! CONSUMER PREFERENCE FOR PACKAGING MATERIALS : WILLINGNESS TO PAY A ND BARRIERS TO RECYCLING By Kimberly Klaiman A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Agricultural, Food, and Resource Economics - Master of Science 2015! ABSTRACT CONSUMER PREFERENCE FOR PACKAGING MATERIAL : WILLINGNESS TO PAY AND BARRIERS TO RECYCLING By Kimberly Klaiman This study focuses on the application of a stated preference elicitation method to asses s consumer preferences for packaging recyclability. In collaboration with top executives from food companies, a survey was developed in order to obtain current recycling habits from an online sample of 2,000 respondents, who self identified as the primary household decision -maker and grocery shopper. Two choice experiments were used to determine consumer willingness to pay (WTP) for packaging recyclability and barriers to recycling. This study contributes to the literature on individual packaging and its co mponents that increase or hinder recycling. The effectiveness of indirect questioning and information treatments on influencing consumer behavior was also analyzed using a between subject design. Our results show an average positive willingness to pay for packaging recyclability, with a fraction of respondents having an estimated negative willingness to pay. Willing ness to accept to clean in order to recycle has significant and positive for all treatment groups . The analysis of the information treatments showed that consumers were responsive to the video treatment by increasing their WTP for packaging recyclability, but not responsive to the infographic treatment. Evidence of social desirability bias was found in this analysis, suggesting more scrutiny sho uld be placed on estimates that do not address social desirability bia s. """!ACKNOWLEDGEMENTS First of all, I would like to express my gratitude to my research advisors, Dr. Cloe Garnache and Dr. David Ortega. Both provided me continuous support and advice needed to complete my research. Second , I want to thank my committee members, Dr. Sue Selke and Dr. Rob ert Shupp for the guidance and advice with my research project . Also, I would like to thank Dr. Frank Lupi for his input on the interpretation of my fi ndings. I would like to especially thank Michigan State UniversityÕs Center for Packaging Innovation and Sustainability for the opportunity to work with their members. Finally, I would like to thank my family , Sammie, Michelle, David, and Yogi for their support and encouragement th rough my masterÕs program. Also, I would like to thank my friends and classmates for their patience over the past year. "#!TABLE OF CONTENTS LIST OF TABLES .........................................................................................................................vi LIST OF FIGURES .....................................................................................................................viii 1. Introduction .................................................................................................................................1 1.1 Motivation .............................................................................................................................1 1.2 Research Questions ...............................................................................................................2 1.3 Objective ...............................................................................................................................2 1.4 Thesis Structure ....................................................................................................................3 2. Literature Review ........................................................................................................................4 2.1 Recycling Trends in the U.S. ................................................................................................4 2.2 Motives and Barriers to Recycling .......................................................................................4 2.3 Valuing Packagin g and Recycling ........................................................................................6 3. Theory .........................................................................................................................................8 3.1 Willingness to Pay ................................................................................................................8 3.2 Contingent Valuation Methods ...........................................................................................10 3.3 Caveats associated with CVM ............................................................................................11 3.3.1 Hypothetical Bias .........................................................................................................11 3.3.2 Social Desirability Bias ................................................................................................13 3.4 Discrete Choice Experiments ..............................................................................................13 3.5 Previous Studies Valuing Recycling With Choice Experiments ........................................14 3.6 Approach .............................................................................................................................14 4. Survey Method ..........................................................................................................................15 4.1 Survey Overview ................................................................................................................15 4.1.1 Sampling plan ..............................................................................................................15 4.2 Effe cts of Information .........................................................................................................16 4.3 Survey Content ....................................................................................................................18 4.4 Willingness to Pay Experiment ...........................................................................................19 4.5 Willingness to Recycle Experiment ....................................................................................21 5. Research Me thod ......................................................................................................................23 5.1 Theoretical Framework .......................................................................................................23 5.2 Empirical Model .................................................................................................................24 6. Data Analysis ............................................................................................................................28 6.1 Willingness to Pay for Recyclability Choice Experime nt..................................................29 6.2 Barriers to Recycling Choice Experiment ..........................................................................36 7. Discussion & Conclusion ..........................................................................................................42 7.1 Summary of Key Results ....................................................................................................42 7.2 Limitations ..........................................................................................................................43 7.3 Implications .........................................................................................................................43 #!APPENDICES ..............................................................................................................................45 Appendix A: Statistics for Select Recycling Habits .................................................................46 Appendix B: Full Survey ..........................................................................................................48 Appendix D: Individual WTP for Packaging Material by treatment group - OLS robust regressions .................................................................................................................................64 Appendix E: Barriers to Recycling Choice Experiment Supplemental Tables .......................68 Appendix F: Differe nces in WTP for packaging materials ......................................................69 Appendix G: Individual WTP for packaging recyclability .......................................................70 REFER ENCES........................ ......................................................................................................71 #"!LIST OF TABLES ! Table 1: Recycling Video Script..........................................................................................18 Table 2: Example choice set for packaging recyclability choice experiment ......................20 Table 3: Example choice set for willingness to recycling choice experiment .....................22 Table 4: Packaging materials choice experiment parameters ÉÉÉÉÉÉÉÉ.....ÉÉ27 Table 5: Barriers to recycling choice experiment parameters ÉÉÉÉÉÉÉÉÉ...É..27 Table 6: Descriptive statis tics of the observed sample population in percentage É...ÉÉ28 Table 7: WTP estimated derived from model estimation in preference space and WTP - space .........................................................................................................................................31 Table 8: Determinants of estimated individual WTP for packaging recyclability: OLS robust regression ÉÉÉÉÉÉÉÉÉÉÉÉÉ...........................................................É..34 Table 9: Comparing negative and positive WTP for packaging recycl ability ....ÉÉÉÉ35 Table 10: WTP for recyclability by material and treatment ....ÉÉÉÉÉÉÉÉÉÉÉ36 Table 11: WTA to Recycle Estimation in WTP -space ......ÉÉÉÉÉÉÉÉÉÉÉÉÉ37 Table 12: Determinates of estimated individual WTA to clean in order t o recycle: OLS robust regression ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉÉÉ.É...40 Table A -1: Packaging recyclability choice experiment in preference s pace ............................63 Table A -2: Packaging recyclability choice experiment in WTP -Spa ce...................................63 Table A -3: Individual WTP for plastic packaging by treatment gr oup....................................64 Table A -4: Individual WTP for glass packaging by treatment group......................................65 Table A -5: Individual WTP for aluminum packaging by treatment group..............................66 Table A -6: Individual WTP for carton packa ging by treatment group....................................67 Table A -7: Barriers to recycling CE estimated parameters in preference spac e.................. ....68 Table A -8: Determinants of Individual WTA to clean by treatment group.............................68 #""!Table A -9: T-test mean di fferences for packaging materials within treatment groups ........... 69 Table A -10: Individual WTP for packaging r ecyclability by treatment group ..........................70 #"""!LIST OF FIGURES ! Figue 1: Municipal Solid Waste Recycling Rates, 1960 -2013ÉÉÉÉÉÉÉ......ÉÉÉ1 Figue 2: Classi fication framework for methods to measure willingness -to-pay ....ÉÉÉÉ9 Figue 3: Recycling I nfographic ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ17 Figue 4: Recycling Video Snapshot .......ÉÉÉÉÉÉÉÉÉÉÉ..ÉÉÉÉÉÉÉÉ18 Figue 5: Density graph of estimated WTP for packaging recyclability in preference space and W TP-space ...É....ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.ÉÉÉÉÉ.............32 Figue 6: Density graph of estimated WTA to clean in order to recycle in preference space and WTP -space ........ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..... 38 Figue A -1: Importance of energy when deciding to recycle.................. ..................................46 Figue A -2: Importance of cost when deciding to recycle........................................................46 Figue A -3: Importance of time when deciding to recycle.......................................................47 Figue A -4: Environmental Warm Glow..................................................................................47 !$!1. Introduction 1.1 Motivation Total household trash generated in the United States is at an all time high (EPA, 2015), which poses environmental problems and wastes resources. Some of this increase can be attributed to the amount of packaging each household consumes. Recycling has been proposed as part of the solution to increasing trash problem. According to the EPA (2015), although the total amount of househ old waste that is recycled has been steadily increasing since the 1960s, the percent of household waste that is recycled has stagnated in recent years (Figure 1). Figure 1: Municipal Solid Waste Recycling Rates, 1960 -2013 Source: EPA, 2015 As the am ount of packaging consumed has increased, companies are pressured to take responsibility for the entire life cycle of their product packaging. Companies have responded by promoting increasing recycling of their packaging materials. For example, Coca -Cola h as released a green leaf recycling logo for their products in an effort to promote recycling behavior . Even though there has been an increase in overall packaging recycling, the percent of packaging !%!that is recycled has remained stagnant in recent years. Previous research has mainly focused on cumulative recycling habits, with relatively little work available on product -specific recycling. 1.2 Research Questions In this study, information was collected about U.S. consumer preference s for packaging recycla bility . In an online survey, answers were collected from consumers concerning their knowledge about recycling, motives for recycling, and socio -demographic information. The following hypotheses guide this study : 1) Consumers receive positive utility from packaging recyclability, resulting in an increase in consumer willingness to pay for the packaging and thus the entire product ; and 2) Consumers are willing to make tradeoffs between the benefits they gain from recycling and the effort and time needed to re cycl e. Since companies and policy makers have used commercials and recycling logos to support increased recycling, we test the effects of a video and an infographic to determine the effectiveness of certain information treatments on consumer valuation of r ecycling. This study also used a survey technique called i ndirect questioning to address issues of hypothetical bias in stated preference choice experiments. 1.3 Objective The objective of this thesis is to estimate demand for packaging recyclability for different materials. The purpose of this analysis is to help food, beverage, and packaging companies develop packaging strategies to improve consumer willingness to recycle. This information can also benefit the recycling industry as a whole by specifyin g factors that negatively affect consumerÕs recycling behavior . With the results of this study, government officials, waste !&!management companies, and packaging companies can select proper channels to incentivize consumers to recycle. 1.4 Thesis Structure This Thesis is comprised of 7 chapters and is organized as follows: Chapter 2 present s background information on consumer preference for recycling and packaging. Chapter 3 provide s theory concerning the experiments used in this thesis. Chapter 4 describe s the data presented in this study. Chapter 5 incorporate s the data methods used in this study. Chapter 6 present s the results and analysis of this study. Chapter 7 review s findings and suggest s discussion points for future research. Supplementa l documents are presented in the appendix. !'!2. Literature Review This chapter reviews both empirical and theoretical research in consumer attitudes and behavior s concerning recycling. Although most studies report consumers receive positive utility from recycling, all report that barriers, such as cost, public perception, and confusion, also affect recycling rates. 2.1 Recycling Trends in the U.S. Recycling rates have steadily increased s ince recycling trends were first measur ed in the U.S. during the 1960s . In 2013, 87.2 million tons of materials were recycled, compared to 15 million ton s in 1980 (EPA, 2012). Since 2005 , waste generat ed that was recycled has remained between 31 and 34% (EPA, 2012). The consumer packaging industry reported selling 4.2 trillion units o f retail products in 2011 (Euromonitor International, 2013). Of this total , food packaging represented a staggering 43% of volume share. Rapid improvement of packaging technology has resulted in growing demand throughout the world for packaging related to produce and ready -made meals. Other trends in food packaging are individualized packaging resulting from on-the -go food products , smaller household sizes, and rising health awareness, (Mintel, 2014). As a result, the food -packaging sector is expected to ha ve an annual growth rate of 3% between 2011 and 2016, thus increasing the need for research in the area of recycling. 2.2 Motives and Barriers to Recycling Using a questionnaire from before and after a recycling scheme, Perrin and Barton (2001) found th at the main reasons for recycling were convenience, concerns for future generations , the environment, and personal satisfaction. The study also found that the main reasons for not recycl ing were inconvenience, storage problems, and distance to recycling ce nters. Donohue !(!(2013) found that participants in a focus group concerning consumer packaging Òfeel better about their purchasing decisions if they believe they are more proactive about protecting the environmentÓ. There are many studies that have focused on motives and barriers to collective recycling rates. One example used panel data of county -level recycling rates in Minnesota, (Sidique, Joshi & Lupi, 2009) to review several policies (e.g. mandatory recycling regulations and increasing recycling educat ion expenditures) and their effects on recycling rates over a period of eight years. They found the biggest increase in recycling rates came from a variable pricing strategy, which is when households are charged more for larger trash cans, which decreases the relative cost of recycling bins. Through a study of recycling drop -off site participants, evidence was found that the distance to the drop -off site had a large impact on its usage (Sidique, Joshi, & Lupi, 2009). This suggests that the effort needed to recycle is a large determinate of consumers willingness to recycle. Their study concluded that higher income, older age, and larger household size were better predict ing factors of usage of a recycling center than gender or marital status (Sidique, Joshi, & Lupi, 2009). In a related study, r esearchers observed a natural experiment that distributed recycling bins to select participants and followed up with a questionnaire (Guagnano, Stern, & Dietz, 1995). As expected, households that were given a recycling bin were shown to increase their recycling rates more than the control households. They concluded that there could be several reasons for this change in behavior including : changed perception of social norms, increased convenience, or reduced cost of recyc ling. Gamba and Oskamp (1994) found that having specific knowledge about recycling was the strongest predictor of high recycling rates , followed by higher family income and higher number of people living in the household. All of these !)!studies suggest tha t consumers do receive positive utility from recycling, while there are barriers that need to be addressed before recycling rates can increase. Halvorsen (2008) used an in -person survey in Norway to study the opportunity cost to of spent recycling. The s tudy concluded that a consumerÕs increased opportunity cost of time recycling has a sever e negative effect on household recycling efforts. They also found that although social norms including warm -glow and the ÔGolden -ruleÕ had a significant effect on recy cling rates , mone tary incentives had the largest effect. 2.3 Valuing Packaging and Recycling There have been few p revious studies that have researched consumer preference for packaging attributes. Rokka and Uusitalo (2008) used a choice experiment to fi nd the relative importance of different packaging at tributes in consumersÕ choices, including resealability, brand, and recyclability. Their results found that respondents receive 34% of their overall product utility from packaging. A portion of their resp ondents (31%) placed environmentally -friendly packaging as the most important factor in their purchasing decision. When determining factors that contributed to valuing packaging sustainability, t hey found no strong correlation to any particular demographic variable; rather valuing packaging sustainability correlated more closely to common interests and preferences (Rokka and Uusitalo, 2008) . Another study used a discrete choice model, with examples of shoes and batteries in six countries, to find consumers place a value on the social aspects of a product (Auger, Devinney, Louviere, & Burke, 2007). This study found that consumers from more developed countries were more influenced by the social attributes (e.g. use of child labor or recyclability) of a produc t. This study also showed that individuals who had previous knowledge about the social attributes of the products were more likely to choose more socially positive products in the experiment. !*! Many studies have examined household willingness t o pay for cu rbside recycling. Wh ile some of these studies have been able to use revealed preference data to determine WTP for curbside recycling (for example, Aadland and Caplan, 2003; Guagnano, Stern, & Dietz, 1995), most of these studies have used choice experiments . Aadland and Caplan (2003) studied 1,000 households in Utah by phone interview, using a choice experiment. They had a direct comparison of revealed and stated preference for WTP for curbside recycling and found a mean WTP of $7.00 when asking in a hypothe tical scenario, and $6.71/month when asking actual users of a recycling program. Karousakis and Birol (2008) used a choice experiment to estimate an average consumer WTP of £2.68/month to have one additional material accepted for recycling in London. More recently, Ferreria and Marques (2015) used a survey in Portugal to find a mean WTP for monthly recycling service of !1.35 and !3.16 depending on whether protest answers were included. Protest answers are when respondents refuse to give an amount they are w illing to pay for a particular reason. They found that many of the protest answers had positive WTP for recycling, but respondents indicated that it was the governmentÕs duty to pay for waste management. These positive results on WTP for curbside recyclin g suggest that households may also display positive WTP for recyclable packaging materials. !+!3. Theory This chapter introduce s, explain s, and evaluate s the theories and instruments used in this study. These instruments are used to elicit consumer willin gness to pay (WTP) for packaging recyclability and recycling attributes. WTP is the maximum amount that a given consumer is willing to pay for (and thus value s) a good or service or a specific trait of a good or service . The advantage of estimating consume r WTP is that it provides pricing information that is meaningful for industry leaders and policymakers . 3.1 Willingness to Pay Willingness to pay measurements can be determined using sev eral different methods (Figure 2 ). The two most general categories a re revealed and stated preference methods. Revealed preference methods assume utility maximizing behavior and infer peopleÕs preferences from analyzing how individuals make choices . Studies using revealed preference often analyze choices made by individual s, including natural and designed experiments. Stated preference methods use answers to questions and surveys to elicit what individual s would have chosen and infer preference s for a non -market good. Stated preference method s are usually criticized because they are prone to bias given their hypothetical nature and may not perfectly replicate real world circumstances. As a result , when data is available, revealed preferences are typically favored because they more accurately reflect an individual Õs true pref erence s. However, revealed preference techniques are not alw ays available to a researcher as data might not be available . Stated preference techniques are usually more affordable and can be more easily utilized by the researcher to examine a specific rese arch question. !"!Figure 2: Classification framework for methods to measure willingness -to-pay Source: Breidert, C., Hahsler, M., & Reutterer, T., 2006 1 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!#!$%&'(&)*&'!+,-. ,'(%&!/*'0%12!,&1!30%(3*!*45*6(7*&'2!8,--!.&1*6!1(236*'*!30%(3*!,&,-92(2 !!"#!3.2 Contingent Valuation Methods Contingent Valuation Method ( CVM ) is a stated -preference techniqu e. This method was first developed to estimate the value of ecosystems and the flow of services that they provide. The method came under scrutiny after the 1989 Exxon Valdez oil spill , when a governmental panel, assembled by the National Oceanic and Atmosp heric Administration, was charged with reviewing the validity of the method for natural resource damage assessments. The panel concluded that CVM could produce reliable estimates as long as a set of guidelines is followed to reduce bias (Arrow et. , al . 1993). These guidelines include : having a strong, large, and representative sample, clear instructions , careful experiment al design, and specificity when describing the good being valued (Kling, Phaneuf, and Zhao, 2012). Types of CVMs include direct surveys s uch as hypothetical referendums , where participants are shown a price for a particular environmental service and aske d whether they would vote to have it provided . Carson (2012), points out that the most important element to consider in a contingent valua tion study is the Òface validityÓ of the survey instrument, which refers to whether the instrument poses a credible proposal to respondents and provide s all the necessary information in a clear format. More simply, the CV study has to be believable and tar get what the researcher wants to learn. Carson states that even though the CVM is not perfect, when done properly it can Òprovide a reliable basis for what the public is willing to trade off to obtain a non -market goodÓ (2012). CVM has been used in a wide variety of studies ranging from a consumersÕ WTP for existence value of environmental services to WTP for governmental services (Carson, 2000). In this study, a stated preference CVM survey is used to measure different consumer responses to certain traits , including the ir WTP for packaging recyclability. We address the issue of face !""!validity of our instrument by collaborating with top executives from food companies when designing our survey. 3.3 Caveats associated with CVM CVM has come under harsh review since the 1989 panel submitted their review of contingent valuation . CVM is prone to many biases , due to its hypothetical nature and lack of incentive compatibility, that must be addressed before experimental results can be accepted because of its hypotheti cal nature and lack of incentive compatibility . Two of the main issues that concern economists are hypothetical and social desirability bias. 3.3.1 Hypothetical Bias Hypothetical bias is the difference between what a person indicates that he/she would do versus what he/she would actually do in real life (Loomis, 2014). Hypothetical bias can be addressed through different survey designs . Possible ex ante survey tools include emphasizing the consequentiality of the respondentÕs choices, cheap talk, including an opt -out alternative, and urging honesty statements. Emphasizing the consequentiality of the respondentÕs choices has been shown to reduce bias by informing th e respondent that his/her answer could ultimately affect their actual wel l being (Carson, Grov es, and List, 2014 ). Cheap talk reduces hypothetical bias by stating the problem of hypothetical bias explicitly to the participant. Often cheap talk is a script that states that in past surveys consumers often overstate d their willingness to pay. Loomis Õ review of studies that used cheap talk, showed that while some studies were able to eliminate or reduce hypothetical bias , the effect was not universal (Loomis, 2014). Cummings and Taylor compared revealed and stated preference experiments in three separ ate studies and were able to eliminate hypothetical bias by using a cheap talk script (1999). However, they presented their cheap talk script in person and the script !"#!was much larger than would be convenient for an online or telephone survey. In a more cri tical review of cheap talk, Aadland and Caplan (2006) compared actual and hypothetical willingness to pay for a curbside recycling pro gram through a telephone survey. They found no reduction in hypothetical bias when using cheap talk. In fact, they found t hat with one of their longer, in -depth cheap talk scripts, hypothetical bias actually increased . They hypothesize that t his increase was caused when because direction of the hypothetical bias was not stated in the cheap talk used . Another study by Aasland and Caplan (2003), found that using cheap talk in their study was more effective on people who would tend to state relatively high WTP. When estimating WTP for curbside recycling, they found using cheap talk resulted with a decrease of 59 cents (8.43%) on average WTP. Despite the potential issues involved with cheap talk methods, many studies that use CVM to elicit stated preference use cheap talk to address potential hypothetical bias (Olynk Tonsor and Wolf, 2010, Lusk, 2003, & Aadland and Caplan, 2003). The u rging honesty method differs from cheap talk by not explicitly stating that people overstate or understate their true choice in a hypothetical scenario. R ather , urging honesty is commonly seen as a statement in the beginning of a survey, in which th e participant swears to answer questions truthfully. A study where students were asked to sign an oath of honesty found that signing the oath was able to eliminate hypothetical bias in the survey (Stevens, Tabatabaei, and Lass, 2013). Loomis (2014) suggest s r esearcher use either the cheap talk or urging honesty methods because there have been cases where using cheap talk along with urging honesty has overco rrected for hypothetical bias. Moreover, i ncluding an opt -out or null alternative in discrete choice experiments avoids forcing the individual to choose between goods and skewing results (Hensher, 2010). This better replicate s real world scenarios, since consumers often choose between similar goods or not purchasing the good at all. !"#!3.3.2 Social Desirabi lity Bias Social desirability bias has also been shown to affect CVM results (Nederhof, 1985). Social desirability bias is the tendency of consumers to resp ond in a manner that they view as socially favorable or in a way that they feel is pleasing to the researcher. Given that recycling can be associated with environmental stewardship and social responsibility , it is conceivable that respondents may exhibit social desirability bias when answering questions about recycling. Indirect questioning is an unobt rusive technique used to mitigate social desirability bias. Rather than ask ing what the respondent would do, as in a typical survey, indirect questioning asks the respondent what they think other people would do . Studies have shown that indirect questionin g has allowed researchers to reduce social desirability bias (Norwood & Lusk, 2011). Past research has shown that indirect questioning affect s respondentsÕ answers to variables that were subject to social influence, but does not affect respondentsÕ answers to variables that bear no social stigma (Fisher, 1993). The main assumption behind thi s technique is that respondents receive positive utility from their own socially desirable behaviors, yet receive no utility from saying others conform to socially desir able behavior (Norwood & Lusk, 2011). Olynk, Tonsor, and Wolf (2010) used direct and indirect questioning in comparing consumer food choices and found that the indirect questioning yield ed more accurate results. 3.4 Discrete Choice Experiments Choice experiments ( CE) are one type of stated preference tool used to elicit consumer preference s. CEs differ from other CVM instruments ( experimental referendums, expert judges , conjoint analysis, and customer surveys) in that consumers are asked to choose from a lternative bundles of attributes instead of ranking or rating them (Adamowicz, Boxall, Williams, and Louviere, 1998). In a typical CE, participants assess various levels of attributes across several !"#!options, similar to what a consumer faces when making dec isions in a real world scenario. In doing so, it elicit s stated preference s for a good. By comparing other CVM instruments and a CE in valuing passive value for protection for a forested area in Canada, researchers found that the welfare values obtained h ad smaller variances (relative to their means) when using CE compared to other CVM instruments (Adamowicz, Boxall, Williams, and Louviere, 1998). A choice experiment design was chosen for this study since it provides the researcher ability to study how con sumers weigh several different product attributes simultaneously. 3.5 Previous Studies Valuing Recycling With Choice Experiments 3.6 Approach Building and expanding on previous research, our study uses two choice experiments to estimate consumer WTP for packaging recyclability. This survey also added a new element by including a time variable to measure consumersÕ willingness to expend effort in order to recycle. The time variable was transformed into a money -metric value in order to compare WTP for packa ging recyclability and willingness to recycle packaging. Our survey uses cheap talk and indirect questioning in order to mitigate the hypothetical and social desirability bias. Two information treatments were used to add to the literature on the effects o f information on consumer WTP for recyclable packaging. !"#! 4. Survey Method 4.1 Survey Overview An online survey was used to understand consumer preferences for packaging recyclability as well as their current recycling habits .2 There were 54 questions in the survey that determined consumer preference for packaging, grocery shopping habits, socio -economic demographics, and stances on environmental issues. The median time to completion was 19 minutes, with 90% of consumers completing the survey within 36 mi nutes. 2,000 surveys were completed between February 25 and March 3 , 2015. The survey was developed in collaboration with sustainabi lity and packaging executives from food companies in collaboration with Michigan State UniversityÕs Center for Packaging Inn ovation and Sustainability. The survey was conducted online through a marketing research and survey company; Decipher Inc. Invitations to participate in the study were sent out from a consumer database maintained by Survey Sampling International (SSI). The survey was pre -tested and improved before conducting the actual survey. The pre -test resulted in some wording refinements and elimination of unclear questions. Additionally, to improve data quality , surveys completed in less than 10 were rejected to imp rove quality. 4.1.1 Sampling plan A marketing firm, Decipher Inc., sent out invitations to the survey to a large opt -in panel maintained by SSI . SSI sources participants through non -probabilistic methods including random digit dia ling telephone recruitme nt, social media, and websites. SSI maintains a subpanel !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!$!The implementation of this study was approved by Michigan State UniversityÕs Internal Review Board. !!"#!of 400,000 members whose demographics roughly match that of the national census . SSI maintains this panel by eliminating duplicates and computer -generated responses (Survey Sampling International, 20 15). Several studies in the past that used choice experiments have chosen respondents from SSIÕs online panel (Loureiro and Umberger, 2007; Olynk Tonsor and Wolf, 2010; Tonsor and Shupp, 2010). A set of survey questions was specified in order to fill pre -set quotas to insure a varied group of responses were accepted and that they reflected the broader U.S. population . The targeted quotas included gender, age, income level, education level, and geographic region (within the United States). Results and demog raphics from this section are shown in Table 1 and discussed in Chapter 6 . Respon ses were only accepted as part of the final analysis if they stated that they made the hou sehold food purchasing decisions. Our sample is representative of the 2010 census in terms of most demographic characteristics ; however respondents reported slightly higher levels of education than the national average. The ability to administer surveys through an online format has allowed researchers to reach a wider and more represent ative pool of respondents with minimal cost and time. Additionally , online surveys facilitate the addition of visual components to surveys , enhancing the respondentÕs ability to conceptualize the choice given to them (Savage & Waldman, 2008). Online surve ys have been found to have higher response rates than mail surveys (Olsen, 2009). Moreover, Olsen (2009) found no significant difference in consumer willingness to pay from choice experiments administered online or through conventional mail. 4.2 Effects of Information This survey used a between subject design to study the effects of information on consumer behavior . Two different information treatments were used to determine whether !"#!additional information on the benefits of recycling increases consumer WTP and the amount of effort participants are willing to put into recycling. Since the percent of news and information that Americans receive from the television is increasing (Gallup, 2013), this study also compares video to written information presentation to assess the impact s of different information delivery strategies on consumer perceptions and behavior . This study used one i nfographic (Figure 3 ) that could be used on a packaging label and one short 30 -second video that could be used as the basis for a c ommercial. The video utilized was the winning entry in a contest run by the EPA to promote household recycling (Figure 4 and Table 1 ). The infographic promoted general waste reduction, while the video provided specific examples on how recycling reduced ene rgy consumption. It is important to keep in mind both the format in which the information is provided as well as the message of each information treatment. If neither information treatment is found to have a positive impact on a consumers recycling behavio r, companies should find new avenues and messages to increase the value of recycling for their consumers. Figure 3: Recycling Infographic Source: EPA, 2015 !"#! Figure 4: Recycling Video Snapshot Source: EPA, 2015 Table 1 : Recycling Video Script Source: EPA, 2015 4.3 Survey Content 3 The survey ha d 4 distinct treatment groups, with 25% of the respondents randomly assigned to each group. The first group was the control group. The second group was the indirect question ing group . This group was asked what the average American would do in the experiments, but no indirect questioning was used for the non -experiment questions. The third group was shown the infographic in Figure 3 before the experiments. Finally, the fourth group was shown the video in Figure 4 with the script in Figure 5. Each treatment group had the same first two sections before the information treatments. The first section of the survey focused on the respondentsÕ household food and grocery shopping habi ts. Respondents were asked where and how often they shopped for groceries. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!3 A copy of the complete survey can be found in Appendix B $%&'&()*+!,-%.!*-/!-*('!.01%!.20&%!)*!(0*,3)((.4!56/!0(.-!&-*.%71%.!%*%7+'!0*,!)/!).! .6727).)*+!8-9!%1%7',0'!)/%:.!&0*!7%0(('!0,,!62;!$%&'&()*+!-*%!.-62!&0*!.01%.! %*-6+8!%*%7+'!/-!2-9%7!0!(02/-2!3-7!/9-!8-67.;!<80/!)3!9%!7%&'&(%,!:-7%=!$%&'&()*+! -*%!>?@-6*&%!2(0./)&!5-//(%!&0*!.01%!%*-6+8!%*%7+'!/-!2-9%7!0*!8-67!-3!AB;!C/D.!-67! 2(0*%/4!-67!./6334!0*,!-67!&8-)&%;! !!!"#!Certain food preferences were revealed in this section by asking what types of food characteristics they looked for while shopping such as on -the -go, GMO -free, vegetarian, packagin g material, etc. The second section of the survey focused on the respondentsÕ packaging preferences , environmental attitude s, and recycling behaviors. Questions pertaining to incentives and barriers to recycling were asked alongside questions about their current habits. A series of matching questions were used to determine respondentsÕ recognition and understanding of common recycling/packaging labels. The third section of the survey was comp rised of two choice experiments: one on WTP for packaging re cyc lability, another on barriers to recycling behavior. Preceding both choice experiments w ere explicit directions and an explanation of each of the attribute s and options available to the respondent . Further explanation of the choice experiments will be presented in Chapter 6 . 4.4 Willingness to Pay Experiment For the choice experiment on willingness to pay for packaging recyclability, consumers were asked to base their decision on fruit juices and fruit juice drinks. We focused on this product because of the increased prevale nce of juice drink consumption. A ccording to a report recently published by Mintel (2014), 84% of American consumers reported purchasing fruit juice or juice drinks in the past six months. For those who bought juice in the past 6 mont hs, 55% bought single -serve juice drinks ; that number increases to 76% in the 18 -24 age range. Furthermore , 32% of respondents reported purchasing more juice drinks than they did a year ago (Mintel, 2014). !"#! For the first choice experiments, labeled choic e sets were created taking into consideration the specifications of discrete choice models. Labeled choice sets shows an attribute consistently throughout all choice sets. In this choice experiment, packaging material was held consistent throughout all cho ice sets. In this choice experiment, respondents chose between one of four products or alongside an opt -out/no choice alternative. ¥ The first attribute, material, had four levels: glass, aluminum, plastic and carton. Since this was a labeled choice experime nt these four materials were always presented as options in the choice sets, alongside an opt -out/no choice alternative. ¥ The second attribute was packaging recyclability: yes or no. Instructions were included on whether the specific material was recyclable in their hometown; the respondent was to base their decision on the option given. ¥ The third attribute was price. Prices were based on observed market prices. Price levels were: $0.75, $1.00, $2.00, and $3.00. Choice sets were created by using a simultan eous orthogonal design in Ngene (ChoiceMetrics, 2011) . With this design type , orthogonality holds within and across alternatives. The experimental design consisted of a total of 20 choice scenarios. The scenarios were blocked so that each respondent evalua ted 5 choice sets. A sample ch oice set is provided in Table 2 : Characteristics Option 1 Option 2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can Recyclable Yes Yes Yes No Price $0.75 $0.75 $1.00 $1.00 I would not purchase any of these products I would choose ! ! ! ! ! Table 2 : Example choice set for packaging recyclability choice experiment !"#!4.5 Willingness to Recycle Experiment For the choice experiment focused on willingness to recycle , an on -the -go sandwich container was us ed as the product example. On -the -go packaging was used since it is one of the fastest growing trends in food and beverage packaging (Mintel, 2014). A recent study showed that on any given day, 49% of Americans eat at least one sandwich (Sebastian, Enns, Goldman, Hoy, & Moshfegh, 2014). Since sandwich consumption is so prevalent in the U.S., it was the basis for the second choice experiment. This choice experiment uses an unlabeled design , with each option representing a hypothetical product to be recycled . For an unlabeled design, all attributes are varied in each choice set. Respondents were asked to cho ose between two products and an opt -out/ no choice option if they would recycle neither option . ¥ The first attribute was packaging material: paperboard or plastic. These materials were chosen because they are common packaging materials used for sandwich containers. ¥ The second attribute was whether the package needs to be cleaned in order to be recycled: yes or no. ¥ The third attribute was how many different parts the packaging must be separated into in order to be recycled: 1 to 4. For example, a sandwich box might be mainly paperboard, but also has a film wrapper, etc. that need to be taken apart before recycling the main packaging. ¥ The fourth attribute was the time required to recycle the packaging with time options: 5, 10, 20, 30 or 60 seconds. !""!Choice sets were chosen by using a sequential orthogonal design in Ngene (ChoiceMetrics, 2011). With this design, orthogonality holds only within alternatives, red ucing the number of choice sets in the design. The experimental design consisted of a total of 20 choice scenarios. The scenarios were blocked so that each respondent evaluated 5 choice sets. A sample choice set is provided in Table 3 . Characteristics Opt ion 1 Option 2 Option 3 Type of material Paper/boxboard Plastic Cleaning required for recycling No Yes Number of Packaging Parts 2 4 Time required to recycle 30 sec 5 sec I would not recycle either of these I would recycle ! ! ! Table 3 : Example choice set for w illingness to recycling choice experiment !"#!5. Research Method 5.1 Theoretical Framework The framework of this study follow s La ncasterÕs Theory of Demand and Consumer Utility from his 1966 paper ÒA New Approach to Consumer Theory Ó. Before this paper, economists studied utility derived from the entire good. Lancaster argued that consumers do not receive utility from consuming the aggregate good; rather consumers receive utility from its different attributes or characteristics. Lancaster Õs (1966) new consumer theory has three main assumptions : 1. The good itself does not give utility to the consumer; it possesses attributes , and these attributes provide the consumer with utility 2. In general, a good possesses many attributes , and these attributes can be shared by many goodsa 3. Goods in combination may possess attributes different from those pertaining to the goods separately. For example , in this study , consumers receive utility from all aspects of consuming fruit juice: the packaging, the price, t he flavor, the color, etc. Choice experiments work by varying the levels of different attributes and asking respondents to choose between them, contrasting with other CVM methods that directly ask WTP. Random u tility theory is used in this study, which ha s been widely applied to studies that value consumer preference for products with multiple attributes (Manski, 1977). Random utility theory assumes that individuals seek to maximize their expected utility, given the budget and potential choices . Also, the individualÕs utility is considered to be a random variable because the researcher has incomplete information (Manski, 1977). !"#!5.2 Empirical Model An individual will evaluate each alternative as represented by alternatives. An individual decision makerÕs rule is that they will compare U1, U2, É.U J and choose the alternative which proved them with maximum utility. Individual i will choose product j from J alternatives in situation t if and only if: (1) Specifically, utility Uijt is a combination of both a deterministic and stochastic term: (2) The utilities associated with each alternative are not directly obser vable in the choice experiment because they include an unobserved component. Therefore, the probability of selecting alternative j is: (3) If we ass ume that the coefficients vary over decision makers in the population with density f( !) then t he resulting probability of individual i choosing alternative j is: (4)! Pijt=eVijteVijtj"f(#)d$# Although there are several model specification that can be used with this theory to es timate utility , e.g., condit ional or multinomial, the Random Parameters Logit (RPL) model was chosen. RPL is more flexible than the other models since it accounts for heterogeneity among individuals (McFadden and Train, 2000) . Since RPL allows for random t aste variation within the set of individuals, RPL is free of the independence of irrelevant alternatives assumption and allows correlation in unobserved factors over time (Train, 2003). In this model , indirect utility is assumed to be linear and can be wr itten as ! Uj:j=1...J! Uijt=Vijt+"ijt! P(j)=P(vij+"j#vk+"k),j$k%j&J!"#!(5) where Xijt is a vector of attributes for the jth alternative. !i is a vector of individual -specific taste parameters and "ijt is a stochastic component of utility that is independently and identically distributed across individuals and alter native choices. This stochastic comp onent of utility captures unobserved variations in tastes and errors in consumer perceptions and optimization. !ijt has a type -I extreme value distribution , also known as the Gumbel distribution . In order to operational ize the RPL model, the researcher must specify the distribution of the random parameters , this allows the researcher to estimate the distribution of preference parameters for each individual . The traditional approach has been to specify the distribution of non-price coefficients as normal , holding the price coefficient constant. Specifying the distribution of the price parameter as normal would be problematic since a normal distribution would allow posit ive values for the price parameter. This is illogical since economic theory predicts that individuals obtain negative utility from an increase in price. Also, a normally distributed price parameter could result in deriving distributions of WTP measures with infinite variances. Because of these reasons most re searchers assume a fixed price coefficient . In line with the literature, distribution of the random parameters is normally distributed for all variables, except for price, which is fixed. Specifying utility as separable in price, p, and as a vector of non -price attributes Xijt, we can rewrite equation 2 as: (6) ! Uijt="#pijt+$'Xijt+%ijt Following model estimation, WTP for each attribute X is derived shown in equation 7 .4 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!$!To eliminate the po tential confounding effects between the intercept and certain attribute levels, the data was effect coded, which means the coefficient is just in co mparison to the opt -out option. With effects coding, the estimated WTP have to be multiplied by 2 to generat e the appropriate marginal WTP measures. !! Vijt="'iXijt+#ijt!"#!(7) ! WTP X="#$ Assuming a fixed price coefficient is analogous to assuming that preferences over prices are homogene ous in the population, and implies that the standard deviation of unobserved utility or the scale parameters is the same for all observations. Louviere (2003) convincingly argues that the scale parameter can, and indeed often does, vary randomly over obser vations, and ignoring this variation can result in erroneous conclusions. In the context of product choice modeling, if the price coefficient is constrained to be fixed, when in fact scale varies over observations, then the variation in scale will be incor rectly attributed to variation in WTP for product characteristics. A solution is this problem is to parameterize the model such that the parameters represent the marginal WTP for each attribute rather than the utility coefficient of each attribute. The app eal of this approach is that it allows the researcher to specify and estimate the distributions of WTP directly, rather than deriving them indirectly from distributions of coefficients in the utility function as in e quation 5 (Scarpa, Thiene, and Train, 20 08). Following Train and Weeks (2005), we divide utility in Equation 6, by a scale parameter kijt to reparam etrize the model resulting in equation 8: (8) Where new coefficient ! is defined as !ijt = ("ijt / kijt) and #ijt = ($ijt /"ijt ) is a vector of WTP for the product attributes that is independent of scale (Train and Weeks, 2005) . The attributes used in the estimation of the model parameters for each choice experiment are given in Tables 4 and 5. ! Vijt="#ijtpijt+(#i$i)'Xijt!"#!Table 4: Packaging material s choice experiment p arameters Material Glass, Plastic, Aluminum, and Carton Packaging Recyclability Yes, indicates the packaging is recyclable, whether or not that material is actually recyclable in your community Price Price for the fruit juice drink Table 5: Barriers to recycling choice experiment parameters Type of Material Paperboard in comparison to plastic. Cleaning Whether the packaging was required to be cleaned in order to be recyclable Values of Time 5 Time required to recycle the packaging transformed into opportunity cost of time Number of Packaging Parts How many parts the packaging must be separated into in order to be recycled: 1 to 4. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!$!Consumer opportunity cost of time was calculated by using their self -reported household annual income. The wage imputation was calculated following previous literature where annual income is divided by 2000 hours of work time (50 weeks at 40 hours per week). This survey did not ask respondents their exact income, rather it provided 8 range options that could be selected. The choices were less than $20K, $20K -40K, $40K -$60K, $60K -$80K, $80K -$100K, $100K -$150K, $150 K-$200K, and more than 200K. To convert the ranges into a continuous variable, we used the midpoint range for the first 7 choices. Respondents reporting an income over 200K were assigned an income of $400K. The final computation of the opportunity cost of a second was taken by taking the hourly wage rate and dividing it by 3600 (60 minutes in an hour, 60 seconds in a minute). !!"#!6. Data Analysis This chapter presents and interprets the results of the econometric estimation of consumer incentives for recycling. First, background data is presented. The choice experiments are analyzed separately, then the implications of the choice experiments are reviewed. Table 6 : Descriptive statistics of the observed sample population in percentage Survey 2010 Census Age 18 to 24 years 11 13 25 to 44 years 37 35 45 to 64 years 36 35 65 to 13 years 16 17 Gender Male 48 49 Female 52 51 Education Did not graduate from high school 4 12 Graduated from high school 30 31 Attend ed College, no degree earned 28 26 Attended college, degree earned 34 19 Graduate/Advanced Degree 4 11 Household Income <$20,000 20 20 $20,000 -$59,999 41 40 $60,000 -$99,999 25 20 $100,000 -$200,000 12 17 >$200,000 2 3 Region of the U.S. South 35 37 West 24 23 Northeast 19 18 Midwest 22 22 Urban Rural Continuum 1) Counties of 1 million or more 54 55 2) 250,000 to 1 million population 25 21 3) metro area < 250,000 8 9 4) population > 20,000 adjacent to metro 4 4 5) population > 20,000 not adjacent to metro 2 2 6) 2,500 < population <19,999 adjacent to metro 4 5 7) 2,500 < population <19,999 not adjacent to metro 2 3 8) population< 2,500 adjacent to metro 1 1 9) population< 2,500 not adjacent to metro 1 1 Source: Survey data . Notes: Age percent is calculated for population above 18, Education percentage for 2010 census is calculated for population older than 25 !"#! A comparison from the observed demographics from our survey with the equivalent 2010 census d ata shows our survey resembles the demographic makeup of the U.S. population. In our survey we have slightly oversampled respondents in the middle of the distribution of education . Statistics for select recycling habits, recycling rates and incentives to r ecycle can b e found in Appendix A . 6.1 Willingness to Pay for Recyclability Choice Experiment The choice experiment estimating WTP for packaging recyclability was first estimated using the ra ndom parameters logit model in preference -space (Appendi x C). All of the estimated coefficients in every treatment were statistically significant at the 10% level. As expected , the coefficients were negative for price and positive for recyclability and material, suggesting respondents received positive utility from re cyclability and packaging materials. Since the estimated coefficients are m easured in utils (units of utility) , a non -cardinal measure , it is more convenient to transform them and analyze them as WTP estimates (Table 7 ). Estimated coeffici ents for every tr eatment in WTP -space can be found in Appendix C . Confidence intervals for the random parameters logit model estimated in preference space were derived using the Krinsky -Robb method .6. Because WTP measures in preference space are non -linear functions of es timated parameters, common methods for determining a confidence interval are not appropriate since they result in symmetric confidence intervals. Confidence intervals were estimated with 1000 draws with the Krinsky -Robb method. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!6 The Krinsky -Robb approach is used to simulate an asymptotic distribution of the WTP by randomly drawing from a multivariate norma l distribution, constructed by the combination of the coefficient estimated and the associated variance -covariance matrix from the RPL model. !!"#! From the results in Table 7, we can see positive average WTP for packaging recyclability for a 12 -ounce juice drink for all treatment groups. In comparison to the other treatment groups, the indirect questioning treatment group reported significantly lower WTP for packaging recycla bility at the 1% level. If the sample were a true representative sample of the U.S. consumer and there were no social desirability bias, we would expect to see the mean WTP for the control group to match that of the indirect questioning group. Since the m ean of the indirect questioning group is statistically significantly lower than that of the control group it means one of two possibilities or a combination of both: the sample is not representative of the U.S. population and we are over -sampling consumers with higher WTP for recycling, and/or respondents in this survey are biased and report higher WTP than for the average American . Likely, our sampleÕs true mean WTP for recyclable packaging lies in between that of the indirect questioning and the control groupÕs mean. A t -test was performed and revealed significant increase in WTP for the video treatment group at the 5% level. Although the variance for each average WTP is large, mean WTP for the video treatment is 31 cents more than that of the control gr oup. Since this is an increase of 24% from the control group estimation, this result represents a considerable increase. The results of the t-test showed that in comparison to the control group, the information treatment did not have a statistically signif icant impact on WTP for packaging recyclability. The difference between these two information treatments indicates that the channel in which media is presented does have an impact on consumer WTP for packaging recyclability. !"#! Table 7 : WTP estimated der ived from model estimation in preference space and WTP -space Preference Space WTP -space Mean 95% CI Mean 95% CI Control Packaging Recyclability 1.46 [1.27, 1.65] 1.31 [1.14, 1.48] Plastic Packaging 2.19 [2.02, 2.37] 2.1 0 [1.96, 2.2 4] Glass Packaging 2.03 [1.85, 2.21] 2.1 0 [1.94, 2.25] Carton Packaging 1.79 [1.61, 1.96] 1.79 [1.64, 1.94] Aluminum Packaging 1.46 [1.28, 1.65] 1.52 [1.67, 1.37] Indirect Questioning Packaging Recyclability 0.54 [0.40, 0.67] 0.49 [0.40, 0.57] Plastic Packaging 2.8 [2.61, 2.99] 2.86 [2.64, 3.09] Glass Packaging 2.2 [2.01, 2.39] 2.29 [2.07, 2.51] Carton Packaging 2.12 [1.93, 2.31] 2.16 [1.93, 2.39] Alu minum Packaging 2.32 [2.13, 2.53] 2.39 [2.17, 2.60] Infographic Treatment Packaging Recyclability 1.51 [1.29, 1.74] 1.41 [1.26, 1.56] Plastic Packaging 2.05 [1.88, 2.23] 1.93 [1.83, 2.04] Glass Packaging 1.89 [1.72, 2.07] 1.74 [1.62, 1.86] Carton Packaging 1.72 [1.55, 1.90] 1.63 [1.52, 1.75] Aluminum Packaging 1.49 [1.31, 1.68] 1.45 [1.32, 1.57] Video Treatment Packaging Recyclability 1.67 [1.44, 1.90] 1.62 [1.45, 1.78] Plastic Packaging 2.13 [1.94, 2.33] 2.05 [1.92, 2.18] Glass Packaging 1.98 [1.79, 2.16] 2.03 [1.90, 2.17] Carton Packaging 1.71 [1.51, 1.92] 1.64 [1.49, 1.78] Aluminum Packaging 2.32 [2.13, 2.53] 1.59 [1.46, 1.73] *WTP estimates are reported in $ per single serve juice drink **Model fit stati stics can be found in Appendix C From Table 7 we can see that the mean for WTP es timated in the two models fall within each otherÕs 95% confidence interval . However, from the density graph in Figure 5 , we can see that estimates in WTP-space tend to be more normally distributed and report less extreme values than estimates in preference space . This is in line with past research (Scarpa, Thiene, & Train, 2008) that has found that estimating respondents values in WTP-space address es the Òfat tailÓ problem of reporting many extreme values, which takes place when estimating WTP in preference space. In recent years estimating willingness to pay in WTP-space has grown in !"#!popularity because it reduces extreme values. Therefore , estimates obtained from estimation in WTP-space will be used to analyze the data . Figure 5 : Density graph of estimat ed WTP for packaging recyclability in preference space and WTP-space Understanding determinants of WTP for packaging recyclability is important for packaging companies in order to promote recyclability in the U.S. The random parameters logit model allow s researchers to estimate i ndividual WTP conditional on average estimated WTP and the specified parameter distribution. Estimated individual WTP for packaging recyclability were estimated by taking the estimated mean WTP and parameter distribution . We regr ess consumer WTP for packaging recyclability on basic demograp hics including age, education, etc. with results shown in Table 8 .7 Beyond the typical demographic variables, the following recycling specific variables were included : ÒPrice sensitiveÓ denotes that the respondent stated that price !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!7 Individual WTP for packaging recyclability regressions by treatment group can be found in Appendix G !""!was an important barrier to recycling (on a scale from least important (1) to most important (5)). ÒTime sensitiveÓ denotes that the respondent stated time was an important barrier to recycling (on a scale from least i mportant (1) to most important (5)). For the Ò Energy reasonsÓ variable, respondents reported that their main reasons for recycling was because recycling materials requires less energy than creating new materials. ÒRecycling ConfusionÓ is a dummy variable that is equal to 1 if respondents did not know if there was a recycling program in their community. ÒEnvironmental Warm -glowÓ denotes respondents attitudes to the statement ÒI feel good when I take steps to help the environmentÓ from disagree completely (1 ) to agree completely (7). ÒBottle Return StatesÓ is a dummy variable for states with a current refundable deposit system on certain containers. Dummy variables for the indirect questioning, infographic, and video treatment were included in this regressio n. Distributions for these vari ables can be found in Appendix A . !"#!Table 8 : Determinants of estimated individual WTP for packaging recyclability : OLS robust regression Indirect Questioning -0.874*** (0.058) Infographic -0.004 (0.058) Video 0.263*** (0.058) Female 0.044 (0.042) Age -0.018** (0.008) Age Squared 1.68E -04** (0.000) Education 0.010 (0.012) White 0.027 (0.053) Income 0.008 (0.012) Urban Continuum 0.019 (0.013) Time Sensitive -0.059*** (0.021) Price Sensitive -0.013 (0.021) Environmental Warm -glow 0.098*** (0.018) Bottle Return States -0.028 (0.046) Water Reasons 0.073** (0.029) Energy Reasons -0.003 (0.029) Democratic Voters 0.025 (0.049) Republican Voters -0.095* (0.053) Constant 0.985*** (0. 203) Observations 2,001 R-squared 0.225 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.10 The results in Table 8 are consistent with past research suggesting socio -economic demographics are not highly correlated with utility derived from r ecycling (Rokka and Uusitalo, !"#!2008). Willingness to pay for packaging recyclability is correlated more with motives than socio -economic demographics. WTP is correlated with age with WTP for packaging minimizing at age 59. Respondents who reported wa rm-glow from participating in environmentally friendly activities were willing to pay more for recyclable packaging than those who did not report environmental warm -glow . From a list of motives for recycling, responden ts who recycle to reduce energy the water pollution consumed showed a higher WTP for recyclable packaging . Republican voters also reported a slightly significant decrease in WTP for packaging recyclability in comparison to independents. A fraction of respondents displayed a negative willingne ss to pay for packaging recyclability. A nonparametric test proposed by Poe et al. (2005) 8 was used to compare the two groups. The Poe test results in a p -value that estimates the likelihood that the null hypothesis is false. The null hypothesis in this ca se is that the demographics of the lowest WTP are equal to that of the highest WTP. Those in the low WTP group consist of respondents with estimated WTP of less than 10 cents, with all other responses in the high/average group. Table 9 : Comparing negative and positive WTP for packaging recyclability High est 85% WTP (n=1708) Lowest 15% WTP (n=293) P-value Mean SD Mean SD Female (%) 0.53 0.50 0.48 0.50 0.20 Age 45.77 16.48 44.68 15.79 0.29 Education (scale) 3.98 1.73 3.72 1.67 0.01*** Income (scale) 3.17 1.76 3.08 1.86 0.44 White (%) 0.77 0.42 0.76 0.43 0.98 Democrats (%) 0.35 0.48 0.31 0.46 0.16 Republicans (%) 0.24 0.43 0.29 0.46 0.06* Note: SD means standard deviation *** p<0.01, ** p<0.05, * p<0.10 From the results in T able 9 , it can be seen t hat the most significant correlation is respondents wit h the lowest estimated WTP were more likely to report lower levels of education. Additionally , respondents who self -reported as Republican were more likely to be in the lowest !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!8 This te st will be referred to as a Poe -test from now on. !"#!WTP group, at the 10% st atistically significant level. Table 10 : WTP for recyclability by material and treatment Control Indirect Questioning Infographic Treatment Video Mean 95% CI Mean 95% CI Mean 95% CI Mean 95% CI Plastic 1.54 [1.30, 1.80] 0.60 [0.46, 0.72] 1.24 [1.02 , 1.44] 1.64 [1.42, 1.86] Glass 1.12 [0.94, 1.32] 0.18 [0.02, 0.32] 0.80 [0.60, 0.98] 1.24 [1.08, 1.42] Carton 1.10 [0.88, 1.34] 0.22 [0.10, 0.34] 0.98 [0.76, 1.18] 1.06 [0.82, 1.32] Aluminum 1.34 [1.08, 1.56] 0.60 [0.44, 0.72] 1.10 [0.96, 1.30] 1.28 [1.06, 1.48] N 2500 2500 2500 2505 Log -likelihood -2992 -2940 -3132 -3153 Pseudo R 2 0.256 0.269 0.221 0.218 AIC 2.41 2.37 2.52 2.53 Note: Number of parameters esti mated is 18. Breaking down WTP for recyclability by material in T able 10 enables us to see that the respondents did report a slight difference for WTP by treatment group . WTP for recyclability for plastic was significantly higher than for glass and carton at the 10% for all treatment groups. This contrasts to consumer W TP for material where respondents reported a significant difference in WTP for material s.9 A possible explanation for this is that consumers are more concerned about plastic packaging ending up in a landfill rather than carton or glass. Consumers may believe that plastic is the most important materials to recycle . Tables for OLS regressions on WTP for packaging mat erial can be found in Appendix D . 6.2 Barriers to Recycling Choice Experiment The choice experiment measuring consumerÕs willingness to recycle was estimated using random parameters logit . Tables showing consumer utility in random parameters logit in preference space can be found in Appendix E . Since consumer utility was measured in seconds , not dollars, consumer opportunity cost of time was calcu lated. Estimation results for consumer opportunity cost to overcom e barriers are shown in Table 11. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!9 T-test results comparing WTP for packaging material within treatment group can be found in Appendix F !"#! Willingness to accept (WTA) is the minimum amount of money that a person is willing to accept to give up something or to put up with something negative, su ch as cleaning. The standard assumptions of economic theory imply that when income effects are small, the gap between WTP and WTA should be negligible. Since measurements in this choice experiment are very small, the negative amount of WTP should be equal to WTA. This choice experiment values burdens, where expected WTP should be negative, therefore results will be presented in a willingness to accept format. The interpretation of the WTA estimates for paper /boxboard is that consumers are willing to accept $X more to recycle paper/boxboard packaging relative to plastic packaging. The cleaning WTA estimates can be interpreted as consumers are willing to accept $X to clean a single unit of packaging in order to recycle it . WTA for parts can be interpreted as consumers are willing to accept $X per part in order to recycle a single unit of packaging. Table 11 : WTA to Recycle Estimation in WTP-space Control Indirect Questioning Infographic Video Mean [95% CI] Mean [95% CI] Mean [95% CI] Mean [95% CI] Paper/boxboard 0.00 [ -0.03, 0.03] 0.00 [ -0.01, 0.01] -0.02 [ -0.04, 0.01 ] -0.01 [ -0.01, 0.03 ] Cleaning 0.06 [0.02, 0.09 ]** 0.08 [0.05, 0.10 ]** 0.08 [0.05, 0.12 ]** 0.07 [0.04, 0.10 ]** Parts 0.01 [ -0.01, 0.02 ] 0.0 1 [0.00 , 0.02 ]** 0.00 [ -0.02, 0.01 ] 0.00 [-0.01, 0.02 ] N 2500 2500 2500 2505 Log -likelihood -1971 -1953 -1968 -1986 Pseudo R 2 0.28 0.29 0.28 0.278 AIC 3962 3926 3957 3993 Note: Number of parameters is 10. ** p<0.05 The WTA estimate s for the material parameter are not significant in any of the treatment groups (Table 11 ). This suggests that consumers do not have a preference for paper/boxboard packaging over plastic packaging when deciding to recycle . The WTA estimate for p arts is negative and significant for the indirect questioning group, but not for the other treatment groups. This can be interpreted as respondents do not report a WTA for the number of parts they have to separate, but think that the average American would have a positive WTA for each part they had !"#!to separat e. Res pondents either reported socially bias ed answers in the control and information treatment groups, or our sample is less likely to view an additional part to recycle as effort compared to the average American. Results presented in Ta ble 11 show that averag e WTA to clean in order to recycle is positive and significan t for all treatment groups (distributions from each treatment group can be seen in Figure 6 ). This is intuitive, since cleaning packa ging takes both time and effort, respondents should report significant WTA to clean. Mean WTA to clean is approximately 7 cents as measured in opportunity cost of time. Differences between estimated WTA for the different treatment groups are not statistically significant . Although respondents in the indirect question ing group did not have a statistically significant different WTA to clean, respondents in the indirect questioning group were more likely to choose to opt -out/ not recycle the packaging. Willingness to accept to clean will be used as the basis for effort si nce it is an interaction between consumer opportunity cost of time and effort needed to recycle. Figure 6 : Density graph of estimated WTA to clean in order to recycle in preference space and WTP-space !"#!OLS robust regressions were used to correlate indi vidual WTA to clean in order to recycle with certain demographics ( Table 12 ). Results show that the coefficients for the indirect questioning, infographic, and video treatment groups were all statistically significant and positive. Since the indirect ques tioning group is correlated with higher WTA , this could be interpreted as consumers in the control group reporting socially biased answers. Since we are seeing counter -intuitive signs for the two information treatments, it may be the case that the informat ion treatments reminded the consumers of the effort required to recycle. Both Democratic and Republican voters reported lower WTA values in comparison to independents. Also, consumers who recycle for energy reasons view cleaning as less of a barrier than those who recycle for other reasons. Consumers who lived in states with bottle -return laws did report a lower WTA to clean than those who live in states without bottle -return laws. Since these states normally report higher recycling rates on average, it m ight be more of a social norm for these consumers to put in the effort in order to recycle. !"#!Table 12 : Determinates of estimated individual WTA to clean in order to recycle: OLS robust regression Variables Regression Coefficients Indirect Questio ning 0.0152** (0.007) Infographic 0.024*** (0.007) Video 0.013* (0.007) Female 0.005 (0.005) Age 0.000 (0.000) Age Squared -0.001 (0.001) Education 0.002* (0.001) White 0.023*** (0.006) Income 0.000 (0.001) Urban Continuum -0.001 (0.001) Time Sensitive 0.002 (0.002) Price Sensitive -0.004 (0.002) Warm -glow 0.001 (0.002) Bottle Return States -0.012** (0.005) Water Reasons 0.003 (0.003) Energy Reasons -0.001*** (0.003) Democratic Voters -0.016*** (0.006) Repu blican Voters -0.014** (0.006) WTP Packaging Recyclability -0.007*** (.006) Constant 0.101*** (0.024) Observations 2,001 R-squared 0.037 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#!Although the WTP for packaging recycl ability and WTA to clean to recycle cannot be compared directly since they use two different products and focus on two different parts of recycling, WTP for packaging recyclability was still used as an independent variable in the WTP to clean regression. WTP for packaging recyclability is correlated with a lower WTA to clean in order to recycle. For every $1 increase in WTP for packaging recyclability, there is an expect ed .07 cent decrease in WTA to clean in ord er to recycle. Since average WTA to clean is 6 cents this represents a 12% decrease in WTA to clean for each additional dollar WTP to packaging recyclability. 10 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!10 These results were robust across several regression specifications. Regression results from alterna tive estimations can be found in the appendix. !"#! 7. Discussion & Conclusion In this study, we use d a nationally representative survey that includes two choice experiments to estimate U.S. consumer WTP for packaging recyclability and willingness to recycle . These findings can help the packaging industry to adjust their recyclable packaging and their messaging for their target consumer. 7.1 Summary of Key Results Average estimated WTP for pa ckaging recyclability is positive for all materials , however, it is the highest for plastic, followed by aluminum can, glass, and then carton. Since consumers are willing to pay the most for plastic packaging recyclability , consumers may view plastic as more detrimental if it is thrown away. Although, aluminum recyclability was only significantly higher at the 10% level, it should be noted that aluminum packaging was valued the least, showing a difference between how consumers value packaging materials and their recyclability. For WTP for packaging recyclability, Republican voters report lower WTP than non -Republicans. Consumers who reported recycling to decrease water pollution were likely to report higher WTP for packaging recyclability. Since recycling do es not directly contribute to decreasing water pollution, packaging companies may choose to either exploit this misconception or clarify to consumers how recycling benefits the environment. It is important to point out that consumers who live in bottle st ate laws did not report an increase in WTP for packaging recyclability. This may result from consumers being accustomed to prices for beverages in their states and not noticing the increase in price due to the deposit laws. Mean w illingness to accept to clean did not change by treatment group. However, when holding demographics constant in the individual WTA to clean regressions, the indirect treatment group did show a significant decrease in WTA . This suggest s that consumers in the !"#!control group did rep ort some type of bias. The decrease in WTA for the video and infographic groups could be a result of the information treatments being a sort of a reminder of the effort of recycling. Additionally , consumers with a higher WTP for packaging recyclability a lso report that cleaning is less of a barrier to recycl ing . White consumers report ed higher willingness to accept to clean packaging for themselves , but not for the average American in comparison to non -white consumers . Consumers who recycle for energy rea sons reported a lower willingness to accept to clean over consumers who recycle for other reasons. Since reducing total energy consumed is one of the main reasons for recycling, consumers who recycle for energy reasons maybe more informed on the topic, the refore willing to put in more effort in order to recycle. 7.2 Limitations As with any stated preference experiment there are limitations to this study . Since preferences are stated and not revealed, there exists the potential for hypothetical bias with t his survey. It can be assumed that the cheap talk before the experiments reduced some of the hypothetical bias, but did not eliminate all of the bias. There also exists an issue not addressed in this study called attribute non -attendance, which is when c onsumers ignore one or more attributes w hen deciding between choices , often when the number of attributes is large . Future research should be done to determine which packaging attributes consumers consider when purchasing a product , as decision making can a ffect choice model estimates. 7.3 Implications As the quantity of packing sold in the U.S. increases, waste from th e packaging has also increased. Since environmental awareness has risen in the U.S., consumers have increasingly pressured companies to produce more environmentally friendly products . Companies have !""!responded with marketing campaigns to enhance recycling. Our results show that consumers already place positive utility of packaging recyclability and that providing additional information only s howed an increase with the video treatment . Since the video showed information on the amount of energy saved in a more easily understandable format , companies should focus on particular reasons to recycle rather than promoting recycling in general . A larg e portion, 38.5%, of respondents reported confusion about which labels indicate that a packaging product is recyclable. Companies should focus their marketing efforts on clearing up confusion on label meanings and recognition , rather than encouraging gener al recycling habits . The study also show s that consumers are more sensitive to cleaning material than the number of parts or the material of the packaging. Identifying barriers to packaging recycling in the U.S. is an essential first step in order for the packaging industry to mitigate these barriers and enhance overall recycling rates. !"#! APPENDICES !"#! Appendix A : Statistics for Select Recycling Habits Figure A -1: Importance of energy when deciding to recycle Figure A -2: Importance of cost when deciding to recycle !"#! Figure A -3: Importance of time when deciding to recycle Figure A -4: Environmental Warm Glow !"#! Appendix B : Full Survey Informed Consent This is a research survey designed to obtain information from U.S. households on recycling behavior and preferences for food packaging. Your participation in this survey is entirely voluntary and your responses will be kept in strict confidence. Food Purchasing Habits 1) Do you make your household food purchasing decisions? ¥ Yes ¥ No [IF NO, THE N INDIVIDUAL WAS SCR EENED OUT] 2) On average how much does your household spend on groceries each week? ¥ Less than $25 ¥ $25-49 ¥ $50-99 ¥ $100-$124 ¥ $125-$149 ¥ $150-$199 ¥ $200-$249 ¥ $250-$299 ¥ $300 or more ¥ Do not know 3) On average how much does your household spend on f ood away from home each week? ¥ Less than $50 ¥ $50-99 ¥ $100-$124 ¥ $125-$149 ¥ $150-$199 ¥ $200-$249 ¥ $250-$299 ¥ $300 or more ¥ Do not know 4) Where do you buy the majority of your householdÕs groceries? ¥ Dollar stores ¥ Mega -stores (ex: super -Wal-Mart or super -target) ¥ Loca l grocery chain ¥ Regional grocery chain ¥ Farmers markets ¥ Specialty Stores ¥ Warehouse clubs ¥ Drugstores ¥ Convenience Stores !"#! 5) On average, how often do you go to a grocery store? ¥ Once a month ¥ Twice a month ¥ Once a week ¥ Two times a week ¥ Three times a week ¥ Four times a week ¥ More than four times a week ¥ Other 6) How often do you buy fast food? ¥ Every day ¥ Several times a week ¥ About once a week ¥ Once or twice a month ¥ A few times a year ¥ Never 7) Please rank your preferences for the following characteristics from most to least important: ¥ Local ¥ Organic ¥ Non -GMO ¥ Humane treatment of animals ¥ ÒGreenÓ packaging ¥ Sustainable ¥ Natural 8) Which of the following categories do you consider while grocery shopping? (choose all that apply) o Quick/no preparation o Inexpensive o Lactose -free o Gluten -free o Organic o Fair-trade o Vegan o Vegetarian o Local o On-the -go o Non -GMO Packaging, Environmental Attitudes and Recycling Behavior !"#!9) How important is packaging in your food purchasing decisions? [LIKERT SCALE FROM 1 (NOT IMPORTANT) -5 (EXTREMELY IMPORTANT)] 10) Please ra nk the following packaging descriptions from most environmentally friendly to least ¥ Packaging made from recycled materials that cannot be recycled ¥ Packaging that can be recycled and that is made from recycled materials ¥ Packaging that can be recycled and th at is made from raw materials ¥ Packaging made from plant -based materials that cannot be recycled ¥ Packaging made from plant -based materials that can also be recycled ¥ Packaging that is biodegradable ¥ Reusable packaging ¥ Reduced/minimalist packaging 11) Please ind icate your attitudes towards the following statements Disagree Completely Disagree Mostly Somewhat Disagree Neither Agree Nor Disagree Somewhat Agree Agree Mostly Agree Completely I am very confused about what is good and bad for the environment New technologies will surely come along to solve environmental problems before they get out of hand Some pollution is inevitable if we are going to continue to make improvements in our standard of living I just donÕt have the time to worry about how all of my actions affect the environment I feel good when I take steps to help the environment I would be embarrassed if people I know caught me not recycling Local governments should provide more incentives for people t o recycle !"#!12) Where have you learned the most about sustainable packaging and the effects of recycling? ¥ Word -of-mouth ¥ News media ¥ Food retailers ¥ TV programming other than news ¥ Advertisements ¥ Social media and blogs ¥ Other 13) How often do you recycle you r household goods? ¥ Never ¥ Rarely ¥ Sometimes ¥ About half of the time ¥ Most of the time ¥ Always 14) How often does your household recycle? ¥ Never ¥ Rarely ¥ Sometimes ¥ About half of the time ¥ Most of the time ¥ Always 15) How often would you say your neighbors recycle their h ousehold goods? ¥ Never ¥ Rarely ¥ Sometimes ¥ About half of the time ¥ Most of the time ¥ Always 16) How often would you say the average American recycles their household goods? ¥ Never ¥ Rarely ¥ Sometimes ¥ About half of the time ¥ Most of the time ¥ Always !"#!17) How often do yo u recycle products made from the following materials? Never Rarely Sometimes About half of the time Most of the time Always Glass Paper Plastic Aluminum/Tin 18) How have your recycling habits changed in comparison to one year ag o? ¥ Increased ¥ Decreased ¥ Stayed the same 19) When you buy fast food, where do you most commonly dispose of the packaging and food waste? ¥ Household ¥ Road -side trash bin ¥ In-store ¥ Work location ¥ Other 20) The following chart lists several packaging characteristic s that someone may consider when deciding to recycle a product. For each product, please indicate which factor most influences your decision to recycle: Plastic beverage bottle Glass pickle jar Cereal box Soda can Shampoo bottle To-go hot beverage cup (pa per) Bulkiness Effort Required to recycle Time needed to recycle the packaging Having to clean residual product from the package Presence of a recycling label Obtaining a rebate for recycling !"#!21) Recycling label s- To the best of your abilities match the following labels to their meaning: A B C D E Please match the symbols to what they represent: Anti -litter symbol ¥ Made o ut of Recyclable Tin ¥ What type of plastic the product is made out of ¥ Made out of recyclable plastic ¥ The manufacturer of this product contributes to recovery and recycling costs ¥ Product made from at least 30% plant -based material ¥ Made from sustainably sourc ed forests ¥ Universal recycling symbol ¥ Product is compostable ¥ Made from recycled materials ¥ Do not know 22) Does your community have a recycling program? Yes No Do not know 23) Does your neighborhood have pick -up or curbside recycling? Yes No Do not know 24) Wha t types of recycling programs do you participate in? ¥ Curbside recycling ¥ Community drop -off recycling ¥ Store take back recycling ¥ In store rebate recycling ¥ Do not recycle ¥ Other 25) What materials does your local recycling program accept? i. Glass ii. Aluminum iii. Plastic iv. Paper v. Cardboard vi. Carton vii. Do not know ¥ !"#!26) How important are the following reasons for why you recycle? Least Important Not Very Important Somewhat Important Very Important Most Important Reduce waste sent to landfills 1 2 3 4 5 Protect wildlife 1 2 3 4 5 Reduce water pollution 1 2 3 4 5 Use less energy than creating new material 1 2 3 4 5 Cheaper goods in the long run 1 2 3 4 5 General sustainability 1 2 3 4 5 Mandated in your area 1 2 3 4 5 Cheaper than trash pickup 1 2 3 4 5 27) How important are th e following reasons for why you do not recycle? Least Important Not Very Important Somewhat Important Very Important Most Important Recycling takes too much time 1 2 3 4 5 Recycling costs too much 1 2 3 4 5 Not remembering to recycle 1 2 3 4 5 Recy cling takes too much effort 1 2 3 4 5 No recycling bin available 1 2 3 4 5 Recycling does not make a difference for the environment 1 2 3 4 5 Recycling guidelines are too confusing 1 2 3 4 5 No recycling program in your area 1 2 3 4 5 No curbside pick ups 1 2 3 4 5 28) Where do you normally buy your single -serve bottled drinks? 1. In a grocery store 2. From a vending machine 3. At a fast -food restaurant 4. At a convenience store !""!5. At a gas -station 6. Other 29) How often do you drink some type of fruit juice? ¥ Several times a day ¥ Once a day ¥ Several times a week ¥ A few times a month ¥ Once a month ¥ A few times a year ¥ Never 30) Which of the following fruit juices do you prefer to buy? ¥ National/named brand ¥ Store/generic brand ¥ Local brand ¥ Generic 31) Which material do you prefer your sing le-serve juice packaging to be made out of? ¥ Plastic ¥ Glass ¥ Carton ¥ Aluminum can ¥ No preference 32) What type of fruit juice do you normally drink? ¥ Orange ¥ Apple ¥ Lemon ¥ Multi -fruit ¥ Grape ¥ Berry ¥ Kiwi & strawberry ¥ Coconut ¥ Mango ¥ Vegetable/fruit blend ¥ Other Informa tion Treatments Were Shown Here The next portion of this survey presents you with five hypothetical fruit juice purchasing scenarios that you could face in a retail store where you typically shop. The four products that will be presented in each scenari o possess the same characteristics (for example, color, brand, flavor, etc.) except for varying levels of the characteristics presented below such as product price. For each scenario, please select the bottle of fruit juice that you would purchase, or ÔI would not !"#!purchase any of these productsÕ, if you would not purchase any of the fruit juice products. For your information in interpreting alternative bottles of fruit juice, keep in mind the following characteristics: Product price refers to the cost per 12-ounce (single serve) unit of fruit juice: ¥ $0.75/12 -ounce unit ¥ $1.00/12 -ounce unit ¥ $2.00/12 -ounce unit ¥ $3.00/12 -ounce unit Packaging material refers to the material of the bottle that the fruit juice is served/packaged in: ¥ Glass ¥ Plastic ¥ Carton ¥ Aluminu m can Recyclable refers to whether the packaging product is recyclable: ¥ Yes means that the product packaging is recyclable ¥ No means that the product packaging is not recyclable The experience from previous similar surveys is that people often state a h igher willingness to pay than what one actually is willing to pay for the good. It is important that you make your selections like you would if you were actually facing these choices in your retail purchase decisions, noting that allocation of funds to th ese products means you will have less money available for other purchases. Please select on the ÒI chooseÓ line below the option that you would purchase from each of the following scenarios: EXAMPLE OF THE CHOICE EXPERIMENT 33) Characteristics Option 1 Option2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can I would not Recyclable Yes No No No purchase any of Price ($/12 -ounce unit) $3.00 $2.00 $3.00 $1.00 these products I would choose ! ! ! ! ! !"#! 34) Characteristics Option 1 Option2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can I would not Recyclable Yes Yes Yes Yes purchase any of Price ($/12 -ounce unit) $3.00 $1.00 $0.75 $0.75 these products I would choose ! ! ! ! ! 35) Characteristics Option 1 Option2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can I would not Recyclable No Yes Yes No purchase any of Price ($/12 -ounce unit) $1.00 $3.00 $2.00 $3.00 these products I would choose ! ! ! ! ! 36) Characteristics Option 1 Option2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can I would not Recyclable No No Yes Yes purchase any of Price ($/12 -ounce unit) $0.75 $3.00 $1.00 $3.00 these products I would choose ! ! ! ! ! 37) Characteristics Option 1 Option2 Option 3 Option 4 Option 5 Material Plastic Glass Carton Aluminum can I would not Recyclable No No Yes No purchase any of Price ($/12 -ounce unit) $0.75 $0.75 $0.75 0.75 these products I would choose ! ! ! ! ! !"#! The next portion of this surv ey presents you with five hypothetical scenarios where you will have to choose which product you would most likely recycle. The two products that will be presented in each scenario will contain the same food prior to disposal (for example, a Ôto goÕ sandw ich) and are assumed to be the same except for varying levels of the characteristics presented below. For each scenario, please select the packaging that you would most likely recycle, or ÔI would not recycle either of theseÕ, if you would not recycle any of the packaging options. For your information in interpreting the difference in product packages: Material refers to the packaging of the to go container: ¥ Plastic ¥ Paper/boxboard Cleaning refers to whether you must clean the packaging in order for it to be recyclable: ¥ Yes- the packaging must be rinsed or cleaned before it can be recycled ¥ No- the packaging does not have to be rinsed or cleaned for it to be recycled Time refers to how long it takes to clean and/or take apart the packaging for recycling: ¥ This is indicated in seconds. Please take the time to picture yourself preparing each of the packages in order to get a more accurate response. Number of parts of the packaging that go into different recycling/trash bins: ¥ 1: the packaging does not have t o be taken apart to be recycled ¥ 2 or more : the packaging requires disassembling in order to be recyclable. o Examples include: removing a cap from a bottle or removing a non -recyclable plastic film from a recyclable cardboard box or other recyclable material . The experience from previous similar surveys is that people often state a higher effort level to recycle than what one will actually put into recycling. It is important that you make your selections like you would if you were actually facing these choi ces as you would in your household, please note that your answers will be used to influence food packaging companies and will have real world effects. Please select on the ÒI would recycleÓ line below the option that you would recycle from each of the fol lowing scenarios: EXAMPLE OF THE CHOICE EXPERIMENT 38) Characteristics Option 1 Option 2 Option 3 Material Plastic Plastic Cleaning No Yes I would not recycle Parts 1 1 either of these Time 5 sec 60 sec I would recycle ! ! ! 39) !"#!Characteristics Option 1 Option 2 Option 3 Material Paper/boxboard Plastic Cleaning No No I would not recycle Parts 3 3 either of these Time 10 sec 20 sec I would recycle ! ! ! 40) Characteristics Option 1 Option 2 Option 3 Material Plastic Paper/boxboard Cleaning Yes Yes I would not recycle Parts 1 3 either of these Time 60 sec 5 sec I would recycle ! ! ! 41) Characteristics Option 1 Option 2 Option 3 Material Paper/boxboard Paper/boxboard Cleaning Yes No I would not recycle Parts 4 4 either of thes e Time 20 sec 60 sec I would recycle ! ! ! 42) Characteristics Option 1 Option 2 Option 3 Material Paper/boxboard Plastic Cleaning Yes No I would not recycle Parts 2 1 either of these Time 5 sec 5 sec I would recycle ! ! ! !"#! 43) How inf luential are the following reasons in increasing your recycling behavior: A major influence A minor influence No influence Do not know I see my friends and people I know recycle My friends and people I know encourage me to recycle Government o fficials encourage me to recycle A non -profit encourages me to recycle A celebrity I respect encourages me to recycle I see news media encouraging recycling I see public notices from my community encouraging recycling I hear people promoting the benefits of recycling I hear people talking about the dangers of not recycling There is a financial incentive to recycle There is a financial penalty to not recycle Demographic Background: 44) I am: Male Female 45) I am _ ______ years old [TYPE IN] 46) Your highest level of education completed is: ¥ Did not graduate from high school ¥ Graduated from high school, did not attend college ¥ Graduated from high school, currently attending college ¥ Attended College, No Degree earned ¥ Attended College, Associates or Trade Degree earned !"#!¥ Attended College, BachelorÕs (B.S. or B.A.) Degree earned ¥ Graduate or Advanced Degree (i.e. M.S., M.A., Ph.D., J.D., M.D., etc. ) ¥ Other 47) Your annual pre -tax household income is: ¥ Less than $20,000 ¥ $20,000- $39,999 ¥ $40,000- $59,999 ¥ $60,000- $79,999 ¥ $80,000- $99,999 ¥ $100,000- $149,999 ¥ $150,000- $200,000 ¥ More than $200,000 48) In which state and ZIP code is your main residence? ¥ State: ZIP code: (TYPE IN) 49) What best describes where you live? ¥ A mobile home ¥ A one -fami ly house detached from any other house (single family home) ¥ A one -family house attached to one or more houses (townhome) ¥ A building with 2 apartments/condos ¥ A building with 3 -5 apartments/condos ¥ A building with more than 6 apartments/condos ¥ Boat, RV, van e tc. ¥ Other 50) What is your main political reference or affiliation: ¥ Democrat ¥ Republican ¥ Independent ¥ Libertarian ¥ Green Party ¥ Tea Party ¥ Other 51) How many people are currently living in your household? Please include yourself in the count ¥ One ¥ Two ¥ Three ¥ Four ¥ Five ¥ Six ¥ Seven ¥ Eight !"#!¥ Nine or more 52) Which of the following sectors would your primary current employment fall under? ¥ Construction ¥ Manufacturing ¥ Utilities ¥ Retail ¥ Transportation ¥ Information and Technology ¥ Financial and business services ¥ Education ¥ Health C are ¥ Social Work ¥ Leisure and Hospitality ¥ Federal Government ¥ State or Local government ¥ Agriculture ¥ Homemaker ¥ Unemployed ¥ Student ¥ Military ¥ Retired ¥ Other 53) What is your race? ¥ White ¥ Hispanic or Latino ¥ Black or African American ¥ Native American ¥ Asian ¥ Pacific Isla nder ¥ Other 54) How do you commute to work? ¥ Car ¥ Walk ¥ Bike ¥ Public transportation ¥ Other Thank you for your time in completing this survey. Your input will strengthen our research and help us obtain more accurate conclusions. If you wish to add any comments pl ease feel free to do so here: !"#! Appendix C : Packaging Recyclability Choice Experiment Estimations Table A -1: Packaging recyclability choice experiment in preference s pace Control Indirect Questioning Infographic Video Mean 95% CI Mean 95% CI Mean 95% CI Mean 95% CI Packaging Recyclability 1.46 [1.27, 1.65] 0.54 [0.40, 0.67] 1.51 [1.29, 1.74] 1.67 [1.44, 1.90] Plastic Packaging 2.19 [2.02, 2.37] 2.8 [2.61, 2.99] 2.05 [1.88, 2.23] 2.13 [1.94, 2.33] Glass Packaging 2.03 [1.85, 2.21] 2.2 [2.01, 2.39] 1.89 [1.72, 2.07] 1.98 [1.79, 2.16] Carton Packaging 1.79 [1.61, 1.96] 2.12 [1.93, 2.31] 1.72 [1.55, 1.90] 1.71 [1.51, 1.92] Aluminum Can Packaging 1.46 [1.28, 1.65] 2.32 [2.13, 2.53] 1.49 [1.31, 1.68] 2.32 [2.13, 2.53] N 2500 2500 2500 2505 No. of parameters 6 6 6 6 Log -likelihood -3246 -3235 -3389 -3409 Pseudo R -Squared 0.16 0.14 0.14 0.13 AIC 2.6 2.59 2.7 2.72 Table A -2: Packaging recyclability choice experiment in WTP-Space Control Indirect Questioning Infographic Video Mean 95% CI Mean 95% CI Mean 95% CI Mean 95% CI 1.31 [1.14, 1.48] 0.49 [0.40, 0.57] 1.41 [1.26, 1.56] 1.62 [1.45, 1.78] 2.1 [1.96, 2.24] 2.86 [2.64, 3.09] 1.93 [1.83, 2.04] 2.05 [1.92, 2.18] 2.1 [1.94, 2.25] 2.29 [2.07, 2.51] 1.74 [1.62, 1.86] 2.03 [1.90, 2.17] 1.79 [1.64, 1.94] 2.16 [1.93, 2.39] 1.63 [1.52, 1.75] 1.64 [1.49, 1.78] 1.52 [1.67, 1.37] 2.39 [2.17, 2.60] 1.45 [1.32, 1.57] 1.59 [1.46, 1.73] N 2500 2500 2500 2505 No. of parameters 18 18 18 18 Log -likelihood -2992 -2940.1 -3132.4 -3153.2 Pseudo R -Squared 0.256 0.269 0.221 0.218 AIC 2.41 2.37 2.52 2.53 !"#! Appendix D : Individual WTP for Packaging Material by treatment group - OLS robust regressions Table A -3: Individual WTP for plastic packaging by treatment group Plastic Packaging Control Indirect Questions Infographic Video Female 0.041 0.055* 0.112*** 0.016 (0.048) (0.029) (0.040) (0.040) Age 0.008 0.000 0.006 -0.012* (0.009) (0.006) (0.007) (0.007) Age Squared 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) Education -0.008 -0.006 -0.029** 0.006 (0.015) (0.009) (0.012) (0.012) Income 0.003 -0.005 0.035*** -0.017 (0.015) (0.010) (0.013) (0.013) White -0.145** -0.006 0.074 0.010 (0.062) (0.039) (0.050) (0.052) Democrats 0.046 -0.039 0.085* 0.045 (0.054) (0.037) (0.046) (0.048) Republicans -0.033 0.007 0.114** 0.087* (0.064) (0.037) (0.053) (0.051) 0.005 0.003 0.010 -0.002 Urban Continuum (0.016) (0.009) (0.013) (0.011) Constant 2.146** * 2.923*** 1.659*** 2.385*** (0.191) (0.140) (0.163) (0.181) Observations 500 500 500 501 R-squared 0.04 0.022 0.055 0.026 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#!Table A -4: Individual WTP for glass packagi ng by treatment group Glass Packaging Control Indirect Questions Infographic Treatment Video Treatment Female 0.068 0.030 0.052 0.049 (0.057) (0.030) (0.044) (0.057) Age -0.005 0.009 0.009 0.012 -0.010 -0.006 -0.008 -0.009 Age Squared 0.000 1.05E -04* 0.000 1.70E04* (0.000) (0.000) (0.000) (0.000) Education 0.055*** 0.005 0.005 -0.001 (0.017) (0.009) (0.013) (0.016) Income -0.008 0.011 0.007 0.030* (0.016) (0.008) (0.014) (0.016) White 0.051 -0.004 -0.085 0.106* (0.068) (0.036) (0.0 61) (0.063) Democrats 0.042 0.032 0.065 -0.044 (0.067) (0.035) (0.053) (0.065) Republicans 0.126* 0.051 0.054 -0.118* (0.075) (0.037) (0.056) (0.070) 0.002 0.002 0.012 -0.021 Urban Continuum (0.017) (0.010) (0.015) (0.016) Constant 1.877*** 2.033 *** 1.573*** 1.725*** (0.205) (0.124) (0.185) (0.210) Observations 500 500 500 501 R-squared 0.029 0.018 0.027 0.034 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !""!Table A -5: Individual WTP for aluminum packaging by treatme nt group Aluminum Packaging Control Indirect Questions Infographic Treatment Video Treatment Female -0.110*** -0.041 -0.036 -0.122*** (0.035) (0.028) (0.042) (0.042) Age 0.001 -0.009* -0.007 0.001 -0.006 -0.005 -0.007 -0.008 Age Squared 0.00 0 1.01E -04* 0.000 0.000 (0.000) (0.000) (0.000) (0.000) Education -0.016 0.000 -0.002 -0.029** (0.011) (0.008) (0.013) (0.013) Income 0.013 -0.0184** 0.000 -0.011 (0.010) (0.008) (0.013) (0.013) White 0.020 -0.012 -0.061 -0.043 (0.042) (0.039) (0.053) (0.052) Democrats 0.119*** 0.002 0.002 0.037 (0.042) (0.034) (0.047) (0.046) Republicans 0.067 -0.004 0.043 0.108* (0.044) (0.036) (0.056) (0.057) -0.002 0.001 -0.017 0.002 Urban Continuum (0.010) (0.009) (0.012) (0.014) Constant 1.605*** 2.653*** 1.755*** 1.720*** (0.130) (0.134) (0.170) (0.163) Observations 500 500 500 501 R-squared 0.053 0.023 0.019 0.043 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#!Table A -6: Individual WTP for carton p ackaging by treatment group Carton Packaging Control Indirect Questions Infographic Treatment Video Treatment Female 0.005 -0.064* -0.055 0.053 (0.048) (0.034) (0.040) (0.048) Age -0.007 0.007 -0.004 0.006 -0.009 -0.006 -0.007 -0.008 Age Squ ared 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) Education -0.003 -0.002 -0.003 0.013 (0.014) (0.011) (0.011) (0.013) Income 0.000 0.018* 0.000 0.014 (0.014) (0.011) (0.013) (0.013) White -0.052 0.025 0.066 0.014 (0.059) (0.039) (0.0 48) (0.060) Democrats -0.098* -0.006 0.011 0.096 (0.056) (0.039) (0.046) (0.059) Republicans -0.063 -0.058 0.008 -0.056 (0.064) (0.042) (0.052) (0.056) -0.018 -0.009 -0.007 0.000 Urban Continuum (0.014) (0.011) (0.012) (0.015) Constant 2.112*** 2.024*** 1.698*** 1.462*** (0.200) (0.135) (0.166) (0.197) Observations 500 500 500 501 R-squared 0.018 0.026 0.01 0.031 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#!Appendix E: Barriers to Recycling Choice Experiment Suppl emental Tables Table A -7: Estimated Parameters of the barriers to recycling CE in preference space Control Indirect Questioning Infographic Video Time Value -3.10*** -4.55*** -3.33*** -3.73*** Material -0.01 0.00 0.02 0.01 Cleaning -0.13*** -0.25*** -0.16*** -0.16*** Parts -0.04 -.06** 0.01 -0.04 No. of parameters 4 4 4 4 Log -likelihood -2324 -2432 -2284 2249 AIC 1.81 1.836 1.77 1.74 Table A -8: Determinants of Individual WTA to clean by treatment group Control Indirect Questioning Infographi c Video Female -0.014 -0.006 0.000 -0.003 (0.011) (0.008) (0.012) (0.010) Age 0.002 0.002 -0.001 0.001 (0.00) (0.00) (0.00) (0.00) Age Squared 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) Education -0.002 -0.001 -0.003 -0.001 (0.003) (0.002) (0.004) (0.003) Income 0.000 -0.001 0.002 0.001 (0.003) (0.002) (0.004) (0.003) White -0.020 -0.009 -0.042*** -0.020 (0.014) (0.010) (0.016) (0.012) Democrats 0.016 0.013 0.013 0.023** (0.013) (0.009) (0.015) (0.012) Republicans -0.005 0.014 0.023 0.029** (0.015) (0.010) (0.016) (0.013) Urban Continuum 0.000 0.000 0.001 0.004 (0.003) (0.003) (0.004) (0.003) Price Sensitive 0.000 0.001 -0.002 -0.005 (0.006) (0.004) (0.006) (0.005) Time Sensitive 0.001 0.004 0.001 0.011** (0.006) (0.004) (0.006) (0.005) Recycle for Energy Reasons 0.019*** 0.009** 0.007 0.003 (0.006) (0.004) (0.007) (0.006) Recycling Confusion -0.001 -0.011 0.020 -0.012 (0.022) (0.014) (0.024) (0.016) Environmental Warm -glow -0.008 -0.007** 0.005 0.004 (0. 005) (0.003) (0.005) (0.004) Constant -0.123** -0.088** -0.117* -0.138*** (0.054) (0.039) (0.064) (0.049) Observations 467 461 460 458 R-squared 0.047 0.052 0.037 0.044 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#! Appendix F: Differences in WTP for packaging materials Table A -9: T-test mean differences for packaging materials within treatment groups Plastic Packaging Glass Packaging Carton Packaging Control Glass Packaging $0.00 (0.994) Carton Packaging $0.31*** $0.31*** (0.00) (0.01) Aluminum Can Packaging $0.58*** $0.58*** $0.27*** (0.00) (0.00) (0.01) Indirect Questioning Glass Packaging $0.57*** (0.00) Carton Packaging $0.70*** $0.13 (0.00) (0.42) Aluminum Can Packaging $0.48*** $0.09 $0.22 (0.00) (0.56) (0.16) Infographic Glass Packaging $0.19** (0.02) Carton Packaging $0.31*** $0.11 (0.00) (0.19) Aluminum Can Packaging $0.49*** $0.30*** $0.19** (0.00) (0.00) (0.03) Video Glass Packaging $0.014 (0.88) Carton Packaging $0.41*** $0.40*** (0.00) (0.00) Aluminum Can Packaging $0.45*** $0.44*** $0.04 (0.00) (0.00) (0.69) *Note: P -values in parentheses *** p<0.01, ** p<0.05, * p<0.1 Mean WTP for packaging materials were tested within treatment groups. Consistently throughout all treatment groups, WTP for plastic packaging was valued the most with significant differences from carton and aluminum packaging. !"#! Appendix G: Individual WTP for packaging recyclability !Tabl e A -10: Individual WTP for packaging r ecyclability by treatment group WTP for packaging Recyclability Control Indirect Questioning Infographic Video Female 0.061 0.091** 0.069 -0.030 (0.094) (0.037) (0.101) (0.101) Age -0.009 0.006 -0.041** -0.027 (0.02) (0.01) (0.02) (0.02) Age Squared 0.000 0.000 3.8E -04** 0.000 (0.000) (0.000) (0.000) (0.000) Education 0.022 -0.010 0.005 -0.006 (0.030) (0.010) (0.030) (0.032) Income 0.032 0.017 0.019 -0.035 (0.030) (0.011) (0.030) (0.031) White -0.003 -0.090* 0.149 0.073 (0.120) (0.046) (0.134) (0.128) Democrats 0.006 -0.032 0.127 -0.031 (0.108) (0.044) (0.119) (0.121) Republicans -0.183 0.023 -0.159 -0.110 (0.126) (0.044) (0.121) (0.135) Urban Continuum 0.020 0.003 0.006 0.018 (0.026) (0.011) (0.032) (0.031) Price Sensitive -0.026 0.025 0.012 -0.109** (0.046) (0.016) (0.048) (0.048) Time Sensitive -0.100** -0.026 -0.092* 0.020 (0.046) (0.018) (0.050) (0.049) Recycle for Energy Reasons 0.0856* 0.021 -0.011 0.128** (0.050) (0.019) (0.052 ) (0.063) Recycling Confusion -0.400* 0.050 -0.305* -0.051 (0.213) (0.061) (0.185) (0.155) Environmental Warm -glow 0.140*** 0.015 0.185*** -0.002 (0.043) (0.015) (0.044) (0.046) Constant 0.432 0.161 1.311** 2.165*** (0.466) (0.199) (0.570) (0.473) Observations 467 461 460 458 R-squared 0.109 0.045 0.094 0.049 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 !"#! REFERENCES !"#!REFERENCES Aadland, D., & Caplan, A. J. (2003). Willingness to pay for curbside rec ycling with detection and mitigation of hypothetical bias. American Journal of Agricultural Economics , 85(2), 492Ð502. Aadland, D., & Caplan, A. J. (2006). Cheap talk reconsidered: New evidence from CVM. Journal of Economic Behavior & Organization , 60(4), 562Ð578. Adamowicz, W., Boxall, P., Williams, M., & Louviere, J. (1998). Stated preference approaches for measuring passive use values: choice experiments and contingent valuation. American Journal of Agricultural Economics ,80(1), 64Ð75. Arrow, Kenneth J., Robert S. Solow, Edward Learner, Paul Portney, Ray Rodner and Howard Schuman (1993). Report of the NOAA -Panel on Contingent Valuation. Federal Register 58(10): 4601 -4614. Auger, P., Devinney, T. M., & Louviere, J. J. (2007). Using best Ðworst scaling me thodology to investigate consumer ethical beliefs across countries. Journal of Business Ethics , 70(3), 299Ð326. Breidert, C., Hahsler, M., & Reutterer, T. (2006). A review of methods for measuring willingness -to-pay. Innovative Marketing ,2(4), 8Ð32. Cars on, R. T. (2000). Contingent valuation: a userÕs guide. Environmental Science & Technology ,34(8), 1413Ð1418. Carson, R. T. (2012). Contingent valuation: A practical alternative when prices arenÕt available. The Journal of Economic Perspectives , 26(4), 27Ð42. Carson, R. T., Groves, T., & List, J. A. (2014). Consequentiality: A theoretical and experimental exploration of a single binary choice. Journal of the Association of Environmental and Resource Economists , 1(1), 171Ð207. Cummings, R. G., & Taylor, L. O. (1999). Unbiased value estimates for environmental goods: a cheap talk design for the contingent valuation method. American Economic Review , 649Ð665. Retrieved from http://www.jstor.org/stable/117038 Don ohue, William. (2013). Creating a Communication Plan to Promote Consumer Awareness of the Value of Packaging: Final Report. Euromonitor International. (2013). Packaging Industry in the US . Retrieved from http://www.euromonitor.com/packaging -industry -in-the -us/report Ferreira, S., & Marques, R. C. (2015). Contingent valuation method applied to waste management. Resources, Conservation and Recycling , 99, 111Ð117. !"#!Fisher, R. J. (199 3). Social desirability bias and the validity of indirect questioning. Journal of Consumer Research , 303Ð315. Gallup. (2013) TV Is AmericansÕ Main Source of News . Retrieved from http://www.gallup.com/poll/163412/americans -main-source -news.aspx Gamba, R. J ., & Oskamp, S. (1994). Factors influencing community residentsÕ participation in commingled curbside recycling programs. Environment and Behavior , 26(5), 587Ð612. Guagnano, G. A., Stern, P. C., & Dietz, T. (1995). Influences on attitude -behavior relation ships a natural experiment with curbside recycling. Environment and Behavior ,27(5), 699Ð718. Halvorsen, B. (2008). Effects of norms and opportunity cost of time on household recycling. Land Economics ,84(3), 501Ð516. Hensher, D. A. (2010). Hypothetical bi as, choice experiments and willingness to pay. Transportation Research Part B: Methodological , 44(6), 735Ð752. Karousakis, K., & Birol, E. (2008). Investigating household preferences for kerbside recycling services in London: A choice experiment approach. Journal of Environmental Management , 88(4), 1099Ð1108. Kling, C. L., Phaneuf, D. J., & Zhao, J. (2012). From Exxon to BP: Has some number become better than no number? The Journal of Economic Perspectives , 3Ð26. Lancaster, K. J. (1966). A new approach t o consumer theory. The Journal of Political Economy , 132Ð157. Loomis, J. B. (2014 ). Strategies for overcoming hypothetical bias in stated preference surveys. Journal of Agricultural and Resource Economics , 39(1), 34Ð46. Loureiro, M. L., & Umberger, W. J. (2007). A choice experiment model for beef: What US consumer responses tell us about relative preferences for food safety, country -of-origin labeling and traceability. Food Policy , 32(4), 496Ð514. Louviere, J. J. (2004). Random utility theory -based state d preference elicitation methods: applications in health economics with special reference to combining sources of preference data. keynote address, Canberra: Australian Health Economics Society Conference. Lusk, J. L. (2003). Effects of cheap talk on consu mer willingness -to-pay for golden rice. American Journal of Agricultural Economics , 85(4), 840Ð856. Manski, C. F. (1977). The structure of random utility models. Theory and Decision , 8(3), 229Ð254. Mintel. (n.d.). Food Packaging Trends - US - July 2014 . Retrieved from http://reports.mintel.com/display/680530/# !"#!Nederhof, A. J. (1985). Methods of coping with social desirability bias: A review. European Journal of Social Psychology , 15(3), 263Ð280. Nor wood, F. B., & Lusk, J. L. (2011). Social desirability bias in real, hypothetical, and inferred valuation experiments. American Journal of Agricultural Economics , aaq142. Olsen, S. B. (2009). Choosing between internet and mail survey modes for choice expe riment surveys considering non -market goods. Environmental and Resource Economics , 44(4), 591Ð610. Olynk, N. J., Ton sor, G. T., & Wolf, C. A. (2010 ). Consumer willingness to pay for livestock credence attribute claim verification. Journal of Agricultural and Resource Economics , 261Ð280. http://www.jstor.org/stable/41960517 Perrin, D., & Barton, J. (2001). Issues associated with transforming household attitudes and opinions into materials recovery: a review of two kerbside recycling schemes. Resources, Conservation and Recycling , 33(1), 61Ð74. http://doi.org/ 10.1016/S0921 -3449(01)00075-1 Rokka, J., & Uusitalo, L. (2008). Preference for green packa ging in consumer product choices ÐDo consumers care? International Journal of Consumer Studies , 32(5), 516Ð525. Savage, S. J., & Waldman, D. M. (2008). Learning and fatigue during choice experiments: a comparison of online and mail survey modes. Journal of Applied Econometrics , 23(3), 351Ð371. Scarpa, R., Thiene, M., & Train, K. (2008). Utility in willingness to pay space: a tool to address confounding random scale effects in destination choice to the Alps. American Journal of Agricultural Economics ,90(4), 994Ð1010. Sebastian, R., Goldman, J., Enns, C. W., Martin, C., Steinfeldt, L., Dwyer, J., É Milner, J. (2014). Food sources of flavonoids for adults in the United States: What We Eat in America, NHANES 2007 -2008 (369.5). The FASEB Journal , 28(1 Supplemen t), 369 Ð5. Sidique, S. F., Joshi, S. V., & Lupi, F. (2010). Factors influencing the rate of recycling: An analysis of Minnesota counties. Resources, Conservation and Recycling , 54(4), 242Ð249. http://doi.org/ 10.1016/j.resconrec.2009.08.006 Sidique, S. F., Lupi, F., & Joshi, S. V. (2010). The effects of behavior and attitudes on drop -off recycling activities. Resources, Conservation and Recycling , 54(3), 163Ð170. http://doi.org/ 10.1016/j.resconrec.2009.07.012 Survey Sampling International. (2015) Retrieved from https://www.surveysampling.com/about/ Stevens, T. H., Tabatabaei, M., & Lass, D. (2013). Oaths and hypothetical bias. Journa l of Environmental Management , 127, 135Ð141. !"#!Tonsor, G. T., Shupp, R., & others. (2009). Val uations of sustainably produced labels on beef, tomato, and apple products. Agricultural & Resource Economics Review , 38(3), 371. U.S. Census Bureau. (2011). US D emographic Profiles. Retrieved from http://www.census.gov/2010census/news/press -kits/demographic -profiles.html US EPA. (2013). Municipal Solid Waste [Overviews & Factsheets]. Retrieved from http://www3.epa.gov/epawaste/nonhaz/municipal/