EXPLAINING CONSERVATION BEHAVIORS TO REDUCE WILDLIFE CRIME: A CROSS- NATIONAL AND THEORETICAL COMPARISON By Apoorva Joshi A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Information and Media – Doctor of Philosophy Environmental Science and Policy – Dual Major 2022 ABSTRACT Wildlife crimes such as poaching and the illegal trade in wild animals and plants are globally prolific and may converge with other serious crimes including drugs and arms trafficking. Wildlife crime is a wicked problem, requiring interdisciplinary approaches to manage its far-reaching impacts on the environment, social justice, public health, the economy, governance, etc. Conservation organizations frequently use public communication and campaigns to draw the public’s attention to issues such as wildlife crime. These campaigns often urge audiences to take various actions such as donating money to support the organizations’ efforts to combat wildlife crime. Despite their reliance on communication to reach and engage their audiences, little is publicly known about the effects of such communication on conservation behaviors. In fact, we do not yet know which social-psychological factors matter most in driving behaviors linked to curbing wildlife crime. This dissertation compares the theory of planned behavior and the value- belief-norm theory – both widely used to explain environmental behaviors – and compares responses from a cross-national group of participants from India and the United States of America to provide new empirical evidence of how well each theory performs in predicting intentions to donate money to conservation organizations to reduce wildlife crime. It also explores the potential for using wildlife value orientations in studies that focus specifically on wildlife-related behaviors, rather than the New Ecological Paradigm scale that is used as part of the value-belief-norm model. Data was collected from a total of 1,820 participants, of which 900 were from the U.S. and 920 were from India. The survey instrument assessed responses to the primary constructs in the planned behavior and value-belief-norm theory. Data were analyzed using correlations, regressions, multi-group confirmatory factor analysis, and structural equation modeling. Results indicate that the theory of planned behavior explained greater variance in donation intentions for both Indian and U.S. groups but attitude toward behavior did not have a significant association with intention in the multivariate analysis in either sample. While the value-belief-norm theory explained lesser variance in donation intentions, and it held up as expected for the U.S. sample with good reliability metrics, scale reliability was low for the Indian sample and convergent validity was poor overall with values and ecological worldview dimensions manifesting unexpectedly in multivariate regressions. Measurement models and structural models were different for both samples, necessitating a parallel analysis. A modified value-belief-norm model with mutualistic wildlife value orientations had slightly better fit for the India sample but slightly lower for the U.S. sample. Perceived behavioral control was the strongest predictor for India while personal norms mattered most in the US sample. Key contributions of this dissertation to advancing theory and building new empirical knowledge in environmental communication and conservation social science research are discussed. In recognizing that several existing scales were developed by and for primarily Western audiences, this dissertation also underscores the importance of cross-validating measures and being inclusive of communities in non-Western emerging economies. Theoretical and practical implications of the results along with directions for future research and limitations are also discussed. Copyright by APOORVA JOSHI 2022 With immense gratitude, I dedicate this dissertation to the people who, with their unwavering support, unconditional love, and fierce belief in me, have empowered me to accomplish this goal. To my village – my fiancé, soulmate, and best friend – Pushkar; my parents – Dr. Suvarna and Prasanna Joshi; my grandparents – Smt. Vasanti and Col. Vidyadhar Dixit; my OG queen – Ginny; my siblings – Parashar and Shruti; my parents-in- law – Pradnya and Pradeep Pandit; our sweet fur baby – Burfee; and my dearest friends – Rahul, Manmeeta, Meenal, Aakash, Suhita, Devashri, Paroma, and Kajal – Thank you for everything! Having you in my corner means the world to me! v ACKNOWLEDGEMENTS Graduate students always advise incoming grad. students to choose mentors and committee members who will be helpful, kind, supportive, will challenge you in the right ways, and set you up for success. I came to MSU with no such advice, very much like a lost sea turtle trying to get to the ocean, but I hit the PhD Committee Jackpot on my way there. I have been advised, taught, and tested by four scholars who are unarguably bonafide experts in their respective fields, but who also happen to be wonderful people and mentors. Right from his response to the cold email I sent when I was considering applying for the Information and Media PhD program at MSU, I have been incredibly fortunate to have an advisor and mentor as kind, empathetic, supportive, and open-minded as Dr. Bruno Takahashi. I cannot count the number of times over the last six years when I have silently thanked my stars that I get to work with an advisor who not only encourages me to pursue my passion and curiosity, but also offers support, advice, and opportunities for me to grow and has shown immense faith in me. Bruno has advocated for me, has encouraged me to pursue professional development opportunities, to explore and expand on my skills as a scholar and an educator, and he has been an unwavering pillar of support not only during “normal” times, but most importantly, during a pandemic that tore through so many of our lives. I don’t know where I will be in 10 years, but I know that even then, I will cherish our equation and will do everything I can do remember and practice what Bruno has taught me about being a good advisor and a wonderful human being. New to the world of theory, academic research, and to Michigan, I felt out of my depth for the entirety of my first year as PhD student. In one theory class, I was questioning my choice of pursuing a PhD because of how complex theory seemed and how difficult it was vi for me to keep up with conversations. Then one afternoon, in the office of the professor teaching that class, I had my first “Aha!” moment. Dr. Serena Miller didn’t have to, but she chose to spend hours talking to me till everything that had been evading my understanding till that point began to make perfect sense. With her unassuming, inclusive, empowering approach to mentorship, she sparked a lifelong curiosity and love for theory in my mind, and made me believe that I could, in fact, do this. Over the years, Dr. Miller has taken the initiative to coach me through some of the most important hurdles of being an academic and has consistently been one of the strongest allies for my peers and I in the program. I consider myself very fortunate to have learned so much from her, especially the high benchmark she sets for service to students, field, science, and society. As a dual major PhD student, and as part of the Conservation Criminology Lab, I took classes with and got to know some amazing faculty members and fellow students in the Environmental Science and Policy Program, and the School of Criminal Justice at MSU as well. Some of the defining moments of my doctoral career have come from classes, brainstorming sessions, conversations, meetings, and research collaborations with Dr. Adam Zwickle, who has always been approachable and generous with his time, advice, and insight. I learned quickly to bring a notebook and pen to every meeting with Adam because, for an as-yet unknown reason, my brain begins generating idea after idea or finds solutions to challenges I’ve been struggling with every time we have a conversation about our interests. I will always credit Adam for helping me identify and develop my interest in and passion for studying risk and decision science, and for helping me turn my interdisciplinary thinking and background into my strengths. vii I would also like to acknowledge the continued support and insight I have received from Dr. John Besley with whom most conversations are like shots of brain energy that provide the kind of sustained intellectual fodder that researchers crave and cherish. His expertise, feedback, and his ability to see patterns or connections – both meta and micro – has helped me not only refine this dissertation but has also taught me to be confident and assertive of my own expertise, and to be unafraid to challenge the status quo or push boundaries as a scholar seeking to benefit both science and society. On this doctoral journey, I have met, worked with, and learned from numerous people who made a lasting impact on me, to whom I would like to express my sincere gratitude for their implicit and explicit support and guidance along the way. Dr. Julie Libarkin, Dr. Mark Axelrod, and Dr. Joe Hamm have been incredibly warm and kind with their time, insights, feedback, and support both as instructors and as mentors. I am very fortunate to have been able to have been able to expand on and nurture my passion for interdisciplinary research through the conversations and ideas I was able to discuss with them over the years. My friends, colleagues, and professors associated with the Environmental Science and Policy Program have been instrumental in shaping both my academic experience and my professional growth. From the warmth and friendliness that Karessa Weir Wheeler extended to me as I was trying to adapt to life in a new place, to the spirited discussions in ESP courses, the numerous coffee chats and lunch sessions, and the annual research conference and colloquium events, I have truly felt welcome, included, and empowered in my pursuit of the ESPP doctoral dual major. viii In the School of Journalism and the College of Communication Arts and Sciences, I have found incredible mentors, friends, teachers, colleagues, students, staunch supporters, and an extended family that I am very grateful to be a part of. I am thankful to Dr. Lucinda Davenport, Dr. Tim Vos, Dr. Eric Freedman, Prof. Dave Poulson, Prof. Judy Walgren, Dr. Esther Thorson, and Dr. Patricia Huddleston for their support and advice. I would also like to thank the School of Journalism, the Information and Media PhD program, the College of Communication Arts and Sciences, and the Environmental Science and Policy Program for supporting my research and this dissertation through scholarships, graduate assistantships, awards, fellowships, travel grants, and by providing frequent opportunities for professional growth. The administrators and staff at the J-School, ComArtSci offices, the MSU Library, and ESPP deserve special thanks for the hours of tireless work they have put in to get so much done, largely behind to scenes. Thank you, also, to those who made this PhD voyage less choppy and a lot more enjoyable with their camaraderie, peer support, and friendship – Ran Duan, Tony van Witsen, Zhao Peng, Tara Jean Ward Mesyn, Sevgi Baykaldi, Soo Young Shin, Swati Misra, Mel Medeiros, the Association for Doctoral Students in Information and Media (ADSIM), and the ESPP Research Colloquium Organizing Committee. I would also like to extend special thanks to the Center for Statistical Training and Consulting (CSTAT) at MSU, specifically, Barry DeCicco, for providing timely and valuable statistical analysis consultations for this dissertation, and to the College of Communication Arts and Sciences for making these services available to ComArtSci graduate students. Finally, while this dissertation is dedicated to them, I would be remiss not to acknowledge the countless sacrifices that my family has made to support me on this ix journey. I do not know of a single person who is as selflessly caring as my fiancé, who has dealt with my stress, anxiety, and a whole lot more while bearing the larger burden for a substantial amount of time and has done so with dignity, love, and mind-boggling levels of resilience and ease, all while building up my spirits and helping me chase my dreams. I would simply not have been able to complete this PhD without his strength and faith, and without the empowered upbringing my parents gave me. On days when I found it difficult to get back up, they rallied around me and reminded me what I’m capable of. They have always led by example and inspired me to be better and do better. Being away from friends and family was already difficult without a pandemic thrown into the mix during which time I lived in perpetual fear of losing those closest to me, and in those moments I learned that even when faced with the harshest of adversities, I don’t lose sight of what’s most important to me and my never-give-up spirit was still intact. This PhD is a result of just that – not giving up. The love, faith, and support of my family and friends is and will always be the fuel that keeps me going. x TABLE OF CONTENTS CHAPTER ONE: INTRODUCTION .................................................................................................................... 1 Background......................................................................................................................................................... 1 Objectives and Intellectual Merit ............................................................................................................... 6 Dissertation Outline ........................................................................................................................................ 7 CHAPTER TWO: LITERATURE REVIEW ....................................................................................................... 9 Conservation Messaging and Intentions to Donate to Wildlife Crime Reduction ................... 9 Comparing Two Theories of Environmental Behavior ................................................................... 12 Cross-national Wildlife Crime .................................................................................................................. 22 Wildlife Value Orientations ....................................................................................................................... 25 Research Questions ...................................................................................................................................... 30 CHAPTER THREE: METHOD, MEASURES, AND ANALYSIS ................................................................ 31 Survey Design, Sampling, and Procedure ............................................................................................ 31 Measures........................................................................................................................................................... 36 Expected Outcomes ...................................................................................................................................... 44 Pilot Test and Rationale .............................................................................................................................. 46 Analysis and Main Test ............................................................................................................................... 47 CHAPTER FOUR: RESULTS ............................................................................................................................. 49 Exploratory Factor Analysis ...................................................................................................................... 49 Value-Belief-Norm Theory......................................................................................................................... 50 Theory of Planned Behavior ..................................................................................................................... 69 Other Hypotheses .......................................................................................................................................... 81 Wildlife Value Orientations in the Value-Belief-Norm Theory .................................................... 83 Relationships That Mattered Most – Graphical Representations ............................................... 93 Predictors That Mattered the Most ........................................................................................................ 97 Broad Overview of Responses and Demographics........................................................................... 98 CHAPTER FIVE: DISCUSSION ...................................................................................................................... 105 Research Questions and Objectives ..................................................................................................... 105 Key Findings .................................................................................................................................................. 106 Theoretical Implications........................................................................................................................... 108 Practical Implications for Conservation Communication............................................................ 116 Key Contributions ....................................................................................................................................... 122 Future Research ........................................................................................................................................... 124 Limitations ..................................................................................................................................................... 126 REFERENCES ..................................................................................................................................................... 129 APPENDIX A: MULTIVARIATE NORMALITY TESTS IN R ................................................................. 142 APPENDIX B: THEORY OF PLANNED BEHAVIOR CFA MODEL FIT STATISTICS .................... 143 xi APPENDIX C: THEORY OF PLANNED BEHAVIOR SEM MODEL - INDIA ..................................... 150 APPENDIX D: THEORY OF PLANNED BEHAVIOR SEM MODEL - USA......................................... 153 APPENDIX E: VALUE-BELIEF-NORM THEORY SEM MODEL - INDIA .......................................... 156 APPENDIX F: VALUE-BELIEF-NORM THEORY SEM MODEL – USA.............................................. 162 APPENDIX G: VBN SEM WITH WILDLIFE VALUE ORIENTATIONS - USA .................................. 169 APPENDIX H: VBN SEM WITH WILDLFIE VALUE ORIENTATIONS - INDIA ............................. 175 APPENDIX I: SURVEY INSTRUMENT ........................................................................................................ 181 APPENDIX J: CROSS-NATIONAL DIFFERENCES IN RESPONSES TO ITEMS .............................. 188 xii CHAPTER ONE: INTRODUCTION Background Across the world, wild animals and plants are traded every day – in-person and online – by sellers, intermediaries, buyers, fixers, etc. with an entire supply chain built to maintain the flow of live wildlife such as snakes, spiders, exotic birds, orchids, turtles, seahorses, etc. and parts or products made from wildlife such as elephant tusk ivory, rhino horn, pangolin scales, tiger or leopard skins, lion bones, mongoose hair, shark fins etc. (Kasper et al., 2020; Nijman et al., 2022). Wildlife or their parts and products are traded for various purposes; some people consume these parts or animals and plants as food; others believe that they hold medicinal properties and use them in traditional medicines; and there are also those who buy them as jewelry, fashion items, trophies or other kinds of status symbols, or to keep as pets (Anagnostou & Doberstein, 2022; UNODC, 2020). A substantial amount of this international trade, however, is illegal. Crimes such as poaching (i.e., illegal hunting) and the trafficking or smuggling of wild animals, plants, and their parts or derivatives are often collectively referred to as the illegal wildlife trade (IWT) or as wildlife crimes (WLC). Globally, wildlife crimes are among the most lucrative and widespread criminal operations sometimes involving not only organized criminal syndicates, but now also known to frequently converge with other serious crimes such as the trafficking of drugs, weapons, and people, as well as crimes such as corruption, money laundering, fraud, murder, etc. (Anagnostou & Doberstein, 2022; Brashares et al., 2014; Dalpane & Baideldinova, 2022; Duffy & Brockington, 2022; Nellemann et al., 2014; UNODC, 2020). 1 Contrary to what the name suggests though, wildlife crimes do not merely have ecological impacts. While they certainly are one of the primary contributors of species extinction and biodiversity loss (O’Donoghue & Rutz, 2016), wildlife crimes also exacerbate issues of governance, law enforcement, policymaking, safeguarding human rights, conflict resolution, economic and social stability, animal welfare, environmental and social justice, and, as became evident during the SARS-CoV-2 (COVID-19) pandemic, they are very much also an issue of public health (Anagnostou & Doberstein, 2022; Bezerra-Santos et al., 2021; Brashares et al., 2014; Duffy & Brockington, 2022; Kahler & Gore, 2015; Nellemann et al., 2014; Wyatt et al., 2021; Zucca et al., 2020). Due to the complex nature of the causes and consequences of wildlife crimes, they are considered ‘wicked problems’ – the kind of issues that are not “solved” but rather, must be studied and managed through holistic interdisciplinary approaches (Giannetta, 2017; Levin et al., 2012). Despite emerging interdisciplinary research on illegal wildlife trade and wildlife crimes from criminologists, economists, conservation biologists, and conservation social scientists (Bennett et al., 2017; Blair et al., 2017; Boratto & Gibbs, 2019), a communication-based perspective is markedly rare (Giannetta, 2017; Glenn et al., 2019; Kachen & Krishen, 2020). This is one of the knowledge gaps that this dissertation addresses. A communication-based perspective is both necessary and important to the interdisciplinary study of wildlife crime because some of the most commonly deployed demand-reduction interventions to prevent and reduce wildlife crimes around the world are persuasive and strategic communication campaigns by conservation organizations, non-profits, and alliances. Conservation organizations use public communication and social 2 marketing campaigns to draw the public’s attention to wildlife crime and its impacts (Duthie et al., 2017; Greenfield & Veríssimo, 2019; Lundberg et al., 2019; Veríssimo, 2019; Veríssimo et al., 2017, 2018; Veríssimo & Wan, 2019). These campaigns may be aimed at educating or influencing audiences, often urging them to take various actions such as donating money to support the organizations’ efforts to combat wildlife crime, or asking people to click on a web link to learn more about the issue or the organization’s work to address it (Joshi et al., under review). Studying how wildlife crime is communicated and how that communication may influence emotional, cognitive, and behavioral responses among target audiences is an important way to assess the success of conservation campaigns, and to identify the salient factors that might explain, drive, or impede audience engagement. Currently, theories of environmental communication and environmental behavior have not been tested in the context of conservation behaviors associated with wildlife crime prevention. If conservation communication professionals and conservation social scientists want to understand what drives these behaviors, researchers will need to apply and test the predictive and explanatory powers of various theories to identify the predictors that matter most for different audience segments. This dissertation is a first step in that direction. Additionally, this dissertation helps meet the need for and underscores the value of publicly available, independent evaluations of the efficacy of such conservation communication campaigns (Khanwilkar et al., 2022; Veríssimo & Wan, 2019). Many campaigns may aim to secure public donations for conservation organizations working on combating wildlife crime, among other issues. Considering that conservation organizations rely on public donations to support their programs, key questions need to be answered: 1) How is wildlife crime communicated as a problem? 2) Which social- 3 psychological factors are salient influencers of conservation behaviors to reduce wildlife crime? 3) Through which mechanisms does wildlife crime-related communication influence behaviors to reduce wildlife crime? While a previous study (Joshi et al., under review) partly answers the first question through a content analysis, this dissertation addresses some of these knowledge gaps by answering the second question using a survey, and it sets the stage for a follow-up study to answer the third question through an experiment. This dissertation is based on a key finding from the previous study (Joshi et al., under review) that conservation organizations most frequently call for information-seeking and monetary donation behaviors in visual wildlife crime prevention ads. Despite valuable work on conservation social marketing and behavior change campaigns (Greenfield & Veríssimo, 2019; Veríssimo, 2019), communication scholars have rarely investigated how, for example, salient social-psychological factors may drive behaviors to reduce wildlife crime that audiences may adopt upon exposure to communication from conservation organizations (Giannetta, 2017). Previous research largely suggests that environmental behaviors may be motivated by factors such as an individual's self-interest or intent to perform a behavior; their attitude toward the behavior; their values, beliefs, or norms, etc. (Bamberg & Möser, 2007; Kaiser et al., 2005; Kidd et al., 2019; Shreedhar & Mourato, 2018; Thomas-Walters et al., 2019). Although factors that drive conservation behaviors to reduce wildlife crime have not specifically been studied before, research from natural resource management and human dimensions of conservation suggests that willingness to donate or pay for conservation may be influenced by factors such as values, subjective norms, attitudes, environmental 4 worldviews, and personal norms (Ateş, 2020; Cárdenas & Lew, 2016; López-Mosquera et al., 2014; López-Mosquera & Sánchez, 2012; Martín-López et al., 2007b; Sánchez-García et al., 2021; Wakita et al., 2019). These are some of the main constructs within two social- psychological theories that have successfully been used to explain environmental behaviors consistently: the theory of planned behavior (TPB) (Ajzen, 1991) and the value- belief-norm theory (VBN) (Dietz et al., 2005; Stern et al., 1999). In this dissertation, I contend that donating money to conservation organizations to help reduce wildlife crime represents the kind of environmental behavior that may be explained by the intention-based premise of the theory of planned behavior (TPB) (Ajzen, 1985, 1991), or by the pro-social motive premise of the value-belief-norm theory (VBN) (Dietz et al., 2005; Stern et al., 1999). In comparing and contrasting the explanatory power of the theory of planned behavior (TPB) with the value-belief-norm theory (VBN), and by using structural equation modeling to evaluate how the main constructs in both these models operate together, I demonstrate whether and to what extent the planned behavior and value-belief-norm models can be applied to studying conservation behaviors for reducing wildlife crime – a novel context and application for both these theories. In addition to testing the full TPB and VBN models, I will examine the relationship between wildlife value orientations and wildlife crime-related environmental behaviors as evidence from research on human dimensions of wildlife suggests that wildlife value orientations (WVOs) can predict wildlife-related attitudes and behaviors (Manfredo, Teel, Carlos, et al., 2020; Teel et al., 2007b; Teel & Manfredo, 2010). I will also examine whether WVOs may be a stronger, more accurate measure of ecological worldviews in this wildlife 5 crime-related context, compared to the New Ecological Paradigm (Dunlap, 2008) largely used to assess ecological worldviews in the VBN model. As most environmental social science research has tended to focus on Western and developed countries, it has left many countries in which issues like wildlife crime are widespread under-represented in scholarship. By testing the two theories in a cross- national survey in India and the U.S.A., I will provide new empirical evidence of how well the theory of planned behavior (TPB) and the value-belief-norm theory (VBN) can explain donation behaviors among Indian and U.S. respondents, specifically, as they relate to reducing wildlife crime. The cross-national nature of this dissertation and the application of a communication-based approach to the study of wildlife crime-reduction help to advance and expand the purview of environmental communication scholarship from a more diverse and comparative perspective (Cordano et al., 2011; Thaker, 2021). This dissertation will also add to the small but growing body of work in conservation social science, particularly as it relates to communication and wildlife crime (Bennett et al., 2017; Giannetta, 2017; Megias et al., 2017; Militz & Foale, 2017; Nijman & Nekaris, 2017). Objectives and Intellectual Merit The knowledge gaps that this dissertation addresses are: 1) The lack of communication-based research on wildlife crime prevention; 2) The lack of representation of non-Western audiences in interdisciplinary social scientific research on wildlife crime; 3) The identification of salient social-psychological factors that may drive wildlife crime- reducing conservation behaviors and may influence the success of conservation communication campaigns aimed at combatting wildlife crime. 6 The broad objectives of this dissertation are to: 1) Determine the theoretical structure of behaviors linked to reducing wildlife crime, such as donating money to conservation organizations ; and 2) Explore how findings differ in an international sample of respondents from India and the U.S.A. The key contributions of this dissertation are that – 1) The survey provides new empirical evidence of the salient beliefs, values, attitudes, norms, and intentions related to wildlife crime prevention which will help determine the theoretical structure of conservation behaviors to reduce wildlife crime; 2) A cross-national comparison adds new knowledge by examining a non-Western perspective on an issue largely under-studied by communication scholars globally (Takahashi, Duan, et al., 2021); 3) The theoretical comparison demonstrates how well two widely-applied theories of environmental behavior – the TPB and VBN – explain donation intentions in the novel context of wildlife crime reduction; and how each theory performs in a cross-national sample; and 4) Whether wildlife value orientations (WVOs) can replace the New Ecological Paradigm scale (NEP) as a measure of ecological worldviews, specifically in the context of wildlife-related behaviors as predicted or explained by the value-belief-norm (VBN) model (Manfredo, Teel, Don Carlos, et al., 2020). Dissertation Outline This dissertation consists of five chapters including this introduction chapter. Chapter two provides an in-depth review of interdisciplinary literature relevant to the primary questions posed in this dissertation. Chapter three presents the methods, measures, and analytical approaches along with expected outcomes. Chapter four presents the results of the statistical analyses, and the last chapter discusses the implications of these findings for 7 theory, practice, and future research. Supplemental materials are included in appendices that follow the discussion, while a list of references cited is provided at the end. 8 CHAPTER TWO: LITERATURE REVIEW In this chapter, I synthesize research from mass communication, natural resource management, criminology, conservation biology, environmental psychology and social psychology that helps set the stage for understanding the scientific foundations of a behavior such as donating money to conservation organizations for reducing wildlife crime. First, I review the limited research there currently is on conservation communication to identify the gaps that this dissertation helps fill. Next, I discuss the two main theoretical frameworks applied and tested in this dissertation: the theory of planned behavior, and the value-belief-norm theory. In comparing the two theories, I identify the variables of interest and provide conceptual definitions for key constructs. I then provide an overview of the state of wildlife crime as a global issue but more specifically, its prevalence in the two countries studied in this dissertation: India and the United States. I demonstrate why it is necessary to study conservation behaviors associated with reducing wildlife crime. Finally, I describe wildlife value orientations and situate them within the broader goal of this dissertation while outlining how this construct can potentially enhance the predictive and explanatory power of the value-belief-norm theory – thereby testing a potential modification of the theory in a cross-national survey. Conservation Messaging and Intentions to Donate to Wildlife Crime Reduction In a meta-analysis on conservation messaging literature, Kidd et al. (2019) found that one of the primary outcome variables in journal articles on conservation messaging was raising funds, and another was encouraging behavior change. Additionally, they found that most experimental studies only measured intentions to donate despite how much conservation organizations rely on actual donation behavior or actions from their 9 stakeholders (Thomas-Walters et al., 2020). This dissertation while also identifying factors that may play a significant role in explaining monetary donation intentions – does so specifically in the context of reducing wildlife crime, so that future research can build on the findings in this dissertation to measure actual monetary donation behavior as an outcome variable and determine which factors drive that behavior in diverse regions and how. In one of the few studies on this topic, Giannetta (2017) used a relatively small convenience sample to administer surveys in the United States to examine whether communicating the complex nature of wicked problems such as wildlife crime can increase the perceived importance of such an issue. The surveys tested messages that included contextual information about how the poaching of African elephants for their ivory has been associated with social conflict, terrorism, violence, and human rights issues etc., and simple messages including a one-line statement on how the poaching of African elephants for their ivory was pushing them towards extinction. While a larger number of respondents ranked the contextual message as being very and highly important, there was no statistical difference between the messages. However, it is worth noting here that this was not an experimental study, and potential confounds were not addressed or accounted for. For instance, U.S. survey respondents could be subject to the effects of psychological distance when considering messages about African elephants and orangutans in Southeast Asia; there could be boomerang or psychological reactance effects or priming effects to the detailed message about converging crimes and threats; and the contextual message on social, political, and economic harms associated with wildlife crime could also have been processed in a way that made respondents feel like there was nothing they could do to help 10 address the issue. While these findings are not generalizable, I contend that Giannetta’s study is a step in the right direction because not only does it acknowledge that communication about issues such as wildlife crime will likely need to acknowledge and address the wicked nature of this issue by making the non-environmental aspects (e.g., economy, health, conflict, justice, politics, etc.) salient to audiences, but it also provides important empirical insight into how respondents – even a small sample of them – are likely to perceive the importance of an issue such as poaching/wildlife crime when communicators make decisions to include or exclude specific information. Giannetta (2017) therefore, provides a helpful block on which future cross-national and experimental research examining the role of communication in influencing the adoption of conservation behaviors linked to reducing wildlife crime can be built, thereby facilitating a greater understanding of the social-psychological factors that may drive or dampen individuals’ intentions to donate money to conservation organizations for reducing wildlife crime in response to conservation messaging. As communication-based research on wildlife crime-reducing conservation behaviors is only beginning to scratch the surface of this area of scholarship, to begin explaining donation intentions to reduce wildlife crime, I turned to research from natural resource management, conservation social science, and human dimensions of wildlife that has examined a ‘close cousin’ pair of constructs – willingness to pay and willingness to donate to wildlife conservation. In doing so, I found that factors from two widely applied social psychological theories of environmental behavior – the theory of planned behavior (TPB) (Ajzen, 1991), and the value-belief-norm theory (VBN) (Stern et al., 1999) – have consistently been associated with these conservation donation intentions. 11 For instance, studies on psychometric predictors of willingness to pay suggest that altruistic and biospheric values, and subjective norms positively influence willingness to pay for the environment (Wakita et al., 2019). Attitudes have also been identified as a significant driver of willingness to donate to conservation (Martín-López et al., 2007a). Environmental worldviews, such as those represented by the New Ecological Paradigm, as well as beliefs about ascribed responsibility have been associated with willingness to pay for conservation (Cárdenas & Lew, 2016). Primary constructs from both the planned behavior and the value-belief-norm models – such as attitudes, perceived behavioral control, subjective norms, value orientations, beliefs, and personal norms – have all been significantly associated with willingness to pay in an environmental context (Ateş, 2021; López-Mosquera, 2016; López-Mosquera et al., 2014; López-Mosquera & Sánchez, 2012; Sánchez-García et al., 2021). Thus, a logical starting point for explaining intentions to donate money to conservation organizations for reducing wildlife crime is to apply these two validated behavioral theories to this novel context of wildlife crime reduction, and assess the explanatory power of each theory, eventually identifying the potential for developing an integrated model that may offer a more enhanced predictive potential (Klöckner, 2013). Comparing Two Theories of Environmental Behavior Environmental behaviors can be planned or unplanned. Primarily, existing explanations of environmental or pro-environmental behaviors have been offered through either the planned behavior (TPB) lens which suggests that intentions are the closest predictors of behaviors (Ajzen, 1991), or the value-belief-norm (VBN) lens which suggests the pro-social motives or norms are the closest determinants of environmental behaviors 12 (Dietz et al., 2005; Stern et al., 1999). Over the years, research has also suggested that environmental behaviors might be a composite of both rational-thinking (intentions) and pro-social motives (Bamberg & Möser, 2007; Klöckner, 2013; Steg et al., 2014; Steg & Nordlund, 2018). Given the lack of prior research in this area, however, we do not currently know which of these factors can explain or predict conservation behaviors to reduce wildlife crime, such as donating money to conservation organizations. Therefore, to build new knowledge on factors explaining such behaviors, I begin by comparing the theory of planned behavior (Ajzen, 1991) and the value-belief-norm theory (Dietz et al., 2005; Stern et al., 1999) to understand how well these two theories explain donation intentions to help reduce wildlife crime in a cross-national sample of participants. In the following section, I present a review of the two theories, their key constructs, the conceptual definitions, and a critical analysis of how these theories relate to and complement one another. Theory of Planned Behavior The theory of planned behavior (TPB) (Ajzen, 1991) suggests that factors related to self-interest, such as positive attitudes toward a behavior, favorable social normative perceptions about a behavior, and greater perceived behavioral control increase an individual’s intentions to perform a behavior (Ajzen, 1991; Kaiser et al., 2005). Figure 1 shows the primary constructs in the TPB and the relationships between these constructs and behavior. 13 Figure 1 - Conceptual model of the theory of planned behavior Behavioral beliefs are defined as “beliefs about the likely consequences of the behavior” (Ajzen, 2006, p.1), and outcome evaluation beliefs are defined as the extent to which the behavioral outcome is evaluated as good or bad (Ajzen, 2006). Together, behavioral and outcome evaluation beliefs influence attitudes. Attitudes towards the behavior are defined as “the degree to which a person has a favorable or unfavorable evaluation or appraisal of the behavior in question,” (Ajzen, 1991, p.188); they reflect the positive or negative evaluations of the target behavior and its outcomes (Ajzen, 2006; Kaiser et al., 2005; López-Mosquera & Sánchez, 2012). Normative beliefs are defined as beliefs about the normative expectations of others (Ajzen, 2006, p.1), and they refer to both injunctive as well as descriptive normative beliefs. The latter describes the extent to which an individual believes that people close to them are performing the behavior in question. If a person believes that those who are important to them would approve of them performing a behavior, they would be more likely to perform it (Kaiser et al., 2005). Motivations to comply are described as the extent to which an 14 individual wants to comply with the injunctive norms of their referent group or the people important to them (Ajzen, 2006; Montaño & Kasprzyk, 2008). Together, normative beliefs and motivations to comply influence subjective norms which can be described as the strength of one’s perceptions that those relevant to them expect them to perform the behavior (Ajzen, 2006; López-Mosquera & Sánchez, 2012). Ajzen (1991, p.188) defines subjective norms as “the perceived social pressure to perform or not to perform the behavior.” Control beliefs, or beliefs about the presence or absence of certain factors that may facilitate or impede behavioral performance and the perception that a salient control factor will enable or inhibit a person’s ability to perform the behavior together determine perceived behavioral control (Ajzen, 2006; Montaño & Kasprzyk, 2015). Perceived behavioral control is described as “the perceived ease or difficulty of performing the behavior,” (Ajzen, 1991, p.188) which is expected to “reflect past experience as well as anticipated impediments and obstacles” to performing the behavior (Ajzen, 1991, p.188; Han, 2015; López-Mosquera & Sánchez, 2012). Behavioral intention, the closest predictor of behavior as per the TPB, is described as the willingness to perform the target behavior (Ajzen, 1991; Kaiser et al., 2005). In this dissertation, behavioral intention is conceptually defined as “an indication of how hard people are willing to try, of how much of an effort they are willing to exert, in order to perform the behavior,” (Ajzen, 1991, p.181; Han, 2015). While this model of intentional behavior has received consistent and strong empirical support, it has been criticized for undervaluing the role of morality or personal norms in predicting environmental behavior (Klöckner, 2013). 15 While this model of intentional behavior has received consistent and strong empirical support, it has been criticized for undervaluing the role of morality or personal norms in predicting environmental behavior (Klöckner, 2013). Value-Belief-Norm Theory The value-belief-norm theory (VBN), an extension of Schwartz's (1977) work on norm activation suggesting that people behave altruistically when they feel morally obligated to do so, posits that environmental behaviors are a product of pro-social factors such as values, ecological worldviews, outcome expectations, problem awareness, and personal norms (Dietz et al., 2005; Schwartz, 1977; Stern et al., 1999). Schwartz (1977) suggested that for people to act out of moral obligation, their personal norms have to be activated which requires that four conditions be met – one must be aware of the need for help; one must be aware of the consequences of the behavior in question for those in need; one must accept responsibility to act; and one must believe they are capable of performing the helpful behavior (Klöckner, 2013). Experts have suggested that these relationships explaining altruistic behavior also apply to explaining environmental behaviors because, as Thøgersen argues, environmental behaviors are a kind of moral behavior so they should not be conceptualized as an outcome of simple cost-benefit analyses, but rather, as also being driven by moral beliefs (Thøgersen, 1996). 16 Figure 2 - Conceptual model of the value-belief-norm theory The VBN, therefore, suggests that environmental behaviors that align with one’s normative pressures or expectations are likely to be performed when the person feels morally responsible for the outcomes and is aware of the consequences of action or inaction (van Riper & Kyle, 2014a; Carfora et al., 2021). Conceptually, values can be defined as beliefs related to a desirable end state or behavior that transcend specific situations, serve as guiding principles when it comes to selecting or assessing behaviors, and vary in their levels of importance (Han, 2015; Schwartz, 1992, p.4). The importance of values to environmental behaviors has been repeatedly established (De Groot & Steg, 2008; Gkargkavouzi et al., 2019; Steg et al., 2005; Stern et al., 1999; van der Werff & Steg, 2016), yet their ability to explain or predict conservation behaviors to reduce wildlife crime remains unknown. According to the VBN, people whose values align with the outcome or behavior, those who believe that valued objects are threatened and that their actions – or lack thereof – can help alleviate those threats, and those who feel morally obligated to take action, are more likely to perform a 17 behavior (Gkargkavouzi et al., 2019; Steg et al., 2005; Stern et al., 1999). The VBN model includes five dimensions of values of which three are tested in this dissertation. These are – a) Biospheric values focus on the benefits of environmental behaviors to non- human species and the biosphere such that individuals with this value orientation believe that protecting nature and environmental conservation are important (De Groot & Steg, 2008; Stern et al., 1999; van Riper & Kyle, 2014a). Dimensions of biospheric values include prevention of pollution, respecting the earth, unity with nature, and protecting nature (De Groot & Steg, 2008); b) Altruistic values focus on other humans and society itself such that individuals with altruistic value orientations value the benefits of environmental behaviors for other people (Stern et al, 1999). The dimensions of altruistic values include equality, world peace, social justice, and social welfare (De Groot & Steg, 2008); and c) Egoistic values are related to self-interest such that individuals with this value orientation value environmental behavior for benefit to oneself, one’s personal well-being, or one’s family and friends. Individuals with egoistic values are likely to act pro-environmentally when the benefits of an environmental behavior outweigh the perceived costs (Jansson et al., 2010; Stern et al., 1999; van Riper & Kyle, 2014a). Dimensions of egoistic values include social power, wealth, authority, and influence (De Groot & Steg, 2008). The VBN theory suggests that biospheric and altruistic values are positively related to ecological worldviews, while egoistic values are negatively related to ecological worldviews (Gkargkavouzi et al., 2019). Ecological worldviews are described as overarching beliefs 18 about the relationship between humans and the natural environment (Steg et al., 2005; Stern et al., 1999) which are positively associated with an individual’s predisposition to act pro-environmentally (Han, 2015). Most prior research has operationalized ecological worldviews by determining the extent to which an individual subscribes to the New Ecological Paradigm (NEP), thereby measuring this concept with the widely used NEP scale (Dunlap et al., 2000; van Riper & Kyle, 2014a). In the VBN model, ecological worldviews are conceptualized as being positively associated with awareness of consequences – a concept I explicate next. Within VBN-related literature, the concept of awareness of consequences has been defined inconsistently. While some researchers have described it as one’s awareness of the adverse consequences of events for valued objects (Kaiser et al., 2005; Stern et al, 1999), the manner in which this conceptualization has been characterized operationally and measured has been very similar to how the concept of attitudes within TPB has been defined, operationalized, and measured at times. Addressing these inconsistencies, Klöckner (2013) says that, “Empirically speaking, awareness of need and awareness of consequences often blend together, which has led to several researchers either using awareness of need or awareness of consequences in their studies,” (pp.10). Since this dissertation is a comparison of the TPB and the VBN, establishing discriminant validity is especially crucial. Therefore, awareness of consequences – in this dissertation – is conceptualized in line with Han’s (2015) definition – as one’s awareness of adverse consequences to valued objects – which has sometimes been labelled, and indeed, operationalized as awareness of problems or problem awareness (Bamberg & Möser, 2007; 19 Han, 2015; Steg & Nordlund, 2018; Ünal et al., 2019; van der Werff & Steg, 2016; van Riper & Kyle, 2014a). In VBN’s sequential chain of factors predicting environmental behaviors, awareness of consequences is positively associated with ascription of responsibility – which refers to one’s belief that they are responsible for negative consequences that may result from not acting pro-environmentally (Gkargkavouzi et al., 2019; Han, 2015). Others have described ascription of responsibility as the belief of the denial that one’s actions have either caused or helped alleviate negative consequences to valued objects or others (Stern et al., 1999). Van Riper & Kyle (2014) suggest that whether a person believes they can make a difference and/or whether they believe that other people are not doing the needful also describes ascription of responsibility. Still others have conceptualized it as ‘felt responsibility,’ described as beliefs about the extent to which a person feels compelled to take action to obtain a desired result (Kaiser et al., 2005; Munson et al., 2021). Finally, VBN suggests that ascription of responsibility is positively associated with personal norms – often interchangeably referred to in the literature as ‘moral obligations.’ In the context of environmental behaviors, personal norms are conceptually defined as a sense of moral obligation to act pro-environmentally (Han, 2015). Influenced by an individual’s value orientations and belief structures, personal norms determine whether or not an individual feels they ought to or are obligated to engage in an environmental behavior (Gkargkavouzi et al., 2019; van Riper & Kyle, 2014a). Unlike subjective norms in TPB which refer to what an individual thinks other people would expect them to do, personal norms within VBN reflect self-expectations (Kaiser et al., 2005; Schwartz, 1977; Stern et al., 1999). 20 Like the TPB model, the VBN has also received substantial empirical support for its ability to explain environmental behaviors. However, with its emphasis on personal norms, values and beliefs, the VBN leaves out crucial other determinants of environmental behaviors such as perceived behavioral control, social norms, and attitudes (Klöckner, 2013). Relationships Between TPB and VBN Constructs In addition to comparing the planned behavior (TPB) and value-belief-norm (VBN) models in their ability to explain intentions to donate money to reduce wildlife crime, this dissertation also considers the relationships between constructs of the two theories that prior research has found to determine the scope for proposing a more comprehensive, if integrated, approach to explaining conservation behaviors aimed at reducing wildlife crime. Previous research has found that TPB’s key constructs – attitudes, subjective norms, and perceived behavioral control – are strongly related to VBN’s constructs – awareness of consequences and personal norms (Han, 2015). It has also been suggested that individuals who are aware of the consequences or seriousness of environmental problems may be more likely to have positive environmental attitudes; awareness of consequences may also have a direct positive impact on subjective norms and perceived behavioral control; and subjective norms may be positively associated with inducing personal norms (Han, 2015). However, research comparing the theories has shown mixed results. While several studies have found that TPB more strongly predicted environmental behaviors (López- Mosquera & Sánchez, 2012), others have underscored the importance of the values and personal norms-based VBN model as well (Kaiser et al., 2005). Even in the context of 21 conservation behaviors specifically, scholars have highlighted the role of values, beliefs, ecological worldviews, and personal norms in predicting environmental behaviors (Jansson et al., 2010). Both the TPB and VBN models are largely missing from the communications-based scholarship on conservation behaviors to reduce wildlife crime. This dissertation is, therefore, grounded in the literature reviewed above, presenting and discussing findings from a survey questionnaires designed based on the constructs in both the planned behavior (Ajzen, 1991) and value-belief-norm models (Stern et al., 1999). In addition to providing new empirical evidence of salient social-psychological factors likely to explain intentions to donate to conservation organizations for reducing wildlife crime, the findings from this dissertation will aid in the future development of an integrated model that can be tested in subsequent experimental research. Cross-national Wildlife Crime As mentioned in the previous chapter, wildlife crime is a kind of wicked problem whose far-reaching effects go beyond just environmental impacts and influence various aspects of our lives such as public health, economic and political stability, governance, development, and peacekeeping efforts (Giannetta, 2017; Kurland et al., 2017; McNamara et al., 2016; Veríssimo & Wan, 2019). Geopolitically, both India and the United States are significant players in the fight against wildlife crime. For instance, the U.S. is a major consumer market for illegal wildlife, parts, and products, including those such as tiger skins, tiger bones etc. that may originate in India when tigers are poached (illegally hunted) in their native wild habitats. India, therefore, is a major source country where wildlife is not only poached and smuggled, but 22 where multiple species including tortoises, marine animals, pangolins, leopards, rhinos, etc. are vulnerable to the illegal wildlife trade, often being trafficked across borders to destination countries (Khanwilkar et al., 2022). In India and the United States, multiple agencies such as the Wildlife Crime Control Bureau and the US Fish and Wildlife Service, respectively, are involved in multiple national and international operations to gather intel, enforce laws, and collaborate globally on efforts to combat illegal wildlife trade and mitigate its effects (UNODC, 2020). Given the presence and the importance of the issue in both these countries, multiple conservation organizations operating within India and the U.S. also run conservation campaigns to educate the public about the issue of wildlife crime and to persuade audiences to engage in the conservation behaviors called for in these campaigns or in other communication. Therefore, I contend that it is crucial to examine how people from both these countries think and feel about the issue of wildlife crime and behaviors linked to reducing it. Given the lack of previous research examining social-psychological factors linked to wildlife crime prevention, this dissertation makes no assumptions about levels of awareness among U.S. and Indian participants regarding wildlife crime itself, nor regarding participants’ prior knowledge of contemporary conservation issues, endangered species, etc. when assessing their responses to the survey. Both theoretically, as it relates to understanding the structure of conservation behaviors that reduce wildlife crime, and empirically, as it relates to discovering drivers of donation intentions to reduce wildlife crime, the cross-national comparison of both empirical data and theoretical models in this dissertation make valuable contributions 23 towards advancing international environmental communication scholarship (Takahashi, Metag, et al., 2021). As some scholars have found, country-level differences such as affluence, income inequality, education levels, and cultural values can influence environmental behaviors either directly, or through other social-psychological drivers such as a perceived behavioral control (Aral & López-Sintas, 2021). Others suggest that cross-national differences in family values, social structure and social norms can influence the extent to which people feel connected to nature – which itself, has been associated with ecological worldviews, environmental concern, and environmental attitudes (Clayton et al., 2021; Oh et al., 2021). In their cross-cultural analysis of the links between modernization (wealth, education, and urbanization), anthropomorphism, and attitudes towards wildlife, Gómez-Melara et al. (2021) found that the association between anthropomorphism, positive attitudes and intentions towards wildlife was universal across the five countries in which they surveyed participants – Brazil, Indonesia, Malaysia, Mexico, and Spain – but there were culturally mediated relationships that contribute to them echoing the underlying argument I make in this dissertation – that generalizing research findings from Western industrialized countries to inform conservation communication, campaigns or policies globally can misfire and should be done with extreme caution. Gómez-Melara et al. (2021) based their work on pioneering previous research establishing the relevance of studying wildlife value orientations – which are associated with conservation attitudes and behaviors – and are reviewed next. 24 Wildlife Value Orientations Teel et al. (2007) define wildlife value orientations as “basic beliefs that give personal meaning of right and wrong and an ideal life to one’s more basic values in relation to wildlife,” (p.300). Broadly, this construct is described as “enduring beliefs regarding wildlife”, (What Are Wildlife Value Orientations?, n.d.). Theoretically, wildlife value orientations are grounded in the value-attitude-behavior or cognitive hierarchy model, and as such, the construct of wildlife value orientations suggests that “individual behavior toward wildlife is driven by specific attitudes, and these attitudes are directed by wildlife value orientations,” (Teel et al., 2007, p.300). It is important to point out here that the characterization of wildlife value orientations as influencing ‘specific attitudes’ which are then responsible for influencing wildlife-related behaviors, aligns with how the theory of planned behavior framework (Gkargkavouzi et al., 2019; Kaiser et al., 2005; López-Mosquera et al., 2014) conceptualizes the relationship between attitudes and behavior – that attitudes must be as specific as possible and must correspond as closely as possible to the behavior being assessed for them to be an effective and reliable predictor of that behavior. Given that one of the objectives of this dissertation are to identify salient social-psychological drivers of wildlife-related behaviors – like donating money to reduce wildlife crime – I contend that wildlife value orientations are likely to play a significant role as drivers of attitudes towards wildlife crime-reducing behaviors, and as a more accurate assessment of how participants from India and the U.S. see their relationship with wildlife compared to the general nature of beliefs captured by the New Ecological Paradigm scale. 25 Broadly, there are two dimensions of wildlife value orientations – a domination value orientation, and a mutualism value orientation – each of which have two facets within them that reflect these constructs theoretically. Domination value orientations emphasize a hierarchical division between humans and non-humans and are suggested to lead to a utilitarian or anthropocentric attitude to wildlife which prioritizes human welfare. Mutualism orientations emphasize that humans and non-human animals are part of the same group, that wildlife deserves rights, and that humans and wildlife should co-exist in harmony. Mutualistic beliefs generally lead to more positive attitudes that prioritize co- existing with wildlife (Gómez-Melara et al., 2021; Teel et al., 2007b). Research has shown that males and older people are more likely to have domination value orientations whereas females and urban residents are more likely to have mutualistic value orientations than rural dwellers and males (Gómez-Melara et al., 2021). In their cross-cultural assessment of wildlife value orientations, Teel et al. (2007) found that mutualism value orientations were more frequently expressed in the Netherlands, Thailand, and Estonia than domination orientations; a mixture of mutualism and domination was found in Mongolia; whereas domination value orientations were most common in China. Prior work suggests that, as part of a larger shift in values, wildlife value orientations in the United States have shifted from a domination-oriented focus to a being more mutualistic and rooted in various cultural influences which, in turn, influences attitudes towards wildlife (Jacobs et al., 2018a; Manfredo et al., 2016; Teel et al., 2007b). Research also indicates that wildlife value orientations may mediate the relationship between general values and wildlife-related attitudes and norms (Jacobs et al., 2018a; Manfredo, Teel, Don Carlos, et al., 2020; Manfredo, Urquiza-Haas, Don Carlos, et al., 2020). 26 However, as Teel et al. (2007) point out, there is very little scholarship on a global, cross- national level on the cognitive determinants of human-wildlife relationships. Moreover, although research on cross-national wildlife value orientations is limited, it has not yet looked at how these value orientations manifest in India – a country in which wildlife- related conservation programs, issues, policies, crimes, and enforcement are by no means trivial in severity and scale, and one in which theories of environmental communication and environmental psychology are not commonly applied or tested in. Manfredo et al. (2009) underscore the importance of understanding global wildlife value orientations and their relationship with culture: “Every culture’s relationship with wildlife is a response to universal human needs (e.g. food, protection, reproduction), and across cultures, both differences and similarities exist in how these needs have been met. Cross-cultural study of human-wildlife relationships reveals these differences and similarities, providing important information for wildlife managers who increasingly operate in a global context,” (p. 31). In this dissertation, I therefore contend that assessing wildlife value orientations is important because without understanding the social context in which conservation programs and communication or campaigns to combat wildlife crime are run, it would be difficult to accurately evaluate the success and impacts of those campaign communications (Teel et al., 2007b). Wildlife Value Orientations as a Measure of Ecological Worldviews Due to the close correspondence between wildlife value orientations and wildlife- related attitudes and behaviors, and given how they are conceptualized and measured, I argue that wildlife value orientations may be a more accurate construct for measuring 27 ecological worldviews and, thereby, may perform better than the New Ecological Paradigm scale (Dunlap et al., 2000) at explaining or predicting those environmental behaviors that are related to wildlife and its conservation. Within the value-belief-norm framework (Stern et al., 1999; Stern & Dietz, 1994), ecological worldviews are conceptualized as general beliefs that people have about their relationships with the environment (van Riper & Kyle, 2014a), or in other words, as our beliefs about human-nature interactions (Gkargkavouzi et al., 2019). As van Riper & Kyle (2014) point out, the New Ecological Paradigm scale (NEP) that is frequently used to measure ecological worldviews within the VBN model, is “theoretically related to principles about living in harmony with or having mastery over natural and social worlds,” (p.289). The worldviews represented by the NEP scale are also described as being “situated along a continuum anchored by biocentric beliefs oriented toward environmental protection and anthropocentric beliefs geared toward people taking precedent over nature,” (van Riper & Kyle, 2014, p.289). The operationalization of the New Ecological Paradigm – a 15-item scale – has been demonstrated to tap into five underlying latent dimensions – balance of nature; limits to growth; anti-exemptionalism; eco-crisis; and human domination – each of which is measured using three items (Amburgey & Thoman, 2012). While a shorter 6-item version of the scale has been used in prior research which taps into three of those five underlying dimensions of ecological worldviews (van Riper & Kyle, 2014a), the key takeaway here is that the New Ecological Paradigm scale does not measure a unidimensional construct although a substantial number of studies have applied it as one. When a confirmatory factor analysis was used to assess the theoretical structure of the NEP scale, not only did 28 the five distinct dimensions mentioned above emerge, but researchers found that the NEP was best represented as a second-order factor structure construct in which the five underlying latent dimensions are correlated scales measured by three indicators each (Amburgey & Thoman, 2012). Wildlife value orientations also have two underlying dimensions – domination and mutualism – that are conceptually and operationally similar to the way in which the NEP scale is organized (Vaske et al., 2011). Respondents scoring higher on the eco-centric belief items on the NEP scale will be identified as having stronger ecological worldviews – which is similar to the mutualism wildlife value orientations, and those with lower NEP scores will be identified as having utilitarian or human-centered beliefs – which is similar to having domination wildlife value orientations (Amburgey & Thoman, 2012; Dayer et al., 2007; Dunlap, 2008; Dunlap et al., 2000; Fulton et al., 1996; Gómez-Melara et al., 2021; Liordos et al., 2021; Ntanos et al., 2019). Essentially, I argue, both wildlife value orientations and the NEP scale represent a general set of beliefs – one about how people perceive their relationship with wildlife (Teel et al., 2007b), and the other about how people perceive their relationship with the environment more broadly (van Riper & Kyle, 2014a). Further, as Vaske et al. (2011) point out, wildlife value orientations that stem from the cognitive hierarchy framework, reflect ideologies, and therefore, I contend, that they are conceptually and theoretically an adequate and appropriate replacement of the NEP scale – itself a set of ecological ideologies or beliefs. Finally, I posit in this dissertation that as wildlife value orientations assess beliefs about wildlife specifically, they will be more closely related to and a better measure of 29 wildlife-related behaviors, and may enhance the explanatory power of the value-belief- norm model (VBN) as a replacement of the NEP’s general environmental beliefs (X. Liu et al., 2018; Stern et al., 1999; Stern & Dietz, 1994). This dissertation examines the scope for wildlife value orientations to explain wildlife crime-reducing conservation behaviors as part of the VBN framework, and further adds new knowledge by providing empirical data on wildlife value orientations cross-culturally which itself is a crucial merit of this dissertation because no information currently exists about wildlife value orientations in India whereas there have been several studies assessing these beliefs in the United States. Research Questions Aligned with the broad objectives outlined in Chapter One, and guided by the relevant literature reviewed in this chapter and by preliminary work, this dissertation seeks to answer the following research questions: • RQ 1: Which social-psychological factors best explain donation intentions to help reduce wildlife crime? • RQ 2: How do the responses of participants from India compare to those of participants from the U.S.A.? 30 CHAPTER THREE: METHOD, MEASURES, AND ANALYSIS Survey Design, Sampling, and Procedure To answer the two research questions aimed at understanding existing values, attitudes, beliefs, and norms related to conservation behaviors for wildlife crime reduction, I conducted an online survey hosted on Qualtrics, recruiting a total of 2,000 participants of which 1,000 participants were recruited from India and 1,000 from the United States. The samples were recruited such that they were representative of latest available census data from both countries; the criteria for quota-matching were age, gender, and income. This dissertation research has been approved by the IRB at Michigan State University. The survey included questions based on both, the theory of planned behavior (Ajzen, 1991), and the value-belief-norm model (Stern et al., 1999; Stern & Dietz, 1994). Specific survey items are discussed in the following section. Participants were first asked to indicate whether they were above the age of 18. Participants were then introduced to the study and its objectives, as well as informed of their data remaining anonymous and their right to withdraw from participating without penalty at any time. Participants continued with the survey only if they granted informed consent. Three other enforced checks were used throughout the survey to ensure data quality – two attention checks and one speeding check. Participants who failed either one of the attention checks and those who completed the survey in under 10 minutes were removed from the final sample retained for analysis. For the U.S. survey, N = 900 participants were retained in the final sample after 615 participants failed the first attention check, 287 failed the second attention check, 583 were removed for speeding, 286 were underage, and 134 did not grant informed consent. For the India survey, N = 920 participants were retained for the final analysis after 770 failed the 31 first attention check, 330 failed the second attention check, 504 were removed for speeding, 156 were underage, and 147 did not consent to participate. For both the India and U.S. surveys, 10% of the study sample was designated for the pilot test. Therefore, of the total N = 1,000 participants recruited from each country by Qualtrics, N = 900 were part of the final study and n = 100 were part of the pilot study for the U.S. survey, whereas N = 920 were part of the final study and n = 98 were part of the pilot study for the India survey. Given that both the theory of planned behavior and value-belief-norm theory have not previously been applied to a study of conservation behaviors specific to reducing wildlife crime, there are no standardized scales, validated measures, or adaptable surveys that could have been used in this dissertation. This necessitated both a pre-test in which the survey instrument was tested among a group of peers, colleagues, and social scientists, and some follow-up interviews with them to gain a deeper understanding of respondents’ perceptions of the survey questions, their ability to understand them in the same way, and to identify potential sources of response error or confusion among participants (Collins, 2003). I began by conducting interviews with peers to pre-test the questionnaire and identify potential issues with order effects, understandability, etc. (Collins, 2003; Haeger et al., 2012). Using feedback from these interviews, I made requisite changes to question and response option wording and phrasing before conducting the pilot test, which I elaborate on later in this chapter. Sample Overall, for the final surveys, the sample consisted of N = 1,820 respondents, all of whom were above the age of 18 and were residents of either India or the United States. An 32 a priori power analysis showed that I would need at least 450 participants from each country. However, this study was conducted using a larger sample of respondents given an existing contract with Qualtrics. Table 1 shows the age and gender distributions of participants from both countries. Table 1 - Age and gender frequency distributions of respondents from India and USA Age India United States of America Total Female Male Female Male 18 – 24 291 136 103 17 547 25 – 34 92 180 106 50 428 35 – 44 26 80 70 68 244 45 – 54 31 27 116 39 213 55 – 64 22 15 75 81 193 More than 65 12 4 8 167 191 Total 474 442 478 422 1,816 Most participants across both countries were between 18 and 24 years old. Cross- nationally, there were slightly more female respondents than male respondents. Most (76%) of the respondents from India – 80.80% of female participants, and 71.49% of male participants were under the age of 35, whereas 43.72% of female respondents in the USA and 15.87% of men were under the age of 35. Among all the cross-national respondents over the age of 65 (n = 191), 87.43% were males from the USA (n = 167). Two respondents from India identified as non-binary whereas two others preferred not to indicate their gender identification. Tables 2 and 3 indicate the income distributions of participants from both India and the United States. 33 Table 2 - Income distribution of Indian participants Income India Total (INR) Female Male Less than ₹2,50,000 36 92 128 ₹2,50,001 to ₹5,00,000 317 124 441 ₹5,00,001 to ₹7,50,000 102 81 183 ₹7,50,001 to ₹10,00,000 8 50 58 ₹10,00,001 to ₹12,50,000 7 47 54 ₹12,50,001 or more 4 48 52 Total 473 442 916 Income categories for India reflect that country’s national tax bracket categories. Each of the two Indian respondents who identified as non-binary and each of two respondents who did not identify their gender earned less than INR 2,50,000, and between INR 2,50,001 and 5,00,000, respectively. Most Indian women earned less than INR 7,50,000 per year, whereas men had higher incomes and predominantly among those who earned more than INR 7,50,000 per year. For context, based on the exchange rate in mid-August 2022, one U.S. dollar is worth 79.75 Indian National Rupees (INR). Contextualized in terms of U.S. Dollars, most Indian female respondents earned less than $9,500 per year. Table 3 - Income distributions of US participants Income USA Total (USD) Female Male Less than $20,000 253 52 305 $20,001 to $40,000 125 114 239 $40,001 to $60,000 60 85 145 $60,001 to $80,000 30 53 83 $80,001 to $100,000 2 39 41 $100,001 or more 8 79 87 Total 478 422 900 Similar to the gender and income breakdown of the sample of Indian respondents, most U.S. participants who earned less than $40,000 per year were women, and most of those who earned higher incomes were men. Most U.S. participants (76.55%) earned up to $60,000 per year. 34 Other demographic characteristics that reflect cross-national differences in the study sample include education, employment, and political ideology. Table 4 shows the highest level of education that participants from both countries had completed. Table 4 - Education levels of participants from India and USA Education Level India USA Total No formal education 1 2 3 Primary education (US) 5 22 27 SSC/ICSE/CBSE/Grade 10 (IND) High school diploma (US) 117 231 348 Junior college/HSC/Grade 12 (IND) Vocational, technical, trade school (US) 37 100 137 Professional diploma (IND) Bachelor’s degree 439 71 510 Master’s degree 292 15 307 Ph.D. or higher 18 3 21 Other 7 27 34 Prefer not to say 4 6 10 Most respondents from India (n = 567) and the United States (n = 273) were employed either full-time or were self-employed. Table 5 shows the distribution of participants’ employment status. Table 5 - Employment distributions of Indian and U.S. participants Employment Status USA India Total Employed full-time/Self-employed 273 567 840 Employed part-time 101 119 220 Intern 2 46 48 Unemployed – looking for work 112 68 180 Unemployed – not looking for work 108 30 138 Retired 210 17 227 Other 86 61 147 Prefer not to say 8 11 19 Most respondents in both India (n = 352) and the United States (n = 359) identified their political ideologies as neither conservative nor liberal. Among those who identified as very conservative, n = 156 were from the United States whereas n = 96 were from India. n = 173 participants from the U.S. and n = 144 participants from India identified as slightly 35 conservative. Among those who identified as slightly liberal, on the other hand, n = 193 were from India and n = 125 were from the U.S. Finally, n = 135 participants from India and n = 87 participants from the U.S. identified as very liberal. Measures In this section, I outline each item used to measure the theory of planned behavior and value-belief-norm theory constructs, respectively. However, it is worth noting here that no prior validated scales and measures of these constructs exist specifically in the context of wildlife crime reduction. Therefore, I rely on previous research in environmental behavior – even if outside the context of wildlife crime-reducing conservation behaviors – to operationalize the constructs in both theoretical models. Further, given this lack of existing previously validated scales to measure wildlife crime-reducing behaviors, I expect that this dissertation will aid in the development and validation of scales that more closely measure attitudes, norms, beliefs, perceived behavioral control, and intentions related to reducing wildlife crime. Dependent Variable: Behavioral Intention Items modified from Carfora et al. (2021b) and Han (2015) were used to measure participants’ intentions to donate money to conservation organizations to help reduce wildlife crime. These items were: “I am willing to donate money to conservation organizations over the next month to help reduce wildlife crime”; “I plan to donate money to conservation organizations over the next month to help reduce wildlife crime”; and “I will definitely donate money to conservation organizations over the next month to help reduce wildlife crime.” Responses to each of these items were assessed on a scale from 1 = Strongly disagree to 7 = Strongly agree (see Table 6). 36 Theory of Planned Behavior While the conceptual definitions of the constructs measured by the items listed below are outlined in Chapter 2, Table 6 presents the operationalizations and scales/items. The table also indicates which previous literature the measures were derived from and how they were modified to apply to this novel context of conservation behaviors aimed at reducing wildlife crime – a category of environmental behaviors that this theory has not previously been used to examine, creating the absence of valid, reliable, and closely corresponding measures. Table 6 - Measures of theory of planned behavior constructs Construct Measures References Attitude towards the “I think the idea of donating money to Modified to include the behavior conservation organizations to help reduce outcome behavior of wildlife crime is” --- donating money to • Very positive reduce wildlife crime • Very responsible López-Mosquera et al., • Very intelligent 2014 • Very useful • Very ecologically helpful (1 = Strongly disagree, 7 = Strongly agree) Subjective norms “Most people who are important to me think Modified to include the that one should donate money to outcome behavior of conservation organizations to help reduce donating money to wildlife crime.” reduce wildlife crime “Most people who are important to me López-Mosquera et al., expect that I will donate money to 2014, and Han, 2015. conservation organizations to help reduce wildlife crime.” Both cited references use these items. “Those people whose opinions I value would donate money to conservation organizations to help reduce wildlife crime.” (1 = Strongly disagree, 7 = Strongly agree) 37 Table 6 (cont’d) Perceived behavioral “I am confident that if I want, I can donate Modified to include the control money to conservation organizations to help outcome behavior of reduce wildlife crime.” (Han, 2015) donating money to reduce wildlife crime; “I have sufficient resources to be able to and modified one item donate money to conservation organizations into three individual to help reduce wildlife crime.” (Han, 2015) items on resources, time and opportunity. “I have enough time to donate money to conservation organizations to help reduce Ajzen, 2006; Han 2015; wildlife crime.” (Han, 2015) and López-Mosquera et al., 2014. “I have enough opportunities to donate money to conservation organizations to help reduce wildlife crime.” (Han, 2015) “Donating money to conservation organizations to help reduce wildlife crime is completely up to me.” (Han, 2015; Ajzen, 2006) “If I donated money to conservation organizations, it would help them to reduce wildlife crime.” (López-Mosquera et al., 2014) (1 = Strongly disagree, 7 = Strongly agree) Intention “I am willing to donate money to conservation Modified to include the organizations within the next month to help outcome behavior of reduce wildlife crime.” (Han 2015; Carfora et donating money to al., 2021) reduce wildlife crime and changed “over the “I plan to donate money to conservation next month” to “within organizations within the next month to help the next month”. reduce wildlife crime.” (Han 2015; Carfora et al., 2021) “I will definitely donate money to conservation Han, 2015 and Carfora organizations within the next month to help et al., 2021 reduce wildlife crime.” (Carfora et al., 2021) (1 = Strongly disagree, 7 = Strongly agree) 38 Value-Belief-Norm Measures Table 7 indicates the items/scales used to measure the constructs of the value-belief- norm theory while also outlining the previous literature from which these measures were adapted and how they were modified to fit this dissertation given that this theory, too, has not previously been applied to examinations of this category of environmental behaviors. Table 7 - Measures of value-belief-norm theory constructs Construct Measures References How important are each of the following to you as guiding principles in your life? (1 = Not at all important, 7 = Very important) Biospheric “How important is it to you ---” All value items values • To prevent environmental pollution. modified to be framed • To protect the environment. as a question instead of • To respect nature. a statement like “It is • To be in unity with nature. important for him/her to…” Bouman et al., 2018 Altruistic “How important is it to you ---” values • That every person has equal opportunities. • To take care of those who are worse off. • That every person is treated justly. • That there is no war or conflict. • To be helpful to others. Egoistic “How important is it to you ---” values • To have control over others’ actions. • To have authority over others. • To be influential. • To have money and possessions. • To work hard and be ambitious. 39 Table 7 (cont’d) Ecological • “We are approaching the limit of the number of NEP scale; Dunlap, worldviews people the earth can support.” 2008 • “Humans have the right to modify the natural environment to suit their needs.” (1 = Strongly disagree, • “When humans interfere with nature, it often 7 = Strongly agree) produces disastrous consequences.” • “Human ingenuity will ensure that we do not make the earth unlivable.” • “Humans are severely abusing the environment.” • “The earth has plenty of natural resources if we just learn how to develop them.” • “Plants and animals have as much right as humans to exist.” • “The balance of nature is strong enough to cope with the impacts of modern industrial nations.” • “Despite our special abilities, humans are still subject to the laws of nature.” • “The so-called ‘ecological crisis’ facing humankind has been greatly exaggerated.” • “The earth is like a spaceship with very limited room and resources.” • “Humans were meant to rule over the rest of nature.” • “The balance of nature is very delicate and easily upset.” • “Humans will eventually learn enough about how nature works to be able to control it.” • “If things continue on their present course, we will soon experience a major ecological catastrophe.” Awareness of “Wildlife crimes cause biodiversity loss, exhaustion of consequences natural resources, and species extinction.” Modified to attribute the problem statement “Wildlife crimes generate environmental impacts on to wildlife crime and to the neighboring areas and wider environment.” mention relevant negative consequences “Wildlife crimes such as illegal international trade in of wildlife crime. exotic species can cause the spread of infectious diseases from animals to humans.” Structure is from Han, 2015 “Conservation organizations help to curb wildlife crime and mitigate its impacts.” (Reverse coded) (1 = Strongly disagree, 7 = Strongly agree) 40 Table 7 (cont’d) Ascription of “Every citizen must take responsibility for mitigating Modified to mention responsibility the environmental, economic, and health issues linked specific impact types of to wildlife crimes by donating money to conservation wildlife crime. organizations.” “The authorities are responsible for mitigating the Structure is from environmental, economic, and health issues linked to Carfora et al., 2021; and wildlife crimes by financially supporting conservation Gkargkavouzi et al., organizations.” 2019. “I am responsible for mitigating the environmental, Both cited references economic, and health issues linked to wildlife crimes use these items by donating money to conservation organizations.” (1 = Strongly disagree, 7 = Strongly agree) Personal “I feel I ought to donate money to conservation Modified the action and norms organizations to help reduce wildlife crime.” outcome from López- Mosquera & Sánchez, “I would feel guilty if I did not donate money to 2012; conservation organizations to help reduce wildlife crime.” “I feel morally obligated to donate money to conservation organizations to help reduce wildlife crime regardless of what others are doing.” Modified the action (donating money) and “I feel that donating money to conservation outcome (reducing organizations to help reduce wildlife crime is the right wildlife crime) from thing to do.” Gkargkavouzi et al., 2019 “I would feel good about myself if I donated money to conservation organizations to help reduce wildlife crime.” (1 = Strongly disagree, 7 = Strongly agree) Other Variables This section outlines the scales and items used to measure wildlife value orientations (WVOs) (Dayer et al., 2007; Fulton et al., 1996; Vaske et al., 2022); as well as items measuring the dependent variable, and demographic variables. 41 Wildlife Value Orientations As mentioned in Chapter 2, this dissertation is also an assessment of the suitability of wildlife value orientations as a valid and reliable measure of environmental behaviors specifically associated with wildlife conservation – such as donating money to conservation organizations to reduce wildlife crime. Given the criticism that the current measure of the ecological worldviews construct in the value-belief-norm model – the New Ecological Paradigm (NEP) – has received (Amburgey & Thoman, 2012), I used the measures outlined in Table 8, that have been developed and used widely for over a decade by scholars studying human dimensions of wildlife conservation. A typology of beliefs about human-wildlife relationships such as domination (also known as utilitarianism) and mutualism is based on value dimensions. People who have a domination wildlife value orientation are more likely to believe that wildlife should be used and managed for human benefit, whereas those who have a mutualistic wildlife value orientation are likely to see wildlife as part of their extended family and as deserving of care and rights equal to what humans receive (Jacobs et al., 2018b; Manfredo, Teel, Carlos, et al., 2020; Teel et al., 2007b). 42 Table 8 - Measures of wildlife value orientations Domination Value Orientations Appropriate use “Humans should manage fish and wildlife populations so that humans belief dimension benefit.” “The needs of humans should take priority over fish and wildlife (1 = Strongly protection.” disagree to 7 = “It is acceptable for people to kill wildlife if they think it poses a threat to Strongly agree) their life.” “It is acceptable for people to kill wildlife if they think it poses a threat to their property.” “Fish and wildlife are on earth primarily for people to use.” Hunting belief “We should strive for a world where there’s an abundance of fish and dimension wildlife for hunting and fishing.” (1 = Strongly “Hunting is cruel and inhumane to the animals.” (R) disagree to 7 = “Hunting does not respect the lives of animals.” (R) Strongly agree) “People who want to hunt should be provided the opportunity to do so.” Mutualism Value Orientation Social affiliation “We should strive for a world where humans and fish and wildlife can belief dimension live side by side without fear.” (1 = Strongly “I view all living things as part of one big family.” disagree to 7 = “Animals should have rights similar to the rights of humans.” Strongly agree) “Wildlife are like my family and I want to protect them.” Caring belief “I care about animals as much as I do other people.” dimension “It would be more rewarding to me to help animals rather than people.” (1 = Strongly “I take great comfort in the relationships I have with animals.” disagree to 7 = “I feel a strong emotional bond with animals.” Strongly agree) “I value the sense of companionship I receive from animals.” Demographic Variables I also measured social-demographic variables such as age, gender, education, income, employment status and political ideology as these have been known to be associated with environmental behaviors (Kollmuss & Agyeman, 2002) and I expected them to relate to the dependent variable – intent to donate money. The questionnaire for Indian participants categorized income levels using the Government of India’s income tax brackets to maintain a sense of uniformity and familiarity for participants rather than simply converting the 43 corresponding U.S. Dollar amounts into Indian National Rupees. For specific demographic measures for U.S. and Indian questionnaires, see Appendix I. Expected Outcomes This section outlines the relationships I expected to see between constructs of the theory of planned behavior and the value-belief-norm theory, as well as inter-theory relationships that I expected to see based on what scholars have found in previous research. The hypotheses outlined in this section were all analyzed using correlations and regressions. However, I also used structural equation modeling (SEM) to explore and analyze multivariate relationships and find out how well the data from both India and U.S. surveys is represented in the structural models that are based on the conceptual models of both theories (see Figures 1 and 2). Value-Belief-Norm Theory Hypotheses H 1a. Biospheric values are positively associated with ecological worldviews. H 1b. Altruistic values are positively associated with ecological worldviews. H 1c. Egoistic values are negatively associated with ecological worldviews. H 2. Ecological worldviews are positively associated with awareness of consequences. H 3. Awareness of consequences is positively associated with ascription of responsibility. H 4. Ascription of responsibility is positively associated with pro-environmental personal norms. H 5. Pro-environmental personal norms are positively associated with donation intentions to reduce wildlife crime. 44 Theory of Planned Behavior Hypotheses H 6. Attitudes towards monetary donation are positively associated with donation intentions to reduce wildlife crime. H 7. Subjective norms are positively associated with donation intentions to reduce wildlife crime. H 8. Perceived behavioral control is positively associated with donation intentions to reduce wildlife crime. Other Expected Relationships Based on the literature reviewed in Chapter 2, I also proposed the following hypotheses that relate concepts from the theory of planned behavior to concepts from the value-belief-norm theory. If supported, these hypotheses can help identify aspects of future conservation communication on wildlife crime reduction, and can contribute to theory testing and building, such as through the development of an integrated model that includes concepts from both theories and may have enhanced power to predict wildlife crime- related conservation behaviors. H 9a. Awareness of consequences will be positively associated with attitudes towards donating money to reduce wildlife crime. (Han 2015; Carfora et al., 2021) H 9b. Awareness of consequences will be positively associated with subjective norms. (Han, 2015) H 9c. Awareness of consequences will be positively associated with perceived behavioral control. (Han, 2015) H 10a. Ecological worldviews are positively associated with wildlife value orientations. 45 H 10b. Ecological worldviews are positively associated with attitudes towards donating money to reduce wildlife crime. (Carfora et al., 2021) H 11a. Mutualistic wildlife value orientations will be positively associated with attitudes. (Teel et al., 2007b) H 11b. Mutualistic wildlife value orientations will be positively associated with donation intentions to reduce wildlife crime. (Manfredo, Teel, Don Carlos, et al., 2020) H 12. Biospheric values will be positively associated with mutualistic wildlife value orientations.(Oh et al., 2021) H 13. Ascription of responsibility is positively associated with perceived behavioral control (Carfora et al., 2021). Pilot Test and Rationale I conducted pre-test interviews with five peers (doctoral students and friends) to test- run the questionnaire and identify potential issues with order effects, understandability, etc. (Collins, 2003; Haeger et al., 2012). After revising minor issues related to item wording or phrasing that were revealed during interviews, I launched a pilot test of the survey questionnaire on a small subset of respondents from India (n = 98) and the United States (n = 100). The pilot test was hosted online using the Qualtrics platform, which is also how participants were recruited. The pilot test helped identify potential issues with the online survey design, the Qualtrics flow, and setup, as well as potentially problematic response behavior from participants such as straight-lining, speeding, not paying attention, etc. Additionally, I was able to use results from the pilot test to determine whether I may have systematic issues with missing data due to incomplete responses etc. 46 In both India and U.S. samples, I observed straight-lining and speeding behavior but any missing data appeared to be rare and random rather than systematic, suggesting that the survey questions and response options were clear to participants. Any responses in which participants failed one or both attention checks in the questionnaire were automatically removed from the final list of “good completes” that Qualtrics provided. This likely helped minimize issues with missing and unreliable data quite substantially. Based on the pilot test, I included a speeding check for the final study such that respondents who completed the survey in less than 10 minutes would not be included in the final data as “good completes.” I also asked Qualtrics to add straight-lining behavior to their data scrubbing and cleaning criteria – which is part of their protocol before survey responses to the researcher after data collection. Finally, in both the pilot test and the final study, the order of questions within blocks was randomized. Items serving as measures of one common construct were considered to be within the same organizational survey ‘block’ as a part of the Qualtrics design element. Analysis and Main Test Upon ensuring that there were no systematic patterns of missing data, I conducted tests of normality and multicollinearity by looking at distributions, histograms, standard deviations, skewness, and kurtosis. I also tested for multicollinearity by examining Variance Inflation Factor (VIF) values. All the values for these tests were within acceptable limits; for example, VIF values did not exceed 5 for any variables, suggesting low to moderate correlations between variables. Additionally, I also conducted the Kaiser-Meyer- Olkin (KMO) test and the Bartlett’s test of sphericity to determine the factorability of the 47 data. All KMO values were higher than 0.6 and all Bartlett’s sphericity tests were significant at p ≤ 0.05. For the statistical analysis, I used SPSS to examine descriptives such as means, SDs, frequencies etc., and ran crosstabs and Chi-square tests. I also ran bivariate correlations, linear regressions, scale reliability, exploratory factor analyses with maximum likelihood estimation as the extraction method and promax rotations (Miller - Carpenter, 2018). Finally, I conducted confirmatory factor analysis (CFA), multivariate normality tests, measurement invariance analysis, and ran structural equation models (SEM) using the lavaan, semTools, semPlot, equaltestMI, MVN, and tidyverse packages in R. Confirmatory factor analyses (CFA) are used to run measurement models which determine the relationships between indicators and the latent factors that they are said to measure, whereas the structural models are used to demonstrate relationships between different latent factors. While running the CFAs, I also examined the correlation matrices within the data, and calculated the Average Variance Extracted (AVE) to assess convergent and discriminant validity. To assess the fit of the measurement models (CFA) and structural models, I assessed different indices and measures of model fit such as Chi-square, Chi- square and degrees of freedom ratio, Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR), while also examining factor loadings and retaining only those indicators (i.e., items) that had loadings greater than 0.5 on their respective latent factors (Carfora et al., 2021; Gkargkavouzi et al., 2019; van Riper & Kyle, 2014a). 48 CHAPTER FOUR: RESULTS In conducting the confirmatory factor analyses and structural equation modelling, I found that different measurement and structural models would necessitate a parallel country-specific modelling approach. Therefore, any results presented to answer RQ1 will also help answer RQ2. To avoid redundancies, this chapter is organized by theory rather than by research question. The sections below, therefore, answer both RQ1 by identifying the social-psychological factors that mattered most in predicting donating intentions to curb wildlife crime, and RQ2 by highlighting cross-national differences in models and findings. A more detailed analysis of variable-specific cross-national differences in responses is available in Appendix J. Exploratory Factor Analysis This section outlines the protocol and rationale I followed for exploratory factor analysis (EFA) of latent factors in both theoretical models. Any theory-specific findings from the EFA are discussed in the theories’ respective sections below. First, I checked all the data for missing values. No more than five missing values were found in both countries’ data, and no patterns of missing data emerged. On examining normality, I found that the standard deviations, histograms, skewness and kurtosis values indicate univariate normal distribution as none of the standard deviation (SD) values were greater than 2, none of the skewness values were greater than 3, and none of the kurtosis values were greater than 10 (X. Liu et al., 2018). To assess multicollinearity, I examined Variance Inflation Factor (VIF) values and on finding that none of them exceeded 5, I concluded that there were weak to moderate correlations between variables in the data, suggesting that multicollinearity was not a concern. 49 I conducted the Bartlett’s test of sphericity and the Kaiser-Meyer-Olkin (KMO) test to validate whether the data from both countries were appropriate for running Exploratory Factor Analyses (EFA) on them. All variables met the criteria for suitability for proceeding with an EFA – i.e., KMO values were above 0.6 and Bartlett’s tests were statistically significant for all variables (Miller - Carpenter, 2018; Obeng & Aguilar, 2018). As I planned to conduct a Confirmatory Factor Analysis (CFA) first, and run structural equation models later on hypothesized conceptual models, I used maximum likelihood estimation as the extraction method for the EFA, and since I did expect the factors in both my theoretical models – TPB and VBN – to correlate with one another, I used the promax oblique rotation for EFAs. Items across both theories’ sets of measures were retained only if they had factor loadings greater than 0.32 in the EFA, and greater than 0.5 in the CFA (Miller - Carpenter, 2018). I followed this procedure for the EFA for exploring factor loadings across constructs in both the theories – which I discuss individually next. Value-Belief-Norm Theory In determining the factorability of the data with respect to the latent factors in the value-belief-norm model (Stern et al., 1999), I followed the process for exploratory factor analysis outlined above. Values for the Kaiser-Meyer-Olkin (KMO) test were above 0.6 for all factors, with the lowest one being .65 for the ‘ascription of responsibility’ factor. Bartlett’s tests of sphericity were also significant for all factor analyses. Items for most latent factors loaded on single-factor solutions representing those respective latent factors. However, ‘egoistic values’ and the ‘NEP’ items measuring ecological worldviews had multi- dimensional outputs. The five items for egoistic values loaded on two different factors in both samples, as demonstrated by the factor loadings, eigenvalues in the EFA output, and 50 the scree plot. Similarly, the 15 items in the NEP scale loaded on three different factors in both samples as shown by their eigenvalues and scree plots. Further, values for the Variance Inflation Factor (VIF) did not exceed 3.8 when assessing multicollinearity, and while there were only rare instances of missing data that did not follow any patterns, I used the full information maximum likelihood estimation method for the SEM analysis involving this VBN model constructs. I also assessed convergent validity by calculating the Cronbach’s alpha and Average Variance Extracted (AVE) values for all latent constructs in the VBN model, and by examining factor loadings of all included items to ensure they were above 0.3 in the EFA and above 0.5 in the CFA (Cheung & Wang, 2017; Tabachnick & Fidell, 2007). Table 9 shows the means, standard deviations, and Cronbach’s alpha for the items and latent variables in the VBN model for both Indian and US samples. The Cronbach’s alpha values indicate that these measures and scales were far more reliable and internally consistent for the U.S. sample than they were for the Indian sample. Table 9 - Means, standard deviations, and reliability for value-belief-norm theory concepts Variables/Measures INDIA USA Mean S.D. Mean S.D. New Ecological Paradigm (NEP) 4.70 .5963 4.84 .8829 (Unidimensional - USA α = .814; India α = .624)** Environmental Concern: (USA α = .795; India α = .693) NEP3: When humans interfere with nature, it often 6.04 1.202 5.54 1.311 produces disastrous consequences.^ NEP5: Humans are severely abusing the environment.^ 6.08 1.205 5.66 1.411 NEP9: Despite our special abilities, humans are still 5.81 1.234 5.92 1.152 subject to the laws of nature.^ NEP13: The balance of nature is very delicate and easily 5.41 1.456 5.38 1.438 upset.^ NEP15: If things continue on their present course, we 6.04 1.149 5.26 1.624 will soon experience a major ecological catastrophe.^ 51 Table 9 (cont’d) Anthropocentricism: (USA α = .761; India α = .701) NEP2: Humans have the right to modify the natural 4.62 2.002 4.73 1.758 environment to suit their needs. NEP8: The balance of nature is strong enough to cope 3.92 1.873 4.65 1.701 with the impacts of modern industrial nations. NEP10: The so-called 'ecological crisis' facing 3.38 1.838 4.65 1.899 humankind has been greatly exaggerated. NEP12: Humans were meant to rule over the rest of 4.87 2.053 4.58 1.974 nature. NEP14: Humans will eventually learn enough about how 2.92 1.599 4.49 1.755 nature works to be able to control it. Limits to Growth: NEP1: We are approaching the limit of the number of 5.32 1.493 4.41 1.745 people the earth can support.~ NEP11: The earth is like a spaceship with very limited 5.02 1.867 4.61 1.803 room and resources.~ Removed from analysis of both samples: NEP4: Human ingenuity will ensure that we do not make 2.93 1.454 4.01 1.695 the earth unlivable.* NEP6: The earth has plenty of natural resources if we 1.91 1.225 2.87 1.578 just learn how to develop them.* NEP7: Plants and animals have as much right as humans 6.39 1.144 5.89 1.389 to exist.* Awareness of Consequences (AC) 5.90 .8222 5.41 .9332 (USA α = .754; India α = .72) AC1: Wildlife crimes cause biodiversity loss, exhaustion 6.18 1.091 5.70 1.210 of natural resources, and species extinction. AC2: Wildlife crimes generate environmental impacts on 5.97 1.048 5.49 1.167 areas in which they occur and on surrounding communities. AC3: Wildlife crimes such as illegal international trade 5.81 1.237 5.26 1.305 in exotic species causes the spread of deadly viruses and pathogens from animals to humans. AC4: Conservation organizations help to curb wildlife 5.63 1.075 5.21 1.236 crime and mitigate its impacts. Ascription of Responsibility (AR) 5.97 .8605 4.66 1.325 (USA α = .783; India α = .685) AR1: Every citizen must take responsibility for 6.18 .962 4.51 1.657 mitigating the environmental, economic, and health issues linked to wildlife crimes by donating money to conservation organizations that work to address it. 52 Table 9 (cont’d) AR2: The authorities must take responsibility for 6.16 .973 5.18 1.442 mitigating the environmental, economic, and health issues linked to wildlife crimes by financially supporting conservation organizations that work to address it. AR3: I am responsible for mitigating the environmental, 5.57 1.322 4.29 1.653 economic, and health issues linked to wildlife crimes by donating money to conservation organizations that work to address it. Personal Norms (PN) 5.60 1.006 4.36 1.483 (USA α = .925; India α = .826) PN1: I feel I ought to donate money to conservation 5.56 1.295 4.31 1.736 organizations to help reduce wildlife crime. PN2: I would feel guilty if I did not donate money to 5.17 1.600 3.64 1.804 conservation organizations to help reduce wildlife crime. PN3: I feel morally obligated to donate money to 5.38 1.362 3.89 1.784 conservation organizations to help reduce wildlife crime regardless of what others are doing. PN4: I feel that donating money to conservation 5.88 1.136 4.99 1.507 organizations to help reduce wildlife crime is the right thing to do. PN5: I would feel good about myself if I donated money 6.01 1.099 4.95 1.607 to conservation organizations to help reduce wildlife crime. Biospheric Values (BioV) 6.55 .5948 5.54 1.130 (USA α = .869; India α = .794) BioV1: How important is it to you to prevent 6.48 .882 5.5 1.379 environmental pollution? BioV2: How important is it to you to protect the 6.58 .701 5.59 1.317 environment? BioV3: How important is it to you to respect nature? 6.66 .626 5.95 1.134 BioV4: How important is it to you to be in unity with 6.46 .792 5.12 1.483 nature? Altruistic Values (AltV) 6.11 .7045 5.76 .9269 (USA α = .803; India α = .673) AltV1: How important is it to you that every person has 6.36 .887 5.93 1.207 equal opportunities? AltV2: How important is it to you to take care of those 5.52 1.358 5.17 1.386 who are worse off? AltV3: How important is it to you that every person is 6.11 1.124 6.09 1.101 treated justly? AltV4: How important is it to you that there is no war or 6.29 1.029 5.85 1.302 conflict? AltV5: How important is it to you to be helpful to others? 6.25 .880 5.74 1.182 53 Table 9 (cont’d) Egoistic Values (EgoV) 5.16 .9489 3.80 1.138 (USA α = .758; India α = .678)** EgoV1: How important is it to you to have control over 4.34 1.822 2.62 1.727 others' actions? EgoV2: How important is it to you to have authority over 4.09 1.839 2.51 1.640 others? EgoV3: How important is it to you to be influential? 5.36 1.336 3.95 1.793 EgoV4: How important is it to you to have money and 5.70 1.221 4.53 1.602 possessions?* EgoV5: How important is it to you to work hard and be 6.32 .887 5.40 1.432 ambitious?* * Note: Items removed from reliability and SEM analysis due to poor EFA and CFA factor loadings ~ Note: NEP items excluded from analysis of reliability, CFA, SEM etc. ^ Note: NEP items measuring the environmental concern dimension. ** Note: Cronbach’s alpha indicates reliability of scale including only retained items The Average Variance Extracted (AVE) for latent factors in the value-belief-norm model was not consistently good. Table 10 shows the AVE values for VBN constructs across both samples. As several factors did not have AVE values exceeding 0.5, these factors would not be considered to have convergent validity. This means that for those constructs in Table 10 that have AVE values below 0.5, the items or measures used do not do a good enough job of closely explaining the latent variable they were meant to measure (Cheung & Wang, 2017; My Easy Statistics, 2020; Tabachnick & Fidell, 2007). Table 10 - Values of average variance extracted from VBN latent factors Latent Variables in VBN Average Variance Extracted India USA Biospheric values .512 .625 Altruistic values .331 .439 Egoistic values .597 .546 Environmental concern (NEP) .257 .701 Anthropocentricism (NEP) .326 .395 Awareness of consequences .393 .45 Ascription of responsibility .439 .488 Personal norms .493 .701 Note: AVE values that are above the 0.5 cut threshold are in bold font. 54 Dimensionality of the New Ecological Paradigm Scale Before getting to the results of the CFA and SEM, in this section I lay out the process I followed to be able to define my measurement models in the CFA accurately and in a manner that aligns closely with what previous literature on the value-belief-norm theory and the NEP scale suggests. As I discuss in Chapter 2, the New Ecological Paradigm scale developed by Dunlap et al. (2000) has been used widely, including by those who developed the value-belief-norm theory (Stern et al., 1999; Stern & Dietz, 1994) as an operationalization of the theory’s ‘ecological worldviews’ construct. A substantial number of studies in environmental psychology, environmental communication, and environmental behavior have used the NEP (Dunlap, 2008; Ntanos et al., 2019; Slimak & Dietz, 2006; Sparks et al., 2021; Tyllianakis & Ferrini, 2021) – and many have used is as a unidimensional scale that measures a single latent factor sometimes referred to as ‘environmental concern’ or just as ecological worldviews. Over the years, shorter versions of the original-revised 15-item scale have also been used (van Riper & Kyle, 2014a) but most studies using the NEP scale have only ever conducted exploratory factor analyses, if that, leaving much to be desired in terms of hard evidence of the factor structure of this larger construct. One of the few scholars to not only conduct several CFAs on this scale, but to actually compare proposed measurement models of NEP, Amburgey & Thoman (2012) compared three different but commonly used approaches to conceptualizing and operationalizing the NEP. They found that the best model fit suggests that the NEP scale was in fact, a second- order latent factor with five correlated underlying dimensions, namely: balance of nature; 55 eco-crisis; anti-exemptionalism; limits to growth; and anti-anthropocentricism or human domination (Amburgey & Thoman, 2012). In determining the factor structure for the NEP latent variable in my own data, therefore, I found through the EFA that all the items in the NEP scale loaded on three different factors and while some had poor loadings, others also cross-loaded on more than one factor. Comparing the items that loaded on my three EFA factors with the items loaded on to the five factors in the study by Amburgey & Thoman (2012), I determined that my data most closely resembled these three underlying factors of a second-order NEP: environmental concern (a combination of ‘balance of nature’ and ‘eco-crisis’ as explicated by Amburgey & Thoman (2012)), limits to growth, and anthropocentricism or human domination (a combination of the anti-exemptionalism and human domination dimensions explicated by Amburgey & Thoman (2012)). I thereby began specifying my measurement model for the NEP CFA by investigating whether these three dimensions – environmental concern, limits to growth, and anthropomorphism – are contributing to a well-fitting model with high internal consistencies. I found that one of the two items measuring ‘limits to growth’ had extremely poor factor loadings across both US and Indian groups, necessitating its removal, and by extension, the exclusion of the single remaining item assessing that dimension as a minimum of two indicators are required for a CFA (Kenny, 2016). From the other two dimensions, some items showed poor loadings in one or both groups of respondents. I omitted these from the country-specific CFAs when specifying the measurement models. Ultimately, when using the NEP latent variable in my larger value-belief-norm model assessments, I conceptualize, and therefore operationalize it as a second-order latent factor 56 with two underlying dimensions – environmental concern and anthropomorphism – which is partly consistent with what Amburgey & Thoman (2012) found. Confirmatory Factor Analysis The VBN latent variables with low internal consistency (i.e., Cronbach’s alpha below 0.7) for the Indian sample were: altruistic values; egoistic values; ascription of responsibility; and the unidimensional operationalization of the NEP scale measures (see Table 9). All latent factors had high reliability for the U.S. sample (i.e., greater than 0.7). Given these inconsistencies in reliability across groups and latent factors, I aimed to answer some guiding questions with this CFA, namely: 1) Does the same measurement model fit data from both groups of respondents?; 2) Can relationships between latent factors in both groups be explained by the same structural model?; 3) How do measurement and structural model fits compare when the NEP scale latent factor is replaced by the mutualistic wildlife value orientations (WVOs) latent factor? Does the model with mutualistic wildlife value orientations have better predictive power than the original VBN model that includes the NEP scale as a measure of ecological worldviews? To answer these guiding questions, I conducted a series of multi-group confirmatory factor analyses (MGCFA) to identify models that fit the data well, to identify cross-national differences between groups, and to test for measurement invariance or equivalence. All CFAs and MGCFAs included models that were specified based on an a priori theoretical model (see Figure 2). I used Rstudio (version 2022.02.3+492) to run all CFA, MGCFA, and SEM analyses, and to plot path diagrams. Specifically, I used the lavaan, haven, semTools, semPlot, equaltestMI, MVN, tidyverse, and magrittr packages in R. As established before, although I could establish univariate normality in the US sample for VBN variables and 57 measures, some measures were not normally distributed within the India sample as evidenced by the higher standard deviations seen in Table 9. I assessed multivariate normality using the MVN package. When looking at multivariate relationships simultaneously, both sets of data – the India and the U.S. data – did not exhibit normality as demonstrated by the Mardia skewness and kurtosis values (see Appendix A). However, the assumption of normality is a requirement of the method of estimation, namely, maximum likelihood, and not a requirement of the SEM analysis itself, and it is also worth keeping in mind that normality assumptions apply not to the exogenous independent variables, but the residuals, and therefore, only apply to endogenous variables. This is why I use the robust maximum likelihood estimation, and report the robust fit statistics and Satorra-Bentler Chi-square and standard error values for CFA and SEM models (CenterStat, 2019; Finney & DiStefano, 2006; Hoyle & Isherwood, 2013; Satorra & Bentler, 1994). Testing the models for equivalence, my steps would involve running CFAs to examine one unrestricted measurement model in which the data were stacked overall without grouping by country; a second unrestricted model in which the stacked data were grouped by country; a third model in which factor loadings were made equal; a fourth model in which regression paths were constrained to be equal; and a fifth model in which both intercepts and factor loadings were constrained to be equal (Boomsma et al., 2012; Kenny, 2016, 2020; Pirralha, 2020b, 2020c, 2020a; Thakkar, 2020). In following these steps and running the CFAs, I found that for the value-belief-norm model, as with the planned behavior model, I did not have even weak invariance – meaning that measurement models were different for both the Indian and U.S. groups. There were 58 several problematic indicator items in the VBN data as evidenced both by the low scale reliability of many of the VBN factors in the Indian sample, and the lack of convergent validity for several VBN latent factors across both samples. Primarily, the items with poor factor loadings in both EFAs and CFAs that were common to both groups were a handful of the NEP items (see Table 9), and two of the items measuring egoistic values. Specifically in the Indian sample, one of the measures for altruistic values, and additional measure of egoistic values, and a couple of the NEP items were excluded from the CFA and SEM analysis due to their poor factor loadings on all models. In the US sample, apart from one or two different NEP items that loaded poorly and were excluded from the analysis, I was able to retain most of the measures for other latent factors. This was not surprising as I could see that the data fit the US sample and distributions more normally than it did the Indian sample. Figure 3 - Measurement model for original value-belief-norm model - USA 59 Figure 4 - Measurement model for original value-belief-norm model – India Structural Equation Modeling Given that the measurement models for both country groups appeared to be different (i.e., not ‘invariant’), I followed a similar process to test the equivalence of structural models, again finding that even weak invariance could not be established. Here, I present the results of the SEM. The fit statistics for the structural equation model that that best describes the ability of item measures to explain latent variables, and the relationships between these latent variables in the Indian sample data are: χ2 = 1316.455, df = 561; p < .001; CFI = .925; TLI = .918; RMSEA = .042; SRMR = .077. Figure 5 shows this model. 60 Figure 5 - Complete SEM and measurement model for value-belief-norm theory - India Figure 6 - Complete SEM and measurement model for value-belief-norm theory – USA 61 Figure 6 visually represents the structural equation and measurement models that fit the U.S. sample: χ2 = 2214.262, df = 576; p < .001; CFI = .912; TLI = .903; RMSEA = .056; SRMR = .109. Hypothesis Testing Based on the SEM analysis, I present results of the proposed relationships between VBN constructs here (Stern et al., 1999; Stern & Dietz, 1994). I also include standardized regression coefficients for linear regressions to demonstrate how these coefficients change from a bivariate linear analysis to a multivariate structural equation modeling analysis. My assessment of whether these hypotheses are supported is based on the SEM results for both countries. H1a. This hypothesis posits that biospheric values are positively associated with ecological worldviews. First, I ran correlations between biospheric values and the two underlying dimensions of the higher-order NEP latent variable – environmental concern and anthropocentricism. For US. respondents, biospheric values are positively and significantly associated with environmental concern (r = .551, p < .001) and with anthropocentricism (r = .259, p < .001). Among Indian respondents, while biospheric values were positively and significantly associated with environmental concern (r = .306, p < .001), their association with anthropocentricism was negative but non-significant. Linear regressions show that biospheric values predicted only 9.4% of the variance in environmental concern in Indian respondents [F (1, 918) = 94.807, p < .001; std. β = .306], but they predicted 30.4% of the variance in environmental concern among U.S. respondents [F (1, 898) = 392.251, p < .001; std. β = .551]. Multivariate regressions from the SEM show that biospheric values are negatively but non-significantly predictive of 62 environmental concern among Indian respondents, however they significantly predicted environmental concern among U.S. respondents (std. β = .989, p < .001). Linear regressions examining the relationship between biospheric values and anthropocentricism show a small negative but non-significant association among Indian respondents but a small and significant association among U.S. respondents [F (1, 898) = 64.33, p < .001; std. β = .259]. Multivariate regressions from the SEM support this positive and significant relationship among U.S. respondents (std. β = .819, p < .001), and confirm no significant association for Indian respondents. Therefore, H1a is partially supported. H1b. This hypothesis posits that altruistic values are positively related to ecological worldviews. For U.S. respondents, altruistic values are positively and significantly correlated with both environmental concern (r = .363, p < .001) and anthropocentricism (r = .106, p < .01). A linear regression shows that altruistic values explain 36.3% of the variance in environmental concerns among U.S. respondents [F (1, 898) = 136.290, p < .001; std. β = .363], and 10.6% of the variance in anthropocentricism [F (1, 898) = 10.134, p < .01; std. β = .106]. However, multivariate regressions from the SEM analysis show negative and significant relationships between altruistic values and both environmental concern (std. β = -.357, p < .001), and anthropocentricism (std. β = -.487, p < .001). For Indian respondents, altruistic values are positively correlated with environmental concern (r = .333, p < .001) but non-significantly associated with anthropocentricism. A linear regression shows that altruistic values explain 33.3% of the variance in environmental concern in India [F (1, 918) = 114.332, p < .001; std. β = .333] and have no significant association with anthropocentricism. However, multivariate regressions from the SEM analysis show that altruistic values positively and, indeed, significantly predict 63 both environmental concern (std. β = .741, p < .001) and anthropocentricism (std. β = .285, p < .01). Therefore, H1b is also partially supported given the unexpected negative associations found among U.S. respondents. H1c. As per this hypothesis, egoistic values are negatively associated with ecological worldviews. For U.S. respondents, egoistic values positively correlated with environmental concern (r = .101, p < 0.01) and negatively with anthropocentricism (r = -.310, p < .001). A linear regression shows that egoistic values explained just about a percent of the variance in environmental concern [F (1, 898) = 9.242, p < .01; std. β = .10] and about 9.5% of the variance in anthropocentricism [F (1, 898) = 95.261, p < .001; std. β = -.31]. The multivariate SEM analysis shows that egoistic values negatively predict environmental concerns (std. β = -.07, p < .05) and anthropocentricism (std. β = -.513, p < .001). For Indian respondents, egoistic values correlated negatively with anthropocentricism (r = -.376, p < .001) but non-significantly with environmental concern. Linear regressions show that egoistic values explain 37.6% of the variance in anthropocentricism [F (1, 918) = 151.323, p < .001; std. β = -.376] but had no significant predictive influence on environmental concern. However, these relationships were, in fact, significant in the SEM multivariate analysis. Egoistic values negatively predicted both environmental concern (std. β = -.142, p < .01) and anthropocentricism (std. β = -.58, p < .001). Therefore, based on the multivariate analyses, this hypothesis was supported. H2. According to this hypothesis, ecological worldviews are posited to be positively associated with awareness to consequences. For U.S. respondents, both environmental concern (r = .489, p < .001) and anthropocentricism (r = .161, p < .001) were positively correlated with awareness of consequences. Linear regressions show that environmental 64 concern and anthropocentricism explain 24.2% of the variance in awareness of consequences [F (2, 897) = 143.479, p < .001; std. β for environmental concern = .517, p < .001; std. β for anthropocentricism = -.064, p < .05]. Multivariate SEM analysis shows that environmental concern positively predicts awareness of consequences (std. β = .723, p < .001), but anthropocentricism negatively predicts awareness of consequences (std. β = - .094, p < .05). For Indian respondents, environmental concern was positively correlated with awareness of consequences (r = .452, p < .001) but anthropocentricism does not significantly correlate with awareness of consequences. The linear regression shows that environmental concern and anthropocentricism explain 20.7% of the variance in awareness of consequences [F (2, 917) = 119.871, p < .001; std. β for environmental concern = .453, p < .001; std. β for anthropocentricism = -.05, p = not significant]. Again, in the multivariate analysis, both these relationships were significant such that environmental concern positively predicted awareness of consequences (std. β = .84, p < .001) whereas anthropocentricism negatively predicted awareness of consequences (std. β = -.153, p < .001). Therefore, based on the negative associations between anthropocentricism and awareness of consequences in both groups, this hypothesis is partially supported. H3. This hypothesis posits that awareness of consequences is positively associated with ascription of responsibility. For U.S. respondents, awareness of consequences correlated positively with ascription of responsibility (r = .453, p < .001) and explained 20.5% of the variance in ascription of responsibility as seen in a linear regression [F (1, 898) = 232.022, p < .001; std. β = .453]. The multivariate SEM analysis shows that 65 awareness of consequences strongly and positively predicted ascription of responsibility (std. β = .651, p < .001). For Indian respondents also, awareness of consequences positively correlated with ascription of responsibility (r = .464, p < .001) and explained 21.6% of the variance in ascription of responsibility [F (1, 918) = 252.407, p < .001; std. β = .464]. As per the multivariate analysis, awareness of consequences strongly predicted ascription of responsibility (std. β = .738, p < .001). Therefore, based on the SEM analyses, H3 is supported. H4. According to this hypothesis, ascription of responsibility is positively associated with personal norms. For U.S. respondents, ascription of responsibility correlated positively with personal norms (r = .652, p < .001) and explained 42.5% of the variance in personal norms [F (1, 898) = 664.832, p < .001; std. β = .652]. The SEM analysis shows that ascription of responsibility strongly and positively predicted personal norms (std. β = .808, p < .001). Among Indian respondents also, ascription of responsibility was positively correlated with personal norms (r = .443, p < .001) and explained 19.6% of the variance in personal norms [F (1, 918) = 224.330, p < .001; std. β = .443]. The multivariate analysis shows that ascription of responsibility positively predicts personal norms (std. β = .623, p < .001). Therefore, this hypothesis is supported. H5. This hypothesis suggests that personal norms will be positively associated with donation intentions to reduce wildlife crime. In the U.S. sample, personal norms positively and strongly correlate with donation intentions (r = .73, p < .001) and explained 53.2% of the variance in donation intentions [F (1, 898) = 1021.899, p < .001; std. β = .73]. The SEM 66 analysis shows that personal norms strongly and positively predicted intentions to donate money to reduce wildlife crime (std. β = .779, p < .001). For Indian respondents also, personal norms positively correlated with donation intentions (r = .644, p < .001) and explains 41.5% of the variance in donation intentions [F (1, 918) = 650.620, p < .001; std. β = .644]. The SEM analysis also shows that personal norms strongly and positively predicted intentions to donate money to reduce wildlife crime (std. β = .75, p < .001). Therefore, this hypothesis is also supported. Table 11 and Table 13 provide a comparative assessment of the multivariate regression outputs for SEMs in which this original VBN model was tested across both countries. Table 11 - Multivariate regressions testing the NEP-based value-belief-norm model in India Relationship between Hypothesis predictor and Estimate Std. Err. z-value Sig. Std.all supported outcome BioVal → EnvConcern -0.017 0.148 -0.112 0.911 -0.011 H1a – Rejected AltVal → EnvConcern 1.082 0.190 5.688 0.000 0.741 H1b – Supported EgoVal → EnvConcern -0.208 0.066 -3.161 0.002 -0.142 H1c – Supported BioVal → Anthro -0.191 0.117 -1.625 0.104 -0.154 H1a – Rejected AltVal → Anthro 0.353 0.124 2.855 0.004 0.285 H1b – Supported EgoVal → Anthro -0.719 0.067 -10.752 0.000 -0.580 H1c – Supported EnvConcern → AC 1.035 0.126 8.224 0.000 0.840 H2 – Supported Anthro → AC -0.222 0.061 -3.664 0.000 -0.153 H2 - Rejected AC → AR 0.607 0.065 9.334 0.000 0.738 H3 – Supported AR → PN 0.538 0.046 11.779 0.000 0.623 H4 – Supported PN → Intention 0.885 0.057 15.590 0.000 0.750 H5 - Supported BioVal, AltVal, EgoVal = Biospheric, Altruistic, and Egoistic values, respectively. EnvConcern = Environmental concern (NEP). Anthro = Anthropocentricism (NEP). AC = Awareness of consequences. AR = Ascription of responsibility. PN = Personal norms. 67 Table 12 - Std. covariances and correlated errors in the NEP-based VBN model in India Relationship between predictor and Std. Estimate z-value Sig. Std.all outcome Err. PN4 ~~ PN5 0.161 0.027 5.928 0.000 0.262 Nep2 ~~ Nep10 -0.507 0.103 -4.926 0.000 -0.210 Nep10 ~~ Nep14 0.390 0.093 4.217 0.000 0.170 Nep5 ~~ Nep15 0.188 0.040 4.651 0.000 0.181 Biospheric values ~~ Altruistic values 0.805 0.026 30.852 0.000 0.805 Biospheric values ~~ Egoistic values 0.098 0.042 2.353 0.019 0.098 Altruistic values ~~ Egoistic values 0.117 0.046 2.549 0.011 0.117 Table 13 - Multivariate regressions testing the NEP-based value-belief-norm model in the USA Relationship between Hypothesis predictor and Estimate Std. Err. z-value Sig. Std.all supported outcome BioVal → EnvConcern 1.476 0.130 11.330 0.000 0.989 H1a – Supported AltVal → EnvConcern -0.533 0.105 -5.095 0.000 -0.357 H1b – Rejected EgoVal → EnvConcern -0.105 0.051 -2.070 0.038 -0.070 H1c – Supported BioVal → Anthro 1.146 0.124 9.257 0.000 0.819 H1a – Supported AltVal → Anthro -0.681 0.112 -6.093 0.000 -0.487 H1b – Rejected EgoVal → Anthro -0.718 0.067 -10.682 0.000 -0.513 H1c – Supported EnvConcern → AC 0.674 0.052 12.893 0.000 0.723 H2 – Supported Anthro → AC -0.094 0.038 -2.449 0.014 -0.094 H2 – Rejected AC → AR 0.616 0.049 12.677 0.000 0.651 H3 – Supported AR → PN 1.040 0.080 12.937 0.000 0.808 H4 – Supported PN → Intention 0.734 0.046 15.800 0.000 0.779 H5 - Supported BioVal, AltVal, EgoVal = Biospheric, Altruistic, and Egoistic values, respectively. EnvConcern = Environmental concern (NEP). Anthro = Anthropocentricism (NEP). AC = Awareness of consequences. AR = Ascription of responsibility. PN = Personal norms. Table 14 - Std. covariances and correlated errors in the NEP-based VBN model in the USA Relationship between predictor and Std. Estimate z-value Sig. Std.all outcome Err. PN4 ~~ PN5 0.558 0.044 12.762 0.000 0.548 AltV1 ~~ AltV3 0.335 0.035 9.562 0.000 0.412 AR1 ~~ AR3 0.470 0.072 6.507 0.000 0.356 68 Table 14 (cont’d) Nep15 ~~ Nep10 0.410 0.062 6.650 0.000 0.273 Biospheric values ~~ Altruistic values 0.769 0.021 36.278 0.000 0.769 Biospheric values ~~ Egoistic values 0.165 0.038 4.294 0.000 0.165 Altruistic values ~~ Egoistic values 0.139 0.041 3.422 0.001 0.139 As pointed out in the previous section of this chapter, there were discrepancies not only in how the relationships between VBN latent factors manifested in the U.S. sample and the Indian sample, but also in how reliable and valid these commonly used VBN scales and measures seemed to be outside the United States. While the relationships between awareness of consequences, ascription of responsibility, and personal norms were expectedly positive and significant in both U.S. and Indian groups, these findings overall suggest that a cross-validation of commonly used measures and scales associated with the VBN model outside the United States, especially in developing economies, is crucial to establish reliability and validity of measures and, by extension, of results. Theory of Planned Behavior Exploratory factor analysis for TPB concepts revealed that all items measuring each individual latent factor loaded on their respective single-factor solutions, thereby suggesting strong convergence with the latent factor. Further, I assessed convergent validity by examining the Average Variance Extracted (AVE) which is calculated using the sum of the squared standardized factor loadings of items, and the number of items. AVE values for all TPB latent factors – attitudes, subjective norms, perceived behavioral control, and intention – were greater than 0.5 in both Indian and USA samples, suggesting that the item measures did closely explain their respective latent variables. In addition to the AVE 69 values, I also ascertained convergent validity by assessing Cronbach’s alpha for internal reliability (alpha for all scales was greater than 0.7), and only retained those items that had factor loadings higher than 0.32 in the EFA and higher than 0.50 in the CFA (Cheung & Wang, 2017; Miller - Carpenter, 2018). I also examined whether the value of the square root of the AVE values was greater than the correlations between latent variables to determine whether discriminant validity could be established. The only two latent variables for which the square root of AVE was not greater than latent variable correlations, were subjective norms and perceived behavioral control but this was the case only for the India dataset. While this suggests that subjective norms and perceived behavioral control measures may be slightly related to other measures instead of being sufficiently unrelated. However, the square root of AVE values for these latent variables only fell short of their correlations with other latent variables by values of 0.02 and 0.03, respectively, potentially suggesting that the technical lack of discriminant validity here may not be as severe. Table 15 shows the means and standard deviations for the theory of planned behavior (TPB) concepts as seen in both India and the United States. Table 15 - Means, standard deviations, and reliability for theory of planned behavior concepts Variables and Measures USA India Mean S.D. Mean S.D. Attitude (Att) 5.27 1.315 5.87 .965 (USA α = .957; India α = .905) Att1: I think the idea of donating money to conservation organizations to help reduce wildlife 5.44 1.378 5.95 1.077 crime is very positive. Att2: I think the idea of donating money to conservation organizations to help reduce wildlife 5.28 1.419 5.90 1.093 crime is very responsible. 70 Table 15 (cont’d) Att3: I think the idea of donating money to conservation organizations to help reduce wildlife 5.15 1.443 5.66 1.253 crime is very intelligent. Att4: I think the idea of donating money to conservation organizations to help reduce wildlife 5.26 1.426 5.95 1.104 crime is very useful. Att5: I think the idea of donating money to conservation organizations to help reduce wildlife 5.24 1.458 5.89 1.134 crime is very ecologically helpful. Subjective Norms (SN) 4.13 1.449 5.54 1.065 (USA α = .884; India α = .809) SN1: Most people who are important to me think that one should donate money to conservation 4.13 1.597 5.56 1.231 organizations to help reduce wildlife crime. SN2: Most people who are important to me expect that I will donate money to conservation 3.77 1.713 5.40 1.346 organizations to help reduce wildlife crime. SN3: Those people whose opinions I value would donate money to conservation organizations to 4.47 1.509 5.68 1.176 help reduce wildlife crime. Perceived Behavioral Control (PBC) 4.44 1.229 5.36 1.082 (USA α = .828; India α = .864) PBC1: I am confident that if I want, I can donate 4.75 1.684 5.65 1.285 money to conservation organizations to help reduce wildlife crime. PBC2: I have sufficient resources to be able to 3.29 1.910 4.98 1.616 donate money to conservation organizations to help reduce wildlife crime. PBC3: I have enough time to donate money to 4.24 1.717 5.17 1.475 conservation organizations to help reduce wildlife crime. PBC4: I have enough opportunities to donate 4.06 1.789 5.11 1.548 money to conservation organizations to help reduce wildlife crime. PBC5: Donating money to conservation 5.53 1.577 5.44 1.400 organizations to help reduce wildlife crime is completely up to me.* PBC6: If I donated money to conservation 4.78 1.495 5.79 1.158 organizations, it would help them to reduce wildlife crime. 71 Table 15 (cont’d) Intention (Int) 3.59 1.701 5.44 1.259 (USA α = .936; India α = 0.911) Int1: I am willing to donate money to conservation 3.94 1.846 5.52 1.319 organizations within the next month to help reduce wildlife crime. Int2: I plan to donate money to conservation 3.49 1.788 5.41 1.353 organizations within the next month to help reduce wildlife crime. Int3: I will definitely donate money to 3.35 1.787 5.38 1.428 conservation organizations within the next month to help reduce wildlife crime. * - Item removed from reliability analysis and SEM due to poor factor loadings in EFA and CFA Confirmatory Factor Analysis While measures used to assess the latent variables in the TPB model may have shown high reliability (Cronbach’s alpha reported in Table 15), convergent validity, and for the most part, discriminant validity as well, it is important to recall that not only were these measures adapted to measure wildlife crime-related conservation behaviors from previous research on different kinds of environmental behaviors, but this is also the only study in which the theory of planned behavior has been applied to understand salient factors that may drive donation intentions to reduce wildlife crime. Given this two-fold contextual setting, I was interested in examining how well the different latent variables load onto their indicator items, and to identify the theoretical structures of these variables when applied to a new context and in a cross-national setting. A confirmatory factor analysis (CFA) was performed to test an a priori defined conceptual model (see Figure 1) using both stacked cross-national data, and a multi-group confirmatory factor analysis (MGCFA) was also performed to test cross-national differences and assess measurement invariance or equivalence testing. Here too, I used R studio (version 2022.02.3+492) to run all CFA and structural equation modeling (SEM) analyses 72 as well as to plot path diagrams. Again, I used the following packages with R: lavaan; haven; semTools; semPlot; equaltestMI; MVN; tidyverse; and magrittr. As multivariate normality had already been assessed and I had established that the data did not have a normal multivariate distribution, I again used robust maximum likelihood estimation and reported robust fit statistics with Satorra-Bentler Chi-square and std. error values for the value-belief-norm CFAs and SEMs (CenterStat, 2019; Finney & DiStefano, 2006; Hoyle & Isherwood, 2013; Satorra & Bentler, 1994). Additionally, although only a handful of questions had missing responses and no systematic pattern of missing data emerged, I used the full information maximum likelihood approach in conducting my CFAs and SEMs (Hoyle & Isherwood, 2013). To examine whether the factor structure for the latent variables in the TPB model was the same across both countries, I applied the measurement invariance or equivalence testing steps (Boomsma et al., 2012; Pirralha, 2020a, 2020b, 2020c), beginning by looking at whether I could establish configural invariance in an unrestricted model in which factors were free to vary. In running the configural invariance model for a multi-group CFA testing the measurement model of the theory of planned behavior, all indicator items except ‘PBC5’, an indicator of perceived behavioral control, were specified as measures of their respective latent variables (see Table 15); ‘PBC5’ was removed because it did not meet factor loading cut-offs during both the EFA and the CFA. The MGCFA indicates that the measurement model fit the data well for both countries: Robust χ2 = 878.601, df = 196, p < .001 (Robust χ2 for India = 284.346; Robust χ2 for USA = 594.256); Robust CFI = .954; Robust TLI = .943; Robust RMSEA = .077; SRMR = .062 (full 73 model output in Appendix B). Figure 7 and Figure 8 visually depict the measurement models assessed in this multi-group CFA, showing how various indicators loaded on to their respective latent variables. In testing equivalence, the next step was to add restrictions to this model, beginning with making item loadings on latent factors equal across both India and USA groups. While this restricted model was also a good fit for the data, a likelihood ratio test (LRT) meant for non-normal data – the scaled Chi-square difference test using the Satorra-Bentler method – was conducted to compare the unrestricted model with the restricted one found that both models were statistically significantly different such that weak invariance could not be established. This suggests that the measurement models for the theory of planned behavior are not similar across both the groups, meaning that values of factor loadings and latent means are likely not directly comparable across groups. Figure 7 - Theory of planned behavior measurement model – USA 74 Figure 8 - Theory of planned behavior measurement model - India Measurement models assessed in the MGCFA demonstrate how well the indicator items measure their latent variables, however, to examine relationships between the latent variables themselves, I used a structural equation model to examine whether it fit the data as well. Structural Equation Model Adding a regression to the above multi-group CFA model, I examined whether the theoretical factor structures – the relationships between latent factors – are the same across both groups. The SEM model was also tested for measurement invariance and since weak invariance could not be established, and it appeared that regression coefficients and factor loadings varied between countries, I conducted a parallel country models analysis. The SEM for India showed good model fit and explained 71.9% of the variance in intentions to donate for wildlife crime reduction: χ2 = 443.524, df = 98, p < .001; CFI = .965; TLI = .958; 75 RMSEA = .062; SRMR = .039. The SEM for USA explained 66.2% of the variance in donation intentions to reduce wildlife crime: χ2 = 912.975, df = 98, p < .001; CFI = .936; TLI = .921; RMSEA = .096; SRMR = .093. For full model output, see Appendix C. Hypothesis Testing Based on the SEM analysis, I present results of the proposed relationships between TPB constructs here (Ajzen, 1991). H6. This hypothesis posited that attitudes towards monetary donation are positively associated with donation intentions to reduce wildlife crime. Correlational analysis showed that attitudes were positively and significantly related to donation intentions but that these correlations were moderate in both India (α = .599, p = .01) and the United States (α = .538, p = .01). A linear regression showed that attitudes significantly predicted 35.9% of the variance in donation intentions in India [F (1, 918) = 513.76, p < .001; std. β = .599] and 29% of the variance in donation intentions in the U.S. [F (1, 898) = 366.14, p < .001; std. β = .538]. However, examining multivariate regressions and relationships in the SEM model for the Indian sample shows that attitudes had no significant effect on donation intentions (std. β = .016, p = .736). Even in the USA sample, attitudes did not significantly predict donation intentions to reduce wildlife crime (std. β = .029, p = .375). Therefore, when examined as a part of multivariate relationships, attitudes were not significantly associated with donation intentions in both samples. H6, is therefore, not supported by the SEM evidence. I include the standardized beta coefficients of the linear regression and the multivariate SEM regression here because they are directly comparable, so assessing the difference between them can demonstrate how the standardized beta coefficient changed from a linear to multivariate analysis, in this case potentially indicating that the linear relationship between 76 attitude and donation intentions may have been capturing some indirect effects given how much smaller the coefficients in the SEM analysis are. H7. This hypothesis suggested that subjective norms are positively associated with donation intentions to reduce wildlife crime. Subjective norms were positively and significantly correlated with donation intentions in both India (α = .635, p < .001) and the U.S. (α = .670, p < .001). Linear regressions show that subjective norms predicted 40.3% of the variance in intentions in India [F (1, 918) = 618.786, p < .001; std. β = .635] and 44.9% of the variance in donation intentions the U.S. [F (1, 898) = 731.271, p < .001; std. β = .67]. When examining multivariate regressions and the SEM output, subjective norms moderately but significantly predicted donation intentions in the U.S. (std. β = .36, p < .001) and this effect was much smaller for the Indian sample (std. β = .183, p < .01). Here, too, comparisons between linear regression standardized beta coefficients and SEM multivariate regression beta coefficients shows that the linear regressions were likely accounting for some indirect effects as the coefficients reduced once examined as part of the SEM. Even so, both in bivariate and multivariate examinations, subjective norms were, indeed, positively associated with donation intentions to reduce wildlife crime in both samples, thereby, supporting H7. H8. According to this hypothesis, perceived behavioral control would be positively associated with donation intentions to reduce wildlife crime. Perceived behavioral control and donation intentions were strongly and positively correlated in India (α = .743, p < .001) and the U.S. (α = .664, p < .001). Linear regressions showed that perceived behavioral control predicted 55.2% of the variance in donation intentions in India [F (1, 918) = 1132.465, p < .001; std. β = .743] and about 44% of the variance in donation intentions in 77 the U.S. [F (1, 898) = 708.057, p < .001; std. β = .664]. The multivariate SEM analysis confirmed that perceived behavioral control strongly and positively predicted donation intentions to reduce wildlife crime in India (std. β = .687, p < .001), and in the U.S. (std. β = .511, p < .001). Comparing the standardized beta coefficients across linear and multivariate regressions assessing the influence of perceived behavioral control on donation intentions shows that most of the effect captured by the linear relationships across both samples continues to be captured in the multivariate analysis as well. While not explicitly hypothesized, the theory of planned behavior also suggests that the three exogenous variables – attitudes, subjective norms, and perceived behavioral control – would be positively correlated with one another. The multivariate SEM model confirms these correlations; for instance, attitude and subjective norms were positively and strongly correlated (India: r = .787, p < .001; USA: r = .671, p < .001), as were attitude and perceived behavioral control (India: r = .727, p < .001; USA: r = .524, p < .001), and subjective norms and perceived behavioral control (India: r = 789, p < .001; USA: r = .653, p < .001). Figure 9 and Figure 10 visually represent the structural model described above. As with the MGCFA model, equivalence testing found that weak invariance could not be established even for the SEM models for the theory of planned behavior. Therefore, the structural model also differed between groups as shown below. Table 16 and Table 18 provide regression coefficients and other information from the SEM analyses for the U.S. and Indian samples, respectively, whereas Table 17 and Table 19 include the covariances for both samples. 78 Table 16 - Multivariate SEM regressions testing the theory of planned behavior in India Relationship between Std. z- Hypothesis Estimate Sig. Std.all predictor and outcome Err. value supported Attitude → Intention 0.027 0.088 0.302 0.763 0.014 H6 – Rejected Subjective norm → Intention 0.316 0.112 2.829 0.005 0.168 H7 – Supported Perceived behavioral control 1.317 0.118 11.171 0.000 0.698 H8 - Supported → Intention Table 17 - Std. covariances in theory of planned behavior model in India Relationship between predictor and Std. z- Estimate Sig. Std.all outcome Err. value Attitude ~~ Subjective Norms 0.787 0.019 42.214 0.000 0.787 Attitude ~~ Perceived behavioral control 0.728 0.020 36.397 0.000 0.728 Subjective norm ~~ Perceived behavioral 0.788 0.020 40.313 0.000 0.788 control Table 18 - Multivariate regressions testing the theory of planned behavior in the USA Relationship between Std. z- Hypothesis Estimate Sig. Std.all predictor and outcome Err. value supported Attitude → Intention 0.051 0.057 0.888 0.375 0.029 H6 – Supported Subjective norm → Intention 0.619 0.074 8.362 0.000 0.360 H7 – Supported Perceived behavioral control 0.879 0.074 11.948 0.000 0.511 H8 - Supported → Intention Table 19 - Std. covariances in the theory of planned behavior model in the USA Relationship between predictor and Std. Estimate z-value Sig. Std.all outcome Err. Attitude ~~ Subjective Norms 0.671 0.021 31.686 0.000 0.671 Attitude ~~ Perceived behavioral control 0.524 0.028 18.579 0.000 0.524 Subjective norm ~~ Perceived behavioral 0.653 0.025 26.482 0.000 0.653 control 79 Figure 9 - Structural model of the theory of planned behavior - USA Figure 10 - Structural model of the theory of planned behavior - India 80 Other Hypotheses In this section, I present results of some inter-theory relationships I hypothesized. These relationships were not subjected to CFAs or SEMs in models that combined constructs from the theory of planned behavior and the value-belief-norm theory. However, the correlations and linear regression results presented in this section can serve as precedent for future research that may be interested in developing integrated theoretical models. Other hypotheses focusing specifically on the wildlife value orientations construct are examined in the next section. H9a. Awareness of consequences will be positively associated with attitudes towards donating money to reduce wildlife crime (Carfora et al., 2021; Han, 2015). A bivariate correlation analysis shows that awareness of consequences positively correlates with attitudes toward donation intentions in the U.S. sample (r = .504, p < .001) and in the Indian sample (r = .403, p < .001). Linear regressions show that awareness of consequences explained 25.4% of the variance in attitude in the U.S. [F (1, 898) = 306.376, p < .001; std. β = .504], and explained 16.3% of the variance in attitude in India [F (1, 918) = 178.484, p < .001; std. β = .403]. Thus, a positive association between these two variables suggests that this hypothesis is supported. H9b. Awareness of consequences will be positively associated with subjective norms (Han, 2015). Positive correlations emerged both among U.S. respondents (r = .416, p < .001) and among Indian respondents (r = .364, p < .001). Awareness of consequences explained 17.3% of the variance in subjective norms in the U.S. [F (1, 898) = 187.382, p < .001; std. β = .416], and 13.3% of the variance in India [F (1, 918) = 140.244, p < .001; std. β = .364]. This hypothesis is, therefore, supported. 81 H9c. Awareness of consequences will be positively associated with perceived behavioral control (Han, 2015). Positive correlations were found among both U.S. respondents (r = .344, p < .001) and Indian respondents (r = .35, p < .001). Awareness of consequences explained 11.8% of the variance in perceived behavioral control among U.S. respondents [F (1, 898) = 120.363, p < .001; std. β = .344], and 12.3% of the variance in Indian responses [F (1, 918) = 128.182, p < .001; std. β = .35]. Findings suggest that these hypotheses are supported. H10b. Ecological worldviews are positively associated with attitudes towards donating money to reduce wildlife crime (Carfora et al., 2021). Both environmental concern (r = .404, p < .001) and anthropocentricism (r = .072, p < .05) are positively correlated with attitudes for U.S. respondents, whereas for Indian respondents, only environmental concern is positively correlated with attitude (r = .248, p < .001) while anthropocentricism is negatively correlated with attitudes (r = -.16, p < .001). As ecological worldviews are conceptualized as preceding attitudes in the causal chain of variables explaining environmental behaviors, I ran a linear regression analysis which shows that environmental concern and anthropocentricism, together, explain 17.7% of the variance in attitude among U.S. respondents [F (2, 897) = 96.4, p < .001], but only about 9% of the variance in attitude among Indian respondents [F (2, 917) = 44.033, p < .001]. This hypothesis is supported based on the regression results. H13. Ascription of responsibility is positively associated with perceived behavioral control (Carfora et al., 2021). Both for U.S. (r = .478, p < .001) and Indian respondents (r = .426, p < .001), ascription of responsibility was correlated positively with perceived behavioral control. In the U.S. sample, ascription of responsibility explains 22.8% of the 82 variance in perceived behavioral control [F (1, 898) = 265.357, p < .001], while for the Indian sample, it explains about 18% of the variance in perceived behavioral control [F (1, 918) = 203.189, p < .001]. Therefore, this hypothesis is supported. Wildlife Value Orientations in the Value-Belief-Norm Theory As mentioned in Chapters 1 and 2, part of this dissertation involves assessing whether the wildlife value orientations construct could be an appropriate substitute for the NEP scale as an operationalization of the value-belief-norm theory’s ‘ecological worldviews’ construct. In this section, I present the results of those hypotheses that involved examining the relationships between wildlife value orientations and some VBN constructs, as informed by an extensive review of interdisciplinary literature that sets the context for these hypotheses. I further present the results of a CFA conducted to assess the goodness of fit of a measurement model in which mutualistic wildlife value orientations replace the NEP scale as a measure of ecological worldviews, and the results of an SEM showing how the latent variables relate to one another in a model including mutualistic wildlife value orientations. Hypothesis Testing H10a. Ecological worldviews are positively associated with wildlife value orientations. More specifically, given that both these constructs are multi-dimensional latent factors, I expected environmental concern to be positively correlated with mutualistic wildlife value orientations but negatively correlated with domination wildlife value orientations. Conversely, I expected anthropocentricism to correlated positively with domination wildlife value orientations and negatively with mutualistic wildlife value orientations. 83 Among U.S. respondents, environmental concern correlated positively with mutualistic wildlife value orientations (r = .493, p < .001) and negatively with domination wildlife value orientations (r = -.376, p < .001). Among Indian respondents, environmental concern correlated positively with mutualistic wildlife value orientations (r = .401, p < .001) and negatively with domination wildlife value orientations (r = -.205, p < .001). Anthropocentricism correlated negatively with domination wildlife value orientations (r = -.50, p < .001) and positively with mutualistic wildlife value orientations (r = .213, p < .001) for U.S. respondents. Among Indian respondents, anthropocentricism correlated negatively with both domination (r = -.528, p < .001) and mutualistic (r = -.088, p < .01) wildlife value orientations. Given that both ecological worldviews and wildlife value orientations are sets of beliefs, I did not run regression analyses because I did not expect to eventually test a causal relationship between these factors. However, based on the mixed correlations, this hypothesis was only partially supported. H11a. Mutualistic wildlife value orientations will be positively associated with attitudes (Teel et al., 2007b). Both for the U.S. (r = .460, p < .001) and Indian respondents (r = .537, p < .001), mutualistic wildlife value orientations were positively correlated with attitudes towards donating money to reduce wildlife crime. In fact, mutualistic wildlife value orientations explained 21.2% of the variance in attitudes among U.S. respondents [F (1, 898) =241.181, p < .001], and 28.8% of the variance in attitudes among Indian respondents [F (1, 918) = 371.054, p < .001]. On the other hand, and as expected, domination wildlife value orientations were negatively correlated with attitudes for both 84 U.S. (r = -.192, p < .001) and Indian (r = -.123, p < .001) respondents. This hypothesis, is therefore, supported. H11b. Mutualistic wildlife value orientations will be positively associated with donation intentions to reduce wildlife crime (Manfredo, Teel, Don Carlos, et al., 2020). Both for U.S. (r = .425, p < .001) and Indian respondents (r = .491, p < .001), mutualistic wildlife value orientations correlated positively with intentions to donate money to reduce wildlife crime. For U.S. respondents, mutualistic wildlife value orientations explained 18% of the variance in donation intentions [F (1, 898) = 197.563, p < .001], whereas for Indian respondents, they explained about 24% of the variance in donation intentions [F (1, 918) = 291.067, p < .001]. This hypothesis is also, therefore, supported. H12. Biospheric values will be positively associated with mutualistic wildlife value orientations (Oh et al., 2021). For both U.S. (r = .596, p < .001) and Indian respondents (r = .516, p < .001), biospheric values were positively correlated with mutualistic wildlife value orientations, and, as expected but not explicitly stated here, were negatively correlated with domination wildlife value orientations (U.S.: r = -.290, p < .001) (India: r = -.226, p < .001). Biospheric values explained 35.5% of the variance in mutualistic wildlife value orientations for U.S. respondents [F (1, 898) = 494.339, p < .001], and 26.6% of the variance for Indian respondents [F (1, 918) = 333.197, p < .001]. Further, results from the multivariate SEM analysis presented below show that biospheric values did, indeed, positively predict mutualistic wildlife value orientations in both U.S. (std. β = .817, p < .001) and Indian (std. β = .42, p < .001) groups, thereby supporting this hypothesis. 85 Measurement Model Before running the confirmatory factor analysis, I conducted an EFA and ran the Bartlett’s sphericity test along with the Kaiser-Meyer-Olkin test on items measuring mutualistic wildlife value orientations. KMO values were above 0.9 for both India and the U.S. group of respondents, and Bartlett’s test of sphericity was significant in both groups as well. The EFA using maximum likelihood estimation for factor extraction and promax rotation showed that items loaded on a single-factor solution for both groups, and that factor loadings were greater than 0.5 for both groups as well. Moreover, a high Cronbach’s alpha further underscored the reliability of these measures. For the Indian sample, Cronbach’s α was .855 and for the U.S. sample, Cronbach’s α was .902. The mutualistic wildlife value orientations construct, therefore, did not have the same problem with scale reliability that NEP measures did. I conducted confirmatory factor analyses to determine the measurement model for mutualistic wildlife value orientations for both groups and found that model fit was better for the Indian sample [Robust χ2 = 103.955, df = 27, p < .001, Robust CFI = .96, Robust TLI =.946, Robust RMSEA = .068, SRMR = .033] than it was for the U.S. sample which required re-specification after including correlated errors between items [Robust χ2 = 228.489, df = 26, p < .001, Robust CFI = .927, Robust TLI =.898, Robust RMSEA = .117, SRMR = .047]. Figure 11 and Figure 12 depict the measurement model for mutualistic wildlife value orientations in the Indian and U.S. groups, respectively. For specific indicators used to measure this construct, see Table 8. 86 Figure 11 - Measurement model for mutualistic wildlife value orientations in India Figure 12 - Measurement model for mutualistic wildlife value orientations in the USA While the model fit for the Indian sample was good without any correlated errors suggested by examining modification indices, this was not true for the U.S. measurement model which required at least one correlated error to be added to the model specification for the fit indices to be within acceptable ranges. 87 Structural Equation Model To compare the path structure and predictive power of models in which the New Ecological Paradigm scale is replaced by mutualistic wildlife value orientations (WVOs) as part of the larger VBN model, I conducted an SEM analysis following the same protocols I used above, such as testing for measurement invariance and comparing equivalence. I ran both CFAs and SEMs for the following models: 1) Overall stacked data without grouping by country and without constraints, 2) Stacked data with country-level grouping and without constraints (i.e., configural invariance), 3) Stacked data with country-level grouping and factor loadings constrained to being equal (i.e., metric invariance), and 4) Stacked data with country-level grouping and factor loadings and intercepts constrained to being equal (i.e., scalar invariance) (Pirralha, 2020a, 2020b, 2020c). I ran a likelihood ratio test in lavaan to compare model fit for testing equivalence. I found that country-level differences mattered such that the grouped model had better fit and was significantly different from the overall stacked model without country-level grouping. This was consistent with the fact that the measurement model for mutualistic wildlife value orientations was itself different for the U.S. and India, as was the initial measurement model that tested the value-belief-norm model. To compare whether the NEP-based VBN model or the Mutualistic WVO-based VBN model was the better fit for Indian and U.S. respondents, I looked at the model fit statistics and R2 value since a likelihood ratio test that compares models can only be used when the models being compared are nested models. For both groups, the model with mutualistic wildlife value orientations (WVOs) had slightly higher predictive power. For Indian respondents, though, the WVO-based VBN 88 model had slightly better fit statistics and explained 56.5% of the variance in intentions to donate money to reduce wildlife crime [χ2 = 1415.618, df = 515, p < .001, CFI = .928, TLI =.922, RMSEA = .044, SRMR = .069]. Figure 13 shows the structural model for the value- belief-norm theory with WVOs replacing NEP in India. Table 20 below shows the fit statistics and R2 values for both value-belief-norm structural equation models as they apply to Indian respondents – the original one with NEP measures, and the one replacing NEP with mutualistic WVOs. Multivariate regressions and standardized covariances for these models are presented in Tables 22 through 25. Table 20 - Fit statistics for value-belief-norm SEMs with NEP and WVOs for India Model Chi-square df AIC BIC CFI TLI RMSEA SRMR R2 VBN_NEP 1321.9 509 91030 91609 .925 .918 .042 .07 .562 VBN_WVO 1415.618 515 83702 84252 .928 .922 .044 .069 .565 Table 21 - Fit statistics for value-belief-norm SEMs with NEP and WVOs for USA Model Chi- df AIC BIC CFI TLI RMSEA SRMR R2 square VBN_NEP 2214.3 576 101868 102473 .912 .903 .056 .109 .608 VBN_WVO 2479.8 579 99206 99797 .907 .899 .06 .128 .612 For U.S. respondents, the fit statistics were better for the NEP-based VBN model. Table 21 provides fit statistics and R2 values for both U.S. structural equation models testing the VBN models – the original one including NEP measures, and the one replacing NEP with mutualistic WVOs. Figure 14 shows the WVO-based VBN structural model for USA. 89 Figure 13 - Value-belief-norm SEM with mutualistic wildlife value orientations in India First, I compared the multivariate regression outputs from the different structural equation models I ran to test the model fit and predictive power of a value-belief-norm model with mutualistic wildlife value orientations replacing the conventional NEP scale as a measure of ecological worldviews. Table 22 and Table 24 show how the relationships between latent factors in this model played out both in the U.S. and in India, and whether the proposed hypotheses were supported in either or both of these samples. 90 Figure 14 - Value-belief-norm SEM with mutualistic wildlife value orientations in the USA Table 22 - Multivariate regressions testing the WVOs-based value-belief-norm model in the USA Relationship between Std. z- Hypothesis Estimate Sig. Std.all predictor and outcome Err. value supported BioVal → MutWVO 1.166 0.100 11.640 0.000 0.817 H1a - Supported AltVal → MutWVO -0.250 0.084 -2.991 0.003 -0.175 H1b – Rejected EgoVal → MutWVO 0.116 0.045 2.558 0.011 0.081 H1c – Rejected MutWVO → AC 0.496 0.038 13.032 0.000 0.578 H2 – Supported AC → AR 0.639 0.049 13.168 0.000 0.617 H3 – Supported AR → PN 0.954 0.061 15.750 0.000 0.771 H4 – Supported PN → Intention 0.797 0.045 17.881 0.000 0.782 H5 - Supported BioVal, AltVal, EgoVal = Biospheric, Altruistic, and Egoistic values, respectively. MutWVO = Mutual wildlife value orientations. AC = Awareness of consequences. AR = Ascription of responsibility. PN = Personal norms. 91 Table 23 - Std. covariances and correlated errors in the WVOs-based VBN model in the USA Relationship between predictor and Std. Estimate z-value Sig. Std.all outcome Err. PN4 ~~ PN5 0.571 0.044 12.905 0.000 0.553 CaB3 ~~ CaB5 0.525 0.043 12.211 0.000 0.504 CaB3 ~~ CaB4 0.426 0.042 10.234 0.000 0.430 CaB4 ~~ CaB5 0.422 0.044 9.608 0.000 0.389 AltV1 ~~ AltV3 0.320 0.035 9.198 0.000 0.401 Biospheric values ~~ Altruistic values 0.746 0.022 33.700 0.000 0.746 Biospheric values ~~ Egoistic values 0.164 0.038 4.268 0.000 0.164 Altruistic values ~~ Egoistic values 0.152 0.041 3.742 0.000 0.152 Table 24 - Multivariate regressions testing the WVOs-based value-belief-norm model in India Relationship between Std. z- Hypothesis Estimate Sig. Std.all predictor and outcome Err. value supported BioVal → MutWVO 0.582 0.109 5.336 0.000 0.420 H1a – Supported AltVal → MutWVO 0.397 0.115 3.440 0.001 0.287 H1b – Supported EgoVal → MutWVO 0.133 0.046 2.907 0.004 0.096 H1c – Rejected MutWVO → AC 0.940 0.069 13.722 0.000 0.793 H2 – Supported AC → AR 0.923 0.093 9.935 0.000 0.835 H3 – Supported AR → PN 0.483 0.046 10.578 0.000 0.659 H4 – Supported PN → Intention 0.858 0.055 15.485 0.000 0.752 H5 - Supported BioVal, AltVal, EgoVal = Biospheric, Altruistic, and Egoistic values, respectively. MutWVO = Mutual wildlife value orientations. AC = Awareness of consequences. AR = Ascription of responsibility. PN = Personal norms. Table 25 - Std. covariances and correlated errors in the WVOs-based VBN model in India Relationship between predictor and Std. Estimate z-value Sig. Std.all outcome Err. PN4 ~~ PN5 0.157 0.027 5.828 0.000 0.257 CaB3 ~~ CaB4 0.135 0.023 5.906 0.000 0.235 Biospheric values ~~ Altruistic values 0.806 0.026 30.946 0.000 0.806 Biospheric values ~~ Egoistic values 0.108 0.041 2.629 0.009 0.108 Altruistic values ~~ Egoistic values 0.122 0.045 2.701 0.007 0.122 92 As the tables above show, while most hypotheses were supported across both samples, altruistic values did not have a positive relationship with mutualistic wildlife value orientations in the U.S., whereas egoistic values did not have the expected negative relationships with the wildlife value orientations either. The absence of a negative association between egoistic value and mutualistic wildlife value orientations was also seen among Indian respondents. However, mutualistic wildlife value orientations had a much stronger association with awareness of consequences among Indian respondents than it did among U.S. respondents. Relationships That Mattered Most – Graphical Representations In conducting the SEM multivariate analysis, my objective was to assess how and to what extent the different latent variables in both the planned behavior and value-belief- norm models related to each other, and what, if any, differences there were in how these relationships manifested in samples from the two different countries. In this section, I present the graphical representations of the different structural models testing the two theories that fit the data for each country. Value-Belief-Norm Model with NEP Dimensions Figure 15 depicts the path model structure that fit the data to show relationships between the latent factors in the original value-belief-norm theory model with the two retained NEP dimensions for the U.S. sample. Figure 16 shows the relationships that emerged in the structural model for the India sample. 93 Figure 15 - Structural model of the value-belief-norm theory with only significant paths – USA Figure 16 - Structural model of the value-belief-norm theory with only significant paths – India Theory of Planned Behavior Figure 17 depicts the path model structure that fits the data to show relationships between the latent factors in the theory of planned behavior model for the Indian sample while Figure 18 shows these relationships for the U.S. sample. 94 Figure 17 - Structural model of the theory of planned behavior with only significant paths – India Figure 18 - Structural model of the theory of planned behavior with only significant paths - USA 95 Value-Belief-Norm Model with Wildlife Value Orientations Figure 19 shows the structural path model for relationships between the latent factors in the modified value-belief-norm theory model with mutualistic wildlife value orientations replacing the NEP dimensions for the India sample whereas Figure 20 shows these relationships for the U.S. sample. Figure 19 - Structural model with significant paths for the value-belief-norm theory with mutualistic wildlife value orientations - India Figure 20 - Structural model with significant paths for the value-belief-norm theory with mutualistic wildlife value orientations – USA 96 Predictors That Mattered the Most In the multivariate analysis, I was interested in identifying the factors that seemed to matter the most in predicting intentions to donate money to reduce wildlife crime. To examine this, I ran a multiple linear regression in SPSS to see which of the four direct predictors of intention – attitude towards the behavior, subjective norms, perceived behavioral control, and personal norms – had the highest regression coefficients cross- nationally in predicting donation intentions to reduce wildlife crime. I found that for Indian respondents, perceived behavioral control was most powerful, followed by subjective norms, personal norms, and attitudes, in that order. However, for U.S. respondents, personal norms were most powerful, followed by perceived behavioral control, subjective norms, and attitudes, in that order. Table 26 shows the results of this multiple linear regression comparison. Implications are discussed in the next chapter. Table 26 - Multiple linear regression comparing salient drivers of donation intentions cross-nationally Standardize Unstandardized Collinearity d Country Model Coefficients t Sig. Statistics Coefficients B Std. Err. Beta Tolerance VIF (Constant) -.982 .165 -5.965 .000 Attitude -.088 .041 -.068 -2.144 .032 .419 2.389 Subjective norms .259 .038 .221 6.755 .000 .399 2.509 0 USA 1 Personal norms .474 .048 .413 9.973 .000 .248 4.032 Perceived .430 .037 .311 11.474 .000 .580 1.723 behavioral control 97 Table 26 (cont’d) (Constant) -.231 .171 -1.353 .176 Attitude .096 .042 .074 2.286 .022 .420 2.379 Subjective norms .201 .038 .170 5.343 .000 .429 2.333 1 India 1 Personal norms .168 .043 .134 3.955 .000 .376 2.656 Perceived .568 .037 .489 15.306 .000 .427 2.342 behavioral control a. Dependent Variable: Intention Intention to donate money to conservation organizations to help reduce wildlife crime Broad Overview of Responses and Demographics While testing the theory of planned behavior, although attitudes had no significant influence on predicting donation intentions to reduce wildlife crime in both India and the U.S., more respondents in both samples agreed with the attitudinal measures than those who disagreed. For instance, approximately 78% of U.S. respondents agreed on some level (from somewhat agree to strongly agree) with the statement that donating money to conservation organizations to help reduce wildlife crime is a positive thing, whereas nearly 91% of Indian respondents agreed with this statement on some level. Subjective norms appeared to be more meaningful among Indian respondents than among U.S. respondents. Considering responses to perceived behavioral control measures, the number of Indian respondents who strongly agreed with having sufficient resources to be able to donate money to conservation organizations for wildlife crime reduction was nearly three times the number of U.S. respondents who strongly agreed with the same statement. Conversely, the number of U.S. respondents who strongly disagreed that they had sufficient resources to donate to conservation organizations was 14 times the number of Indian participants who also strongly disagreed. However, when asking about resources, I wanted to be clear 98 that I was not only asking about financial resources but was also interested in knowing whether time or the presence of opportunities to engage in such conservation behaviors are proving to be barriers for behavioral engagement in these groups, and if they are, what that would mean for those tasked with designing conservation campaigns and communications to reduce wildlife crime. More participants from India agreed that the recommended action – donating money to conservation organizations – would help in reducing wildlife crime than U.S. participants, who did not seem to agree quite as strongly as participants from India did. Most of the participants who expressed strong intentions to donate money to conservation organizations to help reduce wildlife crime were from India. Responses to the other two items measuring donation intentions also aligned with this finding. For instance, of the number of participants who strongly agreed that they plan to donate money to conservation organizations within the next month to help reduce wildlife crime, n = 206 were from India while n = 56 were from the U.S, and of the number of participants who agreed with that statement, n = 299 were from India while n = 78 were from the U.S. Most participants from the U.S., in fact, either disagreed or strongly disagreed with the statement that they would definitely donate money to conservation organizations within the next month to help reduce wildlife crime. The implications of these findings are discussed in the next chapter. From the value-belief-norm model, despite biospheric values not significantly predicting either of the two NEP dimensions in the India sample, they did significantly predict mutualistic wildlife value orientations, as expected, and were not one of the VBN constructs that had issues with convergent validity. While U.S. participants did also largely 99 agree with biospheric value items, their responses were distributed between ‘moderately important’, ‘very important’, and ‘extremely important’, whereas most Indian participants rated all biospheric value items as either ‘very’ or ‘extremely important’). While biospheric values fail to predict NEP dimensions among Indian participants, altruistic values played a significant role, and they also significantly and positively predict mutualistic wildlife value orientations among Indian respondents. Notably however, altruistic values failed to predict both dimensions of NEP – environmental concern and anthropocentricism – and mutualistic wildlife value orientations among U.S. respondents. It is immediately clear, however, that there are far more participants from India for whom having control over others’ actions is important whereas for most respondents from the U.S., this is not that important. Further, across both groups combined, most participants who said it was either ‘extremely important’ or ‘very important’ to them to have control over others’ actions, and to have authority over others were between 18 and 34 years old. Nep10: The so called 'ecological crisis' facing humankind has been greatly exaggerated Strongly agree 141 75 Agree 222 66 Somewhat agree 186 108 Neither 135 169 Somewhat disagree 64 111 Disagree 99 175 Strongly disagree 73 196 0 50 100 150 200 250 India USA Figure 21 - Cross-national distribution of responses to Nep10 100 Most respondents from both countries largely agreed with the statements measuring the environmental concern dimension of the NEP scale. Among items measuring the second NEP dimension of anthropocentricism, the items on which both groups did not consistently agree or disagree had to do with the belief that the current ecological crisis is being greatly exaggerated, and the belief that humans will eventually learn enough about nature to be able to control it. More Indian respondents think that over time, people will be able to control nature than U.S respondents, who largely disagreed with this belief statement. Nep14: Humans will eventually learn enough about how nature works to be able to control it Strongly agree 173 53 Agree 273 77 Somewhat agree 205 138 Neither 115 185 Somewhat disagree 64 136 Disagree 60 174 Strongly disagree 30 137 0 50 100 150 200 250 300 India USA Figure 22 - Cross-national distribution of responses to Nep14 For nearly all the items measuring mutualistic wildlife value orientations, more participants from India either agreed or strongly agreed with statements than U.S. respondents did. While most participants across both groups seem to agree that wildlife crimes have consequences such as biodiversity loss, depletion of natural resources, and species extinction, a substantially higher number of participants from India strongly agreed with that statement than U.S. participants did. When it comes to the scale of ecological 101 impacts that wildlife crimes have, participants from both groups agreed that these impacts spillover into surrounding communities and areas, but not as many participants agreed strongly with this statement even though among those who did, there were higher numbers of participants from India than the U.S. Moreover, a somewhat similar number of U.S. respondents ‘strongly agreed’, ‘agreed’, ‘somewhat agreed’, and ‘neither agreed nor disagreed’ with the statement that wildlife crimes such as the illegal trade in exotic species can cause the spread of deadly viruses and pathogens from animals to humans. This was not the case for Indian respondents who mostly either agreed or strongly agreed with that statement. When indicating whether they thought that conservation organizations help to actually combat the issue of wildlife crime and reduce its impacts, agreement did not seem as strong even though most participants across both groups did agree or somewhat agree with the statement. Those who strongly agreed that every citizen is responsible for mitigating the impacts of wildlife crime were mostly Indian participants. For U.S. participants, the most common responses were either ‘somewhat agree’ or ‘neither agree nor disagree’ on all three ascription items. While Indian participants rated ‘every citizen’ and ‘the authorities’ as somewhat equal in who they thought should take responsibility, U.S. respondents primarily agreed that the authorities should be responsible, followed by the public, and then the self. While most U.S. and Indian participants largely agreed that they ought to donate money to conservation organizations to help reduce wildlife crime, most U.S. respondents indicated that they would not feel guilty if they did not donate money to conservation organizations whereas most Indian respondents largely agreed that they would feel guilty if they didn’t donate money to conservation organizations for wildlife crime reduction. 102 Most U.S. respondents disagreed that they felt morally obligated to donate money to conservation organizations for wildlife crime reduction, but most Indian respondents largely agreed with this statement. Although most U.S. participants did not feel morally obligated to donate money for wildlife crime reduction, as seen in Figure 45, most of them, in fact, indicated that they felt donating money to conservation organizations for this cause was the ‘right thing to do’. Another item whose responses diverge from U.S. participants’ responses to PN3 – which assesses whether they would feel guilty by not donating money – is the final item in the scale measuring personal norms in which most participants from both countries agreed that they would feel good about themselves if they donated money to conservation organizations to help reduce wildlife crime. The number of respondents from India who strongly agreed with this statement was more than twice the number of U.S. respondents who expressed strong agreement with the statement. Visual representations of cross-national responses to these individual items from both the planned behavior and value-belief-norm models are provided in Appendix J. Demographic Factors Most of the respondents across both groups who expressed stronger intentions to donate money to conservation organizations were between 18 and 34 years old. Those who identified as female slightly outnumbered those who identified as male in their willingness to donate money, however among those who said they would definitely be donating money within the next month, male participants slightly outnumbered female participants. Further, most respondents who indicated a high willingness to donate money reported earning between $20,000 and $40,000 per year in the U.S. and between INR 2,50,000 and INR 5,00,000 in India. Also, most of the respondents across both groups who expressed 103 higher willingness to donate money reported having completed at least a bachelor's degree, with the next most frequent category being those who had completed at least a master's degree. Moreover, respondents who indicated they were employed full-time or self-employed were the ones who expressed most willingness to donate money. Finally, most of the participants who were more willing to donate money reported their political ideologies as being neither conservative nor liberal, with the next most frequent category being those who considered themselves to be ‘very liberal’. 104 CHAPTER FIVE: DISCUSSION Research Questions and Objectives This dissertation is set against the backdrop of a noticeable lack of three things that, due to their absence, prevent the development of a holistic understanding of a wicked problem such as wildlife crime. First, there is remarkably little communication-based research on wildlife crime. Second, there is a dearth of representation of non-Western communities, especially those in developing economies, in environmental communication and conservation social science research. Third, interdisciplinary research on wildlife crime is growing but still occurs primarily outside the domain of environmental communication and environmental psychology, which explains the lack of evidence of salient social-psychological factors that may drive or impede the adoption of conservation behaviors related to curbing wildlife crime, especially factors that may influence the success of conservation communication campaigns. Against this backdrop, my broad objectives in conducting this research were to: 1) Determine the theoretical structure of behaviors linked to reducing wildlife crime – such as donating money to conservation organizations working to curb it; and 2) Determine whether the same factors predicted these behaviors in two different national samples. To meet these objectives, I asked two overarching research questions: • RQ 1: Which social-psychological factors best explain donation intentions to help reduce wildlife crime? • RQ 2: How do the responses of participants from India compare to those of participants from the U.S.A.? 105 To answer both these questions, I tested two major theories of environmental behavior and conducted confirmatory factor analyses and structural equation modeling to determine how well existing scales measure the latent constructs in these theoretical models, and to determine the relationships between those latent constructs, thereby identifying factors from the two theories that most strongly predict donation intentions to help reduce wildlife crime. Secondly, I conducted this research in the form of a cross-national survey to demonstrate and compare how well each of the two theories performs in predicting these intentions. Key Findings The primary finding in this dissertation is that the theory of planned behavior performs better than the value-belief-norm theory at predicting intentions to donate money to conservation organizations to reduce wildlife crime. The theory of planned behavior explained greater variance in intentions to donate money to conservation organizations to help reduce wildlife crime in both U.S. (R2 = 66.2%) and Indian (R2 = 71.9%) samples, than either the original value-belief-norm theory model (USA R2 = 60.8%; India R2 = 56.2%), or the modified value-belief-norm model with wildlife value orientations did (USA R2 = 61.1%; India R2 = 56.5%). Despite its better performance, the manner in which the TPB’s exogenous variables predicted the outcome variable was surprising because, as the multivariate analysis revealed, attitudes toward the behavior did not significantly predict behavioral intention in either group. Further, had this study only presented correlations and linear regressions as findings, the relationship between attitudes and intentions would have been significant as they were both positively and strongly correlated in both samples. Therefore, this 106 dissertation also underscores the importance of path analysis approaches such as structural equation modeling to demonstrate precisely how each predictor variable behaves in a multivariate model. Most U.S. respondents largely disagreed that they planned to or would definitely donate money to conservation organizations within the next month whereas most Indian respondents largely agreed that they were willing to and would definitely donate money. The value-belief-norm models – both the original one with New Ecological Paradigm measures and the modified one with mutualistic wildlife value orientations – were a lot more complicated to identify and fit. They were also affected by issues of low scale reliability and lack of convergent validity, particularly when it came to the Indian sample. The structural models for both Indian and U.S. groups for the value-belief-norm model were different; this was not based on whether hypothesized relationships were found, but rather on the non-significance of regression coefficients in the multivariate SEM analyses. Across both the theories, there were four direct predictors of intention – attitude toward the behavior, subjective norms, perceived behavioral control, and personal norms. For Indian participants, perceived behavioral control mattered the most, followed by subjective norms, personal norms, and attitude – in that order. For U.S. participants, personal norms mattered the most when it came to predicting intentions to donate money to conservation organizations to help reduce wildlife crime, followed by perceived behavioral control, subjective norms, and then attitude – in that order. The most notable differences between group responses either because there were divergent responses or because one group expressed much stronger agreement than the other, were related to 107 subjective norms, egoistic values, ascription of responsibility, personal norms, and intention to donate money. Theoretical Implications In this section, I discuss issues related to conceptualization, operationalization, measurement, scale development, etc. relevant to both the theories tested in this dissertation. Implications of the findings discussed here are aimed primarily at other scholars who may be able to address some of these concerns in their future work or may be able to solve some of the problems I ran into. Practical implications for conservation and communication practitioners are discussed in a later section. Theory of Planned Behavior For the most part, given the simplicity of the TPB model, the analysis and results were straightforward. Given that all latent factors in the TPB model met the criteria for convergent and discriminant validity, and had high scale reliability, the absence of a significant relationship between attitude toward the behavior and intention to donate was surprising but looking at the significant positive correlations between the three exogenous variables in this model (attitude, subjective norms, and perceived behavioral control), it is plausible that the influence of attitude on intention could have been dissipated through its covariance with the two other exogenous latent factors, leaving no significant direct association with intention. Other researchers have tried to explain the gap between attitudes and behavior by suggesting that if the social normative, familiar, or cultural environmental of an individual espouses views that are antithetical to the behavior being measured, attitudes are unlikely to influence behavior strongly, or that assessing attitudes and behaviors in different time 108 periods can cause this gap, and attitudinal measures that are either too broad or too narrow based on the behavior being measured are also likely to impede the attitude— behavior relationship (Kollmuss & Agyeman, 2002). However, these issues would be unlikely to afflict the relationship between these two factors in this dissertation because behavior-specific attitudinal measures were used which showed high scale reliability as well as convergent validity. While the majority of research using the theory of planned behavior tends to demonstrate a significant relationship between attitudes and intentions, the impact of attitudes on behaviors has been known to be small as well (Kollmuss & Agyeman, 2002). In their low-cost/high-cost model explaining the gap between attitudes and environmental behaviors, Diekmann & Preisendörfer (1992) suggest that people are likely to engage in environmental behaviors that have the least cost – whether financial, psychological, or in the form of time and effort needed to perform the behavior. Given that, in this study, the outcome behavior being assessed would have required respondents to bear financial, cognitive, and resource-related costs, it is plausible – based also on the large number of U.S. respondents who did not intent to donate money – that these associated costs of performing the behavior were unacceptable or out of line with respondents’ priorities. Additionally, this is not the only study in which attitudes did not relate significantly to intentions/behaviors (Valle et al., 2005). In their cross-national comparison of the theory of reasoned action, the norm activation model, and the VBN, Cordano et al. (2011) found that attitudes did not significantly predict intentions (p. 645) for Chilean respondents, and this relationship was just barely significant for U.S. respondents. In explaining this finding, Cordano et al. cite Ajzen (1985, 1991), who suggested that whether or not a particular TPB predictor is likely to play a significant role varies with the kind of 109 behavior being studied. Therefore, although it is not possible to say from cross-sectional survey data what may have caused this unexpected finding, the fact that the attitude— intention relationship was non-significant in both the U.S. and the Indian context merits further investigation, particularly in the context of other wildlife-related conservation behaviors. The measurement and structural equation models for both countries were the same for the TPB when it came to the indicators and associations between latent factors, however regression coefficients and factor loadings were different across both groups. This suggests that while it is likely that the items being used to measure the latent factors are doing a good enough job of measuring those latent variables in both countries, some of the items may be contributing more or less to the latent factor, and that the strength of association between the indictor items and their latent factors, and between the latent factors themselves may be different in different samples of respondents. That said, the TPB holds up well even outside the predominantly Western and developed nations context in which it is largely applied. The findings from the SEM for the TPB help to cross-validate and, in fact, demonstrate, that not only is this theory able to predict this unique type of conservation behavior (related to wildlife crime reduction), but it is also able to retain its measurement and theoretical structure cross-nationally. In answering RQ1 then, I found that from the theory of planned behavior, perceived behavioral control and subjective norms were the salient factors that predicted intentions to donate money to conservation organizations to help reduce wildlife crime, and while attitude did not play a salient role, their significant positive covariance with these two factors suggests that may matter but not directly, as initially expected. 110 In addressing RQ2, I was not surprised that subjective norms measures seemed to have stronger agreement from Indian respondents than from U.S. respondents given that social norms play a significant role in the collective type of society that exists in India (R. W. Liu et al., 2022; Shalender & Sharma, 2021; Taufique & Vaithianathan, 2018). Moreover, the stronger agreement with self- and response-efficacy items suggests that more Indian participants were confident in their ability to donate money to conservation organizations, and had more availability of sufficient resources, time, and opportunities than U.S. participants. This self-confidence and availability of resources, time, and opportunities seemed to play an important role in Indian participants’ strong intentions to donate money, along with the belief that making these donations would actually help conservation organizations reduce wildlife crime. On the other hand, most U.S. respondents said they did not have sufficient resources to donate money and showed low response efficacy compared to Indian respondents. Practical implications of these cross-national findings and differences are discussed in a later section. Overall, the primary finding that the TPB performs much better than the VBN at predicting intentions to engage in a wildlife crime-related conservation behavior cross- nationally, raises important questions about the type of behavior this might be. One may suggest that the act of donating money, being charitable, lets us describe donating money to conservation organizations to help reduce wildlife crime as a pro-social or values- beliefs-or-norm-driven behavior, and while a part of that argument may well be valid given the role of personal norms in this dissertation, it is important to note that the strongest predictors were TPB constructs. Considering that the theory of planned behavior is rooted in the premise of self-interest or rational-thought-based-behaviors, the findings from this 111 dissertation would suggest that donating money to conservation organizations to help reduce wildlife crime is a rational and planned behavior, adopted more out of self-interest than out of a pro-social motive that underlies the value-belief-norm model. More specifically, the difference between how much variance each of these two theories explain in the donation intentions outcome is larger for India (15%) and smaller for the United States (5%) which may indicate that this behavior could be characterized as more of a planned, rational act for Indian respondents but that it may be more of a mixed bag for U.S. respondents as personal norms play a larger role in the U.S. Original Value-Belief-Norm Model A handful of interesting findings regarding the value-belief-norm model emerged in this dissertation, especially when it comes to cross-national comparisons. First, although the SEM for this model did produce an acceptable fit for the Indian sample, the consistent lack of scale reliability and convergent validity makes it difficult to generalize the findings from the Indian sample. While the theory held up for U.S. respondents, showing clear relationships between values, beliefs, personal norms, and intention, I would say the theory – in its simplest form – did not hold up Indian respondents as well. For instance, biospheric values – which did demonstrate convergent validity for both samples – had no significant influence on ecological worldviews dimensions for Indian respondents, but this relationship was significant for U.S. respondents. Similarly, although altruistic values were the strongest predictors of ecological worldviews in the Indian sample, their measures did not meet the requirements for establishing convergent validity. With respect to egoistic values, two of the five items had to be removed from the analysis for both groups, and a third had to be removed for the Indian sample, leaving a two-item latent factor which is not 112 ideal in CFAs. Interestingly, however, egoistic values were among those variables on which cross-national responses differed substantially. For instance, having control over others’ action, having authority over others, and being influential were overwhelmingly rated as “extremely important” by Indian participants who far outweighed U.S. respondents who agreed, whereas most U.S. participants largely disagreed with those values statements. Tables 10 and 11 show that scale reliability is consistently above the 0.7 threshold for Cronbach’s alpha in the U.S. sample whereas several scales have low internal consistency in the Indian group. The inconsistency and lack of reliability of measures raises important questions about where and how these measures were developed, and further underscores the need for cross-validating scales that are so widely used in environmental communication and environmental psychology research. This brings up another two-fold concern – most studies that apply or test this theory, like many other social-psychological theories of environmental behavior and communication, are conducted in Western, industrialized, rich countries; and these are also the countries where scholars develop and validate their scales. In most cases when these theories are applied in non-Western countries or emerging economies, scholars do not resort to confirmatory factor analyses or path analyses etc. that could validate both the measurement and structural models that have been accepted in the Western countries where these scales and theories were developed and tested (R. W. Liu et al., 2022). Within the original VBN model, while there were items measuring altruistic and egoistic values that did have poor factor loadings and were eventually removed from one or both of the countries’ SEM analyses, the most challenging hurdle to navigate was presented by the New Ecological Paradigm scale which, despite being criticized for its 113 reliability and validity as a measure of general environmental beliefs, has continued to be applied in studies using the VBN theory as a unidimensional construct or in various permutations and combinations of the complete 15-item NEP scale (Carfora et al., 2021; Dunlap, 2008; Dunlap et al., 2000; Gkargkavouzi et al., 2019; Han, 2015; Ntanos et al., 2019; van Riper & Kyle, 2014b). Following one of the rare instances in which scholars assessed the dimensionality of the NEP scale using confirmatory factor analysis, a primary takeaway from the results of this dissertation research has to do with the inherent multi-dimensionality of the NEP scale (Amburgey & Thoman, 2012). Ecological worldviews, as operationalized by the NEP scale, emerged as having three dimensions – environmental concern or ecocentrism, anthropocentricism, and limits to growth (Ntanos et al., 2019). Despite demonstrating this multi-dimensionality, scholars did not appear to modify the hypothesis associated with ecological worldviews in the VBN model. For instance, if the NEP is multi-dimensional with each dimension reflecting a different view of human-nature relationships (e.g., human domination vs. anti-exemptionalism), why would each of those dimensions continue to relate positively to awareness of consequences in the VBN chain of effects, or why would biospheric and egoistic values have the same valence of effect on the different NEP dimensions? In my review of the literature, I found that these questions are not accounted for in the way we use these scales. For example, anthropocentricism among Indian respondents is negatively associated with a positive attitude toward the behavior, which would logically be expected. However, this relationship was positive for US respondents, even if minutely so. Additionally, if anthropocentricism is a dimension of NEP, then would we not expect biospheric values to be negatively related to anthropocentricism, as was the 114 case in the Indian sample in this dissertation? Moreover, building a 15-item scale into a questionnaire only to discard a third of those items from subsequent factor analyses is an inefficient approach to measuring and including ecological worldviews in this theoretical model. Essentially, based on what I found, the challenges I faced, what I have read and learned about cross-validating measurement and structural models, I believe it is entirely possible to develop a more helpful measure of ecological worldviews – even if it is the multi-dimensional NEP scale – to better predict and explain environmental behaviors without continuing with business-as-usual wherein it is assumed that all values (biospheric, altruistic, egoistic) will be positively or negatively associated with a singular unidimensional concept of NEP. Comparing the Original and Modified VBN Models According to the SEM for the original VBN model, altruistic and egoistic values predicting the two NEP dimensions I found and included in the model, but biospheric values had no significant relationship with either NEP dimension. Moreover, having dropped one indicator of altruistic values and three indicators of egoistic values from the Indian sample, and having eliminated two egoistic values indicators from the U.S. sample, I expected to end up with two different measurement models for the two groups. When comparing the original NEP-based value-belief-norm model to the modified version in which mutualistic wildlife value orientations replaced NEP as a measure of ecological worldviews, I found that while the variance explained by the WVO-based model was not that different from the NEP-based model, the structural models were, in fact, different. For instance, biospheric values played a significant role in predicting mutualistic wildlife value orientations which, themselves, significantly and positively predicted 115 awareness of consequences, and so on, until personal norms positively and significantly predicted intention to donate money to conservation organizations. The finding that the WVO-based VBN model was not only a well-fitting model but that it had slightly better model fit goes to show that there is, indeed, potential to replace the New Ecological Paradigm scale as a measure of ecological worldviews and, instead, use a behavior-specific measure of value orientations, and while there has been cross-validation of the wildlife value orientations measures, these findings underscore the importance of cross-validating these measures and models in non-Western countries and emerging economies as well (De Groot & Steg, 2007; Liordos et al., 2021; Teel et al., 2007). Despite the roadblocks with reliability, validity, and cross-validation of measures in the VBN models – both NEP-based and WVO-based – the theory does predict a type of conservation behavior linked to reducing wildlife crime, and it does so in two different countries. There is certainly scope for improving the VBN’s ability to predict and explain these behaviors cross-nationally in the future by being more selective about measures, and by consistently confirming and revising measurement and structural models with findings from non-Western countries and emerging economies where this theory is tested (Sharma & Gupta, 2020). Practical Implications for Conservation Communication Based on the results that answered both the research questions, in this section, I will translate some of the findings and theoretical implications into practical recommendations for conservation communication designers or professionals and practitioners in conservation tasked with public communication and campaign message design. 116 As I point out above, donating money to conservation organizations for reducing wildlife crime may sound like and even be framed as a charitable, pro-social behavior in conservation campaigns and social marketing materials, but the predominance of the TPB’s constructs in predicting intentions to donate money for this cause should prompt conservation practitioners in strategic communication to pause and consider whether their messages to diverse, cross-national audiences are appealing to or evoking the right factors. For example, such campaigns in the U.S. may tend to focus more heavily on personal norms and that may be warranted, as the results show, but outside the U.S. and even within it, organizations risk alienating audience members who are not drive by personal norms but would be willing to donate money for other reasons such as having the time, money, confidence, and opportunities to donate, or believing that doing so is likely to really help conservation organizations combat wildlife crime. Conservation professionals tasked with designing campaign communications for Indian audiences would do well to include normative messaging that highlights social normative engagement with the behavior they are calling for in campaigns. For example, including a message in a campaign against wildlife crime that tells audience members that those who they look up to or those who matter most to them would want and expect them to perform the target behavior may enhance the effectiveness of that message and influence the success of the campaign. This social normative aspect of such a message may matter more to audiences in India than in the U.S. but it would still be important for U.S. audiences regardless. Conservation campaign messages seeking behavioral engagement in India and the United States should also build efficacy among potential donors such as by suggesting that 117 requested donations are of a value that is acceptable to those audience segments, or by having an adjustable sliding scale so that donors can choose the amount they feel most comfortable donating. Relatedly, given the low response efficacy among U.S. participants in this dissertation, I would recommend that conservation organizations be specific about targeting audience segments that are not resource-deficient, and that these organizations should explicitly communicate how they would use the donations they receive to take tangible action or make changes or improvements that will actually help reduce wildlife crime eventually. Showing donors that their time, resource, and action actually builds towards something can be crucial for the success of a campaign to prevent wildlife crime. Further, by making the process of donating money as efficient and easy as possible to use minimal time, conservation organization can further enhance self-efficacy among potential donors. Additionally, it is important to consider that this study was conducted during the COVID-19 pandemic. In the early days of this pandemic, news stories were published that attributed the origins of the pandemic to illegal wildlife trade, specifically, the trafficking of wild animals such as pangolins or bats to wet markets in Wuhan, China, where the virus responsible for the COVID-19 disease was traced back to by the WHO. The international coverage and unprecedented attention that illegal wildlife trade was getting in this context, allowed some members of the public to connect the dots between wildlife trade and the spread of deadly pathogens and zoonotic diseases. It is not possible to gauge the extent to which participants from both countries were aware of or had been exposed to news coverage linking the COVID-19 pandemic to international wildlife trade, however, based on their responses to the item featured in Figure 40, a relatively similar number of U.S. 118 respondents ‘strongly agreed’, ‘agreed’, ‘somewhat agreed’, and ‘neither agreed nor disagreed’ with the statement that wildlife crimes such as the illegal trade in exotic species can cause the spread of deadly viruses and pathogens from animals to humans. This was not the case for Indian respondents who mostly either agreed or strongly agreed with that statement. I mention this not because I am able to report a measured impact of information about the COVID-19 pandemic on participants’ responses in these surveys, but more so because the extent to which audiences are aware of the range of impacts an issue has and what problems it causes is likely to contribute to the success or failure of a conservation campaign message about that issue. For instance, although most participants from both groups largely agreed that wildlife crimes cause biodiversity loss, species extinction, depletion of natural resources, and have far-reaching environmental consequences on surrounding areas and communities, fewer Indian and U.S. American respondents agreed that conservation organizations help curb wildlife crime and mitigate its impacts. If audiences do not believe that conservation organizations are effectively addressing the issue and impacts of wildlife crime, I would argue that is likely to impede the efficacy of any campaign communication from these organizations seeking donations from such audience members. To pre-emptively avoid a loss of donor engagement, conservation organizations should, whenever possible, include evidence of the impact their work has had on combatting wildlife crime, or provide evidence of how donations from contributing individuals have been used to mitigate some of the impacts of wildlife crime. This is not to say that such impact evaluations may not already exist, but they are hardly made public. If conservation organizations do not transparently disclose the impact of their work to reduce wildlife crime, for instance, audiences may not consider the issue or that 119 organization’s work to manage it as important. Organizations must, therefore, ask themselves – Would their prospective supporters donate money to an organization whose work, impact, effectiveness, or success they are not aware of? Despite the myriad issues associated with the NEP scale, looking at item-level responses provided some valuable insight that could have practical relevance. One of the NEP items states that the ongoing ecological crisis has been greatly exaggerated (see Table 10 for specifics). Respondents from India overwhelmingly agreed with that statement while respondents from the U.S. largely disagreed with it. This was surprising because most participants from both countries had largely agreed on most of the biospheric values and mutualistic wildlife value orientations measures. Considering this particular NEP item is similar in the sense that it is seeking to reflect an eco-centric belief for those who disagree with the item, I would have expected to see disagreement from both groups. However, based on those findings, it then becomes important to acknowledge that if audiences believe that the current ecological crisis has been exaggerated, conservation organizations communicating about wildlife crime and trying to gain support through donations may need to focus their messaging on other salient factors that predict intention to donate within this community – such as by centering their messages on building self- and response efficacy, by invoking social norms, appealing to biospheric and altruistic values, and by evoking mutualistic wildlife value orientations. It is important to also note that guilt-shaming people into donating to such causes may not work for all audiences. For instance, it was mostly Indian respondents who agreed that they would feel guilty if they did not donate money to conservation organizations to help reduce wildlife crime, and that they feel morally obligated to donate money for this cause. 120 U.S. respondents largely disagreed that they would feel guilty if they did not donate money, and also disagreed that they feel morally obligated to do so. This was despite overall agreement from both groups that they felt they ought to donate money for this purpose, and that doing so would be the right thing to do, and it would make them feel good about themselves. In most cases, strong agreement largely came from Indian respondents whereas U.S. respondents tended to opt for a neutral response more often. When appealing to personal norms in their campaigns or messaging, therefore, conservation organizations might benefit from highlighting the ‘feel good’ aspect of charitable donation behaviors, but should not universally aim to make people feel guilty or obligated to donate money unless they have a well-rounded understanding of their target audience’s beliefs. It might also be helpful for conservation organizations interested in running campaigns on wildlife crime prevention to get an idea of the socio-demographic details of their target audiences. In this dissertation, for example, those who expressed strong intentions to donate money to conservation organizations were mostly between the ages of 18 and 34; earned between U.S. $20,000 - $40,000 or INR 2,50,000 – INR 5,00,000 per year; had completed at least a bachelor’s degree, were either employed full-time or self-employed, and indicated a political ideology that was neither conservative nor liberal. It is also worth noting that although most Indian respondents made between INR 2,50,000 and INR 5,00,000 per year – which, based on current exchange rates, is the equivalent of U.S. $3,140 and $6,282 per year – the number of Indian respondents who strongly agreed with having sufficient resources to be able to donate money to conservation organizations for wildlife crime reduction was nearly three times the number of U.S. respondents who strongly agreed with the same statement (see Figure 23). For 121 context, most U.S. participants earned less than $20,000 per year. However, since I did not collect information on participants’ place of work or residence, it is not possible to estimate the average cost of living for either group. This is not to say that those who reported earning higher incomes did not intend to donate money, but rather to say that those who seemed most willing to donate money did not appear to have the most disposable income which suggests that income may not always be a barrier to engaging in donation behavior. Key Contributions The major contributions of this dissertation are two-fold: 1) The application and rigorous testing of two major theories of environmental behavior in an entirely different context of wildlife crime-related conservation behaviors, thereby identifying the factors and theoretical models that are salient to predicting intentions to donate money to conservation organizations to reduce wildlife crime, and 2) The cross-national test of these theories in the context of wildlife crime prevention, which leads to the identification and confirmation of reliable measurement models, and the identification of the theoretical structure of these behaviors as predicted by the two theories in two different countries, thereby building new empirical knowledge, advancing theory, and underscoring the value of having more diverse and inclusive environmental social science research. Not only does the cross-national comparison advance empirical and theoretical research in environmental communication (Takahashi, Duan, et al., 2021; Takahashi, Metag, et al., 2021), but it provides the first-known empirical data on social-psychological determinants of conservation behaviors related to wildlife crime from India and, to my knowledge, the United States as well. This dissertation demonstrates how well each of the two theories – the theory of planned behavior (Ajzen, 1991), and the value-belief-norm 122 theory (Stern et al., 1999) – performed across two national samples to predict the same outcome. Moreover, this dissertation also demonstrates that mutualistic wildlife value orientations (WVOs) can replace the New Ecological Paradigm scale (NEP) as a measure of ecological worldviews without affecting the model fit, specifically in the context of wildlife- related behaviors as predicted or explained by the value-belief-norm (VBN) model (Manfredo, Teel, Don Carlos, et al., 2020). With both theoretical and empirical contributions, this dissertation expands the purview of environmental communication and environmental psychology research to include wildlife crime-related behaviors and aspects and underscores the need for scholars to move beyond predominantly Western, industrialized, established economies and be more inclusive in the communities and regions they study. This research also emphasizes the importance of developing and maintaining long- term cross-functional partnerships between academic researchers and conservation practitioners so that such mutually beneficial relationships can lead to effective management strategies to combat wildlife crime – a wicked problem that requires such integrated approaches to problem-solving. Sharing my findings with conservation organizations and practitioners as an accessible toolkit will be my way of taking a step towards building such a collaboration. In the future, I hope that a two-way relationship can develop through which social scientists can help practitioners better understand their audiences and better strategize their communication, while practitioners can provide valuable real-world data to help advance theory, science, and conservation goals. 123 Future Research While identifying salient drivers of intentions to donate money to reduce wildlife crime is certainly a necessary first step towards building a theory of wildlife conservation behaviors aimed at combatting wildlife crime, it is also important to keep in mind that this dissertation only measures predictors of behavioral intention and not behavior itself. As prior research has shown, just as there may be a set of factors that facilitate the adoption of such behaviors, there are also barriers to these behaviors (Kollmuss & Agyeman, 2002). As future research builds on and further develops our understanding of the factors that drive wildlife crime-related conservation behaviors, it would also be helpful, not just for theorizing, but also for making practical recommendations to conservation professionals, to know which factors may act as barriers to engaging in such conservation behaviors, and which communities or populations are especially likely to encounter those barriers. Additionally, future research could expand on this dissertation by applying and testing these theories and models in other countries as well, especially in parts of the world that tend to get overlooked in environmental communication and environmental psychology research. Not only is this likely to produce interesting empirical data from understudied regions, but it will go a long way in improving the reliability and validity of some of the most widely used measures and scales in environmental social science research. On the other hand, it could lead to the development of new scales that are more specific to wildlife-related behaviors. For example, this dissertation uses measures and scales adapted from other research – albeit research on environmental behaviors – but not from work that has focused specifically on and used validated measures to assess wildlife 124 crime-related conservation behaviors. Therefore, future research will likely have to dig its heels into scale development and cross-validation. Using the cross-sectional survey as a steppingstone, future research could design and conduct experiments to test causal chain of effects and determine whether the measurement and structural models found to fit the data in this dissertation would also be reproduced in an experiment. An experiment would be able to assess not only the paths through which direct and indirect causal effects occur but would also be able to test an actual conservation communication campaign message on wildlife crime prevention for its cognitive, affective, and behavioral effects, and could share the results with conservation practitioners or organizations to help them optimize their strategic communication and message design. A path analysis from experimental data could demonstrate direct, indirect, mediation, and moderation effects which could provide much deeper insight to both scholars and conservation professionals tasked with public communication and strategizing. Those interested in the theoretical or methodological implications of the findings in this dissertation could, for instance, investigate whether the ability of attitude to predict intention was being dissipated through its covariances with subjective norms and perceived behavioral control or whether attitudes, even though they’re known to sometimes have small effects, are necessary to include in the model. That said, it would be interesting to develop an integrated model along the lines of what others have tested before to demonstrate whether factors that are strong and known predictors of wildlife crime-related conservation behaviors can be integrated into a single model that retains parsimony but has great predictive potential (Bamberg & Möser, 2007; Carfora et al., 2021; 125 Gkargkavouzi et al., 2019; Klöckner, 2013; Montaño & Kasprzyk, 2008). Personally, I would like to expand on this research by examining other kinds of wildlife-related conservation behaviors, or specifically, other behaviors linked to reducing wildlife crime such as information-seeking behaviors, or signing petitions, publicly showcasing support for the cause etc. Such assessments would add to the recommendations and findings from this dissertation and would make for a more comprehensive guide for conservation organizations who wish to understand their audiences better and who seek to optimize their communication strategies and messages for different audiences. Finally, and this may be far-fetched or idealistic, but my hope is that as conservation social science is embracing its interdisciplinarity and as social scientists and practitioners are now working together more frequently than they did before, there may be room for academic researchers and conservation organizations to collaborate on such projects. This dissertation already demonstrates to conservation organizations working in India and the U.S.A. how a sample of their audiences currently think and feel about wildlife crime and its reduction. I believe the findings from this dissertation will aid these organizations in identifying factors they can leverage in their approach to designing wildlife crime prevention communication materials and campaigns in these communities. Limitations One of the limitations of this dissertation is inherent to the method – surveys rely on self-reported data which can be prone to social desirability and other kinds of biases that prompt respondents to select options that they think the researcher would want them to select or to avoid seeming ‘undesirable’. Questions about pro-environmental values, beliefs, attitudes, etc. may evoke such biases though so I paid special attention to my question 126 wording in designing the survey. Furthermore, the use of Qualtrics panels have recently sparked concerns about their vulnerability to automated bots that produce unreliable data. However, to avoid this problem, I used two attention checks and a speeding check to ensure that I only got complete responses. Even so, within the complete responses returned by Qualtrics, I noticed frequent straight-lining behaviors, especially among U.S. respondents, and a large number of U.S. respondents selecting the ‘neither agree nor disagree’ option while answering questions. Relatedly, if I was to do this all over again, I would consider using item-specific response options rather than the standard Likert-type 7-point or 5-point agree-disagree scales (Saris et al., 2010). Using item-specific response options may improve not only reliability and validity, but also the quality of data interpretations. Additionally, surveys can be subject to measurement error, but they have higher external validity and generalizability. Considering that no previously validated scales specifically to measure wildlife crime-related behaviors existed, this dissertation draws from prior research on environmental behavior, natural resource management, environmental tourism, and other studies that have examined determinants of willingness to pay or willingness to donate, and studies that have compared the theory of planned behavior to the value-belief-norm theory. While the pilot test and pre-test interviews helped identify potential issues and make revisions to the survey instrument, the low reliability and convergent validity for VBN measures for the Indian sample could be an indication of measurement error. This survey was only accessible to participants from both countries who were fluent in English. In the future, I hope to work with collaborators to develop regional language 127 translations of such questionnaires as well, for a more in-depth look at these relationships. By nature of the method and data, findings from this dissertation are restricted to making correlational statements about relationships between constructs which, although not as powerful as establishing causality, is an important first step towards building theory to explain and predict wildlife crime-related conservation behaviors. Finally, it is important to keep in mind that this dissertation only measures predictors of behavioral intentions, and not behavior itself. Therefore, I cannot extrapolate from these findings and endeavor to explain behaviors without knowing what might impede the transition from intention to behavior in the context of wildlife crime reduction. 128 REFERENCES Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In Action control (pp. 11–39). Springer. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179–211. Ajzen, I. (2006). Constructing a TPB Questionnaire: Conceptual and Methodological Considerations. http://www.people.umass.edu/aizen/pdf/tpb.measurement.pdf Amburgey, J. W., & Thoman, D. B. (2012). Dimensionality of the new ecological paradigm issues of factor structure and measurement. Environment and Behavior, 235–256. Anagnostou, M., & Doberstein, B. (2022). Illegal wildlife trade and other organised crime: A scoping review. Ambio, 51(7), 1615–1631. https://doi.org/10.1007/s13280-021- 01675-y Aral, Ö. H., & López-Sintas, J. (2021). Is pro-environmentalism a privilege? Country development factors as moderators of socio-psychological drivers of pro- environmental behavior. Environmental Sociology, 0(0), 1–17. https://doi.org/10.1080/23251042.2021.2018123 Ateş, H. (2020). Merging Theory of Planned Behavior and Value Identity Personal norm model to explain pro-environmental behaviors. Sustainable Production and Consumption, 24, 169–180. https://doi.org/10.1016/j.spc.2020.07.006 Ateş, H. (2021). Understanding Students’ and Science Educators’ Eco-Labeled Food Purchase Behaviors: Extension of Theory of Planned Behavior with Self-Identity, Personal Norm, Willingness to Pay, and Eco-Label Knowledge. Ecology of Food and Nutrition, 60(4), 454–472. https://doi.org/10.1080/03670244.2020.1865339 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. Bennett, N. J., Roth, R., Klain, S. C., Chan, K., Christie, P., Clark, D. A., Cullman, G., Curran, D., Durbin, T. J., Epstein, G., Greenberg, A., Nelson, M. P., Sandlos, J., Stedman, R., Teel, T. L., Thomas, R., Veríssimo, D., & Wyborn, C. (2017). Conservation social science: Understanding and integrating human dimensions to improve conservation. Biological Conservation, 205, 93–108. https://doi.org/10.1016/j.biocon.2016.10.006. Bezerra-Santos, M. A., Mendoza-Roldan, J. A., Thompson, R. C. A., Dantas-Torres, F., & Otranto, D. (2021). Illegal Wildlife Trade: A Gateway to Zoonotic Infectious Diseases. Trends in Parasitology, 37(3), 181–184. https://doi.org/10.1016/j.pt.2020.12.005 129 Blair, M. E., Le, M. D., Sethi, G., Thach, H. M., Nguyen, V. T. H., Amato, G., Birchette, M., & Sterling, E. J. (2017). The Importance of an Interdisciplinary Research Approach to Inform Wildlife Trade Management in Southeast Asia. BioScience, 67(11), 995–1003. https://doi.org/10.1093/biosci/bix113 Boomsma, A., Hoyle, R., & Panter, A. (2012). The structural equation modeling research report. In Handbook of Structural Equation Modeling (pp. 341–358). Boratto, R., & Gibbs, C. (2019). Advancing interdisciplinary research on illegal wildlife trade using a conservation criminology framework. European Journal of Criminology, 1477370819887512. https://doi.org/10.1177/1477370819887512 Brashares, J. S., Abrahms, B., Fiorella, K. J., Golden, C. D., Hojnowski, C. E., Marsh, R. A., McCauley, D. J., Nuñez, T. A., Seto, K., & Withey, L. (2014). Wildlife decline and social conflict. Science, 345(6195), 376–378. https://doi.org/10.1126/science.1256734 Cárdenas, S. A., & Lew, D. K. (2016). Factors Influencing Willingness to Donate to Marine Endangered Species Recovery in the Galapagos National Park, Ecuador. Frontiers in Marine Science, 3, 60. https://doi.org/10.3389/fmars.2016.00060 Carfora, V., Cavallo, C., Catellani, P., Del Giudice, T., & Cicia, G. (2021). Why Do Consumers Intend to Purchase Natural Food? Integrating Theory of Planned Behavior, Value- Belief-Norm Theory, and Trust. Nutrients, 13(6), 1904. https://doi.org/10.3390/nu13061904 CenterStat. (2019, January 23). Can I estimate an SEM if the sample data are not normally distributed? CenterStat. https://centerstat.org/can-i-estimate-an-sem-if-the- sample-data-are-not-normally-distributed/ Cheung, G. W., & Wang, C. (2017). Current Approaches for Assessing Convergent and Discriminant Validity with SEM: Issues and Solutions. Academy of Management Proceedings, 2017(1), 12706. https://doi.org/10.5465/AMBPP.2017.12706abstract Clayton, S., Czellar, S., Nartova-Bochaver, S., Skibins, J. C., Salazar, G., Tseng, Y.-C., Irkhin, B., & Monge-Rodriguez, F. S. (2021). Cross-Cultural Validation of A Revised Environmental Identity Scale. Sustainability, 13(4), 2387. https://doi.org/10.3390/su13042387 Collins, D. (2003). Pretesting survey instruments: An overview of cognitive methods. Quality of Life Research, 12(3), 229–238. Cordano, M., Welcomer, S., Scherer, R. F., Pradenas, L., & Parada, V. (2011). A Cross-Cultural Assessment of Three Theories of Pro-Environmental Behavior: A Comparison Between Business Students of Chile and the United States. Environment and Behavior, 43(5), 634–657. https://doi.org/10.1177/0013916510378528 130 Dalpane, F., & Baideldinova, M. (2022). Poaching and Wildlife Trafficking as Threats to International Peace and Security. In S. Sayapin, R. Atadjanov, U. Kadam, G. Kemp, N. Zambrana-Tévar, & N. Quénivet (Eds.), International Conflict and Security Law: A Research Handbook (pp. 861–883). T.M.C. Asser Press. https://doi.org/10.1007/978-94-6265-515-7_40 Dayer, A. A., Stinchfield, H. M., & Manfredo, M. J. (2007). Stories about Wildlife: Developing an Instrument for Identifying Wildlife Value Orientations Cross-Culturally. Human Dimensions of Wildlife, 12(5), 307–315. https://doi.org/10.1080/10871200701555410 De Groot, J. I. M., & Steg, L. (2007). Value Orientations and Environmental Beliefs in Five Countries: Validity of an Instrument to Measure Egoistic, Altruistic and Biospheric Value Orientations. Journal of Cross-Cultural Psychology, 38(3), 318–332. https://doi.org/10.1177/0022022107300278 De Groot, J. I. M., & Steg, L. (2008). Value orientations to explain beliefs related to environmental significant behavior: How to measure egoistic, altruistic, and biospheric value orientations. Environment and Behavior, 40(3), 330–354. Diekmann, A., & Preisendörfer, P. (1992). Persónliches umweltverhalten: Diskrepanzen zwischen Anspruch und Wirklichkeit. [Ecology in everyday life: Inconsistencies between environmental attitudes and behavior.]. Kölner Zeitschrift für Soziologie und Sozialpsychologie, 44(2), 226–251. Dietz, T., Fitzgerald, A., & Shwom, R. (2005). Environmental Values. Annual Review of Environment and Resources, 30(1), 335–372. https://doi.org/10.1146/annurev.energy.30.050504.144444 Duffy, R., & Brockington, D. (2022). Political ecology of security: Tackling the illegal wildlife trade. Journal of Political Ecology, 29, 1–22. https://doi.org/10.2458/jpe.2201 Dunlap, R. E. (2008). The New Environmental Paradigm Scale: From Marginality to Worldwide Use. The Journal of Environmental Education, 40(1), 3–18. https://doi.org/10.3200/JOEE.40.1.3-18 Dunlap, R. E., Van Liere, K. D., Mertig, A. G., & Jones, R. E. (2000). New trends in measuring environmental attitudes: Measuring endorsement of the new ecological paradigm: A revised NEP scale. Journal of Social Issues, 56(3), 425–442. Duthie, E., Veríssimo, D., Keane, A., & Knight, A. T. (2017). The effectiveness of celebrities in conservation marketing. PLoS One, 12(7), 0180027. https://doi.org/10.1371/journal.pone.0180027 Finney, S. J., & DiStefano, C. (2006). Non-normal and categorical data in structural equation modeling. In G. Hancock & R. Mueller (Eds.), Structural Equation Modeling: A Second Course (pp. 269–314). Information Age Publishing. 131 Fulton, D. C., Manfredo, M. J., & Lipscomb, J. (1996). Wildlife value orientations: A conceptual and measurement approach. Human Dimensions of Wildlife, 1(2), 24–47. https://doi.org/10.1080/10871209609359060 Giannetta, C. (2017). Increasing the effectiveness of conservation messaging by drawing connections with related political, social, and economic issues. Applied Environmental Education & Communication, 17(3), 1–15. https://doi.org/10.1080/1533015X.2017.1398689 Gkargkavouzi, A., Halkos, G., & Matsiori, S. (2019). Environmental behavior in a private- sphere context: Integrating theories of planned behavior and value belief norm, self- identity and habit. Resources, Conservation and Recycling, 148, 145–156. https://doi.org/10.1016/j.resconrec.2019.01.039 Glenn, I., Ferreira, S. M., & Pienaar, D. (2019). Communication on rhino poaching: Precautionary lessons about backfires and boomerangs. South African Journal of Science, 115(3–4), 1–4. https://doi.org/10.17159/sajs.2019/5585 Gómez-Melara, J., Acosta-Naranjo, R., Patricia Izar, Shahrul Anuar Mohd Sah, Pladevall, J., Maulany, R. I., Ngakan, P. O., Majolo, B., Romero, T. R., & Amici, F. (2021). A Cross- Cultural Comparison of the Link between Modernization, Anthropomorphism and Attitude to Wildlife. Sustainability. https://doi.org/10.3390/su132313095 Greenfield, S., & Veríssimo, D. (2019). To What Extent Is Social Marketing Used in Demand Reduction Campaigns for Illegal Wildlife Products? Insights From Elephant Ivory and Rhino Horn. Social Marketing Quarterly, 25(1), 40–54. https://doi.org/10.1177/1524500418813543 Haeger, H., Lambert, A. D., Kinzie, J., & Gieser, J. (2012). Using Cognitive Interviews to Improve Survey Instruments. 15. Han, H. (2015). Travelers’ pro-environmental behavior in a green lodging context: Converging value-belief-norm theory and the theory of planned behavior. Tourism Management, 47, 164–177. https://doi.org/10.1016/j.tourman.2014.09.014 Hoyle, R. H., & Isherwood, J. C. (2013). Reporting Results from Structural Equation Modeling Analyses in Archives of Scientific Psychology. Archives of Scientific Psychology, 1(1), 14–22. https://doi.org/10.1037/arc0000004 Jacobs, M. H., Vaske, J. J., Teel, T. L., & Manfredo, M. J. (2018a). Human Dimensions of Wildlife. In L. Steg & J. I. M. Groot (Eds.), Environmental psychology: An Introduction (Second, pp. 85–94). Wiley-Blackwell. Jacobs, M. H., Vaske, J. J., Teel, T. L., & Manfredo, M. J. (2018b). Human Dimensions of Wildlife. In Environmental Psychology (pp. 85–94). John Wiley & Sons, Ltd. https://doi.org/10.1002/9781119241072.ch9 132 Jansson, J., Marell, A., & Nordlund, A. (2010). Green consumer behavior: Determinants of curtailment and eco‐innovation adoption. Journal of Consumer Marketing, 27(4), 358–370. https://doi.org/10.1108/07363761011052396 Kachen, S., & Krishen, A. S. (2020). Exploring online comments from a strategic marketing stance to reduce wildlife crime. Psychology & Marketing, 37(12), 1771–1780. https://doi.org/10.1002/mar.21420 Kahler, J. S., & Gore, M. L. (2015). Local perceptions of risk associated with poaching of wildlife implicated in human-wildlife conflicts in Namibia. Biological Conservation, 189, 49–58. https://doi.org/10.1016/j.biocon.2015.02.001. Kaiser, F. G., Hübner, G., & Bogner, F. X. (2005). Contrasting the Theory of Planned Behavior with the Value‐Belief‐Norm Model in Explaining Conservation Behavior 1. Journal of Applied Social Psychology, 35(10), 2150–2170. Kasper, K., Schweikhard, J., Lehmann, M., Ebert, C. L., Erbe, P., Wayakone, S., Nguyen, T. Q., Le, M. D., & Ziegler, T. (2020). The extent of the illegal trade with terrestrial vertebrates in markets and households in Khammouane Province, Lao PDR. Nature Conservation, 41, 25–45. https://doi.org/10.3897/natureconservation.41.51888 Kenny, D. A. (2016, April 9). Multiple Latent Variable Models: Confirmatory Factor Analysis. https://davidakenny.net/cm/mfactor.htm Kenny, D. A. (2020, June 5). Measuring Model Fit. Structural Equation Modeling. https://davidakenny.net/cm/causalm.htm Khanwilkar, S., Sosnowski, M., & Guynup, S. (2022). Patterns of illegal and legal tiger parts entering the United States over a decade (2003–2012). Conservation Science and Practice, 4(3). https://doi.org/10.1111/csp2.622 Kidd, L. R., Garrard, G. E., Bekessy, S. A., Mills, M., Camilleri, A. R., Fidler, F., & Louis, W. (2019). Messaging matters: A systematic review of the conservation messaging literature. Biological Conservation, 236, 92–99. Klöckner, C. A. (2013). A comprehensive model of the psychology of environmental behaviour—A meta-analysis. Global Environmental Change, 23(5), 1028–1038. https://doi.org/10.1016/j.gloenvcha.2013.05.014 Kollmuss, A., & Agyeman, J. (2002). Mind the Gap: Why do people act environmentally and what are the barriers to pro-environmental behavior? Environmental Education Research, 8(3), 239–260. https://doi.org/10.1080/13504620220145401 Kurland, J., Pires, S. F., McFann, S. C., & Moreto, W. D. (2017). Wildlife crime: A conceptual integration, literature review, and methodological critique. Crime Science, 6(1), 1. https://doi.org/10.1186/s40163-017-0066-0. 133 Levin, K., Cashore, B., Bernstein, S., & Auld, G. (2012). Overcoming the tragedy of super wicked problems: Constraining our future selves to ameliorate global climate change. Policy Sciences, 45(2), 123–152. https://doi.org/10.1007/s11077-012- 9151-0. Liordos, V., Kontsiotis, V. J., Eleftheriadou, I., Telidis, S., & Triantafyllidis, A. (2021). Wildlife Value Orientations and Demographics in Greece. Earth, 2(3), 457–467. https://doi.org/10.3390/earth2030027 Liu, R. W., Lapinski, M. K., Kerr, J. M., Zhao, J., Bum, T., & Lu, Z. (2022). Culture and Social Norms: Development and Application of a Model for Culturally Contextualized Communication Measurement (MC3M). Frontiers in Communication, 6. https://www.frontiersin.org/article/10.3389/fcomm.2021.770513 Liu, X., Zou, Y., & Wu, J. (2018). Factors Influencing Public-Sphere Pro-Environmental Behavior among Mongolian College Students: A Test of Value–Belief–Norm Theory. Sustainability, 10(5), 1384. https://doi.org/10.3390/su10051384 López-Mosquera, N. (2016). Gender differences, theory of planned behavior and willingness to pay. Journal of Environmental Psychology, 45, 165–175. https://doi.org/10.1016/j.jenvp.2016.01.006 López-Mosquera, N., García, T., & Barrena, R. (2014). An extension of the Theory of Planned Behavior to predict willingness to pay for the conservation of an urban park. Journal of Environmental Management, 135, 91–99. https://doi.org/10.1016/j.jenvman.2014.01.019 López-Mosquera, N., & Sánchez, M. (2012). Theory of Planned Behavior and the Value- Belief-Norm Theory explaining willingness to pay for a suburban park. Journal of Environmental Management, 113, 251–262. https://doi.org/10.1016/j.jenvman.2012.08.029 Lundberg, P., Vainio, A., MacMillan, D. C., Smith, R. J., Veríssimo, D., & Arponen, A. (2019). The effect of knowledge, species aesthetic appeal, familiarity and conservation need on willingness to donate. Animal Conservation, 22(5), 432–443. https://doi.org/10.1111/acv.12477 Manfredo, M. J., Teel, T. L., Carlos, A. W. D., Sullivan, L., Bright, A. D., Dietsch, A. M., Bruskotter, J., & Fulton, D. (2020). The changing sociocultural context of wildlife conservation. Conservation Biology, 34(6), 1549–1559. https://doi.org/10.1111/cobi.13493 Manfredo, M. J., Teel, T. L., & Dietsch, A. M. (2016). Implications of human value shift and persistence for biodiversity conservation. Conservation Biology, 30(2), 287–296. 134 Manfredo, M. J., Teel, T. L., Don Carlos, A. W., Sullivan, L., Bright, A. D., Dietsch, A. M., Bruskotter, J., & Fulton, D. (2020). The changing sociocultural context of wildlife conservation. Conservation Biology, 34(6), 1549–1559. Manfredo, M. J., Teel, T., & Zinn, H. (2009). Understanding Global Values toward Wildlife. In Wildlife and Society: The Science of Human Dimensions (pp. 31–43). Island Press. Manfredo, M. J., Urquiza-Haas, E. G., Don Carlos, A. W., Bruskotter, J. T., & Dietsch, A. M. (2020). How anthropomorphism is changing the social context of modern wildlife conservation. Biological Conservation, 241, 108297. https://doi.org/10.1016/j.biocon.2019.108297 Martín-López, B., Montes, C., & Benayas, J. (2007a). The non-economic motives behind the willingness to pay for biodiversity conservation. Biological Conservation, 139(1–2), 67–82. Martín-López, B., Montes, C., & Benayas, J. (2007b). The non-economic motives behind the willingness to pay for biodiversity conservation. Biological Conservation, 139(1), 67– 82. https://doi.org/10.1016/j.biocon.2007.06.005 McNamara, J., Rowcliffe, M., Cowlishaw, G., Alexander, J. S., Ntiamoa-Baidu, Y., Brenya, A., & Milner-Gulland, E. J. (2016). Characterising Wildlife Trade Market Supply-Demand Dynamics. PLoS One, 11(9), 0162972. https://doi.org/10.1371/journal.pone.0162972. Megias, D. A., Anderson, S. C., Smith, R. J., & Veríssimo, D. (2017). Investigating the impact of media on demand for wildlife: A case study of Harry Potter and the UK trade in owls. PLoS ONE, 12(10), 1–13. https://doi.org/10.1371/journal.pone.0182368 Militz, T. A., & Foale, S. (2017). The “Nemo Effect”: Perception and reality of Finding Nemo’s impact on marine aquarium fisheries. Fish and Fisheries, 18(3), 596–606. https://doi.org/10.1111/faf.12202 Miller - Carpenter, S. (2018). Ten Steps in Scale Development and Reporting: A Guide for Researchers. Communication Methods and Measures, 12, 25–44. https://doi.org/10.1080/19312458.2017.1396583 Montaño, D. E., & Kasprzyk, D. (2008). Theory of reasoned action, theory of planned behavior, and the integrated behavioral model. In Health Behavior and Health Education: Theory, Research, and Practice (4th ed., pp. 67–96). Jossey-Bass. https://books.google.com/books?hl=en&lr=&id=9BQWCgAAQBAJ&oi=fnd&pg=PA9 5&dq=The+theory+of+reasoned+action+and+the+theory+of+planned+behavior+an d+the+integrated+model&ots=eeKb7guS95&sig=MgBoNswg5sF64ydLxYgEfkYdSKs Munson, S., Kotcher, J., Maibach, E., Rosenthal, S. A., & Leiserowitz, A. (2021). The Role of Felt Responsibility in Climate Change Political Participation. Oxford Open Climate Change, kgab012. https://doi.org/10.1093/oxfclm/kgab012 135 My Easy Statistics (Director). (2020, January 22). How to test the Convergent Validity and Discriminant Validity of SEM Model? https://www.youtube.com/watch?v=5esq4D5ZG4s Nellemann, C., Henriksen, R., Raxter, P., Ash, N., & Mrema, E. (2014). Environmental crime crisis: Threats to sustainable development from illegal exploitation and trade in wildlife and forest resources. United Nations Environment Programme (UNEP). Nijman, V., Ardiansyah, A., Langgeng, A., Hendrik, R., Hedger, K., Foreman, G., Morcatty, T. Q., Siriwat, P., van Balen, S. (Bas), Eaton, J. A., Shepherd, C. R., Gomez, L., Imron, M. A., & Nekaris, K. A. I. (2022). Illegal Wildlife Trade in Traditional Markets, on Instagram and Facebook: Raptors as a Case Study. Birds, 3(1), 99–116. https://doi.org/10.3390/birds3010008 Nijman, V., & Nekaris, K. A. I. (2017). The Harry Potter effect: The rise in trade of owls as pets in Java and Bali, Indonesia. Global Ecology and Conservation, 11, 84–94. https://doi.org/10.1016/j.gecco.2017.04.004 Ntanos, S., Kyriakopoulos, G., Skordoulis, M., Chalikias, M., & Arabatzis, G. (2019). An Application of the New Environmental Paradigm (NEP) Scale in a Greek Context. Energies, 12(2), 239. https://doi.org/10.3390/en12020239 Obeng, E. A., & Aguilar, F. X. (2018). Value orientation and payment for ecosystem services: Perceived detrimental consequences lead to willingness-to-pay for ecosystem services. Journal of Environmental Management, 206, 458–471. https://doi.org/10.1016/j.jenvman.2017.10.059 O’Donoghue, P., & Rutz, C. (2016). Real‐time anti‐poaching tags could help prevent imminent species extinctions. The Journal of Applied Ecology, 53(1), 5–10. https://doi.org/10.1111/1365-2664.12452 Oh, R. R. Y., Fielding, K. S., Nghiem, L. T. P., Chang, C. C., Carrasco, L. R., & Fuller, R. A. (2021). Connection to nature is predicted by family values, social norms and personal experiences of nature. Global Ecology and Conservation, 28, e01632. https://doi.org/10.1016/j.gecco.2021.e01632 Pirralha, A. (2020a). Chapter 3 Measurement Invariance. In Testing for Measurement Invariance with Many Groups. https://bookdown.org/content/5737/invariance.html Pirralha, A. (2020b). Chapter 4 Checklist for Measurement Invariance with MGCFA. In Testing for Measurement Invariance with Many Groups. https://bookdown.org/content/5737/check.html Pirralha, A. (2020c). Chapter 5 Testing for measurement invariance with lavaan in R. In Testing for Measurement Invariance with Many Groups. https://bookdown.org/content/5737/lavaan.html 136 Sánchez-García, M., Zouaghi, F., Lera-López, F., & Faulin, J. (2021). An extended behavior model for explaining the willingness to pay to reduce the air pollution in road transportation. Journal of Cleaner Production, 314, 128134. https://doi.org/10.1016/j.jclepro.2021.128134 Saris, W. E., Revilla, M., Krosnick, J. A., & Shaeffer, E. M. (2010). Comparing questions with agree/disagree response options to questions with construct-specific response options. https://doi.org/10.18148/srm/2010.v4i1.2682 Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance structure analysis. In A. von Eye & C. C. Clogg (Eds.), Latent variables analysis: Applications for developmental research. (pp. 399–419). Sage Publications, Inc. Schwartz, S. H. (1977). Normative Influences on Altruism11This work was supported by NSF Grant SOC 72-05417. I am indebted to L. Berkowitz, R. Dienstbier, H. Schuman, R. Simmons, and R. Tessler for their thoughtful comments on an early draft of this chapter. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 10, pp. 221–279). Academic Press. https://doi.org/10.1016/S0065-2601(08)60358-5 Schwartz, S. H. (1992). Universals in the Content and Structure of Values: Theoretical Advances and Empirical Tests in 20 Countries. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 25, pp. 1–65). Academic Press. https://doi.org/10.1016/S0065-2601(08)60281-6 Shalender, K., & Sharma, N. (2021). Using extended theory of planned behaviour (TPB) to predict adoption intention of electric vehicles in India. Environment, Development and Sustainability, 23(1), 665–681. https://doi.org/10.1007/s10668-020-00602-7 Sharma, R., & Gupta, A. (2020). Pro-environmental behaviour among tourists visiting national parks: Application of value-belief-norm theory in an emerging economy context. Asia Pacific Journal of Tourism Research, 25(8), 829–840. https://doi.org/10.1080/10941665.2020.1774784 Shreedhar, G., & Mourato, S. (2018). Do biodiversity conservation videos cause pro- environmental spillover effects? Grantham Research Institute on Climate Change and the Environment. Working Paper, 302. http://www.bioecon- network.org/pages/20th%202018/Mourato.pdf Slimak, M. W., & Dietz, T. (2006). Personal Values, Beliefs, and Ecological Risk Perception. Risk Analysis, 26(6), 1689–1705. https://doi.org/10.1111/j.1539- 6924.2006.00832.x Sparks, A. C., Henderson, G. L., Sriram, S. K., & Smith, E. R. A. N. (2021). Measuring Environmental Values and Identity. Society & Natural Resources, 34(3), 291–310. https://doi.org/10.1080/08941920.2020.1817644 137 Steg, L., Bolderdijk, J. W., Keizer, K., & Perlaviciute, G. (2014). An Integrated Framework for Encouraging Pro-environmental Behaviour: The role of values, situational factors and goals. Journal of Environmental Psychology, 38, 104–115. https://doi.org/10.1016/j.jenvp.2014.01.002 Steg, L., Dreijerink, L., & Abrahamse, W. (2005). Factors influencing the acceptability of energy policies: A test of VBN theory. Journal of Environmental Psychology, 25(4), 415–425. https://doi.org/10.1016/j.jenvp.2005.08.003 Steg, L., & Nordlund, A. (2018). Theories to Explain Environmental Behaviour. In Environmental Psychology (pp. 217–227). John Wiley & Sons, Ltd. https://doi.org/10.1002/9781119241072.ch22 Stern, P. C., & Dietz, T. (1994). The value basis of environmental concern. Journal of Social Issues, 50(3), 65–84. Stern, P. C., Dietz, T., Abel, T., Guagnano, G. A., & Kalof, L. (1999). A Value-Belief-Norm Theory of Support for Social Movements: The Case of Environmentalism. Human Ecology Review, 6(2), 81–97. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Allyn and Bacon. Takahashi, B., Duan, R., Joshi, A., Witsen, A. V., & Li, W. (2021). Covering the Environment Beat: A Systematic Review of Research on News Media Coverage of the Environment. In The Handbook of International Trends in Environmental Communication (1st ed., pp. 17–30). Routledge. Takahashi, B., Metag, J., Thaker, J., & Comfort, S. E. (2021). Expanding Conceptualizations of Environmental Communication Research. In The Handbook of International Trends in Environmental Communication (1st ed., pp. 3–13). Routledge. Taufique, K. Md. R., & Vaithianathan, S. (2018). A fresh look at understanding Green consumer behavior among young urban Indian consumers through the lens of Theory of Planned Behavior. Journal of Cleaner Production, 183, 46–55. https://doi.org/10.1016/j.jclepro.2018.02.097 Teel, T. L., & Manfredo, M. J. (2010). Understanding the Diversity of Public Interests in Wildlife Conservation. Conservation Biology, 24(1), 128–139. https://doi.org/10.1111/j.1523-1739.2009.01374.x Teel, T. L., Manfredo, M. J., & Stinchfield, H. M. (2007a). The need and theoretical basis for exploring wildlife value orientations cross-culturally. Human Dimensions of Wildlife, 12(5), 297–305. 138 Teel, T. L., Manfredo, M. J., & Stinchfield, H. M. (2007b). The Need and Theoretical Basis for Exploring Wildlife Value Orientations Cross-Culturally. Human Dimensions of Wildlife, 12(5), 297–305. https://doi.org/10.1080/10871200701555857 Thaker, J. (2021). Environmentalism of the Poor: Global South Perspectives on Environmental Communication. In The Handbook of International Trends in Environmental Communication. Routledge. Thakkar, J. J. (2020). Procedural Steps in Structural Equation Modelling. In J. J. Thakkar (Ed.), Structural Equation Modelling: Application for Research and Practice (with AMOS and R) (pp. 29–34). Springer. https://doi.org/10.1007/978-981-15-3793-6_3 Thøgersen, J. (1996). Recycling and Morality: A Critical Review of the Literature. Environment and Behavior, 28(4), 536–558. https://doi.org/10.1177/0013916596284006 Thomas-Walters, L., McNulty, C., & Veríssimo, D. (2019). A scoping review into the impact of animal imagery on pro-environmental outcomes. Ambio, 49, 1135–1145. Thomas-Walters, L., McNulty, C., & Veríssimo, D. (2020). A scoping review into the impact of animal imagery on pro-environmental outcomes. Ambio, 49(6), 1135–1145. Tyllianakis, E., & Ferrini, S. (2021). Personal attitudes and beliefs and willingness to pay to reduce marine plastic pollution in Indonesia. Marine Pollution Bulletin, 173, 113120. https://doi.org/10.1016/j.marpolbul.2021.113120 Ünal, A. B., Steg, L., & Granskaya, J. (2019). “To support or not to support, that is the question”. Testing the VBN theory in predicting support for car use reduction policies inRussia. Transportation Research Part A: Policy and Practice, 119, 73–81. https://doi.org/10.1016/j.tra.2018.10.042 UNODC. (2020). World Wildlife Crime Report: Trafficking in protected species. (pp. 1–136). United Nations Office on Drugs and Crime. Valle, P. O. D., Rebelo, E., Reis, E., & Menezes, J. (2005). Combining Behavioral Theories to Predict Recycling Involvement. Environment and Behavior, 37(3), 364–396. https://doi.org/10.1177/0013916504272563 van der Werff, E., & Steg, L. (2016). The psychology of participation and interest in smart energy systems: Comparing the value-belief-norm theory and the value-identity- personal norm model. Energy Research & Social Science, 22, 107–114. https://doi.org/10.1016/j.erss.2016.08.022 van Riper, C. J., & Kyle, G. T. (2014a). Understanding the internal processes of behavioral engagement in a national park: A latent variable path analysis of the value-belief- norm theory. Journal of Environmental Psychology, 38, 288–297. https://doi.org/10.1016/j.jenvp.2014.03.002 139 van Riper, C. J., & Kyle, G. T. (2014b). Understanding the internal processes of behavioral engagement in a national park: A latent variable path analysis of the value-belief- norm theory. Journal of Environmental Psychology, 38, 288–297. https://doi.org/10.1016/j.jenvp.2014.03.002 Vaske, J. J., Don Carlos, A. W., Manfredo, M. J., & Teel, T. L. (2022). Evaluating alternative survey methodologies in human dimensions of wildlife research. Human Dimensions of Wildlife, 0(0), 1–15. https://doi.org/10.1080/10871209.2022.2057622 Vaske, J. J., Jacobs, M. H., & Sijtsma, M. T. J. (2011). Wildlife value orientations and demographics in The Netherlands. European Journal of Wildlife Research, 57(6), 1179–1187. https://doi.org/10.1007/s10344-011-0531-0 Veríssimo, D. (2019). The Past, Present, and Future of Using Social Marketing to Conserve Biodiversity. Social Marketing Quarterly, 25(1), 3–8. https://doi.org/10.1177/1524500419825545. Veríssimo, D., Campbell, H. A., Tollington, S., MacMillan, D. C., & Smith, R. J. (2018). Why do people donate to conservation? Insights from a “real world” campaign. PLoS One, 13(1), 0191888. https://doi.org/10.1371/journal.pone.0191888. Veríssimo, D., Vaughan, G., Ridout, M., Waterman, C., MacMillan, D., & Smith, R. J. (2017). Increased conservation marketing effort has major fundraising benefits for even the least popular species. Biological Conservation, 211, 95–101. https://doi.org/10.1016/j.biocon.2017.04.018. Veríssimo, D., & Wan, A. K. Y. (2019). Characterizing efforts to reduce consumer demand for wildlife products. Conservation Biology, 33(3), 623–633. https://doi.org/10.1111/cobi.13227. Wakita, K., Kurokura, H., Oishi, T., Shen, Z., & Furuya, K. (2019). Exploring the effect of psychometric variables on willingness to pay for marine ecosystem services: A survey in Japan. Ecosystem Services, 35, 130–138. https://doi.org/10.1016/j.ecoser.2018.12.003 What are wildlife value orientations? (n.d.). America’s Wildlife Values. Retrieved September 4, 2022, from https://sites.warnercnr.colostate.edu/wildlifevalues/home- page/what-are-value-orientations/ Wyatt, T., Maher, J., Allen, D., Clarke, N., & Rook, D. (2021). The welfare of wildlife: An interdisciplinary analysis of harm in the legal and illegal wildlife trades and possible ways forward. Crime, Law and Social Change. https://doi.org/10.1007/s10611-021- 09984-9 Zucca, P., Rossmann, M.-C., Osorio, J. E., Karem, K., De Benedictis, P., Haißl, J., De Franceschi, P., Calligaris, E., Kohlweiß, M., Meddi, G., Gabrutsch, W., Mairitsch, H., Greco, O., Furlani, R., Maggio, M., Tolomei, M., Bremini, A., Fischinger, I., Zambotto, P., … 140 Zamaro, G. (2020). The “Bio-Crime Model” of Cross-Border Cooperation Among Veterinary Public Health, Justice, Law Enforcements, and Customs to Tackle the Illegal Animal Trade/Bio-Terrorism and to Prevent the Spread of Zoonotic Diseases Among Human Population. Frontiers in Veterinary Science, 7. https://www.frontiersin.org/articles/10.3389/fvets.2020.593683 141 APPENDIX A: MULTIVARIATE NORMALITY TESTS IN R $multivariateNormality (USA) Test Statistic p value Result 1 Mardia Skewness 154637.914481546 0 NO 2 Mardia Kurtosis 186.295829703372 0 NO 3 MVN NO $multivariateNormality Test Statistic p value Result 1 Mardia Skewness 163327.797601619 0 NO 2 Mardia Kurtosis 188.924204987438 0 NO 3 MVN NO 142 APPENDIX B: THEORY OF PLANNED BEHAVIOR CFA MODEL FIT STATISTICS Raw output from RStudio ## lavaan 0.6-11 ended normally after 112 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 108 ## ## Number of observations per group: ## USA 900 ## India 920 ## Number of missing patterns per group: ## USA 1 ## India 2 ## ## Model Test User Model: ## Standard Robust ## Test Statistic 1349.825 878.601 ## Degrees of freedom 196 196 ## P-value (Chi-square) 0.000 0.000 ## Scaling correction factor 1.536 ## Yuan-Bentler correction (Mplus variant) ## Test statistic for each group: ## USA 912.975 594.256 ## India 436.850 284.346 ## ## Model Test Baseline Model: ## ## Test statistic 22974.689 13301.927 ## Degrees of freedom 240 240 ## P-value 0.000 0.000 ## Scaling correction factor 1.727 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.949 0.948 ## Tucker-Lewis Index (TLI) 0.938 0.936 ## ## Robust Comparative Fit Index (CFI) 0.954 ## Robust Tucker-Lewis Index (TLI) 0.943 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -40966.728 -40966.728 143 ## Scaling correction factor 1.698 ## for the MLR correction ## Loglikelihood unrestricted model (H1) NA NA ## Scaling correction factor 1.594 ## for the MLR correction ## ## Akaike (AIC) 82149.457 82149.457 ## Bayesian (BIC) 82744.169 82744.169 ## Sample-size adjusted Bayesian (BIC) 82401.057 82401.057 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.080 0.062 ## 90 Percent confidence interval - lower 0.076 0.059 ## 90 Percent confidence interval - upper 0.085 0.065 ## P-value RMSEA <= 0.05 0.000 0.000 ## ## Robust RMSEA 0.077 ## 90 Percent confidence interval - lower 0.072 ## 90 Percent confidence interval - upper 0.082 ## ## Standardized Root Mean Square Residual: ## ## SRMR 0.062 0.062 ## ## Parameter Estimates: ## ## Standard errors Sandwich ## Information bread Observed ## Observed information based on Hessian ## ## ## Group 1 [USA]: ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa =~ ## Att1 1.000 1.223 0.888 ## Att2 1.037 0.031 33.068 0.000 1.269 0.894 ## Att3 1.069 0.034 31.879 0.000 1.308 0.907 ## Att4 1.079 0.030 35.754 0.000 1.320 0.926 ## Att5 1.071 0.036 29.360 0.000 1.310 0.900 ## SubNorms_cfa =~ ## SN1 1.000 1.417 0.887 ## SN2 1.003 0.029 34.506 0.000 1.422 0.830 ## SN3 0.889 0.029 31.023 0.000 1.259 0.835 144 ## PBC_cfa =~ ## PBC1 1.000 1.143 0.679 ## PBC2 1.251 0.073 17.079 0.000 1.429 0.749 ## PBC3 1.129 0.055 20.526 0.000 1.290 0.752 ## PBC4 1.144 0.064 17.919 0.000 1.308 0.731 ## PBC6 0.799 0.053 14.959 0.000 0.913 0.611 ## Intention_cfa =~ ## Int1 1.000 1.558 0.845 ## Int2 1.083 0.027 40.481 0.000 1.688 0.944 ## Int3 1.088 0.028 38.573 0.000 1.695 0.949 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa ~~ ## SubNorms_cfa 1.164 0.090 12.974 0.000 0.671 0.671 ## PBC_cfa 0.732 0.090 8.107 0.000 0.524 0.524 ## SubNorms_cfa ~~ ## PBC_cfa 1.058 0.093 11.387 0.000 0.653 0.653 ## Attitude_cfa ~~ ## Intention_cfa 1.027 0.089 11.591 0.000 0.539 0.539 ## SubNorms_cfa ~~ ## Intention_cfa 1.575 0.102 15.475 0.000 0.713 0.713 ## PBC_cfa ~~ ## Intention_cfa 1.356 0.101 13.420 0.000 0.762 0.762 ## ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 5.441 0.046 118.495 0.000 5.441 3.950 ## .Att2 5.280 0.047 111.651 0.000 5.280 3.722 ## .Att3 5.150 0.048 107.162 0.000 5.150 3.572 ## .Att4 5.257 0.048 110.662 0.000 5.257 3.689 ## .Att5 5.236 0.049 107.816 0.000 5.236 3.594 ## .SN1 4.133 0.053 77.669 0.000 4.133 2.589 ## .SN2 3.774 0.057 66.149 0.000 3.774 2.205 ## .SN3 4.468 0.050 88.851 0.000 4.468 2.962 ## .PBC1 4.747 0.056 84.594 0.000 4.747 2.820 ## .PBC2 3.291 0.064 51.721 0.000 3.291 1.724 ## .PBC3 4.239 0.057 74.114 0.000 4.239 2.470 ## .PBC4 4.057 0.060 68.055 0.000 4.057 2.268 ## .PBC6 4.777 0.050 95.934 0.000 4.777 3.198 ## .Int1 3.944 0.061 64.141 0.000 3.944 2.138 ## .Int2 3.489 0.060 58.562 0.000 3.489 1.952 ## .Int3 3.350 0.060 56.272 0.000 3.350 1.876 ## Attitude_cfa 0.000 0.000 0.000 ## SubNorms_cfa 0.000 0.000 0.000 ## PBC_cfa 0.000 0.000 0.000 145 ## Intention_cfa 0.000 0.000 0.000 ## ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 0.401 0.052 7.685 0.000 0.401 0.211 ## .Att2 0.403 0.052 7.716 0.000 0.403 0.200 ## .Att3 0.368 0.040 9.237 0.000 0.368 0.177 ## .Att4 0.288 0.031 9.405 0.000 0.288 0.142 ## .Att5 0.405 0.050 8.030 0.000 0.405 0.191 ## .SN1 0.542 0.063 8.592 0.000 0.542 0.213 ## .SN2 0.910 0.083 10.977 0.000 0.910 0.310 ## .SN3 0.689 0.066 10.423 0.000 0.689 0.303 ## .PBC1 1.528 0.096 15.849 0.000 1.528 0.539 ## .PBC2 1.602 0.119 13.427 0.000 1.602 0.440 ## .PBC3 1.281 0.096 13.284 0.000 1.281 0.435 ## .PBC4 1.488 0.116 12.795 0.000 1.488 0.465 ## .PBC6 1.398 0.110 12.713 0.000 1.398 0.626 ## .Int1 0.976 0.094 10.330 0.000 0.976 0.287 ## .Int2 0.346 0.042 8.276 0.000 0.346 0.108 ## .Int3 0.318 0.045 7.043 0.000 0.318 0.100 ## Attitude_cfa 1.497 0.123 12.146 0.000 1.000 1.000 ## SubNorms_cfa 2.007 0.116 17.329 0.000 1.000 1.000 ## PBC_cfa 1.306 0.123 10.629 0.000 1.000 1.000 ## Intention_cfa 2.427 0.135 17.943 0.000 1.000 1.000 ## ## R-Square: ## Estimate ## Att1 0.789 ## Att2 0.800 ## Att3 0.823 ## Att4 0.858 ## Att5 0.809 ## SN1 0.787 ## SN2 0.690 ## SN3 0.697 ## PBC1 0.461 ## PBC2 0.560 ## PBC3 0.565 ## PBC4 0.535 ## PBC6 0.374 ## Int1 0.713 ## Int2 0.892 ## Int3 0.900 ## ## ## Group 2 [India]: 146 ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa =~ ## Att1 1.000 0.875 0.814 ## Att2 1.000 0.044 22.855 0.000 0.875 0.801 ## Att3 1.167 0.051 23.088 0.000 1.021 0.815 ## Att4 1.064 0.049 21.713 0.000 0.931 0.844 ## Att5 1.017 0.054 18.846 0.000 0.890 0.785 ## SubNorms_cfa =~ ## SN1 1.000 0.980 0.796 ## SN2 0.982 0.045 21.710 0.000 0.962 0.715 ## SN3 0.953 0.039 24.538 0.000 0.934 0.794 ## PBC_cfa =~ ## PBC1 1.000 0.972 0.757 ## PBC2 1.284 0.062 20.627 0.000 1.248 0.773 ## PBC3 1.111 0.060 18.474 0.000 1.080 0.733 ## PBC4 1.225 0.064 19.156 0.000 1.190 0.769 ## PBC6 0.869 0.049 17.599 0.000 0.845 0.730 ## Intention_cfa =~ ## Int1 1.000 1.139 0.872 ## Int2 1.033 0.035 29.139 0.000 1.177 0.870 ## Int3 1.130 0.039 29.203 0.000 1.287 0.909 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa ~~ ## SubNorms_cfa 0.675 0.050 13.525 0.000 0.787 0.787 ## PBC_cfa 0.619 0.051 12.219 0.000 0.728 0.728 ## SubNorms_cfa ~~ ## PBC_cfa 0.751 0.061 12.281 0.000 0.789 0.789 ## Attitude_cfa ~~ ## Intention_cfa 0.658 0.055 11.997 0.000 0.659 0.659 ## SubNorms_cfa ~~ ## Intention_cfa 0.823 0.066 12.519 0.000 0.737 0.737 ## PBC_cfa ~~ ## Intention_cfa 0.933 0.076 12.332 0.000 0.842 0.842 ## ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 5.953 0.035 167.830 0.000 5.953 5.533 ## .Att2 5.901 0.036 163.840 0.000 5.901 5.402 ## .Att3 5.658 0.041 137.014 0.000 5.658 4.517 ## .Att4 5.946 0.036 163.500 0.000 5.946 5.390 ## .Att5 5.890 0.037 157.626 0.000 5.890 5.197 ## .SN1 5.555 0.041 136.932 0.000 5.555 4.515 147 ## .SN2 5.400 0.044 121.759 0.000 5.400 4.014 ## .SN3 5.679 0.039 146.594 0.000 5.679 4.833 ## .PBC1 5.654 0.042 133.513 0.000 5.654 4.402 ## .PBC2 4.977 0.053 93.458 0.000 4.977 3.081 ## .PBC3 5.171 0.049 106.417 0.000 5.171 3.508 ## .PBC4 5.114 0.051 100.242 0.000 5.114 3.305 ## .PBC6 5.793 0.038 151.806 0.000 5.793 5.005 ## .Int1 5.526 0.043 128.265 0.000 5.526 4.229 ## .Int2 5.407 0.045 121.282 0.000 5.407 3.999 ## .Int3 5.389 0.047 115.376 0.000 5.389 3.804 ## Attitude_cfa 0.000 0.000 0.000 ## SubNorms_cfa 0.000 0.000 0.000 ## PBC_cfa 0.000 0.000 0.000 ## Intention_cfa 0.000 0.000 0.000 ## ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 0.391 0.047 8.325 0.000 0.391 0.338 ## .Att2 0.428 0.045 9.461 0.000 0.428 0.359 ## .Att3 0.526 0.049 10.769 0.000 0.526 0.335 ## .Att4 0.349 0.045 7.838 0.000 0.349 0.287 ## .Att5 0.493 0.061 8.033 0.000 0.493 0.384 ## .SN1 0.555 0.053 10.445 0.000 0.555 0.366 ## .SN2 0.884 0.101 8.735 0.000 0.884 0.489 ## .SN3 0.509 0.057 8.905 0.000 0.509 0.369 ## .PBC1 0.706 0.056 12.566 0.000 0.706 0.428 ## .PBC2 1.052 0.084 12.512 0.000 1.052 0.403 ## .PBC3 1.006 0.087 11.551 0.000 1.006 0.463 ## .PBC4 0.977 0.085 11.550 0.000 0.977 0.408 ## .PBC6 0.626 0.047 13.240 0.000 0.626 0.467 ## .Int1 0.409 0.047 8.659 0.000 0.409 0.240 ## .Int2 0.444 0.047 9.361 0.000 0.444 0.243 ## .Int3 0.349 0.044 7.872 0.000 0.349 0.174 ## Attitude_cfa 0.766 0.067 11.365 0.000 1.000 1.000 ## SubNorms_cfa 0.960 0.081 11.780 0.000 1.000 1.000 ## PBC_cfa 0.944 0.090 10.470 0.000 1.000 1.000 ## Intention_cfa 1.298 0.097 13.395 0.000 1.000 1.000 ## ## R-Square: ## Estimate ## Att1 0.662 ## Att2 0.641 ## Att3 0.665 ## Att4 0.713 ## Att5 0.616 ## SN1 0.634 148 ## SN2 0.511 ## SN3 0.631 ## PBC1 0.572 ## PBC2 0.597 ## PBC3 0.537 ## PBC4 0.592 ## PBC6 0.533 ## Int1 0.760 ## Int2 0.757 ## Int3 0.826 149 APPENDIX C: THEORY OF PLANNED BEHAVIOR SEM MODEL - INDIA ## lavaan 0.6-11 ended normally after 33 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 38 ## ## Number of observations 920 ## ## Model Test User Model: ## ## Test statistic 443.524 ## Degrees of freedom 98 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 10105.973 ## Degrees of freedom 120 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.965 ## Tucker-Lewis Index (TLI) 0.958 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -19652.593 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 39381.186 ## Bayesian (BIC) 39564.512 ## Sample-size adjusted Bayesian (BIC) 39443.829 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.062 ## 90 Percent confidence interval - lower 0.056 ## 90 Percent confidence interval - upper 0.068 ## P-value RMSEA <= 0.05 0.000 ## ## Standardized Root Mean Square Residual: ## ## SRMR 0.039 150 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa_IND =~ ## Att1 0.875 0.030 29.275 0.000 0.875 0.813 ## Att2 0.875 0.031 28.587 0.000 0.875 0.801 ## Att3 1.021 0.035 29.377 0.000 1.021 0.815 ## Att4 0.931 0.030 31.009 0.000 0.931 0.844 ## Att5 0.890 0.032 27.771 0.000 0.890 0.785 ## SubNorms_cfa_IND =~ ## SN1 0.980 0.036 27.415 0.000 0.980 0.797 ## SN2 0.961 0.041 23.608 0.000 0.961 0.715 ## SN3 0.933 0.034 27.308 0.000 0.933 0.794 ## PBC_cfa_IND =~ ## PBC1 0.974 0.037 26.187 0.000 0.974 0.758 ## PBC2 1.251 0.046 27.008 0.000 1.251 0.775 ## PBC3 1.077 0.043 24.885 0.000 1.077 0.731 ## PBC4 1.188 0.045 26.678 0.000 1.188 0.768 ## PBC6 0.844 0.034 24.783 0.000 0.844 0.729 ## Intention_cfa_IND =~ ## Int1 0.609 0.025 23.959 0.000 1.149 0.871 ## Int2 0.617 0.026 23.722 0.000 1.163 0.860 ## Int3 0.687 0.028 24.620 0.000 1.296 0.908 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Intention_cfa_IND ~ ## Attitud_cf_IND 0.027 0.088 0.302 0.763 0.014 0.014 ## SubNrms_cf_IND 0.316 0.112 2.829 0.005 0.168 0.168 ## PBC_cfa_IND 1.317 0.118 11.171 0.000 0.698 0.698 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa_IND ~~ ## SubNrms_cf_IND 0.787 0.019 42.214 0.000 0.787 0.787 ## PBC_cfa_IND 0.728 0.020 36.397 0.000 0.728 0.728 ## SubNorms_cfa_IND ~~ ## PBC_cfa_IND 0.788 0.020 40.313 0.000 0.788 0.788 ## ## Variances: 151 ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 0.392 0.022 17.647 0.000 0.392 0.338 ## .Att2 0.428 0.024 17.981 0.000 0.428 0.359 ## .Att3 0.526 0.030 17.594 0.000 0.526 0.335 ## .Att4 0.349 0.021 16.611 0.000 0.349 0.287 ## .Att5 0.492 0.027 18.332 0.000 0.492 0.383 ## .SN1 0.553 0.036 15.586 0.000 0.553 0.365 ## .SN2 0.886 0.049 18.004 0.000 0.886 0.489 ## .SN3 0.510 0.032 15.678 0.000 0.510 0.369 ## .PBC1 0.702 0.038 18.313 0.000 0.702 0.425 ## .PBC2 1.044 0.058 17.960 0.000 1.044 0.400 ## .PBC3 1.011 0.054 18.793 0.000 1.011 0.466 ## .PBC4 0.982 0.054 18.107 0.000 0.982 0.410 ## .PBC6 0.628 0.033 18.827 0.000 0.628 0.469 ## .Int1 0.420 0.027 15.734 0.000 0.420 0.241 ## .Int2 0.475 0.029 16.301 0.000 0.475 0.260 ## .Int3 0.356 0.028 12.939 0.000 0.356 0.175 ## Attitud_cf_IND 1.000 1.000 1.000 ## SubNrms_cf_IND 1.000 1.000 1.000 ## PBC_cfa_IND 1.000 1.000 1.000 ## .Intentn_cf_IND 1.000 0.281 0.281 ## ## R-Square: ## Estimate ## Att1 0.662 ## Att2 0.641 ## Att3 0.665 ## Att4 0.713 ## Att5 0.617 ## SN1 0.635 ## SN2 0.511 ## SN3 0.631 ## PBC1 0.575 ## PBC2 0.600 ## PBC3 0.534 ## PBC4 0.590 ## PBC6 0.531 ## Int1 0.759 ## Int2 0.740 ## Int3 0.825 ## Intentn_cf_IND 0.719 152 APPENDIX D: THEORY OF PLANNED BEHAVIOR SEM MODEL - USA ## lavaan 0.6-11 ended normally after 35 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 38 ## ## Number of observations 900 ## ## Model Test User Model: ## ## Test statistic 912.975 ## Degrees of freedom 98 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 12813.915 ## Degrees of freedom 120 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.936 ## Tucker-Lewis Index (TLI) 0.921 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -21363.353 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 42802.706 ## Bayesian (BIC) 42985.197 ## Sample-size adjusted Bayesian (BIC) 42864.515 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.096 ## 90 Percent confidence interval - lower 0.090 ## 90 Percent confidence interval - upper 0.102 ## P-value RMSEA <= 0.05 0.000 ## ## Standardized Root Mean Square Residual: ## ## SRMR 0.093 153 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa_USA =~ ## Att1 1.223 0.036 33.899 0.000 1.223 0.888 ## Att2 1.269 0.037 34.304 0.000 1.269 0.894 ## Att3 1.308 0.037 35.142 0.000 1.308 0.907 ## Att4 1.320 0.036 36.458 0.000 1.320 0.926 ## Att5 1.310 0.038 34.641 0.000 1.310 0.900 ## SubNorms_cfa_USA =~ ## SN1 1.417 0.043 32.848 0.000 1.417 0.887 ## SN2 1.422 0.048 29.646 0.000 1.422 0.830 ## SN3 1.259 0.042 29.888 0.000 1.259 0.835 ## PBC_cfa_USA =~ ## PBC1 1.143 0.052 21.803 0.000 1.143 0.679 ## PBC2 1.429 0.057 24.876 0.000 1.429 0.749 ## PBC3 1.290 0.052 25.015 0.000 1.290 0.752 ## PBC4 1.308 0.054 24.086 0.000 1.308 0.731 ## PBC6 0.913 0.048 19.075 0.000 0.913 0.611 ## Intention_cfa_USA =~ ## Int1 0.906 0.034 26.474 0.000 1.558 0.845 ## Int2 0.982 0.033 29.961 0.000 1.688 0.944 ## Int3 0.986 0.033 30.068 0.000 1.695 0.949 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Intention_cfa_USA ~ ## Attitud_cf_USA 0.051 0.057 0.888 0.375 0.029 0.029 ## SubNrms_cf_USA 0.619 0.074 8.362 0.000 0.360 0.360 ## PBC_cfa_USA 0.879 0.074 11.948 0.000 0.511 0.511 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## Attitude_cfa_USA ~~ ## SubNrms_cf_USA 0.671 0.021 31.686 0.000 0.671 0.671 ## PBC_cfa_USA 0.524 0.028 18.579 0.000 0.524 0.524 ## SubNorms_cfa_USA ~~ ## PBC_cfa_USA 0.653 0.025 26.482 0.000 0.653 0.653 ## ## Variances: 154 ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .Att1 0.401 0.023 17.756 0.000 0.401 0.211 ## .Att2 0.403 0.023 17.507 0.000 0.403 0.200 ## .Att3 0.368 0.022 16.895 0.000 0.368 0.177 ## .Att4 0.288 0.019 15.577 0.000 0.288 0.142 ## .Att5 0.405 0.023 17.278 0.000 0.405 0.191 ## .SN1 0.542 0.042 12.757 0.000 0.542 0.213 ## .SN2 0.910 0.056 16.130 0.000 0.910 0.310 ## .SN3 0.689 0.043 15.937 0.000 0.689 0.303 ## .PBC1 1.528 0.083 18.426 0.000 1.528 0.539 ## .PBC2 1.602 0.094 17.028 0.000 1.602 0.440 ## .PBC3 1.281 0.076 16.949 0.000 1.281 0.435 ## .PBC4 1.488 0.085 17.446 0.000 1.488 0.465 ## .PBC6 1.398 0.073 19.277 0.000 1.398 0.626 ## .Int1 0.976 0.052 18.630 0.000 0.976 0.287 ## .Int2 0.346 0.030 11.510 0.000 0.346 0.108 ## .Int3 0.318 0.029 10.775 0.000 0.318 0.100 ## Attitud_cf_USA 1.000 1.000 1.000 ## SubNrms_cf_USA 1.000 1.000 1.000 ## PBC_cfa_USA 1.000 1.000 1.000 ## .Intentn_cf_USA 1.000 0.338 0.338 ## ## R-Square: ## Estimate ## Att1 0.789 ## Att2 0.800 ## Att3 0.823 ## Att4 0.858 ## Att5 0.809 ## SN1 0.787 ## SN2 0.690 ## SN3 0.697 ## PBC1 0.461 ## PBC2 0.560 ## PBC3 0.565 ## PBC4 0.535 ## PBC6 0.374 ## Int1 0.713 ## Int2 0.892 ## Int3 0.900 ## Intentn_cf_USA 0.662 155 APPENDIX E: VALUE-BELIEF-NORM THEORY SEM MODEL - INDIA ## lavaan 0.6-11 ended normally after 61 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 120 ## ## Number of observations 920 ## ## Model Test User Model: ## ## Test statistic 1316.455 ## Degrees of freedom 509 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 11387.171 ## Degrees of freedom 561 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.925 ## Tucker-Lewis Index (TLI) 0.918 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -45394.886 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 91029.771 ## Bayesian (BIC) 91608.696 ## Sample-size adjusted Bayesian (BIC) 91227.590 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.042 ## 90 Percent confidence interval - lower 0.039 ## 90 Percent confidence interval - upper 0.044 ## P-value RMSEA <= 0.05 1.000 ## ## Standardized Root Mean Square Residual: ## ## SRMR 0.077 156 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## BioVal_cfa =~ ## BioV1 0.551 0.028 19.457 0.000 0.551 0.625 ## BioV2 0.516 0.021 24.028 0.000 0.516 0.736 ## BioV3 0.446 0.019 23.064 0.000 0.446 0.714 ## BioV4 0.616 0.024 25.880 0.000 0.616 0.778 ## AltVal_cfa =~ ## AltV1 0.534 0.030 17.772 0.000 0.534 0.602 ## AltV3 0.615 0.039 15.919 0.000 0.615 0.547 ## AltV4 0.589 0.035 16.765 0.000 0.589 0.572 ## AltV5 0.509 0.030 16.973 0.000 0.509 0.579 ## EgoVal_cfa =~ ## EgoV1 1.482 0.072 20.583 0.000 1.482 0.814 ## EgoV2 1.332 0.070 18.920 0.000 1.332 0.724 ## EnviroConcern =~ ## Nep3 0.439 0.036 12.251 0.000 0.642 0.534 ## Nep5 0.373 0.035 10.620 0.000 0.545 0.453 ## Nep9 0.418 0.036 11.579 0.000 0.611 0.495 ## Nep15 0.427 0.035 12.284 0.000 0.623 0.542 ## Anthropocentric =~ ## Nep2 1.064 0.066 16.168 0.000 1.319 0.659 ## Nep8 0.874 0.058 15.056 0.000 1.083 0.579 ## Nep10 0.725 0.062 11.667 0.000 0.898 0.489 ## Nep12 1.080 0.065 16.730 0.000 1.339 0.652 ## Nep14 0.573 0.050 11.372 0.000 0.710 0.444 ## AwareCon_cfa =~ ## AC1 0.403 0.034 11.732 0.000 0.725 0.665 ## AC2 0.411 0.034 11.964 0.000 0.739 0.705 ## AC3 0.364 0.034 10.581 0.000 0.655 0.530 ## AC4 0.358 0.032 11.242 0.000 0.644 0.599 ## AscResp_cfa =~ ## AR1 0.472 0.028 16.801 0.000 0.699 0.727 ## AR2 0.452 0.028 16.293 0.000 0.669 0.687 ## AR3 0.505 0.036 14.066 0.000 0.747 0.566 ## PersNorm_cfa =~ ## PN1 0.773 0.034 22.424 0.000 0.989 0.764 ## PN2 0.801 0.043 18.634 0.000 1.024 0.641 ## PN3 0.721 0.036 19.782 0.000 0.922 0.678 157 ## PN4 0.642 0.031 20.982 0.000 0.821 0.723 ## PN5 0.603 0.030 20.307 0.000 0.771 0.702 ## Intent_cfa =~ ## Int1 0.766 0.029 26.059 0.000 1.157 0.877 ## Int2 0.772 0.030 25.649 0.000 1.167 0.863 ## Int3 0.850 0.032 26.627 0.000 1.284 0.900 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## EnviroConcern ~ ## BioVal_cfa -0.017 0.148 -0.112 0.911 -0.011 -0.011 ## AltVal_cfa 1.082 0.190 5.688 0.000 0.741 0.741 ## EgoVal_cfa -0.208 0.066 -3.161 0.002 -0.142 -0.142 ## Anthropocentric ~ ## BioVal_cfa -0.191 0.117 -1.625 0.104 -0.154 -0.154 ## AltVal_cfa 0.353 0.124 2.855 0.004 0.285 0.285 ## EgoVal_cfa -0.719 0.067 -10.752 0.000 -0.580 -0.580 ## AwareCon_cfa ~ ## Anthropocentrc -0.222 0.061 -3.664 0.000 -0.153 -0.153 ## EnviroConcern 1.035 0.126 8.224 0.000 0.840 0.840 ## AscResp_cfa ~ ## AwareCon_cfa 0.607 0.065 9.334 0.000 0.738 0.738 ## PersNorm_cfa ~ ## AscResp_cfa 0.538 0.046 11.779 0.000 0.623 0.623 ## Intent_cfa ~ ## PersNorm_cfa 0.885 0.057 15.590 0.000 0.750 0.750 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .PN4 ~~ ## .PN5 0.161 0.027 5.928 0.000 0.161 0.262 ## .Nep2 ~~ ## .Nep10 -0.507 0.103 -4.926 0.000 -0.507 -0.210 ## .Nep10 ~~ ## .Nep14 0.390 0.093 4.217 0.000 0.390 0.170 ## .Nep5 ~~ ## .Nep15 0.188 0.040 4.651 0.000 0.188 0.181 ## BioVal_cfa ~~ ## AltVal_cfa 0.805 0.026 30.852 0.000 0.805 0.805 ## EgoVal_cfa 0.098 0.042 2.353 0.019 0.098 0.098 ## AltVal_cfa ~~ ## EgoVal_cfa 0.117 0.046 2.549 0.011 0.117 0.117 ## ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 6.482 0.029 222.897 0.000 6.482 7.349 158 ## .BioV2 6.583 0.023 284.938 0.000 6.583 9.394 ## .BioV3 6.655 0.021 322.812 0.000 6.655 10.643 ## .BioV4 6.462 0.026 247.565 0.000 6.462 8.162 ## .AltV1 6.362 0.029 217.755 0.000 6.362 7.179 ## .AltV3 6.108 0.037 164.955 0.000 6.108 5.438 ## .AltV4 6.292 0.034 185.649 0.000 6.292 6.121 ## .AltV5 6.247 0.029 215.531 0.000 6.247 7.106 ## .EgoV1 4.342 0.060 72.324 0.000 4.342 2.384 ## .EgoV2 4.086 0.061 67.424 0.000 4.086 2.223 ## .Nep3 6.042 0.040 152.582 0.000 6.042 5.030 ## .Nep5 6.075 0.040 153.033 0.000 6.075 5.045 ## .Nep9 5.807 0.041 142.793 0.000 5.807 4.708 ## .Nep15 6.036 0.038 159.386 0.000 6.036 5.255 ## .Nep2 4.615 0.066 69.957 0.000 4.615 2.306 ## .Nep8 3.918 0.062 63.482 0.000 3.918 2.093 ## .Nep10 3.378 0.061 55.795 0.000 3.378 1.839 ## .Nep12 4.867 0.068 71.947 0.000 4.867 2.372 ## .Nep14 2.917 0.053 55.370 0.000 2.917 1.825 ## .AC1 6.180 0.036 171.861 0.000 6.180 5.666 ## .AC2 5.971 0.035 172.775 0.000 5.971 5.696 ## .AC3 5.808 0.041 142.422 0.000 5.808 4.696 ## .AC4 5.633 0.035 158.994 0.000 5.633 5.242 ## .AR1 6.177 0.032 194.879 0.000 6.177 6.425 ## .AR2 6.162 0.032 192.073 0.000 6.162 6.332 ## .AR3 5.571 0.044 127.896 0.000 5.571 4.217 ## .PN1 5.564 0.043 130.341 0.000 5.564 4.297 ## .PN2 5.166 0.053 98.010 0.000 5.166 3.231 ## .PN3 5.376 0.045 119.815 0.000 5.376 3.950 ## .PN4 5.878 0.037 157.000 0.000 5.878 5.176 ## .PN5 6.014 0.036 166.002 0.000 6.014 5.473 ## .Int1 5.518 0.043 126.931 0.000 5.518 4.185 ## .Int2 5.407 0.045 121.276 0.000 5.407 3.998 ## .Int3 5.382 0.047 114.391 0.000 5.382 3.771 ## BioVal_cfa 0.000 0.000 0.000 ## AltVal_cfa 0.000 0.000 0.000 ## EgoVal_cfa 0.000 0.000 0.000 ## .EnviroConcern 0.000 0.000 0.000 ## .Anthropocentrc 0.000 0.000 0.000 ## .AwareCon_cfa 0.000 0.000 0.000 ## .AscResp_cfa 0.000 0.000 0.000 ## .PersNorm_cfa 0.000 0.000 0.000 ## .Intent_cfa 0.000 0.000 0.000 ## ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 0.474 0.025 18.793 0.000 0.474 0.610 159 ## .BioV2 0.225 0.014 16.463 0.000 0.225 0.458 ## .BioV3 0.192 0.011 17.090 0.000 0.192 0.491 ## .BioV4 0.247 0.017 14.972 0.000 0.247 0.394 ## .AltV1 0.501 0.028 17.981 0.000 0.501 0.638 ## .AltV3 0.883 0.047 18.886 0.000 0.883 0.700 ## .AltV4 0.711 0.038 18.506 0.000 0.711 0.672 ## .AltV5 0.514 0.028 18.404 0.000 0.514 0.665 ## .EgoV1 1.119 0.165 6.796 0.000 1.119 0.337 ## .EgoV2 1.605 0.147 10.944 0.000 1.605 0.475 ## .Nep3 1.031 0.055 18.610 0.000 1.031 0.715 ## .Nep5 1.152 0.059 19.415 0.000 1.152 0.795 ## .Nep9 1.148 0.060 19.149 0.000 1.148 0.755 ## .Nep15 0.931 0.051 18.305 0.000 0.931 0.706 ## .Nep2 2.264 0.152 14.865 0.000 2.264 0.565 ## .Nep8 2.332 0.131 17.755 0.000 2.332 0.665 ## .Nep10 2.566 0.144 17.811 0.000 2.566 0.761 ## .Nep12 2.418 0.153 15.836 0.000 2.418 0.574 ## .Nep14 2.050 0.106 19.397 0.000 2.050 0.802 ## .AC1 0.664 0.038 17.350 0.000 0.664 0.558 ## .AC2 0.552 0.034 16.319 0.000 0.552 0.503 ## .AC3 1.101 0.057 19.452 0.000 1.101 0.720 ## .AC4 0.740 0.040 18.563 0.000 0.740 0.641 ## .AR1 0.436 0.030 14.724 0.000 0.436 0.471 ## .AR2 0.499 0.031 16.069 0.000 0.499 0.527 ## .AR3 1.187 0.064 18.673 0.000 1.187 0.680 ## .PN1 0.699 0.043 16.341 0.000 0.699 0.417 ## .PN2 1.507 0.080 18.942 0.000 1.507 0.590 ## .PN3 1.001 0.054 18.378 0.000 1.001 0.541 ## .PN4 0.616 0.036 17.088 0.000 0.616 0.478 ## .PN5 0.613 0.035 17.494 0.000 0.613 0.507 ## .Int1 0.400 0.027 14.614 0.000 0.400 0.230 ## .Int2 0.468 0.030 15.560 0.000 0.468 0.256 ## .Int3 0.386 0.030 12.782 0.000 0.386 0.190 ## BioVal_cfa 1.000 1.000 1.000 ## AltVal_cfa 1.000 1.000 1.000 ## EgoVal_cfa 1.000 1.000 1.000 ## .EnviroConcern 1.000 0.469 0.469 ## .Anthropocentrc 1.000 0.650 0.650 ## .AwareCon_cfa 1.000 0.309 0.309 ## .AscResp_cfa 1.000 0.456 0.456 ## .PersNorm_cfa 1.000 0.611 0.611 ## .Intent_cfa 1.000 0.438 0.438 ## ## R-Square: ## Estimate ## BioV1 0.390 160 ## BioV2 0.542 ## BioV3 0.509 ## BioV4 0.606 ## AltV1 0.362 ## AltV3 0.300 ## AltV4 0.328 ## AltV5 0.335 ## EgoV1 0.663 ## EgoV2 0.525 ## Nep3 0.285 ## Nep5 0.205 ## Nep9 0.245 ## Nep15 0.294 ## Nep2 0.435 ## Nep8 0.335 ## Nep10 0.239 ## Nep12 0.426 ## Nep14 0.198 ## AC1 0.442 ## AC2 0.497 ## AC3 0.280 ## AC4 0.359 ## AR1 0.529 ## AR2 0.473 ## AR3 0.320 ## PN1 0.583 ## PN2 0.410 ## PN3 0.459 ## PN4 0.522 ## PN5 0.493 ## Int1 0.770 ## Int2 0.744 ## Int3 0.810 ## EnviroConcern 0.531 ## Anthropocentrc 0.350 ## AwareCon_cfa 0.691 ## AscResp_cfa 0.544 ## PersNorm_cfa 0.389 ## Intent_cfa 0.562 161 APPENDIX F: VALUE-BELIEF-NORM THEORY SEM MODEL – USA ## lavaan 0.6-11 ended normally after 51 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 126 ## ## Number of observations 900 ## ## Model Test User Model: ## ## Test statistic 2214.262 ## Degrees of freedom 576 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 19154.477 ## Degrees of freedom 630 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.912 ## Tucker-Lewis Index (TLI) 0.903 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -50807.926 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 101867.853 ## Bayesian (BIC) 102472.954 ## Sample-size adjusted Bayesian (BIC) 102072.799 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.056 ## 90 Percent confidence interval - lower 0.054 ## 90 Percent confidence interval - upper 0.059 ## P-value RMSEA <= 0.05 0.000 ## ## Standardized Root Mean Square Residual: ## 162 ## SRMR 0.109 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## BioVal_cfa =~ ## BioV1 1.103 0.039 28.024 0.000 1.103 0.801 ## BioV2 1.096 0.037 29.724 0.000 1.096 0.833 ## BioV3 0.865 0.033 26.134 0.000 0.865 0.763 ## BioV4 1.126 0.043 25.980 0.000 1.126 0.760 ## AltVal_cfa =~ ## AltV1 0.716 0.040 17.842 0.000 0.716 0.594 ## AltV2 1.022 0.043 23.670 0.000 1.022 0.737 ## AltV3 0.715 0.036 20.040 0.000 0.715 0.650 ## AltV4 0.702 0.044 16.055 0.000 0.702 0.540 ## AltV5 0.889 0.037 24.306 0.000 0.889 0.752 ## EgoVal_cfa =~ ## EgoV1 1.345 0.056 24.109 0.000 1.345 0.779 ## EgoV2 1.409 0.052 26.879 0.000 1.409 0.860 ## EgoV3 0.968 0.060 16.009 0.000 0.968 0.540 ## EnviroConcern =~ ## Nep3 0.600 0.033 18.156 0.000 0.895 0.683 ## Nep5 0.720 0.036 19.930 0.000 1.074 0.762 ## Nep13 0.645 0.036 17.844 0.000 0.963 0.670 ## Nep15 0.818 0.040 20.456 0.000 1.221 0.773 ## Anthropocentric =~ ## Nep2 0.867 0.051 17.136 0.000 1.212 0.690 ## Nep8 0.858 0.049 17.433 0.000 1.200 0.706 ## Nep10 0.819 0.052 15.745 0.000 1.145 0.607 ## Nep12 0.848 0.055 15.281 0.000 1.185 0.601 ## Nep14 0.663 0.049 13.634 0.000 0.927 0.529 ## AwareCon_cfa =~ ## AC1 0.682 0.032 21.112 0.000 0.948 0.781 ## AC2 0.643 0.031 20.749 0.000 0.894 0.764 ## AC3 0.489 0.034 14.311 0.000 0.680 0.520 ## AC4 0.514 0.032 15.910 0.000 0.716 0.578 ## AscResp_cfa =~ ## AR1 0.941 0.047 20.016 0.000 1.239 0.747 ## AR2 0.717 0.040 18.046 0.000 0.944 0.655 ## AR3 0.866 0.047 18.358 0.000 1.140 0.689 ## PersNorm_cfa =~ 163 ## PN1 0.917 0.045 20.567 0.000 1.555 0.895 ## PN2 0.922 0.046 20.167 0.000 1.564 0.866 ## PN3 0.934 0.046 20.469 0.000 1.584 0.888 ## PN4 0.691 0.037 18.769 0.000 1.171 0.777 ## PN5 0.706 0.039 18.219 0.000 1.198 0.745 ## Intent_cfa =~ ## Int1 0.978 0.035 28.025 0.000 1.562 0.846 ## Int2 1.063 0.033 32.326 0.000 1.697 0.949 ## Int3 1.057 0.033 32.182 0.000 1.688 0.944 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## EnviroConcern ~ ## BioVal_cfa 1.476 0.130 11.330 0.000 0.989 0.989 ## AltVal_cfa -0.533 0.105 -5.095 0.000 -0.357 -0.357 ## EgoVal_cfa -0.105 0.051 -2.070 0.038 -0.070 -0.070 ## Anthropocentric ~ ## BioVal_cfa 1.146 0.124 9.257 0.000 0.819 0.819 ## AltVal_cfa -0.681 0.112 -6.093 0.000 -0.487 -0.487 ## EgoVal_cfa -0.718 0.067 -10.682 0.000 -0.513 -0.513 ## AwareCon_cfa ~ ## Anthropocentrc -0.094 0.038 -2.449 0.014 -0.094 -0.094 ## EnviroConcern 0.674 0.052 12.893 0.000 0.723 0.723 ## AscResp_cfa ~ ## AwareCon_cfa 0.616 0.049 12.677 0.000 0.651 0.651 ## PersNorm_cfa ~ ## AscResp_cfa 1.040 0.080 12.937 0.000 0.808 0.808 ## Intent_cfa ~ ## PersNorm_cfa 0.734 0.046 15.800 0.000 0.779 0.779 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .PN4 ~~ ## .PN5 0.558 0.044 12.762 0.000 0.558 0.548 ## .AltV1 ~~ ## .AltV3 0.335 0.035 9.562 0.000 0.335 0.412 ## .AR1 ~~ ## .AR3 0.470 0.072 6.507 0.000 0.470 0.356 ## .Nep15 ~~ ## .Nep10 0.410 0.062 6.650 0.000 0.410 0.273 ## BioVal_cfa ~~ ## AltVal_cfa 0.769 0.021 36.278 0.000 0.769 0.769 ## EgoVal_cfa 0.165 0.038 4.294 0.000 0.165 0.165 ## AltVal_cfa ~~ ## EgoVal_cfa 0.139 0.041 3.422 0.001 0.139 0.139 ## 164 ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 5.497 0.046 119.667 0.000 5.497 3.989 ## .BioV2 5.586 0.044 127.259 0.000 5.586 4.242 ## .BioV3 5.954 0.038 157.677 0.000 5.954 5.256 ## .BioV4 5.119 0.049 103.632 0.000 5.119 3.454 ## .AltV1 5.926 0.040 147.344 0.000 5.926 4.911 ## .AltV2 5.171 0.046 111.969 0.000 5.171 3.732 ## .AltV3 6.093 0.037 166.093 0.000 6.093 5.536 ## .AltV4 5.852 0.043 134.947 0.000 5.852 4.498 ## .AltV5 5.737 0.039 145.648 0.000 5.737 4.855 ## .EgoV1 2.621 0.058 45.570 0.000 2.621 1.519 ## .EgoV2 2.511 0.055 45.952 0.000 2.511 1.532 ## .EgoV3 3.949 0.060 66.091 0.000 3.949 2.203 ## .Nep3 5.539 0.044 126.798 0.000 5.539 4.227 ## .Nep5 5.661 0.047 120.437 0.000 5.661 4.015 ## .Nep13 5.377 0.048 112.222 0.000 5.377 3.741 ## .Nep15 5.257 0.053 99.886 0.000 5.257 3.330 ## .Nep2 4.733 0.059 80.825 0.000 4.733 2.694 ## .Nep8 4.651 0.057 82.056 0.000 4.651 2.735 ## .Nep10 4.649 0.063 73.940 0.000 4.649 2.465 ## .Nep12 4.576 0.066 69.569 0.000 4.576 2.319 ## .Nep14 4.493 0.058 76.852 0.000 4.493 2.562 ## .AC1 5.699 0.040 140.898 0.000 5.699 4.697 ## .AC2 5.488 0.039 140.627 0.000 5.488 4.688 ## .AC3 5.261 0.044 120.791 0.000 5.261 4.026 ## .AC4 5.209 0.041 126.276 0.000 5.209 4.209 ## .AR1 4.512 0.055 81.615 0.000 4.512 2.720 ## .AR2 5.179 0.048 107.720 0.000 5.179 3.591 ## .AR3 4.293 0.055 77.866 0.000 4.293 2.596 ## .PN1 4.306 0.058 74.352 0.000 4.306 2.478 ## .PN2 3.636 0.060 60.409 0.000 3.636 2.014 ## .PN3 3.892 0.060 65.411 0.000 3.892 2.180 ## .PN4 4.989 0.050 99.265 0.000 4.989 3.309 ## .PN5 4.953 0.054 92.428 0.000 4.953 3.081 ## .Int1 3.944 0.062 64.095 0.000 3.944 2.137 ## .Int2 3.489 0.060 58.509 0.000 3.489 1.950 ## .Int3 3.350 0.060 56.221 0.000 3.350 1.874 ## BioVal_cfa 0.000 0.000 0.000 ## AltVal_cfa 0.000 0.000 0.000 ## EgoVal_cfa 0.000 0.000 0.000 ## .EnviroConcern 0.000 0.000 0.000 ## .Anthropocentrc 0.000 0.000 0.000 ## .AwareCon_cfa 0.000 0.000 0.000 ## .AscResp_cfa 0.000 0.000 0.000 ## .PersNorm_cfa 0.000 0.000 0.000 165 ## .Intent_cfa 0.000 0.000 0.000 ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 0.681 0.040 17.031 0.000 0.681 0.359 ## .BioV2 0.532 0.033 15.910 0.000 0.532 0.307 ## .BioV3 0.536 0.030 17.961 0.000 0.536 0.417 ## .BioV4 0.927 0.051 18.026 0.000 0.927 0.422 ## .AltV1 0.943 0.050 18.685 0.000 0.943 0.648 ## .AltV2 0.876 0.055 15.999 0.000 0.876 0.456 ## .AltV3 0.700 0.039 17.929 0.000 0.700 0.578 ## .AltV4 1.200 0.062 19.455 0.000 1.200 0.709 ## .AltV5 0.606 0.039 15.499 0.000 0.606 0.434 ## .EgoV1 1.169 0.094 12.402 0.000 1.169 0.393 ## .EgoV2 0.702 0.089 7.863 0.000 0.702 0.261 ## .EgoV3 2.275 0.117 19.375 0.000 2.275 0.708 ## .Nep3 0.916 0.051 18.132 0.000 0.916 0.533 ## .Nep5 0.835 0.051 16.307 0.000 0.835 0.420 ## .Nep13 1.138 0.062 18.348 0.000 1.138 0.551 ## .Nep15 1.001 0.063 15.899 0.000 1.001 0.402 ## .Nep2 1.617 0.099 16.313 0.000 1.617 0.524 ## .Nep8 1.451 0.091 15.853 0.000 1.451 0.502 ## .Nep10 2.246 0.124 18.100 0.000 2.246 0.631 ## .Nep12 2.488 0.137 18.190 0.000 2.488 0.639 ## .Nep14 2.216 0.116 19.147 0.000 2.216 0.720 ## .AC1 0.573 0.040 14.310 0.000 0.573 0.389 ## .AC2 0.571 0.038 15.018 0.000 0.571 0.416 ## .AC3 1.245 0.063 19.627 0.000 1.245 0.729 ## .AC4 1.019 0.053 19.055 0.000 1.019 0.666 ## .AR1 1.215 0.086 14.095 0.000 1.215 0.442 ## .AR2 1.189 0.068 17.465 0.000 1.189 0.572 ## .AR3 1.436 0.092 15.681 0.000 1.436 0.525 ## .PN1 0.601 0.039 15.268 0.000 0.601 0.199 ## .PN2 0.815 0.049 16.793 0.000 0.815 0.250 ## .PN3 0.677 0.043 15.726 0.000 0.677 0.212 ## .PN4 0.901 0.048 18.959 0.000 0.901 0.396 ## .PN5 1.150 0.059 19.336 0.000 1.150 0.445 ## .Int1 0.969 0.052 18.561 0.000 0.969 0.284 ## .Int2 0.320 0.030 10.613 0.000 0.320 0.100 ## .Int3 0.347 0.031 11.341 0.000 0.347 0.109 ## BioVal_cfa 1.000 1.000 1.000 ## AltVal_cfa 1.000 1.000 1.000 ## EgoVal_cfa 1.000 1.000 1.000 ## .EnviroConcern 1.000 0.449 0.449 ## .Anthropocentrc 1.000 0.511 0.511 ## .AwareCon_cfa 1.000 0.517 0.517 ## .AscResp_cfa 1.000 0.577 0.577 166 ## .PersNorm_cfa 1.000 0.348 0.348 ## .Intent_cfa 1.000 0.392 0.392 ## ## R-Square: ## Estimate ## BioV1 0.641 ## BioV2 0.693 ## BioV3 0.583 ## BioV4 0.578 ## AltV1 0.352 ## AltV2 0.544 ## AltV3 0.422 ## AltV4 0.291 ## AltV5 0.566 ## EgoV1 0.607 ## EgoV2 0.739 ## EgoV3 0.292 ## Nep3 0.467 ## Nep5 0.580 ## Nep13 0.449 ## Nep15 0.598 ## Nep2 0.476 ## Nep8 0.498 ## Nep10 0.369 ## Nep12 0.361 ## Nep14 0.280 ## AC1 0.611 ## AC2 0.584 ## AC3 0.271 ## AC4 0.334 ## AR1 0.558 ## AR2 0.428 ## AR3 0.475 ## PN1 0.801 ## PN2 0.750 ## PN3 0.788 ## PN4 0.604 ## PN5 0.555 ## Int1 0.716 ## Int2 0.900 ## Int3 0.891 ## EnviroConcern 0.551 ## Anthropocentrc 0.489 ## AwareCon_cfa 0.483 ## AscResp_cfa 0.423 ## PersNorm_cfa 0.652 167 ## Intent_cfa 0.608 168 APPENDIX G: VBN SEM WITH WILDLIFE VALUE ORIENTATIONS - USA ## lavaan 0.6-11 ended normally after 45 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 123 ## ## Number of observations 900 ## ## Model Test User Model: ## ## Test statistic 2479.781 ## Degrees of freedom 579 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 21071.551 ## Degrees of freedom 630 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.907 ## Tucker-Lewis Index (TLI) 0.899 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -49479.905 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 99205.810 ## Bayesian (BIC) 99796.505 ## Sample-size adjusted Bayesian (BIC) 99405.877 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.060 ## 90 Percent confidence interval - lower 0.058 ## 90 Percent confidence interval - upper 0.063 ## P-value RMSEA <= 0.05 0.000 ## ## Standardized Root Mean Square Residual: ## 169 ## SRMR 0.128 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## BioVal_cfa =~ ## BioV1 1.080 0.040 27.067 0.000 1.080 0.783 ## BioV2 1.104 0.037 29.945 0.000 1.104 0.838 ## BioV3 0.880 0.033 26.728 0.000 0.880 0.777 ## BioV4 1.166 0.043 27.226 0.000 1.166 0.787 ## AltVal_cfa =~ ## AltV1 0.728 0.040 18.127 0.000 0.728 0.603 ## AltV2 1.026 0.043 23.725 0.000 1.026 0.741 ## AltV3 0.724 0.036 20.253 0.000 0.724 0.657 ## AltV4 0.700 0.044 15.949 0.000 0.700 0.538 ## AltV5 0.895 0.037 24.437 0.000 0.895 0.758 ## EgoVal_cfa =~ ## EgoV1 1.400 0.059 23.610 0.000 1.400 0.811 ## EgoV2 1.359 0.056 24.102 0.000 1.359 0.829 ## EgoV3 0.971 0.061 15.934 0.000 0.971 0.542 ## MutualWVO_cfa =~ ## SAB1 0.640 0.036 17.767 0.000 0.913 0.605 ## SAB2 0.729 0.035 20.991 0.000 1.040 0.707 ## SAB3 0.916 0.042 21.593 0.000 1.307 0.726 ## SAB4 0.862 0.035 24.855 0.000 1.230 0.828 ## CaB1 0.930 0.040 23.194 0.000 1.328 0.776 ## CaB2 0.801 0.042 19.274 0.000 1.144 0.653 ## CaB3 0.639 0.032 20.143 0.000 0.912 0.683 ## CaB4 0.797 0.036 22.204 0.000 1.137 0.746 ## CaB5 0.565 0.032 17.617 0.000 0.807 0.603 ## AwareCon_cfa =~ ## AC1 0.752 0.034 22.389 0.000 0.921 0.762 ## AC2 0.711 0.032 21.928 0.000 0.871 0.746 ## AC3 0.553 0.038 14.734 0.000 0.677 0.519 ## AC4 0.601 0.035 17.144 0.000 0.736 0.596 ## AscResp_cfa =~ ## AR1 1.078 0.044 24.422 0.000 1.369 0.827 ## AR2 0.702 0.039 17.976 0.000 0.891 0.619 ## AR3 1.014 0.044 23.075 0.000 1.288 0.780 ## PersNorm_cfa =~ ## PN1 0.986 0.038 25.842 0.000 1.550 0.893 170 ## PN2 0.995 0.040 25.128 0.000 1.564 0.867 ## PN3 1.008 0.039 25.708 0.000 1.584 0.888 ## PN4 0.740 0.033 22.402 0.000 1.163 0.772 ## PN5 0.758 0.035 21.507 0.000 1.191 0.742 ## Intent_cfa =~ ## Int1 0.973 0.035 27.963 0.000 1.560 0.846 ## Int2 1.058 0.033 32.243 0.000 1.695 0.949 ## Int3 1.052 0.033 32.106 0.000 1.686 0.944 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## MutualWVO_cfa ~ ## BioVal_cfa 1.166 0.100 11.640 0.000 0.817 0.817 ## AltVal_cfa -0.250 0.084 -2.991 0.003 -0.175 -0.175 ## EgoVal_cfa 0.116 0.045 2.558 0.011 0.081 0.081 ## AwareCon_cfa ~ ## MutualWVO_cfa 0.496 0.038 13.032 0.000 0.578 0.578 ## AscResp_cfa ~ ## AwareCon_cfa 0.639 0.049 13.168 0.000 0.617 0.617 ## PersNorm_cfa ~ ## AscResp_cfa 0.954 0.061 15.750 0.000 0.771 0.771 ## Intent_cfa ~ ## PersNorm_cfa 0.797 0.045 17.881 0.000 0.782 0.782 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .PN4 ~~ ## .PN5 0.571 0.044 12.905 0.000 0.571 0.553 ## .CaB3 ~~ ## .CaB5 0.525 0.043 12.211 0.000 0.525 0.504 ## .CaB4 0.426 0.042 10.234 0.000 0.426 0.430 ## .CaB4 ~~ ## .CaB5 0.422 0.044 9.608 0.000 0.422 0.389 ## .AltV1 ~~ ## .AltV3 0.320 0.035 9.198 0.000 0.320 0.401 ## BioVal_cfa ~~ ## AltVal_cfa 0.746 0.022 33.700 0.000 0.746 0.746 ## EgoVal_cfa 0.164 0.038 4.268 0.000 0.164 0.164 ## AltVal_cfa ~~ ## EgoVal_cfa 0.152 0.041 3.742 0.000 0.152 0.152 ## ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 5.497 0.046 119.666 0.000 5.497 3.989 ## .BioV2 5.586 0.044 127.259 0.000 5.586 4.242 ## .BioV3 5.954 0.038 157.677 0.000 5.954 5.256 171 ## .BioV4 5.119 0.049 103.632 0.000 5.119 3.454 ## .AltV1 5.926 0.040 147.345 0.000 5.926 4.911 ## .AltV2 5.171 0.046 111.969 0.000 5.171 3.732 ## .AltV3 6.093 0.037 166.093 0.000 6.093 5.536 ## .AltV4 5.852 0.043 134.947 0.000 5.852 4.498 ## .AltV5 5.737 0.039 145.648 0.000 5.737 4.855 ## .EgoV1 2.621 0.058 45.570 0.000 2.621 1.519 ## .EgoV2 2.511 0.055 45.952 0.000 2.511 1.532 ## .EgoV3 3.949 0.060 66.091 0.000 3.949 2.203 ## .SAB1 5.200 0.050 103.314 0.000 5.200 3.444 ## .SAB2 5.147 0.049 104.985 0.000 5.147 3.499 ## .SAB3 4.452 0.060 74.192 0.000 4.452 2.473 ## .SAB4 5.039 0.050 101.772 0.000 5.039 3.392 ## .CaB1 5.062 0.057 88.797 0.000 5.062 2.960 ## .CaB2 4.231 0.058 72.485 0.000 4.231 2.416 ## .CaB3 5.596 0.045 125.712 0.000 5.596 4.190 ## .CaB4 5.257 0.051 103.542 0.000 5.257 3.451 ## .CaB5 5.712 0.045 128.027 0.000 5.712 4.268 ## .AC1 5.699 0.040 141.418 0.000 5.699 4.714 ## .AC2 5.488 0.039 141.122 0.000 5.488 4.704 ## .AC3 5.261 0.043 120.988 0.000 5.261 4.033 ## .AC4 5.209 0.041 126.530 0.000 5.209 4.218 ## .AR1 4.512 0.055 81.731 0.000 4.512 2.724 ## .AR2 5.179 0.048 107.837 0.000 5.179 3.595 ## .AR3 4.293 0.055 77.960 0.000 4.293 2.599 ## .PN1 4.306 0.058 74.451 0.000 4.306 2.482 ## .PN2 3.636 0.060 60.484 0.000 3.636 2.016 ## .PN3 3.892 0.059 65.496 0.000 3.892 2.183 ## .PN4 4.989 0.050 99.364 0.000 4.989 3.312 ## .PN5 4.953 0.054 92.514 0.000 4.953 3.084 ## .Int1 3.944 0.061 64.142 0.000 3.944 2.138 ## .Int2 3.489 0.060 58.563 0.000 3.489 1.952 ## .Int3 3.350 0.060 56.272 0.000 3.350 1.876 ## BioVal_cfa 0.000 0.000 0.000 ## AltVal_cfa 0.000 0.000 0.000 ## EgoVal_cfa 0.000 0.000 0.000 ## .MutualWVO_cfa 0.000 0.000 0.000 ## .AwareCon_cfa 0.000 0.000 0.000 ## .AscResp_cfa 0.000 0.000 0.000 ## .PersNorm_cfa 0.000 0.000 0.000 ## .Intent_cfa 0.000 0.000 0.000 ## ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 0.734 0.042 17.321 0.000 0.734 0.386 ## .BioV2 0.515 0.033 15.389 0.000 0.515 0.297 172 ## .BioV3 0.509 0.029 17.494 0.000 0.509 0.397 ## .BioV4 0.837 0.049 17.237 0.000 0.837 0.381 ## .AltV1 0.926 0.050 18.457 0.000 0.926 0.636 ## .AltV2 0.866 0.055 15.731 0.000 0.866 0.451 ## .AltV3 0.688 0.039 17.675 0.000 0.688 0.568 ## .AltV4 1.203 0.062 19.423 0.000 1.203 0.711 ## .AltV5 0.595 0.039 15.132 0.000 0.595 0.426 ## .EgoV1 1.018 0.112 9.123 0.000 1.018 0.342 ## .EgoV2 0.840 0.103 8.167 0.000 0.840 0.313 ## .EgoV3 2.270 0.118 19.204 0.000 2.270 0.706 ## .SAB1 1.446 0.072 19.960 0.000 1.446 0.634 ## .SAB2 1.081 0.057 19.027 0.000 1.081 0.500 ## .SAB3 1.532 0.082 18.774 0.000 1.532 0.473 ## .SAB4 0.693 0.042 16.419 0.000 0.693 0.314 ## .CaB1 1.163 0.065 17.890 0.000 1.163 0.398 ## .CaB2 1.759 0.090 19.594 0.000 1.759 0.574 ## .CaB3 0.952 0.050 19.171 0.000 0.952 0.534 ## .CaB4 1.028 0.056 18.382 0.000 1.028 0.443 ## .CaB5 1.141 0.058 19.837 0.000 1.141 0.637 ## .AC1 0.613 0.043 14.361 0.000 0.613 0.419 ## .AC2 0.603 0.040 14.970 0.000 0.603 0.443 ## .AC3 1.243 0.064 19.436 0.000 1.243 0.730 ## .AC4 0.984 0.053 18.540 0.000 0.984 0.645 ## .AR1 0.869 0.066 13.191 0.000 0.869 0.317 ## .AR2 1.281 0.068 18.973 0.000 1.281 0.617 ## .AR3 1.071 0.070 15.407 0.000 1.071 0.392 ## .PN1 0.608 0.040 15.369 0.000 0.608 0.202 ## .PN2 0.806 0.048 16.737 0.000 0.806 0.248 ## .PN3 0.670 0.043 15.677 0.000 0.670 0.211 ## .PN4 0.916 0.048 19.027 0.000 0.916 0.404 ## .PN5 1.161 0.060 19.376 0.000 1.161 0.450 ## .Int1 0.970 0.052 18.565 0.000 0.970 0.285 ## .Int2 0.320 0.030 10.642 0.000 0.320 0.100 ## .Int3 0.346 0.031 11.340 0.000 0.346 0.109 ## BioVal_cfa 1.000 1.000 1.000 ## AltVal_cfa 1.000 1.000 1.000 ## EgoVal_cfa 1.000 1.000 1.000 ## .MutualWVO_cfa 1.000 0.491 0.491 ## .AwareCon_cfa 1.000 0.666 0.666 ## .AscResp_cfa 1.000 0.620 0.620 ## .PersNorm_cfa 1.000 0.405 0.405 ## .Intent_cfa 1.000 0.389 0.389 ## ## R-Square: ## Estimate ## BioV1 0.614 173 ## BioV2 0.703 ## BioV3 0.603 ## BioV4 0.619 ## AltV1 0.364 ## AltV2 0.549 ## AltV3 0.432 ## AltV4 0.289 ## AltV5 0.574 ## EgoV1 0.658 ## EgoV2 0.687 ## EgoV3 0.294 ## SAB1 0.366 ## SAB2 0.500 ## SAB3 0.527 ## SAB4 0.686 ## CaB1 0.602 ## CaB2 0.426 ## CaB3 0.466 ## CaB4 0.557 ## CaB5 0.363 ## AC1 0.581 ## AC2 0.557 ## AC3 0.270 ## AC4 0.355 ## AR1 0.683 ## AR2 0.383 ## AR3 0.608 ## PN1 0.798 ## PN2 0.752 ## PN3 0.789 ## PN4 0.596 ## PN5 0.550 ## Int1 0.715 ## Int2 0.900 ## Int3 0.891 ## MutualWVO_cfa 0.509 ## AwareCon_cfa 0.334 ## AscResp_cfa 0.380 ## PersNorm_cfa 0.595 ## Intent_cfa 0.611 174 APPENDIX H: VBN SEM WITH WILDLFIE VALUE ORIENTATIONS - INDIA ## lavaan 0.6-11 ended normally after 58 iterations ## ## Estimator ML ## Optimization method NLMINB ## Number of model parameters 114 ## ## Number of observations 920 ## ## Model Test User Model: ## ## Test statistic 1415.618 ## Degrees of freedom 515 ## P-value (Chi-square) 0.000 ## ## Model Test Baseline Model: ## ## Test statistic 13144.138 ## Degrees of freedom 561 ## P-value 0.000 ## ## User Model versus Baseline Model: ## ## Comparative Fit Index (CFI) 0.928 ## Tucker-Lewis Index (TLI) 0.922 ## ## Loglikelihood and Information Criteria: ## ## Loglikelihood user model (H0) -41737.103 ## Loglikelihood unrestricted model (H1) NA ## ## Akaike (AIC) 83702.205 ## Bayesian (BIC) 84252.184 ## Sample-size adjusted Bayesian (BIC) 83890.133 ## ## Root Mean Square Error of Approximation: ## ## RMSEA 0.044 ## 90 Percent confidence interval - lower 0.041 ## 90 Percent confidence interval - upper 0.046 ## P-value RMSEA <= 0.05 1.000 ## ## Standardized Root Mean Square Residual: ## 175 ## SRMR 0.069 ## ## Parameter Estimates: ## ## Standard errors Standard ## Information Expected ## Information saturated (h1) model Structured ## ## Latent Variables: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## BioVal_cfa =~ ## BioV1 0.546 0.028 19.344 0.000 0.546 0.619 ## BioV2 0.518 0.021 24.344 0.000 0.518 0.739 ## BioV3 0.447 0.019 23.286 0.000 0.447 0.715 ## BioV4 0.617 0.024 26.122 0.000 0.617 0.779 ## AltVal_cfa =~ ## AltV1 0.542 0.030 17.839 0.000 0.542 0.611 ## AltV3 0.597 0.039 15.207 0.000 0.597 0.532 ## AltV4 0.579 0.036 16.236 0.000 0.579 0.563 ## AltV5 0.520 0.030 17.191 0.000 0.520 0.592 ## EgoVal_cfa =~ ## EgoV1 1.605 0.199 8.068 0.000 1.605 0.882 ## EgoV2 1.230 0.157 7.811 0.000 1.230 0.669 ## MutualWVO_cfa =~ ## SAB1 0.447 0.029 15.648 0.000 0.619 0.532 ## SAB2 0.452 0.023 19.699 0.000 0.626 0.659 ## SAB3 0.529 0.029 18.417 0.000 0.732 0.619 ## SAB4 0.518 0.023 22.100 0.000 0.718 0.732 ## CaB1 0.520 0.025 20.417 0.000 0.719 0.681 ## CaB2 0.505 0.035 14.436 0.000 0.700 0.493 ## CaB3 0.514 0.026 20.025 0.000 0.712 0.673 ## CaB4 0.532 0.025 21.193 0.000 0.737 0.708 ## CaB5 0.454 0.025 18.511 0.000 0.628 0.622 ## AwareCon_cfa =~ ## AC1 0.415 0.027 15.516 0.000 0.681 0.624 ## AC2 0.426 0.026 16.311 0.000 0.699 0.667 ## AC3 0.356 0.029 12.277 0.000 0.584 0.473 ## AC4 0.395 0.026 15.111 0.000 0.648 0.604 ## AscResp_cfa =~ ## AR1 0.381 0.029 13.331 0.000 0.691 0.719 ## AR2 0.357 0.028 12.994 0.000 0.649 0.667 ## AR3 0.412 0.034 11.951 0.000 0.747 0.566 ## PersNorm_cfa =~ ## PN1 0.741 0.034 21.966 0.000 0.986 0.761 ## PN2 0.769 0.042 18.386 0.000 1.023 0.640 ## PN3 0.691 0.036 19.429 0.000 0.918 0.675 176 ## PN4 0.619 0.030 20.718 0.000 0.823 0.725 ## PN5 0.582 0.029 20.078 0.000 0.774 0.704 ## Intent_cfa =~ ## Int1 0.763 0.029 26.043 0.000 1.157 0.878 ## Int2 0.769 0.030 25.625 0.000 1.166 0.863 ## Int3 0.847 0.032 26.601 0.000 1.284 0.900 ## ## Regressions: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## MutualWVO_cfa ~ ## BioVal_cfa 0.582 0.109 5.336 0.000 0.420 0.420 ## AltVal_cfa 0.397 0.115 3.440 0.001 0.287 0.287 ## EgoVal_cfa 0.133 0.046 2.907 0.004 0.096 0.096 ## AwareCon_cfa ~ ## MutualWVO_cfa 0.940 0.069 13.722 0.000 0.793 0.793 ## AscResp_cfa ~ ## AwareCon_cfa 0.923 0.093 9.935 0.000 0.835 0.835 ## PersNorm_cfa ~ ## AscResp_cfa 0.483 0.046 10.578 0.000 0.659 0.659 ## Intent_cfa ~ ## PersNorm_cfa 0.858 0.055 15.485 0.000 0.752 0.752 ## ## Covariances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .PN4 ~~ ## .PN5 0.157 0.027 5.828 0.000 0.157 0.257 ## .CaB3 ~~ ## .CaB4 0.135 0.023 5.906 0.000 0.135 0.235 ## BioVal_cfa ~~ ## AltVal_cfa 0.806 0.026 30.946 0.000 0.806 0.806 ## EgoVal_cfa 0.108 0.041 2.629 0.009 0.108 0.108 ## AltVal_cfa ~~ ## EgoVal_cfa 0.122 0.045 2.701 0.007 0.122 0.122 ## ## Intercepts: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 6.482 0.029 222.897 0.000 6.482 7.349 ## .BioV2 6.583 0.023 284.938 0.000 6.583 9.394 ## .BioV3 6.655 0.021 322.812 0.000 6.655 10.643 ## .BioV4 6.462 0.026 247.565 0.000 6.462 8.162 ## .AltV1 6.362 0.029 217.755 0.000 6.362 7.179 ## .AltV3 6.108 0.037 164.955 0.000 6.108 5.438 ## .AltV4 6.292 0.034 185.649 0.000 6.292 6.121 ## .AltV5 6.247 0.029 215.531 0.000 6.247 7.106 ## .EgoV1 4.342 0.060 72.325 0.000 4.342 2.384 ## .EgoV2 4.086 0.061 67.424 0.000 4.086 2.223 177 ## .SAB1 6.082 0.038 158.448 0.000 6.082 5.224 ## .SAB2 6.257 0.031 199.859 0.000 6.257 6.589 ## .SAB3 5.983 0.039 153.499 0.000 5.983 5.061 ## .SAB4 6.270 0.032 194.092 0.000 6.270 6.399 ## .CaB1 6.136 0.035 176.233 0.000 6.136 5.810 ## .CaB2 5.315 0.047 113.550 0.000 5.315 3.744 ## .CaB3 6.000 0.035 171.997 0.000 6.000 5.671 ## .CaB4 6.104 0.034 177.773 0.000 6.104 5.861 ## .CaB5 6.000 0.033 180.234 0.000 6.000 5.942 ## .AC1 6.180 0.036 171.905 0.000 6.180 5.668 ## .AC2 5.971 0.035 172.824 0.000 5.971 5.698 ## .AC3 5.808 0.041 142.445 0.000 5.808 4.696 ## .AC4 5.633 0.035 159.027 0.000 5.633 5.243 ## .AR1 6.177 0.032 194.912 0.000 6.177 6.426 ## .AR2 6.162 0.032 192.101 0.000 6.162 6.333 ## .AR3 5.571 0.044 127.909 0.000 5.571 4.217 ## .PN1 5.564 0.043 130.350 0.000 5.564 4.298 ## .PN2 5.166 0.053 98.015 0.000 5.166 3.231 ## .PN3 5.376 0.045 119.822 0.000 5.376 3.950 ## .PN4 5.878 0.037 157.010 0.000 5.878 5.176 ## .PN5 6.014 0.036 166.012 0.000 6.014 5.473 ## .Int1 5.518 0.043 126.937 0.000 5.518 4.185 ## .Int2 5.407 0.045 121.282 0.000 5.407 3.999 ## .Int3 5.382 0.047 114.397 0.000 5.382 3.772 ## BioVal_cfa 0.000 0.000 0.000 ## AltVal_cfa 0.000 0.000 0.000 ## EgoVal_cfa 0.000 0.000 0.000 ## .MutualWVO_cfa 0.000 0.000 0.000 ## .AwareCon_cfa 0.000 0.000 0.000 ## .AscResp_cfa 0.000 0.000 0.000 ## .PersNorm_cfa 0.000 0.000 0.000 ## .Intent_cfa 0.000 0.000 0.000 ## ## Variances: ## Estimate Std.Err z-value P(>|z|) Std.lv Std.all ## .BioV1 0.480 0.025 19.046 0.000 0.480 0.617 ## .BioV2 0.223 0.013 16.711 0.000 0.223 0.453 ## .BioV3 0.191 0.011 17.351 0.000 0.191 0.489 ## .BioV4 0.247 0.016 15.377 0.000 0.247 0.394 ## .AltV1 0.492 0.028 17.468 0.000 0.492 0.626 ## .AltV3 0.905 0.048 18.870 0.000 0.905 0.717 ## .AltV4 0.722 0.039 18.390 0.000 0.722 0.683 ## .AltV5 0.502 0.028 17.871 0.000 0.502 0.650 ## .EgoV1 0.739 0.622 1.188 0.235 0.739 0.223 ## .EgoV2 1.865 0.375 4.977 0.000 1.865 0.552 ## .SAB1 0.972 0.048 20.370 0.000 0.972 0.717 178 ## .SAB2 0.510 0.026 19.341 0.000 0.510 0.565 ## .SAB3 0.861 0.044 19.742 0.000 0.861 0.616 ## .SAB4 0.445 0.024 18.266 0.000 0.445 0.464 ## .CaB1 0.598 0.031 19.071 0.000 0.598 0.536 ## .CaB2 1.526 0.074 20.572 0.000 1.526 0.757 ## .CaB3 0.613 0.032 18.976 0.000 0.613 0.548 ## .CaB4 0.542 0.029 18.517 0.000 0.542 0.499 ## .CaB5 0.625 0.032 19.716 0.000 0.625 0.613 ## .AC1 0.726 0.039 18.413 0.000 0.726 0.611 ## .AC2 0.610 0.035 17.618 0.000 0.610 0.555 ## .AC3 1.188 0.059 20.095 0.000 1.188 0.777 ## .AC4 0.734 0.039 18.728 0.000 0.734 0.636 ## .AR1 0.447 0.029 15.643 0.000 0.447 0.484 ## .AR2 0.526 0.031 17.132 0.000 0.526 0.556 ## .AR3 1.187 0.063 18.943 0.000 1.187 0.680 ## .PN1 0.704 0.043 16.495 0.000 0.704 0.420 ## .PN2 1.510 0.079 18.995 0.000 1.510 0.591 ## .PN3 1.009 0.055 18.477 0.000 1.009 0.545 ## .PN4 0.612 0.036 17.100 0.000 0.612 0.474 ## .PN5 0.609 0.035 17.505 0.000 0.609 0.504 ## .Int1 0.400 0.027 14.612 0.000 0.400 0.230 ## .Int2 0.468 0.030 15.577 0.000 0.468 0.256 ## .Int3 0.387 0.030 12.805 0.000 0.387 0.190 ## BioVal_cfa 1.000 1.000 1.000 ## AltVal_cfa 1.000 1.000 1.000 ## EgoVal_cfa 1.000 1.000 1.000 ## .MutualWVO_cfa 1.000 0.522 0.522 ## .AwareCon_cfa 1.000 0.371 0.371 ## .AscResp_cfa 1.000 0.304 0.304 ## .PersNorm_cfa 1.000 0.566 0.566 ## .Intent_cfa 1.000 0.435 0.435 ## ## R-Square: ## Estimate ## BioV1 0.383 ## BioV2 0.547 ## BioV3 0.511 ## BioV4 0.606 ## AltV1 0.374 ## AltV3 0.283 ## AltV4 0.317 ## AltV5 0.350 ## EgoV1 0.777 ## EgoV2 0.448 ## SAB1 0.283 ## SAB2 0.435 179 ## SAB3 0.384 ## SAB4 0.536 ## CaB1 0.464 ## CaB2 0.243 ## CaB3 0.452 ## CaB4 0.501 ## CaB5 0.387 ## AC1 0.389 ## AC2 0.445 ## AC3 0.223 ## AC4 0.364 ## AR1 0.516 ## AR2 0.444 ## AR3 0.320 ## PN1 0.580 ## PN2 0.409 ## PN3 0.455 ## PN4 0.526 ## PN5 0.496 ## Int1 0.770 ## Int2 0.744 ## Int3 0.810 ## MutualWVO_cfa 0.478 ## AwareCon_cfa 0.629 ## AscResp_cfa 0.696 ## PersNorm_cfa 0.434 ## Intent_cfa 0.565 180 APPENDIX I: SURVEY INSTRUMENT Informed Consent “Researchers at Michigan State University are studying wildlife crime. Results of this survey will help researchers better understand how people think and feel about this issue. Your participation is completely anonymous and voluntary. Your responses are not tied to any personally identifying information. You can withdraw from the survey at any time or refuse to answer any question without penalty. Only the researchers associated with this study will have access to the survey data. The survey data will be maintained on a secure electronic device and only the lead researcher will have access to that device. By completing this questionnaire, you are acknowledging that any data collected in this study can be used in research, and for related publications and presentations. We estimate that this survey should take no longer than 20 minutes to complete. You must be 18 years of age or older to participate in this study. For further information about this study, please contact Apoorva Joshi at joshiap1@msu.edu or MSU’s IRB at irb@msu.edu • Are you 18 years of age or older? (Yes/No) • Do you consent to participate in this survey? (Yes/No) Introduction The term ‘wildlife crime’ includes illegal acts such as poaching (illegal hunting of wildlife), as well as the trafficking, smuggling, and illegal trade in wild plants and animals – both living and dead. This illegal trade in exotic live animals and wildlife parts and products occurs across international and local borders. Globally, wildlife crime is regarded as one of the largest organized criminal operations behind the trafficking of drugs, weapons, and humans. In the following sections, please pay close attention to each question and respond as honestly as possible. Questionnaire Order will be randomized within each question block – i.e. each construct’s set of items. In the following section, you will be asked about your thoughts about the natural environment. Please answer all questions in this survey as honestly as you can. For each of the following questions, please select from the options below that most closely resembles your honest answer. New Ecological Paradigm Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • Nep1: “We are approaching the limit of the number of people the earth can support.” • Nep2: “Humans have the right to modify the natural environment to suit their needs.” 181 • Nep3: “When humans interfere with nature, it often produces disastrous consequences.” • Nep4: “Human ingenuity will ensure that we do not make the earth unlivable.” • Nep5: “Humans are severely abusing the environment.” • Nep6: “The earth has plenty of natural resources if we just learn how to develop them.” • Nep7: “Plants and animals have as much right as humans to exist.” • Nep8: “The balance of nature is strong enough to cope with the impacts of modern industrial nations.” • Nep9: “Despite our special abilities, humans are still subject to the laws of nature.” • Nep10: “The so-called ‘ecological crisis’ facing humankind has been greatly exaggerated.” • Nep11: “The earth is like a spaceship with very limited room and resources.” • Nep12: “Humans were meant to rule over the rest of nature.” • Nep13: “The balance of nature is very delicate and easily upset.” • Nep14: “Humans will eventually learn enough about how nature works to be able to control it.” • Nep15: “If things continue on their present course, we will soon experience a major ecological catastrophe.” In the following section, you will be asked about your thoughts on the relationship between people and wildlife. Wildlife Value Orientations Please indicate the extent to which you agree or disagree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • AUB1: “Humans should manage fish and wildlife populations so that humans benefit.” • AUB2: “The needs of humans should take priority over fish and wildlife protection.” • AUB3: “It is acceptable for people to kill wildlife if they think it poses a threat to their life.” • AUB4: “It is acceptable for people to kill wildlife if they think it poses a threat to their property.” • AUB5: “Fish and wildlife are on earth primarily for people to use.” • HuB1: “We should strive for a world where there’s an abundance of fish and wildlife for hunting and fishing.” • HuB2: “Hunting is cruel and inhumane to the animals.” (R) • HuB3: “Hunting does not respect the lives of animals.” (R) • HuB4: “People who want to hunt should be provided the opportunity to do so.” • SAB1: “We should strive for a world where humans and fish and wildlife can live side by side without fear.” • SAB2: “I view all living things as part of one big family.” • SAB3: “Animals should have rights similar to the rights of humans.” 182 • SAB4: “Wildlife are like my family and I want to protect them.” • CaB1: “I care about animals as much as I do other people.” • CaB2: “It would be more rewarding to me to help animals rather than people.” • CaB3: “I take great comfort in the relationships I have with animals.” • CaB4: “I feel a strong emotional bond with animals.” • CaB5: “I value the sense of companionship I receive from animals.” The following sections include questions about the problems associated with wildlife crime and who you think is responsible for dealing with them. Awareness of Consequences Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • AC1: “Wildlife crimes cause biodiversity loss, exhaustion of natural resources, and species extinction.” • AC2: “Wildlife crimes generate environmental impacts on the neighboring areas and wider environment.” • AC3: “Wildlife crimes such as illegal international trade in exotic species can cause the spread of deadly zoonotic pathogens and viruses.” • AC4: ““Conservation organizations help to curb wildlife crime and mitigate its impacts.” Ascription of Responsibility Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • AR1: “Every citizen must take responsibility for mitigating the environmental, economic, and health issues linked to wildlife crimes by donating money to conservation organizations.” • AR2: “The authorities are responsible for mitigating the environmental, economic, and health issues linked to wildlife crimes by financially supporting conservation organizations.” • AR3: “I am responsible for mitigating the environmental, economic, and health issues linked to wildlife crimes by donating money to conservation organizations.” In the following few sections, you will be asked about your thoughts on donating money to conservation organizations to help reduce wildlife crime. Please read questions carefully. Attitude For each of the following questions, please select the option that most closely resembles your honest answer. 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree 183 • “I think the idea of donating money to conservation organizations to help reduce wildlife crime is ________.” o Att1: Very positive o Att2: Very responsible o Att3: Very intelligent o Att4: Very useful o Att5: Very ecologically helpful Subjective Norms Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • SN1: “Most people who are important to me think that one should donate money to conservation organizations to help reduce wildlife crime.” • SN2: “Most people who are important to me expect that I will donate money to conservation organizations to help reduce wildlife crime.” • SN3: “Those people whose opinions I value would donate money to conservation organizations to help reduce wildlife crime.” Personal Norms Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • PN1: “I feel I ought to donate money to conservation organizations to help reduce wildlife crime.” • PN2: “I would feel guilty if I did not donate money to conservation organizations to help reduce wildlife crime.” • PN3: “I feel morally obligated to donate money to conservation organizations to help reduce wildlife crime regardless of what others are doing.” • PN4: “I feel that donating money to conservation organizations to help reduce wildlife crime is the right thing to do.” • PN5: “I would feel good about myself if I donated money to conservation organizations to help reduce wildlife crime.” Perceived Behavioral Control Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • PBC1: “I am confident that if I want, I can donate money to conservation organizations to help reduce wildlife crime.” • PBC2: “I have sufficient resources to donate money to conservation organizations to help reduce wildlife crime.” • PBC3: “I have enough time to donate money to conservation organizations to help reduce wildlife crime.” 184 • PBC4: “I have enough opportunities to donate money to conservation organizations to help reduce wildlife crime.” • PBC5: “Donating money to conservation organizations to help reduce wildlife crime is completely up to me.” • PBC6: “If I donated money to conservation organizations, it would help them to reduce wildlife crime.” Intention Please indicate the extent to which you agree with each statement below: 1 = Strongly disagree; 2 = Disagree; 3 = Somewhat disagree; 4 = Neither agree nor disagree; 5 = Somewhat agree; 6 = Agree; 7 = Strongly agree • Int1: “I am willing to donate money to conservation organizations within the next month to help reduce wildlife crime.” • Int2: “I plan to donate money to conservation organizations within the next month to help reduce wildlife crime.” • Int3: “I will definitely donate money to conservation organizations within the next month to help reduce wildlife crime.” In the following section, you will be asked about what you consider generally important in your life. Values How important are each of the following to you as guiding principles in your life? 1 = Not at all important; 2 = Low importance; 3 = Slightly important; 4 = Neutral; 5 = Moderately important; 6 = Very important; 7 = Extremely important • How important is it to you ________? o Bio1: To prevent environmental pollution. o Bio2: To protect the environment. o Bio3: To respect nature. o Bio4: To be in unity with nature. • How important is it to you ________? o Alt1: That every person has equal opportunities. o Alt2: To take care of those who are worse off. o Alt3: That every person is treated justly. o Alt4: That there is no war or conflict. o Alt5: To be helpful to others. • How important is it to you ________? o Ego1: To have control over others’ actions. o Ego2: To have authority over others. o Ego3: To be influential. o Ego4: To have money and possessions. o Ego5: To work hard and be ambitious. 185 In this final section, you will be asked for general information about yourself. Please remember that none of your responses in this survey can directly identify you since all responses are anonymous. Other Questions and Demographic Module • Gen: What gender do you identify as? Please select one of the options below: o Female o Male o Non-binary o Transgender o Other o Prefer not to answer • Age: What is your age? o 18 – 29 years old o 30 – 45 years old o 46 – 60 years old o 61 – 75 years old o More than 75 years old • Edu: What is the highest degree or level of education you have completed? o No formal education o (For India – SSC/ICSE/CBSE = Std. 10 equivalent) o High School Diploma (For India - HSC/Grade 12 equivalent) o Vocational/technical/trade school (For India – Professional diploma) o Bachelor's Degree o Master's Degree o Ph.D. or higher o Other (blank to include participant’s response) o Prefer not to say • Employ: What is your current employment status? o Employed Full-Time/Self-employed o Employed Part-Time o Interning o Unemployed – Looking for work o Unemployed – Not looking for work o Retired o Other o Prefer not to say • Income: What is your total annual income? o Less than $20,000 - India: Less than ₹2,50,000 o $20,001 to $40,000 - India: ₹2,50,001 to ₹5,00,000 o $40,001 to $60,000 - India: ₹5,00,001 to ₹7,50,000 o $60,001 to $80,000 - India: ₹7,50,001 to ₹10,00,000 o $80,001 to $100,000 - India: ₹10,00,001 to ₹12,50,000 o $100,001 or over - India: ₹12,50,001 or more • Polit: Generally speaking, how would you describe your political viewpoint? 186 o Very conservative o Slightly conservative o Neither conservative nor liberal o Slightly liberal o Very liberal Conclusion Thank you for participating in this survey. Your responses have been successfully recorded. For questions or concerns related to this study, please contact Apoorva Joshi at joshiap1@msu.edu 187 APPENDIX J: CROSS-NATIONAL DIFFERENCES IN RESPONSES TO ITEMS Attitude I think the idea of donating money to conservation organizations to help reduce wildlife crime is _______ Very ecologically helpful 190 326 Very useful 183 337 Very intelligent 178 272 Attitude items Very responsible 191 317 Very positive 225 345 0 50 100 150 200 250 300 350 400 Frequency of "Strongly agree" responses India USA Figure 23 - Number of respondents who strongly agreed with attitude items Subjective Norms SN1: Most people who are important to me think that one should donate money to conservation organizatons to help reduce wildlife crime Strongly agree 220 69 Agree 319 126 Somewhat agree 217 125 Neither 106 344 Somewhat disagree 37 78 Disagree 16 80 Strongly disagree 5 70 0 50 100 150 200 250 300 350 400 India USA Figure 24 - Cross-national distribution of responses to item SN1 188 SN2: Most people who are important to me expect that I will donate money to conservation organizatons to help reduce wildlife crime Strongly agree 192 62 Agree 322 95 Somewhat agree 194 112 Neither 137 294 Somewhat disagree 39 100 Disagree 21 120 Strongly disagree 15 117 0 50 100 150 200 250 300 350 India USA Figure 25 - Cross-national distribution of responses to item SN2 SN3: Those people whose opinions I value would donate money to conservation organizations to help reduce wildlife crime Strongly agree 86 238 Agree 143 349 Somewhat agree 181197 Neither 99 342 Somewhat disagree 16 43 Disagree 15 54 Strongly disagree 6 51 0 50 100 150 200 250 300 350 400 India USA Figure 26 - Cross-national distribution of responses to item SN3 189 Perceived Behavioral Control PBC1: I am confident that if I want, I can donate money to conservation organizations to help reduce wildlife crime Strongly agree 116 264 Agree 236 319 Somewhat agree 192206 Neither 82 172 Somewhat disagree 35 54 Disagree 19 48 Strongly disagree 9 68 0 50 100 150 200 250 300 350 India USA Figure 27 - Cross-national distribution of responses to item PBC1 PBC2: I have sufficient resources to be able to donate money conservation organizations to help reduce wildlife crime Strongly agree 176 60 Agree 237 87 Somewhat agree 194 102 Neither 128 157 Somewhat disagree 89 118 Disagree 80 152 Strongly disagree 16 224 0 50 100 150 200 250 India USA Figure 28 - Cross-national distribution of responses to item PBC2 190 PBC3: I have enough time to donate money to conservation organizations to help reduce wildlife crime Strongly agree 182 84 Agree 253 153 Somewhat agree 229 157 Neither 127 259 Somewhat disagree 67 79 Disagree 49 83 Strongly disagree 13 85 0 50 100 150 200 250 300 India USA Figure 29 - Cross-national distribution of responses to item PBC3 PBC4: I have enough opportunities to donate money to conservation organizations to help reduce wildlife crime Strongly agree 181 83 Agree 266 137 Somewhat agree 198 150 Neither 112 215 Somewhat disagree 96 108 Disagree 49 107 Strongly disagree 18 100 0 50 100 150 200 250 300 India USA Figure 30 - Cross-national distribution of responses to item PBC4 191 PBC6: If I donated money to conservation organizations, it would help them to reduce wildlife crime Strongly agree 289 117 Agree 319 171 Somewhat agree 206 242 Neither 67 257 Somewhat disagree 19 35 Disagree 18 33 Strongly disagree 2 45 0 50 100 150 200 250 300 350 India USA Figure 31 - Cross-national distribution of responses to item PBC6 Intentions to Donate Money to Conservation Organizations Int1: I am willing to donate money to conservation organizations within the next month to help reduce wildlife crime Strongly agree 230 75 Agree 304 128 Somewhat agree 212 164 Neither 106 217 Somewhat disagree 32 68 Disagree 28 117 Strongly disagree 7 131 0 50 100 150 200 250 300 350 India USA Figure 32 - Cross-national distribution of responses to item Int1 192 Int3: I will definitely donate money to conservation organizations within the next month to help reduce wildlife crime Strongly agree 240 53 Agree 251 68 Somewhat agree 193 91 Neither 145 249 Somewhat disagree 48 95 Disagree 34 156 Strongly disagree 8 188 0 50 100 150 200 250 300 India USA Figure 33 - Cross-national distribution of responses to item Int3 Biospheric Values Biospheric Values: How important is it to you to ____? Be in unity with nature 546 189 Biospheric values Respect nature 656 344 Protect the environment 618 257 Prevent environmental pollution 587 235 0 100 200 300 400 500 600 700 Frequency of "Extremely important" responses India USA Figure 34 - Cross-national distribution of responses to biospheric value items 193 Altruistic Values Altruistic Values: How important are each of the following to you? Being helpful to others 418 263 Altruistic values Having no war or conflict 502 363 Treating every person justly 429 406 Taking care of the less fortunate 224 157 Equal opportunities for all 496 357 0 100 200 300 400 500 600 Frequencies of 'Extremely important' responses India USA Figure 35 - Cross-national distribution of responses to altruistic value items Egoistic Values EgoV1: How important is it to you to have control over others' actions? Extremely important 113 35 Very important 174 35 Moderately important 162 58 Neutral 218 175 Slightly important 66 45 Low importance 93 227 Not at all important 94 325 0 50 100 150 200 250 300 350 India USA Figure 36 - Cross-national distribution of responses to item EgoV1 194 EgoV2: How important is it to you to have authority over others? Extremely important 100 21 Very important 126 31 Moderately important 158 52 Neutral 244 191 Slightly important 74 47 Low importance 98 204 Not at all important 119 354 0 50 100 150 200 250 300 350 400 India USA Figure 37 - Cross-national distribution of responses to item EgoV2 EgoV3: How important is it to you to be influential? Extremely important 191 79 Very important 291 109 Moderately important 224 163 Neutral 134 235 Slightly important 41 74 Low importance 31 130 Not at all important 8 110 0 50 100 150 200 250 300 350 India USA Figure 38 - Cross-national distribution of responses to item EgoV3 195 Mutualistic Wildlife Value Orientations (example items) CaB5: I value the sense of companionship I receive from animals Strongly agree 329 321 Agree 355 248 Somewhat agree 172 174 Neither 49 105 Somewhat disagree 5 24 Disagree 6 16 Strongly disagree 4 12 0 50 100 150 200 250 300 350 400 India USA Figure 39 - Cross-national distribution of responses to CaB5 SAB3: Animals should have rights similar to the rights of humans Strongly agree 407 144 Agree 247 137 Somewhat agree 162 190 Neither 65 165 Somewhat disagree 29 103 Disagree 6 97 Strongly disagree 4 64 0 50 100 150 200 250 300 350 400 450 India USA Figure 40 - Cross-national distribution of responses to SAB3 196 Awareness of Consequences AC1: Wildlife crimes cause biodiversity loss, exhaustion of natural reousources, and species extinction Strongly agree 457 273 Agree 294 294 Somewhat agree 93 175 Neither 47 124 Somewhat disagree 15 22 Disagree 11 5 Strongly disagree 3 7 0 50 100 150 200 250 300 350 400 450 500 India USA Figure 41 - Cross-national distribution of responses to item AC1 AC2: Wildlife crimes generate environmental impacts on areas in which they occur and on surrounding communities Strongly agree 319 191 Agree 371 296 Somewhat agree 155 222 Neither 48 160 Somewhat disagree 15 19 Disagree 10 7 Strongly disagree 2 5 0 50 100 150 200 250 300 350 400 India USA Figure 42 - Cross-national distribution of responses to item AC2 197 AC3: Wildlife crimes such as illegal international trade in exotic species causes the spread of deadly viruses and pathogens from animals to humans Strongly agree 185 319 Agree 232 303 Somewhat agree 173 204 Neither 85 223 Somewhat disagree 16 31 Disagree 15 18 Strongly disagree 79 0 50 100 150 200 250 300 350 India USA Figure 43 - Cross-national distribution of responses to item AC3 AC4: Conservation organizations help to curb wildlife crime and mitigate its impacts Strongly agree 200 125 Agree 356 279 Somewhat agree 225 251 Neither 117 188 Somewhat disagree 11 29 Disagree 9 17 Strongly disagree 2 11 0 50 100 150 200 250 300 350 400 India USA Figure 44 - Cross-national distribution of responses to item AC4 198 Ascription of Responsibility Who should take responsibility for mitigating the environmental, economic, and health issues linked to wildlife crimes? 450 411 399 400 350 350 332 314 300 246 250 226 184 200 155 141 150 126 95 100 50 0 USA India USA India Strongly agree Agree Every citizen Authorities Myself Figure 45 - Cross-national distribution of responses to ascription of responsibility items Personal Norms PN1 I feel I ought to donate money to conservation organizations to help reduce wildlife crime. Strongly agree 229 95 Agree 333 134 Somewhat agree 192 214 Neither agree nor disagree 101 219 Somewhat disagree 34 77 Disagree 21 69 Strongly disagree 10 91 0 50 100 150 200 250 300 350 India USA Figure 46 - Cross-national distribution of responses to item PN1 199 PN2 I would feel guilty if I did not donate money to conservation organizations to help reduce wildlife crime. Strongly agree 74 203 Agree 79 263 Somewhat agree 127 184 Neither agree nor disagree 135 206 Somewhat disagree 60 123 Disagree 39 161 Strongly disagree 36 130 0 50 100 150 200 250 300 India USA Figure 47 - Cross-national distribution of responses to item PN2 PN3: I feel morally obligated to donate money to conservation organizations to help reduce wildlife crime regardless of what others are doing. Strongly agree 77 200 Agree 10 299 Somewhat agree 144 199 Neither agree nor disagree 140 222 Somewhat disagree 44 108 Disagree 27 133 Strongly disagree 11 106 0 50 100 150 200 250 300 350 India USA Figure 48 - Cross-national distribution of responses to item PN3 200 PN4: I feel that donating money to conservation organizations to help reduce wildlife crime is the right thing to do. Strongly agree 318 144 Agree 322 224 Somewhat agree 181 217 Neither agree nor disagree 68 218 Somewhat disagree 16 30 Disagree 10 24 Strongly disagree 5 43 0 50 100 150 200 250 300 350 India USA Figure 49 - Cross-national distribution of responses to item PN4 PN5: I would feel good about myself if I donated money to conservation organizations to help reduce wildlife crime. Strongly agree 367 157 Agree 324 221 Somewhat agree 147 204 Neither agree nor disagree 59 196 Somewhat disagree 10 38 Disagree 6 31 Strongly disagree 7 53 0 50 100 150 200 250 300 350 400 India USA Figure 50 - Cross-national distribution of responses to item PN5 201