects, and Total

Effects...... 42

Table 8. Descriptive Statistics for Key Chapter 3 Variables...... 66

Table 9. Summary of Chapter 3 Hypothesis Testing...... 66

Table 10. OLS Regressions with Bootstrapping on Pride and Guilt (Mediators) and Pro- environmental Behavior (Dependent Variable) for Moderated-Mediation Analysis...... 70

Table 11. Summary of ANCOVA Results (Dependent Variable = Internal Attribution).

...... 124

Table 12. Summary of Path Coefficients, Indirect Effects, Direct Effects, and Total

Effects (Negative Feedback Frame Reference Group)...... 125

vii

List of Figures

Figure 1. Attribution Theory (Weiner, 1986)...... 13

Figure 2. Study 1 Concept Map...... 16

Figure 3. Procedures...... 23

Figure 4. Path Coefficients...... 41

Figure 5. Study 2 Concept Map...... 68

Figure 6. Pride as a Function of Environmentalist Identity for Positive vs. Negative

Feedback...... 72

Figure 7. Guilt as a Function of Environmentalist Identity for Positive vs. Negative

Feedback...... 73

Figure 8. Direct and Indirect Effects of Feedback Frame on Pro-Environmental Behavior at Values of Environmentalist Identity...... 75

Figure 9. Scree Plot for Causal Attribution Items...... 121

Figure 10. Pride Distribution...... 122

Figure 11. Guilt Distribution...... 123

viii

Chapter 1: Introduction

Individual-level consumption is a major contributor to greenhouse gas emissions.

Specifically, scholars estimate that households are responsible for roughly 30 to 40 percent of United States (U.S.) greenhouse gas emissions (Vandenbergh, Barkenbus, &

Gilligan, 2008). Changing everyday behaviors can play a significant role in reducing one’s environmental impact. There are numerous ways to reduce one’s carbon footprint by engaging in environmentally-friendly, pro-environmental behaviors (PEBs). For example, researchers project that increasing energy-reducing household thermostat adjustments (e.g., higher temperature settings in warm seasons and lower settings in cooler seasons) and increasing carpooling have the potential to eliminate 4.5 and 6.4 millions of metric tons of carbon per year, respectively, from total US emissions (Dietz,

Gardner, Gilligan, Stern, & Vandenbergh, 2009). The present research focuses on recurring PEBs (Brick, Sherman, & Kim, 2017), which are behaviors that one must do over and over to have a meaningful impact, rather than one-shot behaviors. For example, other recurring PEBs include eating local and vegetarian foods, walking, biking, or using mass transit rather than driving a personal vehicle, and turning off electronics when they are not in use. These behaviors, especially when combined, can greatly reduce one’s carbon footprint (Vandenbergh et al., 2008).

1

Feedback, the process of giving people information regarding their past behavior (Karlin,

Zinger, & Ford, 2015), can be an effective intervention tool for reducing individual-level consumption and promoting PEB (Abrahamse, Steg, Vlek, & Rothengatter, 2005;

Schultz, 2014). However, little is known about the psychological processes by which receiving a feedback message translates to engaging in pro-environmental action, as much existing research only 1) implements a feedback intervention and then 2) measures a behavioral dependent variable. Understanding the cognitive and emotional processes by which feedback translates to pro-environmental action is especially critical given the advancement and proliferation of consumption monitoring technologies, which can output their data into feedback. As of 2016, smart meters, which facilitate near real-time data collection and thereby delivery of near real-time feedback, covered 47% of US households, nearly doubling the 2010 penetration rate (Energy Information

Administration, 2018). Increasingly, utilities are working with software companies, like

Opower and EnergyHub, to take advantage of the increasing consumption data and provide consumer-facing feedback products that promote energy-efficient behaviors

(Ehrhardt-Martinez, Donnelly, & Laitner, 2010). Utilities and software companies are relying on psychological research in the design of these products to maximize their efficacy in reducing consumption (e.g., Allcott, 2011). Much of the current psychological research investigating behavioral response to consumption feedback is rooted in the influence of social norms (e.g., Schultz et al., 2016), with more research needed to understand the cognitive and emotional processes that can occur in the translation of feedback to action.

2

Self-conscious emotions could play a critical role in the translation between feedback message and action. Self-conscious emotions, such as pride and guilt, are emotions that require some degree of antecedent self-reflection or self-evaluation (Tangney, 2005). It is therefore critical to consider self-conscious emotions in feedback contexts since these emotions only occur after implicitly or explicitly reflecting on one’s actions and behaviors, which feedback facilitates through provision of information regarding one’s behavior (Tangney, 2005). In contrast, “basic” emotions (e.g., , fear, joy) can be experienced with a higher degree of automaticity, meaning that they do not require self- reflection to be elicited.

Pride and guilt, which are also referred to as moral emotions, have seen increasing presence in empirical work on PEB, given that engaging in PEB is often conceptualized as a moral action (Bamberg & Möser, 2007). Pride is defined as a positive emotion that reinforces prosocial and adaptive behaviors after one feels responsible for a positive outcome (Tracy & Robins, 2007), while guilt is a negative emotion that reinforces prosocial and reparative behaviors after one feels responsible for a negative outcome

(Tangney & Dearing, 2002). There are two main types of pride: achievement-oriented and hubristic (Lewis, 2000; Tracy & Robins, 2004). Achievement-oriented pride is the result of successful evaluation of specific actions, whereas hubristic pride is the result of successful evaluation of the self as a whole and is generally a more stable construct

(Lewis, 2000). Only achievement-oriented pride is measured and focused on in this

3 research, given that pride will be evaluated following a specific event (i.e., feedback reception). Going forward, mentions of pride in this research refer to achievement- oriented pride.

Lewis (2000) defines guilt and as closely-linked self-conscious emotions, with guilt being the result of failure evaluation of specific actions and shame the result of failure evaluation of the self as a whole. He explains that shame is a highly negative state that results in loss of action and confusion in thought (which makes it detrimental to promoting behavior), while guilt is not as intensely negative and does not lead to loss of action and confusion. In sum, pride and guilt are the focus of this self-conscious emotion research as they draw attention to one’s specific actions and foster tendency toward action related to the emotion-eliciting event. In comparison, other self-conscious emotions (i.e., hubristic pride and shame) draw attention to one’s total self and do not foster tendency toward action (Lewis, 2000; Onwezen, 2014; Tracy & Robins, 2004).

Self-conscious emotions occur when people attribute an event, like a feedback message, to internal causes (Tangney & Dearing, 2002; Weiner, 1985; Weiner, Graham, &

Chandler, 1982). For example, attribution theory predicts that internal attributions for success are likely to induce pride whereas internal attributions for failure are likely to induce guilt (Tracy & Robins, 2004; Weiner, 1985; Weiner et al., 1982). When events are attributed to external causes, basic emotions are more likely to be elicited (Tracy &

Robins, 2004). To elaborate, say that you receive a high grade on a college exam. You

4 may feel immediate joy after reading your exam score, which is a basic emotion.

However, you may feel pride if you internally attribute the cause of the good exam grade

(e.g., “I got a good grade because of my intelligence”). You would be less likely to feel pride if you externally attribute the cause of the good grade (e.g., “I got a good grade because the professor is a lenient grader”).

Additionally, self-serving attribution biases, in which people tend to internalize responsibility for success outcomes and externalize responsibility for failure outcomes, are well documented in psychological literature (e.g., Greenberg, Pyszczynski, &

Solomon, 1982; Greenwald, 1980; Harvey & Weary, 1984; Heider, 1958; Miller, 1976).

According to Taylor’s (1991) mobilization-minimization hypothesis, negative events

(like negative feedback) evoke strong cognitive responses (i.e., mobilization), and individuals are motivated to minimize the impact of the negative event (i.e., minimization). Since this pattern is more significant for negative events (compared to positive and neutral events), people may be more likely to look for ways to discount negative feedback, such as through external attribution of the feedback cause.

In addition to attributions, self-conscious emotions can also be impacted by one’s self- identities, which refer to how an individual subjectively sees oneself (Gatersleben,

Murtagh, & Abrahamse, 2014). According to identity theory, the emotions that one experiences after an event reflect the congruence or incongruence of that event with one’s identities (Stryker, 2004). For example, an identity-congruent event may be receiving a

5 feedback message that confirms an identity one has, which would lead to positive emotions. In contrast, an identity-incongruent event may be receiving a feedback message that disconfirms an identity one has, which would lead to negative emotions. In the present research, we examine how the extent to which one identifies as an environmentalist influences responses to carbon footprint feedback that either confirms or disconfirms their affiliation with environmentalist identity. Given identity congruence may impact the degree to which one experiences emotions, like pride and guilt, after receiving feedback, it is critical to examine how environmentalist identity may influence emotion in the translation from receiving a feedback message to pro-environmental action.

To better understand the cognitive and emotional processes that occur between feedback message reception and pro-environmental action, this research was designed to answer the following questions:

RQ1: What factors influence the degree to which people make internal attributions for their carbon footprint feedback?

RQ2: Can internal attributions and emotions (specifically, pride and guilt) explain the relationship between receiving a feedback message and engaging in PEB?

RQ3: How does congruence of feedback with one’s environmentalist identity impact experienced pride and guilt?

6

In Chapter 2, we investigate the first two research questions through the use of a survey administered to 397 participants from the online participant pool, Prolific Academic. An experimental design is used to manipulate aspects of the feedback hypothesized to influence internal attribution. After experimental results are examined to address RQ1,

RQ2 is examined by testing our conceptual model using the path analysis modeling tool,

PROCESS (Hayes, 2018).

In Chapter 3 we investigate the third research question using the same sample and survey data as Chapter 2. We examine how identity congruence impacts pride and guilt following feedback and may moderate the mediating role of emotion on the relationship between feedback and PEB. Similar to Chapter 2, we use PROCESS to assess our research question.

In Chapter 4 we provide an overview of our research findings and highlight implications of our work for feedback designers and practitioners.

7

Chapter 2. The Influence of Carbon Footprint Feedback Framing on Cognitive and Emotional Pathways to Pro-environmental Behavior

Abstract

Feedback on one’s consumption, for instance, via a carbon footprint calculator, is a common strategy used in attempts to promote pro-environmental behavior. Although evidence suggests that feedback can be effective in reducing consumption, little research examines the psychological processes that occur between feedback reception and subsequent behavior. To illuminate this gap, the present study uses attribution theory to examine the cognitive and emotional processes that occur upon receipt of such feedback.

Three hundred ninety-seven adults received a positively-, neutrally-, or negatively- framed bogus feedback message from a source high or low in credibility. Feedback frame did not impact pro-environmental behavior directly, but instead impacted the extent to which participants made internal attributions for their feedback and their pride and guilt responses. Independent of internal attribution, those who received negative feedback experienced more guilt than those who received neutral feedback, which resulted in greater pro-environmental behavior. Similarly, independent of internal attribution, those who received neutral feedback experienced more guilt than those who received positive feedback, which in turn resulted in greater pro-environmental behavior. Experienced pride did not mediate the relationship between feedback and pro-environmental behavior.

While experienced guilt positively impacts pro-environmental behavior, guilt-inducing feedback frames do not necessarily have to be overly negative to be effective. Based on 8 the findings, I advise that caution should be taken when using positive, pride-inducing feedback frames, as once people get such feedback they mail fail to progress positively in the direction of the desired behavior.

9

Introduction

Background

Individual-level consumption is a major contributor to greenhouse gas emissions, with scholars estimating that households are responsible for roughly 30 to 40 percent of U.S. greenhouse gas emissions (Vandenbergh et al., 2008). Feedback, the process of giving people information regarding their past behavior (Karlin et al., 2015), can be an effective intervention tool for reducing individual-level consumption and promoting pro- environmental behavior (PEB), and has received significant empirical attention

(Abrahamse et al., 2005; Schultz, 2014). As a case in point, a recent meta-analysis of 42 energy feedback studies supports the effectiveness of feedback in promoting PEB (Karlin et al., 2015). Although feedback interventions can be effective, a given individual who receives feedback may not always go on to change her or his consumption behavior, raising questions about the processes that guide post-feedback behavior. Given increases in sensing technology, digitization, and investment in updating utility infrastructure to smart grids, consumption data is likely to continue to proliferate in society. Such data can be crafted into consumption feedback, which highlights the need to understand the psychological processes that can occur in the translation of feedback to action for effective feedback design.

Several sets of factors – both internal and external to a person – may influence the effectiveness of feedback, including characteristics of the feedback message itself, characteristics of the recipient, and dynamic psychological processing of the feedback

10 message. The following is not intended as an exhaustive review of these factors, but rather as a snapshot of key findings from prior research. Current research on message characteristics finds that frequent feedback presented in engaging mediums (e.g., computers vs. utility bills) tends to be more impactful in reducing consumption

(Abrahamse et al., 2005; Karlin et al., 2015; Schultz, 2014). Regarding recipient characteristics, feedback tends to be a more effective tool among recipients who are already motivated or primed to change their behavior (Katzev & Johnson, 1987; Schultz,

2014). Finally, there is little research on the psychological processes by which feedback may translate to behavior. This is in part because most feedback studies take the following simple approach: provide feedback, then examine subsequent behavior. While a number of meta-analyses and reviews support the efficacy of feedback interventions

(Abrahamse et al., 2005; Karlin et al., 2015; Osbaldiston & Schott, 2012), these studies are inherently limited because they do not offer insights into how individuals process feedback. Wide-spread reliance on this limited approach has resulted in a glaring literature gap on the psychological processes that occur upon receipt of feedback. Both

Karlin et al. (2015) and Abrahamse et al. (2005) call for future research to investigate mechanisms by which feedback operates to improve understanding of how best to develop and deploy feedback interventions.

Attribution Theory and Self-Conscious Emotions

Weiner’s (1986) attribution theory of motivation and emotion explains that people ascribe causes to events along several core dimensions, which subsequent research

11 categorizes into four main dimensions of attributions: locus of causality (i.e., internal vs. external), stability, personal control, and external control (McAuley, Duncan, & Russell,

1992). Internal vs. external locus of causality is a critical dimension of attributions for determining whether self-conscious emotions emerge (Weiner, 1986). To better contextualize these processes, we first review self-conscious emotions.

Self-conscious emotions are emotions that require some degree of antecedent self- reflection or self-evaluation (Tangney, 2005). Pride and guilt are among the most widely studied self-conscious emotions in empirical work on PEB. Pride is defined as a positive emotion that reinforces prosocial and adaptive behaviors after one feels responsible for a positive outcome (Tracy & Robins, 2007), while guilt is a negative emotion that reinforces prosocial and reparative behaviors after one feels responsible for a negative outcome (Tangney & Dearing, 2002). According to attribution theory, when people attribute an event to internal causes (i.e., ascribe the causes of an event to something within themselves), they are more likely to experience self-conscious emotions (Tangney

& Dearing, 2002; Weiner, 1985; Weiner et al., 1982) than if they make external attributions for events (i.e., ascribe the causes of an event to something outside of themselves) (Figure 1), which is more likely to elicit basic emotions (e.g., anger, fear, joy) (Tracy & Robins, 2004). For example, internal attributions for success are likely to induce pride, whereas internal attributions for failure are likely to induce guilt (Tracy &

Robins, 2004; Weiner, 1985; Weiner et al., 1982).

12

Event

Something Something What caused the event? outside of within me me

Feelings of Feelings of Internal Pride and Pride and External Attribution Guilt Guilt Attribution Increased Decreased

Future Behavior

Figure 1. Attribution Theory (Weiner, 1986).

Self-serving attribution bias, whereby people internalize responsibility for successful outcomes and externalize responsibility for failed outcomes, is well documented, and offers some insight into how feedback messages may be processed and how people may differentially respond to feedback of varying valence (i.e., positivity or negativity) (e.g.,

Greenberg, Pyszczynski, & Solomon, 1982; Greenwald, 1980; Harvey & Weary, 1984;

Heider, 1958; Miller, 1976). For example, in a feedback context, this bias may influence people such that they are more likely to externalize responsibility for negative feedback compared to positive feedback. The self-serving attribution bias has a robust effect on how people attribute causality to events, as it has been found to occur both following private events, in which people are concerned with protecting their self-image (Greenberg

13 et al., 1982), and public events, in which people are additionally concerned with self- presentation (Bradley, 1978).

Self-Conscious Emotions and Pro-Environmental Behavior

Prior literature has established that pride is positively related to pro-environmental intention (PEI) and PEB (e.g., Antonetti & Maklan, 2014; Bissing-Olson et al., 2016;

Harth et al., 2013). The relationship between guilt and pro-environmental action is mixed, with studies finding both positive relationships (e.g., Rees, Klug, & Bamberg, 2015) and no relationships (e.g., Bissing-Olson et al., 2016) between guilt and PEI/PEB. However, this work has largely occurred outside of feedback contexts. Because self-conscious emotions are more likely to occur after reflecting on one’s actions and behaviors, it is critical to consider them in feedback contexts, which explicitly encourage reflection on one’s past behavior (Tangney, 2005). Minimal research has examined emotions elicited upon receipt of consumption feedback, and among the three studies we identified on this topic, findings are somewhat conflicting. Some evidence suggests that collective pride

(Ferguson, Becker, & Branscombe, 2017) and collective guilt (Ferguson et al., 2017;

Mallett, Melchiori, & Strickroth, 2013) following consumption feedback are related to higher PEB, whereas other research has found that personal guilt does not predict PEB

(Toner, Gan, & Leary, 2014).

Some scholars have called for work that integrates attributions into research on the determinants of PEB (Bamberg & Möser, 2007). Attributions can clarify the

14 circumstances under which different emotions are elicited and can provide more context for the mixed findings on the relationship between guilt and PEI/PEB. Although a few studies examine the role of attributions in environmentally-relevant behavior, they focus on attributions people make for the actions of others (e.g., Guckian, Chapman, Lickel, &

Markowitz, 2018; Markowitz & Malle, 2012). Because people often make different attributions for themselves than they do for others, this prior work falls short of providing a clear understanding of how individuals make attributions for their own behaviors

(Pieters, Bijmolt, van Raaij, & de Kruijk, 1998). To our knowledge, no prior work investigating the influence of pride and guilt on PEB has incorporated self-attributions.

We argue that attributions should not be ignored in this research given attribution theory

(Weiner, 1986) suggests internal attribution is a necessary pre-condition for pride and guilt.

Additionally, this research explores another potential influence on internal attribution relevant to a feedback context: source credibility. To our knowledge, no other study has directly tested for the impacts of message source credibility on internal attribution. We expect that sources low in credibility will cause people to internally attribute responsibility for their feedback less, given the low credibility source can be seen as an external reason why one received a particular feedback message. Expertise (i.e., competence and knowledge of the source; McGinnies & Ward, 1980) and trustworthiness

(i.e., honesty and integrity of the source; McGinnies & Ward, 1980)) are the two most important components of credibility (Giffin, 1967; Hovland, Janis, & Kelley, 1953).

15

Components of credibility, like trustworthiness, can be influenced by perceived motives of a source, with sources perceived as having informing motives being more highly trusted than those perceived to have persuading motives (Rabinovich, Morton, & Birney,

2012).

Figure 2. Study 1 Concept Map.

Contributions

At a broad scale, the present study advances understanding of the cognitive and emotional processes that can occur in the translation of feedback message reception to

PEB (See Figure 2 for our concept map). From a practical standpoint, improved understanding of the psychological processes implicated in feedback can inform marketing, communication, and intervention design.

Next, to our understanding, no work yet has examined the role of self-attributions in promoting environmentally-relevant outcomes, as attributions are typically studied in other domains, like academics (e.g., Baumgardner & Arkin, 1988; Russell & McAuley,

16

1986) and athletic performance (e.g., Spink & Roberts, 1980). Current work on attributions and PEB has only investigated attributions people make for other people

(e.g., Markowitz & Malle, 2012), despite the potential for attributions people make for the own current actions to have a significant impact on their future PEB, and despite calls for future research into self-attributions and PEB (e.g., Bamberg & Möser, 2007).

Our study also helps to clarify the mixed findings on the influence of guilt on PEB by including a number of unique factors in our research design. For example, much of the existing literature examining the links between pride, guilt, and PEI/PEB relies on research paradigms that elicit anticipated emotional states, as well as eliciting emotions through articles, stories, and hypothetical scenarios (e.g., Antonetti & Maklan, 2014;

Harth et al., 2013; Onwezen, Antonides, & Bartels, 2013; Rees et al., 2015; Schneider,

Zaval, Weber, & Markowitz, 2017). Our study builds on this body of work by investigating experienced pride and guilt due to personal action, which are emotions subject to being weakened or strengthened by self-serving attribution biases; such biases could diminish the power of guilt as a motivating force given tendencies to discount negative information about oneself.

Next, our research uses a repeated measures design whereby self-reported PEB is measured one week after feedback is given, which allows us to expand on the mixed findings on guilt and demonstrate if it is associated with future engagement in PEB rather than PEB measured in the same sitting as other survey variables.

17

Finally, the three studies we could identify that have previously investigated emotions elicited upon receipt of consumption feedback (i.e., Ferguson et al., 2017; Mallett et al.,

2013; Toner et al., 2014) imbedded social comparison into their feedback assessments

(e.g., messages stating one’s carbon footprint is larger or smaller than the average US citizen; Mallett et al., 2013), and cannot clarify what relationships between emotion and

PEB emerge without a clear comparison to others. Our research creates feedback messages that purposefully do not include language involving social comparison.

Overview of Research and Hypotheses

The study used a 3 x 2 between-subjects experimental design. Specifically, after reporting baseline PEB and taking a carbon footprint quiz, participants were provided with false feedback whereby feedback frame (i.e., positive, neutral, or negative) was experimentally manipulated. A similar false footprint feedback procedure has been used in previous research (e.g., Brook, 2011; Ferguson et al., 2017; Mallett et al., 2013; Toner et al.,

2014). Source credibility (i.e., high or low) of the organization purportedly providing the feedback was also manipulated in our study. Immediately after receiving false feedback, participants reported attributions for their feedback as well as pride and guilt. A one-week follow-up survey was used to measure PEB. We advance the following hypotheses:

H1: Follow-up PEB will be higher than baseline PEB.

18

H2a: The negative feedback condition will result in weaker internal attributions for the feedback relative to the positive and neutral feedback conditions.

H2b: The high source credibility condition will result in stronger internal attributions relative to the low source credibility condition.

H3a: Internal attribution of the cause of feedback will be positively related to pride.

H3b: Internal attribution of the cause of feedback will be positively related to guilt.

Consistent with much of the existing literature on emotion and pro-environmental action, we anticipate that post-feedback pride will be associated with PEB. The impacts of guilt on PEB are more mixed in the existing research that relies on anticipated guilt and guilt elicited from hypothetical scenarios. We believe that experienced guilt will be associated with PEB in our study because experiences of guilt due to personal action are likely more motivating for PEB (e.g., because of a desire to assuage that guilt).

H4a: Pride will be positively related to PEB.

H4b: Guilt will be positively related to PEB.

Methods

Participants

Participants were recruited from an online research platform, Prolific Academic (ProA).

ProA participants have been found to be more demographically diverse, more naïve, and

19 less dishonest than Amazon Mechanical Turk (MTurk) participants (Peer, Brandimarte,

Samat, & Acquisti, 2017). Palan and Schitter (2018) explain that ProA is a valuable crowdsourcing platform relative to others given its transparency to participants (e.g., knowing they are being recruited for research) and to researchers (e.g., knowing key details about subject pool with ability to screen a wide range of participant attributes).

The platform has seen increasing use in research (e.g., Hafner, Elmes, & Read, 2017;

Meleady & Crisp, 2017; Unsworth & McNeill, 2017) and has been found to replicate lab- based results (Peer et al., 2017).

Eligibility criteria for our study included U.S. citizenship, U.S. as current country of residence, and age of at least 18 years. Additionally, only ProA members who had agreed to participate in studies involving when creating their ProA profiles (roughly half of the total ProA participant pool) could view the study posting. This research was conducted under the overview and compliant with the policies of The Ohio State

University’s Institutional Review Board (Protocol # 2018B0168).

A power analysis with G*Power (Faul, Erdfelder, Lang, & Buchner, 2007) recommended a sample size of 324 participants to provide 95% power to detect medium effect sizes (η²

= 0.06) at α = 0.05 with 6 experimental groups. A large η2 effect size is generally regarded as > 0.14, a medium > 0.06, and a small > 0.01 (Cohen, 1988). To account for attrition and allow us to exclude inattentive participants, we planned to recruit 450 participants. A total of 457 participants enrolled in the study, of which 419 completed the

20 follow-up survey, yielding a follow-up rate of 92%. Of those who completed the follow- up, 19 participants were removed from analyses for failing an attention check (e.g.,

“Please select ‘Sometimes’ as your response to this item”). One participant was removed for not answering manipulation checks, and two participants were removed because their follow-up data could not be linked with their baseline responses. This left us with a final dataset of N = 397 participants, for an overall retention rate of 87%. See Table 1 for participant demographics.

Table 1. Sample vs. U.S. Demographic Characteristics.

Full Sample U.S. N 397 325,719,178a % Male 59.70 49.25a Median Age 30.0 38.1a % White 77.58 72.30a % Liberalc 46.10 26b a U.S. Census Bureau (2017) b Gallup (2018) c Political orientation in Gallup (2018) was measured using a five-point scale (i.e., very conservative, conservative, moderate, liberal, very liberal). In our survey, political orientation was measured using a similar seven-point scale that included “slightly conservative/liberal” options. For comparability to the Gallup results, “% Liberal” in Table 1 does not include those in our sample who identified as “slightly liberal”. Including the “slightly liberal” option, liberals made up 60.71% of the full sample.

21

Procedures

See Figure 3 for a visual representation of all study procedures. Using the ProA prescreening function, only eligible participants were able to view the study posting on

ProA. After agreeing to participate, participants were directed to an online platform hosted by Qualtrics to complete the time 1 study procedures, which took approximately

20 minutes, and for which participants were compensated $2.70.

22

Figure 3. Procedures.

Baseline PEB Assessment. First, participants completed a series of questions assessing their baseline PEB, using a scale adapted from the Recurring Pro-environmental Behavior

Scale (Brick, Sherman, & Kim, 2017). 23

Carbon Footprint Quiz. Next, to serve as a basis for receiving carbon footprint feedback, participants answered questions adapted from the CoolClimate Carbon Footprint

Calculator (The Berkeley Institute of the Environment, 2008). To enhance believability of feedback, adapted from procedures used in prior work (Lacasse, 2016), answer choices for the CoolClimate questions depended on frame condition. For example, in the negative feedback condition, response choices were anchored such that the average participant would be more likely to select responses on the higher end of response scales, which was intended to serve as a cue of relatively high consumption. For instance, for a question that asked how many flights participants took annually, the top end of the response scale among those in negative condition was “2+,” while top scale points among the neutral and positive conditions were “4+” and “6+”, respectively.

Feedback Source Description. After taking the carbon footprint quiz, all participants were told that the consumption questions they just answered (i.e., baseline PEB and carbon footprint quiz) were part of an “Environmental Behavior Test” used by an organization called Environment United to give people feedback on their consumption.

Prior to this point, participants had not been told that they would receive feedback on their answers to the baseline PEB and carbon footprint quiz questions. We chose

Environment United, a fictional organization, as the source, in order to minimize any potential pre-existing beliefs about a particular feedback source. Participants were randomly assigned to read a statement in which Environment United was described as

24 either high or low in credibility. The high credibility description emphasized the key trustworthiness and expertise components of credibility as described by McGinnies and

Ward (1980), while the description of the low credibility source was designed to undermine these concepts. Source motives (i.e., persuading or informing motives) were also manipulated to further differentiate the conditions in credibility in accordance with the findings of Rabinovich et al. (2012). Specifically, in the high credibility condition,

Environment United was described as being an established organization high in expertise, as well as having informing motives, no conflicts of interest, and high trustworthiness. In the low credibility condition, Environment United was described as being a new organization low in expertise, as well as having persuading motives, multiple conflicts of interest, and low trustworthiness. The source descriptions shown to participants are presented in Appendix A.

False Feedback. Next, participants were presented with a feedback message, which they were told was based on their answers to the previous Environmental Behavior Test questions. All participants were told that their carbon dioxide footprint was 17 metric tons per year (roughly the mean per capita carbon dioxide emissions for the U.S. in 2014;

Carbon Dioxide Information Analysis Center, 2014). Additionally, participants were randomly assigned to receive a positively, negatively, or neutrally framed message that accompanied the numeric feedback, each of which included affective language adapted from Lacasse (2016). The feedback messages shown to participants are presented in

Appendix A.

25

Survey. Key variables, including causal attributions, pride, and guilt, were measured immediately after participants received false feedback. The presentation order of the causal attribution and emotions measures was randomized. Participants then reported their PEI. Finally, covariates were assessed, including environmentalist identity and a range of demographic variables, and participants responded to manipulation check questions. Before covariates were assessed, participants completed a word association filler task to minimize the influence of the manipulation on these measures.

Follow-up PEB Assessment. One week after completion of the time 1 measures, the opportunity to participate in the follow-up survey was made available via ProA. The survey was hosted on Qualtrics and assessed PEB engaged in over the past week (since the first survey), using the same behaviors assessed for baseline PEB and PEI. After completing the follow-up, which took approximately 5 minutes, participants were debriefed, paid $0.75 for completing the follow-up, and entered into a drawing to win one of three $25 bonus payments.

Measures

Causal Attributions. Causal attributions for the feedback were measured using an adapted

12-item version of the Revised Causal Dimension Scale (Table 2; McAuley, Duncan, &

Russell, 1992). Three items assessed each of four causal dimensions: locus of causality, personal control, external control, and stability. Participants were asked to think about the

26 reasons they received their feedback message. Using a nine-point bipolar scale, they rated the extent to which reasons could be described between sets of two poles (e.g., “Is the reason(s) something: 1= ‘inside of you’ to 9= ‘outside of you’? ”). Our focus for the present study is on internal attribution, which is consistent with the locus of causality

(internal vs. external) dimension items.

Promax rotated principal component analysis constrained to four factors, per the four factors previously identified using the Revised Causal Dimension Scale (McAuley et al.,

1992), was run to examine the factor structure of the scale in our sample. The output revealed that the causal dimensions items did not load neatly onto the four factors as predicted by McAuley et al. (1992) (Table 2). A 3-factor solution was extracted, supported both by Cattell’s Scree test (Cattell, 1966) (Figure 9 in Appendix B) and the

K1 criterion (Kaiser, 1960), with the eigenvalues of the first three factors being greater than one. We used a minimum loading cutoff of 0.40, as recommended by Osborne and

Costello (2009), to determine whether an item loaded onto a given factor. For internal attributions, the focus of this research, two of the predicted locus of causality (i.e., all predicted locus of causality items except “Inside of you / Outside of you”) and all three predicted personal control items loaded together onto a single factor, which we label as

“internal attribution.” The “Unchangeable / Changeable” item cross-loaded onto both the internal attribution and stability factor, but the difference in magnitude between both loadings was less than 0.20, warranting dropping this items as it is difficult to discern which factor it represents (Ferguson & Cox, 1993). The five internal attribution items

27 were reverse scored (higher values indicate higher internal attribution of cause of feedback), and the mean of these items was taken to form a scale for internal attribution

(Cronbach’s α = 0.87).

Table 2. Promax Rotated Factor Loadings for Causal Attributions Scale Items.

Factor 1: Factor 2: Internal External Factor 3: Item Attribution Control Stability Unsupported Uniqueness Predicted Locus of Causality That reflects an 0.56 0.05 0.35 0.22 0.41 aspect of yourself / Reflects an aspect of the situation

Inside of you / 0.04 0.02 -0.12 0.96 0.08 Outside of you

Something about 0.63 -0.11 -0.03 0.22 0.38 you / Something about others Predicted Personal Control Manageable by you 0.91 0.01 0.04 -0.04 0.21 / Not manageable by you

You can regulate / 0.86 0.02 0.00 0.03 0.25 You cannot regulate

Over which you 0.92 0.04 0.02 -0.10 0.25 have power / Over which you have no power

Continued

28

Table 2. Continued

Item Factor 1: Factor 2: Factor 3: Unsupported Uniqueness Internal External Stability Attribution Control Predicted Stability Permanent / 0.04 0.04 0.87 -0.03 0.26 Temporary

Stable over time / 0.14 -0.06 0.88 -0.15 0.28 Variable over time

Unchangeable / -0.61 0.00 0.52 0.19 0.33 Changeable Predicted External Control Over which others 0.03 0.85 0.00 0.00 0.29 have control / Over which others have no control

Under the power of -0.14 0.71 0.06 -0.17 0.35 other people / Not under the power of other people

Other people can 0.08 0.88 -0.07 0.14 0.26 regulate / Other people cannot regulate Eigenvalues 4.08 1.98 1.88 0.71 Table Note. Based on principal component analysis constrained to four factors. Italics indicate item loading above 0.40 that belong to a unique factor. Items are grouped according to the predicted causal dimension factor loadings as outlined by McAuley et al. (1992).

Pride and Guilt. Post-feedback pride and guilt were measured with three items each adapted from The State Shame and Guilt Scale (Marschall, Sanftner, & Tangney, 1994) 29 previously used by Bissing-Olson et al. (2016). Participants were asked to rate the extent to which they were experiencing each of six emotions on a unipolar five-point scale ranging from 1= “not at all” to 5= “completely.” The mean of responses to “proud,”

“content,” and “pleased with myself” was taken to form a pride scale (Cronbach’s α =

0.87). Mean responses across “guilty,” “remorseful,” and “regretful” were combined to form a guilt scale (Cronbach’s α = 0.91). Similar to Bissing-Olson et al. (2016), pride and guilt scale scores were non-normally distributed. See Table 3 for descriptive information on pride and guilt and see Figures 10 and 11 (Appendix B) for histograms of pride and guilt distributions, respectively.

To examine potential sequencing effects of the causal attribution and emotions measures, a one-way ANOVA was used to test for the effects of presentation order (i.e., causal attributions measures first or emotions measures first) on internal attribution, PEI, and

PEB, and linear regression with a bootstrap resampling routine with 5000 samples was used to test for effects of presentation order on pride and guilt. Results revealed no significant differences due to presentation order for any of these constructs, ps > 0.12.

Pro-environmental Behavior. To measure baseline PEB, post-feedback behavioral intentions, and PEB at one-week follow-up, we adapted the Recurring Pro-environmental

Behavior Scale (Brick et al., 2017), which assesses frequency of performing a set of 16 repeated environmentally-relevant behaviors (e.g., composting, recycling, conserving water, turning off lights), rated on a five-point unipolar frequency scale ranging from 1=

30

“never” to 5= “very often”. For the baseline and follow-up scales, participants were asked to rate how often they performed each of the behaviors over the last week. For the intentions scale, participants rated how often they intended to perform each of the behaviors over the next week. Intentions were not used in key analyses and were collected as a back-up dependent measure in case of high attrition for the one-week follow-up PEB measure. For each set of items (i.e., baseline, intentions, follow-up), the mean of the respective sixteen items was taken to form a scale (Cronbach’s α = 0.74,

0.81, and 0.78 for baseline PEB, PEI, and follow-up PEB, respectively). See Table 3 for descriptive statistics of key outcome variables.

Table 3. Descriptive Statistics for Key Chapter 2 Outcome Variables.

Mean SD Skewness Kurtosis

Internal Attribution 6.02 1.69 -.58 .11

Pride 2.29 1.08 .56 -.67

Guilt 1.64 .87 1.45 1.51

Follow-up Pro-environmental Behavior 2.87 .60 .22 -.30

Covariates. Given that identity can influence emotion (e.g., Burke, 1991), we measured for a relevant identity (i.e., environmentalist identity). Environmentalist identity was measured with three items (e.g., “Acting environmentally friendly is an important part of 31 who I am.”) adapted from van der Werff, Steg, and Keizer (2013) answered on a seven- point Likert scale ranging from 1 = “strongly disagree” to 7 = “strongly agree”

(Cronbach’s α = 0.91). A range of other demographic variables were measured and demonstrated that our sample was skewed relative to U.S. averages (Table 1), including age, gender, race, and political orientation (1 = “extremely liberal,” 7 = “extremely conservative”). Given a high percentage of the sample identified as white (77.6%), race is coded dichotomously as 0 = “not white” and 1 = “white”). Participants who identified as male made up a majority of the sample (59.7%); gender is also coded dichotomously, with 0 = “not male” and 1 = “male,” as 1.0% of participants did not identify as male or female.

Manipulation Check. To assess perceived feedback valence (i.e., perceived positivity or negativity of the feedback), participants rated the extent to which they believed

“Environment United describes my carbon footprint as good” to be true. Perceived source credibility was assessed using four items adapted from Ter Mors et al. (2010) that asked participants to rate the extent to which they believed statements to be true (e.g.,

“Environment United is a trustworthy organization,” “Environment United is an expert on the environmental impacts of consumption”). Both perceived source credibility and perceived feedback valence items were rated on a seven-point scale ranging from 1= “not at all true” to 7= “extremely true.”

32

Manipulation Check

Pilot. Before conducting the full study, a pilot test (N = 72 after removing 4 participants who missed an attention check item) was conducted to ensure the efficacy of the experimental frame and source credibility manipulations. The pilot also recruited participants from ProA and used the same measures and procedures as the main study, excluding the follow-up component. Findings (detailed in Table 4) indicated that the frame manipulation yielded the predicted effects, whereby perceptions of valence were significantly more negative for the negative condition compared to the positive (t = -9.42, p < 0.01) and neutral (t = -4.81, p < 0.01) conditions, and were significantly more positive for the positive condition compared to the neutral condition (t = -3.85, p < 0.01).

The source credibility manipulation also yielded the predicted effects, whereby perceptions of source credibility were significantly lower in the low vs. high conditions (t

= -3.18, p < 0.01).

Main Study. Findings from the main study manipulation checks (also detailed in Table 4) were similar to those of the pilot. The frame manipulations yielded the predicted effects, whereby perceptions of valence were significantly more negative for the negative condition compared to the positive (t = -26.21, p < 0.01) and neutral (t = -9.59, p < 0.01) conditions, and were significantly more positive for the positive condition compared to the neutral condition (t = -15.33, p < 0.01). The source credibility manipulation also yielded the predicted effects, whereby perceptions of source credibility were significantly lower in the low vs. high conditions (t = -6.05, p < 0.01).

33

Table 4. Manipulation Check Results.

Perceived Credibility M (SD) Perceived Valence M (SD) High Low Positive Neutral Negative Credibility Credibility Valence Valence Valence Condition Condition Condition Condition Condition Pilot 4.50a (1.05) 3.62b (1.29) 5.33a (1.37) 3.71b (1.55) 1.75c (1.26) Main 4.57a (1.25) 3.80b (1.29) 5.81a (1.20) 3.34b (1.47) 1.64c (1.37) Table Note. Means were compared (1) within the pilot sample and (2) within the main study sample. For each sample, means for perceived credibility and means for perceived valence with unshared superscripts differ at the p < 0.01 level.

Dropout Analyses

Dropout analyses were conducted to identify any significant differences in internal attribution, pride, guilt, PEI, PEB, and experimental condition between those retained for analyses and those dropped from the full study sample for any reason. Specifically, one- way ANOVAs were used to test for differences between dropped and retained participants on internal attribution, PEI, and PEB. Linear regression with a bootstrap resampling routine with 5000 samples was used to test for differences between dropped vs. retained participants on pride and guilt, and logistic regression was used to test for differences between dropped vs. retained participants on frame condition and credibility condition.

Dropped participants had higher PEI scores (M = 3.35, SD = 0.65) than those retained [M

= 3.15, SD = 0.62; F(1, 455) = 5.63, p = 0.02]. Additionally, dropped participants had higher guilt scores (M = 2.12, SD = 1.26) than those retained (M = 1.64, SD = 0.87; b =

34

0.47; 95% bootstrap CI: 0.15, 0.81), which is likely driven by the fact participants in the negative frame condition were also more likely to be dropped compared to those in the positive (OR = 2.34; 95% CI: 1.18, 4.64) and neutral (OR = 1.92; 95% CI: 1.00, 3.66) conditions. Of 151 total participants assigned to the negative condition, 29 were dropped from the final sample, compared to 17 of 154 assigned to the neutral condition and 14 of

152 assigned to the positive condition. Most of the individuals dropped from the negative condition were removed for not completing the follow-up survey. A total of twenty of the

151 participants assigned to the negative condition did not complete the follow-up, compared to 7 of the 154 assigned to the neutral condition and 13 of the 152 assigned to the positive condition. Follow-up logistic regression testing for the effects of feedback frame condition on whether or not a participant completed the follow-up revealed that those in the negative condition were less likely to complete the follow-up than those in the positive (OR = 0.60; 95% CI: 0.29, 1.26) and neutral conditions (OR = 0.31; 95% CI:

0.13, 0.75). No other significant differences in follow-up completion were identified.

The systematic drop-out by condition, along with higher PEI scores among dropped vs. retained participants, provides reason to suspect that those in the negative condition may have gone on to engage in more PEB than those in the positive and neutral conditions, but leaving us unable to detect this. To further investigate this possibility, a one-way

ANOVA was conducted to discern whether PEI among those who did not take the follow-up differed by experimental condition. PEI scores among those who did not follow-up marginally differed between the negative (M = 3.29, SD = 0.68), neutral (M =

35

3.86, SD = 0.60), and positive conditions [M = 3.25, SD = 0.56; F(2, 37) = 2.49, p =

0.097]. Further, independent samples t-tests were conducted to test if those who did vs. did not take the follow-up differed in PEI within each feedback frame condition. Results revealed that there were no differences in PEI between those who did vs. did not follow- up within the negative condition, t(149) = 0.80, p = 0.42, nor within the positive condition, t(150) = 0.90, p = 0.37. Within the neutral condition, those who did not follow- up (M = 3.86, SD = 0.60) were higher in PEI than those who did (M = 3.20, SD = 0.62), t(152) = 2.76, p = 0.006. Takeaways from these dropout analyses are discussed in the limitations.

Results

Table 5. Summary of Chapter 2 Hypothesis Testing.

Hypothesis Supported? H1 Follow-up PEB will be higher than baseline PEB. Yes H2a The negative feedback condition will result in weaker internal Yes attributions for the feedback relative to the positive and neutral feedback conditions. H2b The high source credibility condition will result in stronger No internal attributions relative to the low source credibility condition. H3a Internal attribution of the cause of feedback will be positively No related to pride. H3b Internal attrition of the cause of feedback will be positively No related to guilt. H4a Pride will be positively related to PEB. No H4b Guilt will be positively related to PEB. Yes

36

See Table 5 for a summary of hypothesis testing outcomes. To test Hypothesis 1 (H1), a paired samples t-test was conducted to compare follow-up PEB scores to baseline PEB scores. There was a significant difference between follow-up PEB scores (M = 2.868, SD

= 0.603) and baseline PEB scores (M = 2.811, SD = 0.568) whereby PEB was significantly higher at the follow-up, t(396) = 3.066, p = 0.002. We also report Cohen’s d effect size to indicate the importance of this mean difference, d = 0.15. A large Cohen’s d effect size is generally regarded as > 0.8, a medium > 0.5, and a small > 0.2 (Cohen,

1988), therefore the difference in means between baseline and follow-up PEB, while significant, was relatively small.

Next, we tested Hypothesis 2 using a two-way ANCOVA to evaluate the effects of feedback frame condition (IV) and source credibility (IV) on internal attribution (DV), controlling for several covariates, including baseline PEB, environmentalist identity, age, political orientation, race, and gender. Baseline PEB was controlled to account for potential differences in attribution of the feedback due to existing levels of engagement in

PEB. Given that previous research has demonstrated relevant identities can influence emotion (e.g., Burke, 1991) and attribution theory suggests that attributions impact emotion (Weiner, 1986), environmentalist identity was controlled to avoid for potential impacts identity may have had in assessing internal attribution. To account for differences between U.S. demographic averages and our sample, age, gender, race, and political orientation were also used as statistical controls.

37

The ANCOVA results demonstrated that there was a statistically significant effect of feedback frame on internal attribution, F(2, 385) = 13.14, p < 0.001, partial η2 = 0.06

(H2a). In contrast to predictions, there was no statistically significant effect of source credibility on internal attribution, F(1, 385) = 0.28, p = 0.594, partial η2 = 0.001 (H2b).

The interaction between feedback frame and source credibility was not significant, F(2,

385) = 0.32, p = 0.728, partial η2 = 0.002. Post-hoc pairwise comparisons with the

Bonferroni adjustment showed that, as predicted, internal attribution was significantly stronger among the positive (adj M = 6.39) vs. negative (adj M = 5.40) condition, p <

0.001, and internal attributions were significantly stronger among the neutral (adj M =

6.19) vs. negative condition, p < 0.001. Internal attribution was not significantly different between the positive and neutral conditions, p = 0.892. See Table 6 for internal attribution descriptive statistics by experimental condition. The full ANCOVA output table can be found in Table 11 (Appendix B).

Consistent with H2a, these results indicate that those who received negatively-framed feedback attributed their feedback less internally than those who received positively- framed and neutrally-framed feedback. The pattern of results did not change when covariates were excluded from the model. Supplemental analyses to examine the impacts of source credibility on other variables of interest (e.g., pride, guilt, PEB) found that source credibility had no significant effect on any variable (Appendix B).

38

Table 6. Mean Internal Attribution and Standard Deviations by Experimental Condition.

Source Credibility High Low Total M(SD) 95% CI M(SD) 95% CI M(SD) 95% CI N N N Lower Upper Lower Upper Lower Upper Positive 6.51 6.13 6.89 6.40 6.00 6.80 6.46 6.18 6.73 (1.60) (1.64) (1.62) 71 67 138 Neutral 6.09 5.74 6.44 6.28 5.90 6.65 6.18 5.93 6.44 (1.47) (1.55) (1.51) 69 68 137 Negative 5.25 4.81 5.70 5.40 4.95 5.86 5.33 5.02 5.65 (1.70) (1.81) (1.75) 59 63 122 Total 5.99 5.76 6.22 6.04 5.80 6.28 (1.66) (1.72) 199 198 Table Note. Means and standard deviations for internal attribution for all experimental conditions: 3 (feedback frame: positive, neutral, negative) x 2 (source credibility: high, low).

The remaining hypotheses (H3-H4) were tested by analyzing the proposed conceptual model (Figure 2) using Hayes’ (2018) PROCESS macro for SPSS (version 3.3).

PROCESS is an OLS and logistic regression path analysis modeling tool that bootstraps indirect effects, a mediation testing technique that is considered superior to the causal steps approach or the Sobel test (Hayes, 2018). Given normality assumption violations,

OLS regression with bootstrapping was used to make inferences about regression coefficients for pathways with pride and guilt as outcomes. Bootstrapping makes no assumptions about the population distribution of measured variables, and allows for inferences to be made about regression coefficients, even when OLS normality

39 assumptions are violated (Darlington & Hayes, 2017). For consistency, bootstrap CIs were used to make inferences for all pathways in the model. Indirect effects and regression coefficients reported used 5,000 bootstrap estimates to generate 95% percentile bootstrap confidence intervals (CIs).

PROCESS model 81 was used with positive feedback frame and negative feedback frame dummy variables, with neutral feedback frame as the reference group. A custom random number seed was used to ensure that the results are replicable with the same pool of bootstrap estimates (seed = 2,000,000). As with testing H2, baseline PEB, environmentalist identity, age, gender, race, and political orientation were set as covariates. Additionally, though source credibility did not influence internal attribution, pride, guilt, or PEB, it was also set as a covariate to control for any potential impacts of this experimental condition.

See Figure 4 and Table 7 for a summary of all path coefficients, including indirect effects, direct effects, and total effects. Contrary to predictions, internal attribution was not related to pride, b = 0.03, 95% CI: -0.03, 0.09 (H3a) or guilt, b = -0.03, 95% CI: -

0.09, 0.03 (H3b). Pride was not related to PEB, b = 0.03, 95% CI: -0.01, 0.07 (H4a), but guilt was positively related to PEB, b = 0.05, 95% CI: 0.004, 0.10 (H4b).

40

Figure Note. Bold paths indicate significance of the regression coefficient based on 95% percentile bootstrap CIs. Figure 4. Path Coefficients.

41

Table 7. Summary of Path Coefficients, Indirect Effects, Direct Effects, and Total Effects.

Independent Mediating Mediating Dependent Effect Effect Effect Effect Effect Direct Indirect effect Total variable (IV) variable variable variable of IV of IV of M1 of M1 of M2 effect effect (M1) (M2) (DV) on M1 on M2 on M2 on DV on DV (c’) (c) (a1) (a2) (a3) (b1) (b2) (a x 95% b) CI Positive IA - PEB .21 - - .01 - -.01 .002 (-.005, -.01 Feedback .01) Negative IA - PEB -.79* - - .01 - .01 -.01 (-.02, .02 Feedback .01) Positive Pride - PEB .96* - - .03 - -.01 .02 (-.01, -.01 Feedback .07) Negative Pride - PEB -.29* - - .03 - .01 -.01 (-.02, .02 Feedback .004) Positive Guilt - PEB -.40* - - .05* - -.01 -.02* (-.04, - -.01 Feedback .002) Negative Guilt - PEB .35* - - .05* - .01 .02* (.001, .02 Feedback .04) Positive IA Pride PEB .21 .96* .03 .01 .03 -.01 .0001 (- -.01 Feedback .0003, .001) Negative IA Pride PEB -.79* -.29* .03 .01 .03 .01 -.001 (-.003, .02 Feedback .001) Positive IA Guilt PEB .21 -.40* -.03 .01 .05* -.01 - (-.002, -.01 Feedback .0003 .001) Negative IA Guilt PEB -.79* .35* -.03 .01 .05* .01 .001 (-.001, .02 Feedback .004) Table Note: 5000 bootstraps. Path coefficients are unstandardized effects. IA = Internal Attribution; PEB = Pro-environmental Behavior. * Indicates significance of 95% percentile bootstrap CI.

42

Next, we go beyond our specific hypotheses and describe the indirect, direct, and total effects in our model. The omnibus tests for total and direct effects of feedback frame on

PEB were not significant, F(2, 387) = 0.13, p = 0.877 and F(2, 384) = 0.10, p = 0.901, respectively, which demonstrated that feedback frame was not directly influencing PEB.

The relative indirect effects of positive frame (with neutral as the reference group) on

PEB were not significant for the serial mediation through internal attribution and pride, point estimate (PE) = 0.0001, 95% CI: -0.0003, 0.001, or through internal attribution and guilt, PE = -0.0003, 95% CI: -0.002, 0.0006. Similarly, the relative indirect effects of negative frame (with neutral as the reference group) on PEB were not significant for the serial mediation through internal attribution and pride, point estimate (PE) = -0.0006,

95% CI: -0.003, 0.001, or through internal attribution and guilt, PE = 0.001, 95% CI: -

0.001, 0.004. These insignificant indirect effects for serial mediation demonstrate that feedback frame, internal attribution, emotion, and PEB were not operating as demonstrated in our conceptual model (Figure 2).

Instead, the results demonstrate that guilt mediated the relationship between feedback frame and PEB without arising due to internal attribution. Specifically, the relative indirect effect of negative frame (with neutral as the reference group) on PEB through guilt was significant, PE = 0.02, 95% CI: 0.001, 0.04, as was the relative indirect effect of positive frame (with neutral as the reference group) on PEB through guilt, PE = -0.02,

95%: -0.04, -0.001. There were no significant indirect effects of feedback frame on PEB

43 through pride alone or internal attribution alone, demonstrating that neither pride nor internal attribution mediated the relationship between feedback frame and PEB.

The pattern of results does not change if our covariates are excluded, with the exception of baseline PEB. Specifically, when baseline PEB is not in the model, pride also becomes a significant mediator of the relationship between feedback frame and PEB. Specifically, the relative indirect effect of positive frame (with neutral as the reference group) on PEB through pride, PE = 0.09, 95% CI: 0.03, 0.15, and the relative indirect effect of negative frame (with neutral as the reference group) on PEB through pride, PE = -0.03, 95% CI: -

0.05, -0.005, are both significant. However, controlling for this variable is critical to rule out any differences in follow-up PEB arising from baseline PEB vs. the feedback intervention. Therefore, we are confident that our results controlling for baseline PEB yield more meaningful results.

Supplemental analyses were conducted to examine the model using positive and neutral dummy variables, with negative feedback as the reference group (Table 12 in Appendix

B). These analyses show the same pattern of findings and demonstrate that guilt is also a significant mediator for the relationship between positive vs. negative feedback, PE = -

0.04, 95% CI: -0.08, -0.003.

44

Discussion and Conclusions

This study examined cognitive and emotional mechanisms that influence PEB after one receives feedback on his or her consumption, building on previous literature in two key ways. First, in examining emotions and cognitions that occurred between feedback reception and behavior in addition to often-studied behavioral outcomes, we advance knowledge on the processes by which feedback works as an intervention tool. Second, by assessing experienced emotions, we contribute to the mixed body of literature on the association between guilt and PEB by showing that experiences of guilt can be influential in increasing PEB, even one-week after the initial guilt-eliciting event. Our key takeaways were that, independent of internal attribution, those who received negative feedback experienced more guilt than those who received neutral feedback, which resulted in more PEB. Additionally, independent of internal attribution, those who received neutral feedback experienced more guilt than those who received positive feedback, which in turn resulted in more PEB.

Supporting our first hypothesis, we found that follow-up PEB was higher overall than baseline PEB, consistent with prior work finding that feedback is effective in boosting pro-environmental action (Abrahamse et al., 2005; Karlin et al., 2015; Schultz, 2014).

However, the different feedback frames in our study did not directly impact PEB; feedback frame instead affected intermediate processes between message reception and action. Immediately after feedback reception, the feedback frame created differences in the extent to which people ascribed causes of the feedback to something about

45 themselves, with those receiving negatively-framed feedback attributing the feedback cause less internally (i.e., deflecting responsibility more) than those in the neutral and positive conditions. This finding aligns with research on self-serving attribution biases

(e.g., Greenberg et al., 1982).

Contrary to predictions from attribution theory (Weiner, 1985), the extent to which people made internal attributions for their feedback did not impact the extent to which they experienced pride and guilt. Bamberg and Möser (2007) called for more research into internal attribution processes as determinants of PEB; this research explored such processes, and found that they do not behave as expected, and ultimately showed no direct or indirect effects on PEB, which demonstrates that internal attribution of responsibility for environmental outcomes was not enough on its own to explain why one engages in PEB.

In this research we attempted to expand on the characteristics of feedback messages outlined by Karlin et al. (2015) that can influence feedback’s efficacy by demonstrating that source credibility is another important factor, which impacts internal attribution of the feedback. Contrary to our prediction, the credibility of the feedback source did not impact internal attribution, despite the fact that our manipulation check demonstrated a significant difference in perceived credibility between the two source conditions. We offer two possible interpretations for this null finding: 1) Though the sources differed in credibility, it is possible that this difference was not extreme enough to create a

46 difference in how people responded to feedback messages (e.g., perhaps a difference in internal attribution would emerge after getting environmental feedback from an oil company described as overtly corrupt vs. an environmental NGO described as being knowledgeable and trustworthy). 2) As demonstrated in Table 4, the mean score for perceived credibility for those in the low source credibility condition was 3.80, which is only 0.20 lower than the scale midpoint of 4, which signifies that one believes it is

“moderately true” that the source is credible; perceived credibility in the low source credibility condition was not in fact all that low.

As long as sources are not extremely different in perceived credibility, our results showed that the valence of the feedback was much more critical to determining how people make internal attributions for their feedback, which previous research has established is critical to how we internalize events of different valences (e.g., Greenberg et al., 1982; Taylor,

1991).

Though the results did not find support for serial mediation through internal attribution and pride or guilt, guilt positively predicted PEB and was a significant mediator of the relationship between feedback frame and PEB. Specifically, as described above, independent of internal attribution, those who received negative feedback experienced more guilt than those who received neutral feedback, which resulted in greater PEB.

Similarly, independent of internal attribution, those who received neutral feedback experienced more guilt than those who received positive feedback, which in turn resulted

47 in greater PEB. As seen in supplemental analyses (Table 12 in Appendix B), when examining the model with positive and neutral dummy variables for feedback (with negative as the reference group instead of neutral), we additionally found that guilt mediated the relationship between positive vs. negative feedback frame and guilt.

In contrast to both our prediction as well as prior work relying on anticipated emotions and hypothetical scenarios to elicit emotions (e.g., Harth et al., 2013; Onwezen et al.,

2013; Schneider et al., 2017), our findings demonstrated that pride did not predict PEB; we also found that pride is not a mediator of the relationship between feedback frame and

PEB. Anticipated pride may be perceived as a desired state that is motivating for engaging in PEB given individuals strive to experience positive emotions (Frijda, 2007), but experienced pride does not have this effect. In this experiential feedback context, pride may signal that everything is okay, and doesn’t facilitate pushing oneself to engage in more PEB. There may be somewhat of a moral self-licensing effect (Merritt, Effron, &

Monin, 2010) occurring with those who experience pride, whereby they do not necessarily go on to decrease their future PEB, but instead become content with their behavioral status quo and do not feel inclined to increase PEB. It is also possible that in order for experiences of personal pride to facilitate PEB, a degree of social comparison or being “better” than relevant others needs to be present, though future work is needed to clarify this. Given negative information (vs. positive information) about oneself is generally attended to longer (Ito, Larsen, Smith, & Cacioppo, 1998; Taylor, 1991), the

48 experience of pride may not have lingered with participants as much as the experience of guilt, which may have facilitated guilt’s stronger influence on PEB.

We conclude our discussion with two main takeaways for feedback designers and practitioners. First, caution should be taken when using positive, pride-inducing feedback frames, as once people get such feedback they may fail to progress positively in the direction of the desired behavior. When creating affectively-laden messaging interventions that are intended to evoke positive emotions, it may be more effective to use anticipated instead of experienced pride. Such an approach would encourage people to think about how much pride they could feel in the future after engaging in PEB, as prior work finds that anticipated pride motivates individuals to engage in PEB to achieve this positive emotion (Onwezen et al., 2013). Second, guilt-inducing feedback frames do not have to be overly negative to be effective. Guilt emerged following both negative and neutral message frames, and then led to more pro-environmental action, so it may not be necessary to use a strongly negative message frame to induce guilt. Though internal attribution of feedback did not impact guilt or PEB in our study, this pattern of discounting responsibility for negatively-framed messages highlights a point of caution with using harshly negative message frames.

Limitations

It is important to consider the findings of our study in light of some limitations. First, dropout analyses found that those who were dropped from the sample were higher in PEI than those retained. This appears to come from the neutral condition specifically, as 49 within the neutral condition, those who did not take the follow-up had higher PEI scores than those retained; no such differences were found among the positive or negative conditions. If results from neutral participants who did not take the follow-up were captured, it seems reasonable to assume that they would likely have translated to higher follow-up PEB scores, which may have resulted in significant differences in PEB across frame conditions. However, only seven participants from the neutral condition did not take the follow-up, so the extent to which a group of this small size would have impacted overall neutral condition PEB scores is questionable. We also found that participants in the negative condition were less likely to take the follow-up survey than those in the positive and neutral conditions. That said, there was no difference in PEI between those in the negative condition who did vs. did not follow-up, which assuages potential concerns that the PEB scores among retained negative condition participants may be biased. Still, appropriate caution should be used in interpreting the results. Second, compared to U.S. demographic averages (Table 1), our sample was generally younger, and over-represented males, those who identified their race as white, and those who identified as liberal. To account for these differences between U.S. demographic averages and our sample, age, gender, race, and political were used as statistical controls.

However, given the demographic skew of our sample, results may not generalize to all populations. Third, PEB was measured using self-report. Future work using actual consumption data is needed to confirm whether our results can be replicated using observed measures of behavior. Fourth, we cannot fully rule out the possibility that the subjective experience of filling out the carbon footprint quiz explained participants’

50 emotional responses, and we cannot confirm that experienced pride and guilt was only due to the feedback message, especially given the scale points for the carbon footprint quiz were altered in each condition. However, we believe that altering the scale points was necessary to make the feedback believable, as this technique has been used in prior work (Lacasse, 2016). Finally, though this research was strong in its experimental design, appropriate caution should be taken with generalizing these findings given they are from a single study.

Future Directions

As mentioned above, future work should further explore explanations for why pride in our study was not associated with increased PEB (e.g., lack of social comparison, increase in behavioral status quo satisfaction). For example, future work could examine post-feedback pride, and additionally manipulate or measure social comparison, and explore how differences in social comparison then impacts whether pride is related to

PEB.

Guilt explained how negative and neutral feedback increased PEB, though these two feedback frames were significantly different in how much participants internally attributed feedback. This difference in internal attribution points out that there could be slightly different reasons why those in the negative vs. neutral conditions experienced guilt, and future work could benefit from exploring what people were feeling guilty about to better understand the process of how guilt was leading to PEB following these different frames. For example, guilt experienced in the negative (vs. neutral) condition 51 may have been more related to feeling bad about trying to discount the negatively-framed message through external attribution. However, guilt experienced by those in the neutral condition may have been due more to reflecting on past behavior, as the neutral frame did not promote externalization of feedback and hence may have permitted more cognitive space to reflect on behavior.

Finally, because internal attribution does not appear to be influencing pride, guilt, or

PEB, and guilt appears to mediate the relationship between feedback frame and PEB on its own, future work should investigate other potential factors, like identity, that impact post-feedback self-conscious emotions (Stets & Trettevik, 2014) to further clarify the translation from feedback message reception to pro-environmental action.

52

Chapter 3. The Influence of Environmentalist Identity on Emotional Response to Carbon Footprint Feedback

Abstract

The congruence of feedback with one’s self-identities is a critical factor in determining individuals’ emotional responses to feedback. Existing research in identity theory finds that positive emotions follow identity-congruent feedback and negative emotions follow identity-incongruent feedback. However, this identity theory research tends to rely on general positive and negative affect, and in a few studies, basic emotions like sadness, anger, and joy. The present research contributes to identity theory work by investigating the self-conscious emotions of pride and guilt, which can emerge following feedback that is congruent or incongruent with a context-relevant identity (i.e., environmentalist). Self- conscious emotions require a degree of self-reflection, which people would likely not engage in as much following a positive event compared to a negative event, even if that positive event did not align with their identity, given self-enhancing motives and the tendency to accept positive feedback about the self without scrutiny. Further, existing research (Chapter 2 of this thesis) demonstrates that guilt mediates the relationship between an environmental feedback message and pro-environmental behavior, but, contrasting with attribution theory, is not influenced by attributional processes. The present research explores other factors, namely environmentalist identity, that may impact this relationship. Using a bogus feedback experimental design, this research finds

53 that the indirect effect of feedback frame on pro-environmental behavior through guilt is stronger for participants with higher levels of environmentalist identity. Our findings demonstrate that guilt is more explanatory of the translation between receiving feedback and engaging in pro-environmental action for those who more strongly identify as environmentalists, which highlights an appropriate personal characteristic that is important for understanding when leveraging guilt in messaging interventions is most effective.

54

Introduction

Background

Emotions can be impacted by one’s self-identities (Stets & Burke, 2000). Identity theory predicts that the emotions one experiences after an event reflect the congruence or incongruence of that event with one’s identities (Stryker, 2004). However, one study has found experimental evidence that is at odds with identity theory predictions, and demonstrates identity incongruence in a positive direction can actually result in positive, and not negative, feelings (Stets, 2005). Identity theory research typically only looks at positive and negative affect or basic emotions (e.g., sadness, anger, joy) following identity-congruent or identity-incongruent events, with no work specifically distinguishing how self-conscious emotions, such as pride and guilt, are elicited after such events. This is a critical gap given self-conscious emotions are unique in that they require one to reflect on past behavior in order to experience the emotion (Tracy &

Robins, 2004), which is less likely after positive vs. negative events given people generally accept positive information about themselves and spend more time evaluating negative information (Ito et al., 1998).

Receiving feedback on one’s behavior occurs when information is given to a person about their past behavior. This information may either be congruent or incongruent with one’s self-identities. Chapter 2 demonstrated that guilt mediates the relationship between feedback frame and pro-environmental behavior (PEB). In Chapter 3, positively-framed feedback is considered to be identity-congruent for those who identified strongly as

55 environmentalists, whereas negatively-framed feedback would be incongruent for these same people. Despite our predictions in Chapter 2, we found no evidence that attributions influence pride and guilt, leading us in Chapter 3 to look to an alternative factor, identity, that may impact how these emotions may be elicited and clarify the translation from feedback message reception to PEB.

Process Model of Self-Conscious Emotions

Self-conscious emotions have received much less attention in emotion research compared to basic emotions, despite the fact models of basic emotions do not sufficiently describe how self-conscious emotions are elicited, as self-conscious emotions involve different attributional antecedents (Robins & Schriber, 2009; Tracy & Robins, 2004). Tracy and

Robins (2004) build a process model that illustrates the conditions necessary for the elicitation of self-conscious emotions, which incorporates research on causal attributions and emotions, cognitive appraisals and emotions, and self-evaluative processes (e.g., assessing identity congruence). We focus specifically on the identity component of self- conscious emotion elicitation in the present study given no evidence that attributions influence self-conscious emotions (Chapter 2). According to the process model, an event must be appraised as relevant to one’s identity for the elicitation of self-conscious emotions. The next step in the process model is for the event to be appraised as congruent or incongruent with one’s identity. Congruence appraisals determine the valence of the self-conscious emotion, with positive emotions (e.g., pride) elicited from identity- congruent events and negative emotions (e.g., guilt) elicited from identity-incongruent

56 events (Lazarus, 1991). However, this model is based on hypothesized relationships – it generates testable hypotheses about self-conscious emotion activation but lacks empirical support and requires future research to substantiate. To further explore research at the intersection of identity and emotion, we turn to work on identity theory.

Identity Theory

Self-identity refers to how an individual subjectively sees oneself (Gatersleben et al.,

2014). Self-identities consist of self-views that stem from self-categorization or identification processes, in which people incorporate meanings, expectations, and standards into these identities (Stets & Burke, 2000). Self-identity can be related to any part of the self, including aspects like preferences, habitual behavior, personality traits, and personal goals (Gatersleben et al., 2014; McAdams, 1995), and can be influenced by personal motivations (e.g., for self-enhancement) and social interaction through demands and expectations assigned to the roles one performs (Stets & Burke, 2000; Stryker &

Burke, 2000; Whitmarsh & O’Neill, 2010). Identity theorists have also used the term

“person identity” to refer to self-identity (Stets, 1995; Stets & Burke, 2000). These identities function across different roles and situations, and can be interpreted as individual characteristics or attributes that people view as representations of “who they are, how they feel, and what they value” (Stets & Biga, 2003, p. 403).

Based on feedback from the environment or social context, people derive different meanings from a given situation based on their identities, and they compare these

57 meanings with their identity standards through self-verification (Burke, 1991). Self- verification theory (Swann, 1983, 1987) emerged from research on self-consistency, which dates back to Festinger’s (1957) theory of cognitive dissonance. At its core, this theory explains that individuals are “invested in preserving their firmly held self- conceptions and that they do so by soliciting self-verifying feedback” (Swann, Pelham, &

Krull, 1989, p. 783). In that way, it is similar to confirmation bias (Nickerson, 1998), but with a focus on information about the self. Identity theory incorporates the process of self-verification, and explains that people are motivated to verify their identities, through means like seeking identity-congruent feedback (Stryker & Burke, 2000).

In the process of self-verification, differences between self-relevant meanings from a situation and one’s identity standard can create an error signal, which would translate to negative subjective experiences when the discrepancy is construed as significant (Burke

& Stets, 1999). However, there are differing findings in research on identity as to when this discrepancy is considered significant enough to lead to negative subjective experience, which are described in the following paragraphs.

Emotions reflect the degree of congruence between the perceived meanings of one’s identity in a situation and one’s identity standard (Burke, 1991). There is general consensus in identity theory with empirical support that positive feelings emerge when identity-relevant information in a situation confirms an identity standard (i.e., identity verification or identity congruence), and negative feelings emerge when identity-relevant

58 information in a situation fails to meet an identity standard (i.e., identity nonverification or identity incongruence) (Averett & Heise, 1987; Burke & Stets, 1999; Cast & Burke,

2002; Stets & Trettevik, 2014). This pattern of findings is echoed in the prediction regarding identity congruence and emotion from the process model of self-conscious emotions (Tracy & Robins, 2004). In particular, Tracy and Robins predict that any form of identity-incongruence will likely elicit negative self-conscious emotions, and do not specify if there are differences in emotion elicitation following positive incongruence

(i.e., feedback is more positive than one’s identity standard) or negative incongruence

(i.e., feedback is more negative than one’s identity standard). Similarly, in identity research, negative feelings have been found to emerge following both positive and negative identity incongruence (Burke, 1991). For example, if one self-identifies as a hard worker and then receives poor feedback from his or her supervisor, the worker would experience negative incongruence and negative emotion. Further, if one self- identifies as a lazy, poor worker and then receives a glowing feedback report from his or her supervisor, the worker would experience positive incongruence and, according to identity theory and the process model of self-conscious emotions, would also experience negative emotion.

However, one study found experimental evidence that demonstrates identity incongruence in a positive direction can actually result in positive, and not negative, feelings (Stets, 2005), and has called for identity theory predictions to be modified to take situational variability into account (detailed below). In situations that reflect the pattern

59 of results found in Stets (2005), it is thought that self-enhancing motives outweigh the desire for accurate feedback that confirms one’s self-identity (i.e., self-verifying motives). Self-enhancement theory can be dated back to Allport (1937), who suggested that people strive to view themselves positively, and that self-enhancement is a central goal of human existence. Overall, this theory states that people prefer positive feedback in order to preserve or enhance a positive self-image. There has been much empirical evidence for self-enhancement theory (Swann et al., 1989), with self-enhancement motives being used to explain phenomena like self-serving attributions (e.g., Greenberg et al., 1982; Greenwald, 1980; Harvey & Weary, 1984; Heider, 1958; Miller, 1976), downward social comparison (e.g., Suls & Wills, 1991), and predictions of future success

(e.g., Weinstein, 1980).

Stets (2005) explains that identity theory needs to be updated to reflect situational variability of identity incongruent events. For example, negative emotions may occur following positive incongruence only when there is ample opportunity to reflect on incongruent information and when degree of “overreward” (i.e., positivity of incongruent event) is high. This can be used as a starting ground for understanding the conditions under which positive identity incongruence may lead to positive emotions, and not negative emotions as identity theory would predict (Burke, 1991). The majority of work looking at emotional response to identity congruence has focused on general positive or negative emotion, only gets as specific as feeling good or bad (i.e., positive or negative

60 affect), or feeling depressed or distressed following congruence or incongruence (Burke,

1991).

To better understand the links between identity and emotion, including when positive incongruence may lead to positive or negative emotions, identity research needs to go beyond consideration of general positive and negative affect, and examine how self- conscious emotions play into this process. Specifically, research ought to examine how self-conscious emotions are elicited following identity-congruent and identity- incongruent events, as identity is a crucial component of the process by which self- conscious emotions emerge (Tracy & Robins, 2004). Focusing on specific, self-conscious emotions can shed light on another dimension of whether incongruent events lead to positive or negative emotions beyond Stets’ (2005) detailing of situational variability.

Specifically, while situational variability can impact whether people experience positive or negative emotion, there may be types of emotion that people are more likely to experience in some cases of identity incongruence than others. Self-conscious emotions require a degree of self-reflection, which people would likely not engage in as much following a positive event compared to a negative event, even if that positive event did not align with their identity, given self-enhancing motives and the tendency to seek positive feedback about the self (e.g., Swann et al., 1989). For this reason, it is likely that people may not experience guilt following a positive event, even if that event is incongruent with their identity.

61

Research Context

Identity congruence and self-conscious emotion, specifically pride and guilt, are researched here in a consumption feedback context, whereby participants receive positive or negative feedback on their carbon footprint. Environmentalist identity is examined given its relevance to carbon footprint feedback. This research focuses on comparing feedback events that are clearly positive or negative so that it is clear whether the feedback is congruent or incongruent with environmentalist identity. This research uses the same sample and data that was detailed in Chapter 2.

Contributions

This research makes three main contributions. First, identity research is lacking in work that looks at how identity congruency and incongruency leads to emotions beyond general positive and negative affect. This study’s focus on self-conscious emotions

(namely pride and guilt) offers a starting point for clarifying the relationships between identity congruency and this class of specific emotions.

Second, existing findings are mixed regarding whether positive self-identity incongruence, in addition to negative self-identity incongruence, elicits negative emotion.

Limited evidence supports theorized predictions (e.g.,Tracy & Robins, 2004) that positive incongruence elicits negative emotion (e.g., Burke & Stets, 1999). However, other findings show that this is not always the case, and begin to outline some conditions that are required for positive identity incongruence to lead to negative emotions (e.g., high

62 degree of “overreward”) (Stets 2005). The present research extends Stets’ preliminary findings on when positive incongruence leads to negative emotions by clarifying whether positive or negative self-conscious emotions (i.e., pride or guilt), which require a degree of internal reflection to experience, are elicited following an event that is positively incongruent with one’s self-identity.

Third, this research expands on work looking at pride and guilt in PEB contexts (e.g.,

Antonetti & Maklan, 2014; Rees et al., 2015) by introducing another critical factor in emotion elicitation, identity. The addition of identity can shed light on the mixed findings of the influence of guilt on PEB. Chapter 2 described building on the mixed guilt findings by examining guilt in a new context (i.e., experienced vs. anticipated guilt). Chapter 3 is still focused on experienced guilt, and also builds on these mixed findings by examining how identity may play an important role in guilt’s indirect effect on PEB, which can help clarify the psychological translation between receiving a feedback message to taking pro- environmental action.

Hypotheses

Consistent with identity theory, we expect that positive, identity-congruent feedback should elicit high levels of pride, a positive emotion, for those at high levels of environmentalist identity. Additionally, we expect that negative, identity-incongruent feedback should elicit guilt for those at high levels of environmentalist identity.

According to identity theory, those low in environmentalist identity should experience 63 positive emotions following identity-congruent events, which in this case would be negative feedback, and negative emotions following identity-incongruent events, which in this case would be positive feedback. However, contrary to identity theory, we expect that those low in environmentalist identity will experience pride and guilt in patterns more consistent with the valence of the feedback. Specifically, we advance the following hypotheses:

H1a: Environmentalist identity will moderate the relationship between feedback frame and pride.

H1b: Specifically, participants will feel pride more when they get positive (vs. negative) feedback when they are higher in environmentalist identity.

H1c: Participants will not feel more pride when they get negative (vs. positive) feedback when they are lower in environmentalist identity.

H2a: Environmentalist identity will moderate the relationship between feedback frame and guilt.

H2b: Specifically, participants will feel guilt more when they get negative (vs. positive) feedback when they are higher in environmentalist identity.

H2c: Participants will not feel more guilt when they get positive (vs. negative) feedback when they are lower in environmentalist identity.

64

H3: Environmentalist identity will moderate guilt’s mediating role (established in

Chapter 2) on the relationship between feedback frame and PEB, such that the size of the indirect effect will be larger for higher levels of environmentalist identity. That is, guilt will be more explanatory of the process by which feedback frame leads to PEB for participants higher in environmentalist identity.

Methods

All methods (including procedures and measures) are the same as Chapter 2. Note that

Chapter 3 focuses on feedback frame comparisons between the positive and negative conditions given this research compares emotional response to identity-congruent and identity-incongruent events, which are clearly defined following overt positive and negative message frames. Comparisons to the neutral frame condition are described as supplemental analyses (Appendix C). See Table 8 for descriptive statistics of key Chapter

3 variables.

65

Table 8. Descriptive Statistics for Key Chapter 3 Variables.

Mean SD Skewness Kurtosis

Environmentalist Identity 4.63 1.32 -.53 .11

Pride 2.29 1.08 .56 -.67

Guilt 1.64 .87 1.45 1.51

Follow-up Pro-environmental Behavior 2.87 .60 .22 -.30

Results

Table 9. Summary of Chapter 3 Hypothesis Testing.

Hypothesis Supported? H1a Environmentalist identity will moderate the relationship between Yes feedback frame and pride. H1b Participants will feel pride more when they get positive (vs. Yes negative) feedback when they are higher in environmentalist identity. H1c Participants will not feel more pride when they get negative (vs. Yes positive) feedback when they are lower in environmentalist identity. H2a Environmentalist identity will moderate the relationship between Yes feedback frame and guilt. H2b Participants will feel guilt more when they get negative (vs. Yes positive) feedback when they are higher in environmentalist identity. H2c Participants will not feel more guilt when they get positive (vs. Yes negative) feedback when they are lower in environmentalist identity. H3 Environmentalist identity will moderate guilt’s mediating role on Yes the relationship between feedback frame and PEB, such that the size of the indirect effect will be larger for higher levels are environmentalist identity.

66

See Table 9 for a summary of hypothesis testing. Hypotheses were tested by analyzing the proposed conceptual model (Figure 5) using Hayes’ (2018) PROCESS macro for

SPSS (version 3.3). PROCESS is an OLS and logistic regression path analysis modeling tool that bootstraps indirect effects, a mediation testing technique that is considered superior to the causal steps approach or the Sobel test (Hayes, 2018). Given normality assumption violations, OLS regression with bootstrapping was used to make inferences about regression coefficients for pathways with pride and guilt as outcomes.

Bootstrapping makes no assumptions about the population distribution of measured variables, and allows for inferences to be made about regression coefficients, even when

OLS normality assumptions are violated (Darlington & Hayes, 2017). For consistency, bootstrap CIs were used to make inferences for all pathways in the model. Indirect effects and regression coefficients reported used 5,000 bootstrap estimates to generate 95% percentile bootstrap confidence intervals (CIs).

67

Figure 5. Study 2 Concept Map.

PROCESS model 7 was used with a custom random number seed to ensure that the results are replicable with the same pool of bootstrap estimates (seed = 2,000,000).

Baseline PEB, political orientation, race, age, and gender were set as covariates, as was done in Chapter 2 analyses. Baseline PEB was controlled to account for potential differences in emotional response and future PEB due to existing levels of engagement in

PEB. These demographic covariates were included given the differences between our sample and U.S. averages on these factors (Table 1). Additionally, though source credibility did not influence internal attribution, pride, guilt, or PEB, it was also set as a covariate to control for any potential impacts of this experimental condition, as was done in Chapter 2 analyses.

68

See Table 10 for a summary of the regression coefficients from testing the model.

Consistent with H1a, the interaction term for positive feedback frame (with negative as the reference group) and environmentalist identity was significant in predicting pride, b =

0.31, 95% CI: 0.11, 0.50. Consistent with H2a, the interaction term for positive feedback frame (with negative as the reference group) and environmentalist identity was significant in predicting guilt, b = -0.21, 95% CI: -0.34, -0.06. Though not a part of this study’s hypotheses, interactions between feedback frame and environmentalist identity in predicting pride and guilt were also examined for positive and negative frames with neutral as the reference group. No significant interactions occurred; results are reported in

Appendix C.

69

Table 10. OLS Regressions with Bootstrapping on Pride and Guilt (Mediators) and Pro- environmental Behavior (Dependent Variable) for Moderated-Mediation Analysis.

Predictor Outcome Pride 95% CI Guilt 95% CI PEB 95% CI PF -.16 (-1.13, .18 (-.50, .001 (-.10, .80) .81) .10) NF -.51 (-1.45, .17 (-.73, -.001 (-.09, .48) 1.13) .09) EI -.07 (-.20, .15 (.01, .26) - - .08) EI x PF .31 (.11, .50) -.21 (-.34, - - - .06) EI x NF .18 (-.03, -.12 (-.33, - - .38) .07) Pride - - - - .04 (-.005, .08) Guilt - - - - .05 (.009, .10) Baseline .26 (.07, .46) .16 (-.005, .80 (.74, .86) PEB .33) Credibility .05 (-.12, -.06 (-.22, -.003 (-.07, .23) .10) .06) Gender .02 (-.16, .01 (-.16, -.04 (-.11, .21) .19) .03) Race .03 (-.20, .02 (-.17, -.06 (-.15, .26) .21) .02) Age .001 (-.007, -.005 (-.01, -.0008 (-.004, .009) .003) .002) Political .11 (.05, .17) -.02 (-.08, -.04 (-.07, - Orientation .03) .02) Constant .87 (.07, 1.20 (.49, .68 (.43, .93) 1.63) 1.94) r2 .37 - .17 - .67 - N 397 - 397 - 397 - Note: Unstandardized effects reported. Bold indicates significance based on 95% percentile bootstrap confidence interval (5000 bootstraps). PF = Positive Feedback. NF = Neutral Feedback. (Feedback variables dummy coded with negative feedback as reference group.) EI = Environmentalist Identity. PEB = Pro-environmental Behavior. The interpretations of the significant interactions with EI x PF for pride and guilt outcomes are: Positive (negative) feedback had more of a positive impact on pride (guilt) for higher levels of environmentalist identity.

To test H1b, H1c, H2b, and H2c, it was necessary to probe the significant interactions.

The pick-a-point approach was used to probe the interaction at low, moderate, and high

70 values of environmentalist identity (Hayes & Matthes, 2009). The Johnson-Neyman technique for finding regions of significance of the effect of the independent variable at levels of the moderator was not able to be used to probe the interaction since the normality assumption of OLS regression is not met when using pride and guilt as outcome variables, as established in Chapter 2 (Hayes & Montoya, 2017).

Values for environmentalist identity at the 16th percentile (3.33; low), 50th percentile

(5.00; moderate), and 84th percentile (6.00; high) were used. Among those low, moderate, and high in environmentalist identity, the conditional effects of positive (vs. negative) feedback frame on pride were 0.87 (95% CI: 0.50, 1.24), 1.38 (95% CI: 1.17, 1.59), and

1.69 (95% CI: 1.38, 2.00), respectively (Figure 6). These conditional effects demonstrate that positive feedback’s impact on pride is more positive when participants are higher in environmentalist identity (H1b). Consistent with H1c, participants low in environmentalist identity do not feel more pride when they get negative (vs. positive) feedback, which can also be visualized in Figure 6 given the positive feedback pride values are consistently higher than negative feedback pride values across low, moderate, and high values of environmentalist identity.

71

Figure Note. Dotted vertical lines indicate low (16th percentile; 3.33), moderate (50th percentile; 5.00), and high (84th percentile; 6.00) values of environmentalist identity.

Figure 6. Pride as a Function of Environmentalist Identity for Positive vs. Negative Feedback.

Further, among those low, moderate, and high in environmentalist identity, the conditional effects of positive (vs. negative) feedback frame on guilt were -0.51 (95% CI:

-0.76, -0.25), -0.85 (95% CI: -1.08, -0.61), and -1.05 (95% CI: -1.38, -0.72), respectively

(Figure 7). These conditional effects demonstrate that negative feedback’s impact on guilt is stronger when participants are higher in environmentalist identity (H2b). Consistent with H2c, participants low in environmentalist identity do not feel more guilt when they get positive (vs. negative) feedback, which can also be visualized in Figure 7 given the

72 negative feedback guilt values are consistently higher than positive feedback guilt values across low, moderate, and high values of environmentalist identity.

Figure Note. Dotted vertical lines indicate low (16th percentile; 3.33), moderate (50th percentile; 5.00), and high (84th percentile; 6.00) values of environmentalist identity.

Figure 7. Guilt as a Function of Environmentalist Identity for Positive vs. Negative Feedback.

As demonstrated in Appendix B, guilt mediates the relationship between positive feedback on PEB (given negative feedback as the reference group). The index of moderated mediation, which tests whether conditional indirect effects at any values of the

73 moderator are statistically different, was used to examine H3. As established in Chapter

2, pride does not mediate the relationship between feedback frame and PEB for any combination of dummy feedback frame variables, so results for moderated mediation with pride were not examined. The index of moderated mediation (I) was significant for positive vs. negative feedback, I = - 0.01, 95% CI: -0.03, -0.001, indicating that conditional indirect effects of feedback frame on PEB through guilt were different among those at different levels of environmentalist identity.

Given the significant index of moderated mediation, it was appropriate to probe the interaction for more clarity on the indirect effect differences between those at differing levels of environmentalist identity. The indirect effect of positive vs. negative feedback frame on PEB through guilt was negative and increased in strength as environmentalist identity strengthens. Specifically, among those low, moderate, and high in environmentalist identity, the conditional indirect effects of positive (vs. negative) feedback frame on guilt were -0.03 (95% CI: -0.06, -0.004), -0.05 (95% CI: -0.09, -

0.007), and -0.06 (95% CI: -0.12, -0.01), respectively (Figure 8). This probing indicated that guilt explained more why negatively framed feedback leads to PEB for those higher in environmentalist identity.

74

Figure Note. Effect on feedback frame refers to the size of the direct effect of positive vs. negative feedback frame on PEB and the size of the indirect effect of positive vs. negative feedback frame on PEB through guilt. Dotted vertical lines indicate low (16th percentile; 3.33), moderate (50th percentile; 5.00), and high (84th percentile; 6.00) values of environmentalist identity. This figure demonstrates that the effect of positive (vs. negative) feedback on PEB operates indirectly through greater feelings of guilt more so among those higher in environmentalist identity. Independent of post-feedback emotion, the framing of feedback does not impact PEB.

Figure 8. Direct and Indirect Effects of Feedback Frame on Pro-Environmental Behavior at Values of Environmentalist Identity.

The pattern of results did not change if covariates were not controlled, with the exception of baseline PEB. As discussed in Chapter 2, when baseline PEB was not in the model, pride also became a significant mediator of the relationship between feedback frame and

PEB. When running the PROCESS model without baseline PEB as a covariate for

Chapter 3 analyses, the index of moderated mediation through pride was significant for 75 positive vs. negative feedback, I = 0.05, 95% CI: 0.02, 0.10, which indicated that the indirect effect of positive vs. negative feedback frame on PEB through pride was larger for those higher in environmentalist identity. However, controlling for baseline PEB was critical to rule out any differences in follow-up PEB arising from baseline PEB vs. the feedback intervention, so we are confident that our results controlling for baseline PEB yield more meaningful findings.

Though not a part of this study’s hypotheses, moderated mediation was also tested using for positive and negative frames with neutral as the reference group. Results are reported in Appendix C.

Discussion and Conclusions

This research used a bogus feedback experiment to examine how self-conscious emotions and environmentalist identity may explain the translation between receiving a feedback message and engaging in PEB. Our findings contribute to identity theory by examining how identity congruency and incongruency lead to self-conscious emotions.

Consistent with identity theory (Burke, 1991; Stryker & Burke, 2000), our results demonstrate that following identity-congruent, positive feedback, those higher in environmentalist identity experience higher levels of pride, and that following identity- incongruent, negative feedback, those higher in environmentalist identity experience

76 more guilt. However, other relationships between identity and emotion do not cleanly follow identity theory predictions.

Specifically, following identity-congruent, negative feedback, those low in environmentalist identity did not experience higher levels of pride compared to when receiving identity-incongruent, positive feedback. This demonstrates that people do not appear to be internally reflecting on their past behavior and feeling proud after receiving negative information about themselves, even though this information is considered to be identity-congruent. Similarly, following identity-congruent, negative feedback, those low in environmentalist identity did not experience lower levels of guilt compared to what they experienced after identity-incongruent, positive feedback. People do not appear to be internally reflecting on their past behavior and feeling guilty after receiving positive information about themselves, even though this information is identity-incongruent. This research assumes that the feedback is incongruent with those who were low in environmentalist identity. Since I only measure environmentalist identity and not an opposing identity (e.g., anti-environmentalist) and do not include a measure of perceived identity congruence, I cannot confirm that those low in environmentalist identity saw the feedback as identity incongruent. However, these findings suggest some interesting contributions to identity theory that deserve exploration in future work.

Specifically, these findings may help to clarify the existing mixed results regarding whether positive identity incongruence, in addition to negative identity incongruence,

77 elicits negative emotion by demonstrating that negative self-conscious emotions in particular may be less likely to occur following positive identity-incongruence. Self- conscious emotions, like guilt, require one to reflect on past behavior, which is less likely to occur if one gets a positive feedback message, even if this message is incongruent with one’s identity. Asymmetric psychological response to differently valenced events are well-documented, with negative events demanding greater attention and more thorough cognitive processing than positive events (e.g., Ito et al., 1998; Taylor, 1991). For this reason, following positive feedback, regardless of identity congruence, individuals may not give equivalent time and attention to think through and process, therefore making them less likely to experience guilt due to positive identity incongruence. Though we did not examine time in our study and cannot confirm this, it is possible that a lower amount of time devoted to evaluating positive events may amplify the conditions outlined for positive incongruence to lead to negative emotions (Stets 2005), as people tend to be systematically biased toward spending less time evaluating positive vs. negative events.

Reflecting on the conditions outlined by Stets’ (2005) as being relevant for positive incongruence to lead to negative emotions, it is possible that the degree of “overreward” of the positive feedback was high enough and the amount of time participants were given to evaluate their feedback was not long enough such that those low in environmentalist identity would then feel guilty, though future work is needed to rule out these alternative explanations. This research is the first to demonstrate that self-conscious emotions may not always conform to predictions of identity theory following congruent and incongruent events. This may be due to self-conscious emotions requiring internal reflection to

78 experience (e.g., Tracy & Robins, 2004); future work should compare basic emotions alongside self-conscious emotions to see if significant differences in emotional response emerge.

Self-verifying motives could explain why following positive, identity-congruent feedback, those higher in environmentalist identity were experiencing higher levels of pride as they were able to verify an identity important to them. Likewise, self-verifying motives could explain why following negative, identity-incongruent feedback, those higher in environmentalist identity were experiencing higher levels of guilt, as an error signal was created given the discrepancy between the feedback and one’s identity standard, which was then construed as a negative subjective experience. However, for those low in environmentalist identity, self-enhancing motives seem to impact the experience of self-conscious emotions post-feedback more than self-verifying motives, given identity congruence is not able to accurately capture the degree to which these people experienced pride or guilt. Overall, confirming an identity via negative feedback does not lead to pride because it may not be construed as a “win” despite identity congruence; the valence of the feedback message itself is more salient in determining how people respond emotionally.

Further, this research provides more clarification on the process by which guilt can indirectly impact PEB. Specifically, the effect of negative (vs. positive) feedback on PEB operates indirectly through greater feelings of guilt more so among those higher in

79 environmentalist identity. Independent of post-feedback emotion, the framing of feedback does not impact PEB.

Overall, guilt is more explanatory in the translation between receiving a feedback message and taking pro-environmental action following negative vs. positive feedback.

Guilt appears to be more explanatory of this translation for those who are higher in environmentalist identity. For this reason, framing feedback messages in guilt-inducing ways may be an effective strategy for promoting PEB, and such a strategy should be used more when the target audience consists of strong environmentalists.

Limitations

Given the research described in this chapter uses the same data as the previous chapter, the same limitations apply to this research as described in Chapter 2. Additionally, this research has another important consideration. I am assuming that for those with low environmentalist identity scores (i.e., those who disagreed that being an environmentalist was an important part of who they are), the positive feedback is incongruent with their level of environmentalist identity, and the negative feedback is congruent with their level of environmentalist identity. The identity measure may only be capturing the extent to which an identity is present and may not directly map onto whether the feedback is construed as congruent or incongruent with one’s identity. Future research would benefit from including questions asking about a relevant but opposite identity (e.g., anti- environmentalist) to have a clearer point of comparison for identity congruence with

80 environmentalist identity. Future research could also ask participants about perceived congruency of feedback to help support the validity of this assumption.

Though we tried to minimize the impact of the feedback manipulation on environmentalist identity through the inclusion of a filler task in the survey, it is possible that the manipulation impacted participant’s responses to environmentalist identity items, which we assume are stable aspects of themselves. A one-way ANOVA testing for differences between frame conditions (IV) in environmentalist identity (DV) revealed a significant omnibus test, F(2, 394) = 3.42, p = 0.034, η2 = 0.017. Post-hoc pairwise comparisons with the Bonferroni adjustment showed that environmentalist identity between the positive and negative conditions was just within the range of significance at the 0.05 level (MD = 0.39, p = 0.049). There were no significant differences between the positive and neutral or negative and neutral conditions, p > 0.12. While it is important to keep the impact of the manipulation on environmentalist identity in mind, we do not feel that this impact has substantively altered the pattern of our findings given the small effect size. Further, the different levels of environmentalist identity that were investigated for conditional effects (3.33, 5.00, and 6.00) were much more spread apart relative to the mean difference between the positive and negative conditions. Including the environmentalist identity measure before the manipulation could have created its own issues, including increased social desirability and consistency biases in reporting baseline

PEB and/or impacting the experience that participants had with the manipulation.

81

Future Directions

The results demonstrated that guilt can explain the translation between feedback and

PEB, but it does not have as strong as an explanatory effect for those low in environmentalist identity. Additionally, the supplemental analyses (Appendix C) demonstrated that, when comparing negative and positive frames with neutral as the reference group, guilt was only explanatory in the translation to PEB for those moderate and high in environmentalist identity, but not those low in environmentalist identity.

Future work should focus on explaining factors that influence why those low in environmentalist identity may choose to engage in PEB after receiving feedback.

Additionally, future work should look at basic emotions (e.g., sadness, anger, joy) alongside pride and guilt to see if identity theory predictions with regard to identity congruence and emotion only apply to basic emotions and not self-conscious emotions, or to establish that it is something about the feedback context (e.g., degree of

“overreward”, time available to evaluate feedback) that caused people to feel more positive emotions after positive feedback and more negative emotions after negative feedback, regardless of degree of environmentalist identity.

82

Chapter 4. Overview and Conclusions

In sum, we find that future PEB does not depend directly on feedback framing. However, the cognitive and emotional processes that occur between feedback and PEB were impacted by feedback framing. Negatively-framed feedback caused participants to attribute responsibility for the cause of their feedback less internally than those who received positive and neutral feedback, but internal attribution did not impact pride, guilt, or PEB. Independent of internal attribution, those who received negative feedback experienced more guilt than those who received neutral feedback, which resulted in more

PEB. Similarly, independent of internal attribution, those who received neutral feedback experienced more guilt than those who received positive feedback, which in turn resulted in more PEB. Pride did not mediate the relationship between feedback frame and PEB.

Guilt’s mediating role on the relationship between feedback frame and PEB was stronger among those higher in environmentalist identity. In other words, experienced guilt, but not experienced pride, explained the translation between a feedback message and engaging in pro-environmental action, and experienced guilt explained this translation more for those higher in environmentalist identity.

Overall, this research is an important first step to providing clarification on the cognitive and emotional translation between receiving a feedback message and engaging in pro- environmental action, which adds to much of the existing research on feedback that

83 simply implements a feedback intervention and then measures a behavioral dependent variable. Another overarching takeaway from this research is that it provides additional context regarding when pride and guilt are effective in promoting PEB. Much of the existing research on the influence of pride and guilt on PEB has focused on anticipated instead of experienced emotions, with the general consensus that anticipated pride positively promotes future PEB (e.g., Schneider et al., 2017), and that anticipated guilt’s influence on future PEB is more mixed, with some research showing that it is positively related to PEB, and others showing no relationship with PEB (e.g., Onwezen et al., 2013 vs. Schneider et al., 2017). As demonstrated by our findings, experienced pride was not related to future PEB.

Our research demonstrates that positive identity incongruence (i.e., getting feedback that is more positive relative to one’s identity) does not always lead to negative emotions, as predicted by identity theory (Burke, 1991). Similarly, we find that identity congruence does not necessarily mean one will experience pride. We expand on existing identity theory research by explaining that the valence of feedback can be more critical in determining emotional response rather than identity congruence, as even those low in environmentalist identity still felt more pride after receiving identity-incongruent, positive feedback (vs. negative feedback) and more guilt after receiving identity- congruent negative feedback (vs. positive feedback). Given self-conscious emotions require one to reflect on their past behavior to experience, we believe that self-conscious emotions may be particularly less likely to follow identity theory predictions regarding

84 congruence compared to much of the current work looking at general positive and negative affect. This is because it seems less likely, for example, that people low in environmentalist identity would think through past experiences of themselves engaging in tasks associated with environmentalist identity after receiving positive carbon footprint rather than feeling a knee-jerk, generally negative emotional reaction. However, future work is needed to further clarify this possibility.

Besides making contributions to scholarly literature, this work also provides important takeaways for practitioners. Practitioners (e.g., utility providers, environmental agencies, environmental advocacy groups, pro-environmental technology marketers) can benefit from a more thorough understanding of the emotion elicitation process in behavior- change contexts in order to improve the efficacy of pro-environmental communication and intervention design. Feedback can be an effective intervention tool to promote environmentally-friendly individual behavior, and it is likely to become more common as sensing technology, digitization, and investment in updating infrastructure to smart grids expands the availability of consumption data. Understanding how people respond to feedback on a cognitive and emotional level is critical for maximizing efficacy of future feedback interventions.

We provide several implications for research and practice:

85

Experiences of guilt are more effective in promoting pro-environmental action than experiences of pride. While some research warns against evoking anticipated guilt in environmental messaging (Schneider et al., 2017), other research concludes that evoking guilt is an important aspect of environmental messaging (Rees et al., 2015). Our research demonstrates that experiencing guilt explains why people go on to increase their PEB after receiving feedback, meanwhile experiencing pride is not motivating. According to past research, anticipating pride seems to be motivating for engaging in PEB as people may be striving to achieve feelings of pride, whereas while they are experiencing the feelings of pride, they may become satisfied with their behavioral status quo, and do not change.

Inducing guilt in consumption feedback does not have to be achieved through overly negative message frames. Our findings demonstrate that guilt can emerge following both negative and neutral message frames, which then leads to pro-environmental action.

While guilt is more explanatory of why negative feedback relative to neutral feedback leads to PEB, guilt still explains why those who received neutral feedback, relative to positive feedback, went on to engage in PEB. Despite the neutral framing, it appears that people generally still interpret feedback on their carbon footprint negatively.

The framing of a feedback message (e.g., positive, neutral, negative) may be more important than environmentalist identity congruence in determining pride and guilt responses to feedback. People do not always experience negative emotions after

86 receiving feedback that may be incongruent with their identity, and they do not always experience positive emotions after receiving feedback that may be congruent with their identity. When people low in environmentalist identity receive positive feedback, they feel more pride than guilt, despite the feedback potentially being incongruent with their affiliation with environmentalist identity. Additionally, when people low in environmentalist identity receive negative feedback, they feel more guilt than pride, despite the feedback potentially being congruent with their affiliation with environmentalist identity. It is important to note that this research cannot definitively claim that those low in environmentalist identity see the feedback as incongruent with their identity given I only measure the extent to which participants see themselves as environmentalists. To provide further support for this takeaway, future work should use the same feedback procedures but additionally measure a relevant but opposite identity

(e.g., anti-environmentalist) to provide a clearer representation of whether someone sees feedback as congruent or incongruent with their identity.

Feedback frames designed to evoke guilt are more effective at promoting pro- environmental action among those with stronger environmentalist identities. It is important to know the audience when designing a feedback intervention, as differences in identity can impact how effective inducing guilt can be in promoting PEB. Framing feedback messages in guilt-inducing ways can be an effective strategy for promoting

PEB, and such a strategy should be used more when the target audience consists of strong environmentalists.

87

References

Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A review of intervention

studies aimed at household energy conservation. Journal of Environmental

Psychology, 25(3), 273–291. https://doi.org/10.1016/j.jenvp.2005.08.002

Allcott, H. (2011). Social norms and energy conservation. Journal of Public Economics,

95(9–10), 1082–1095. https://doi.org/10.1016/j.jpubeco.2011.03.003

Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt.

Antonetti, P., & Maklan, S. (2014). Exploring Postconsumption Guilt and Pride in the

Context of Sustainability: Postconsumption Emotions and Sustainability.

Psychology & Marketing, 31(9), 717–735. https://doi.org/10.1002/mar.20730

Averett, C., & Heise, D. R. (1987). Modified social identities: Amalgamations,

attributions, and emotions. The Journal of Mathematical Sociology, 13(1–2), 103–

132. https://doi.org/10.1080/0022250X.1987.9990028

Bamberg, S., & Möser, G. (2007). Twenty years after Hines, Hungerford, and Tomera: A

new meta-analysis of psycho-social determinants of pro-environmental behaviour.

Journal of Environmental Psychology, 27(1), 14–25.

https://doi.org/10.1016/j.jenvp.2006.12.002

88

Baumgardner, A. H., & Arkin, R. M. (1988). Affective state mediates causal attributions

for success and failure. Motivation and Emotion, 12(2), 99–111.

https://doi.org/10.1007/BF00992167

Bissing-Olson, M. J., Fielding, K. S., & Iyer, A. (2016). Experiences of pride, not guilt,

predict pro-environmental behavior when pro-environmental descriptive norms

are more positive. Journal of Environmental Psychology, 45, 145–153.

https://doi.org/10.1016/j.jenvp.2016.01.001

Bradley, G. W. (1978). Self-Serving Biases in the Attribution Process: A Reexamination

of the Fact or Fiction Question. Journal of Personality and Social Psychology, 36,

56–71.

Brick, C., Sherman, D. K., & Kim, H. S. (2017). “Green to be seen” and “brown to keep

down”: Visibility moderates the effect of identity on pro-environmental behavior.

Journal of Environmental Psychology, 51, 226–238.

https://doi.org/10.1016/j.jenvp.2017.04.004

Brook, A. (2011). Ecological footprint feedback: Motivating or discouraging? Social

Influence, 6(2), 113–128. https://doi.org/10.1080/15534510.2011.566801

Burke, P. J. (1991). Identity Processes and Social Stress. American Sociological Review,

56(6), 836. https://doi.org/10.2307/2096259

Burke, P. J., & Stets, J. E. (1999). Trust and Commitment through Self-Verification.

Social Psychology Quarterly, 62(4), 347. https://doi.org/10.2307/2695833

Carbon Dioxide Information Analysis Center. (2014). CO2 emissions (metric tons per

capita). Oak Ridge National Laboratory, Tennessee, United States.

89

Cast, A. D., & Burke, P. J. (2002). A Theory of Self-Esteem. Social Forces, 80(3), 1041–

1068.

Cattell, R. B. (1966). The Scree Test For The Number Of Factors. Multivariate

Behavioral Research, 1(2), 245–276.

https://doi.org/10.1207/s15327906mbr0102_10

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).

Hillsdale, NJ: Lawrence Earlbaum Associates.

Darlington, R. B., & Hayes, A. F. (2017). Regression analysis and linear models:

concepts, applications, and implementation. In Methodology and the Social

Sciences. New York: Guilford Press.

Dietz, T., Gardner, G. T., Gilligan, J., Stern, P. C., & Vandenbergh, M. P. (2009).

Household actions can provide a behavioral wedge to rapidly reduce US carbon

emissions. Proceedings of the National Academy of Sciences, 106(44), 18452–

18456. https://doi.org/10.1073/pnas.0908738106

Ehrhardt-Martinez, K., Donnelly, K. A., & Laitner, J. (2010). Advanced Metering

Initiatives and Residential Feedback Programs: A Meta-Review for Household

Electricity-Saving Opportunities (p. 140). Retrieved from American Council for

an Energy-Efficient Economy website: http://www.aceee.org/research-report/e105

Energy Information Administration. (2018). Annual Electric Power Industry Report.

Retrieved from https://www.eia.gov/electricity/data/eia861/

Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible

statistical power analysis program for the social, behavioral, and biomedical

90

sciences. Behavior Research Methods, 39(2), 175–191.

https://doi.org/10.3758/BF03193146

Ferguson, Becker, & Branscombe. (2017). Collective Emotions Help Us to Understand

Carbon Feedback Interventions. Presented at the The Society for the

Psychological Study of Social Issues, Albuquerque, NM.

Ferguson, E., & Cox, T. (1993). Exploratory Factor Analysis: A Users’ Guide.

International Journal of Selection and Assessment, 1(2), 84–94.

https://doi.org/10.1111/j.1468-2389.1993.tb00092.x

Festinger, L. (1957). A theory of cognitive dissonance. Stanford University Press.

Frijda, N. H. (2007). The laws of emotion. Mahwah, NJ: Erlbaum.

Gallup. (2018). Americans’ Ideological Views, by Year. Retrieved from

https://news.gallup.com/poll/225074/conservative-lead-ideology-down-single-

digits.aspx

Gatersleben, B., Murtagh, N., & Abrahamse, W. (2014). Values, identity and pro-

environmental behaviour. Contemporary Social Science, 9(4), 374–392.

https://doi.org/10.1080/21582041.2012.682086

Giffin, K. (1967). The contribution of studies of source credibility to a theory of

interpersonal trust in the communication process. Psychological Bulletin, 68(2),

104–120. https://doi.org/10.1037/h0024833

Greenberg, J., Pyszczynski, T., & Solomon, S. (1982). The self-serving attributional bias:

Beyond self-presentation. Journal of Experimental Social Psychology, 18(1), 56–

67. https://doi.org/10.1016/0022-1031(82)90081-6

91

Greenwald, A. G. (1980). The Totalitarian Ego: Fabrication and Revision of Personal

History. American Psychologist, 35(7), 603–618.

Guckian, M. L., Chapman, D. A., Lickel, B., & Markowitz, E. M. (2018). “A few bad

apples” or “rotten to the core”: Perceptions of corporate culture drive brand

engagement after corporate scandal. Journal of Consumer Behaviour, 17(1), e29–

e41. https://doi.org/10.1002/cb.1672

Hafner, R., Elmes, D., & Read, D. (2017). Exploring the Role of Messenger Effects and

Feedback Frames in Promoting Uptake of Energy-Efficient Technologies.

Current Psychology. https://doi.org/10.1007/s12144-017-9717-2

Harth, N. S., Leach, C. W., & Kessler, T. (2013). Guilt, anger, and pride about in-group

environmental behaviour: Different emotions predict distinct intentions. Journal

of Environmental Psychology, 34, 18–26.

https://doi.org/10.1016/j.jenvp.2012.12.005

Harvey, J. H., & Weary, G. (1984). Current Issues in Attribution Theory and Research.

Annual Review of Psychology, 35(1), 427–459.

https://doi.org/10.1146/annurev.ps.35.020184.002235

Hayes, A. F. (2018). Introduction to mediation, moderation, and conditional process

analysis: a regression-based approach (Second edition). In Methodology in the

Social Sciences (Second edition). New York: Guilford Press.

Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in

OLS and logistic regression: SPSS and SAS implementations. Behavior Research

Methods, 41(3), 924–936. https://doi.org/10.3758/BRM.41.3.924

92

Hayes, A. F., & Montoya, A. K. (2017). A Tutorial on Testing, Visualizing, and Probing

an Interaction Involving a Multicategorical Variable in Linear Regression

Analysis. Communication Methods and Measures, 11(1), 1–30.

https://doi.org/10.1080/19312458.2016.1271116

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.

Hovland, C. I., Janis, I. L., & Kelley, H. H. (1953). Communication and ;

psychological studies of opinion change. In Communication and Persuasion;

Psychological Studies of Opinion Change. New Haven, CT, US: Yale University

Press.

International Energy Agency. (2018). Atlas of Energy. Retrieved from

http://energyatlas.iea.org/

Ito, T. A., Larsen, J. T., Smith, N. K. R., & Cacioppo, J. T. (1998). Negative Information

Weighs More Heavily on the Brain: The Negativity Bias in Evaluative

Categorizations. Journal of Personality and Social Psychology, 75(4), 887–900.

Kaiser, H. F. (1960). The Application of Electronic Computers to Factor Analysis.

Educational and Psychological Measurement, 20(1), 141–151.

https://doi.org/10.1177/001316446002000116

Karlin, B., Zinger, J. F., & Ford, R. (2015). The effects of feedback on energy

conservation: A meta-analysis. Psychological Bulletin, 141(6), 1205–1227.

https://doi.org/10.1037/a0039650

Katzev, R., & Johnson, T. (1987). Promoting energy conservation: An analysis of

behavioral research. Boulder, CO: Westview.

93

Lacasse, K. (2016). Don’t be satisfied, identify! Strengthening positive spillover by

connecting pro-environmental behaviors to an “environmentalist” label. Journal

of Environmental Psychology, 48, 149–158.

https://doi.org/10.1016/j.jenvp.2016.09.006

Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press.

Lewis, M. (2000). Self-conscious emotions: Embarrassment, pride, shame, and guilt. In

M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of emotions (2nd ed., pp.

623–636). New York: Guilford.

Mallett, R. K., Melchiori, K. J., & Strickroth, T. (2013). Self-Confrontation via a Carbon

Footprint Calculator Increases Guilt and Support for a Proenvironmental Group.

Ecopsychology, 5(1), 9–16. https://doi.org/10.1089/eco.2012.0067

Markowitz, E. M., & Malle, B. F. (2012). Did You Just See That? Making Sense of

Environmentally Relevant Behavior. Ecopsychology, 4(1), 37–50.

https://doi.org/10.1089/eco.2011.0044

Marschall, D., Sanftner, J., & Tangney, J. P. (1994). The state shame and guilt scale.

Fairfax, VA: George Mason University.

McAdams, D. P. (1995). What Do We Know When We Know a Person? Journal of

Personality, 63(3), 365–396. https://doi.org/10.1111/j.1467-6494.1995.tb00500.x

McAuley, E., Duncan, T. E., & Russell, D. W. (1992). Measuring Causal Attributions:

The Revised Causal Dimension Scale (CDSII). Personality and Social

Psychology Bulletin, 18(5), 566–573. https://doi.org/10.1177/0146167292185006

94

McGinnies, E., & Ward, C. D. (1980). Better Liked than Right: Trustworthiness and

Expertise as Factors in Credibility. Personality and Social Psychology Bulletin,

6(3), 467–472. https://doi.org/10.1177/014616728063023

Meleady, R., & Crisp, R. J. (2017). Redefining climate change inaction as temporal

intergroup bias: Temporally adapted interventions for reducing prejudice may

help elicit environmental protection. Journal of Environmental Psychology, 53,

206–212. https://doi.org/10.1016/j.jenvp.2017.08.005

Merritt, A. C., Effron, D. A., & Monin, B. (2010). Moral Self-Licensing: When Being

Good Frees Us to Be Bad: Moral Self-Licensing. Social and Personality

Psychology Compass, 4(5), 344–357. https://doi.org/10.1111/j.1751-

9004.2010.00263.x

Miller, D. T. (1976). Ego Involvement and Attributions for Success and Failure. Journal

of Personality and Social Psychology, 34(5), 901–906.

Nickerson, R. S. (1998). Confirmation Bias: A Ubiquitous Phenomenon in Many Guises.

Review of General Psychology, 2(2), 175–220.

Onwezen, M. C. (2014). How Pride and Guilt Guide Pro-Environmental Behaviour.

Wageningen University, Wageningen, NL.

Onwezen, M. C., Antonides, G., & Bartels, J. (2013). The Norm Activation Model: An

exploration of the functions of anticipated pride and guilt in pro-environmental

behaviour. Journal of Economic Psychology, 39, 141–153.

https://doi.org/10.1016/j.joep.2013.07.005

95

Osbaldiston, R., & Schott, J. P. (2012). Environmental Sustainability and Behavioral

Science: Meta-Analysis of Proenvironmental Behavior Experiments. Environment

and Behavior, 44(2), 257–299. https://doi.org/10.1177/0013916511402673

Osborne, J. W., & Costello, A. B. (2009). Best practices in exploratory factor analysis:

Four recommendations for getting the most from your analysis. Pan-Pacific

Management Review, 12(2), 131–146. Retrieved from Scopus.

Palan, S., & Schitter, C. (2018). Prolific.ac—A subject pool for online experiments.

Journal of Behavioral and Experimental Finance, 17, 22–27.

https://doi.org/10.1016/j.jbef.2017.12.004

Peer, E., Brandimarte, L., Samat, S., & Acquisti, A. (2017). Beyond the Turk: Alternative

platforms for crowdsourcing behavioral research. Journal of Experimental Social

Psychology, 70, 153–163. https://doi.org/10.1016/j.jesp.2017.01.006

Pieters, R., Bijmolt, T., van Raaij, F., & de Kruijk, M. (1998). Consumers’ Attributions

of Proenvironmental Behavior, Motivation, and Ability to Self and Others.

Journal of Public Policy & Marketing, 17(2), 215–225.

Rabinovich, A., Morton, T. A., & Birney, M. E. (2012). Communicating climate science:

The role of perceived communicator’s motives. Journal of Environmental

Psychology, 32(1), 11–18. https://doi.org/10.1016/j.jenvp.2011.09.002

Rees, J. H., Klug, S., & Bamberg, S. (2015). Guilty conscience: motivating pro-

environmental behavior by inducing negative moral emotions. Climatic Change,

130(3), 439–452. https://doi.org/10.1007/s10584-014-1278-x

96

Richardson, J. T. E. (2011). Eta squared and partial eta squared as measures of effect size

in educational research. Educational Research Review, 6(2), 135–147.

https://doi.org/10.1016/j.edurev.2010.12.001

Robins, R. W., & Schriber, R. A. (2009). The Self-Conscious Emotions: How are they

Experienced, Expressed, and Assessed? Social and Personality Psychology

Compass, 3(6), 887–898. https://doi.org/10.1111/j.1751-9004.2009.00217.x

Russell, D., & McAuley, E. (1986). Causal attributions, causal dimensions, and affective

reactions to success and failure. Journal of Personality and Social Psychology,

50(6), 1174–1185. https://doi.org/10.1037/0022-3514.50.6.1174

Schneider, C. R., Zaval, L., Weber, E. U., & Markowitz, E. M. (2017). The influence of

anticipated pride and guilt on pro-environmental decision making. PLOS ONE,

12(11), e0188781. https://doi.org/10.1371/journal.pone.0188781

Schultz, P. W. (2014). Strategies for Promoting Proenvironmental Behavior: Lots of

Tools but Few Instructions. European Psychologist, 19(2), 107–117.

https://doi.org/10.1027/1016-9040/a000163

Schultz, P. W., Messina, A., Tronu, G., Limas, E. F., Gupta, R., & Estrada, M. (2016).

Personalized normative feedback and the moderating role of personal norms: A

field experiment to reduce residential water consumption. Environment and

Behavior, 48(5), 686–710.

Spink, K. S., & Roberts, G. C. (1980). Ambiguity of Outcome and Causal Attributions.

Journal of Sport Psychology, 2(3), 237–244. https://doi.org/10.1123/jsp.2.3.237

97

Stets, J. E. (1995). Role Identities and Person Identities: Gender Identity, Mastery

Identity, and Controlling One’s Partner. Sociological Perspectives, 38(2), 129–

150. https://doi.org/10.2307/1389287

Stets, J. E. (2005). Examining Emotions in Identity Theory. Social Psychology Quarterly,

68(1), 39–56. https://doi.org/10.1177/019027250506800104

Stets, J. E., & Biga, C. F. (2003). Bringing Identity Theory into Environmental

Sociology. Sociological Theory, 21(4), 398–423. https://doi.org/10.1046/j.1467-

9558.2003.00196.x

Stets, J. E., & Burke, P. J. (2000). Identity Theory and Social Identity Theory. Social

Psychology Quarterly, 63(3), 224. https://doi.org/10.2307/2695870

Stets, J. E., & Trettevik, R. (2014). Emotions in Identity Theory. In J. E. Stets & J. H.

Turner (Eds.), Handbook of the Sociology of Emotions: Volume II (pp. 33–49).

https://doi.org/10.1007/978-94-017-9130-4_3

Stryker, S. (2004). Integrating Emotion into Identity Theory. In Advances in Group

Processes (Vol. 21, pp. 1–23). https://doi.org/10.1016/S0882-6145(04)21001-3

Stryker, S., & Burke, P. J. (2000). The Past, Present, and Future of an Identity Theory.

Social Psychology Quarterly, 63(4), 284. https://doi.org/10.2307/2695840

Suls, J., & Wills, T. A. (1991). Social comparison: Contemporary theory and research.

Hillsdale, NJ: Erlbaum.

Swann, W. B. (1983). Self-verification: Bringing social reality into harmony with the

self. In A. G. Greenwald & J. Suls (Eds.), Social psychological perspectives on

the self (Vol. 2, pp. 33–66). Hillsdale, NJ: Erlbaum.

98

Swann, W. B. (1987). Identity negotiation: Where two roads meet. Journal of Personality

and Social Psychology, 53(6), 1038–1051. https://doi.org/10.1037/0022-

3514.53.6.1038

Swann, W. B., Pelham, B. W., & Krull, D. S. (1989). Agreeable fancy or disagreeable

truth? Reconciling self-enhancement and self-verification. Journal of Personality

and Social Psychology, 57(5), 782–791. https://doi.org/10.1037/0022-

3514.57.5.782

Tangney, J. P. (2005). The Self-Conscious Emotions: Shame, Guilt, Embarrassment and

Pride. In T. Dalgleish & M. J. Power (Eds.), Handbook of Cognition and Emotion

(pp. 541–568). https://doi.org/10.1002/0470013494.ch26

Tangney, J. P., & Dearing, R. L. (2002). Shame and guilt. New York: Guilford.

Taylor, S. E. (1991). Asymmetrical Effects of Positive and Negative Events: The

Mobilization-Minimization Hypothesis. Psychological Bulletin, 110(1), 67–85.

Ter Mors, E., Weenig, M. W. H., Ellemers, N., & Daamen, D. D. L. (2010). Effective

communication about complex environmental issues: Perceived quality of

information about carbon dioxide capture and storage (CCS) depends on

stakeholder collaboration. Journal of Environmental Psychology, 30(4), 347–357.

https://doi.org/10.1016/j.jenvp.2010.06.001

The Berkeley Institute of the Environment. (2008). CoolClimate Carbon Footprint

Calculator. Retrieved from http://coolclimate.berkeley.edu/

99

Toner, K., Gan, M., & Leary, M. R. (2014). The Impact of Individual and Group

Feedback on Environmental Intentions and Self-Beliefs. Environment and

Behavior, 46(1), 24–45. https://doi.org/10.1177/0013916512451902

Tracy, J. L., & Robins, R. W. (2004). Putting the Self into Self-Conscious Emotions: A

Theoretical Model. Psychological Inquiry, 15(2), 103–125.

Tracy, J. L., & Robins, R. W. (2007). The psychological structure of pride: A tale of two

facets. Journal of Personality and Social Psychology, 92(3), 506–525.

https://doi.org/10.1037/0022-3514.92.3.506

Unsworth, K. L., & McNeill, I. M. (2017). Increasing pro-environmental behaviors by

increasing self-concordance: Testing an intervention. Journal of Applied

Psychology, 102(1), 88–103. https://doi.org/10.1037/apl0000155

U.S. Census Bureau. (2017). 2017 American Community Survey 1-Year Estimates.

Retrieved from

https://factfinder.census.gov/faces/nav/jsf/pages/searchresults.xhtml?refresh=t van der Werff, E., Steg, L., & Keizer, K. (2013). The value of environmental self-

identity: The relationship between biospheric values, environmental self-identity

and environmental preferences, intentions and behaviour. Journal of

Environmental Psychology, 34, 55–63.

https://doi.org/10.1016/j.jenvp.2012.12.006

Vandenbergh, M. P., Barkenbus, J., & Gilligan, J. (2008). Individual carbon emissions:

the low-hanging fruit. UCLA Law Review, 55, 1701–1758.

100

Weiner, B. (1985). An Attributional Theory of Achievement Motivation and Emotion.

Psychological Review, 92(4), 548–573.

Weiner, B. (1986). An attributional theory of motivation and emotion. New York:

Springer-Verlag.

Weiner, B., Graham, S., & Chandler, C. (1982). Pity, Anger, and Guilt: An Attributional

Analysis. Personality and Social Psychology Bulletin, 8(2), 226–232.

https://doi.org/10.1177/0146167282082007

Weinstein, N. D. (1980). Unrealistic Optimism About Future Life Events. Journal of

Personality and Social Psychology, 39(5), 806–820.

Whitmarsh, L., & O’Neill, S. (2010). Green identity, green living? The role of pro-

environmental self-identity in determining consistency across diverse pro-

environmental behaviours. Journal of Environmental Psychology, 30(3), 305–

314. https://doi.org/10.1016/j.jenvp.2010.01.003

101

Appendix A: Experimental Instrument

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

Appendix B: Chapter 2 Supplemental Information

Figure 9. Scree Plot for Causal Attribution Items.

121

Figure 10. Pride Distribution.

122

Figure 11. Guilt Distribution.

123

Table 11. Summary of ANCOVA Results (Dependent Variable = Internal Attribution).

Effect F df p η2 Frame Condition 13.14 2 <.01 .06 Credibility .28 1 .59 <.01 Condition Frame x .32 2 .73 <.01 Credibility Baseline PEB .30 1 .58 <.01 Environmentalist 9.41 1 <.01 .02 Identity Age 1.12 1 .29 <.01 Gender 2.60 1 .11 <.01 Race 1.23 1 .27 <.01 Political .85 1 .36 <.01 Orientation

(Insignificant) Effects of Source Credibility Manipulation on Pride, Guilt, PEI, and PEB

Two separate linear regression models with a bootstrap resampling routine with 5000 samples were used to evaluate the effect of source credibility condition (IV) on (1) pride

(DV) and (2) guilt (DV). Results demonstrated that there was no significant effect of source credibility condition on either pride, b = 0.13, percentile bootstrap 95% CI: -0.08,

0.34, or guilt, b = -0.09, percentile bootstrap 95% CI: -0.25, 0.08. Further, a one-way

ANOVA was used to compare the effect of source credibility condition (IV) on PEI (DV) and found the omnibus test was not significant, F(1, 395) = 0.30, p = 0.58. Similarly, a one-way ANOVA was used to compare the effect of source credibility condition (IV) on

PEB (DV) and found the omnibus test was not significant, F(1, 395) = 0.09, p = 0.77.

124

Table 12. Summary of Path Coefficients, Indirect Effects, Direct Effects, and Total Effects (Negative Feedback Frame Reference Group).

Independent Mediating Mediating Dependent Effect Effect Effect Effect Effect Direct Indirect Total variable (IV) variable variable variable of IV on of IV on of M1 of M1 of M2 effect effect effect (M1) (M2) (DV) M1 (a1) M2 (a2) on M2 on DV on DV (c’) (c) (a3) (b1) (b2) (a x 95% b) CI Positive IA - PEB .99* - - .01 - -.02 .007 (-.02, -.02 Feedback .03) Neutral IA - PEB .79* - - .01 - -.01 .01 (-.01, -.02 Feedback .02) Positive Pride - PEB 1.25* - - .03 - -.02 .03 (-.02, -.02 Feedback .09) Neutral Pride - PEB .29* - - .03 - -.01 .01 (- -.02 Feedback .004, .02) Positive Guilt - PEB -.75* - - .05* - -.02 - (-.08, -.02 Feedback .04* -.003) Neutral Guilt - PEB -.35* - - .05* - -.01 - (-.04, -.02 Feedback .02* -.001) Positive IA Pride PEB .99* 1.25* .03 .01 .03 -.02 .001 (- -.02 Feedback .001, .004) Neutral IA Pride PEB .79* .29* .03 .01 .03 -.01 .001 (- -.02 Feedback .001, .003) Positive IA Guilt PEB .99* -.75* -.03 .01 .05* -.02 - (- -.02 Feedback .001 .005, .001) Neutral IA Guilt PEB .79* -.35* -.03 .01 .05* -.01 - (- -.02 Feedback .001 .004, .001) Table Note: 5000 bootstraps. Path coefficients are unstandardized effects. IA = Internal Attribution; PEB = Pro-environmental Behavior. * Indicates significance of 95% percentile bootstrap CI. 125

Appendix C: Chapter 3 Supplemental Information

H1 and H2 Interaction Results with Neutral Feedback Frame Reference Group

The interaction term for positive feedback frame (with neutral as the reference group) and environmentalist identity was not significant in predicting pride, b = 0.12, 95% CI: -0.09,

0.34, or guilt, b = -0.08, 95% CI: -0.26, 0.10. The interaction term for negative feedback frame (with neutral as the reference group) and environmentalist identity was not significant in predicting pride, b = -0.18, 95% CI: -0.38, 0.03, or guilt, b = 0.12, 95% CI:

-0.07, 0.33.

H3 Moderated Mediation Results with Neutral Feedback Frame Reference Group

As established in Chapter 2, guilt mediates the relationships that positive feedback and negative feedback (with neutral feedback as the reference group for both) have with PEB.

For this reason, supplemental analyses were conducted to test for moderated mediation for these relationships. Though the overall index of moderated mediation was not significant for positive vs. neutral feedback, I = -0.004, 95% CI: -0.02, 0.005, or negative vs. neutral feedback, I = 0.006, 95% CI: -0.004, 0.02, the pattern of conditional indirect effects was similar. Specifically, among those low, moderate, and high in environmentalist identity, the conditional indirect effects of positive (vs. neutral) feedback frame on guilt were -0.01 (95% CI: -0.04, 0.001), -0.02 (95% CI: -0.05, -0.003), 126 and -0.03 (95% CI: -0.06, -0.003), respectively. Among those low, moderate, and high in environmentalist identity, the conditional indirect effects of negative (vs. neutral) feedback frame on guilt were 0.01 (95% CI: -0.005, 0.04), 0.02 (95% CI: 0.003, 0.05), and 0.03 (95% CI: 0.003, 0.06), respectively. Overall, these supplemental analyses demonstrate that the conditional indirect effects of positive (vs. neutral) and negative (vs. neutral) feedback on PEB through guilt are significant for those at moderate and high values of environmentalist identity only. In other words, guilt only explains the translation from feedback frame to PEB for those moderate and high in environmentalist identity, when using neutral feedback frame as the reference group.

127