Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigating the role of cause-and-effect simulation design in persuading online consumers to go green Tangwaragorn, Pattharin 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_september_tangwaragorn_pattharin.pdf [ 2.57MB ]
Metadata
JSON: 24-1.0132742.json
JSON-LD: 24-1.0132742-ld.json
RDF/XML (Pretty): 24-1.0132742-rdf.xml
RDF/JSON: 24-1.0132742-rdf.json
Turtle: 24-1.0132742-turtle.txt
N-Triples: 24-1.0132742-rdf-ntriples.txt
Original Record: 24-1.0132742-source.json
Full Text
24-1.0132742-fulltext.txt
Citation
24-1.0132742.ris

Full Text

INVESTIGATING THE ROLE OF CAUSE-AND-EFFECT SIMULATION DESIGN IN PERSUADING ONLINE CONSUMERS TO GO GREEN by  Pattharin Tangwaragorn  B.B.A., Chulalongkorn University, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Business Administration)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2015  © Pattharin Tangwaragorn, 2015 ii Abstract This study examines the role of technology in motivating online consumers to purchase green products. The cause-and-effect simulation proposed by Fogg (2002) and the construal-level theory (CLT, Liberman and Trope 1998) are employed to develop two website designs: 1) the low-level and 2) the high-level cause-and-effect simulation. Both simulation designs show the relationship between consumers’ decision-making on the product attributes, the cause, and its impact on resources (e.g., energy) the product consumes, the effect. A recommendation agent (RA) is used to reflect the cause part of both designs. The main difference between the two designs is the effect part developed based on CLT. Specifically, the low-level simulation presents a more concrete, short-term effect, the utility cost per load, and the high-level simulation provides a more abstract, long-term effect, the 10-year utility cost. We compare these two designs against two control conditions—1) the no simulation which does not provide the RA and the utility cost and 2) the partial simulation which has only the RA. An online experiment with 79 participants was conducted to evaluate the effectiveness of the simulation design on the green selection. Specifically, we assess whether the simulation design could persuade people to pay more for green products. The experimental results show both low-level and high-level simulations successfully motivate participants to choose greener products than the no simulation and the partial simulation. Moreover, the results suggest both full designs persuade participants to go green by enhancing the desirability consideration associated with the outcome resulting from the green purchase. This consideration was found to influence self-efficacy which leads to greener choices. Self-efficacy iii was also found to have a greater impact on the green product selection than participants’ attitudes. This is evident by the fact that participants generally have a pre-existing positive intention to buy green products. Thus, the role of the cause-and-effect simulation design is not so much to change people’s attitudes, but rather to reinforce those positive attitudes and thus help participants to abide by their good intentions. In other words, it helps increase self-efficacy that would be the key to promote the green purchase.iv Preface This thesis is an original intellectual work of the author, P. Tangwaragorn, in consultation with members of the supervisory committee. The online experiment described in Chapter 1 – 6 as well as Appendix C – D received approval from the Behavioral Research Ethics Board of the University of British Columbia, certificate number: H15-00948. The online experiment for the pilot study explained in Appendix A was covered by UBC certificate number:  H14-03068. v Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ...........................................................................................................................v List of Tables ................................................................................................................................ xi List of Figures ............................................................................................................................. xiii Acknowledgements .................................................................................................................... xvi Dedication .................................................................................................................................. xvii Chapter 1: Introduction ................................................................................................................1 1.1 Research Questions for this Study .................................................................................. 8 Chapter 2: Literature Review .......................................................................................................9 2.1 Pro-environmental Behaviors ......................................................................................... 9 2.2 Persuasion ..................................................................................................................... 10 2.3 Persuasive Technology ................................................................................................. 12 2.4 Simulation ..................................................................................................................... 15 vi 2.4.1 Instructional Simulation and Persuasion ................................................................... 17 2.4.2 Mental Simulation ..................................................................................................... 20 2.4.2.1 Process- and Outcome-Focused Simulation ..................................................... 21 2.4.2.2 Process and Outcome Focus in Related Literature ........................................... 24 2.5 Construal-Level Theory (CLT) ..................................................................................... 28 2.5.1 CLT and Preferences, Evaluation, and Judgment on Choices .................................. 31 2.5.2 CLT and Attitudes, Self-Control, and Intention ....................................................... 33 2.5.3 CLT and Persuasion .................................................................................................. 34 2.6 Framing Effect .............................................................................................................. 38 Chapter 3: Theoretical Framework ...........................................................................................42 3.1 Value-Belief-Norm Theory (VBN)............................................................................... 42 3.2 Theory of Planned Behavior ......................................................................................... 45 3.3 Summary of Literature Foundational to this Study....................................................... 48 Chapter 4: Hypotheses Development .........................................................................................50 4.1 The Impact of Cause-and-Effect Simulation on Desirability Consideration ................ 53 vii 4.2 The Impact of Desirability Consideration on Attitudes toward Purchasing a Green Product and Self-efficacy .......................................................................................................... 55 4.3 Attitudes, Perceived Behavioral Control, Intention, and Behavior .............................. 60 4.4 The Impact of Cause-and-Effect Simulation on a Green Product Choice .................... 60 Chapter 5: Research Method ......................................................................................................63 5.1 Task Product ................................................................................................................. 64 5.2 Experimental Website Design....................................................................................... 69 5.3 Cause-and-Effect Simulation Manipulation (Condition 3 and Condition 4) ................ 74 5.4 Experimental Procedure ................................................................................................ 77 5.4.1 Self-Report Measures................................................................................................ 81 5.4.2 Objective Measures ................................................................................................... 86 5.4.3 Manipulation Check Measurement ........................................................................... 87 Chapter 6: Data Analysis ............................................................................................................90 6.1 Participant Background Information ............................................................................. 90 6.2 Time Spent on the Experimental Website and the Entire Study, Involvement, and Website Evaluation ................................................................................................................... 91 viii 6.3 Manipulation Check ...................................................................................................... 93 6.4 The Effect of Cause-and-Effect Simulation Design on Intention to Purchase a Green Product ...................................................................................................................................... 94 6.5 The Effect of Cause-and-Effect Simulation Design on Persuasion .............................. 99 6.6 The Effect of Cause-and-Effect Simulation on Desirability Consideration ............... 113 6.7 Desirability Consideration, Attitudes toward Purchasing a Green Product, Self-efficacy, and Intention to Purchase a Green Product ............................................................................. 115 6.8 Additional Evidence of the Awareness of Cause-and-Effect Relationship, Feasibility and Desirability Trade-off, and Environmental Concern .............................................................. 120 Chapter 7: Conclusion ...............................................................................................................124 7.1 Discussion ................................................................................................................... 124 7.2 Summary ..................................................................................................................... 128 7.3 Contributions............................................................................................................... 130 7.4 Limitations .................................................................................................................. 135 7.5 Future Research .......................................................................................................... 136 References ...................................................................................................................................139 ix Appendices ..................................................................................................................................152 Appendix A Scenario Study.................................................................................................... 153 A.1 Task Product ........................................................................................................... 153 A.2 Manipulation ........................................................................................................... 155 A.3 Experimental Procedure .......................................................................................... 159 A.4 Measurement ........................................................................................................... 160 A.5 Data Analysis .......................................................................................................... 162 Appendix B Pilot Test for Feasibility and Desirability Consideration with respect to Product Attributes................................................................................................................................. 173 B.1 Results on the Product Attribute Evaluation ........................................................... 174 Appendix C Participant Background Information .................................................................. 176 Appendix D Scenarios of the Main Study .............................................................................. 182 D.1 Scenario for the No Simulation Condition.............................................................. 182 D.2 Scenario for the Partial Simulation, the Low-Level Simulation, and the High-Level Simulation ........................................................................................................................... 183 Appendix E RA’s Algorithm and Utility Cost Formula ......................................................... 185 x E.1 RA’s Algorithm ...................................................................................................... 185 E.2 Utility Cost Calculation .......................................................................................... 187 Appendix F Measurement Items of the Main Study ............................................................... 189 F.1 Pre-Questionnaire Survey ....................................................................................... 189 F.2 Post-Questionnaire Survey...................................................................................... 199  xi List of Tables Table 1: Comparison between process and outcome goal focus (Freund et al. 2012).................. 25 Table 2: Comparison between low-level and high-level construals (Trope and Liberman 2003, p. 405) ............................................................................................................................................... 30 Table 3: Experimental conditions ................................................................................................. 64 Table 4: Washing machine attributes ............................................................................................ 67 Table 5: Product attribute correlations .......................................................................................... 68 Table 6: Measurement flow .......................................................................................................... 81 Table 7: Adjusted mean score on intention to purchase a green product ..................................... 99 Table 8: Preference on price ....................................................................................................... 109 Table 9: Means and standard deviations of the product attribute ranking and preferences ........ 112 Table 10: Product attribute ranking and preferences based on the simulation conditions.......... 113 Table 11: Adjusted mean and standard deviation on the desirability consideration ................... 114 Table 12: Loadings and cross loadings ....................................................................................... 116 Table 13: Internal consistency and discriminant validity ........................................................... 116 xii Table 14: Hypotheses testing results........................................................................................... 120 Table 15: Awareness of cause-and-effect relationship, feasibility-desirability trade-off, and environmental concern ................................................................................................................ 123 Table 16: Two product models used in the scenario study ......................................................... 154 Table 17: The scenario study's measurement flow ..................................................................... 162 Table 18: Means of the product attribute ranking ....................................................................... 170 Table 19: Simple mixed design for product attribute ranking .................................................... 171 Table 20: The evaluation of the feasibility and the desirability of the product attributes .......... 175 Table 21: Participant background information ........................................................................... 181 Table 22: Best and worst attribute values ................................................................................... 186 Table 23: Assumption and Data for the Utility Cost Calculation Energy Star (2014) ............... 187 Table 24: Utility cost formula ..................................................................................................... 188 Table 25: Pre-questionnaire survey ............................................................................................ 198 Table 26: Post-questionnaire survey ........................................................................................... 206 xiii List of Figures Figure 1: Value-belief-norm theory (Stern 2000; Stern et al. 1999) ............................................. 43 Figure 2: Theory of planned behavior (Ajzen 1991) .................................................................... 46 Figure 3: Research model ............................................................................................................. 50 Figure 4: No cause-and-effect simulation (condition 1) ............................................................... 70 Figure 5: Example of the product attribute explanation ............................................................... 71 Figure 6: Partial cause-and-effect simulation (condition 2) ......................................................... 72 Figure 7: Low-level cause-and-effect simulation (condition 3) ................................................... 73 Figure 8: High-level cause-and-effect simulation (condition 4) ................................................... 74 Figure 9: Experimental procedure ................................................................................................ 78 Figure 10: The effect of simulation design on temporal distance ................................................. 94 Figure 11: The effect of simulation design on mental construal .................................................. 94 Figure 12: Intention to purchase a green product (green score) ................................................... 95 Figure 13: Number of products participants clicked to see more details ...................................... 97 Figure 14: Adjusted mean score on intention to purchase a green product .................................. 99 xiv Figure 15: Energy use ranking .................................................................................................... 101 Figure 16: Spin speed ranking .................................................................................................... 102 Figure 17: Water use ranking ...................................................................................................... 103 Figure 18: Price ranking.............................................................................................................. 104 Figure 19: Preference on energy use ........................................................................................... 106 Figure 20: Preference on spin speed ........................................................................................... 107 Figure 21: Preference on water use ............................................................................................. 108 Figure 22: Preference on price .................................................................................................... 109 Figure 23: Green attributes ranking ............................................................................................ 111 Figure 24: Adjusted mean and standard deviation on the desirability consideration ................. 115 Figure 25: Structural path model ................................................................................................ 117 Figure 26: Awareness of cause-and-effect relationship .............................................................. 121 Figure 27: Awareness of trade-off between price and green attributes ...................................... 122 Figure 28: Environmental concern .............................................................................................. 123 Figure 29: The screenshot of two product models used in the scenario study ........................... 155 xv Figure 30: The product models chosen by participants in the scenario study ............................ 167 xvi Acknowledgements First, I would like to express my appreciation to my supervisor, Prof. Ronald Cenfetelli. Ron has coached and supported me throughout this study. He has dedicated his valuable time and resources to my research work and encouraged me to focus my work on the design characteristic of technology that is the key of IS research. I would also like to thank Prof. Izak Benbasat for inspiring my research work and has given the constructive suggestions which considerably helped improve my study. His suggestions were important to my research completion. Additionally, I offer my gratitude to the committee of my thesis, Prof. Hasan Cavusoglu, for his support. I am thankful to Clara (Ye) Chen for her patience in developing the experimental website and Lior Shmueli for sharing me the ideas to do research and assisting me to publish the website. To all MIS fellows, I feel grateful for helping me with the pilot studies.  Lastly, I would like to convey special thanks to my parents for their love and support throughout my life.xvii Dedication  To my parents 1 Chapter 1: Introduction Concern for the environment is a prevalent and important issue. According to the World Economic Forum’s Global Risks 2014 report, water crises and climate change are two of the ten global risks of highest concern for 2014 from the perspectives of 700 experts around the world (The World Economic Forum 2014). The general public, particularly consumers, often share these concerns. Many consumers strive to help the environment through consideration of the types of products they purchase, including what are termed “green” products which are designed to be less damaging to the environment as compared to traditional alternatives. Many companies offer green products for their consumers. They may also seek various green certifications such as Green Seal and Energy Star as a means of signaling environmental concerns to their consumers. However, green products often invoke a trade-off as compared to their non-green alternatives on offer. In particular, green products are often more expensive thus dampening consumer demand for those products (Farrell 2012). Nevertheless, there are studies supporting that some consumers will pay more for green products. AYTM Market Research (2013) reveals that almost a half of US Internet users would pay 5 to 10% more for eco-friendly products. Laroche et al. (2001) found that the target customer who tends to pay more for environmental-friendly product is a female who get married and has at least one child living at home. Despite the importance of protecting the environment, there are indications that green purchasing is becoming less prevalent and individuals are now less worried about the environment. A poll conducted by GlobalScan (2013) found that concern for the environment has been decreasing in recent years, despite scientific reports showing that the environment is being increasingly 2 degraded. Another poll from Huffington Post and YouGov (2013) also indicates that Americans are less worried about the environment compared to Americans in 1971, a year after Earth Day was founded. For example, 63% of respondents in 1971 reported that to help restore and fortify the national environment was very important, whereas only 39% of respondents in 2013 reported that it was very important. In line with these polls, Gallup Poll (2014) revealed that people today prioritize the environment less. In 2000, 70% of people gave the environmental protection a priority even at the risk of curtailing economic growth, whereas in 2013 only 40% did so (Gallup Poll 2014). The academic literature has also extensively studied pro-environmental behaviors defined as “behaviors that consciously seek to minimize the negative impact of one’s actions on the natural and built world” (Kollmuss and Agyeman 2002, p. 240) and how to promote such behaviors (Kim and Choi 2005; Loock et al. 2013; Mainieri et al. 1997; Pichert and Katsikopoulos 2008; Stern 2000). An individual’s pro-environmental behavior is related to purchasing and consuming a green product with the goal of reducing the impact to the environment associated with the production or the consumption of that product. Several psychology and marketing studies have investigated the factors contributing to a green behavior (Bickart and Ruth 2012; Griskevicius et al. 2010; Mainieri et al. 1997; Olsen et al. 2014). For example, Bickart and Ruth (2012) studied eco-seals in terms of consumers’ characteristics and advertising characters in order to identify the conditions under which green eco-seals positively influence consumers’ attitudes and intentions to purchase a green product. Even in cases where consumers possess strong positive attitudes toward the environment and strong intentions to pay more for green products, those attitudes and intentions do not necessarily 3 lead to an actual green purchase (Kollmuss and Agyeman 2002). Forrester Research (2008) suggests that “green attitudes do not guarantee green actions”. It found that although 60% of consumers are concerned about the environment, they are less likely to purchase green products if they perceive any inconvenience in purchasing green products. Forrester provides the example of energy-efficient appliance purchases which only 7% of consumers actually pay more for the green alternative. Consistent with Forrester, Kollmuss and Agyeman (2002) summarize pro-environmental behavior frameworks and present the low-cost high-cost model (Diekmann and Preisendörfer 1992). This model proposes that people will perform pro-environmental behaviors if those behaviors require the least cost. In addition, there is evidence supporting that environmental concerns will not significantly affect individuals’ actions. Stern et al. (1993) and Stern (2000) found that people will more likely perform pro-environmental behaviors if they perceive that performing those behaviors can avoid harm for themselves, not for others. It is less likely that they perform pro-environmental behaviors to reduce harm for other people or for the environment. As a result, there are factors other than individual attitudes and intentions influencing pro-environmental behaviors. Factors contributing to pro-environmental behaviors include status and reputation. Griskevicius et al. (2010) propose that status and reputation play an important role in motivating pro-environmental behaviors. In other words, if an individual is led to believe that their pro-environmental purchase will go unnoticed by others, it is less likely that the individual will purchase a green product. In their study, they compared in-store shopping and online shopping. The results suggest that status did not influence individuals who shopped online to pay more for green products than for luxurious and better efficient non-green products. Griskevicius et al. (2010) 4 make reference to a New York Times report (Maynard 2007) on purchases of Prius, a hybrid car—buyers of the Prius want everyone to know they are driving a hybrid—suggesting that individuals value the reputational effects of green purchases more than the pro-environmental effects. These status and reputation effects are born out in other studies. The product type which includes privately-consumed and publicly-consumed products, has an impact on consumers’ decision making. This product type is based on the degree to which other people can notice individuals’ product. If the product an individual makes a purchase is for individual use, the individual’s purchase decision will unlikely be affected by status and reputation. On the other hand, the role of status and reputation will influence consumers greater if they buy products which can be identifiable by other people, especially their reference group. Therefore, the product type can moderate the relationship between status and reputation, and an individual’s purchase decision. Bearden and Etzel (1982) explore the relationship between reference group influence on product and brand purchase decision with the consideration of two dimensions of product types—necessity vs. luxury, and publicly- vs. privately-consumed products. In publicly- and privately-consumed products based on how others can identify the product individuals buy, they found that there were stronger influences of value-expressive and utilitarian influences on product selection decisions with participants in public dimensions reporting greater influences. Value-expressive influence is related to psychological association that an individual tries to imitate her own reference group, while utilitarian influence refers to an individual’s compliance with her reference group’s expectation in order to be rewarded or to avoid punishments. This suggests that people will be less concerned about how others think about their purchase decision if the products are privately-consumed. 5 In line with the findings from Griskevicius et al. (2010) and from Bearden and Etzel (1982), Internet users who make an online purchase are less likely to go green. According to Harris Interactive (2012), less than 10% of US Internet users perceived that the environment issue was critical when purchasing products online during 2012. Another report from Cone Communications (2013) revealed that 7% of respondents said they considered the environment every time they shop. Accordingly, only a small number of Internet users do concentrate much on the effect of their purchase decision toward the environment when shopping online. More and more buying is being done online. As a result, green buying may be curtailed, because so much of buying online is private, not public. This may substantially lessen the opportunities for status and reputation motivations for buying green and thus lead to even less pro-environmental purchasing. Nowadays online shopping is substantially growing. Forrester Research (2014) reported that online retails in the US will reach $294 billion in 2014 and $414 billion by 2018. This marks the significant growth of online market which allows consumers to shop privately and will lead to a substantial decrease in the opportunities for status and reputation to induce the online green purchase. Additionally, the opportunities would be aggravated even further if the product is privately-consumed, since consumers will pay less attention to status and reputation as well as the group influence. As a consequence, it is challenging to persuade individuals to make an online purchase of privately-consumed products. In this case, technology will help motivate people to buy green products in the absence of the reputation and status effects as well as the normative influence. Technology can provide ways of fostering interaction and communication with individuals that may increase the likelihood of the individuals to engage in green purchasing online. For example, Yu (2012) designed an e-commerce website which provided a 6 recommendation agent that incorporates conditioning and evaluation cues (e.g., a “smiley” face) to promote online green purchases. The results from her study support that online consumers were more likely to choose green products when they were provided with these cues. Thus, her study reveals that we can leverage the capability of technology to reinforce green purchasing behavior. This study will explore the role of technology in increasing the likelihood of purchasing green products that are privately-consumed in the online context. Specifically, this study will investigate the role of “simulation” as a tool to enable persuasion. Simulation, which refers to an attempt to imitate the real world, has long been studied in education research to facilitate learning (Barry Issenberg et al. 2005; Duchastel 1990; Reigeluth and Schwartz 1989; Rieber et al. 2004). Two key characteristics of simulation are manipulation and interactivity (Gredler 2004; Rieber 1996). Simulation allows individuals to change inputs and observe outcomes correspondingly. It harnesses the capability of technology to allow people to interact with and manipulate the simulation tool in order to learn the relationship between inputs and outputs. In the absence of technology, individuals would require higher cognitive effort to observe and learn about the cause-and-effect relationships. Therefore, individuals learn the relationships through interacting with and manipulating a simulation tool (Reigeluth and Schwartz 1989). We propose that this will increase an individuals’ awareness of the consequences of their purchase decision on the environment. Learning is proposed to affect individuals’ attitudes and to influence long-term persuasion (Greenwald 1968; Kraiger et al. 1993; Love and Greenwald 1978). Simulation does not only influence individuals to be aware of the outcomes, but also allows them to repeat their actions (Fogg 2002). Consequently, they tend to perceive more control over their behavior and thus to perform the green purchase. 7 To best of our knowledge, despite its benefits, the simulation tool lacks a theoretical foundation to support persuasion and is not well studied in IS. Thus, the current study’s objectives are to evaluate whether the cause-and-effect simulation can persuade individuals to buy a green product that is privately-consumed and to figure out the better design features of the simulation tool of persuasion. In order to develop the simulation tool, we need to identify the important features that the tool will have. In the context of the current study, the simulation tool will convey the relationship between individuals’ decision-making on the product, the cause, and the impact of their decision-making on resources the product consumes (e.g., electricity), the effect.  The recommendation agent which refers to “software agents eliciting the interest or preferences of individual users for products, either explicitly or implicitly, and make recommendations accordingly” (Xiao and Benbasat 2007, p. 137) will be employed to reflect the cause part of the cause-and-effect simulation tool, as it helps individuals to make a product decision more effectively. In designing the effect part, we adopt the “construal-level theory” (CLT).  Construal-level theory suggests that psychological distance (e.g., temporal distance) will influence mental construal which refers to mental representation with respect to the event, the action, or the object (Liberman et al. 2002; Liberman and Trope 1998; Trope and Liberman 2000). In other words, how individuals perceive the distance between themselves and the event, the action, or the object will change the way they construe their thoughts toward that event, action, or object. Two levels of construal, low-level and high-level construal, are proposed based on the level of abstraction. Two construal levels are defined as follows:  “Low-level construals are relatively unstructured, contextualized representations that include subordinate and incidental features of events. High-level construals, in contrast, are schematic, decontextualized representations that 8 extract the gist from the available information. These construals consist of a few superordinate core features of events.” (Trope et al. 2007, p. 83). CLT proposes that the farther away the event is, the more likely individuals will construe their thoughts toward the events at a higher level. Prior CLT studies found a significant relationship between temporal distance and construal level as well as a significant effect of temporal distance and construal level on consumers’ preferences, judgments, and evaluations (e.g., Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). We expect that implementing CLT in the cause-and-effect simulation tool would persuade online consumers to purchase green products more. This will reveal not only concrete design of persuasive simulation to promote green purchasing online, but also the theoretical lens to evaluate its effect. 1.1 Research Questions for this Study 1) What are the key characteristics of a simulation tool to enable the persuasion of online consumers?  2) Can cause-and-effect simulation persuade individuals to be more likely to purchase green products that are privately-consumed online? 3) How does the simulation of cause and effect influence online consumers’ green purchasing decision?9 Chapter 2: Literature Review 2.1  Pro-environmental Behaviors Pro-environmental behaviors are the central issues in environmental psychology. Pro-environmental behaviors refer to “behaviors that consciously seek to minimize the negative impact of one’s actions on the natural and built world” (Kollmuss and Agyeman 2002, p. 240).  Several environmental psychology researchers develop frameworks, models, and theories to explain how and why people act environmentally (Kollmuss and Agyeman 2002; Schwartz 1977; Stern et al. 1999). These works can be sorted into two main streams—self-interest and pro-social motives—based on the motive of pro-environmental behaviors (Bamberg and Möser 2007). Bamberg and Möser (2007) provide norm-activation model (NAM, Schwartz 1977) and theory of planned behavior (TPB, Ajzen 1991) as the examples of pro-environmental models motivated by pro-social and self-interest motives respectively. In NAM, Schwartz (1977) proposes that an individual’s moral norm will trigger the altruistic behavior if the individual perceives that there are threats to others and that her action can mitigate those threats. This reflects that the individual’s action is motivated by the need to help the society and the environment. On the other hand, TPB, the extended version of theory of reasoned action (TRA, Fishbein and Ajzen 1975), mainly assumes that an individual will trade-off between the costs and the benefits associated with the behavior before performing that behavior (Ajzen 1985, 1991). In this view, the underlying motive is self-interest to maximize the benefits to one-self. According to the two main streams, various models are developed by integrating both streams to increase the explanatory power. The integrated view recognized that people do things to help both 10 themselves and others. These integrated models found that people act pro-environmentally because of both rationale and altruistic reasons. For instance, Hines et al. (1987) identify three significant determinants of pro-environmental behaviors including attitudes, locus of control, and personal responsibility. Attitudes and locus of control are influenced by self-interest motive, whereas personal responsibility is motivated by pro-social perspective. Another example is Stern et al. (1999)’s value-belief-norm theory (VBN). Stern et al. (1999) develop theory based on NAM. Unlike NAM, which focuses on how to reduce threats to others, VBN captures altruistic orientation into two main dimensions—collective goods and private benefits. Originally Stern et al. (1993) differentiate three types of value orientations—egoistic, social, and biospheric orientations. In VBN, these orientations are then rearranged into egoistic, altruistic, and biospheric values. Altruistic values and biospheric values focus on the benefits to other people and the environment. On the contrary, egoistic values highlight the importance of individual benefits. Three values are hypothesized to activate an individual’s environmental concern, beliefs, norms, and thus pro-environmental behaviors (Stern 2000; Stern et al. 1999). 2.2 Persuasion Persuasion has long been studied in psychology, marketing, and IS for more than three decades (e.g. Bhattacherjee and Sanford 2006; Grier and Deshpandé 2001; Petty and Cacioppo 1986b). The term “persuasion” is defined by many researchers. Consistent with persuasion literature, Petty and Cacioppo (1986b) define persuasion as how to change attitude, which in turn affects behavior. Attitude refers to “the evaluative dimension of a concept” and differs from beliefs, defined as “the probability dimension of a concept” (Fishbein 1963, p. 233). Attitude plays a critical role in persuasion, since it links to action (Ajzen and Fishbein 1977). To aim at motivating people through 11 technology, Fogg (2002) defines persuasion as “the attempt to change attitudes or behaviors or both”. Thus, persuasion is related to attitude change and/or behaviors in order to motivate the desired behavior.  In psychology, many persuasion theories have been developed. Two major theories widely accepted and used are the elaboration likelihood model (ELM, Petty and Cacioppo 1986a; Petty and Cacioppo 1986b) and the heuristic systematic model (HSM, Chaiken 1980). Both are dual-process models of persuasion that mainly focus on two routes of persuasion: central/systematic versus peripheral/heuristic routes. When an individual adopts the central/systematic route, the quality of argument plays the greatest role in persuasion. On the other hand, an individual who relies on peripheral/heuristic route are enticed through cues such as source expertise, source attractiveness, and source credibility. Additionally, both theories are studied in the context of advertising of brand and of product (e.g. Kirmani and Campbell 2004; Meyers-Levy and Malaviya 1999; Peck and Wiggins 2006). Marketing research examines the effect of message characteristics (Aaker and Lee 2001), the cues (Meyvis et al. 2012), and the consumer factors on consumers’ attitudes (Friestad and Wright 1994).  Another theory in persuasion is protection motivation theory that is studied in the context of health-related issues (Maddux and Rogers 1983; Rogers 1975). This theory differs from elaboration likelihood model and heuristic systematic model in that it mainly concentrates on a fear appeal to enable persuasion. A message that provokes fear in individuals will motivate them to reduce their negative emotion through suggestions from the message. 12 2.3 Persuasive Technology Persuasion has also been studied in the context of technology, particularly how technology can persuade people. Fogg (2002) defines the term “persuasive technology” as an interactive technology designed to change attitudes and behaviors of users. The paradigm of persuasive technology is called captology, which is “the study of computers as persuasive technology” and refers to “how people are motivated or persuaded when interacting with computing products rather than through them” (Fogg 2002, p. 16). He suggests the design of technology to persuade individuals in terms of a functional triad, which consists of a tool, a medium, and a social actor. As a tool, persuasive technology may strengthen an individual’s ability to perform the target behavior. As a medium, persuasive technology provides an individual an experience to perform the behavior. Finally, as a social actor, persuasive technology may be used to directly interact and communicate with an individual in order to create the relationship between individuals and a technology by giving a positive feedback, proposing a target behavior or attitude, and providing social support. Prior IS literature examines the impact of persuasion on attitude change in various contexts and employ some persuasive design characteristics such as personalization, language, self-monitoring, and conditioning. For example, Tam and Ho (2005) study the web personalization to persuade people using three elements—level of preference matching, recommendation set size, and sorting cue. In this study, they adopt the elaboration likelihood model of persuasion as an underlying framework. They categorize the level of preference matching, defined as “the extent to which the web content generated by the personalization agent appeals to users” (Tam and Ho 2005, p. 276) into a cause of the central route of persuasion, because people will be more likely to read the 13 content more thoroughly and thus to buy the recommended products if the content fits their preferences and interests. On the other hand, they capture recommendation set size, and sorting cue into factors contributing to the peripheral route of persuasion. Recommendation set size, referring to the number of products recommended to consumers, is proposed that a large set will draw people’s attention more than a small set. The presence of sorting cue as the order of recommended products will lead to higher elaboration.  According to this study, the personalization strategy that provides the recommended products to consumers will positively influence their product decision-making. As another example, Angst and Agarwal (2009) investigate how people adopt electronic health records when the privacy concerns are present. Grounded within ELM, their work concentrates on the effect of argument framing—a positive or neutral-framed message—and the degree of issue involvement. They found that despite the high privacy concerns using positive-framed message, which presents the advantages of the electronic health record, will induce people to adopt the system. This study uses the framing language to communicate the message to users in order to enable persuasion. In addition, Loock et al. (2013) adopt the notion of goal setting theory, default, and feedback in persuading people to perform energy-efficient behavior. In their study, a default refers to a reference point provided for consumers to help them evaluate other options.  In other words, they usually set the goal of the energy-saving level with reference to the default. Loock et al. (2013) thus suggests that setting the default in the realistic level will positively affect consumers’ goal. They also propose that feedback that presents whether consumers accomplish their goals, will moderate the impact of the default on a consumer’s goal. This indicates that feedback that provides 14 self-monitoring of an individual’s behavior has an indirect effect on a consumer’s energy consumption. Yu (2012) also adopts the notion of persuasive technology to promote green purchases. In her study, she investigates the effect of the conditioning mechanism proposed by Fogg (2002) on persuading individuals to purchase products which are more green. According to her study, this mechanism incorporated a recommendation agent which provided an array of cues ranging from positive (e.g., a “smiley” face) to negative (e.g., “sad” face), and more neutral emotional signals in between. These cues were presented to online consumers towards influencing their product alternative selection. The cues displayed to consumers varied not only in their valence, but also in terms of their evaluability—the degree of the cue’s valence precision (Yu et al., forthcoming). For example, presenting a user with a numbered cue, such as a 1 to 7 scale, is somewhat imprecise in that either 1 or 7 can be interpreted as either positive or negative, if no anchors are provided for that scale. The face cues, in contrast, are high in evaluability (e.g., a smiley face is universally positive). When participants value the green product attributes by indicating the high value toward the green attributes (e.g. energy use) on the RA, the appropriate cue (e.g., smiley face) will be shown. Conversely, the sad face cue will be presented if participants indicate the low value of the green attributes on the RA. Yu found that this conditioning mechanism successfully drove online consumers to more likely select products that are greener. One of the design elements of persuasive technology as a medium is a simulation of cause and effect. Fogg (2002) purports that the cause-and-effect simulation allows individuals to observe and to explore the relationship between the cause and the effect which can persuade individuals without forcing. Consistent with Fogg (2002), Oinas-Kukkonen and Harjumaa (2009) define persuasive 15 systems as “computerized software or information systems designed to reinforce change or shape attitudes or behaviors or both without using coercion or deception” (Oinas-Kukkonen and Harjumaa 2009, p. 486). They suggest the design of persuasive systems based on the functional triad. One of their design features is simulation, which refers to “a system enabling individuals to immediately observe the relationship between cause and effect” (Oinas-Kukkonen and Harjumaa 2009, p. 492). This relationship should be related to the behavior of individuals. Though both Fogg (2002) and Oinas-Kukkonen and Harjumaa (2009) identify the role of the simulation to enable persuasion, they do not explicitly demonstrate why simulation could be a persuasive tool and what the key features of simulation to help change people’s attitudes. 2.4 Simulation Simulation is extensively studied in educational research in order to facilitate learning (Barry Issenberg et al. 2005; Duchastel 1990; Reigeluth and Schwartz 1989; Rieber et al. 2004). According to this research, simulation refers to the attempt to replicate a real world phenomenon (Barry Issenberg et al. 2005; Reigeluth and Schwartz 1989; Rieber 1996; Thurman 1993). This definition is consistent with other researchers (Law et al. 1991; Nelson et al. 2001). Although the term “simulation” is not clearly defined, simulation in an educational purpose is regarded in terms of fidelity, manipulation, and interactivity (Barry Issenberg et al. 2005; Gredler 2004; McGuire 1976; Reigeluth and Schwartz 1989; Rieber 1996). The fidelity has been examined in medical education (Barry Issenberg et al. 2005; Leigh 2008) as well as in the marketing.  In the marketing, Hershfield et al. (2011) found that creating the avatar which represents future selves of individuals will positively influence their retirement saving 16 behavior. That is, the more they perceive the avatar looks like them, the more they allocate money into their retirement saving. These results suggest that in order to persuade individuals to save for their future the simulation which possesses the high degree of fidelity can be used. Although the degree of fidelity of the simulation plays a key role in some situations as described previously, Reigeluth and Schwartz (1989) posit that reduced fidelity, such as by excluding superficial variables and maintaining only key variables, helps learners to focus more on important contents and thus learn more efficiently. Reigeluth and Schwartz (1989) study the design of instructional overlay, the component which optimizes learning and motivation. Various dimensions of instructional overlay designs—such as types, functions, features, and behavior—are proposed. Causal simulation is one of the types. Causal simulation aims at instructing the causal relationship between two or more factors. Also, simulation can harness the power of technology allowing users to interact with and to manipulate many variables more efficiently. Learning the cause-and-effect relationships may result in persuasion. For instance, De Jong and Van Joolingen (1998) adopt this notion to facilitate learners to learn the scientific rules. In their study, learners were assigned to figure out the underlying scientific rules, the causal relationships between causes and effects, presenting by a computer simulation. They were allowed to directly enter input variables and thereby observe outputs in the form of graphics and animations. The findings suggest that a computer simulation helps learners to study the scientific rules more effectively. Consequently, the notion of educational simulation is applicable to enable persuasion. 17 2.4.1 Instructional Simulation and Persuasion Prior literature indicates the correlation between learning and persuasion. For instance, Greenwald (1968) posits that cognitive learning is important for persuasion. He states that to effectively change an individual’s attitude partly requires learning and retention of the message content. In marketing research, Krugman (1965) examines the effect of television advertising which allows viewers to learn the advertising messages despite low degree of viewers’ involvement regarding the advertised items. He points out that people learn the messages from advertising and that gradual exposure to the advertising messages will result in overlearning, a transition of information from short-term memory to long-term memory system, that may affect the attitude change. Simulation as an instructional method effectively influences learners’ attitude development. This is evident by the meta-analysis regarding the effectiveness of a simulation game as a teaching tool conducted by Dekkers and Donatti (1981). Their findings indicate that simulation helped learners to develop attitudes more than traditional lecture, while it did not show superior impact on cognitive development and retention of knowledge. Another study by Schumacher (1997) supports the relationship between simulation and attitude change. He adopted simulation to train corporate employees in order to teach attitudes pertaining to organizational cultures. His findings indicate that simulation led to employees’ attitude changes with respect to the new frame of cultural reference. Simulation will also help learners to develop positive evaluation regarding the learning objects or issues, as it allows individuals to learn thing by themselves without explanation. According to education research, discovery or exploratory learning mode may lead learners to develop positive 18 judgment on what they learn. For example, Ajewole (1991) examines the effect modes of instruction—discovery and expository—on learners’ attitudes on biology. Findings suggested that teaching science in a discovery method resulted in more favorable attitudes toward problem recognition and solving than teaching in an expository method. Learning the cause-and-effect relationships can change a user’s attitudes and therefore behaviors, since learning has an impact on attitudes (Kraiger et al. 1993). Learning does not affect only cognitive outcomes captured by retention, but also affective consequences. According to Kraiger et al. (1993), learning results in three types of outcomes—cognitive, skill-based, and affective outcomes. As the first type of cognitive learning outcomes, a cognitive outcome pertains to knowledge that a learner acquires, organizes, and applies. The second type of the outcomes is a skill-based outcome, which is associated with the development of learners’ skills and includes three stages of the skill development: 1) initial skill acquisition, 2) skill compilation, and 3) skill automaticity (Kraiger et al. 1993).  Finally, an affective outcome is related to an internal state that influences attitude and motivation thereby shaping an individual’s action. Kraiger et al. (1993) define two types of affective outcomes—attitudinal and motivational outcomes. Attitudinal outcomes pertain to learners’ attitudes toward learning objects. The attitudinal outcomes are in line with thoughts in response to a message. That is a learner, when learning a message conveyed by an instructor or a tool, develops attitudes toward that message. Kraiger et al. (1993) assert that these outcomes directly affect attitudes. This is captured through the direction of learners’ feelings with respect to the learning object and through the attitude strength, the degree of persistence of an individual’s attitude. Specifically, internalization captured as the attitude strength is important in persuasion (Kelman 1958). It reflects that an individual accepts attitudes 19 which, in turn, shape the individual’s judgment and determine action. Thus, it might be possible that attitudinal outcomes may lead to persuasion. As another type of affective outcomes, motivational outcomes focus on how to increase learners’ competence in performing tasks. Unlike attitudinal outcomes, motivational outcomes are proposed to influence attitudes indirectly (Kraiger et al. 1993). They include the concept of self-efficacy which refers to an individual’s perception of her own capabilities to achieve a certain activity (Bandura 1977). Motivation is one of the key factors affecting an individual’s degree of elaboration. It is evident by various persuasion studies that high motivation will result in the central route to persuasion (e.g. Homer and Kahle 1990; Petty and Cacioppo 1984). In addition, cognitive and affective learning are investigated in the education literature (e.g. Vermunt 1996). For example, Vermunt (1996) defines cognitive learning as an activity which learners process learning information and refers affective learning to an activity that correlates with feelings resulting from learning such as motivation. Cognitive learning includes various activities such as relating. Relating pertains to a practice that a learner figures out the relationships among different objects (Vermunt 1996). On the other hand, affective learning may stimulate the learner’s emotional state that will influence learning. Thus, the learner may be motivated to learn the content more if she has positive feelings. Moreover, learning is the key to long-term persuasion, since it requires learners to process information that will affect their cognitive knowledge and affective outcomes. This indicates that individuals have a greater degree of elaboration to assess the argument. Many persuasion studies assert that to change individuals’ attitude in the long run requires the central route to persuasion 20 (Chaiken and Stangor 1987; Petty and Cacioppo 1986a; Petty et al. 1995; Petty et al. 1997). For instance, Petty et al. (1995) study how to generate new strong attitudes, the attitude strength. This refers to persistence in attitude leading to the actual behavior. They postulate that “…when an attitude changes as a result of careful thinking about the merits of the attitude object (central route to attitude change), the resulting attitude will be stronger than if the attitude changes because of a relatively simple cue in the persuasion setting…(peripheral route to attitude change)” (Petty et al. 1995, p. 94). In other words, the higher degree of elaboration people use to evaluate the messages to form the new attitude, the more likely people maintain that attitude and translate it into their behavior. Consequently, the concept of instructional simulation which facilitates an individual’s learning can be applied to persuade people in the long term. This lends support to the effectiveness of the simulation tool in enabling persuasion. 2.4.2 Mental Simulation While education research adopts simulation to facilitate learning, psychology and marketing research highlight the importance of mental simulation (Escalas 2004; Pham and Taylor 1999; Wells and Gavanski 1989). Mental simulation is defined as “the imitative representation of real or hypothetical events” (Pham and Taylor 1999, p. 250). This refers to a cognitive activity (Taylor and Pham 1996). Mental simulation allows the individual to rehearse the future situations, to observe the present ones, and to review and to reconstruct the past ones (Pham and Taylor 1999; Taylor and Pham 1996). Thus, it will increase an individual’s self-regulatory leading to the actual action. This is also evident by Bandura (1989) asserting that mental simulation enhances perceived self-efficacy. Accordingly, it is important in transforming thoughts to actions by increasing the likelihood of the actual action. 21 The concept of mental simulation is used to enable persuasion as found in marketing literature. Marketing studies employ mental simulation to persuade consumers through advertising message and found that mental simulation can persuade people by transporting consumers to immerse in the advertising message (Castaño et al. 2008; Escalas and Luce 2003, 2004). It was also evident that mental simulation may increase positive attitudes. For instance, Crisp and Turner (2009) found that imagination about contacting other people in different groups could reduce bias and develop positive perception in the social interactions. This suggests that mental simulation has an impact on individuals’ attitudes and thus persuasion. 2.4.2.1 Process- and Outcome-Focused Simulation The role of mental simulation in persuasion can be categorized into two types: process- and outcome-focused simulation. Process simulation provokes individuals to think about the steps to reach a goal, while outcome simulation leads individuals to think about a desired outcome if achieving a goal (Zhao et al. 2009; Zhao et al. 2007). Pham and Taylor (1999) suggest that outcome simulation might enhance self-efficacy perception, while process simulation may strengthen goal achievement. This is consistent with their findings that process-focused simulation, compared to outcome-focused simulation, leads to better study performance. They reason that when an individual carefully thinks about the steps required to achieve good scores, following those steps will facilitate her action. However, process-focused simulation will not always lead to better performance. This is supported by Thompson et al. (2009). They examine the impact of the process- and the outcome-focused simulation on consumers’ product decision and found that the process simulation led to higher 22 trade-off difficulties in selecting a product choice from product alternatives. In their study, the process simulation resulted in the focus on both means and ends, while the outcome simulation led to the single focus on ends. Means pertain to the feasibility such as the quality and the price of the product. On the contrary, ends are associated with the desirability or benefits of the product. The results of their study indicate that participants in the process condition showed more negative thought toward the product than those in the outcome condition, since process-focused thinking induced more salience of the trade-off decision between means and ends, compared to outcome-focused thinking, which focuses only on ends. In addition, according to marketing research, there are other factors such as temporal frame and product type that moderate the effect of simulation types on individuals’ performance (Zhao et al. 2007, 2011). Zhao et al. (2007) assert that process-focused and outcome-focused simulations can be used to reduce consumers’ preference inconsistency in a product choice. When a consumer plans to buy a specific product, she naturally thinks about the product’s performance to achieve her goal at first and thus develops a product alternative in her mind. Nevertheless, at the time of purchasing she tends to think about how to use the product to accomplish that goal rather than about the performance aspects of product. This will result in the preference inconsistency in her product alternatives. Thus, Zhao et al. (2007) propose that in order to reduce preference inconsistency the less salient thought would be stimulated. That is outcome-focused simulation should be provided at the near-future point, whereas process-focused simulation should be given at the distant-future point. Moreover, Zhao et al. (2011) extend prior research on two types of simulation to examine their role with respect to the modes of evaluation—cognitive and affective modes—on consumers’ 23 evaluation of the product for which consumers lack of experience. They also investigated the moderating effects of product types—a utilitarian product, for which consumers focus on the functionality aspects, and a hedonic product, for which consumers concentrate on the enjoyment aspects, and of time frame—near- and distant-future events. Their findings indicate that consumers positively evaluated the product under a cognitive mode if they were encouraged to think about the outcome regarding that product, whereas they had favorable evaluation under an affective mode if they were stimulated to think about the process to use that product. Also, the product types have an effect on consumers’ product evaluation in that for a utilitarian product consumers with outcome-focused simulation developed higher evaluation under the cognitive mode, while consumers with process-focused simulation favored the product under the affective mode. The results for a hedonic product are reverse to those for a utilitarian condition. In terms of time frame, for a near-future evaluation consumers who focused on the cognitive mode had positive evaluation if they imagined about the outcome, while consumers who concentrated on the affective mode developed favorable evaluation if they thought about the process. The results are opposite for a distant-future evaluation. The relationship between mental simulation and temporal frame has been revealed in another study by Ülkümen and Thomas (2013). They found that the high degree of involvement induces process-focused thoughts and this moderates the effect of temporal frame on intention to perform a behavior. However, they did not find that involvement will elicit outcome-focused thoughts.  In their study, the intention to adopt the diet plan is a target behavior. Participants with perceived high involvement toward the plan in the short-term temporal frame (a 12-month diet plan) reported the higher intention to stay with the plan than those in the long-term temporal frame (a one-year 24 diet plan), since those in the short-term temporal frame felt that the 12-month plan was shorter and thus easier to adopt. These results indicate that temporal frame bias together with the process-focused simulation will affect individuals’ intention to perform a behavior. In order to develop the understanding of the differences between the process and the outcome, we need to explore the concept of process and outcome in other research streams. 2.4.2.2 Process and Outcome Focus in Related Literature The concept of process and outcome focus has been examined in the goal and the organizational control literature. In the goal stream, many studies distinguish between a process and an outcome goal (Zimmerman and Schunk 2006; Zimmerman and Kitsantas 1997, 1999; Zimmerman and Schunk 2004). A process goal is associated with the means needed to complete a certain action, whereas an outcome goal is related to the ends associated with the performed action (Woolley 2009). Freund et al. (2012) distinguish between process goal focus and outcome goal focus in that outcome goal focus emphasizes the higher level related to end state, while process goal focus concentrates on the lower level associated with the steps required to accomplish a certain action. Table 1 summarizes the key differences between the two types of goal focus (Freund et al. 2012). 25 Process Goal Focus Outcome Goal Focus Action/means End state Subordinate goals (concrete) Superordinate goals (abstract) Contextualized Decontextualized Provides vague or no standard of comparison Provides clear standard Provides guidelines for action Provides direction, meaning Table 1: Comparison between process and outcome goal focus (Freund et al. 2012) Zimmerman and Kitsantas (1999) study the impact of process and outcome goal on writing performance and self-motivation with the moderating effect of self-monitoring. They operationalized the process goal as “a three-step method for combining kernel sentences” and the outcome goal as “minimizing the number of words in the combined sentence” (Zimmerman and Kitsantas 1999, p. 241). Their findings indicate that learners in the shifting goal from process to outcome condition outperformed those in the process goal condition, which was superior to only those in the outcome goal condition, in both writing performance and self-motivation. Also, there was an interaction effect of goal with self-monitoring. Additionally, the goal focus is influenced by age. Freund et al. (2010) explore the impact of age—younger and older adults—on the goal focus to quit smoking. In their domain, the process focus goal involved how to quit smoking such as throwing away cigarettes and spending time with non-smokers, whereas the outcome focus goal pertained to the benefits of quitting smoking such as 26 saving money and improving health. Their findings indicate that a younger age group was more likely to focus on the process goal rather than the outcome goal. However, the reverse findings were found for an older age group. The organizational control research adopts the notion of process and outcome focus to evaluate employees’ performance and to give them feedback (Earley et al. 1990; Eisenhardt 1985; Kim 1984; Sawyer 1992). This helps the organization to structure the organizational strategy to manage human resources. For example, Anderson and Oliver (1987) investigated two types of salesforce control systems: outcome-based and behavior-based control systems. The outcome-based system focuses on “monitoring the final outcomes of a process”, while the behavior-based system emphasizes on “monitoring individual stages (e.g., behaviors) in the process” (Anderson and Oliver 1987, p. 76). In this sense, the behavior-based system is in line with the process focus goal and the outcome-based system parallels to the outcome focus goal. The authors propose that a decision to select a control system over the other depends on environmental context. The outcome control is suggested under the condition that the performance measures are objective and the behavioral control involves high costs; otherwise the authors recommend the adoption of the behavior control. Nevertheless, they postulate that the behavior control is more likely to better motivate employees and increase their performance, compared to the outcome control.  As another example, Eisenhardt (1985) examined the effect of the task characteristic and the ability to measure the outcomes on these two types of control systems based on organizational and economics theories. The findings suggest that there was a strong relationship between the task characteristic and organizational control strategy. That is, when the tasks can be structured, the organization should adopt the behavior-based control. On the contrary, if the tasks are less likely 27 to be programmed, the organization requires to use the outcome-based control. These two factors are in line with the differentiation between the process and outcome goal focus suggested by Freund et al. (2012) in that the process focuses on the task characteristic which is able to be structured into the detailed procedures, whereas the outcome pays much attention on the end result which can be measured. In addition, there are studies exploring the role of types of feedbacks based on the process and outcome focus on employees’ performance. Process feedback refers to "information concerning the manner in which an individual implements a work strategy" and outcome feedback pertains to "information concerning performance outcomes" (Earley et al. 1990, p. 88). Earley et al. (1990) found that the feedback types moderated the effect of goal on employees’ performance with process feedbacks showing the stronger impact on task performance in terms of the quality of information search and task strategy. However, they found that the moderating effect of goal and outcome feedback was stronger on employees’ self-confidence and effort. Moreover, the integration of process and outcome goal and feedback is evidenced to influence the performance and satisfaction of employees. Kim (1984) compared the impact of adopting either process or outcome and adopting both process and outcome goal setting and feedback. The findings support that there was a stronger influence of adopting both process and outcome goal and feedback than that of using either process or outcome goal and feedback on employees’ performance and satisfaction. In sum, the concept of process and outcome focus demonstrates the relationship between individuals’ performance and motivation that are related to persuasion context. It also parallels to 28 the notion of process- and outcome-focused simulation in mental simulation literature. This lends support that mental simulation may enable persuasion. Additionally, the impact of temporal distant on the process- and outcome-focused simulation (Ülkümen and Thomas 2013; Zhao et al. 2011) is in line with the well-known psychology theory, the construal-level theory (CLT) (Liberman and Trope 1998; Trope and Liberman 2010).  CLT (Trope and Liberman 2010), which will be discussed in more detail in the next section, proposes that the more the distant is, the more likely individuals rely on high-level construal. High-level construal refers to “relatively abstract, coherent, and superordinate mental representations” (Trope and Liberman 2010, p. 441) and is associated with individuals’ goal. On the contrary, low-level construal involves concrete representations which delineates “the details of how the action is to be performed” (Trope and Liberman 2010, p. 441). This type provides the information such as the specific actions, resources required to do those actions, and the context. Based on the definition of these two levels of construal, the higher-level construal focuses on the abstract goal, which is the outcome-focused thought, while the lower-level construal highlights the concrete information, which is about the process-focused thought. 2.5 Construal-Level Theory (CLT) Construal-level theory highlights the importance of psychological distance to an individual’s thoughts (Liberman et al. 2002; Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000, 2003, 2010). Psychological distance, defined as “a subjective experience that something is close or far away from the self, here, and now.” (Trope and Liberman 2010, p. 440), is composed of four dimensions: temporal, spatial, social, and hypotheticality (probability) (see 29 Liberman et al. (2007a) for more details about these four dimensions) . Liberman and Trope (1998) mainly focus their study on the effect of temporal distance on mental construal. In their study, temporal distance refers to the distance in time including the near-future event and the distant-future event and construal refers to mental representation of the event. They differentiate two levels of construal—low-level and high-level construals—based on whether individuals construe their thoughts regarding the near-future event or the distant-future event in the concrete or the abstract manner. The high-level construals are “relatively simple and coherent representations” including “general, superordinate, and essential features of objects or events” (Trope and Liberman 2000, p. 877). On the contrary, the low-level construals contain “more specific, subordinate, and incidental features of object or events” (Trope and Liberman 2000, p. 877). In other words, the high-level construals refer to the abstract mental representation, whereas the low-level construals refer to the concrete mental representation. Trope et al. (2007) found that there was the significant relationship between the temporal distance and the mental construals. That is, the low-level and the high-level construals are associated with the proximal and the distant future respectively. The level of construal is also distinguished in terms of the goal relevance (Liberman and Trope 1998). This is consistent with the Action Identification Theory (Vallacher and Wegner 1987, 1989), which suggests that there are two types of goals in representing the actions, subordinate and superordinate goals. The first goal involves “how” to perform the action, while the other regards the reason “why” to do the action. Liberman and Trope (1998) propose that the subordinate goal (means) is related to the low-level construals and the superordinate goal (end state) is associated with the high-level construals. People who engage in the near future event are more likely to think 30 about the goal-irrelevant information. On the other hand, those who engage in the distant future event are more likely to think about the goal-relevant information. Table 2 from Trope and Liberman (2003) summarizes the differences between both levels of construals.   Low-Level Construals High-Level Construals Concrete Abstract Complex Simple Unstructured, incoherent Structured, coherent Decontextualized Contextualized Secondary, surface Primary, core Subordinate Superordinate Goal irrelevant Goal relevant Table 2: Comparison between low-level and high-level construals (Trope and Liberman 2003, p. 405) CLT posits that temporal distance is the factor governing the level of construals (e.g., Liberman et al. 2002; Liberman and Trope 1998; Trope and Liberman 2000, 2003). Specifically, the more distant-future event is associated with the high-level construals and the near-future event is related to the low-level construals. As Trope et al. (2007) suggest, the relationship between temporal distance and mental construal is bi-directional. That is, temporal distance will induce mental construal in the same way as mental construal will influence temporal distance. 31 In addition, CLT demonstrates that temporal distance and mental construal affect individuals’ preferences, evaluations, and judgments on the event, the action, and the object. Prior CLT studies assert that with the same amount of the information, individuals make a decision with respect to the event, the action, and the object differently, as they focus on different pieces of information (e.g., low-level vs. high-level features related to the event or the object) when considering these things (Eyal et al. 2009; Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). Emphasis on different pieces of information is influenced by temporal distance as well as the level of mental construal. As a consequence, CLT may have an impact on individuals’ attitudes and behaviors by activating psychological distance and/or mental construal. 2.5.1 CLT and Preferences, Evaluation, and Judgment on Choices As temporal construal affects people’s thoughts, it has implications on effects to an individual’s preferences, evaluations, and judgment on choices (Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). CLT literature found overwhelming support that temporal construal influences consumers’ behavior, since it affects individuals’ preferences and decision-making on choices which involve trade-off between two components (e.g., feasibility vs. desirability consideration, primary vs. secondary feature or the product).  For example, according to Liberman and Trope (1998), individuals involve two types of consideration in choices—feasibility and desirability. Feasibility is defined as “the ease or difficulty of reaching the end state” and desirability refers to “the valence of an action’s end state” (Liberman and Trope 1998, p. 7). In this regard, feasibility and desirability are associated with low-level and high-level construals respectively. They hypothesized that feasibility consideration 32 will decrease over time, while the desirability consideration will increase over time. In other words, in the near-future situation individuals weight their decision more on feasibility consideration, whereas in the distant-future situation they are more likely to make a decision based on desirability consideration.  To illustrate this point, in one of their studies participants were asked to select two task assignments, one assigned today (near future) and another assigned 9 weeks later (distant future), from four options based on their preferences. Four assignments were manipulated in terms of feasibility (easy vs. difficult) and desirability (interesting topic vs. uninteresting topic). Consistent with their expectation, the easy-interesting task was chosen the most and the difficult-uninteresting task was chosen the least. However, the focus is on the results for the trade-off options—easy-uninteresting and difficult-interesting assignment. Consistent with their hypotheses, the results show that participants decreased their preferences toward the easy-uninteresting assignment over time and increased their preferences toward the difficult-interesting assignment over time. This lends support that when facing the trade-off decision between feasibility and desirability, individuals will prefer the more feasible but less desirable option in the near-future event and be more likely to choose the more desirable but less feasible option in the far-future event. As another study, Trope and Liberman (2000) argue that the features of the objects or the events consist of central features associated with the goal of using the objects or doing the actions and peripheral features not related to the goal. They showed in one study which asked participants to buy a radio tomorrow (near future) or one year from now (distant future). Two options were available including 1) a radio with good sound but bad built-in clock and 2) a radio with bad sound but good built-in clock. In this case, the sound quality is the central feature of purchasing a radio, 33 whereas the clock quality is the peripheral one which is irrelevant to the goal. The preferences toward the first option is higher for both the near and the distant future. However, consistent with the predictions of CLT, the results suggest that participants increased their preferences toward the first option over time and decreased their preferences toward the second option over time.  In sum, both studies support that temporal construal (temporal distance and mental construal) has a significant impact on individuals’ judgments. That is, in the near-future event people are more likely to weight more on the (feasibility) consideration and the (peripheral) feature associated with low-level construals, while in the distant-future event they tend to focus more on the (desirability) consideration and the (central) feature associated with high-level construals. 2.5.2 CLT and Attitudes, Self-Control, and Intention Although most CLT studies (except Fujita et al. (2008)) have not directly examined the effect of temporal construal on attitudes, self-efficacy, and intention, extant literature proposes that temporal construal may influence these variables. For instance, Trope and Liberman (2003) argue that unlike low-level construals, high-level construals may positively influence self-confidence, attitudes, and intentions. They reason that high-level construals are more decontextualized than low-level ones. This leads individuals who have less information with respect to the actions to perceive less confidence to predict the action in the near-future event which involves a lot of contextualized information than in the distant-future one. Accordingly, individuals who construe their thoughts at higher level may perceive more confident to perform the action than those who construe their thoughts at lower level. Also, the authors posit that attitudes have general structures which shape individuals’ behavior in general and that attitudes are the key predictor of intentions. 34 Thus, the high-level construals can better invoke individuals’ attitudes and thus intentions than the low-level ones.  In addition, Fujita et al. (2006) investigate the relationship between mental construals and self-control. In one of their studies, they primed participants with low-level vs. high-level construal mindsets and then asked participants to do another separate handgrip task. For the handgrip task, participants were told that they would receive the measure of their psychophysiological ability and also told that the accuracy of this measure depending on how long they could hold the handgrip. In this study, the accurate results of the measure is the delayed outcome associated with the high-level consideration, and the relief of the discomfort resulting from holding the handgrip is the immediate outcome relating to the low-level consideration. Consistent with the CLT’s prediction, those primed with high-level construal were more likely to hold the handgrip significantly longer than those primed with low-level construal. The results indicate that the high-level construal exerts higher self-control. Thus, mental construal will positively affect the level of self-control. In summary, to best of our knowledge, less attention has been paid to the relationships between temporal construal and attitudes, and between temporal construal and intention. Nevertheless, the results from prior studies imply that it is likely to influence people’s evaluation on choices that implicitly reflect their attitudes, and intentions by manipulating temporal construal. 2.5.3 CLT and Persuasion With the implication on influencing individuals’ preferences and judgment on choices, CLT implies that temporal construal may enable persuasion. For instance, Liberman et al. (2002) argue that “…the perceived value of an event derives from its construal, and that if the value of a high-35 level aspect of a target object is different from the value of its low-level aspects, then changing the level of the representation of the target object (e.g., by changing temporal distance) would result in a corresponding change in its perceived value.” (Liberman et al. 2002, p. 351). This suggests that changing the representation of the object that fits temporal distance may change individuals’ preferences toward the object.  As another example, Eyal et al. (2004) explore the effect of temporal distance on pros and cons considerations. They defined pros as “reasons for taking the action” and cons as “reasons against taking the action” (Eyal et al. 2004, p. 781).  In their study, consideration of pros is associated with high-level construal, whereas considerations of cons is related with low-level construal, because pros refer to desirability rather than feasibility and cons are contingent on pros. They found that participants considered more pros-related arguments than cons-related arguments for the distant-future event, while they recognized more cons-related arguments for the near-future event. Thus, they suggest that “…in order to make people view the festival more favorably, the advertisement should emphasize advantages (e.g., the quality of the movies) of the festival when still in the distant future and deemphasize the disadvantages of the festival (e.g., cost) when the festival is closer in time.” (Eyal et al. 2004, p. 795). In other words, for the far-future event pros should be emphasized, and for the near-future event cons should not be emphasized. To best of our knowledge, Fujita et al. (2008) is one study which explicitly examines CLT and attitudes in the context of persuasion. In their study, the persuasive arguments with low-level features and high-level features are the focus. They test the effect of temporal distance and the type of messages on attitudes toward the object. In one of their studies, they employed feasibility and desirability features of the object as low-level and high-level features respectively. Arguments 36 with feasibility feature and with desirability feature were operationalized by adding the statement indicating that the task product, a DVD player, having the easy-to-understand manual and the product made of environmentally-friendly materials respectively. The results reveal that there was a significant interaction effect between temporal distance and argument type on attitudes. Participants who made the purchase decision in the far-future event were persuaded more by the arguments with positive desirability feature than the argument with positive feasibility feature. However, there was no difference between those who read the positive desirability arguments and those who read the positive feasibility arguments in the near-future event condition. As a result, CLT can explain how to influence individuals’ attitudes by using the persuasive arguments.  Moreover, CLT is directly employed to enable persuasion in marketing literature. White et al. (2011) drew on CLT and the frame effect to persuade people to do recycling. To induce people to engage in the recycling behavior, they developed the message framed in terms of a positive frame, which states the benefits of doing recycling, and a negative frame, which suggests the consequences of not engaging in recycling. They also present the information regarding low-level construal, which focuses on how to undertake recycling, and high-level construal, which presents the reasons why people engage in this behavior, as well as the message primed with the near-future event (“Recycle for a better Calgary Today”) and the far-future event (“Recycle for a better Calgary Tomorrow”) (White et al. 2011, p. 477). The results of the study suggest that the match between high-level construal and positive-framed messages helped motivate people to engage in the recycling behavior, whereas the match between low-level construal and negative-framed messages induce people to do recycling as well. In other words, to motivate individuals to perform a specific action, the high-level construals should be presented with the positive frame, while the 37 low-level construals should be tied together with the negative frame. However, the researchers did not find the effect of mismatch between a construal type and a frame type on people’s intention to go recycling. Both mental simulation and CLT literature support that process-focused mental simulation/low-level construal and outcome-focused mental simulation/high-level construal influences how individuals think and behave. Both types of mental representation will engender positive attitudes and motivation to perform a certain behavior. Taken together with this concept, the cause-and-effect simulation, which presents the relationship between individuals’ decision-making on the product and its effect on resources (e.g., energy), will better persuade individuals, as it influences the individuals’ attitudes and behavior.  When we compare CLT and mental simulation, we found that CLT can apply to “any object or event, not only to instrumental actions” (Fujita et al. 2006, p. 352), whereas mental simulation focuses on “how” and “why” to perform an action. This suggests CLT is a broader concept than the mental simulation. Accordingly, we decided to adopt CLT to increase online consumers’ positive attitudes and motivations to purchase green products.  Based on CLT, the effect of the cause-and-effect simulation, enabled by technology, can be described in terms of the low-level and the high-level construals and the cause of the simulation would be captured by users’ decision on the input to the simulation, operationalized by the recommendation agent (RA). The low-level simulation highlights the proximal future and represents the concrete, short-term outcomes related to purchasing a product, whereas the high-38 level simulation emphasizes the distant future and provides the information regarding the abstract, long-term outcomes associated with buying a product. In our context, the green purchasing behavior involves trade-offs between prices and green attributes which pertain to energy and water consumption. As suggested by GreenBiz (2010) and Farrell (2012), prices of green products are critical factors dampening consumers’ purchasing green. We expect that presenting the concrete content through simulation may help online consumers reduce their cognitive effort to imagine the short-term (low-level) and the long-term (high-level) effect resulting from their decision regarding each product attribute (e.g., prices, and green attributes), and ensure that they elaborate on the correct information. 2.6 Framing Effect The frame of the message is hypothesized to affect individuals’ decision making. Tversky and Kahneman (1981) studied the impact of frame on decision making on choice. They defined “a decision frame” as “the decision-maker’s conception of the acts, outcomes, and contingencies associated with a particular choice” (Tversky and Kahneman 1981, p. 453) and found that the frame influences people’s decision making on choice. In framing of outcomes, outcomes can be presented in either positive or negative way. The positive-outcome frame focuses on gains, while the negative-outcome frame concentrates on losses. The underlying assumption for this framing is that people are rationale decision-makers who evaluate gains and losses. In persuasion, the message frame draws researchers’ attention. Extant research indicates that there is the effect of message frame on individuals’ attitudes. Marketing researchers study how to effectively frame the message to advertise products. Several frames are investigated such as 39 regulatory fit effect including promotion and prevention focus (Wang and Lee 2006); functional and hedonic frame (Belei et al. 2012); positive- and negative-outcome-focused frame (Roney et al. 1995); and gain and loss frame (Keller et al. 2003). Despite different terms, most of them refer to the similar concept which focuses on positive vs. negative effects.  For instance, Wang and Lee (2006) seek to clarify how the regulatory focus of consumers influences their search and decision-making behaviors. Promotion focus concentrates on “the presence or absence of positive outcomes”, whereas prevention focus delivers the information regarding “the presence or absence of negative outcomes” (Wang and Lee 2006, p. 28). They posit that the fit between consumers’ goal and the message frame draws greater attention. Belei et al. (2012) study messages focusing on either functional attributes or hedonic attributes such as “extra antioxidants” and “low fat” respectively to persuade a healthy consumption. They found the effect of framing on a consumption behavior. While health messages concentrating on hedonic attributes negatively affect a healthy consumption, those featuring functional attributes lead to the healthy consumption. Roney et al. (1995) examine the impact of outcome-focused frame on individuals’ emotion and motivation. They refer positive-outcome-focused frame to “the presence or absence of positive outcomes” and defined negative-outcome-focused frame in terms of “the presence or absence of negative outcomes” (Roney et al. 1995, p. 1152).  In addition, Keller et al. (2003) investigate the moderation effect on the relationship between the effectiveness of gain- and loss-framed messages on persuasion. They propose the message recipients’ state as the moderator. The loss frame has a greater effect on recipients under a positive mood, while the gain frame has a stronger impact on recipients under a negative mood (Keller et al. 2003). IS studies also examine the effect of gain and loss frame. For example, Lee and Benbasat 40 (2011) extend the effort-accuracy framework to evaluate the effects of recommendation agents which highlight trade-off difficulty on users’ intention to use recommendation agents. They found that decision context—loss and gain conditions—moderates the effects. There was a stronger effect of trade-off difficulty in a loss context than in a gain context. As another example, Xiao and Benbasat (forthcoming) investigate the positive- and the negative-framed messages in the context of online product recommendations. Specifically, they focus on designing warning messages to detect biased product recommendations. Moreover, the framing effect is studied in pro-environmental psychology aiming at promoting pro-environmental behaviors. For example,  Lindenberg and Steg (2007) identified three types of goal-frames which contribute to pro-environmental behaviors: hedonic, gain, and normative frames. The first type, a hedonic goal-frame, targets at how to improve an individual’s feeling. The second one is a gain goal-frame, which concentrates on how to improve an individual’s resources in terms of money, time, and status. This frames involves trade-off between costs and benefits that is a fundamental assumption of the theory of planned behavior (TPB) (Ajzen 1985, 1991). Finally, a normative goal-frame aims at promoting an individual to appropriately perform the behavior. It is hypothesized that if people have high environmental concerns, they are more likely to be influenced by this type of frame than others.  Lindenberg and Steg (2007) also determined the appropriate type of a goal-frame for a specific condition. They suggest that under high-gain decisions that involve high benefits to engage in the pro-environmental behavior, gain goal-frames are more appropriate, whereas under low-gain decisions that require less costs, normative goal-frames are more suitable to stimulate the pro-environmental behavior. However, they propose that pro-environmental behavior is shaped not by 41 a single goal, but multiple goals and that only if engaging in the behavior involves high gain, people are more likely to perform that behavior. In order to motivate pro-environmental behaviors, they suggest that it is required to fortifying normative goal-frames. This is in line with Stern et al. (1999) finding that suppressing individuals’ egoistic values will help promote pro-environmental behaviors.  In the current study, the loss or the negative frame is used to enable persuasion, although the cause-and-effect simulation can be framed in either gain or loss and White et al. (2011) found the fit between mental construal and the frame was better in persuading individuals to recycle more. We believe that negative-framed messages are more likely to enable persuasion than positive-framed ones. According to prospect theory (Kahneman and Tversky 1979; Tversky and Kahneman 1992), individuals are more sensitive to losses than gains. Therefore, the cause-and-effect simulation framed in loss term will be more likely to affect individuals’ decision-making. Also, Rogers (1975) argues that fear can induce persuasion in the protection motivation theory.  The more recent findings from Hardisty et al. (2015) reveal that the energy cost of operating the appliance will better inform consumers’ decision-making than the energy saving. In their project with BC Hydro, they tagged the 10-year operating cost of the appliance and observed its effect on consumers’ purchasing decision. The results indicate that consumers were more likely to purchase the greener or more energy-efficient appliance (e.g., light bulbs), as they considered this cost. This lends support that the cost which is framed in terms of loss can influence consumers’ decision-making. As a result, the focus of the current study is to develop the cause-and-effect simulation framed in terms of the lose frame which is the cost associated with operating the privately-consumed product, namely utility cost.42 Chapter 3: Theoretical Framework 3.1 Value-Belief-Norm Theory (VBN) For more than two decades, social science research has examined pro-environmental behaviors (Ellen et al. 1991; Kim and Choi 2005; Mainieri et al. 1997; Pichert and Katsikopoulos 2008; Stern 2000; Stern et al. 1999). Recently Froehlich et al. (2010), Loock et al. (2013), and Mithas et al. (2010) investigate how technology can promote the green behavior such as energy saving. According to Stern (2000), purchasing green is categorized into private-sphere environmentalism, which refers to “the purchase, use, and disposal of personal and household products that have environmental impact” (Stern 2000, p. 409). The subgroup of private-sphere environmentalism consists of the purchase of environmental-impact products, the use and maintenance of environmentally products, the disposal of household waste, the green consumerism, and other environmentally significant behaviors. Purchasing green is related to the purchase of environmental friendly products and the green consumerism.  Stern et al. (1999) developed a value-belief-norm theory (VBN) theory, a well-accepted pro-environmental theory, to motivate pro-environmental behaviors. The theory is developed based on Schwartz (1977)’s moral norm-activation model (NAM). Like NAM, VBN focuses on awareness of consequences, ascription of responsibility, and personal norms. The difference between NAM and VBN is that VBN differentiates three value orientations—egoistic, altruistic, and biospheric orientations, while NAM concentrates on an altruistic value orientation.  43 Figure 1 presents VBN. In VBN, an individual possesses three important values—egoistic, biospheric, and altruistic—which influence new environmental paradigm (NEP). NEP refers to the individual’s environmental concern belief that human beings make a negative impact on the environment (Stern et al. 1999). The environmental concern can be viewed as a general belief or as a specific belief which has a direct link to intention to perform a specific behavior (Fransson and Gärling 1999). According to Stern et al. (1999), NEP captures the broad relationship between human beings and the environment. Thus, it does not directly determine intention, but rather reflects a general belief.  Figure 1: Value-belief-norm theory (Stern 2000; Stern et al. 1999) Biospheric and altruistic values have positive impacts on NEP, while egoistic value negatively influences the belief. The belief will activate awareness of consequences, the belief that the individual perceives threats to others, and thus ascription of responsibility, the belief that she can act to diminish these threats.  Ascription of responsibility is in line with the notion of perceived 44 consumer effectiveness (Ellen et al. 1991; Kim and Choi 2005). Perceived consumer effectiveness (PCE) is defined as “a domain specific belief that the efforts of an individual can make a difference in the solution to a problem” (Ellen et al. 1991, p. 103). In other words, if a consumer beliefs that her action will create a positive impact, she will perform that behavior (Kim and Choi 2005). This is also in line with the concept of self-efficacy (Kim and Choi 2005). VBN postulates that when an individual has the responsibility belief, pro-environmental personal norms are formed that will positively affect pro-environmental behaviors.  Pro-environmental personal norms refer to “feelings of personal obligation that are linked to one’s self-expectations” (Stern et al. 1999, p. 83). VBN has the explanatory power to explain the pro-environmental behaviors when there are constraints preventing an individual from performing those behaviors. In case of few or lack of constraints, VBN may lose the power. This is in a similar vein of NAM. Guagnano et al. (1995) study recycling behavior and found that in the condition that recycle bins were provided for people, NAM showed less explanatory power, compared to in the condition that recycle bins were not provided. In other words, if people found that performing recycling behavior was convenient, they were likely to perform the behavior. Thus, NAM will not help explain their behavior.  In the context of motivating online consumers to purchase green products that are privately consumed, VBN would explain how and why consumers purchase green products. The major constraint which prevents purchasing green is price (Farrell 2012). This may result in the difficulty to purchase green products. Therefore, VBN will help explain this green behavior. 45 3.2 Theory of Planned Behavior Although VBN can explain how individuals will form pro-environmental behaviors, Stern (2000) points out personal capabilities affect the behaviors. Consequently, the integration of VBN with the theory of planned behavior (TPB) will fully reveal the persuasion process. In TPB, Ajzen (1985) posits that only attitudes toward behavior and subjective norms do not guarantee a behavior. Perceived behavioral control influenced by control beliefs is added to the theory of reasoned action (TRA) in order to explain individuals’ behavior, especially one which requires control to achieve. Perceived behavioral control is defined as “the degree to which individual perceive the difficulty in performing the behavior” (Ajzen 1985, p. 189). Without volitional control over performing a certain behavior, an individual will be less likely to perform that behavior although she has positive attitudes toward performing that behavior and believes that her group reference favors that behavior. Volitional control refers to the situation that people have enough behavioral control, thereby leading to the success in performing a behavior (Ajzen 1991). Later perceived behavioral control is redefined as  “perceived control over performance of a behavior” (Ajzen 2002, pp. 667-668). Ajzen (2002) frames the construct to include two components—self-efficacy and controllability. Self-efficacy refers to an individual’s beliefs regarding her own capabilities to successfully perform a certain action (Bandura 1977, 1989). Another component, controllability, is defined as an individual’s evaluation with respect to the availability of resources and opportunities required to perform a certain action (Ajzen 2002). Figure 2 depicts the key constructs of TPB. 46  Figure 2: Theory of planned behavior (Ajzen 1991) TPB proposes that attitude toward performing a behavior, defined as the degree to which an individual has a positive assessment on a certain behavior, subjective norm, referring to the degree to which an individual perceives social pressure to perform the behavior, and perceived behavioral control will positively influence intention to perform the behavior. Intention will predict the behavior. Ajzen and Madden (1986) support the direct impact of perceived behavioral control on the behavior. However, they found partial support of the direct impact of perceived behavioral control on the behavior. Thus, they reason that only in case that an individual accurately evaluates her control and the target behavior is non-volitional, intention will not directly lead to the behavior, but instead intention together with perceived behavioral control will cause the behavior (Ajzen and Madden 1986). TPB is extensively adopted in various fields including IS. For example, Mathieson (1991) compares the power of TPB with that of technology acceptance model (TAM) (Davis 1989; Davis et al. 1989) to explain individuals’ adoption of information systems. He suggests that though TAM is easier to apply, TPB gives richer details. As another example, Pavlou and Fygenson (2006) use TPB to predict e-commerce adoption by individual users. 47 In pro-environmental research, TPB is adopted to explain various pro-environmental behaviors. For instance, Cordano and Frieze (2000) adopted TPB to investigate factors contributing to a pollution reduction behavior of US environmental managers. They found significant positive effects of pollution prevention attitudes and of perceived norms for environmental regulation. As another example, Han et al. (2010) examine the behavior of choosing green hotels. They used TPB to explain this behavior and found that attitudes, subjective norms, and perceived behavioral control have an impact on people’s behavior to stay in green hotels. To integrate VBN with TPB to motivate purchasing green online, the subjective norm is removed, since the domain of the current study is to study how individuals purchase green products which are privately-consumed online. This behavior is private and the product is privately-consumed, and thus the normative influence does not affect individuals’ attitudes and behaviors (Bearden and Etzel 1982). Combining these two theories might help improve our capability to explain the pro-environmental behavior.  Based on Lindenberg and Steg (2007), VBN is good at predicting the behavior under low-gain conditions, while TPB is more appropriate to explain the behavior under high-gain decisions. That is, when engaging in the pro-environmental behavior involves high gains, TPB has more powerful to predict the behavior. In contrast, if performing that behavior requires low costs and efforts, VBN will have more power than TPB.  In our study, purchasing a green product involves costs as well as trade-off difficulties between costs and green attributes. Although TPB can explain purchasing behavior, it does not assist to promote the behavior associated with low gains and high costs. Thus, complemented with TPB, 48 VBN which is good at predicting pro-environmental behaviors may reveal this phenomenon more completely. 3.3 Summary of Literature Foundational to this Study The current study aims to promote an online green purchase of a privately-consumed product in order to mitigate environmental problems. To motivate online consumers to develop positive attitudes and behaviors with respect to the green purchase, we will develop and test a cause-and-effect simulation design. Fogg (2002) proposed that one form of persuasive technology can be interactive simulations that allow users to explore cause-and-effect relationships. According to Fogg (2002), users can learn the relationship between the cause and the effect, thereby leading to persuasion without forcing. Simulation was found by the prior literature that it will positively affect learning and therefore learners’ attitudes which are the key of persuasion  (e.g., Ajewole 1991). This cause-and-effect simulation will be  developed based on the Construal-Level Theory (CLT, Liberman and Trope 1998) as  temporal distance as well as mental construals can influence individuals’ preferences and evaluations on choices (Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). This suggests that the cause-and-effect simulation created based on CLT could enable persuasion.  According to Liberman and Trope (1998), those who think about far-future decisions will focus their decisions more on desirability considerations than on the feasibility considerations. In the context of online green purchasing, the price of the product parallels with the feasibility as it involves the ease or the difficulty to purchase a product. Conversely, green attributes (e.g., energy 49 use) is associated with the outcomes resulting from purchasing a product and thus is in line with the desirability consideration.  In addition, the key information of the effect part of the cause-and-effect simulation will be framed in terms of loss. Prior studies found that individuals are more sensitive to loss than gains (Kahneman and Tversky 1979; Tversky and Kahneman 1992). This effect part will also be framed based on the construal level (low-level vs. high-level). As a result, presenting the effect part in terms of loss will greater influence individuals’ decision-making about the product and thus have an impact on the green product purchase. For the cause part of the cause-and-effect simulation, the recommendation agent will be used as it presents how individuals make a decision about the product. It also facilitates individuals’ decision-making by reducing people’s effort (Lee and Benbasat 2010; Xiao and Benbasat 2007).  VBN and TPB will be used to explain why and how the cause-and-effect simulation will positively influence the online green purchase. Both theories are used in the prior studies to explain the pro-environmental behaviors (e.g., Cordano and Frieze 2000; Stern et al. 1999). In the next chapter, we will discuss the cause-and-effect simulation design as well as the research model in more detail. 50 Chapter 4: Hypotheses Development In the context of the current study, we employ an online simulation which demonstrates the cause-and-effect relationship between an individual’s decision on product attributes and its effect on utility cost associated with operating the product. We purport that such a cause-and-effect simulation will promote online green purchasing. The individual’s decision with respect to the product reflects the cause of this relationship and the utility costs reported back (and how they are reported back) is termed as the effect part of the relationship. Integrating the VBN theory and TPB, we obtain our research model, as depicted in Figure 3.  Figure 3: Research model We will mainly focus on motivating online consumers who do not have strong product preferences. If consumers have a strong preference regarding the product, they are more likely to buy the product based on their preference. This is supported by Zhao et al. (2011) proposing that “For products for which consumers have stable existing preferences, consumers often do not need to undertake a deliberative evaluation stage before choice and are less susceptible to the evaluation context.” (Zhao et al. 2011, p. 827). This indicates that these consumers will be less likely to be 51 persuaded to buy the product that is not consistent with their preference. Thus, the target of our study is online consumers who do not have specific product choices in their minds. Also, the brand and the look of the products used in the current study are concealed and controlled such that they will not influence consumers’ decision.  In order to help these consumers to make a product decision and persuade them to buy green products, we provide them with a recommendation agent (RA). The RA refers to “software agents eliciting the interest or preferences of individual users for products, either explicitly or implicitly, and make recommendations accordingly” (Xiao and Benbasat 2007, p. 137). Also, Fogg (2002) suggests that the cause-and-effect simulation which clearly and immediately presents the causal link enables people to explore and to experiment, thereby changing their attitudes and behaviors.  Thus, incorporating with RA, the simulation will persuade online consumers to buy green products. According to Escalas (2004), mental simulation will influence positive evaluation, since mental simulation is found to result in persuasion through transporting individuals to simulate their thoughts toward the action, thereby reducing negative cognitive response while increasing positive affect simultaneously. In a similar vein, mental construals referring to mental representation is evident that it has an impact on consumers’ judgment and decision-making (Eyal et al. 2009; Liberman et al. 2002; Trope and Liberman 2000, 2003). Liberman et al. (2002) propose that manipulating the level of mental presentation will change how individuals evaluate the action, the event, or the object. As a result, both mental representation concepts (mental simulation and mental construals) will explain why the cause-and-effect simulation will enable persuasion. That is, it will influence individuals’ evaluation through the way they construe their thought. 52 Unlike mental representation which manipulates an individual to imagine the situation by herself based on the messages exposed, the simulation of cause-and-effect facilitates the mental presentation process by providing clear relationships between the causes and the effects rather than allowing individuals to develop thoughts by themselves. This will reduce an individual’s cognitive effort and thus transport the individual to the situation created by a simulation tool. This also helps ensure that the individual will think about the targeted situation more effectively. The cause-and-effect simulation will help change users’ attitude toward buying green products and green purchasing behavior. Simulation intended to display the relationship between users’ product decision and its impact on resources in terms of utility cost will positively influence users’ judgment and decision-making on product alternatives. Two designs of the cause-and-effect simulation are developed based on CLT. According to Trope et al. (2007), the relationship between temporal distance and mental construals is bi-directional. This means that temporal distance can trigger mental construals in the same way as mental construals influence temporal distance with the distant future associated with low-level construals and the proximal future related with high-level construals. Therefore, we combine both temporal distance and mental construals to design the cause-and-effect simulation.  The first design is the low-level simulation which focuses on the proximal future outcomes. This design parallels with the near future and low-level construals. In this design, the RA and the utility cost (e.g., utility cost per load) are presented in terms of a load of laundry or a proximal time frame.  53 The other design is the high-level simulation which concentrates on the distant future outcomes and high-level construals. This design presents the RA and the utility cost in a longer time frame (e.g., 10-year utility cost). In addition, to establish the baseline of online consumers’ behavior and to control for the effect of the RA itself, we create additional two designs, one without the RA and the utility cost and another only with the RA. The former is called the no simulation, as it does not show the cause-and-effect relationship. The latter is called the partial simulation, since it is incorporated with the RA which reveals the cause part without the effect part of the relationship. 4.1 The Impact of Cause-and-Effect Simulation on Desirability Consideration In our green purchasing context, online consumers generally face the trade-off between price and green attributes. According to Farrell (2012), green products cost more than non-green alternatives and the higher prices of the green products obstruct consumers to go green. Thus, price is the attribute which is most related to feasibility considerations, as it reflects how difficult it is to buy a product. Products with a higher price generally make consumers perceive as being more difficult to purchase, while products with a lower price may not invoke the same difficulty to buy in the consumers’ view. Green attributes (e.g., energy use, water use) which reveal the end-state of purchasing a green product are associated with desirability consideration.1                                                   1 We ran the pilot study to test whether how individuals perceive each product attribute in terms of feasibility and desirability. Consistent with our expectation, price was perceived more on feasibility (less desirability) and green attributes (e.g., energy use and water use) lied more toward desirability attributes. 54 As one study, Fujita et al. (2008) use a green attribute of a DVD player to indicate the desirability. In their study, the DVD player made of environmentally-friendly materials is considered to be the desirability aspect of the product. This reveals that green attributes indicate the desirability. The purpose of our study is to motivate online consumers to sacrifice the price for the green attributes (e.g., energy use and water use). In other words, we expect the proposed design of the cause-and-effect simulation will persuade online consumers to pay more for green products. CLT suggests that the feasibility consideration reflects the low-level feature of an event, an action, or an object, whereas the desirability consideration indicates the high-level feature (Liberman and Trope 1998; Trope and Liberman 2003; Trope et al. 2007). Liberman and Trope (1998) found that “Participants generally attached greater significance to desirability considerations than to feasibility considerations, but this difference was more apparent for distant future decisions than for near future decisions.” (p. 12). In other words, individuals who think about the far future will be more likely to make a decision based on the desirability consideration than feasibility consideration, while for those who think about the near future there is no different weight between feasibility and desirability consideration.  However, several CLT studies found that for the alternatives which involve trade-offs between feasibility and desirability aspect temporal distance which invokes mental construals will affect how individuals consider these two aspects (e.g., Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). That is, individuals who make a decision for the near future are more likely to base their decision more on feasibility than desirability, whereas the preference is reverse for those who made a decision for the far future event.  55 As the green purchase is associated with high-level feature which is a desirability consideration, we focus on the desirability aspect and hypothesize that online consumers who are exposed with the high-level simulation will be more likely to have higher desirability consideration than any other designs, while those who are given the low-level simulation will be less likely to have desirability consideration than any other conditions. There will be no difference in the desirability consideration between the no simulation and the partial simulation. H1a: The low-level cause-and-effect simulation will negatively affect an online consumer’s desirability consideration toward purchasing a green product more than other designs. H1b: The high-level cause-and-effect simulation will positively affect an online consumer’s desirability consideration toward purchasing a green product more than other designs. H1c: There will be no difference between the no cause-and-effect simulation and the partial cause-and-effect simulation in an online consumer’s desirability consideration toward purchasing a green product. 4.2 The Impact of Desirability Consideration on Attitudes toward Purchasing a Green Product and Self-efficacy According to TPB, Ajzen (1985) indicates that “The attitude toward the behavior is determined by the person’s evaluation of the outcomes associated with the behavior and by the strength of these associations.” (Ajzen 1985, p.13). This reflects the link between an individual’s evaluation of the consequences regarding the behavior and attitude toward the behavior. Under TPB’s assumptions, people are rationale decision-makers and weight between benefits and costs associated with 56 performing a behavior. In other words, perceived positive outcomes resulting from performing a behavior will positively lead to positive attitudes toward performing that behavior, whereas perceived negative outcomes caused by performing the behavior will negatively result in negative attitudes toward performing the behavior (Ajzen 1991). If the benefits outweigh the costs, these people will have more positive attitudes with respect to the action. In a similar vein, Bandura (1977) indicates that “the capacity to represent future consequences in thought provides one cognitively based source of motivation. Through cognitive representation of future outcomes individuals can generate current motivators of behavior.” (Bandura 1977, p. 193). This means recognizing future consequences motivates people to perform a behavior. Both studies highlight the importance of the outcome recognition in motivating individual’s behavior. Consistent with TPB and Bandura (1977), the VBN theory and perceived consumer effectiveness (PCE) propose that awareness of consequences of an individual’s action will influence the individual’s attitudes, intentions, and behaviors (Ellen et al. 1991; Kim and Choi 2005; Stern 2000; Stern et al. 1999). Thus, the cause-and-effect simulation can exert the power to create an individual’s awareness toward a green purchase through providing the relationship between the individual’s decision about the product and its impact on the resources consumed by that product, thereby influencing attitudes toward purchasing green.  In our study, the low-level simulation which focuses on the immediate outcome in a very short time frame will less likely to promote the desirability consideration. On the other hand, the high-level simulation which emphasizes the delayed outcome in a very distant time frame will lead to higher consideration of desirability. Desirability is related to the outcome associated with the action, whereas feasibility is associated with how hard it is to accomplish the outcome (Liberman 57 and Trope 1998). As the desirability consideration, like awareness of consequence, is associated with the end state, the cause-and-effect simulation developed based on the CLT concept will influence the desirability consideration and thus attitudes toward purchasing green in the same way as awareness of consequences does. This is also supported by CLT which asserts that there is the relationship among temporal distance, mental construals, and attitudes (Trope and Liberman 2003).  As one CLT study, Fujita et al. (2008) investigate the relationships between temporal distance, mental construals, and attitudes, and found significant support. The results from their study indicate that participants in the far-future event focusing more on the desirability had more positive attitudes toward the product than those in the near-future event emphasizing the feasibility more than the desirability. This shows the desirability consideration will positively influence attitudes. In sum, the cause-and-effect simulation provides the information about the outcome in terms of the utility cost associated with the purchasing behavior. This will affect an online consumer’s desirability consideration and also allow the consumer to observe the impact of her decision making on the utility cost. As a result, the consumer will be more likely to develop positive attitudes toward buying a product which consumes less utility (e.g., energy and water). Accordingly, the cause-and-effect simulation that presents the impact of people’s decision making on the utility consumption will help persuade them to buy green products which consume less utility. Thus, we postulate that online consumers’ desirability consideration will increase positive attitudes toward a pro-environmental behavior. H2a: Desirability consideration will lead to attitudes toward purchasing a green product. 58 However, only positive attitudes toward purchasing green does not warrant actual buying (Forrester Research 2008). According to Stern et al. (1999) and Stern (2000), personal capabilities such as knowledge, skills, time, and resources can hinder pro-environmental behaviors. Personal capabilities are related to an individual’s perceived behavioral control.  In a similar fashion, Bandura (1994) developed the notion of self-efficacy asserting that an individual will be more likely to perform a certain action if the individual perceives that her action can generate the positive outcomes. Without such perception, the individual will have less incentive to perform that action or experience difficulties to perform the action. Perceived behavioral control (Ajzen 1991, 2002) is developed based on this premise.  Ajzen (1991) identified perceived behavioral control as a key factor which helps explain the relationship between attitude-intention-behavior. Self-efficacy also underlies the notion of ascription of responsibility and the concept of perceived consumer effectiveness proposed by Ellen et al. (1991). The NAM and the VBN theory support that awareness of consequences will positively influence ascription of responsibility. Both awareness of consequences and ascription of responsibility are two key factors contributing to the pro-environmental behaviors. In line with ascription of responsibility, perceived behavioral control is evident to affect consumers’ green purchase (Ellen et al. 1991; Kim and Choi 2005). As another example, White et al. (2011) pointed out the importance of an individual’s perceived control in influencing their recycling behavior. This suggests that self-control plays an important role in persuading individuals to perform an action. 59 Moreover, several studies suggest that simulation will positively affect learners’ control.  For example, Fogg (2002) mentions that rehearsal is the benefit of simulation to persuade individuals to perform the target behavior. Allowing individuals to practice or to rehearse behaviors results in an individual’s confidence to perform such behaviors in reality. Consistent with Fogg (2002), a research in nursing education reveals that instructional simulations affect learners’ self-efficacy (Goldenberg et al. 2005), which is one of perceived behavioral control’s components (Ajzen 2002). Various studies of mental simulation lend support that mental simulation allows rehearsals of the events leading to enhance self-regulatory behavior (Escalas 2004; Pham and Taylor 1999; Taylor and Pham 1996). This means that the simulation allows learners to iterate their actions or decision making, thereby leading learners to feel that they possess capabilities to achieve certain practices.  In line with TPB, environmental studies, and simulation studies, CLT posits that mental construals will influence individuals’ self-control (Fujita et al. 2006; Trope and Liberman 2003; Trope et al. 2007). That is, the high-level construals will induce higher self-control than the low-level ones. For instance, Fujita et al. (2006) found support for this proposition that participants with high-level construals are more likely to control their actions than those with low-level construals. As a result, we propose that if online consumers are aware of the outcome resulting from their purchase, the desirability consideration, they are more likely to have a greater control over their behavior. To capture perceived behavioral control in our study, the concept of self-efficacy which is the ground of other control constructs, is chosen. H2b: Desirability consideration will lead to self-efficacy. 60 4.3 Attitudes, Perceived Behavioral Control, Intention, and Behavior As suggested by TPB, attitudes toward performing the behavior and perceived behavioral control will positively lead to intention and thus the actual behavior. This receives considerable support from IS research in predicting the users’ intention to adopt the technology (e.g., Mathieson 1991; Pavlou and Fygenson 2006). This notion is used to explain the pro-environmental behaviors as well  (Cordano and Frieze 2000; Han et al. 2010). Thus, we hypothesize that attitudes toward purchasing a green product and self-efficacy will lead to intention to buy a green product. H3a: Attitudes toward purchasing a green product will affect intention to purchase a green product. H3b: Self-efficacy will positively influence intention to purchase a green product. Additionally, intention is evident by various studies that it is a strong predictor of an actual behavior (Ajzen 1985; Ajzen and Madden 1986; Fishbein and Ajzen 1975). In line with this proposition, Bamberg and Möser (2007) purport that behavioral control, as well as attitudes and personal norm, is a predictor of pro-environmental behavioral intention which, in turn, influences the pro-environmental behavior. Thus, intention will lead to an actual green purchase. However, this study did not measure participants’ actual purchase. 4.4 The Impact of Cause-and-Effect Simulation on a Green Product Choice CLT predicts that although people in general consider the desirability feature associated with choices more than the feasibility feature, this consideration will be more salient for those who 61 involve the high-level temporal construal (e.g., Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000, 2003). In other words, individuals who construe their thoughts at higher-level or think about the future behavior will be concerned about the desirability regarding the goal of performing the behavior more than the feasibility. To promote the green purchase online, providing consumers with the cause-and-effect simulation which highlights the long-term outcomes will allow consumers to construe their thoughts at higher-level and thus focus more on the desirability aspect of the product which in this case is the green attributes. On the contrary, the cause-and-effect simulation which provides the low-level information will be less likely to induce consumers’ desirability consideration compared to the high-level simulation. This results from the less salient effect of desirability for the low-level construal. As a result, the high-level construal will be reinforced by the presence of the high-level simulation. Thus, we hypothesize that compared to any other designs the high-level simulation will motivate online consumers to buy more green products. H4a: The high-level cause-and-effect simulation will motivate an online consumer to purchase a green product more than other designs. According to TPB, the presence of new information and timing may affect the correlation between people’s intention and actual behavior. Ajzen (1985) proposes that when the new information is available, people are influenced by the information and more likely to modify their attitudes, intentions, and behavior. He identifies timing which may influence this relationship, since people will be more likely to receive the new information overtime. The shorter lag between the point people form their attitudes and the time to perform the behavior is, the more likely people perform 62 that behavior. Thus, presenting the utility cost information associated with purchasing behavior through simulation approximately at the time of purchase may directly influence online consumers’ behavior.  Though the low-level simulation will be not as powerful as the high-level simulation in reinforcing the desirability consideration which, in turn, increases attitudes, self-efficacy, and intention to buy a green product, it presents the useful information about the resources needed to run the product. Thus, it will influence online consumers to choose a green product more so than the design which does not provide this useful piece of information. Based on this, we propose that the low-level simulation design will be better than both the no-simulation and the partial simulation that do not give this information.  H4b: The low-level cause-and-effect simulation will motivate an online consumer to purchase a green product more than the partial and the no simulation. As the partial simulation does not present the cause-and-effect relationship between online consumers’ decision-making on the product and its impact on utility cost, it will not differ from the no simulation design which does not provide both cause and effect parts of this relationship. Therefore, we argue that no difference in online consumers’ product choice will be found between these two designs. H4c: There will be no difference in the degree of a green purchase between the partial cause-and-effect simulation and the no simulation. 63 Chapter 5: Research Method The first seven hypotheses (H1a – H3b) proposed in the current study were tested by a website experiment with a 4 between-subject design (4 types of the simulation design) and the last three hypotheses (H4a – H4b) were assessed by the a 4 x 2 mixed design (between-subject: 4 types of the simulation design, within-subject: 2 points of time including pre and post website exposure).  The types of the simulation design involves the website interface: 1) without the RA and the utility cost, 2) with the RA but without the utility cost, 3) with the RA and low-level utility cost (the utility cost per load), and 4) with the RA and high-level utility cost (the 10-year utility cost). Table 3 presents all four conditions regarding the design manipulation.  The two time points, before and after the website interface manipulation, were captured to measure whether the website interface can change participants’ thoughts regarding purchasing a green product. Participants were randomly assigned into four conditions: 1) no cause-and-effect simulation (condition 1), 2) partial cause-and-effect simulation (condition 2), 3) low-level cause-and-effect simulation (condition 3), and 4) high-level cause-and-effect simulation (condition 4). The detailed design is described below. 64   Utility Cost (Effect of the Cause-and-Effect Simulation Tool)   Without With    Low-Level High-Level RA (Cause of the Cause-and-Effect Simulation Tool) Without Condition 1 - - With Condition 2 Condition 3 Condition 4 Table 3: Experimental conditions 5.1 Task Product In all conditions, participants were required to select a washing machine for themselves. This context provides for a privately-consumption experience, as noted in the introduction section. That is, the purchase and use of the product will not be observed by others. Washing machines are a functional product that involves water and energy consumption. According to American Water Works Association Research Foundation (1999), washing machines accounted for 21.7% of household water use. In a similar fashion, Environment Canada (2014) supports that laundry makes up 20% of water use in the home. Washing machines do not consume only water, but also electricity. Data from Wikipedia (2014) suggest that their use of electricity to heat water accounts for a large amount of household energy usage. Also, Korn and Mattison (2012) indicate that residential washing machine operation (e.g., agitation) consumes 6% of total energy used in doing laundry, whereas heating water accounts for 13%. However, the greatest proportion of energy usage, 81%, is spent in drying clothes. As a result, the moisture content of clothes washed by the 65 machine plays an important role in determining total energy used to do laundry, since it will influence the electricity use of the electric dryer. Owing to these reasons, we chose the washing machine as a product in our study. 60 models of the washing machine were used in our study based on the information of the real washing machines in the marketplace and manipulated the information for the purpose of our study. That is, all washing machine models involved the trade-off between price (feasibility consideration) and green attributes which constitute the utility cost (desirability) associated with running the washing machine and the electric dryer. In other words, the model that has a lower price (more feasibility) consumes more electricity and water (less desirability) as well as negatively influences energy used by the electric dryer, while the one that costs more (less feasibility) causes less electricity and water (more desirability) in the overall process.  Seven product attributes were selected, as individuals can process approximately seven chunks of information simultaneously (Miller 1956). Also, seven attributes were not too easy for participants to process the information without IT. The attributes include price, energy use, cycle time, spin speed, sound level, water use, and warranty. The information regarding these seven attributes were derived from real product information actually available in the market and manipulated such that it involved a trade-off between the feasibility attribute (price) and the desirability attribute (green attributes). Energy use, spin speed, and water use are the green attributes which are the input to calculate the utility cost. In case of spin speed, Schmitz and Stamminger (2014) suggest that there is the relationship between spin speed and energy consumption in the drying process of laundry. They 66 found that the higher the spin speed is, the less moisture left in laundry and thus the less energy used in drying process is.  Price was chosen to reflect the feasibility to buy a washing machine. According to the results of the scenario study we conducted to test the concept of CLT without the role of IT (see Appendix A), price was perceived as the most important factor influencing how individuals made a decision on the washing machine (M = 9.02, SD = 1.50). Other attributes—cycle time, sound level, and warranty—were manipulated such that they have no significant relationships with the green attributes or price. They were included to fill up the seven attributes. The unit of some product attributes can be presented in terms of either low-level or high-level construals. Table 4 presents the seven attributes of the washing machine and Table 5 presents the relationships between the seven washing machine attributes. 67 Product Attributes Low-Level Unit High-Level Unit Price* $ $ Energy use** kWh/load kWh/year Cycle time^ minutes hours Spin speed** round per minute round per minute Sound level^ decibel decibel Water use** liters/load liters/year Warranty^ months years * reflects feasibility; ** reflects desirability; ^ does not reflect feasibility and desirability Table 4: Washing machine attributes 68  Energy Use Cycle Time Sound Level Spin Speed Water Use Warranty Price Energy Use 1       Cycle Time 0.11 1      Sound Level -0.07 .04 1     Spin Speed -0.94** -0.16 0.02 1    Water Use 0.83** 0.24 -0.21 -0.78** 1   Warranty 0.10 0.08 -0.12 -0.11 -0.03 1  Price -0.95** -0.16 0.02 0.94** -0.82** -0.07 1 **Correlation is significant at the 0.01 level (2-tailed). Table 5: Product attribute correlations In order to reduce participants’ preference on the product brand, we concealed the brands of the product and referred to each model by the model number. The product images were also randomly presented to reduce the impact of the product look on participants’ decision-making.2 Also, the                                                  2 Six product images which were very similar in terms of design and color were used and randomly presented across all conditions for 60 products. 69 scenario (see Appendix D) was employed to mitigate any other effects on people’s decision-making such as the capacity and the load access of the washing machine. 5.2 Experimental Website Design In all conditions, participants were required to select a washing machine for themselves, within the context using an online retail interface. This context provided for a privately-consumption experience, as noted above. That is, the purchase and use of the product will not be observed by others.  The no simulation condition (condition 1) did not provide the RA to participants. The website allowed them to see the information for each product model. They were free to choose whatever product model they wanted. The product models were randomly presented for each participant in order to reduce the product order effect. Figure 4 presents the website interface for the no simulation design. 70  Figure 4: No cause-and-effect simulation (condition 1) The website provided the RA to three conditions (condition 2, 3, and 4) to help participants find the right model based on their preferences. The RA is created in terms of attribute-driven design which elicits participants’ product attribute preferences and then gives the product recommendations accordingly (Lee and Benbasat 2010). There are two stages associated with the RA—the input and the output stage. The input stage allows participants to assign the value and the weight for each product attribute in terms of the ranking (1 = the most important attribute to 7 = the lease important attributes). Participants could hover over each product attribute to see the brief explanation (Figure 5). 71  Figure 5: Example of the product attribute explanation The output stage provides the product recommendations based on participants’ preferences and ranked by fit scores in descending order. We focus on the input stage and aim at motivating participants without force to assign high values and weights for green product attributes such as energy use, water use, and spin speed. The difference among three conditions is the cause-and-effect simulation tool provided by the website. In the partial simulation (condition 2), participants were able to use the RA to input value for and rank each product attribute without the simulated effect resulting from their decision, the utility cost3. On the other hand, participants in other two conditions (the low-level and the high-level simulation) could observe the effect resulting from their decision, the utility cost, regarding the input to the RA. The utility cost for both full cause-and-effect simulation is described in the                                                  3 We randomly presented the low-level and the high-level RA for participants in condition 2 to mitigate the effect of the unit of some product attributes (e.g., energy use, cycle time, water use, and warranty). The low-level RA presented the product information in terms of low-level unit. On the other hand, the high-level RA showed the product information in terms of high-level unit. 72 next section. Figure 6, Figure 7, and Figure 8 represent the website interface for the partial simulaton, the low-level simulation, and the high-level simulation respectively.   Figure 6: Partial cause-and-effect simulation (condition 2) 73  Figure 7: Low-level cause-and-effect simulation (condition 3) 74  Figure 8: High-level cause-and-effect simulation (condition 4) 5.3 Cause-and-Effect Simulation Manipulation (Condition 3 and Condition 4) Two types of the full cause-and-effect simulation are developed based on CLT. As Liberman and Trope (1998) proposed, temporal distance will trigger mental construals influencing an individual’s preferences toward the object or the action. They found that priming participants with the far future induces high-level construals, thus increasing the importance toward the desirability consideration. On the other hand, priming participants with the near future leads to low-level 75 construals, therefore decreasing the significance of the desirability consideration. This consideration has the direct impact on the individual’s preferences on choices. In our study, we design the simulation tool which is governed by the RA and presents the effect caused by participants’ decision with respect to product attributes. Before collecting data to test the effect of the simulation design, the scenario study was conducted to evaluate the effect of low-level and high-level construals without IT (see Appendix A). The results from our scenario study reveal that many participants considered energy and water consumption (8.9% of total number of thoughts) as well as energy and water cost (26.41% of total number of thoughts) when making a purchase decision. In addition, the study in progress by Hardisty et al. (2015) and his colleagues found that the loss-framed message as the 10-year energy cost motivates consumers to buy the energy-efficient choice better than the cost saving.  As a result, the utility cost (energy and water costs) associated with running the machines (a washing machine and an electric dryer) to do laundry is employed to reflect the effect part of the simulation tool.  The RA is regarded as the cause of the simulation tool. Designed based on CLT, the effect presents the utility cost in terms of either low-level, the utility cost per load ($), or high-level, the 10-year utility cost ($).The utility cost per load refers to “What you might pay to run the washing machine and the electric dryer per a load of laundry, based on the overall electricity and water use and the national average cost of energy. This utility cost is compared to the cost of operating other similar washing machine models.” The 10-year utility cost refers to “What you might pay to run the washing machine and the electric dryer for 10 years, based on the overall electricity and water use and the national average cost of energy. This utility cost is compared to the cost of operating other similar washing machine models.”  76 The utility cost is calculated based on participants’ product preferences toward and ranking for each product attribute (see Appendix E). We transformed the thought manipulation employed in the previous studies (e.g., Liberman et al. 2002; Liberman and Trope 1998; Trope and Liberman 2000) into more specific pieces of information conveyed by the simulation. This might help ensure that all participants see and digest the same pieces of information and thus think in the similar way. Next, we propose the design of the cause-and-effect simulation. In the low-level cause-and-effect simulation (condition 3), we provided participants with the utility cost per load. The low-level simulation explains the electricity and water consumption for washing a full load of clothes in normal service, and the required electricity used to dry the clothes in terms of the utility cost per load. On the contrary, the high-level simulation conveys the energy and water used to operate the washing machine in normal service as well as the energy used to run the electric drying for 10 years.  In these two conditions, there is the utility bar above the RA (see Figure 7 and Figure 8). This utility bar presents how much the washing model based on participants’ preference may cost. When a participant indicates her preference on energy use, spin speed, and water use, the utility bar will be changed accordingly. This indicates the interactive characteristic of the simulation as suggested by Gredler (2004) and Rieber (1996). In addition, as temporal distance can influence mental construal and thus consumers’ decision-making (Fujita et al. 2008; Liberman and Trope 1998; Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007), we put the temporal distance statement into the effect part of the simulation. The statement “Purchase the washing machine TODAY” and “Purchase the washing machine in the FUTURE” were implemented in the low-level simulation and the high-level simulation respectively. Also, the unit of some product attributes including energy use, cycle time, 77 water use, and warranty was shown in terms of either low-level or high-level (see Table 4). For the low-level simulation, the low-level unit is incorporated with the RA. On the other hand, the high-level unit is employed for the high-level simulation tool. 5.4 Experimental Procedure We used Amazon Mechanical Turk (MTurk) to run our website experiment. A total of 80 participants were recruited and randomly assigned into four types of the simulation design, with 20 in each. The randomization would help mitigate the differences in participants’ past experiences with respect to purchasing a washing machine. The experiment took approximately 30 minutes and required all participants to choose the model of washing machine they intended to buy for themselves. There are two questionnaire surveys: the pre and the post questionnaire. The pre-questionnaire survey asks participants with respect to their demographics, product expertise, product purchasing experience, product using experience, and attitudes toward buying a green product. Additionally, they were asked to read the scenario to understand the experimental task. Two scenarios were employed with the first one for participants in the no simulation and the other for those in the partial simulation, the low-level simulation, and the high-level simulation (see Appendix D). Participants in the no simulation condition were not provided with the RA, while those in other conditions were allowed to use the RA and adjust their preferences and ranking of each product attribute until they were satisfied with the results. After answered all questions and read the scenario, they were redirected to the experimental website for their own condition. 78 In the experimental website, they were asked to choose one washing machine model without the RA or with the RA. They could spend as much time as they liked. After choosing the product to purchase by click “Checkout”, they were redirected to the post-questionnaire survey to answer the questions and receive $5 as a participation reward. In the post-questionnaire survey, they were asked to answer the questions regarding manipulation check, dependent variables, and control variables. In order to motivate participants to take an experimental task seriously, they were told before the experiment that if they provided very detail, serious responses, they would receive the extra reward of $5. Figure 9 presents the flow of the experiment.  Figure 9: Experimental procedure Respond to the pre-questionnaire surveyRead scenarioEnter the experimental websiteEvaluate the product alternatives with/without the RASelect a washing machine Respond to the post-questionnaire surveyReceive the participation reward79 We employed a questionnaire self-report to measure our variables as well as the objective measures. Questionnaire items were adapted from existing scales and newly developed if existing scales are unavailable (see Appendix F). These items were investigated in terms of construct validity (discriminant validity), and internal consistency reliability.  All items were randomly presented in order to reduce mono-method bias. Table 6 presents the flow of measurement items. 80 Measurements Descriptions Pre-questionnaire survey 1. Demographics Age, education, marital status, gender, and income 2. Control variables Product expertise, product purchasing experience, and product use experiences  3. Pre desirability Consideration Product attribute ranking (ranking the product attributes in order of the importance to participants’ product decision) Experimental website  4. Intention to purchase a green product The green score of the product model which participants chose to checkout Post-questionnaire survey 4. Post desirability Consideration Product attribute ranking (ranking the product attributes in order of the importance to participants’ product decision) 5. Manipulation check Temporal distance, mental construal, and feasibility and desirability associated with each product attribute 6. Other dependent variables Attitudes toward purchasing a green product and self-efficacy (seven-point Likert scales) 7. Control variables Involvement, environmental concern, general green purchasing behavior, and website evaluation 81 Measurements Descriptions 8. Open-ended question The reason why participants chose the product Table 6: Measurement flow 5.4.1 Self-Report Measures To capture the desirability consideration, we asked participants to rank each attribute in order of the importance to their purchasing decision. As the desirability consideration in this study is associated green attributes, the desirability consideration is constructed based on the average of the green attribute ranking. The question is as follow: “Please rank the following washing machine attributes in order of the importance to your purchasing decision on the washing machine (the most important attribute at the top):” We adapted two items of attitudes toward purchasing from Pavlou and Fygenson (2006) and create one additional item. In our study, we capture an online consumer’s attitudes toward purchasing a green product using a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). 82 Attitudes toward purchasing a green product (Adapted from Pavlou and Fygenson (2006)) 1. For me, purchasing the washing machine which consumes less utilities (e.g., electricity) would be a good idea. 2. For me, purchasing the washing machine which uses less utilities (e.g., electricity) would be desirable. 3. I think that purchasing the washing machine which consumed less utilities (e.g., electricity) would be a waste of my money. Two self-efficacy items are borrowed from Pavlou and Fygenson (2006) and one is created. As CLT suggested, self-control will be influenced by temporal distance and mental construal (Fujita et al. 2006; Trope et al. 2007). Self-efficacy which is in line with the concept of self-control will be induced by the cause-and-effect simulation design and thus lead to an online green purchase. All items is measured based on a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). 83 Self-efficacy (Adapted from Pavlou and Fygenson (2006)) 1. If I wanted to, I would be able to purchase the product which consumes less utilities (e.g., electricity). 2. If I wanted to, I am confident I could buy the product which consumes less utilities (e.g., electricity). 3. I am uncertain I could purchase the washing machine which consumes less utilities (e.g., electricity) if I wanted to. Besides the main variables in our model, control variables—product expertise, product purchasing and using experiences, involvement, environmental concern, and green purchase behavior—were measured. We controlled for product expertise and product experiences (see Appendix F).  Expertise items are used to control for participants who had high degree of expertise on the product. The high expertise participants might have bias toward the product or the specific attributes more than the low expertise ones. Also, we measure participants’ product experiences consisting of purchasing and using experiences. Like the product expertise, the recent experience of purchasing a washing machine might influence participants’ bias toward the product. As suggested by Petty and Cacioppo (1984), individuals will elaborate the message in a greater degree under the high involvement condition. Consequently, we measure the degree of involvement participants demonstrated. Involvement refers to the extent to which an individual perceives that the issue presented is relevant to her. Three items are adapted from Bhattacherjee 84 and Sanford (2006) and Sussman and Siegal (2003) using a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). Involvement (Adapted from Bhattacherjee and Sanford (2006) and Sussman and Siegal (2003)) 1. I am involved in making decisions about purchasing a washing machine. 2. Purchasing a washing machine is relevant for me. 3. Purchasing a washing machine is important for me. Environmental concern is evident by prior studies that it influences people’s pro-environmental behavior (Kim and Choi 2005; Stern et al. 1999). Environmental concern is defined as “an individual’s green orientation toward the environment and an individual’s concern level as to environmentally conscious behavior” (Kim and Choi 2005, p. 593). That is, if an individual has a high degree of environmental concern, the individual is more likely to perform pro-environmental behavior. Five items are adapted from Kim and Choi (2005) employing a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). 85 Environmental concern (Adapted from Kim and Choi (2005)) 1. I am extremely worried about the state of the world’s environment and what it will mean for my future. 2. Mankind is severely abusing the environment. 3. When humans interfere with nature, this often produces disastrous consequences. 4. Humans must live in harmony with nature in order to survive. Green purchasing behavior reflects an individual’s general behavior of purchasing green products. It is possible that participants who report high degree of a green purchase behavior will be more likely to purchase a green product. Four items are adapted from Kim and Choi (2005) using a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree). 86 General green purchasing behavior (Adapted from Kim and Choi (2005)) 1. I make a special effort to buy products that are made from recycled materials. 2. When I have a choice between two products, I purchase the one less harmful to other people and the environment. 3. I make a special effort to buy products that are environmental friendly. 4. I have avoided buying a product because it had potentially harmful environmental effects. We also capture the reasons why participants chose the model over others using the open-ended question, “Why did you select the washing machine you chose to checkout? Please explain your reasons.” 5.4.2 Objective Measures We record the ranking and the initial stated preference of each product attribute participants assigned to the RA, website search and use behavior, and time participants spent on the entire study as well as on the experimental website. Product attribute ranking and preferences are indirect measures to capture participants’ desirability consideration. We asked participants to rank the seven product attributes in order of the importance to their product decision-making. We also capture their initial preference on each product attribute. Both ranking and initial stated preferences are used to evaluate whether participants are concerned about the desirability.  In addition, the use of the website is captured to analyze the pattern of product search and the product decision-making. All activities (e.g., the number of times participants used the RA, the 87 number of products participants clicked to see more details) are recorded. For instance, the number of times participants used the RA reflect the degree to which they seriously did the task. Intention to purchase a green product is measured by the green score of the product participants chose. This green score refers to the utility cost which is the cost of operating the product (see Appendix E). We calculate the green score by normalizing the utility cost such that 0 indicating the lowest green score and 1 indicating the highest green score. This normalization is calculated by the same formula used by the RA. 5.4.3 Manipulation Check Measurement Manipulation check was conducted to examine the success of implementing two levels of construal as well as temporal distance. This will ensure that participants perceived each simulation design differently.  The manipulation of temporal distance is measured by six items which are newly created to capture how participants perceived the time distance related to purchasing a washing machine. These items employ a seven-point semantic differential scale with the low anchor reflecting the near future and the high anchor indicating the far future. 88 Temporal distance 1. The website presents the information which focuses on today’s purchase. – The website presents the information which focuses on future’s purchase. 2. I thought that purchasing a washing machine would be done very recently. – I thought that purchasing a washing machine would be done very long time from now. 3. I thought that the time frame for when I would buy a washing machine was very soon. – I thought that the time frame for when I would buy a washing machine was very distant. 4. In my opinion, purchasing a washing machine is a shorter-term action. - In my opinion, purchasing a washing machine is a longer-term action. 5. I think that purchasing a washing machine has a shorter-term impact. - I think that purchasing a washing machine has a longer-term impact. 6. I believe that in the short run purchasing a washing machine does matter. - I believe that in the long run purchasing a washing machine does matter. The success of the mental construal manipulation is evaluated by six newly created items. Six items are developed to capture how participants thought in terms of concrete vs. abstract representation, irrelevant vs. relevant to the goal of purchasing a washing machine, and how vs. why mental representation. These items implement a seven-point semantic differential scale with the low anchor presenting the low-level construal and the high anchor showing the high-level construal. 89 Mental construal Concrete vs. abstract 1. The website presents very concrete information. - The website presents very abstract information. 2. The information the website provides is very specific. - The information the website provides is very general. 3. The website gives very detailed information. - The website gives very broad information. Goal irrelevant vs. goal relevant 4. The information the website provides is very superficial to my goals in purchasing a washing machine. - The information the website provides is very central to my goals in purchasing a washing machine. 5. The information the website provides is very irrelevant to my goals in purchasing a washing machine. - The information the website provides is very relevant to my goals in purchasing a washing machine. How vs. why 6. The website presents the information which focuses on "how" to do laundry. - The website presents the information which focuses on "why" to do laundry. 90 Chapter 6: Data Analysis 6.1 Participant Background Information The 80 participants were recruited from Amazon’s Mechanical Turk. One participant did not complete the website task and the post-questionnaire survey and three participants failed the attention check questions. These four participants were excluded from the analysis resulting in the total usable sample size of 76 with 19 participants for the no simulation condition, 20 participants for the partial simulation condition, 19 for the low-level simulation condition, and 18 for the high-level simulation condition, representing diverse backgrounds (see Appendix C). Among the usable sample, 40 (52.6 percent) were 25 – 34 years old, 29 (38.2 percent) had a bachelor’s degree, 45 (59.2 percent) were single, 41 (53.9 percent) were female, 25 (32.9 percent) had an income less than $25,000, 58 (76.3%) had an experience in purchasing a washing machine, and 75 (98.7 percent) had experience in using a washing machine. There was no significant difference in the participant background across the four conditions. We also captured participants’ product expertise. The scale reliability of this measure was greater than 0.70, thereby having the adequate reliability (Nunnally and Bernstein 1994). The results from a one-way ANOVA indicate that participants did not have different product expertise across the four conditions, F(3, 72) = 0.63, p = .60, ηp2 = .03. Participants had a medium level of product expertise (M = 4.21, SD = 1.06) thereby understanding the attributes of the washing machine to some extent. Thus, the product expertise will not influence participants’ decision-making on the 91 product choice such that individuals with more product expertise are more likely to be biased toward some product attributes than others than those with less product expertise. In order to ensure that there is no effect of individual-level concern for the environment and general propensity for green purchasing, we measured how participants were concerned about the environment and to what degree they purchased green products at the end of the post-questionnaire survey after they were exposed to the experimental website A one-way ANOVA was performed. The results show that there was no difference among the treatment conditions in either environmental concern, F(3, 72) = 0.24, p = .87, ηp2 = .01, or green purchasing behavior, F(3, 72) = 0.60, p = .62, ηp2 = .02. Both the environmental concern scale and green purchasing behavior scale showed adequate reliability, with Cronbach’s alpha greater than 0.80. On average, participants were relatively highly concerned about the environment (M = 5.46, SD = 1.18) and were slightly likely to buy green products in general (M = 4.72, SD = 1.48). Thus, we expected that environmental concern and green purchasing behavior in general will not affect participants’ decision-making on the product choice. In sum, the overall results regarding participants’ backgrounds reveal that participants had similar backgrounds across the four conditions. As a result, we anticipated that the difference in the product-decision will result from our manipulation. 6.2 Time Spent on the Experimental Website and the Entire Study, Involvement, and Website Evaluation We measured the time participants spent on the experimental website as well as throughout the entire study. A one-way ANOVA was conducted to test whether there were differences in either 92 the amount of time participants spent on the experimental website or the time participants spent on the entire study. The results reveal that there were no significant differences in time participants spent on the website, F(3, 72) = 1.22, p = .31, ηp2 = .05, and in time participants spent on the entire study, F(3, 72) = 0.81, p = .49, ηp2 = .03. On average, participants spent 4.28 minutes (SD = 3.22) on the website and 19.01 minutes (SD = 8.58) on the entire study.  The involvement scale (Bhattacherjee and Sanford 2006) was employed to evaluate participants’ involvement with the experimental task (Cronbach’s alpha 0.80). The one-way ANOVA’s results show that there was no significant difference in the degree of involvement across the four conditions, F(3, 72) = 0.32, p = .81, ηp2 = .01. Participants on average had slightly high degree of involvement with the task (M = 5.90, SD = 1.00). Taken together with the results with respect to the time participants on the website and the entire study, the results regarding the involvement indicate that participants in all conditions engaged in the task at the similar extent. In addition, participants’ evaluation of website was captured. The results indicate that there was no difference in participants’ evaluation of the website, F(3, 72) = 1.26, p = .29, ηp2 = .05. Overall, participants had a positive evaluation of the website (M = 5.95, SD = 0.83).  In sum, according to the non-significant results on time spent on the website, time spent on the entire study, involvement, and the evaluation of the website, we will not analyze these data further.  93 6.3 Manipulation Check To evaluate the success of our design manipulation, we employed self-report items, with six items to measure temporal distance and six items to capture mental construal (see Appendix F). The Cronbach’s alpha of temporal distance was 0.74 well above 0.70 indicating acceptable reliability. However, the Cronbach’s alpha of mental construal was about 0.40 suggesting that the reliability was not achieved. A one-way ANOVA was conducted to test the effect of the simulation design on temporal distance and mental construal. The results indicate that there was a marginally significant effect of the design on temporal distance, F(3, 72) = 2.64, p = .06, ηp2 = .10 with a medium effect size. The post-hoc analysis using Bonferroni correction reveals that participants in the high-level simulation perceived significantly higher temporal distance (M = 5.01, SD = 1.13) than those in the low-level simulation (M = 4.01, SD = 1.03), p = .04. No significant difference was found for the other conditions. Figure 10 presents the differences in temporal distance across the four conditions. However, the manipulation of mental construal was not successful, F(3, 72) = 0.21, p = .89, ηp2 = .01. This suggests that participants did not construe their thoughts differently when they were exposed to the experimental website. On average they did not think at either a low- or high-level of construal (M = 3.86, SD = 0.76). However, when considering the internal consistency reliability of this scale was low (Cronbach’s alpha = 0.40). This indicates the problem with the scale rather than the unsuccessful manipulation of mental construal. Figure 11 presents the means of mental construal across all four conditions.  94  Figure 10: The effect of simulation design on temporal distance  Figure 11: The effect of simulation design on mental construal 6.4 The Effect of Cause-and-Effect Simulation Design on Intention to Purchase a Green Product A one-way ANOVA was conducted on the green score associated with participants’ product choice that reflects an individual’s intention to buy a green product. This objective measure can reduce the social desirability bias that may be inherent to the self-report measures which may, in turn, 95 bias the results (Cook et al. 1979)The results show that the effect of the simulation design was marginally significant, F(3, 72) = 2.31, p = .08, ηp2 = .09 (see Figure 12). The medium effect size indicates there might be the difference in intention to buy a green product between conditions. The follow-up analysis using Bonferroni correction did not reveal the difference among conditions.  Figure 12: Intention to purchase a green product (green score) However, when considering how participants interacted with the experimental website, we found that participants differed in how seriously they inspected the products. This was evident by the number of products participants clicked the “more details” button to see the product details as the details were not shown at the main product page. A one-way ANOVA supported a significant effect of simulation design, F(3, 72) = 9.82, p < .001, ηp2 = .29 (see Figure 13). The follow-up 96 analysis using Games-Howell correction4 suggests a significant difference in the number of products participants viewed between the no simulation condition (MN = 14.84, SEN = 1.89) and the partial simulation condition (MP = 3.35, SEP = 1.84), between the no simulation condition and the low-level simulation condition (ML = 2.37, SEL = 1.89), and the no simulation condition and the high-level simulation condition (MH = 3.50, SEH = 1.94). Thus, the number of products participants viewed for more details might be a confound, as those in the no simulation condition did research on the product more than those in other conditions. It is possible that the more products people viewed, the more likely they gained the product knowledge and thus chose more green products. However, this variable was not the target of the current study and therefore was controlled rather than being used as a moderator.                                                  4 Games-Howell correction was used as the assumption of homogeneity of variances was violated. 97  Figure 13: Number of products participants clicked to see more details According to the significant difference in the number of products participants viewed for more details, a one-way between subjects ANCOVA was conducted to examine whether the design influenced participants’ intention to purchase a green product. Intention was measured by the green score which was calculated based on energy usage, spin speed, and water usage associated with the product model that participants chose. As the number of products participants viewed for more details played a role in participants’ intention, it was used as a covariate.5 The covariate, the number of products viewed was found to be marginally significantly associated with intention,                                                  5 Although the assumption of independence of an independent variable and a covariate was not met, the ANCOVA could be used in the current study. According to Miller, G. A., and Chapman, J. P. 2001. "Misunderstanding analysis of covariance," Journal of abnormal psychology (110:1), p 40., the assumption was concerned in case that the quasi-experiment was employed leading to the preexisting groups. However, this current study employed the random assignment. This problem would be mitigated. Also, even if ANCOVA was conducted, the main effect of an independent variable was significant and its effect size was relatively large. Thus, ANCOVA would be justified to be used in this current study. 98 F(1, 71) = 9.3.58, p = .06, ηp2 = .05, with number of products viewed associated with greener intentions. A statistically significant effect of the design on intention was also found, F(3, 71) = 9.82, p = .03, ηp2 = .12. The post-hoc analysis based on LSD reveals that the significant difference in intention for the following conditions: 1) those in the no simulation chose less green products than those in the low-level simulation (p < .05), 2) the no simulation selected fewer green models than those in the high-level simulation (p = .02), 3) those in the partial simulation chose less green products than those in the low-level simulation (p = .05), and 4) the partial simulation selected less green models than those in the high-level simulation (p = .02). However, there was no significant difference in intention between the low-level and the high-level conditions. The adjusted mean score on intention to buy a green product is shown in Table 7 and Figure 14. As a result, H4a was partially supported as the difference in intention between the high-level and the low-level was not significant. H4b and H4c were fully supported. The results on intention to purchase a green product lend some support that our simulation design worked better than the partial and the no simulation design in motivating people to buy a green product. Also, this suggests that without the capability of the technology, people were less likely to buy a green product and took more effort to find the product they wanted as evident by the number of products participants clicked to see for more details.  99 Simulation Design M SE No Simulation 0.49* 0.06 Partial Simulation 0.52* 0.05 Low-Level Simulation 0.66* 0.05 High-Level Simulation 0.69* 0.05 *Adjusted means based on the following values: the number of products participants viewed was 6.01. Table 7: Adjusted mean score on intention to purchase a green product  Figure 14: Adjusted mean score on intention to purchase a green product 6.5 The Effect of Cause-and-Effect Simulation Design on Persuasion A 4 x 2 simple-mixed ANOVA was conducted to evaluate the effect of the simulation design (no simulation, partial simulation, low-level simulation, and high-level simulation) and the time 100 [before the manipulation (time 1) and after the manipulation (time 2)] on the product attribute ranking and the product attribute preferences. In other words, we assessed whether participants’ initial ranking of and initial preferences toward the price and the green attributes (energy use, spin speed, and water use) differed from their ranking and preferences after they were exposed to the experimental website. A lower rank indicates the attribute is not as important to participants decision on the product and a higher rank indicates the attribute is perceived as more important to their decision (1 = the least important to 7 = the most important). Here we focused only on the price and the green attributes, as the former reflects the feasibility consideration and the latter is associated with desirability considerations. All three green attributes have a direct impact on the utility costs used in doing laundry, including the drying process. This analysis captures the change in participants’ implicit attitudes toward buying a green product reflecting the role of the design in enabling persuasion. For the energy use ranking, there were no effect of the simulation design, F(3, 72) = 1.08, p = .36, ηp2 = .04, time, F(1, 72) = 0.79, p = .38, ηp2 = .01, and no interaction effect between the design and time, F(3, 72) = 0.46, p = .71, ηp2 = .02. Figure 15 presents the means of the energy use rank with the higher ranking indicating the more important the energy use was. 101  Figure 15: Energy use ranking For the spin speed ranking, the results indicate that there were no effect of the design, F(3, 72) = 0.76, p = .52, ηp2 = .03, time, F(1, 72) = 1.62, p = .21, ηp2 = .02, and interaction effect between the design and time, F(3, 72) = 1.79, p = .16, ηp2 = .07. However, the medium effect size of the interaction effect might suggest that there will be the impact of the design with time. The simple effect analysis using Bonferroni correction reveals that participants in the no simulation condition perceived that the spin speed was significantly more important after they were exposed to the experimental website (MT1 = 2.11, SET1 = 0.34, MT2 = 2.79, SET2 = 0.34), p = .04. Figure 16 presents the means of the spin speed ranking with the higher ranking indicating the more important the spin speed was. 102  Figure 16: Spin speed ranking The results regarding the water use ranking show that there was a significant effect of the design, F(3, 72) = 3.51, p = .02, ηp2 = .13. The post-hoc analysis using Games-Howell correction indicates that participants in the high-level condition ranked the water use attribute significantly higher than those in the partial simulation (MH = 4.32, SEH = 0.30, ML = 3.60, SEL = 0.29), p = .05. The effect of time, F(1, 72) = 0.59, p = .45, ηp2 = .01, and the interaction effect between the design and time, F(3, 72) = 0.23, p = .88, ηp2 = .01, were not found. However, the simple effect analysis using Bonferroni correction indicates that after being exposed to the website, those in the high-level simulation ranked water usage significantly more important as compared to those in the partial simulation  (MH = 4.94, SEH = 0.38, MP = 3.50, SEP = 0.36),p = .05. Figure 17 presents the means of the water use ranking with the higher ranking indicating greater importance.  103  Figure 17: Water use ranking For the price, the results show that there was no impact of the simulation design, F(3, 72) = 0.60, p = .62, ηp2 = .02, time, F(1, 72) = 0.38, p = .54, ηp2 = .01, and interaction effect of the design with time, F(3, 72) = 0.88, p = .46, ηp2 = .04. Figure 18 presents the means of the price ranking with the higher ranking indicating the more important the price was. 104  Figure 18: Price ranking In addition to the product attribute ranking, the preference toward each attribute (the value) was analyzed to evaluate whether the simulation design has an impact on the change in preferences. Here we focus only on the price’s and the green attributes’ preferences, as we manipulated the products to induce the trade-off between these two aspects. However, participants in the no simulation condition were not provided with the RA. Thus, we could only evaluate the change in the attribute preferences for the partial simulation, the low-level simulation, and the high-level simulation.  To assess the preference change, we compared participants’ initial stated preference toward the price and the three green attributes participants indicated on the RA with the attribute score of the chosen product. The score was normalized with 0 indicating the worst value and 1 indicating the best value. For instance, the energy use score of 0 indicates that the washing machine consumes the most energy compared to the similar model, whereas the energy use score of 1 indicates that 105 the machine consumes the least energy. Simple mixed design with a 3 x 2 mixed design (3 types of the website design (between-subject) x 2 points of time (within-subject)). For the energy use, the results show that there was a significant effect of time, F(1, 54) = 3.94, p = .05, ηp2 = .07. The effect of the design, F(2, 54) = 1.59, p = .21, ηp2 = .06, and the interaction effect between the design and time, F(2, 54) = 0.50, p = .61, ηp2 = .02 were not found. However, the medium effect size for the design indicates that there might be the effect of the design on the preference on the energy use. The follow-up analysis based on Bonferroni correction reveals that there was a significant difference between the initial preference and the preference of the chosen product, p = .05 (MT1 = 0.64, SET1 = 0.02, MT2 = 0.55, SET2 = 0.03). Specifically, the initial stated preference on the energy use was lower than the actual energy use of the product participants selected. This means that they chose the product model which consumed more energy use than they previously preferred. Also, the significant difference was found in the partial simulation, p = .05, with those in the partial simulation choosing the product model which consumed more energy than they previously stated (MT1 = 0.60, SET1 = 0.04, MT2 = 0.47, SET2 = 0.06). Figure 19 presents the means of the energy use preference before and after the manipulation.  106  Figure 19: Preference on energy use The results on the preference toward the spin speed reveal that there was a marginal significant effect of time, F(1, 54) = 3.75, p = .06, ηp2 = .07. However, the significant effect of the design, F(2, 54) = 0.91, p = .41, ηp2 = .03, and the significant interaction effect of the design and time, F(2, 54) = 0.83, p = .44, ηp2 = .03, were not found. The follow-up analysis using Bonferroni was conducted and shows that there was a marginal significant difference between the initial preference on the spin speed and the spin speed value of the chosen product, p = .06 (MT1 = 0.56, SET1 = 0.02, MT2 = 0.50, SET2 = 0.03). The significant difference of the spin speed’s preference in the partial simulation was also found, p = .04. Specifically, those in the partial simulation selected the product that had a lower spin speed than what they previously indicated (MT1 = 0.55, SET1 = 0.03, MT2 = 0.43, SET2 = 0.06). Figure 20 shows the means of the initial stated preference of the spin speed and the spin speed’ specification of the product participants chose. 107  Figure 20: Preference on spin speed Results on the water use preference show that there were a marginal significant effect of the design, F(2, 54) = 2.60, p = .08, ηp2 = .09, and a marginal significant interaction effect between the design and time, F(2, 54) = 2.60, p = .08, ηp2 = .09. The effect of time was not found, F(1, 54) = 0.84, p = .36, ηp2 = .02. The pairwise comparison based on Bonferroni correction reveals that after participants were exposed to the experimental website, there were a significant difference in the water use’s preference of the chosen product between those in the partial simulation and those in the high-level simulation, at p = .03, and a marginal significant difference between the water use’s preference on the chosen product between those in the partial simulation and the low-level simulation, p = .06. That is, participants in the partial simulation selected the product which consumed more water than the one those in the low-level and the high-level simulation chose (MP = 0.48, SEP = 0.04, ML = 0.63, SEL = 0.04, MH = 0.65, SEH = 0.05). Figure 21 presents the means of the initial stated preference for water use and the water use score of the actual product. 108  Figure 21: Preference on water use The analysis on the preference toward the price reveals that there was a statistically significant effect of time, F(1, 54) = 22.34, p < .001, ηp2 = .29. The effect of the design, F(2, 54) = 1.71, p = .19, ηp2 = .06, and the interaction effect between the design and time, F(2, 54) = 0.10, p = .091, ηp2 < .005, were not significant. However, the medium effect size for the design suggests that there might be effect of the design. The post-hoc analysis based on Bonferroni shows that participants initially preferred the cheaper price model than what they chose to pay for the selected product, p < .001 (MT1 = 0.69, SET1 = 0.03, MT2 = 0.55, SET2 = 0.04). This significant difference also holds for those in the partial simulation condition, p = .02, the low-level simulation, p = .01, and the high-level simulation, p = .01. Table 8 and Figure 22 present the marginal means and standard error of the preference on the price. 109 Simulation Design Initial Stated Preference (T1) Actual Attribute Value of the Chosen Product (T2) M SE M SE Partial Simulation 0.74 0.05 0.61 0.06 Low-Level Simulation 0.72 0.05 0.56 0.07 High-Level Simulation 0.62 0.05 0.46 0.07 Table 8: Preference on price  Figure 22: Preference on price Interestingly, taken the results on the product attribute rankings and the preferences together, we found that the full simulation design (low-level and high-level simulation) helped keep participants 110 on their own track without diverting them to value the less green products. On average the initial rankings and the preferences with respect to the green attributes were somewhat high (greater than 4.00 for ranking, except the spin speed, and greater than 0.50). This suggests that in general all participants were concerned about the price of the product as well as the green attributes, especially energy use and water use.  However, when they were exposed to the product alternatives, they faced the trade-off between price and greenness such that greener washing machine alternatives are more costly. Those in the full simulation design were indifferent in valuing the green attributes (e.g., the water use ranking) and valued some green attributes as more important (e.g., the preference on water use). However, this did not hold for the partial simulation design. To demonstrate this point, we averaged the rankings of the three green attributes and compare this average ranking of the green attributes between the partial simulation and the full simulation which includes the low-level and the high-level simulation. The results show that a significant main effect of the design was found, F(1, 55) = 5.58, p = .02, ηp2 = .09, with the full simulation perceiving the green attributes more important than the partial simulation (MFull = 4.15, SEFull = 0.12, MPartial = 3.68, SEPartial = 0.16). The effect of the time, F(1, 55) < .001, p = .95, ηp2  < .001, and the interaction effect between the design and time, F(1, 55) = 0.13, p = .72, ηp2 < .003 were not found. However, the simple effect analysis based on Bonferroni correction reveals that there was a significant difference between the full simulation and the partial simulation only after the manipulation of the design, p = .03 (MFull = 4.17, SEFull = 0.14, MPartial = 3.65, SEPartial = 0.19). Figure 23 presents the means of the average ranking of the green attributes between the partial simulation and the full simulation. 111  Figure 23: Green attributes ranking Although there was no significant difference in the no simulation condition, the values participants in this condition assigned to the green attributes seemed mixed.  Table 9 and Table 10 presents means and standard deviations of the average rankings and preferences for the overall designs and for each design condition respectively. Note that there was no information about the stated preference for the no simulation condition. 112 Attribute Time 1 Time 2  M SD M SD Ranking* Energy Use 4.93 1.68 5.08 1.53 Spin Speed 2.53 1.47 2.72 1.47 Water Use 4.38 1.48 4.24 1.66 Price 6.18 1.33 6.12 1.36 Preference** Energy Use 0.62 0.17 5.44 0.25 Spin Speed 0.56 0.13 0.49 0.25 Water Use 0.62 0.16 0.58 0.21 Price 0.69 0.22 5.49 0.29 *Ranking is on the scale between 1 and 7 inclusively. 1 demonstrates the least important attribute and 7 shows the most important attribute. **Preference score is normalized on the scale between 0 and 1 inclusively, with the 0 indicating the worst value of the attribute and 1 indicating the best value of the attribute. For instance, the score 0 of energy use suggests that the product consumes the most energy relative to other similar models and the score 1 of energy use indicates the product consumed the least water compared to other similar models. Table 9: Means and standard deviations of the product attribute ranking and preferences 113 Product Attribute No Simulation Partial Simulation Low-Level Simulation High-Level Simulation  Time 1 Time 2 Time 1 Time 2 Time 1 Time 2 Time 1 Time 2 Ranking Energy Use 4.79 5.21 4.45 4.6 5.16 5.26 5.39 5.28 Spin Speed 2.11 2.79 2.95 2.85 2.47 2.26 2.56 3.00 Water Use 4.63 4.26 3.70 3.50 4.32 4.32 4.94 4.94 Price 6.05 5.63 6.15 6.25 6.26 6.37 6.33 6.22 Preferences Energy Use n/a 0.52 0.60 0.47 0.63 0.57 0.64 0.60 Spin Speed n/a 0.46 0.55 0.43 0.56 0.50 0.57 0.55 Water Use n/a 0.53 0.61 0.48 0.61 0.63 0.62 0.65 Price n/a 0.55 0.74 0.61 0.72 0.56 0.62 0.46 Table 10: Product attribute ranking and preferences based on the simulation conditions 6.6 The Effect of Cause-and-Effect Simulation on Desirability Consideration A one-way between subjects ANCOVA was conducted to determine whether the design affected participants’ desirability consideration. The desirability consideration was calculated based on the average rankings of energy use, spin speed, and water use after participants were exposed to the 114 experimental website. This will be a surrogate of how participants perceived the desirability features operationalized by the three green product attributes. As each green attribute has a direct impact on the utility cost, the ranking of each green attribute was averaged. The higher score indicates that participants valued the desirability more. The number of products participants viewed for more details was used as a covariate. The covariate was not found to be significantly correlated with the desirability consideration, F(1, 71) = 1.93, p = .17, ηp2 = .03. The marginally significant effect of the design was found to be associated with the desirability consideration, F(3, 71) = 2.56, p = .06, ηp2 = .10, with a medium effect size, lending support that there are difference in desirability consideration between conditions. The post-hoc analysis was performed based on LSD. The results show that those in the high-level simulation were more concerned about the desirability feature than those in partial simulation (p = .01). This lends partial support to H1b and full support to H1c. However, H1a was not supported. Adjusted mean and standard deviation on the desirability consideration were presented in Table 11 and Figure 24. Simulation Design M SE No Simulation 3.94* 0.22 Partial Simulation 3.70* 0.19 Low-Level Simulation 4.01* 0.20 High-Level Simulation 4.45* 0.20 Adjusted means based on the following values: The number of products participants viewed was 6.01. Table 11: Adjusted mean and standard deviation on the desirability consideration 115  Figure 24: Adjusted mean and standard deviation on the desirability consideration 6.7 Desirability Consideration, Attitudes toward Purchasing a Green Product, Self-efficacy, and Intention to Purchase a Green Product SmartPLS 3.0 was used to examine the structural model proposed in this current study. According to Barclay et al. (1995), the measurement model was assessed in terms of the internal consistency reliability and the discriminant validity. Most of the items tapped on their respective latent variables, with loadings greater than or equal to 0.70 (see Table 12). AP3 which had the loading equal to 0.53 was excluded from the analysis. The overall loadings support adequate discriminant validity. The internal consistency was evident by the composite reliability and Cronbach’s alpha greater than or equal to 0.70 (see Table 13). The AVE of each latent variable was greater than the correlations between itself and others (see Table 13) as well as there was no loading higher than the loadings of the respective latent variables (see Table 12). This supports acceptable discriminant validity.  116   Attitude Desirability Intention Self-Efficacy AP1 0.95 0.17 0.43 0.39 AP2 0.93 0.10 0.37 0.43 Average Green Ranking 0.15 1.00 0.47 0.20 Green Score of a Chosen Product 0.42 0.47 1.00 0.51 SE1 0.26 0.24 0.41 0.79 SE2 0.47 0.09 0.46 0.84 SE3 0.27 0.14 0.29 0.70 Table 12: Loadings and cross loadings   Composite reliability Cronbach’s Alpha Attitude Desirability Intention Self-Efficacy Attitude 0.94 0.87 0.89    Desirability 1.00 1.00 0.15 1.00   Intention 1.00 1.00 0.42 0.47 1.00  Self-Efficacy 0.82 0.68 0.44 0.20 0.51 0.61 Off-diagonal – correlations, diagonal – the square root of AVE (Average Variance Extracted) Table 13: Internal consistency and discriminant validity 117 The path significance of the structural path model was evaluated by using bootstrap resampling. The results show that the desirability consideration was marginal significantly and positively related to self-efficacy, t = 1.83, p = .08, R2 = 0.04, whereas had no significant relationship with attitudes toward purchasing a green products, t = 1.28, p = .20, R2 = 0.02. Thus, H3a was not supported and H3b was supported. Attitudes toward purchasing a green product, t = 3.04, p < .004, as well as self-efficacy, t = 3.04, p < .001, had a significant and positive impact on intention to purchase a green product, thereby supporting H3c. The overall constructs could explained 31% of the variances in intention. Figure 25 shows the path model.  * Significant at the level .1, ** significant at level .05, ***significant at level .001, dotted line – non-significant path Figure 25: Structural path model Table 14 summarizes the hypotheses testing results. Tested in PLS 118 Hypothesis Results H1a: The low-level cause-and-effect simulation will negatively affect an online consumer’s desirability consideration toward purchasing a product more than other designs. Not supported.  Post-hoc test reveals no difference between the low-level simulation and other conditions. H1b: The high-level cause-and-effect simulation will positively affect an online consumer’s desirability consideration toward purchasing a product more than other designs. Partially supported. Post-hoc test show the significant difference between the high-level simulation and the partial simulation. There was no difference between the high-level simulation and other conditions. H1c: There will be no difference between the no cause-and-effect simulation and the partial cause-and-effect simulation in an online consumer’s desirability consideration toward purchasing a product. Supported. Post-hoc analysis indicates no difference between the no simulation and the partial simulation. H2a: Desirability consideration will lead to attitudes toward purchasing a green product. Not supported. Bootstrap resampling indicates non-significant path between the desirability consideration and attitudes. H2b: Desirability consideration will lead to self-efficacy. Partially supported. Bootstrap resampling indicates a marginal significant relationship between the desirability consideration and attitudes. 119 Hypothesis Results H3a: Attitudes toward purchasing a green product will affect intention to purchase a green product. Supported. Bootstrap resampling shows the significant path between attitudes and intention.  H3b: Self-efficacy will positively influence intention to purchase a green product. Supported.  Bootstrap resampling shows the significant path between self-efficacy and intention. H4a: The high-level cause-and-effect simulation will motivate an online consumer to purchase a green product more than other designs. Partially supported. Post-hoc analysis reveals the significant difference between the high-level simulation and the no simulation design, and the partial simulation design. No significant difference was found between the high-level and the low-level simulation. H4b: The low-level cause-and-effect simulation will motivate an online consumer to purchase a green product more than the partial and the no simulation. Supported. Post-hoc analysis suggests the significant difference between the low-level simulation and the no simulation design, and the partial simulation design. 120 Hypothesis Results H4c: There will be no difference in the degree of a green purchase between the partial cause-and-effect simulation and the no simulation. Supported. Post-hoc analysis demonstrates non-significant difference between the no simulation design and the partial simulation design. Table 14: Hypotheses testing results 6.8 Additional Evidence of the Awareness of Cause-and-Effect Relationship, Feasibility and Desirability Trade-off, and Environmental Concern Open-ended responses to the question “Why did you select the washing machine you chose to checkout? Please explain your reasons.” were analyzed. A single coder was employed. This helps clarify the reason why participants chose the product they selected and also reveals some insights regarding whether the simulation design worked. A chi-square test was conducted. First, we evaluated whether participants were aware of the relationship between their product decision on the product attributes, the cause, and its impact on the utility cost, the effect. The results show the effect of the design on awareness of cause-and-effect relationship was not found, χ² = 3.95, p = 0.27. This suggests that participants did not differ in concerning the relationship between the cause and the effect the simulation tool provided. However, when the percentage of those who were concerned about this relationship was compared, the low-level and the high-level simulation had a higher percentage of those who were aware of the relationship greater than the other two conditions (see Figure 26). There was comparable percentage of those who recognized 121 this relationship between the two full simulation designs (PercentageL = 63.16, PercentageH = 61.11).  Figure 26: Awareness of cause-and-effect relationship Secondly, the awareness of the feasibility and the desirability consideration was analyzed by coding the open-ended responses and counting the number of participants who reported that they were aware of the trade-off. We assessed whether participants realized this trade-off. As price was associated with feasibility and green attributes were related to desirability, any responses which described the sacrifice of one to another would support that a participant is aware of the trade-off. The results indicate that the effect of the design on the trade-off awareness was significant, χ² = 7.76, p = 0.05 (see Figure 27). The post-hoc test based on Bonferroni correction reveals the difference in this trade-off awareness between the partial simulation and the low-level simulation, p < .05 with those in the low-level simulation recognizing the trade-off more than those in the partial simulation. 122  Figure 27: Awareness of trade-off between price and green attributes Finally, the environmental concern was examined as this reflects whether the simulation design could induce participants’ environmental concern when making a decision about the product. The results on environmental concern reveal that there was no relationship between the design and the environmental concern, χ² = 0.30, p = 0.96 (see Figure 28). Across the four conditions, participants had the similar concern about the environment. Thus, the environmental concern might be less likely to induce green product purchasing. The summarized statistics test of open-ended response regarding awareness of cause-and-effect relationship, awareness of the feasibility and the desirability trade-off, and environmental concern was shown in Table 15. 123  Figure 28: Environmental concern Awareness No Simulation Partial Simulation Low-Level Simulation High-Level Simulation χ² p Cause-and-Effect Relationship 4 3 7 7 3.95 0.27 Trade-Off between Feasibility and Desirability 4 1* 8* 4 7.76 0.05 Environmental Concern 4 3 3 3 0.30 0.96 * Significant at the level .05. Table 15: Awareness of cause-and-effect relationship, feasibility-desirability trade-off, and environmental concern 124 Chapter 7: Conclusion 7.1 Discussion The overall results of our study show that the full cause-and-effect simulation, the low-level and the high-level simulation design, helped motivate people to choose more green products than the partial simulation and the no simulation design. According to the green score associated with the product participants chose to checkout, the score of both low-level and high-level simulation was significantly greater than that of the no simulation and the partial simulation design. However, contrary to our expectation, people given the high-level simulation did not select more green products than those with the low-level.  One possible explanation for this is participants in the low-level simulation realized the trade-off relationship between price and green attributed more than those in any other conditions based on the open-ended responses. This might suggest that people in the low-level simulation understood the product alternatives more than others, thereby leading them to be less concerned about the feasibility (price). Also, although they were not provided with the 10-year utility cost which focused more on the long-term outcome, they were likely to predict the long-term benefit from energy and water saving by themselves as evident by the open-ended responses indicating no difference in the percentage of people who recognized the cause-and-effect relationship between the low-level and high-level simulation. These findings are consistent with what Thompson et al. (2009) found. In their study, they explored the process-focused and the outcome-focused simulation and found that the process-focused 125 simulation which parallels with the low-level simulation caused people to make a decision harder. They reasoned that as the processed-focus simulation invoked people’s trade-off more salient, people found the decision required more effort. As a result, this notion might explain why those in the high-level simulation did not perceive the trade-off as much as those in the low-level simulation. However, both the full simulation designs invoked more trade-off between the feasibility and the desirability more salient than other condition, though statistically significant results were not found. This lends partial support to the effectiveness of the full simulation design which invoked the trade-off between feasibility and desirability more salient. Another explanation for no difference between the two full cause-and-effect simulations might results from the nature of the website experiment. As the manipulation of mental construals was not successful, we suspected the effect of mental construals was transient. If participants did not engage much in the experimental website, the impact of mental construals on participants’ preference would not be salient.  This received slightly support from time participants spent on the website experiment. On average all participants spent 4.28 minutes, thus suggesting low engagement on the experimental website. Consequently, the significant difference between these two conditions was not found. As a result, the role of technology in allowing individuals to interact with and manipulate the RA and to observe the full cause-and-effect relationship between people’s decision regarding the green product attributes and the utility cost influenced individuals to make more green purchase. The construal levels did not make significant change in people’s decision-making. 126 To explain how the full cause-and-effect simulation made the difference in persuading participants to go green, the comparisons of the ranking and the preference regarding the green product attributes (energy use, spin speed, and water use) between pre- and post-experimental website were conducted. Both green attribute ranking and preferences captured individuals’ desirability considerations. For the comparison of the product attribute preference, we could analyze data with three conditions, except the no simulation design. For those in the no simulation, they ranked the spin speed significantly higher after being exposed to the experimental website. Nevertheless, no one in this condition reported the importance of the spin speed for the open-ended response. As the ranking of one attribute was not independent on the ranking of others, we suspected that those in the no simulation design might not perceive spin speed as more important. Also, participants were exposed to more products as compared to the other treatment conditions. As a result, we did not compare this design with others.  The results of both analyses for the three conditions (the partial simulation, the low-level simulation, and the high-level simulation) were analyzed. We found interesting and relatively consistent findings. According to the relatively high ranking of the green attributes, especially energy use and water use (greater than 4 out of 7), people generally were concerned about energy and water the appliance consumed. However, when the product alternatives which were manipulated such that they invoked the trade-off between the feasibility (price) and the desirability (green attributes) with respect to purchasing a product, individuals needed to sacrifice one for the other.  Inconsistent with our expectation, providing people with the full cause-and-effect simulation did not significantly increase their preferences toward the green attributes. However, interestingly, this reinforced people not to divert from their own preference when they faced the trade-off. The results 127 indicate no significant difference was found between pre- and post-experimental website for those in the full simulation design. On the contrary, without demonstrating the full cause-and-effect relationship, people showed lower preferences toward the green as in the case of those in the partial simulation design. In the partial simulation design, participants perceived water as less important and chose the product which consumed more energy and water as well as had less spin speed than what they previously preferred. Therefore, these findings support that we could leverage the capability of IT to keep online consumers stay on their preference track. This was considered as the way to persuade individual to buy a green product. In addition, the results on the desirability consideration reveal the relatively consistent findings that the full simulation design, as the high-level simulation, enhanced people’s desirability consideration to some extent. Specifically, those provided with the high-level simulation valued the green attributes significantly more than those provided with the partial simulation. As in the partial simulation, people did not observe the full cause-and-effect relationship, they would less likely to be aware of the outcome resulting from their decision-making. Therefore, this analysis partially supports that the cause-and-effect simulation led individuals to be more concerned regarding green attributes than any other conditions and also suggests that there might be the effect of mental construals in influencing individuals’ desirability consideration. Moreover, the structural path analyses reveal the consistent findings with those on green attribute ranking and preference. That is, the role of the full cause-and-effect simulation reinforced people to abide by their initial and more environmental preferences. This was evident by the partial significant path coefficient between the desirability consideration and the self-efficacy. The path between the desirability consideration and attitudes toward purchasing a green product was not 128 significant, lending some support to our findings that generally people were concerned about how much an appliance consumed resources. As TPB suggests, attitudes and self-efficacy which is the ground of perceived behavioral control used in TPB were found to be significantly related with intention to purchase a green product. In our study, we found the stronger impact of self-efficacy on intention to purchase than attitudes. Thus, the structural path model provides considerable support that the full cause-and-effect simulation motivated individuals to choose more green products by reinforcing desirability consideration and self-efficacy. In sum, we found evidence to support our cause-and-effect simulation design could enable persuasion such that the full cause-and-effect simulation induced people to keep track on their desirability consideration, thereby leading to a green purchase.  7.2 Summary In this study, we aim at promoting an online green purchase to carve out a better future for the world. The state of the planet’s environment is an important issue that affects the future of mankind. As green products have less environmental impact compared to non-green counterparts, persuading people to buy these green products would help to mitigate this environmental issue.  As supported by several studies, online consumers are less likely to buy green products (Cone Communications 2013; Harris Interactive 2012).  We reviewed the literature and found the factors which would explain why online consumers are not concerned about the environment when they make an online purchase. Two factors, proposed by Griskevicius et al. (2010), are status and reputation. They found that these two factors play a key role in influencing individuals to go green. However, in the online context, there is an inherent lack of the role of status and reputation. In 129 addition, we suspected that if online consumers buy the product which privately consumed without other people’s recognition. This will aggravate the likelihood the online consumers buy green even further, as the normative influence will not affect individuals’ decision-making in case of the privately-consumed products (Bearden and Etzel 1982). Accordingly, designing the e-commerce website to persuade online consumers to buy green privately-consumed products will be challenging. In order to overcome this challenge, we explored the concept of the cause-and-effect simulation suggested by Fogg (2002) to design an e-commerce website which promotes online consumers to buy green products. This would be a useful concept, as it can present the relationship between online consumers’ decision on the product and its impact on the environment in terms of resource efficient. As the RA can facilitate online consumers in making decisions about products (Xiao and Benbasat 2007), it was chosen to present the cause part of the relationship. In order to specify the effect part of this cause-and-effect simulation, we adopted Construal-Level Theory (CLT) (Liberman and Trope 1998). CLT proposed that individuals attach more to the desirability consideration than to the feasibility consideration if they construe their thoughts at a higher level as opposed to a lower level, thereby influencing consumers’ preferences and evaluation on choices (Liberman et al. 2007b; Trope and Liberman 2000; Trope et al. 2007). Thus, the desirability feature of the product that is associated with the outcome resulting from the purchasing behavior, the utility cost (e.g., energy and water cost) was selected as the effect part of the simulation design. 130 Our study tested the effect of two types of the cause-and-effect simulation design, the low-level and the high-level simulation, developed based on CLT, on the desirability consideration as well as intention to choose a green choice. We compared these two website designs against the no simulation and the partial simulation design which presents only the cause part. We found both types of the cause-and-effect simulation design could effectively persuade people to choose more green products than the other two conditions. Additionally, we found interesting findings which would explain how and why the two cause-and-effect simulation design works better than the partial simulation and the no simulation. In particular, the full cause-and-effect simulation design could increase self-efficacy toward buying green. Providing online consumers with this type of the simulation keeps them focus more on the desirability features of the products. As green is associated with the desirability (versus feasibility), the full cause-and-effect simulation design will help online consumers to stay on the “path of good intentions” to buy green. Accordingly, our study refers persuasion to behavior change rather than attitude change. In summary, our study leveraged the capability of technology to enable persuasion by keeping online consumers on the desirability path increasing the likelihood to go green. This study also provided the theoretical lens which helps to explain how and why implementing the full cause-and-effect simulation reinforces the green purchasing behavior. 7.3 Contributions This study examines the effect of the e-commerce website design on intention to purchase a green product and explains how and why this design influences the green intention. Two types of the 131 website design—the low-level and the high-level cause-and-effect simulation—were developed based on the concept of the cause-and-effect simulation, proposed by Fogg (2002) and the temporal construals advocated by Liberman and Trope (1998). The theoretical contributions of our study are four-fold. First, to best of our knowledge, our study is the first study which implements the cause-and-effect simulation (Fogg 2002) and CLT (Liberman and Trope 1998) to enable green persuasion in an online context. According to CLT, the desirability consideration is associated with the end-state or the outcome that is in line with the effect part of the cause-and-effect simulation. Also, Liberman and Trope (1998) found that this consideration will be more salient for individuals who construe their thought at a high level induced by the temporal distance than for those who focus their thought at a low level, especially in the case that individuals need to trade-off between feasibility features and desirability features. Feasibility features are associated with a more immediate outcome involving how difficult it is to perform the action. In contrast, desirability features are related to delayed outcomes (Fujita et al. 2006; Liberman and Trope 1998), the green features lie more toward on the desirability and price indicates the feasibility.  In a similar vein, Farrell (2012) suggests green products constitute higher price than non-green alternatives, thus dampening the likelihood of green purchase. Therefore, if we can motivate online consumers to value desirability more, they will be more likely to buy green products.  Our results on the desirability consideration indicate that the high-level cause-and-effect simulation design was successful in invoking individuals’ desirability consideration more than the partial simulation design. Also, the high-level simulation design showed the higher desirability consideration than the other conditions, the no simulation and the low-level simulation. Contrary 132 to our hypothesis, no significant difference between the low-level and the high-level simulation was found.  Nevertheless, this might be explained by the misfit between construal level and the message frame.  As suggested by White et al. (2011), the fit between construal level and the frame affects intention to perform an action (e.g., recycling). In particular, matching the low-level construal with the loss-framed message and paring the high-level construal with the gain-framed message will positively affect people’s intention to perform a behavior.  In our study, only loss framing, utility cost, was used. Thus, it might be possible that the misfit between the high-level construal and the loss framing reduces the impact of the high-level simulation on the desirability consideration and thus intention to buy a green product, resulting in no difference between the two full cause-and-effect simulations. Overall, the results reveal that regardless of the construal level the full simulation which presents both cause and effect reinforced people to select the greener products. This can be partly explained by the fact that the full simulation design, especially the high-level simulation, induced people’s desirability consideration more than the design without or with partial simulation. As a result, our study provides the reason why the cause-and-effect simulation can enable persuasion that Fogg (2002) does not give the clear explanation. Secondly, we found people in general valued the green attributes (e.g., energy use and water use) relatively highly. However, when they faced the situation that they needed to sacrifice the feasibility attribute to the desirability attribute, they tended to compromise the desirability as observed in the partial simulation condition. Interestingly, with the full cause-and-effect simulation 133 design, people did not change their values toward the green attributes much and sometimes increased their desirability consideration toward some green attributes. This suggests that the cause-and-effect simulation design helps promote green purchasing by keeping people’s desirability consideration on track.  The above findings are consistent with the findings of Zhao et al. (2007)  and Zhao et al. (2011). They propose that consumers have inconsistent preferences on their product choices and this results from temporal distance. In their study, they found in the near the time of purchase condition individuals valued more the product features associated with the process and in the far from the time of purchase condition they were concerned more about the product attributes related with the outcome. Specifically, at the point near the time of purchase the outcome thought is less salient, while at the point far from the time of purchase the process thought is less salient. Thus, in order to resolve this inconsistent preference, they propose that the less salient thought should be reinforced. The desirability consideration and the feasibility consideration parallel with the outcome-focused thought and the process-focused thought, described in Zhao et al. (2007) and Zhao et al. (2011), respectively. Our study offers another explanation to their predictions based on the view of CLT. Consequently, we empirically reconcile CLT and the concept of mental simulation. Thirdly, our study explains why green attitudes do not lead to pro-environmental behaviors. Forrester Research (2008) shows that positive attitudes toward green do not always lead to the actual green behaviors. We tested our research model based on TPB framework which includes attitudes toward purchasing a green product and self-efficacy. Our findings indicate that the 134 desirability consideration partially affect self-efficacy and did not significantly influenced attitudes.  It seemed that people initially had favorable attitudes toward purchasing green. However, when they were involved in making trade-offs for decision-making, self-efficacy reinforced intention to purchase to a greater extent than attitudes. Based on these results, we argue that without self-efficacy green attitudes might not be transformed into intention and thus behavior. This prediction is consistent with the original TPB’s proposition that perceived behavioral control which is constructed based on self-efficacy helps explain people’s intention and behavior (Ajzen 1985, 1991, 2002). However, in the context of green purchasing, self-efficacy plays more important role than attitudes. Fourthly, our study has provided additional support to the findings reported by Yu (2012). In her study, she found that a conditioning mechanism which serves as a positive or a negative cue could promote online green purchases. Her study mainly focuses on the affective route to persuasion by providing cues which induce people’s affect. Although our study focuses on the cognitive path by providing the additional information, the utility cost, to the users, it will serve as a cue to enable persuasion as well. This provides support that both affective and cognitive cues can persuade online consumers to select greener product models. This can be explained by Ajzen (1985). He proposes that the availability of new information influences people’s attitudes, intention, and behavior. Thus, both affective and cognitive cues which provide the additional piece of information to individuals will positively affect their green purchase behaviors.  135 In addition to the theoretical contribution, the overall results from our study have provided concrete guideline in developing an e-commerce website to persuade online consumers to buy green products. Although the significant difference between the low-level and the high-level cause-and-effect simulation was not achieved, we propose that providing the cause-and-effect simulation which demonstrates the relationship between the feasibility attribute (e.g., price) and the desirability attribute (e.g., utility cost) will reinforce online consumers’ to be more concerned about the desirability which, in turn, increase their self-efficacy and thus intention to buy green products. In sum, this study provides the effective design to enable persuasion in the online green purchase context and also the theoretical explanations of how and why this design works to promote online consumers to purchase a privately-consumed product such as a washing machine. 7.4 Limitations The major concern of our study is the unsuccessful mental construal manipulation. Prior CLT studies found the bi-directional relationship between temporal distance and mental construals (Trope and Liberman 2010). In other words, invoking people’s perception of temporal distance will influence how they construe their thoughts and this holds for the reverse as well. However, with the successful manipulation of temporal distance, the manipulation of mental construal was not successful.  This may be caused by the use of the website experiment. This brings up the issue regarding whether our manipulation of mental construal was not successful because the transient effect of the mental construal itself. It might also be possible that participants on MTurk were multitasking and thus did not engage in our task much. Although we could measure the time they 136 spent on our study, we could not control them to do only our task. Accordingly, this might diminish the effectiveness of our website manipulation. Another limitation of this study is the generalizability of the results. We are unsure whether the results we obtained from the MTurk’s sample could be generalized to online consumers in general. As participants on MTurk have a lot of experiences in responding to the survey and other human-related tasks (e.g., past experiences with the MTurk’s tasks greater than 5,000 tasks), it is possible that they did not respond to our instrument as normal online consumers do. They just know how to respond to the questionnaire survey. Finally, this study used a relatively small sample as suggested by non-significant and partial significant effects with the medium to large effect size. Collecting more data will help better target the true effect of our cause-and-effect simulation design that will give more considerable support to the effectiveness of our design in enabling persuasion. 7.5 Future Research This study has found evidence to support the use of the full cause-and-effect simulation. However, inconsistent with our expectation, the mental construal shows no difference in the desirability consideration and in intention to buy green product. According to White et al. (2011), they found that pairing low-level construals with loss-framed messages would positively influence the recycling behavior, while match high-level construals with gain-framed messages would be better to persuade people to do recycling. Based on these findings, the non-significant difference between the low-level and the high-level simulation might be caused by the misfit between the construal level and the message frame. As we employed only loss framing (utility cost) in our study, the 137 future study may explore the moderating effect of framing on the relationship between the simulation design and the desirability consideration. It is expected that pairing the high-level simulation with the utility saving framed in terms of gain may greater influence online consumers’ desirability consideration which, in turn, might increase their self-efficacy and thus intention to buy a green product. In addition, the future research may introduce the trade-off tool to the RA as studied by Xu et al. (2014). As in the current study most of participants did not recognize the trade-off as suggested by the open-ended responses, incorporating the RA with the trade-off tool which allows individuals to observe the correlations between attributes will bring trade-off between the feasibility and the desirability features more salient. Consequently, we expect that with the trade-off tool the high-level simulation will increase the desirability consideration than the other conditions. The results of our research imply that intention to purchase a green product measured by the green score associated with the product participants chose will predict the actual behavior. However, to ensure that the strong intention will lead to the actual behavior, the future study may capture the real behavior by offering the discounts on the product participants choose. This will give considerable support to the impact of the cause-and-effect simulation design on green purchasing. Moreover, the design of the cause-and-effect simulation will not be restricted to motivate the green purchasing behavior. The future research may examine the role of the simulation design in other contexts (e.g., recycling, exercising, quitting smoking). According to the results we found, the cause-and-effect simulation design will enhance self-efficacy that will increase the likelihood 138 people perform a certain behavior by highlighting the desirability feature to be more salient at a time they perform that action. 139 References Aaker, J. L., and Lee, A. Y. 2001. "“I” seek pleasures and “we” avoid pains: The role of self‐regulatory goals in information processing and persuasion," Journal of Consumer Research (28:1), pp 33-49. Ajewole, G. A. 1991. "Effects of discovery and expository instructional methods on the attitude of students to biology," Journal of Research in Science Teaching (28:5), pp 401-409. Ajzen, I. 1985. From intentions to actions: A theory of planned behavior, (Springer. Ajzen, I. 1991. "The theory of planned behavior," Organizational behavior and human decision processes (50:2), pp 179-211. Ajzen, I. 2002. "Perceived Behavioral Control, Self‐Efficacy, Locus of Control, and the Theory of Planned Behavior," Journal of Applied Social Psychology (32:4), pp 665-683. Ajzen, I., and Fishbein, M. 1977. "Attitude-behavior relations: A theoretical analysis and review of empirical research," Psychological bulletin (84:5), p 888. Ajzen, I., and Madden, T. J. 1986. "Prediction of goal-directed behavior: Attitudes, intentions, and perceived behavioral control," Journal of experimental social psychology (22:5), pp 453-474. American Water Works Association Research Foundation. 1999."Residential End Uses of Water," AWWA Research Foundation and American Water Works Association. Retrieved 9 July 2014 from http://www.waterrf.org/PublicReportLibrary/RFR90781_1999_241A.pdf Anderson, E., and Oliver, R. L. 1987. "Perspectives on behavior-based versus outcome-based salesforce control systems," The Journal of Marketing), pp 76-88. Angst, C. M., and Agarwal, R. 2009. "Adoption of electronic health records in the presence of privacy concerns: the elaboration likelihood model and individual persuasion," Mis Quarterly (33:2), pp 339-370. AYTM Market Research. 2013."US Internet users who would pay more for eco-friendly products, May 2013," AYTM Market Research. Retrieved 29 May 2013 from http://totalaccess.emarketer.com/Chart.aspx?R=139386&dsNav=Ntk:basic|green+AYTM+Market+Research+|1|,Ro:-1 Bamberg, S., and 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), pp 14-25. 140 Bandura, A. 1977. "Self-efficacy: toward a unifying theory of behavioral change," Psychological review (84:2), p 191. Bandura, A. 1989. "Regulation of cognitive processes through perceived self-efficacy," Developmental psychology (25:5), p 729. Bandura, A. 1994. Self‐efficacy, (Wiley Online Library. Barclay, D., Higgins, C., and Thompson, R. 1995. "The partial least squares (PLS) approach to causal modeling: Personal computer adoption and use as an illustration," Technology studies (2:2), pp 285-309. Barry Issenberg, S., McGaghie, W. C., Petrusa, E. R., Lee Gordon, D., and Scalese, R. J. 2005. "Features and uses of high-fidelity medical simulations that lead to effective learning: a BEME systematic review*," Medical teacher (27:1), pp 10-28. Bearden, W. O., and Etzel, M. J. 1982. "Reference group influence on product and brand purchase decisions," Journal of Consumer research), pp 183-194. Belei, N., Geyskens, K., Goukens, C., Ramanathan, S., and Lemmink, J. 2012. "The Best of Both Worlds? Effects of Attribute-Induced Goal Conflict on Consumption of Healthful Indulgences," Journal of Marketing Research (49:6), pp 900-909. Bhattacherjee, A., and Sanford, C. 2006. "Influence processes for information technology acceptance: an elaboration likelihood model," MIS quarterly), pp 805-825. Bickart, B. A., and Ruth, J. A. 2012. "Green eco-seals and advertising persuasion," Journal of Advertising (41:4), pp 51-67. Castaño, R., Sujan, M., Kacker, M., and Sujan, H. 2008. "Managing consumer uncertainty in the adoption of new products: temporal distance and mental simulation," Journal of Marketing Research (45:3), pp 320-336. Chaiken, S. 1980. "Heuristic versus systematic information processing and the use of source versus message cues in persuasion," Journal of personality and social psychology (39:5), p 752. Chaiken, S., and Stangor, C. 1987. "Attitudes and attitude change," Annual review of psychology (38:1), pp 575-630. Cone Communications. 2013."Frequency with which US Internet users consider the environmental impacts of their shopping, 2008-2013," 2013 Green Gap Trend Tracker. Retrieved 2 April 2013 from http://totalaccess.emarketer.com/Chart.aspx?R=137906&dsNav=Ntk:relevance|green+Cone+communications+|1|,Ro:2 141 Cook, T. D., Campbell, D. T., and Day, A. 1979. Quasi-experimentation: Design & analysis issues for field settings, (Houghton Mifflin Boston. Cordano, M., and Frieze, I. H. 2000. "Pollution reduction preferences of US environmental managers: Applying Ajzen's theory of planned behavior," Academy of Management Journal (43:4), pp 627-641. Crisp, R. J., and Turner, R. N. 2009. "Can imagined interactions produce positive perceptions?: Reducing prejudice through simulated social contact," American Psychologist (64:4), p 231. Davis, F. D. 1989. "Perceived usefulness, perceived ease of use, and user acceptance of information technology," MIS quarterly), pp 319-340. Davis, F. D., Bagozzi, R. P., and Warshaw, P. R. 1989. "User acceptance of computer technology: a comparison of two theoretical models," Management science (35:8), pp 982-1003. De Jong, T., and Van Joolingen, W. R. 1998. "Scientific discovery learning with computer simulations of conceptual domains," Review of educational research (68:2), pp 179-201. Dekkers, J., and Donatti, S. 1981. "The integration of research studies on the use of simulation as an instructional strategy," The Journal of Educational Research), pp 424-427. Diekmann, A., and Preisendörfer, P. 1992. "Persónliches umweltverhalten: Diskrepanzen zwischen Anspruch und Wirklichkeit," Kölner Zeitschrift für Soziologie und Sozialpsychologie). Duchastel, P. 1990. "Instructional strategies for simulation-based learning," Journal of Educational Technology Systems (19:3), pp 265-276. Earley, P. C., Northcraft, G. B., Lee, C., and Lituchy, T. R. 1990. "Impact of process and outcome feedback on the relation of goal setting to task performance," Academy of Management Journal (33:1), pp 87-105. Eisenhardt, K. M. 1985. "Control: Organizational and economic approaches," Management science (31:2), pp 134-149. Ellen, P. S., Wiener, J. L., and Cobb-Walgren, C. 1991. "The role of perceived consumer effectiveness in motivating environmentally conscious behaviors," Journal of Public Policy & Marketing), pp 102-117. Energy Star. 2014."Appliance Calculator." Retrieved September 25 2014 from www.energystar.gov/sites/default/files/asset/document/appliance_calculator.xlsx 142 Environment Canada. 2014."Wise Water Use." Retrieved 29 September 2014 from https://www.ec.gc.ca/eau-water/default.asp?lang=En&n=F25C70EC-1 Escalas, J. E. 2004. "Imagine yourself in the product: Mental simulation, narrative transportation, and persuasion," Journal of Advertising (33:2), pp 37-48. Escalas, J. E., and Luce, M. F. 2003. "Process versus outcome thought focus and advertising," Journal of Consumer Psychology (13:3), pp 246-254. Escalas, J. E., and Luce, M. F. 2004. "Understanding the Effects of Process‐Focused versus Outcome‐Focused Thought in Response to Advertising," Journal of Consumer Research (31:2), pp 274-285. Eyal, T., Liberman, N., Trope, Y., and Walther, E. 2004. "The pros and cons of temporally near and distant action," Journal of Personality and Social Psychology (86:6), p 781. Eyal, T., Sagristano, M. D., Trope, Y., Liberman, N., and Chaiken, S. 2009. "When values matter: Expressing values in behavioral intentions for the near vs. distant future," Journal of Experimental Social Psychology (45:1), pp 35-43. Farrell, M. H. J. 2012."Higher prices prevent some consumers from going green," Consumer reports. Retrieved 24 September 2012 from http://www.consumerreports.org/cro/news/2012/09/higher-prices-prevent-some-consumers-from-going-green/index.htm Fishbein, M. 1963. "An investigation of the relationship between beliefs about an object and the attitude toward that object," Human relations). Fishbein, M., and Ajzen, I. 1975. Belief, attitude, intention and behavior: An introduction to theory and research, ( Fogg, B. J. 2002. "Persuasive technology: using computers to change what we think and do," Ubiquity (2002:December), p 5. Forrester Research. 2008."Green attitudes don't guarantee green actions," Forrester Research. Retrieved 11 August 2008 from http://www.forrester.com/Green+Attitudes+Dont+Guarantee+Green+Actions/fulltext/-/E-RES46818 Forrester Research. 2014."US eCommerce forecast: 2013 to 2018," Forrester Research. Retrieved 12 May 2014 from http://www.forrester.com/US+eCommerce+Forecast+2013+To+2018/fulltext/-/E-RES115513?intcmp=blog:forrlink 143 Fransson, N., and Gärling, T. 1999. "Environmental concern: Conceptual definitions, measurement methods, and research findings," Journal of environmental psychology (19:4), pp 369-382. Freund, A. M., Hennecke, M., and Mustafić, M. 2012. "16 On Gains and Losses, Means and Ends: Goal Orientation and Goal Focus Across Adulthood," The Oxford handbook of human motivation). Freund, A. M., Hennecke, M., and Riediger, M. 2010. "Age-related differences in outcome and process goal focus," European Journal of Developmental Psychology (7:2), pp 198-222. Friestad, M., and Wright, P. 1994. "The persuasion knowledge model: How people cope with persuasion attempts," Journal of consumer research), pp 1-31. Froehlich, J., Findlater, L., and Landay, J. Year. "The design of eco-feedback technology," Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, ACM2010, pp. 1999-2008. Fujita, K., Eyal, T., Chaiken, S., Trope, Y., and Liberman, N. 2008. "Influencing attitudes toward near and distant objects," Journal of Experimental Social Psychology (44:3), pp 562-572. Fujita, K., Trope, Y., Liberman, N., and Levin-Sagi, M. 2006. "Construal levels and self-control," Journal of personality and social psychology (90:3), p 351. Gallup Poll. 2014."Environment," Gallup Poll. Retrieved July 9 2014 from http://www.gallup.com/poll/1615/environment.aspx GlobalScan. 2013."Environmental concerns "at record lows": global poll," GlobalScan. Retrieved 25 Febuary 2013 from http://www.globescan.com/images/images/pressreleases/2013-Enviro-Radar/globescan_press_release_enviroconcern_03-25-2013.pdf Goldenberg, D., Andrusyszyn, M.-A., and Iwasiw, C. 2005. "The effect of classroom simulation on nursing students' self-efficacy related to health teaching," The Journal of nursing education (44:7), pp 310-314. Gredler, M. E. 2004. "Games and simulations and their relationships to learning," Handbook of research on educational communications and technology (2), pp 571-581. GreenBiz. 2010."Price, Performance Still Obstacles to Increased Sales of Green Products." Retrieved 18 October 2014 from http://www.greenbiz.com/news/2010/05/24/price-performance-obstacles-increased-sales-green-products Greenwald, A. G. 1968. "Cognitive learning, cognitive response to persuasion, and attitude change," Psychological foundations of attitudes), pp 147-170. 144 Grier, S. A., and Deshpandé, R. 2001. "Social dimensions of consumer distinctiveness: The influence of social status on group identity and advertising persuasion," Journal of Marketing Research (38:2), pp 216-224. Griskevicius, V., Tybur, J. M., and Van den Bergh, B. 2010. "Going green to be seen: status, reputation, and conspicuous conservation," Journal of personality and social psychology (98:3), p 392. Guagnano, G. A., Stern, P. C., and Dietz, T. 1995. "Influences on attitude-behavior relationships a natural experiment with curbside recycling," Environment and behavior (27:5), pp 699-718. Han, H., Hsu, L.-T. J., and Sheu, C. 2010. "Application of the theory of planned behavior to green hotel choice: Testing the effect of environmental friendly activities," Tourism Management (31:3), pp 325-334. Hardisty, D. J., Griffin, D., Shim, Y., and Sun, D. 2015. "Tackling Temporal Tradeoffs: Using Product Labels to Activate Latent Goals,"). Harris Interactive. 2012."Importance of environmental issues when making purchase decisions according to US Internet users, 2009-2012," The Harris Poll. Retrieved 30 May 2012 from http://totalaccess.emarketer.com/chart.aspx?R=122378 Hershfield, H. E., Goldstein, D. G., Sharpe, W. F., Fox, J., Yeykelis, L., Carstensen, L. L., and Bailenson, J. N. 2011. "Increasing saving behavior through age-progressed renderings of the future self," Journal of Marketing Research (48:SPL), pp S23-S37. Hines, J. M., Hungerford, H. R., and Tomera, A. N. 1987. "Analysis and synthesis of research on responsible environmental behavior: A meta-analysis," The Journal of environmental education (18:2), pp 1-8. Homer, P. M., and Kahle, L. R. 1990. "Source expertise, time of source identification, and involvement in persuasion: An elaborative processing perspective," Journal of Advertising (19:1), pp 30-39. Huffington Post and YouGov. 2013."Poll finds Americans less concerned about the environment now than when Earth Day began." Retrieved 22 April 2013 from http://www.huffingtonpost.com/2013/04/22/environment-poll-earth-day_n_3117003.html Kahneman, D., and Tversky, A. 1979. "Prospect theory: An analysis of decision under risk," Econometrica: Journal of the Econometric Society), pp 263-291. Keller, P. A., Lipkus, I. M., and Rimer, B. K. 2003. "Affect, framing, and persuasion," Journal of Marketing Research (40:1), pp 54-64. 145 Kelman, H. C. 1958. "Compliance, identification, and internalization: Three processes of attitude change," Journal of conflict resolution), pp 51-60. Kim, J. S. 1984. "Effect of behavior plus outcome goal setting and feedback on employee satisfaction and performance," Academy of Management Journal (27:1), pp 139-149. Kim, Y., and Choi, S. M. 2005. "Antecedents of Green Purchase Behavior: An Examination of Collectivism, Environmental Concern, and PCE," Advances in Consumer Research (32:1). Kirmani, A., and Campbell, M. C. 2004. "Goal seeker and persuasion sentry: How consumer targets respond to interpersonal marketing persuasion," Journal of Consumer Research (31:3), pp 573-582. Kollmuss, A., and 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), pp 239-260. Korn, D., and Mattison, L. 2012. "Do Savings Come Out in the Wash?," in Home Energy. Kraiger, K., Ford, J. K., and Salas, E. 1993. "Application of cognitive, skill-based, and affective theories of learning outcomes to new methods of training evaluation," Journal of applied psychology (78:2), p 311. Krugman, H. E. 1965. "The impact of television advertising: Learning without involvement," Public opinion quarterly (29:3), pp 349-356. Laroche, M., Bergeron, J., and Barbaro-Forleo, G. 2001. "Targeting consumers who are willing to pay more for environmentally friendly products," Journal of consumer marketing (18:6), pp 503-520. Law, A. M., Kelton, W. D., and Kelton, W. D. 1991. Simulation modeling and analysis, (McGraw-Hill New York. Lee, Y. E., and Benbasat, I. 2010. "Interaction design for mobile product recommendation agents: Supporting users' decisions in retail stores," ACM Transactions on Computer-Human Interaction (TOCHI) (17:4), p 17. Lee, Y. E., and Benbasat, I. 2011. "Research Note-The Influence of Trade-off Difficulty Caused by Preference Elicitation Methods on User Acceptance of Recommendation Agents Across Loss and Gain Conditions," Information Systems Research (22:4), pp 867-884. Leigh, G. T. 2008. "High-fidelity patient simulation and nursing students' self-efficacy: A review of the literature," International Journal of Nursing Education Scholarship (5:1), pp 1-17. 146 Liberman, N., Sagristano, M. D., and Trope, Y. 2002. "The effect of temporal distance on level of mental construal," Journal of experimental social psychology (38:6), pp 523-534. Liberman, N., and Trope, Y. 1998. "The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory," Journal of personality and social psychology (75:1), p 5. Liberman, N., Trope, Y., and Stephan, E. 2007a. "Psychological distance," Social psychology: Handbook of basic principles (2), pp 353-383. Liberman, N., Trope, Y., and Wakslak, C. 2007b. "Construal level theory and consumer behavior," Journal of Consumer Psychology (17:2), pp 113-117. Lindenberg, S., and Steg, L. 2007. "Normative, gain and hedonic goal frames guiding environmental behavior," Journal of Social issues (63:1), pp 117-137. Loock, C.-M., Staake, T., and Thiesse, F. 2013. "MOTIVATING ENERGY-EFFICIENT BEHAVIOR WITH GREEN IS: AN INVESTIGATION OF GOAL SETTING AND THE ROLE OF DEFAULTS," Mis Quarterly (37:4). Love, R. E., and Greenwald, A. G. 1978. "Cognitive responses to persuasion as mediators of opinion change," The Journal of Social Psychology (104:2), pp 231-241. Maddux, J. E., and Rogers, R. W. 1983. "Protection motivation and self-efficacy: A revised theory of fear appeals and attitude change," Journal of experimental social psychology (19:5), pp 469-479. Mainieri, T., Barnett, E. G., Valdero, T. R., Unipan, J. B., and Oskamp, S. 1997. "Green buying: The influence of environmental concern on consumer behavior," The Journal of social psychology (137:2), pp 189-204. Mathieson, K. 1991. "Predicting user intentions: comparing the technology acceptance model with the theory of planned behavior," Information systems research (2:3), pp 173-191. Maynard, M. 2007."Say 'hybrid' and many people will hear 'Prius'," The New York Times. Retrieved 4 July 2007 from http://www.nytimes.com/2007/07/04/business/04hybrid.html?_r=1& McGuire, C. H. 1976. "Construction and Use of Written Simulations,"). Meyers-Levy, J., and Malaviya, P. 1999. "Consumers' Processing of Persuasive Advertisements: An Integrative Framework of Persuasion Theories," Journal of marketing (63:4). 147 Meyvis, T., Goldsmith, K., and Dhar, R. 2012. "The importance of the context in brand extension: how pictures and comparisons shift consumers' focus from fit to quality," Journal of Marketing Research (49:2), pp 206-217. Miller, G. A. 1956. "The magical number seven, plus or minus two: some limits on our capacity for processing information," Psychological review (63:2), p 81. Miller, G. A., and Chapman, J. P. 2001. "Misunderstanding analysis of covariance," Journal of abnormal psychology (110:1), p 40. Mithas, S., Khuntia, J., and Roy, P. K. 2010. "Green information technology, energy efficiency, and profits: Evidence from an emerging economy,"). Nelson, B. L., Carson, J. S., and Banks, J. 2001. Discrete event system simulation, (Prentice hall. Nunnally, J. C., and Bernstein, I. 1994. "The assessment of reliability," Psychometric theory (3), pp 248-292. Oinas-Kukkonen, H., and Harjumaa, M. 2009. "Persuasive Systems Design: Key Issues, Process Model, and System Features," Communications of the Association for Information Systems (24). Olsen, M. C., Slotegraaf, R. J., and Chandukala, S. R. 2014. "Green Claims and Message Frames: How Green New Products Change Brand Attitude," Journal of Marketing). Pavlou, P. A., and Fygenson, M. 2006. "Understanding and predicting electronic commerce adoption: an extension of the theory of planned behavior," MIS quarterly), pp 115-143. Peck, J., and Wiggins, J. 2006. "It just feels good: customers' affective response to touch and its influence on persuasion," Journal of Marketing (70:4), pp 56-69. Petty, R. E., and Cacioppo, J. T. 1984. "The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion," Journal of personality and social psychology (46:1), p 69. Petty, R. E., and Cacioppo, J. T. 1986a. "Communication and persuasion: Central and peripheral routes to attitude change,"). Petty, R. E., and Cacioppo, J. T. 1986b. The elaboration likelihood model of persuasion, (Springer. Petty, R. E., Haugtvedt, C. P., and Smith, S. M. 1995. "Elaboration as a determinant of attitude strength: Creating attitudes that are persistent, resistant, and predictive of behavior," Attitude strength: Antecedents and consequences (4), pp 93-130. 148 Petty, R. E., Wegener, D. T., and Fabrigar, L. R. 1997. "Attitudes and attitude change," Annual review of psychology (48:1), pp 609-647. Pham, L. B., and Taylor, S. E. 1999. "From thought to action: Effects of process-versus outcome-based mental simulations on performance," Personality and Social Psychology Bulletin (25:2), pp 250-260. Pichert, D., and Katsikopoulos, K. V. 2008. "Green defaults: Information presentation and pro-environmental behaviour," Journal of Environmental Psychology (28:1), pp 63-73. Reigeluth, C. M., and Schwartz, E. 1989. "An instructional theory for the design of computer-based simulations," Journal of Computer-based instruction). Rieber, L. P. 1996. "Seriously considering play: Designing interactive learning environments based on the blending of microworlds, simulations, and games," Educational technology research and development (44:2), pp 43-58. Rieber, L. P., Tzeng, S.-C., and Tribble, K. 2004. "Discovery learning, representation, and explanation within a computer-based simulation: Finding the right mix," Learning and instruction (14:3), pp 307-323. Rogers, R. W. 1975. "A protection motivation theory of fear appeals and attitude change1," The Journal of Psychology (91:1), pp 93-114. Roney, C. J., Higgins, E. T., and Shah, J. 1995. "Goals and framing: How outcome focus influences motivation and emotion," Personality and Social Psychology Bulletin (21:11), pp 1151-1160. Sawyer, J. E. 1992. "Goal and process clarity: Specification of multiple constructs of role ambiguity and a structural equation model of their antecedents and consequences," Journal of Applied Psychology (77:2), p 130. Schmitz, A., and Stamminger, R. 2014. "Usage behaviour and related energy consumption of European consumers for washing and drying," Energy Efficiency), pp 1-18. Schumacher, T. Year. "Simulation and attitude change: Evolving the research model," System Sciences, 1997, Proceedings of the Thirtieth Hawaii International Conference on, IEEE1997, pp. 648-654. Schwartz, S. H. 1977. "Normative influences on altruism," Advances in experimental social psychology (10), pp 221-279. Stern, P. C. 2000. "New environmental theories: toward a coherent theory of environmentally significant behavior," Journal of social issues (56:3), pp 407-424. 149 Stern, P. C., Dietz, T., Abel, T., Guagnano, G. A., and Kalof, L. 1999. "A value-belief-norm theory of support for social movements: The case of environmentalism," Human ecology review (6:2), pp 81-98. Stern, P. C., Dietz, T., and Kalof, L. 1993. "Value orientations, gender, and environmental concern," Environment and behavior (25:5), pp 322-348. Sussman, S. W., and Siegal, W. S. 2003. "Informational influence in organizations: an integrated approach to knowledge adoption," Information Systems Research (14:1), pp 47-65. Tam, K. Y., and Ho, S. Y. 2005. "Web personalization as a persuasion strategy: An elaboration likelihood model perspective," Information Systems Research (16:3), pp 271-291. Taylor, S. E., and Pham, L. B. 1996. "Mental Simulation, Mativation, and Action," The psychology of action: Linking cognition and motivation to behavior), p 219. The World Economic Forum 2014. "Global Risks 2014, Ninth Edition." Thompson, D. V., Hamilton, R. W., and Petrova, P. K. 2009. "When Mental Simulation Hinders Behavior: The Effects of Process‐Oriented Thinking on Decision Difficulty and Performance," Journal of Consumer Research (36:4), pp 562-574. Thurman, R. A. 1993. "Instructional simulation from a cognitive psychology viewpoint," Educational technology research and development (41:4), pp 75-89. Trope, Y., and Liberman, N. 2000. "Temporal construal and time-dependent changes in preference," Journal of personality and social psychology (79:6), p 876. Trope, Y., and Liberman, N. 2003. "Temporal construal," Psychological review (110:3), p 403. Trope, Y., and Liberman, N. 2010. "Construal-level theory of psychological distance," Psychological review (117:2), p 440. Trope, Y., Liberman, N., and Wakslak, C. 2007. "Construal levels and psychological distance: Effects on representation, prediction, evaluation, and behavior," Journal of consumer psychology: the official journal of the Society for Consumer Psychology (17:2), p 83. Tversky, A., and Kahneman, D. 1981. "The framing of decisions and the psychology of choice," Science (211:4481), pp 453-458. Tversky, A., and Kahneman, D. 1992. "Advances in prospect theory: Cumulative representation of uncertainty," Journal of Risk and uncertainty (5:4), pp 297-323. 150 Ülkümen, G., and Thomas, M. 2013. "Personal Relevance and Mental Simulation Amplify the Duration Framing Effect," Journal of Marketing Research (50:2), pp 194-206. Vallacher, R. R., and Wegner, D. M. 1987. "What do people think they're doing? Action identification and human behavior," Psychological review (94:1), p 3. Vallacher, R. R., and Wegner, D. M. 1989. "Levels of personal agency: Individual variation in action identification," Journal of Personality and Social Psychology (57:4), p 660. Vermunt, J. D. 1996. "Metacognitive, cognitive and affective aspects of learning styles and strategies: A phenomenographic analysis," Higher education (31:1), pp 25-50. Wang, J., and Lee, A. Y. 2006. "The role of regulatory fit on information search and persuasion," Journal of Marketing Research (43:1), pp 28-38. Wells, G. L., and Gavanski, I. 1989. "Mental simulation of causality," Journal of Personality and Social Psychology (56:2), p 161. White, K., MacDonnell, R., and Dahl, D. W. 2011. "It's the mind-set that matters: The role of construal level and message framing in influencing consumer efficacy and conservation behaviors," Journal of Marketing Research (48:3), pp 472-485. Wikipedia. 2014."Washing Machine." Retrieved 9 July 2014 from http://en.wikipedia.org/wiki/Washing_machine#Spinning Woolley, A. W. 2009. "Means vs. ends: Implications of process and outcome focus for team adaptation and performance," Organization Science (20:3), pp 500-515. Xiao, B., and Benbasat, I. "Designing Warning Messages for Detecting Biased Online Product Recommendations: An Empirical Investigation," Information Systems Research). Xiao, B., and Benbasat, I. 2007. "E-commerce product recommendation agents: use, characteristics, and impact," Mis Quarterly (31:1), pp 137-209. Xu, J., Benbasat, I., and Cenfetelli, R. T. 2014. "The nature and consequences of trade-off transparency in the context of recommendation agents," Mis Quarterly (38:2), pp 379-406. Yu, T. 2012. Design of a Persuasive Recommendation Agent to Promote Environmentally Friendly Products, University of British Columbia. Yu, T., Benbasat, I., and Cenfetelli, R. "How to Design Interfaces for Product Recommendation Agents to Influence the Purchase of Environmenally-Friendly Products,"). 151 Zhao, M., Hoeffler, S., and Dahl, D. W. 2009. "The role of imagination-focused visualization on new product evaluation," Journal of Marketing Research (46:1), pp 46-55. Zhao, M., Hoeffler, S., and Zauberman, G. 2007. "Mental simulation and preference consistency over time: The role of process-versus outcome-focused thoughts," Journal of Marketing Research (44:3), pp 379-388. Zhao, M., Hoeffler, S., and Zauberman, G. 2011. "Mental simulation and product evaluation: The affective and cognitive dimensions of process versus outcome simulation," Journal of Marketing Research (48:5), pp 827-839. Zimmerman, B., and Schunk, D. 2006. "Competence and control beliefs: Distinguishing the means and ends," Handbook of educational psychology), pp 349-367. Zimmerman, B. J., and Kitsantas, A. 1997. "Developmental phases in self-regulation: Shifting from process goals to outcome goals," Journal of educational psychology (89:1), p 29. Zimmerman, B. J., and Kitsantas, A. 1999. "Acquiring writing revision skill: Shifting from process to outcome self-regulatory goals," Journal of educational Psychology (91:2), p 241. Zimmerman, B. J., and Schunk, D. H. 2004. "Self-regulating intellectual processes and outcomes: A social cognitive perspective," Motivation, emotion, and cognition: Integrative perspectives on intellectual functioning and development), pp 323-349.  152 Appendices 153 Appendix A  Scenario Study In order to evaluate the effect of low-level and high-level construals on individuals’ decision-making and to understand what information individuals were concerned when making a product decision, we conducted the scenario experiment with a between-subject design. We expected that the results of this scenario study reveal the important information individuals consider when they involved either the low-level or the high-level mental construals. The cause of the cause-and-effect mental simulation was individuals’ decision on the product’s attributes. At this stage, we focused on how individuals thought about the process, the low-level construals, and the outcome, the high-level construals, both of which are considered the effect part of the cause-and-effect relationship. The mental simulation involved two conditions: 1) the low-level construal and 2) the high-level construal. Participants were randomly assigned into three conditions including 1) no construal, 2) the low-level construal, and 3) the high-level construal. The detailed design was described below. A.1 Task Product Two models of the washing machine were used in our scenario study based on the information of the real washing machines in the marketplace and we adjusted some information for the purpose of our manipulation. In order to reduce participants’ preference of the brand, we concealed the brands of the product and referred to each model as either model 1 or model 2 for the less-green and the greener washing machine respectively. Before we ran the scenario study, we conducted the product evaluation test to ensure that these two models are comparable in terms of desirability and quality. There would be non-dominated products. 154 Product Attributes Model 1 (Less Green) Model 2 (Greener) Price ($) 1,099.99 1,199.99 Model Front-load Front-load Capacity (cu.ft.) 5.0 5.0 Energy usage (kWh/load) 0.33 0.29 Sound reduction system Best Standard Automatic temperature control Yes Yes Water usage (Liters/load) 24.5 9.0 Spin speed (rpm) 1,300 1,400 Fast wash Yes – 15 minutes No Warrantee labor/parts (yr.) 1/2 1/1 Table 16: Two product models used in the scenario study We conducted the product evaluation survey with eight judges. The results suggest that judges considered model 2 a greener product than model 1 (MModel 1 = 2.81, SDModel1 = 0.84, MModel2 = 5.56, SDModel2 = 0.82). Thus, our green product manipulation was successful. In addition, the results reveal that there was no dominant product. Although the intention to purchase the green model (Model 2) based on a seven-point Likert scale indicate that people were slightly more likely to purchase the greener choice (M = 4.25, SD = 1.75), half of the judges chose 155 model 1 and the other half selected model 2. However, model 1 showed slightly higher overall evaluation scores based on a seven-point Likert scale than model 2 (MModel 1 = 5.61, SDModel1 = 0.80, MModel2 = 5.30, SDModel2=1.01). Based on these results, the two product models were comparable. Figure 29 presents the screenshot used in the scenario study. The product image of the two models were randomly presented to reduce the effect of the product appearance on participants’ decision-making.  Figure 29: The screenshot of two product models used in the scenario study A.2 Manipulation We developed one scenario for the control condition and two scenarios to manipulate participants’ thought toward either the process or the outcome associated with the washing machine purchase. Those in the control condition read as follows: 156 “Purchasing a washing machine Assume that you plan to purchase a washing machine. After your evaluation of all the models available in the market, you have found the following two models, which generally match your preferences. As a result, you are now making a choice between these two models. Once you choose one of the following two models, you will be asked to answer some questions regarding your choice. The information for the two models are as follows: (Note: You can find the explanation when you hover over a blue product attribute.)” Those in the low-level condition which focused on the process associated with buying and doing laundry read as follows: “Purchasing a washing machine Assume that you plan to purchase a washing machine. After your evaluation of all the models available in the market, you have found the following two models, which generally match your preferences. As a result, you are now making a choice between these two models. While you are looking at the information of each model’s specifications, we would like you to focus on the specific features of each model and think about the process of using each model to do your laundry. As you consider these two alternatives, focus on how you would use each washing machine to do your laundry. 157 For example:     What actions or steps would you take to do your laundry? Please take some time below to explain how you would go about using each washing machine to do your laundry. The information for the two models are as follows: (Note: You can find the explanation when you hover over a blue product attribute.)” Those in the high-level condition which were asked to think about the outcome related with purchasing and using the washing machine to do laundry read as follows: “Purchasing a washing machine Assume that you plan to purchase a washing machine. After your evaluation of all the models available in the market, you have found the following two models, which generally match your preferences. As a result, you are now making a choice between these two models. While you are looking at the information of each model’s specifications, we would like you to focus on the reasons for buying and using each model. Think about the outcomes of using each model to do your laundry. As you consider these two alternatives, focus on why you would buy and use each washing machine to do your laundry. For example: 158      What are the underlying reasons for buying and using either washing machine to do your laundry?      For what purposes would you buy and use either washing machine to do your laundry? Please take some time below to explain why you would go about using either washing machine to do your laundry. The information for the two models are as follows: (Note: You can find the explanation when you hover over a blue product attribute.)” In the control condition, there was no manipulation. We followed the manipulation of the low-level and the high-level construals as well as the process and the outcome mental simulation based on prior studies (e.g. Trope and Liberman 2010; Woolley 2009; Zhao et al. 2011; Zimmerman and Schunk 2004). The instruction of low-level and high-level construal manipulation was derived from Zhao et al. (2011) and White et al. (2011) to create our study’s scenarios. The low-level condition focused on the procedures associated with using a washing machine to do laundry. The procedures included both tasks and resources. Performance of the two alternatives varied among three dimensions: operating the washing machine, heating the water, and drying clothes. These dimensions are the most salient to overall performance (Korn and Mattison 2012) —involve doing laundry and require electricity and water to complete.  On the other hand, the high-level condition emphasizes on the outcomes of buying and using the washing machine to do laundry. We adopted the concept of high-level construal and 159 operationalized it in terms of the reasons why participants buy and utilize the washing machine. If the low-level condition is operationalized based on the definition of the outcome simulation in mental simulation literature, this type of simulation will present the benefits of buying and using the product or be framed in terms of gain. This may cause the inequality between the process and the outcome simulation, since the process is neutral and not shaped by the positive frame. As a result, using the concept of high-level construal which focused on the abstract level of representation might reduce the positive-frame effect and thus enabled the equivalence between the two conditions apart from the type of simulation. A.3 Experimental Procedure We used Amazon Mechanical Turk (MTurk) to recruit our participants. Participants who lived in the US and were willing to join our online survey were asked to sign the consent form and to complete the survey questionnaire. Before we ran the scenario study, we ran the two pilot studies to test the experimental procedure and the measurement items. In the pilot tests, two surveys were employed. The first survey involved demographic information and control variables including product expertise, product experience, environmental concern, and green purchasing behavior. To reduce the hypotheses guessing problem, a week later we provided participants with the link to access the second survey. Participants were randomly assigned into three scenarios. The second survey required participants to choose the model of washing machine they intend to buy. They read their scenarios and answered the questions about the dependent and the control variables.  160 After conducting the pilot studies, we adjusted the price of model 2, the experimental procedure, and measurement items. The price of model 2 was changed from $1,199 to $1,399 (the real price of model 2), as most participants in the pilot studies stated that the price difference between the two model was minimal, and approximately 73% of participants chose model 2 without the main effect of the simulation type.  We ran the main scenario study with 64 participants. Four participants who failed the attention check questions were excluded. The final sample was 60 (control = 20, low-level = 19, and high-level = 21). In the main scenario study, one survey employed, since we found that implementing the implicit and objective measures would mitigate the hypotheses guessing and social desirability biases.  They were asked to answer the questions capturing demographic and control variables, and were randomly assigned into three conditions. Their task was to choose a product model they were interested in buying. They read the scenarios and answered the questions regarding the dependent and the control variables. After completing the survey, they received $6 as a participation reward. In order to motivate participants to take an experimental task seriously, they are told before the experiment that the participants who provide very detailed, serious, and diligent responses will receive an additional reward of $6. A.4 Measurement We employed a questionnaire self-report to measure our variables as well as the objective measures. Questionnaire items were adapted from existing scales and newly developed if existing scales are unavailable. These items were investigated in terms of content validity, construct validity (convergent and discriminant), and internal consistency reliability. The factor analysis was 161 conducted. The results indicate that all items loaded highly on the respective constructs, with the loadings greater 0.7, had adequate reliability, with the Cronbach’s Alpha of all constructs greater than 0.7, composite reliability greater than 0.9, and the square roots of average variance extracted (AVE) greater than 0.7. Thus, the measurement show acceptable reliability, convergent, and discriminant validity.  All measurement items were randomly presented in order to reduce mono-method bias. Table 17 presents the flow of measurement items. 162 Measurements Descriptions 1. Demographic Age, education, marital status, gender, and income 2. Control variables Product expertise, product purchase and use experiences, and product decision factors 3. Scenario questions Either explain the process or the outcome 4. Product decision Intention to purchase a green product 5. Product decision factors Ranking the product attributes in order of the importance to participants’ product decision 6. Manipulation check Process- and outcome-focused simulation 7. Control variable Involvement 8. Dependent variables Awareness of consequences, perceived consumer effectiveness, self-efficacy, attitudes toward purchasing a green product 9. Control variables Environmental concern, and green purchase behavior Table 17: The scenario study's measurement flow A.5 Data Analysis The results reveal that there were no significant differences in terms of age, education, marital status, gender, and income among three conditions (p > .05). Also, the Chi-square test for the purchasing experience and the frequency of doing laundry indicate non-significant difference 163 among three conditions (p > .05). These results suggest that participants in all three conditions were comparable. The process thought manipulation was not successful (MControl = 5.42, MLow = 5.91, MHigh = 6.06; F(2, 57) = 1.45, p = .24, ηp2 = .05). The outcome thought was also not successful (MControl = 5.88, MLow = 6.09, MHigh = 6.41; F(2, 57) = 1.45, p = .17, ηp2 = .06). However, the medium effect size of the outcome manipulation check suggests that there may be a difference between conditions. It appears that participants in the outcome condition thought about the reasons of buying and choosing the product more than the other two conditions. The unsuccessful manipulation may result from the less salient effect of the thought manipulation in the online study. Although we could not capture how participants on MTurk did our task, they were suspected of being multitasking. This might influence their responses to the manipulation check items. As a result, the objective data were analyzed to assess the success of the thought manipulation. The number of thoughts regarding the process of using each of the two washing machine models and the reason of buying and using each of the two models were analyzed. One-way ANOVA was conducted to evaluate the differences in the number of process thoughts and the outcome thoughts between the process and the outcome condition. The assumption regarding the homogeneity of variances was not met. The results reveal that there was a significant difference in the number of process thoughts between the two conditions (MLow = 8.37, MHigh = 0.10; F(1, 38) = 124.59, p < .001, ηp2 = .77). Participants in the low-level condition thought about how to use each of the washing machine models more than those in the high-level condition. In terms of the number of the outcome thoughts, the significant difference between the two condition was found (MLow = 164 0.89, MOutcome = 6.48; F(1, 38) = 97.57, p < .001, ηp2 = .72). Participants in the high-level condition thought about the reasons why they would buy and use each of the two models than those in the process condition. As a results, the process and the outcome thought manipulation were successful. Furthermore, the total number of thoughts which combine both process and outcome thoughts were analyzed using ANOVA. The assumption regarding the homogeneity of variances was violated. The significant difference between the two conditions was found with participants in the low-level condition had greater number of total thoughts than those in the high-level condition (MLow = 9.26, MHigh = 6.57; F(1, 38) = 8.56, p = .01, ηp2 = .18). Though the results from the objective measures supported that our manipulation was achieved, we could not rule out the possibility that people in general (the control condition) may exert some thoughts about the process and the outcome. Most of studies in CLT and the mental simulation tested only two conditions—the low-level and the high-level. Therefore, we did not have the baseline to assess participants in the control condition. The relationships between the manipulation and attitudes toward purchasing a green product as well as awareness of consequences were analyzed. ANOVA was conducted on attitudes toward purchasing a green product and awareness of consequences separately.6 The results show that there were no significant effect of the manipulation of thought on attitudes (MControl = 4.95, MLow = 5.77,                                                  6 MANOVA was not conducted, because of a significant Box’s test which suggests that the homogeneity of variance-covariance matrix was not met. 165 MHigh = 5.27; F(2, 57) = 2.06, p = .14, ηp2 = .07) and awareness of consequences (MControl = 5.70, MLow = 5.67, MHigh = 5.87; F(2, 57) = 0.27, p = .77, ηp2 = .01). However, the medium effect size for attitudes reveals that there may be the difference between conditions. Participants in the low-level condition had more positive attitudes than those in the other two conditions, whereas those in the high-level condition had more positive attitudes than those in the control condition. This lends slight support that the thought manipulation may influence attitudes, thought the results were not consistent with our expectation that those in the high-level had the positive attitudes than any other condition. Additionally, we assessed the control variables to ensure that there was no confounds. We conducted ANOVA to evaluate the differences in product expertise, involvement, time spent in the entire study, time spent in the scenario, environmental concern, and green purchasing behavior between conditions. The results reveal no differences in product expertise (MControl = 3.84, MLow = 3.74, MHigh = 3.69; F(2,57) = 0.11, p = .89, ηp2 = .00), involvement (MControl = 5.98, MLow = 5.72, MHigh = 5.94; F(2,57) = 0.36, p = .70, ηp2 = .01), time spent in the entire study (MControl = 13.74, MLow = 17.31, MHigh = 17.41; F(2,57) = 1.43, p = .25, ηp2 = .05), environmental concern (MControl = 5.33, MLow= 5.14, MHigh = 5.19; F(2,57) = 0.20, p = .82, ηp2 = .01), and green purchasing behavior (MControl = 4.35, MLow = 4.28, MHigh = 3.93; F(2,57) = 0.62, p = .54, ηp2 = .02).  However, the significant difference in time spent in the scenario was found (MControl = 1.11, MLow = 6.26, MHigh = 6.40; F(2, 57) = 18.82, p < .001, ηp2 = .40). The post-hoc analysis using Games-166 Howell7 indicate that those in the control condition spent significantly less amount of time in the scenario than the other two conditions, while there was no significant difference between those in the low-level and the high-level condition. However, these results would be interpreted with caution, as the study did not take place in the laboratory where participants did only a task. Participants on MTurk might did other tasks while they participated in our study. To capture the green product decision-making (intention to purchase a green product), we implemented two measures: 1) the discrete measure (model 1 or model 2) and 2) the scale measure (1 = model 1 to 7 = model 2). According to Figure 30, there might be no difference in the product choice between the control and the process condition, as the proportions of the product choices regarding model 1 and model 2 were quite close between the two conditions. On the contrary, the product choice in the control condition might be significantly different from that of the other two conditions. However, Chi-square test was conducted to evaluate the difference in the product choice between conditions. The results suggest that there was no significant difference between conditions (χ2 (2, 57) = 1.55, p > .05). In addition, the scale measure was evaluated using ANOVA. The results indicate that there was no significant difference between conditions with the small effect size (MControl = 3.65, MLow = 4.00, MHigh = 3.05; F(2, 57) = 1.15, p = .34, ηp2 = .04).                                                   7 Homogeneity of variance assumption was not met. 167  Process – low-level condition, outcome – high-level condition Figure 30: The product models chosen by participants in the scenario study Taking the results from both analyses together, we concluded that in all conditions participants were less likely to buy the greener model (model 2) and thus there was no effect of the thought manipulation on people’s decision-making. Nevertheless, when taking the closer look at the means of the scale measure, we found that participants in the control and the process condition perceived no difference between buying model 1 and model 2, as their group means were close to 4, whereas those in the outcome condition were less likely to buy the greener model. These results were contradict to our expectations that people in the low-level and the high-level condition will be more likely to buy the greener choice that those in the control condition, and that those in the high-level purchased the greener one than those in the low-level. As mentioned earlier, one possible explanation might be the less salient effect of the thought manipulation itself. Another explanation might be that some participants did not like both choices, but were forced to choose one. In this case, the product choice did not reflect their true preference. 168 In order to evaluate the effect of the construal level on a green product decision, we assessed the change of the ranking scores of the product attributes. We assumed that if people rank the green product attributes (e.g., energy usage, spin speed, water usage) more important than other attributes (e.g., price, fast wash, etc.), the thought manipulation may be successful in persuading people to value the green. Next, we assess the ranking scores. Participants were asked to rank the ten product attributes in order of their importance to their decision regarding washing machine purchase before the manipulation and after the manipulation (1 = the most important to 10 = the least important). To evaluate the baseline of the ranking score, the ranking scores before the manipulation were analyzed using ANOVA. There were no differences in the rankings of all attributes between conditions (p > .05). This suggests that participants considered product attributes in the similar way when they make a washing machine choice. Top five attributes were price, capacity, energy, water, and warranty in descending order of the importance. This suggests that participants considered green attributes—energy and water usage—important in some extents. The simple mixed design was conducted to assess the effect of the simulation type (3 conditions) and time (before and after the manipulation) on the ranking. See Table 18 for means and the significant difference based on the simple effect analysis, and Table 19 for the results. The results reveal that for most attributes there were significant main effects of time as well as interaction effects between time and construal level. The main effect of the construal level was found significant only for price. While there was no difference in the ranking between before and after the manipulation for the high-level condition, the significant differences were found for the 169 control and the low-level condition with participants in both condition perceived prices less important after the manipulation. Additionally, according to the simple effect analysis using Bonferroni correction, we found the interesting findings which were inconsistent with our predictions in the high-level condition. Participants in the outcome condition perceived that green attributes (energy usage and water usage) were less important and that non-green attributes (fast wash and sound reduction) were more important. When considering water usage, although there was no significant simple effect, only participants in the high-level condition perceived water usage more important.  170 Attribute Control condition Process condition Outcome condition Before After Before After Before After Auto temp control 6.55 5.85a 6.53 6.58 7.52 7.38a Energy usage 5.05 5.25 4.47 3.84 3.86a 5.86a Fast wash 6.25 5.60 7.11 6.53 7.29a 5.38a Model 6.70 7.45 5.84 6.53 6.52 6.62 Price 1.95a 3.10a 2.21b 4.32b 1.81 1.81 Sound reduction 7.90a 6.70 a 8.37b 7.05b 8.67c 5.10c Spin speed 7.40 6.80 7.74a 6.32a 7.19 6.76 Warranty 4.85 4.45 5.11a 6.32a 4.76 4.86 Water usage 5.10 5.60 4.37 4.00 4.38 a 6.38 a Capacity 3.25a 4.20a 3.26 3.53 3.00b 4.86b a, b, c Significant at the .05 level, same letter indicates the significant difference based on the simple effect analysis using Bonferroni. Table 18: Means of the product attribute ranking  171 Attribute Construal Level Time Time x Construal Level F p-value ηp2 F p-value ηp2 F p-value ηp2 Auto temp control 2.67 .08* .09 1.52 .22 .03 1.10 .34 .03 Energy usage 0.95 .39 .03 1.95 .17 .03 4.36 .02** .13 Fast wash 0.99 .38 .03 6.33 .02** .10 1.10 .34 .04 Model 0.55 .58 .02 2.03 .16 .03 0.34 .71 .01 Price 3.19 .05* .10 12.24 .00** .18 3.86 .03** .12 Sound reduction 1.00 .37 .03 45.00 <.001** .44 6.69 .00** .19 Spin speed 0.04 .96 .00 5.71 .02** .09 0.79 .46 .03 Warranty 1.20 .31 .04 0.96 .33 .02 2.33 .11 .08 Water usage 1.71 .19 .06 3.24 .08* .05 3.10 .05* .10 Capacity 0.63 .54 .02 16.31 <.001** .22 3.33 .04** .11 * Significant at the .1 level, ** Significant at the .05 level Table 19: Simple mixed design for product attribute ranking In summary, the results from the scenario study reveal that the concept of construals might influence individuals’ decision-making on the products to some extent. However, the mixed results 172 which were not consistent with our expectation suggest the thought manipulation was not salient enough to get participants engaged in the manipulation. Thus, we hoped that leveraging the role of technology assists individuals in the low-level and in the high-level to be involved in the construal manipulation, thereby motivating them to choose more green products. 173 Appendix B  Pilot Test for Feasibility and Desirability Consideration with respect to Product Attributes In order to ensure that the price attribute reflects the feasibility and the three green attributes which include energy use, spin speed, and water use are perceived as the desirability regarding the washing machine purchase.  In this pilot study, we evaluated how each product attribute was associated with the feasibility or the desirability feature associated with the purchasing behavior. This will ensure that participants perceived that price was related more to feasibility and the three green attributes were associated more with the desirability. Therefore, five items were developed employing a seven-point semantic differential scale for each product attribute. The low anchor and the high anchor indicate the feasibility and the desirability consideration associated with the product attribute respectively. 174 Feasibility vs. desirability consideration Consider the <product attribute> of the washing machine. Please evaluate this washing machine attribute according to the following aspects: 1. Feasibility (the ease of buying the washing machine) - Desirability (the values of buying the washing machine) 2. How you would buy the washing machine (the steps taken to buy the washing machine) - Why you would buy the washing machine (the reasons for buying the washing machine) 3. The outcome of buying the washing machine - The process of buying the washing machine (Reverse) 4. The means of buying the washing machine - The ends of buying the washing machine 5. Short-term outcomes resulting from buying the washing machine - Long-term outcomes resulting from buying the washing machine B.1 Results on the Product Attribute Evaluation Ten participants were recruited using Amazon Mechanical Turk. They were asked to rate seven washing machine attributes in terms of the feasibility and the desirability. The results indicate that relative to other attributes price was perceived as the least desirability (M = 3.72, SD = 1.74). On the contrary, water use and energy use were the top two attributes which participants perceived as the most desirable (MW = 5.43, SDW = 1.22, ME = 5.36, SDE = 1.18). The spin speed lied relatively toward the desirability (MS = 4.93, SDS = 1.17). The overall results lend support to our proposition that price was associated with less the desirability with respect to purchasing a washing machine and the three green attributed were related more to the feasibility regarding buying a product. 175 Scale Item Attribute  Cycle Time Energy Use Price Sound Level Spin Speed Warranty Water Use Feasibility – Desirability 4.49 4.08 2.75 4.97 4.88 5.23 4.62 How – Why 4.20 5.09 4.23 4.77 5.11 4.63 5.64 Process – Outcome 5.17 5.97 3.89 5.08 5.75 4.97 5.56 Means – Ends 5.11 5.67 4.00 4.79 4.29 4.95 5.36 Short-Term Outcomes – Long-Term Outcomes 4.27 5.97 3.73 4.34 4.59 6.42 5.96 Average 4.65 5.36 3.72 4.93 4.79 5,24 5.43 Table 20: The evaluation of the feasibility and the desirability of the product attributes 176 Appendix C  Participant Background Information  177  Frequency Percentage Age   18 - 24 years 10 13.2 25 - 34 years 40 52.6 35 - 44 years 19 25.0 45 - 54 years 6 7.9 55 - 64 years 1 1.3 Education Level   High school graduate 12 15.8 Completed some college 19 25.0 Associate degree 10 13.2 Bachelor's degree 29 38.2 Master's degree 4 5.3 Professional degree 2 2.6 Marital Status   Single (never married) 45 59.2 178  Frequency Percentage Married 26 34.2 Separated 1 1.3 Divorced 4 5.3 Gender   Female 41 53.9 Male 35 46.1 Income   Less than $25,000 25 32.9 $25,000 - $34,999 17 22.4 $35,000 - $49,999 17 22.4 $50,000 - $74,999 11 14.5 $75,000 - $99,999 4 5.3 $100,000 - $149,999 1 1.3 $150,000 - $199,999 1 1.3 Have you ever purchased a washing machine?   179  Frequency Percentage Yes 58 76.3 No 18 23.7 If “Yes”, when did you last purchase a washing machine?   Less than a month ago 1 1.3 In the last 1 - 3 months 1 1.3 In the last 4 - 6 months 10 13.2 In the last 7 - 9 months 2 2.6 In the last 10 - 12 months 9 11.8 More than 12 months ago 35 46.1 If “Yes”, how much did you pay for the washing machine last time?   Less than $500 13 17.1 $500 - $699 23 30.3 $700 - $899 11 14.5 $900 - $1,099 6 7.9 180  Frequency Percentage $1,100 - $1,299 3 3.9 $1,300 - $1,499 2 2.6 If “No”, when do you intend to purchase a washing machine?   Never 1 1.3 In the next 4 - 6 months 4 5.3 In the next 7 - 9 months 1 1.3 In the next 10 - 12 months 2 2.6 More than 12 months from now 10 13.2 Have you ever used a washing machine to do laundry?   Yes 75 98.7 No 1 1.3 If “Yes”, about how often do you use a washing machine to do laundry?   Less than once a week 3 3.9 About once a week 27 35.5 181  Frequency Percentage About twice a week 21 27.6 About three times a week 13 17.1 About four times a week 4 5.3 About five times a week 3 3.9 About six times a week 2 2.6 More than seven times a week 2 2.6 Table 21: Participant background information 182 Appendix D  Scenarios of the Main Study D.1 Scenario for the No Simulation Condition Participants in the no simulation condition read the following scenario: “Purchasing a washing machine Assume that you plan to purchase a washing machine for yourself and your family of four. You are looking for a 4.2 Cu.Ft. front-load washing machine which is appropriate for the family of four like yours. After your evaluation of all available online websites, you have found a website called Home Appliance Group (homeappliancegroup.com), which provides a wide selection of 4.2 Cu.Ft. washing machine models that fit your space. As a result, assume you are now entering the Home Appliance Group website to buy a washing machine. You can spend as much time as you like on the website. Once you choose the washing machine model you seriously consider purchasing for you and your family, you will be asked to answer the questions regarding the website design and your purchasing decision. Please click next to visit the Home Appliance Group website.  In order to enter the website and access the post-questionnaire survey, please copy the verification number provided on the next page.” 183 D.2 Scenario for the Partial Simulation, the Low-Level Simulation, and the High-Level Simulation Participants in the partial simulation, the low-level simulation, and the high-level simulation read the following scenario: “Purchasing a washing machine Assume that you plan to purchase a washing machine for yourself and your family of four. You are looking for a 4.2 Cu.Ft. front-load washing machine which is appropriate for the family of four like yours. After your evaluation of all available online websites, you have found a website called Home Appliance Group (homeappliancegroup.com), which provides a wide selection of 4.2 Cu.Ft. washing machine models that fit your space. The website provides a recommendation tool which will more effectively help you find the right model based on your preferences.  As a result, assume you are now using the recommendation tool on the Home Appliance Group website to buy a washing machine. The recommendation tool will ask you to indicate your preferences toward each washing machine attribute, such as spin speed and warranty. You will also be asked to rank these attributes as to how important these are to you (e.g., is spin speed more important than warranty). Ranking will be used to generate the washing machine alternatives recommended to you when you click on “submit”. The tool will then give the product recommendations which fit your preferences and ranking. The product recommendations are ordered based on the fit to your needs in a descending sequence, with the model best fitting your needs at the top, followed by models which fit your needs but to a lesser degree. 184  You will be asked to use the recommendation tool to help you find the washing machine model you want. You can use the recommendation tool as many times as you like. Once you choose the model you seriously consider purchasing for you and your family, you will be asked to answer the questions regarding the recommendation tool design and your purchasing decision. Please click next to visit the Home Appliance Group website.  In order to enter the website and access the post-questionnaire survey, please copy the verification number provided on the next page.” 185 Appendix E  RA’s Algorithm and Utility Cost Formula The RA’s algorithm and the utility cost were calculated based on the low-level unit. To derive the output of the high-level unit, the algorithm and the formula was calculate in the same way. E.1 RA’s Algorithm Step 1: Normalize the stated preference value and the product model’s value for each product attribute (see Table 22) Normalize value = (Value – Worst Value)/(Best Value – Worst Value) Attribute Worst Value Best Value Price 1199.99 499.99 Energy use 1.5 0.3 Cycle time 60 30 Sound level 60 40 Spin speed 400 1600 Water use 85 14 Warranty 12 24 Worst value – the value of the product attribute that is worst to a consumer’s decision-making compared to that of other alternatives (e.g., the most expensive price, the shortest warranty), best value – the value of the product attribute that is the best to a consumer’s decision-making relative to that of other alternatives (e.g., the least expensive price, the longest warranty) 186 Table 22: Best and worst attribute values Step 2: Calculate the misfit score of each product model j (j = 1 to 60) for each product attribute i (i = 1 to 7) Case 1: For a product attribute i (i = 1 to 7), if the normalized stated preference for a product attribute is less than the normalized value of the product model j (j = 1 to 60), then the misfit score of the product model j for the product attribute i is 0. Case 2: For a product attribute i (i = 1 to 7), if the normalized stated preference for a product attribute is greater than the normalized value of the product model j (j = 1 to 60), then the misfit score of the product model j for the product attribute i is calculated by the following formula:  𝑀𝑖𝑠𝑓𝑖𝑡 𝑆𝑐𝑜𝑟𝑒𝑗𝑖= (𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑆𝑡𝑎𝑡𝑒𝑑 𝑃𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑖 −𝑁𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑉𝑎𝑙𝑢𝑒𝑗𝑖)  × (𝑅𝑒𝑣𝑒𝑟𝑠𝑒 𝑅𝑎𝑛𝑘𝑖  ÷ 𝑇𝑜𝑡𝑎𝑙 𝑅𝑎𝑛𝑘) Remark: Reverse rank is calculated such that the most important attribute which was ranked 1 had the reverse rank value of 7 and the least important attribute which was assigned 7 had the reverse rank value of 1. The total rank is calculated by summing 1, 2, 3, …, and 7, that equals 28. Step 3: Calculate the fit score of each product model j (j = 1 to 60) by the following formula: 𝐹𝑖𝑡 𝑆𝑐𝑜𝑟𝑒 (%) = ∑ (1 −𝑀𝑖𝑠𝑓𝑖𝑡 𝑆𝑐𝑜𝑟𝑒𝑖)  × 1007𝑖=1   187 E.2 Utility Cost Calculation To calculate the utility cost per load and the 10-year utility cost for the low-level simulation and the high-level simulation, the assumptions and data from Energy Star (2014) were used as references. However, annual residential default loads were adjusted to 300 loads per year in our current study. Table 23 presents the assumptions and reference data.  Table 24 represents how to calculate the utility cost. Electricity used in the drying process was calculated based on the relationship between spin speed and residual moisture content. The basic premise is the faster the spin speed is, the less the residual moisture content left in clothes and thus the less energy the electric dryer consumes. The formula was derived from Schmitz and Stamminger (2014). Assumption and Data Value US average electricity cost ($) 0.1128 per kWh US average water cost ($) 0.0317 per liter Residential default loads per year 295 Table 23: Assumption and Data for the Utility Cost Calculation Energy Star (2014) 188  Electricity Cost (Washing + Drying) Water Cost Total Utility Cost Utility Cost per Load [Energy Use + (14.936 – (0.007 x Spin Speed)] x 0.1128 Water Use x 0.0317 Electricity cost + water cost 10-year Utility Cost [Energy Use + (14.936 – (0.007 x Spin Speed)] x 0.1128 x 300 x 10 Water Use x 0.0317 x 300 x 10 Electricity cost + water cost Table 24: Utility cost formula 189 Appendix F  Measurement Items of the Main Study F.1 Pre-Questionnaire Survey 190 Construct Item Response Source Demographics Age What is your age? Under 18 years    18 - 24 years    25 - 34 years    35 - 44 years    45 - 54 years    55 - 64 years    65 years or older  Education What is your education level? Completed some high school    High school graduate    Completed some college    Associate degree    Bachelor's degree    Master's degree  191 Construct Item Response Source   Professional degree    Doctoral degree  Marital Status What is your marital status? Single (never married)    Married    Separated    Widowed    Divorced  Gender What is your gender? Female    Male  Income What was your income before taxes during the past 12 months? Less than $25,000    $25,000 - $34,999    $35,000 - $49,999    $50,000 - $74,999    $75,000 - $99,999  192 Construct Item Response Source   $100,000 - $149,999    $150,000 - $199,999    $200,000 or more  Product Purchase and Use Experience Product Purchase Experience Have you ever purchased a washing machine? Yes    No  If “Yes” When did you last purchase a washing machine? Less than a month ago    In the last 1 - 3 months    In the last 4 - 6 months    In the last 7 - 9 months    In the last 10 - 12 months  193 Construct Item Response Source   More than 12 months ago   Which brand of the washing machine did you purchase last time? Bosch    Danby    Electrolux    GE    Haier    LG    Samsung    Whirlpool    Other - please specify:   How much did you pay for the washing machine last time? Less than $500    $500 - $699    $700 - $899  194 Construct Item Response Source   $900 - $1,099    $1,100 - $1,299    $1,300 - $1,499    $1,500 or more  If “No” When do you intend to purchase a washing machine? Never    Less than a month from now    In the next 1 - 3 months    In the next 4 - 6 months    In the next 7 - 9 months    In the next 10 - 12 months    More than 12 months from now  Product Use Experience Have you ever used a washing machine to do laundry? Yes  195 Construct Item Response Source   No  If “Yes” About how often do you use a washing machine to do laundry? Less than once a week    About once per week    About twice per week    About three times per week    About four times per week    About five times per week    About six times per week    About seven times per week    More than seven times a week  Product Expertise Please evaluate yourself in the following aspects: Seven semantic differential scales  196 Construct Item Response Source  I do not know about washing machines at all. - I know much more about washing machines than anyone else.    I have no idea about the attributes of washing machines. - I am knowledgeable about all attributes of washing machines.    If people ask me for help with buying washing machines, I am not confident that I can give them good advice. - If people ask me for help with buying washing machines, I am confident that I can give them good advice.   197 Construct Item Response Source  Other people know that I am not an expert about washing machines. - Other people know that I am an expert about washing machines.   Trap 1 I am not a human. – I am a human.   Pre Product Attribute Ranking  Please rank the following washing machine attributes in order of the importance to your purchasing decision on the washing machine (the most important attribute at the top): Cycle Time    Energy Use    Price    Sound Level  198 Construct Item Response Source   Spin Speed    Warranty    Water Use  Table 25: Pre-questionnaire survey 199 F.2 Post-Questionnaire Survey 200 Construct Item Response Source Manipulation Check Temporal Distance Please evaluate the website in the following aspects: Seven-semantic differential scales  TD1 The website presents the information which focuses on today's purchase. - The website presents the information which focuses on future's purchase.   TD2 I thought that purchasing a washing machine would be done very recently. - I thought that purchasing a washing machine would be done very long time from now.   TD3 I thought that the time frame for when I would buy a washing machine was very soon. - I thought that the time frame for when I would buy a washing machine was very distant.   TD4 In my opinion, purchasing a washing machine is a shorter-term action. - In my opinion, purchasing a washing machine is a longer-term action.   TD5 I think that purchasing a washing machine has a shorter-term impact. - I think that purchasing a washing machine has a longer-term impact.   201 Construct Item Response Source TD6 I believe that in the short run purchasing a washing machine does matter. - I believe that in the long run purchasing a washing machine does matter.   Mental Construal  Please evaluate the website in the following aspects: Seven-semantic differential scales  MC1 The website presents very concrete information. - The website presents very abstract information.   MC2 The information the website provides is very specific. - The information the website provides is very general.    MC3 The website gives very detailed information. - The website gives very broad information.   MC4 The information the website provides is very superficial to my goals in purchasing a washing machine. - The information the website provides is very central to my goals in purchasing a washing machine.   202 Construct Item Response Source MC5 The information the website provides is very irrelevant to my goals in purchasing a washing machine. - The information the website provides is very relevant to my goals in purchasing a washing machine.   MC6 The website presents the information which focuses on “how” to do laundry. – The website presents the information which focuses on “why” to do laundry.   Trap2 I do not read this statement. – I read this statement.   Dependent Variables Post Product Attribute Ranking Please rank the following washing machine attributes in order of the importance to your purchasing decision on the washing machine (the most important attribute at the top): Cycle Time    Energy Use    Price    Sound Level    Spin Speed    Warranty  203 Construct Item Response Source   Water Use    Cycle Time  Self-Efficacy To what extent to you agree or disagree with the following statements: 1 = Strongly disagree to 7 = Strongly agree Pavlou and Fygenson (2006) SE1 If I wanted to, I would be able to purchase the product which consumes less utilities (e.g., electricity).   SE2 If I wanted to, I am confident I could buy the product which consumes less utilities (e.g., electricity).   SE3 I am uncertain I could purchase the washing machine which consumes less utilities (e.g., electricity) if I wanted to. (Reverse)   Attitudes toward Purchasing a Green Product To what extent to you agree or disagree with the following statements:  1 = Strongly disagree to 7 = Strongly agree Pavlou and Fygenson (2006) AP1 For me, purchasing the product which consumes less utilities (e.g., electricity) would be a good idea.   AP2 For me, purchasing the product which consumes less utilities (e.g., electricity) would be desirable.   204 Construct Item Response Source AP3 I think that purchasing the product which consumes less utilities (e.g., electricity) would be a waste of my money. (Reverse)   Trap3 If you read this statement, please select 'disagree'.   Control Variables Involvement To what extent do you agree or disagree with the following statements: 1 = Strongly disagree to 7 = Strongly agree Bhattacherjee and Sanford (2006) and Sussman and Siegal (2003) INV1 I am involved in making decisions about purchasing a washing machine.   INV2 Purchasing a washing machine is relevant for me.   INV3 Purchasing a washing machine is important for me.   Environmental Concern To what extent do you agree or disagree with the following statements: 1 = Strongly disagree to 7 = Strongly agree Kim and Choi (2005) EC1 I am extremely worried about the state of the world's environment and what it will mean for my future.   205 Construct Item Response Source EC2 Mankind is severely abusing the environment.   EC3 When humans interfere with nature, this often produces disastrous consequences.   EC4 Humans must live in harmony with nature in order to survive.   Green Purchasing Behavior To what extent do you agree or disagree with the following statements: 1 = Strongly disagree to 7 = Strongly agree Kim and Choi (2005)  I make a special effort to buy products that are made from recycled materials.    When I have a choice between two products, I purchase the one less harmful to other people and the environment.    I make a special effort to buy products that are environmentally friendly.    I have avoided buying a product because it had potentially harmful environmental effects.   Trap4 If you read this statement, please select 'disagree'.   206 Construct Item Response Source Website Evaluation Please evaluate the recommendation tool in the following aspects: Seven semantic differential scales   Bad - Good    Dislike - Like    Not useful - Useful    Undesirable - Desirable    Unfavorable - Favorable    Not informative - Informative    Bad looking - Good looking   Reasons to Purchase a Product Reasons Why did you select the washing machine you chose to checkout? Please explain your reasons. Open-Ended  Comments Comments Do you have any other comments? Please add any additional questions, comments, concerns, and/or suggestions you may wish to share with us. Open-Ended  Table 26: Post-questionnaire survey 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0132742/manifest

Comment

Related Items