Examining Household Food Waste Decisions: A Vignette Approach

Examining Household Food Waste Decisions: A Vignette Approach Abstract Although food waste is increasingly recognized as an environmental and food security problem, there remains uncertainty over its primary contributors. Some food waste analyses seem to treat household food waste as a “mistake” or careless decision; however, consumer decisions to waste also likely reflect trade-offs and economic incentives. These issues were explored in large surveys of U.S. food consumers using both within- and between-subject designs, where we study consumers’ decisions to discard food in different scenarios that vary safety, price, and opportunity costs. We find that food waste is a function of consumers’ demographic characteristics, and that decisions to discard food vary with contextual factors. Food waste, consumer waste, vignette methodology, household behavior Food waste is argued to be a problem at virtually every point along the supply chain, and is capturing the attention of policymakers worldwide. While many efforts are focused on food waste reduction, the waste decision is rarely framed as an economic decision. However, there are likely instances where the decision to waste (as opposed to the decision to save or keep) may be optimal, depending on one’s preferences, incentives, and resource constraints. Becker (1965), for example, suggested that Americans should be more wasteful than people in lower-income countries because Americans’ opportunity cost of time is much greater than that in other countries. Daniel (2016) posited a different rationale; she found that low-income households in the United States bought fewer fruits and vegetables than higher-income consumers because they were more risk-averse regarding waste. High-income households were more willing to let food be wasted in hopes that their children would eventually acquire a taste for more healthy foods. As these and other studies suggest, the decision to waste may not always be a “mistake” or due to a lack of information, but rather results from legitimate economic incentives and trade-offs. None of this is to say that food waste is not a serious issue. There is mounting concern over the loss of scarce natural resources such as land, water, and energy that are inputs in the food production system (Gunders 2012; Buzby, Wells, and Hyman 2014; Thyberg and Tonjes 2016). Gunders (2012) reported that 10% of the total U.S. energy budget, 50% of U.S. land, and 80% of U.S. freshwater consumed is used to move food from farm to fork, yet when food is wasted, such inputs are considered to be wasted as well. With the global population expected to reach 9.3 billion by 2050, there is also an urgency to reduce food waste in hopes of (a) increasing the amount of food available to consume and (b) decreasing food prices (Buzby, Wells, and Hyman 2014). The cost of food waste has driven efforts in both the private and public sectors to reduce food waste along the supply chain, though Bellemare and colleagues (2017) note that cost estimates are likely overstated due to (a) overestimates in the quantity of food wasted, and (b) the use of the retail price of food, regardless of where the food was lost or wasted along the supply chain. Public policies are likely to be made more effective by a better understanding of the economic forces driving decisions to discard food. At the farmer-producer level, much academic research has been devoted to reducing postharvest losses, particularly in developing countries (see Hodges, Buzby, and Bennett 2011 and Affognon et al. 2015). At the foodservice (restaurant) level, food tracking technologies have been introduced that help kitchens track the quantity of food wasted before it reaches consumers’ plates. The most prominent example is the LeanPath food waste tracking software (www.leanpath.com). In addition, initiatives have been formed to bring food industry leaders together to share knowledge and identify best practices to reduce food waste in their operations. The Food Waste Reduction Alliance (FWRA) is one such effort that unites three of the food sector’s main trade associations: the Grocery Manufacturers Association, the Food Marketing Institute, and the National Restaurant Association (FWRA 2013). Other initiatives that aim to share best practices for waste reduction and food recovery across the supply chain include the U.S. Food Waste Challenge (USDA 2013); the Waste and Resources Action Programme (WRAP) in the United Kingdom (WRAP 2017); and the global SAVE FOOD initiative (FAO 2016). Despite these efforts, there has been less attention on food waste at the household level. The U.S. Food Waste Challenge and the SAVE FOOD initiative posit that food waste awareness and knowledge need to increase in households, but fewer efforts have been made to understand how households (and the consumers in them) actually make waste decisions.1 The academic research to date has primarily been descriptive in nature, gauging consumers’ knowledge of and attitudes toward food waste, motivations for wasting food, and their performance of waste-promoting or waste-reducing behaviors (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Parizeau, von Massow, and Martin 2015; Mondéjar-Jiménez et al. 2016; Qi and Roe 2016; Stancu, Haugaard, and Lähteenmäki 2016). However, to our knowledge there has been little empirical work considering economic factors that influence consumers’ utility maximizing decisions to throw out food (Lusk and McCluskey 2018). The purpose of this research is to examine household (consumer) food waste decisions. Because measuring food waste is fraught with difficulty, our first contribution is the application of vignette methodology to the issue of food waste. Our second contribution is to systematically determine how decisions to waste food vary with factors such as price, location, cost of replacement, and freshness, among other factors. The empirical analysis is concentrated on specific food waste decisions: one focused on leftovers from a fully prepared meal and a second related to a single product (milk). The empirical results show that decisions to discard food are a function of consumers’ demographic characteristics and some of the factors experimentally varied in the vignette design. Background and Literature Review Food Waste at the Household (Consumer) Level The current literature on household food waste is largely descriptive in nature. Researchers have worked to identify and understand several constructs related to food waste, including consumers’ knowledge and awareness, attitudes, motivations, and behaviors.2 Much of this work has taken place in European countries (Refsgaard and Magnussen 2009; Williams et al. 2012; Quested et al. 2013; Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2014; Lazell 2016; Mondéjar-Jiménez et al. 2016; Stancu, Haugaard, and Lähteenmäki 2016), with only a handful of studies to our knowledge examining consumers in the United States (Neff, Spiker, and Truant 2015; Qi and Roe 2016; Wilson et al. 2017). Attitudes toward food waste have been studied most extensively. Several studies (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Mondéjar-Jiménez et al. 2016; Stancu, Haugaard, and Lähteenmäki 2016) have explored food waste behavior using the Theory of Planned Behavior (TPB; Ajzen 1991), where attitudes are the central construct. In these studies, consumers exhibited positive attitudes toward reducing food waste. Within the TPB framework, attitudes were positively related to intention not to waste, as well as planning routines (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Stancu, Haugaard, and Lähteenmäki 2016; Mondéjar-Jiménez et al. 2016). Motivations have been conceptualized in two different ways in the food waste literature: (a) motivations for throwing out food and (b) motivations for reducing food waste. Food safety concerns are a key reason U.S. and European consumers throw out food. Namely, consumers are worried about the possibility of food poisoning, which could adversely affect both work and home responsibilities (Quested, Ingle, and Parry 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Qi and Roe 2016; Wilson et al. 2017). This concern is often tied to confusion over label dates such as “use by”, “sell by”, or “best by” (Gunders 2012; Quested, Ingle, and Parry 2013; Newsome et al. 2014; Wilson et al. 2017). Conversely, a primary motivation for reducing food waste is saving money (Quested, Ingle, and Parry 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Thyberg and Tonjes 2016). The literature to date has provided a broad understanding of consumers’ knowledge, attitudes, motivations, and behaviors related to food waste; however, there is a knowledge gap when it comes to understanding individual consumers’ waste decisions. When contemplating throwing food out, a consumer may consider different attributes for a banana than they do for yesterday’s leftovers. In the present study, we aim to fill this gap by exploring behaviors for two distinct waste decisions—one for leftovers from a fully-prepared meal and one for a carton of milk in a context where waste is clearly defined and where we can experimentally manipulate economic variables of interest. We examine consumers’ value of the different factors in each decision context when determining the likelihood of wasting the food in question; further, we explore the potential for heterogeneity in these decisions by interacting each decision factor with sociodemographic variables. The Vignette Method Our empirical research relies on the vignette method. Vignettes are a type of stated-preference experiment, similar to conjoint analysis, where people are asked to choose or rate hypothetical profiles (of products, situations, or other people) with varying attributes. This process allows the researcher to identify the relative importance of each decision attribute (Hainmueller, Hangartner, and Yamamoto 2015). The vignette methodology has its origins in the field of social psychology (Alexander and Becker 1978), but has been extended to research in marketing and management (see Aguinis and Bradley 2014 for a review) and economics (Kapteyn, Smith, and van Soest 2007; Epstein, Mason, and Manca 2008; Kristensen and Johansson 2008). In some cases, survey/interview questions may be too vague or difficult for respondents to answer. With food waste, for example, several studies have asked consumers to estimate the proportion of food thrown out in their household (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Neff, Spiker, and Truant 2015; Stancu, Haugaard, and Lähteenmäki 2016). The question is conceptually straightforward, but it can be challenging for respondents to answer (and for researchers to interpret) because definitions of food waste vary across consumers, meaning responses will reflect each individual’s own characterization of food waste. Further, from this question, it is impossible to know which criteria consumers use when deciding whether or not a food should be thrown out. The vignette methodology can help to overcome these limitations by providing a more concrete scenario that accounts for the most likely decision criteria and holds these criteria constant across respondents, allowing for standardization (Alexander and Becker 1978). In this study, we utilize both between-subject and within-subject vignette designs. In the between-subjects design, respondents are randomly assigned different versions of the same basic vignette. For the within-subjects design, respondents are presented with multiple vignette scenarios and asked to make decisions between them. Empirical Study 1: Leftovers Vignette In the first study, we examined a waste decision related to leftovers from a fully prepared meal. This waste decision may be different for consumers relative to a single product like juice or milk because this is a value-added product rather than a single-ingredient; therefore, the time cost of preparation may be a factor in the decision—though the importance of this factor could depend on whether or not the consumer is the one actually incurring that cost. Further, Stancu, Haugaard, and Lähteenmäki (2016) note that the reuse of leftovers may be an especially important behavior to target in terms of reducing food waste. The basic vignette shown to respondents is provided below; variables that were experimentally varied across vignettes are in brackets. Imagine you just finished eating dinner [at home; out at a restaurant]. The meal cost about [$8; $25] per person. You’re full, but there is still food left on the table – enough for [a whole; half a] lunch tomorrow. Assuming you [don’t; already] have meals planned for lunch and dinner tomorrow, what would you do? Data collection for study 1 took place in the fall of 2015 via an online survey. For this study, there were 1,016 participants, with 904 individuals randomly assigned to the between-subject design and 112 randomly assigned to the within-subject design (see table 1 for participant socio-demographic information). Table 1 Socio-Demographic Variables and Definitions Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Table 1 Socio-Demographic Variables and Definitions Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Methods: Between-Subject Design The leftovers vignette had four attributes (preparation location; price; amount left; and future meal plans) varied at two levels each. From the 16 possible vignettes (24 = 16), we selected a subset of eight vignettes such that each variable was uncorrelated with the others (an orthogonal, fractional factorial design). Respondents were randomly assigned to evaluate one (and only one) of the eight vignettes, with approximately 113 respondents per scenario. For the vignette presented, respondents were first presented with two response options: “Throw away the remaining dinner” or “Save the leftovers to eat tomorrow”. As a follow-up, we asked, “Thinking more precisely about your actions, what would you do?”, where respondents could choose one of the following five categories: (a) I’d definitely throw away what’s left of dinner; (b) I’d probably throw away what’s left of dinner; (c) I’m not sure whether I’d throw away what’s left of dinner or save the leftovers to eat tomorrow; (d) I’d probably save the leftovers to eat tomorrow; or (e) I’d definitely save the leftovers to eat tomorrow. Methods: Within-Subject Design In the within-subject design, each participant was presented with the eight vignettes used in the between-subject design. Rather than evaluating each one individually, however, they were asked to rank each of the eight scenarios from one to eight, where one was the most likely to save the leftovers and eight was the most likely to throw away the remaining dinner. The order of the appearance of the scenarios was randomized across participants. Within this design, it is important to note that we cannot ascertain the overall propensity for food waste; rather, we can only obtain information on the relative likelihood of wasting in one scenario vs. another. Results: Between-Subject Design Table 2 presents the summary statistics for the between-subject design. For each of the eight vignettes, the percentage who said they would throw out the leftovers (on the dichotomous choice question), the waste score (on the 5-point scale where 1=definitely save and 5=definitely throw out), the attributes of the vignette scenario, and the number of participants who were assigned to the vignette are provided. From the table we see that overall, participants were unlikely to waste the leftovers, with the percent wasting ranging from only 7.1% to 19.5%. Further, the mean likelihood of waste scores were well below the midpoint for all eight vignettes, leaning toward “definitely save”. Table 2 Summary Statistics for Study 1, Leftovers from a Fully Prepared Meal Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111–114. For the within-subject design, 112 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Throw out the leftovers” and 0=“Save the leftovers”; b indicates based on 5-point scale response where 1=“Definitely save” and 5=“Definitely throw out”; c indicates that vignettes were ranked such that 1=Most likely to save; 8=Most likely to throw out (waste). Table 2 Summary Statistics for Study 1, Leftovers from a Fully Prepared Meal Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111–114. For the within-subject design, 112 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Throw out the leftovers” and 0=“Save the leftovers”; b indicates based on 5-point scale response where 1=“Definitely save” and 5=“Definitely throw out”; c indicates that vignettes were ranked such that 1=Most likely to save; 8=Most likely to throw out (waste). To determine which attributes impacted the waste decision for leftovers, we estimated an ordered logit regression for the scale waste question.3 In the first model specification, we isolate the effects of the vignette experimental variables. The regression estimates are shown below. Likelihood of Waste = Intercept – 0.051*Home – 0.013* Cost per Person– 0.324*Whole Meal Leftover + 0.084*No Future Meal Plans (1) In the ordered logit regression, the intercept is represented by four threshold parameters (not reported for brevity), and significant coefficients are bolded. Home, Whole Meal Leftover, and No Future Meal Plans are indicator variables, with the effects relative to Restaurant, Half Meal Leftover, and Future Meal Plans, respectively. From the equation, we can see that there is a negative relationship between the amount of food left and waste, such that consumers were less likely to waste when there was enough for a whole meal left, rather than a half meal. To explore heterogeneity in waste behavior, we extend equation (1) by interacting each socio-demographic variable with each vignette attribute. These results are shown in table 3. Table 3 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Between-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors appear in parentheses. Table 3 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Between-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors appear in parentheses. The table reveals that meals from home had a lower likelihood of waste than meals from a restaurant; however, Democrats and medium-income households were significantly more likely to throw out meals from home relative to individuals in other political parties and low-income households, respectively. Surprisingly, the table shows that SNAP recipients were generally more likely to throw out the leftovers relative to non-recipients (intercept column), but this effect could be offset by the meal cost per person. We observed that SNAP households were significantly more price sensitive, such that they were less likely to waste high-cost meals. Related to cost, we also observed heterogeneity on the basis of age. Respondents aged 45–54 were less price sensitive (and thus, more likely to throw out the leftovers even when the cost of the meal is high) than those participants in the 65 and older category. Though there were few differences in waste preferences for the amount of leftovers and future meal plans, our results revealed that individuals with kids in the household were more likely to throw out the leftovers even though they had no future meal plans (relative to people with no children in the household). Results: Within-Subject Design The within-subject design presented each respondent with all eight vignette scenarios, and they were asked to rank the vignettes on a relative waste scale (1=most likely to keep; 8=most likely to throw out). Table 2 presents the mean ranking for each of the eight vignettes. Here, we see that respondents were most likely to save the leftovers from a meal cooked at home when the meal cost $25 per person, provided enough leftovers for a whole meal, and there were no future meal plans (mean ranking = 2.866); in contrast, respondents were most likely to throw out leftovers from a restaurant meal when the meal cost $8 per person, provided leftovers for only half a meal, and there were future meal plans in place (mean ranking = 6.027). To further examine waste decisions, we first estimate the effects of the vignette attributes on waste ranking. Equation (2) presents the results Waste Ranking =6.828 – 0.134*Home – 0.093* Cost per Person – 0.522*Whole Meal Leftover – 0.915*No Future Meal Plans. (2) From equation (2) we see that three of the four decision factors significantly impacted the waste/save decision. In particular, respondents were less likely to throw out the leftovers when (a) the meal had a higher cost per person, (b) there were enough leftovers for a whole meal rather than half a meal, and (c) there were no future meal plans in place. By estimating subject-specific regressions, we can take the coefficients estimated for each individual and use a second-stage regression to model each coefficient as a function of socio-demographic variables to account for heterogeneity in preferences.4 These results are presented in table 4. From the table, we can see in the intercept column that younger participants (aged 18–44) were overall less likely to throw out the leftovers relative to those 65 years and older. However, it should be noted that these same younger participants were also significantly less price sensitive compared to their older counterparts, meaning they were more likely to throw out higher-priced leftovers. Based on the range of prices used in this study ($8 to $25), we calculated that participants 65 years and older are more likely to throw out leftovers up to a certain dollar amount ($18.95, $12.49, and $12.65 when compared to 18–24, 25–34, and 35–44 year-olds, respectively), yet once the meal cost exceeds this amount, the younger group becomes more likely to throw out the leftovers, all else held constant. We also observed that medium-income households were overall more likely to throw out the leftovers relative to low-income households, but the reverse was true when neither group had future meal plans. Lastly, we found that respondents with children in the home were less likely to throw out higher-priced leftovers but more likely to throw out leftovers when there was enough for a whole meal compared to individuals with no children in the home. Table 4 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Within-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 4 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Within-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Empirical Study 2: Milk Vignette For the second study, we considered the waste decision process for a single product, milk. We chose milk because it is a commonly purchased product in U.S. households, and it has been identified as a product that is regularly thrown out. Gunders (2012) estimated that 20% of milk is lost along the supply chain, with the largest losses occurring at the household level. The vignette presented was about a carton of milk in the participant’s refrigerator. The variables that were experimentally varied are generally comparable to those in the leftovers vignette; however, we replaced the source attribute (home vs. restaurant) with a sensory attribute which indicated whether the milk smelled fine or slightly sour. The basic vignette shown to survey respondents is provided below; variables that varied across vignettes are in brackets. Imagine this evening you go to the refrigerator to pour a glass of milk. While taking out the carton of milk, which is [one quarter; three quarters] full, you notice that it is one day past the expiration date. You open the carton and the milk smells [fine; slightly sour]. [There is another unopened carton of milk in your refrigerator that has not expired; no statement about replacement]. Assuming the price of a half-gallon carton of milk at stores in your area is [$2.50; $5.00], what would you do? Data collection for study 2 took place in the fall of 2015 via an online survey. In total, 1,003 individuals participated; 894 were randomly assigned to the between-subject design, and 109 were randomly assigned to the within-subject design. Participant characteristics are provided in table 1. Methods: Between-Subject Design Like the leftovers vignette, the milk vignette had four attributes (fullness of carton, smell, presence of an unopened carton, and price) varied at two levels each. From the 16 possible vignettes, we selected an orthogonal, fractional factorial design of eight vignettes. Respondents were randomly assigned to one of the eight vignettes; thus, there were approximately 112 respondents per vignette. For the vignette presented, respondents were first presented with two response options: “Pour the expired milk down the drain” or “Go ahead and drink the expired milk”. Following this question, there was a follow-up that asked, “Thinking more precisely about your actions, what would you do?” Respondents could choose between the following five response options: (a) I’d definitely pour the expired milk down the drain; (b) I’d probably pour the expired milk down the drain; (c) I’m not sure whether I’d discard the milk or drink it; (d) I’d probably drink the expired milk; or (e) I’d definitely drink the expired milk. Methods: Within-Subject Design Each participant in the within-subject design was presented with the eight vignettes used in the between-subject design. They were asked to rank each of the eight scenarios from one to eight, where one was the most likely to drink and eight was the most likely to pour down the drain. The order of the appearance of the scenarios was randomized across participants. Results: Between-Subject Design Table 5 provides the summary statistics for the between-subject design. From these results, it appears that there are at least some scenarios where consumers are much more likely to pour out the milk relative to others. The four vignettes with the highest probability of waste had one attribute in common: milk that smells slightly sour. It should also be noted, though, that the milk vignettes overall had a higher likelihood of waste relative to the leftovers vignette, suggesting that the expiration date factor likely also contributed to the waste decision. Table 5 Summary Statistics for Study 2, Carton of Milk Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111-113. For the within-subject design, 109 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Pour out the milk” and 0=“Drink the milk”; b indicates based on 5-point scale response where 1=“Definitely drink” and 5=“Definitely pour out”; c indicates that vignettes were ranked such that 1=Most likely to drink; 8=Most likely to pour out (waste). Table 5 Summary Statistics for Study 2, Carton of Milk Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111-113. For the within-subject design, 109 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Pour out the milk” and 0=“Drink the milk”; b indicates based on 5-point scale response where 1=“Definitely drink” and 5=“Definitely pour out”; c indicates that vignettes were ranked such that 1=Most likely to drink; 8=Most likely to pour out (waste). To further examine which factors are likely to lead consumers to pouring out the milk (i.e., food waste), we estimated an ordered logit regression for the 5-point likelihood of waste scale. Equation (3) presents the results, with significant coefficients bolded. Likelihood of Waste= Intercept – 0.045*Price – 0.159*Three-quarters Full– 1.575*Smells Fine + 0.007*Replacement Present. (3) As in study 1, the intercept term is replaced with four threshold parameters in the ordered logit model. Further, Three-quarters Full, Smells Fine, and Replacement Present are indicator variables; their effects are relative to One-quarter Full, Smells Slightly Sour, and Replacement Absent, respectively. Looking at equation (3), it is clear that the smell variable drives the waste decision in the case of milk. When milk smells fine (as opposed to slightly sour), consumers were significantly less likely to pour out the milk. The price, fullness, and replacement variables had no statistically significant impact on the waste decision. Table 6 extends equation (3) to explore the potential for heterogeneity in preferences for the vignette attributes. We again observed that the smell attribute dominated the waste decision, where consumers were generally less likely to pour out milk that smells fine. This finding did not hold across all consumers, though. Younger participants (aged 18–44) and Democrats were significantly more likely to throw out milk even when it smelled fine compared to those who were 65 years or older and non-Democrats, respectively. In addition, Democrats were less likely to pour out the milk when the carton is fuller relative to non-Democrats. Table 6 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Between-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 6 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Between-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Results: Within-Subject Design Table 5 presents the summary statistics for the within-subject design. Similar to the between-subject design, we observe two clusters of means. The cluster that is more likely to drink the milk (rankings closer to 1) all share the characteristic that the milk smells fine, while the other cluster that is more likely to pour out the milk (rankings closer to 8) all share milk that smells slightly sour. We confirmed the impact of the smell variable on the decision to waste by running a regression on the waste rankings as a function of the vignette variables. Equation (4) shows that consumers are significantly less likely to throw out milk that smells fine relative to milk that smells slightly sour Waste Ranking =5.431+ 0.009*Price – 0.050*Three-quarters Full– 1.798*Smells Fine – 0.083*Replacement Present. (4) Using subject-specific regression estimates, we can examine heterogeneity in preferences by modeling the coefficient for each decision factor as a function of socio-demographic variables (see table 7). The table shows there was significant heterogeneity in the smell attribute. Here, males and younger participants (with the exception of the 45–54 year olds) were more likely to pour out milk when it smells fine relative to females and older participants (ages 65 and up), respectively. Within the replacement category, our results revealed that females and higher-income consumers were more likely to pour out the milk when a replacement was readily available. Though there was less variation in preferences based on price and fullness, we found that individuals who were 55–64 years old were less likely to waste when prices were high (relative to those 65 years and older) and that SNAP recipients were less likely to pour out the milk when the carton was fuller (relative to non-recipients). Table 7 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Within-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 7 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Within-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Discussion This study revealed that the keep/waste decision is not always a straightforward one. It is an economic decision, with both costs and benefits; the outcome depends on several contextual factors as well as individual-level characteristics. For waste reduction efforts to be effective, it is critical to understand the household decision process as well as the potential heterogeneities that exist across households. In the case of meal leftovers, respondents were generally less likely to waste the leftovers when the meal cost was high, when there were leftovers for a whole meal, when there were no future meal plans, and when the meal was prepared at home. Many of these relationships have a very obvious time component. Leftovers can save individuals time when there is enough for a whole meal and there are no future meal plans; further, when a meal is prepared at home, there is already a time cost for that meal (albeit a sunk cost) that people do not want to discount by throwing the leftovers out. With milk, the decision to waste was heavily impacted by food safety considerations as reflected in the smell of the product. Not surprisingly, milk that smelled slightly sour was more likely to be thrown out than milk that smelled fine—signaling individuals’ aversion to consuming a product they believe could make them or their family members ill (Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015). However, a subset of consumers opted to throw out the milk even when it smelled fine. This may be due to the expiration date information given in the vignette. In all scenarios, the milk was one day past its expiration, which may have caused some consumers to throw it out regardless of sensory properties. Our results have important implications for policymakers and the food manufacturing and retailing industries. For example, the milk results provide further evidence that more consumer education is needed on date labeling—particularly among those consumers who are likely to throw out milk even when it smells fine. Sensory properties are touted as an important signal for discerning product quality and safety, yet some consumers strictly rely on (potentially misleading) date labels. A new voluntary labeling initiative encourages food manufacturers and retailers to streamline the labeling terms used down to these two: “BEST If Used By” and “USE by” to signal product quality (e.g., the product may not taste as expected, but is safe to consume) and safety (e.g., product should not be consumed after this date), respectively (GMA 2017). Secondly, our findings suggest that there is heterogeneity in how consumers approach waste decisions. Consistent with past research (Quested, Ingle, and Parry 2013; Thyberg and Tonjes 2016), we found that younger individuals (18–44 years) were more likely to waste food than older consumers. Interestingly, this group was more likely to waste in contexts where a decision to waste was less intuitive—when the meal was higher-priced and when the milk smelled fine. One possible explanation for this may be that individuals with lower marginal productivities in meal preparation are likely to waste more (Lusk and Ellison 2017). It is likely that older individuals have acquired more skill in food preparation, and that retired individuals have more time for such activities. It may also be the case that younger consumers purchase more convenience-oriented items (frozen, microwavable, etc.) that are not well-suited for leftovers. We also found that political affiliation, sex, income level, SNAP participation, and children in the home may impact how a household approaches a waste decision, but the effects are not always straightforward or intuitive. For example, we cannot simply conclude that households with children waste less because there are more mouths to feed, or that Democrats waste less because they tend to be more environmentally-conscious. Rather, we find that consumer segments respond differently to different decision attributes. As educational campaigns are identified as a critical solution in reducing food waste (Rethink Food Waste through Economics and Data 2016), understanding the heterogeneity in waste behaviors can enable policymakers or other advocacy groups to better target educational efforts to the households most susceptible for high levels of food waste. Limitations and Areas for Future Research While this study is one of the first to examine specific food waste decisions, more work is needed to fully understand food waste at the household level. One limitation of the current study is that waste behaviors are self-reported, which may underestimate food waste (Food Loss and Waste Protocol 2016). However, as long as the self-report bias is constant across treatments, we can still identify the marginal effects of our decision attributes, even if our overall estimated level of waste is underestimated. Future studies should work to replicate these findings where behaviors are non-hypothetical and do not rely on self-reported waste data. Regarding the vignette design, this study was limited in that the outcomes for food were narrowly defined. Particularly with the milk vignette in study 2, the only outcomes proposed were for the individual to drink the milk or pour it down the drain. There was no option to use the milk for cooking, for example, which may have been more acceptable to individuals in some cases (e.g., when the milk smelled slightly sour) than drinking the milk. Footnotes 1 One notable exception is the WRAP program in the United Kingdom. This research group has undertaken rigorous household food waste audits (to learn more about the methodology used, see Quested, Ingle, and Parry 2013) and has developed an extensive consumer-facing education campaign, “Love Food Hate Waste”. For an evaluation of the impact of this campaign on household food waste behavior, see Quested and Ingle (2013). 2 We focus our discussion on attitudes toward and motivations for food waste. For more information on consumers’ knowledge and awareness of food waste, refer to Neff, Spiker, and Truant (2015); Stefan et al. (2013); and Parizeau, von Massow, and Martin (2015). For a discussion on waste-promoting and waste-reducing behaviors, see Neff, Spiker, and Truant (2015). 3 We present results for the waste scale outcome variable, rather than the dichotomous keep/waste variable because the scale allows for more variation across respondents. Logistic regression results for the dichotomous outcome variable are available upon request. 4 In total, five regressions are estimated: one for the intercept and one for each of the four vignette attributes. References Affognon H. , Mutungi C. , Sanginga P. , Borgemeister C . 2015 . Unpacking Postharvest Losses in Sub-Saharan Africa: A Meta-Analysis . World Development 66 : 49 – 68 . Google Scholar CrossRef Search ADS Aguinis H. , Bradley K.J . 2014 . Best Practice Recommendations for Designing and Implementing Experimental Vignette Methodology Studies . Organizational Research Methods 17 ( 4 ): 351 – 71 . http://dx.doi.org/10.1177/1094428114547952 Google Scholar CrossRef Search ADS Ajzen I. 1991 . The Theory of Planned Behavior . Organizational Behavior and Human Decision Processes 50 ( 2 ): 179 – 211 . http://dx.doi.org/10.1016/0749-5978(91)90020-T Google Scholar CrossRef Search ADS Alexander C.S. , Becker H.J . 1978 . The Use of Vignettes in Survey Research . Public Opinion Quarterly 42 ( 1 ): 93 – 104 . http://dx.doi.org/10.1086/268432 Google Scholar CrossRef Search ADS Becker G.S. 1965 . A Theory on the Allocation of Time . The Economic Journal 75 ( 299 ): 493 – 517 . http://dx.doi.org/10.2307/2228949 Google Scholar CrossRef Search ADS Bellemare M.F. , Çakir M. , Peterson H.H. , Novak L. , Rudi J . 2017 . On the Measurement of Food Waste . American Journal of Agricultural Economics . 99 ( 5 ): 1148 – 1158 . Google Scholar CrossRef Search ADS Buzby J.C. , Wells H.F. , Hyman J . 2014 . The Estimated Amount, Value, and Calories of Postharvest Food Losses at the Retail and Consumer Levels in the United States . Washington DC : U.S. Department of Agriculture, Economic Research Service . Daniel C. 2016 . Economic Constraints on Taste Formation and the True Cost of Healthy Eating . Social Science & Medicine 148 : 34 – 41 . Google Scholar CrossRef Search ADS Epstein D. , Mason A. , Manca A . 2008 . The Hospital Costs of Care for Stroke in Nine European Countries . Health Economics 17 ( S1 ): S21 – S31 . Google Scholar CrossRef Search ADS PubMed Food and Agriculture Organization of the United Nations . 2016 . FAO and Messe Düsseldorf Lead the SAVE FOOD – Global Initiative on Food Loss and Waste Reduction. Available at: http://www.fao.org/save-food/background/en/. Food Loss and Waste Protocol . 2016 . Food Loss and Waste Accounting and Reporting Standard: Version 1.0. Available at: http://www.wri.org/sites/default/files/FLW_Standard_final_2016.pdf. Food Waste Reduction Alliance . 2013 . About the Food Waste Reduction Alliance. Available at: http://www.foodwastealliance.org/about-us-page-2/. Graham-Rowe E. , Jessop D.C. , Sparks P . 2014 . Identifying Motivations and Barriers to Minimising Household Food Waste . Resources, Conservation and Recycling 84 : 15 – 23 . Google Scholar CrossRef Search ADS Graham-Rowe E. , Jessop D.C. , Sparks P. 2015 . Predicting Household Food Waste Reduction Using an Extended Theory of Planned Behavior . Resources, Conservation and Recycling 101 : 194 – 202 . Google Scholar CrossRef Search ADS Grocery Manufacturers Association . 2017 . Grocery Industry Launches New Initiative to Reduce Consumer Confusion on Product Date Labels. Available at: http://www.gmaonline.org/news-events/newsroom/grocery-industry-launches-new-initiative-to-reduce-consumer-confusion-on-pr/. Gunders D. 2012 . Wasted: How America is Losing Up to 40 Percent of Its Food from Farm to Fork to Landfill. Natural Resources Defense Council (NRDC) Issue Paper: 12-06-B. Available at: http://www.indianasna.org/content/indianasna/documents/NRDC_Wasted_Food_Report.pdf. Hainmueller J. , Hangartner D. , Yamamoto T . 2015 . Validating Vignette and Conjoint Survey Experiments against Real-World Behavior . Proceedings of the National Academies of Science 112 ( 8 ): 2395 – 400 . http://dx.doi.org/10.1073/pnas.1416587112 Google Scholar CrossRef Search ADS Hodges R.J. , Buzby J.C. , Bennett B . 2011 . Postharvest Losses and Waste in Developed and Less Developed Countries: Opportunities to Improve Resource Use . Journal of Agricultural Science 149 : 37 – 45 . Google Scholar CrossRef Search ADS Kapteyn A. , Smith J.P. , van Soest A . 2007 . Vignettes and Self-Reports of Work Disability in the United States and the Netherlands . The American Economic Review 97 ( 1 ): 461 – 73 . http://dx.doi.org/10.1257/aer.97.1.461 Google Scholar CrossRef Search ADS Kristensen N. , Johansson E . 2008 . New Evidence on Cross-Country Differences in Job Satisfaction Using Anchoring Vignettes . Labour Economics 15 ( 1 ): 96 – 117 . http://dx.doi.org/10.1016/j.labeco.2006.11.001 Google Scholar CrossRef Search ADS Lazell J. 2016 . Consumer Food Waste Behaviour in Universities: Sharing as a Means of Prevention . Journal of Consumer Behavior 15 ( 5 ): 430 – 39 . http://dx.doi.org/10.1002/cb.1581 Google Scholar CrossRef Search ADS Lusk J.L. , Ellison B . 2017 . A Note on Modelling Household Food Waste Behaviour . Applied Economics Letters 24 ( 16 ): 1199 – 202 . http://dx.doi.org/10.1080/13504851.2016.1265070 Google Scholar CrossRef Search ADS Lusk J.L. , McCluskey J . Forthcoming. Understanding the Impacts of Food Consumer Choice and Food Policy Outcomes . Applied Economic Perspectives and Policy . Mondéjar-Jiménez J.-A. , Ferrari G. , Secondi L. , Principato L . 2016 . From the Table to Waste: An Exploratory Study on Behavior Towards Food Waste of Spanish and Italian Youths . Journal of Cleaner Production 138 ( 1 ): 8 – 18 . Google Scholar CrossRef Search ADS Neff R.A. , Spiker M.L. , Truant P.L . 2015 . Wasted Food: U.S. Consumers’ Reported Awareness, Attitudes, and Behaviors . PLoS ONE 10 ( 6 ): e0127881 . Google Scholar CrossRef Search ADS PubMed Newsome R. , Balestrini C.G. , Baum M.D. , Corby J. , Fisher W. , Goodburn K. , Labuza T.P. , et al. 2014 . Applications and Perceptions of Date Labeling of Food . Comprehensive Reviews in Food Science and Food Safety 13 ( 4 ): 745 – 69 . Google Scholar CrossRef Search ADS Parizeau K. , von Massow M. , Martin R . 2015 . Household-level Dynamics of Food Waste Production and Related Beliefs, Attitudes, and Behaviors in Guelph, Ontario . Waste Management 35 : 207 – 17 . Google Scholar CrossRef Search ADS PubMed Rethink Food Waste through Economics and Data . 2016 . A Roadmap to Reduce U.S. Food Waste by 20 Percent. Available at: http://www.refed.com/downloads/ReFED_Report_2016.pdf. Qi D. , Roe B.E . 2016 . Household Food Waste: Multivariate Regression and Principal Components Analyses of Awareness and Attitudes among U.S. Consumers . PLoS ONE 11 ( 7 ): e0159250 . Google Scholar CrossRef Search ADS PubMed Quested T. , Ingle R . 2013 . West London Food Waste Prevention Campaign Evaluation Report. WRAP Research Report, Project Code: CFP104-030. Available at: http://www.wrap.org.uk/sites/files/wrap/West%20London%20Food%20Waste%20Campaign%20Evaluation%20Report_1.pdf. Quested T. , Ingle R. , Parry A . 2013 . Household Food and Drink Waste in the United Kingdom 2012. WRAP Research Report, Project Code: CFP102. Available at: http://www.wrap.org.uk/sites/files/wrap/hhfdw-2012-main.pdf.pdf. Quested T.E. , Marsh E. , Stunell D. , Parry A.D . 2013 . Spaghetti Soup: The Complex World of Food Waste Behaviors . Resources, Conservation and Recycling 79 : 43 – 51 . Google Scholar CrossRef Search ADS Refsgaard K. , Magnussen K . 2009 . Household Behaviour and Attitudes with Respect to Recycling Food Waste – Experiences from Focus Groups . Journal of Environmental Management 90 : 760 – 71 . Google Scholar CrossRef Search ADS PubMed Stancu V. , Haugaard P. , Lähteenmäki L . 2016 . Determinants of Consumer Food Waste Behavior: Two Routes to Food Waste . Appetite 96 : 7 – 17 . Google Scholar CrossRef Search ADS PubMed Stefan V. , van Herpen E. , Tudoran A.A. , Lähteenmäki L . 2013 . Avoiding Food Waste by Romanian Consumers: The Importance of Planning and Shopping Routines . Food Quality and Preference 28 : 375 – 81 . Google Scholar CrossRef Search ADS Thyberg K.L. , Tonjes D.J . 2016 . Drivers of Food Waste and their Implications for Sustainable Policy Development . Resources, Conservation and Recycling 106 : 110 – 23 . Google Scholar CrossRef Search ADS U.S. Department of Agriculture . 2013 . USDA and EPA Launch U.S. Food Waste Challenge. Washington DC: USDA Press Release Number 0112.13. Available at: http://www.usda.gov/wps/portal/usda/usdahome?contentidonly=true&contentid=2013/06/0112.xml. Waste and Resources Action Programme . 2017 . Food Waste Reduction. Available at: http://www.wrap.org.uk/food-waste-reduction. Williams H. , Wikström F. , Otterbring T. , Löfgren M. , Gustafsson A . 2012 . Reasons for Household Food Waste with Special Attention to Packaging . Journal of Cleaner Production 24 : 141 – 8 . Google Scholar CrossRef Search ADS Wilson N.L.W. , Rickard B.J. , Saputo R. , Ho S.-T . 2017 . Food Waste: The Role of Date Labels, Package Size, and Product Category . Food Quality and Preference 55 : 35 – 44 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Economic Perspectives and Policy Oxford University Press

Examining Household Food Waste Decisions: A Vignette Approach

Loading next page...
 
/lp/ou_press/examining-household-food-waste-decisions-a-vignette-approach-08T0N03y9q
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com
ISSN
2040-5790
eISSN
2040-5804
D.O.I.
10.1093/aepp/ppx059
Publisher site
See Article on Publisher Site

Abstract

Abstract Although food waste is increasingly recognized as an environmental and food security problem, there remains uncertainty over its primary contributors. Some food waste analyses seem to treat household food waste as a “mistake” or careless decision; however, consumer decisions to waste also likely reflect trade-offs and economic incentives. These issues were explored in large surveys of U.S. food consumers using both within- and between-subject designs, where we study consumers’ decisions to discard food in different scenarios that vary safety, price, and opportunity costs. We find that food waste is a function of consumers’ demographic characteristics, and that decisions to discard food vary with contextual factors. Food waste, consumer waste, vignette methodology, household behavior Food waste is argued to be a problem at virtually every point along the supply chain, and is capturing the attention of policymakers worldwide. While many efforts are focused on food waste reduction, the waste decision is rarely framed as an economic decision. However, there are likely instances where the decision to waste (as opposed to the decision to save or keep) may be optimal, depending on one’s preferences, incentives, and resource constraints. Becker (1965), for example, suggested that Americans should be more wasteful than people in lower-income countries because Americans’ opportunity cost of time is much greater than that in other countries. Daniel (2016) posited a different rationale; she found that low-income households in the United States bought fewer fruits and vegetables than higher-income consumers because they were more risk-averse regarding waste. High-income households were more willing to let food be wasted in hopes that their children would eventually acquire a taste for more healthy foods. As these and other studies suggest, the decision to waste may not always be a “mistake” or due to a lack of information, but rather results from legitimate economic incentives and trade-offs. None of this is to say that food waste is not a serious issue. There is mounting concern over the loss of scarce natural resources such as land, water, and energy that are inputs in the food production system (Gunders 2012; Buzby, Wells, and Hyman 2014; Thyberg and Tonjes 2016). Gunders (2012) reported that 10% of the total U.S. energy budget, 50% of U.S. land, and 80% of U.S. freshwater consumed is used to move food from farm to fork, yet when food is wasted, such inputs are considered to be wasted as well. With the global population expected to reach 9.3 billion by 2050, there is also an urgency to reduce food waste in hopes of (a) increasing the amount of food available to consume and (b) decreasing food prices (Buzby, Wells, and Hyman 2014). The cost of food waste has driven efforts in both the private and public sectors to reduce food waste along the supply chain, though Bellemare and colleagues (2017) note that cost estimates are likely overstated due to (a) overestimates in the quantity of food wasted, and (b) the use of the retail price of food, regardless of where the food was lost or wasted along the supply chain. Public policies are likely to be made more effective by a better understanding of the economic forces driving decisions to discard food. At the farmer-producer level, much academic research has been devoted to reducing postharvest losses, particularly in developing countries (see Hodges, Buzby, and Bennett 2011 and Affognon et al. 2015). At the foodservice (restaurant) level, food tracking technologies have been introduced that help kitchens track the quantity of food wasted before it reaches consumers’ plates. The most prominent example is the LeanPath food waste tracking software (www.leanpath.com). In addition, initiatives have been formed to bring food industry leaders together to share knowledge and identify best practices to reduce food waste in their operations. The Food Waste Reduction Alliance (FWRA) is one such effort that unites three of the food sector’s main trade associations: the Grocery Manufacturers Association, the Food Marketing Institute, and the National Restaurant Association (FWRA 2013). Other initiatives that aim to share best practices for waste reduction and food recovery across the supply chain include the U.S. Food Waste Challenge (USDA 2013); the Waste and Resources Action Programme (WRAP) in the United Kingdom (WRAP 2017); and the global SAVE FOOD initiative (FAO 2016). Despite these efforts, there has been less attention on food waste at the household level. The U.S. Food Waste Challenge and the SAVE FOOD initiative posit that food waste awareness and knowledge need to increase in households, but fewer efforts have been made to understand how households (and the consumers in them) actually make waste decisions.1 The academic research to date has primarily been descriptive in nature, gauging consumers’ knowledge of and attitudes toward food waste, motivations for wasting food, and their performance of waste-promoting or waste-reducing behaviors (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Parizeau, von Massow, and Martin 2015; Mondéjar-Jiménez et al. 2016; Qi and Roe 2016; Stancu, Haugaard, and Lähteenmäki 2016). However, to our knowledge there has been little empirical work considering economic factors that influence consumers’ utility maximizing decisions to throw out food (Lusk and McCluskey 2018). The purpose of this research is to examine household (consumer) food waste decisions. Because measuring food waste is fraught with difficulty, our first contribution is the application of vignette methodology to the issue of food waste. Our second contribution is to systematically determine how decisions to waste food vary with factors such as price, location, cost of replacement, and freshness, among other factors. The empirical analysis is concentrated on specific food waste decisions: one focused on leftovers from a fully prepared meal and a second related to a single product (milk). The empirical results show that decisions to discard food are a function of consumers’ demographic characteristics and some of the factors experimentally varied in the vignette design. Background and Literature Review Food Waste at the Household (Consumer) Level The current literature on household food waste is largely descriptive in nature. Researchers have worked to identify and understand several constructs related to food waste, including consumers’ knowledge and awareness, attitudes, motivations, and behaviors.2 Much of this work has taken place in European countries (Refsgaard and Magnussen 2009; Williams et al. 2012; Quested et al. 2013; Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2014; Lazell 2016; Mondéjar-Jiménez et al. 2016; Stancu, Haugaard, and Lähteenmäki 2016), with only a handful of studies to our knowledge examining consumers in the United States (Neff, Spiker, and Truant 2015; Qi and Roe 2016; Wilson et al. 2017). Attitudes toward food waste have been studied most extensively. Several studies (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Mondéjar-Jiménez et al. 2016; Stancu, Haugaard, and Lähteenmäki 2016) have explored food waste behavior using the Theory of Planned Behavior (TPB; Ajzen 1991), where attitudes are the central construct. In these studies, consumers exhibited positive attitudes toward reducing food waste. Within the TPB framework, attitudes were positively related to intention not to waste, as well as planning routines (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Stancu, Haugaard, and Lähteenmäki 2016; Mondéjar-Jiménez et al. 2016). Motivations have been conceptualized in two different ways in the food waste literature: (a) motivations for throwing out food and (b) motivations for reducing food waste. Food safety concerns are a key reason U.S. and European consumers throw out food. Namely, consumers are worried about the possibility of food poisoning, which could adversely affect both work and home responsibilities (Quested, Ingle, and Parry 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Qi and Roe 2016; Wilson et al. 2017). This concern is often tied to confusion over label dates such as “use by”, “sell by”, or “best by” (Gunders 2012; Quested, Ingle, and Parry 2013; Newsome et al. 2014; Wilson et al. 2017). Conversely, a primary motivation for reducing food waste is saving money (Quested, Ingle, and Parry 2013; Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015; Thyberg and Tonjes 2016). The literature to date has provided a broad understanding of consumers’ knowledge, attitudes, motivations, and behaviors related to food waste; however, there is a knowledge gap when it comes to understanding individual consumers’ waste decisions. When contemplating throwing food out, a consumer may consider different attributes for a banana than they do for yesterday’s leftovers. In the present study, we aim to fill this gap by exploring behaviors for two distinct waste decisions—one for leftovers from a fully-prepared meal and one for a carton of milk in a context where waste is clearly defined and where we can experimentally manipulate economic variables of interest. We examine consumers’ value of the different factors in each decision context when determining the likelihood of wasting the food in question; further, we explore the potential for heterogeneity in these decisions by interacting each decision factor with sociodemographic variables. The Vignette Method Our empirical research relies on the vignette method. Vignettes are a type of stated-preference experiment, similar to conjoint analysis, where people are asked to choose or rate hypothetical profiles (of products, situations, or other people) with varying attributes. This process allows the researcher to identify the relative importance of each decision attribute (Hainmueller, Hangartner, and Yamamoto 2015). The vignette methodology has its origins in the field of social psychology (Alexander and Becker 1978), but has been extended to research in marketing and management (see Aguinis and Bradley 2014 for a review) and economics (Kapteyn, Smith, and van Soest 2007; Epstein, Mason, and Manca 2008; Kristensen and Johansson 2008). In some cases, survey/interview questions may be too vague or difficult for respondents to answer. With food waste, for example, several studies have asked consumers to estimate the proportion of food thrown out in their household (Stefan et al. 2013; Graham-Rowe, Jessop, and Sparks 2015; Neff, Spiker, and Truant 2015; Stancu, Haugaard, and Lähteenmäki 2016). The question is conceptually straightforward, but it can be challenging for respondents to answer (and for researchers to interpret) because definitions of food waste vary across consumers, meaning responses will reflect each individual’s own characterization of food waste. Further, from this question, it is impossible to know which criteria consumers use when deciding whether or not a food should be thrown out. The vignette methodology can help to overcome these limitations by providing a more concrete scenario that accounts for the most likely decision criteria and holds these criteria constant across respondents, allowing for standardization (Alexander and Becker 1978). In this study, we utilize both between-subject and within-subject vignette designs. In the between-subjects design, respondents are randomly assigned different versions of the same basic vignette. For the within-subjects design, respondents are presented with multiple vignette scenarios and asked to make decisions between them. Empirical Study 1: Leftovers Vignette In the first study, we examined a waste decision related to leftovers from a fully prepared meal. This waste decision may be different for consumers relative to a single product like juice or milk because this is a value-added product rather than a single-ingredient; therefore, the time cost of preparation may be a factor in the decision—though the importance of this factor could depend on whether or not the consumer is the one actually incurring that cost. Further, Stancu, Haugaard, and Lähteenmäki (2016) note that the reuse of leftovers may be an especially important behavior to target in terms of reducing food waste. The basic vignette shown to respondents is provided below; variables that were experimentally varied across vignettes are in brackets. Imagine you just finished eating dinner [at home; out at a restaurant]. The meal cost about [$8; $25] per person. You’re full, but there is still food left on the table – enough for [a whole; half a] lunch tomorrow. Assuming you [don’t; already] have meals planned for lunch and dinner tomorrow, what would you do? Data collection for study 1 took place in the fall of 2015 via an online survey. For this study, there were 1,016 participants, with 904 individuals randomly assigned to the between-subject design and 112 randomly assigned to the within-subject design (see table 1 for participant socio-demographic information). Table 1 Socio-Demographic Variables and Definitions Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Table 1 Socio-Demographic Variables and Definitions Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Study 1 Study 2 Variable Definition Sample Proportion Sample Proportion Female 1 if female; 0 if male 0.535 0.500 Age 18–24 1 if 18–24 years old; 0 otherwise 0.115 0.125 Age 25–34 1 if 25–34 years old; 0 otherwise 0.228 0.227 Age 35–44 1 if 35–44 years old; 0 otherwise 0.189 0.199 Age 45–54 1 if 45–54 years old; 0 otherwise 0.157 0.154 Age 55–64 1 if 55–64 years old; 0 otherwise 0.152 0.171 Age 65 and older 1 if 65 years or older; 0 otherwise 0.158 0.124 SNAP 1 if current SNAP recipient; 0 otherwise 0.143 0.168 College degree 1 if obtained college degree; 0 otherwise 0.495 0.507 Democrat 1 if identifies as a Democrat; 0 for all other parties 0.440 0.466 Obese 1 if BMI ≥ 30; 0 otherwise 0.279 0.244 Kids in household 1 if children under age 12 living in the household; 0 otherwise 0.325 0.362 Low Income 1 if annual income is less than $40,000; 0 otherwise 0.263 0.281 Medium Income 1 if annual income is $40,000–$99,999; 0 otherwise 0.459 0.471 High Income 1 if annual income is $100,000 or more; 0 otherwise 0.279 0.248 Number of Observations 1,016 1,003 Methods: Between-Subject Design The leftovers vignette had four attributes (preparation location; price; amount left; and future meal plans) varied at two levels each. From the 16 possible vignettes (24 = 16), we selected a subset of eight vignettes such that each variable was uncorrelated with the others (an orthogonal, fractional factorial design). Respondents were randomly assigned to evaluate one (and only one) of the eight vignettes, with approximately 113 respondents per scenario. For the vignette presented, respondents were first presented with two response options: “Throw away the remaining dinner” or “Save the leftovers to eat tomorrow”. As a follow-up, we asked, “Thinking more precisely about your actions, what would you do?”, where respondents could choose one of the following five categories: (a) I’d definitely throw away what’s left of dinner; (b) I’d probably throw away what’s left of dinner; (c) I’m not sure whether I’d throw away what’s left of dinner or save the leftovers to eat tomorrow; (d) I’d probably save the leftovers to eat tomorrow; or (e) I’d definitely save the leftovers to eat tomorrow. Methods: Within-Subject Design In the within-subject design, each participant was presented with the eight vignettes used in the between-subject design. Rather than evaluating each one individually, however, they were asked to rank each of the eight scenarios from one to eight, where one was the most likely to save the leftovers and eight was the most likely to throw away the remaining dinner. The order of the appearance of the scenarios was randomized across participants. Within this design, it is important to note that we cannot ascertain the overall propensity for food waste; rather, we can only obtain information on the relative likelihood of wasting in one scenario vs. another. Results: Between-Subject Design Table 2 presents the summary statistics for the between-subject design. For each of the eight vignettes, the percentage who said they would throw out the leftovers (on the dichotomous choice question), the waste score (on the 5-point scale where 1=definitely save and 5=definitely throw out), the attributes of the vignette scenario, and the number of participants who were assigned to the vignette are provided. From the table we see that overall, participants were unlikely to waste the leftovers, with the percent wasting ranging from only 7.1% to 19.5%. Further, the mean likelihood of waste scores were well below the midpoint for all eight vignettes, leaning toward “definitely save”. Table 2 Summary Statistics for Study 1, Leftovers from a Fully Prepared Meal Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111–114. For the within-subject design, 112 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Throw out the leftovers” and 0=“Save the leftovers”; b indicates based on 5-point scale response where 1=“Definitely save” and 5=“Definitely throw out”; c indicates that vignettes were ranked such that 1=Most likely to save; 8=Most likely to throw out (waste). Table 2 Summary Statistics for Study 1, Leftovers from a Fully Prepared Meal Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Location Cost per Person Amount of Meal Leftover Future Meal Plans 1 14.90% 1.667 4.696 (2.044) restaurant $8 whole no 2 19.50% 1.973 6.027 (2.252) restaurant $8 half yes 3 8.00% 1.545 3.902 (2.135) restaurant $25 whole yes 4 11.70% 1.721 3.643 (2.008) restaurant $25 half no 5 12.30% 1.623 5.491 (2.110) home $8 whole yes 6 12.30% 1.930 4.964 (2.079) home $8 half no 7 7.10% 1.752 2.866 (2.064) home $25 whole no 8 8.80% 1.602 4.411 (2.025) home $25 half yes Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111–114. For the within-subject design, 112 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Throw out the leftovers” and 0=“Save the leftovers”; b indicates based on 5-point scale response where 1=“Definitely save” and 5=“Definitely throw out”; c indicates that vignettes were ranked such that 1=Most likely to save; 8=Most likely to throw out (waste). To determine which attributes impacted the waste decision for leftovers, we estimated an ordered logit regression for the scale waste question.3 In the first model specification, we isolate the effects of the vignette experimental variables. The regression estimates are shown below. Likelihood of Waste = Intercept – 0.051*Home – 0.013* Cost per Person– 0.324*Whole Meal Leftover + 0.084*No Future Meal Plans (1) In the ordered logit regression, the intercept is represented by four threshold parameters (not reported for brevity), and significant coefficients are bolded. Home, Whole Meal Leftover, and No Future Meal Plans are indicator variables, with the effects relative to Restaurant, Half Meal Leftover, and Future Meal Plans, respectively. From the equation, we can see that there is a negative relationship between the amount of food left and waste, such that consumers were less likely to waste when there was enough for a whole meal left, rather than a half meal. To explore heterogeneity in waste behavior, we extend equation (1) by interacting each socio-demographic variable with each vignette attribute. These results are shown in table 3. Table 3 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Between-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors appear in parentheses. Table 3 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Between-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a a −1.344* (0.574)† −0.02 (0.034) −0.863 (0.576) −0.234 (0.573) Female vs. Male −0.590 (0.400) −0.268 (0.297) 0.004 (0.017) 0.113 (0.299) −0.355 (0.297) Age 18–24 vs. 65 and older −0.552 (0.835) 0.821 (0.596) 0.042 (0.035) 0.795 (0.598) 0.998 (0.602) Age 25–34 vs. 65 and older 0.308 (0.765) 0.662 (0.543) 0.040 (0.032) 0.138 (0.544) −0.448 (0.546) Age 35–44 vs. 65 and older −0.125 (0.824) 0.083 (0.578) 0.021 (0.034) 0.829 (0.579) 0.478 (0.584) Age 45–54 vs. 65 and older −0.545 (0.799) 0.032 (0.582) 0.078* (0.034) −0.063 (0.589) −0.492 (0.583) Age 55–64 vs. 65 and older −0.871 (0.86) 0.720 (0.590) 0.046 (0.035) 0.341 (0.593) −0.343 (0.593) SNAP vs. Non-SNAP 2.131* (0.591) 0.135 (0.438) −0.107* (0.026) 0.175 (0.428) 0.485 (0.429) College degree vs. No degree 0.659 (0.455) −0.279 (0.347) −0.016 (0.020) 0.348 (0.348) −0.076 (0.349) Democrat vs. Other parties −0.225 (0.395) 0.698* (0.297) 0.009 (0.017) −0.107 (0.298) 0.183 (0.295) Obese vs. Non-obese −0.241 (0.435) 0.404 (0.332) −0.016 (0.020) 0.319 (0.333) 0.143 (0.331) Kids in household vs. No kids −0.259 (0.47) −0.286 (0.349) 0.012 (0.021) 0.380 (0.352) 0.998* (0.352) Medium vs. Low income 0.006 (0.533) 1.061* (0.412) 0.001 (0.024) −0.149 (0.414) −0.268 (0.411) High vs. Low income 1.066 (0.599) 0.724 (0.472) −0.046 (0.028) −0.541 (0.474) 0.255 (0.474) Number of Observations 904 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors appear in parentheses. The table reveals that meals from home had a lower likelihood of waste than meals from a restaurant; however, Democrats and medium-income households were significantly more likely to throw out meals from home relative to individuals in other political parties and low-income households, respectively. Surprisingly, the table shows that SNAP recipients were generally more likely to throw out the leftovers relative to non-recipients (intercept column), but this effect could be offset by the meal cost per person. We observed that SNAP households were significantly more price sensitive, such that they were less likely to waste high-cost meals. Related to cost, we also observed heterogeneity on the basis of age. Respondents aged 45–54 were less price sensitive (and thus, more likely to throw out the leftovers even when the cost of the meal is high) than those participants in the 65 and older category. Though there were few differences in waste preferences for the amount of leftovers and future meal plans, our results revealed that individuals with kids in the household were more likely to throw out the leftovers even though they had no future meal plans (relative to people with no children in the household). Results: Within-Subject Design The within-subject design presented each respondent with all eight vignette scenarios, and they were asked to rank the vignettes on a relative waste scale (1=most likely to keep; 8=most likely to throw out). Table 2 presents the mean ranking for each of the eight vignettes. Here, we see that respondents were most likely to save the leftovers from a meal cooked at home when the meal cost $25 per person, provided enough leftovers for a whole meal, and there were no future meal plans (mean ranking = 2.866); in contrast, respondents were most likely to throw out leftovers from a restaurant meal when the meal cost $8 per person, provided leftovers for only half a meal, and there were future meal plans in place (mean ranking = 6.027). To further examine waste decisions, we first estimate the effects of the vignette attributes on waste ranking. Equation (2) presents the results Waste Ranking =6.828 – 0.134*Home – 0.093* Cost per Person – 0.522*Whole Meal Leftover – 0.915*No Future Meal Plans. (2) From equation (2) we see that three of the four decision factors significantly impacted the waste/save decision. In particular, respondents were less likely to throw out the leftovers when (a) the meal had a higher cost per person, (b) there were enough leftovers for a whole meal rather than half a meal, and (c) there were no future meal plans in place. By estimating subject-specific regressions, we can take the coefficients estimated for each individual and use a second-stage regression to model each coefficient as a function of socio-demographic variables to account for heterogeneity in preferences.4 These results are presented in table 4. From the table, we can see in the intercept column that younger participants (aged 18–44) were overall less likely to throw out the leftovers relative to those 65 years and older. However, it should be noted that these same younger participants were also significantly less price sensitive compared to their older counterparts, meaning they were more likely to throw out higher-priced leftovers. Based on the range of prices used in this study ($8 to $25), we calculated that participants 65 years and older are more likely to throw out leftovers up to a certain dollar amount ($18.95, $12.49, and $12.65 when compared to 18–24, 25–34, and 35–44 year-olds, respectively), yet once the meal cost exceeds this amount, the younger group becomes more likely to throw out the leftovers, all else held constant. We also observed that medium-income households were overall more likely to throw out the leftovers relative to low-income households, but the reverse was true when neither group had future meal plans. Lastly, we found that respondents with children in the home were less likely to throw out higher-priced leftovers but more likely to throw out leftovers when there was enough for a whole meal compared to individuals with no children in the home. Table 4 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Within-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 4 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 1, Within-Subject Design) Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Interaction with … Intercept Home vs. Restaurant Cost per Person Whole vs. Half Meal Leftover No Future Meal Plans vs. Plans n/a 7.180* (1.079) 0.872 (0.630) −0.181* (0.052) 0.016 (0.666) −0.283 (0.696) Female vs. Male 0.058 (0.489) −0.498 (0.285) 0.019 (0.024) −0.244 (0.302) −0.016 (0.315) Age 18–24 vs. 65 and older −2.501* (1.056) 0.338 (0.616) 0.132* (0.051) 0.047 (0.651) 0.271 (0.681) Age 25–34 vs. 65 and older −1.774 (0.949) −0.612 (0.554) 0.142* (0.046) −0.399 (0.585) −0.115 (0.612) Age 35–44 vs. 65 and older −2.062* (0.935) −0.649 (0.546) 0.163* (0.045) −0.593 (0.577) 0.002 (0.603) Age 45–54 vs. 65 and older −0.177 (1.002) −0.909 (0.585) 0.050 (0.048) −0.184 (0.618) −0.189 (0.646) Age 55–64 vs. 65 and older −0.806 (0.954) −0.571 (0.557) 0.086 (0.046) −0.280 (0.589) −0.364 (0.615) SNAP vs. Non-SNAP −0.865 (0.760) −0.582 (0.444) 0.071 (0.037) 0.306 (0.469) −0.327 (0.490) College degree vs. No degree −0.124 (0.562) 0.108 (0.328) 0.008 (0.027) −0.110 (0.347) −0.018 (0.363) Democrat vs. Other parties −0.444 (0.555) −0.510 (0.324) 0.035 (0.027) −0.043 (0.342) 0.282 (0.358) Obese vs. Non-obese 0.503 (0.551) −0.303 (0.321) −0.014 (0.027) −0.133 (0.340) −0.114 (0.355) Kids in household vs. No kids 0.237 (0.622) 0.173 (0.363) −0.066* (0.030) 1.119* (0.384) 0.425 (0.401) Medium vs. Low income 1.431* (0.631) 0.164 (0.368) −0.046 (0.030) −0.472 (0.389) −1.027* (0.407) High vs. Low income 1.508 (0.807) −0.263 (0.471) −0.039 (0.039) −0.611 (0.498) −0.842 (0.520) Number of Observations 112 112 112 112 112 R-Squared 0.23 0.14 0.25 0.13 0.12 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Home vs. Restaurant, for example, the variable is coded such that 1=Meal Prepared at Home and 0=Meal Prepared at a Restaurant. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Empirical Study 2: Milk Vignette For the second study, we considered the waste decision process for a single product, milk. We chose milk because it is a commonly purchased product in U.S. households, and it has been identified as a product that is regularly thrown out. Gunders (2012) estimated that 20% of milk is lost along the supply chain, with the largest losses occurring at the household level. The vignette presented was about a carton of milk in the participant’s refrigerator. The variables that were experimentally varied are generally comparable to those in the leftovers vignette; however, we replaced the source attribute (home vs. restaurant) with a sensory attribute which indicated whether the milk smelled fine or slightly sour. The basic vignette shown to survey respondents is provided below; variables that varied across vignettes are in brackets. Imagine this evening you go to the refrigerator to pour a glass of milk. While taking out the carton of milk, which is [one quarter; three quarters] full, you notice that it is one day past the expiration date. You open the carton and the milk smells [fine; slightly sour]. [There is another unopened carton of milk in your refrigerator that has not expired; no statement about replacement]. Assuming the price of a half-gallon carton of milk at stores in your area is [$2.50; $5.00], what would you do? Data collection for study 2 took place in the fall of 2015 via an online survey. In total, 1,003 individuals participated; 894 were randomly assigned to the between-subject design, and 109 were randomly assigned to the within-subject design. Participant characteristics are provided in table 1. Methods: Between-Subject Design Like the leftovers vignette, the milk vignette had four attributes (fullness of carton, smell, presence of an unopened carton, and price) varied at two levels each. From the 16 possible vignettes, we selected an orthogonal, fractional factorial design of eight vignettes. Respondents were randomly assigned to one of the eight vignettes; thus, there were approximately 112 respondents per vignette. For the vignette presented, respondents were first presented with two response options: “Pour the expired milk down the drain” or “Go ahead and drink the expired milk”. Following this question, there was a follow-up that asked, “Thinking more precisely about your actions, what would you do?” Respondents could choose between the following five response options: (a) I’d definitely pour the expired milk down the drain; (b) I’d probably pour the expired milk down the drain; (c) I’m not sure whether I’d discard the milk or drink it; (d) I’d probably drink the expired milk; or (e) I’d definitely drink the expired milk. Methods: Within-Subject Design Each participant in the within-subject design was presented with the eight vignettes used in the between-subject design. They were asked to rank each of the eight scenarios from one to eight, where one was the most likely to drink and eight was the most likely to pour down the drain. The order of the appearance of the scenarios was randomized across participants. Results: Between-Subject Design Table 5 provides the summary statistics for the between-subject design. From these results, it appears that there are at least some scenarios where consumers are much more likely to pour out the milk relative to others. The four vignettes with the highest probability of waste had one attribute in common: milk that smells slightly sour. It should also be noted, though, that the milk vignettes overall had a higher likelihood of waste relative to the leftovers vignette, suggesting that the expiration date factor likely also contributed to the waste decision. Table 5 Summary Statistics for Study 2, Carton of Milk Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111-113. For the within-subject design, 109 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Pour out the milk” and 0=“Drink the milk”; b indicates based on 5-point scale response where 1=“Definitely drink” and 5=“Definitely pour out”; c indicates that vignettes were ranked such that 1=Most likely to drink; 8=Most likely to pour out (waste). Table 5 Summary Statistics for Study 2, Carton of Milk Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Between-Subject Design Within-Subject Design Scenario % Wastinga Likelihood of Wasteb Mean Rankingc (std. dev.) Price Fullness Smell Replacement 1 50.50% 2.838 5.486 (2.234) $2.50 one-quarter sour absent 2 15.30% 1.622 3.688 (2.044) $2.50 one-quarter fine present 3 51.80% 2.857 5.450 (1.808) $2.50 three-quarters sour present 4 14.20% 1.646 3.422 (2.070) $2.50 three-quarters fine absent 5 52.70% 2.857 5.229 (2.058) $5.00 one-quarter sour present 6 14.30% 1.598 3.495 (2.154) $5.00 one-quarter fine absent 7 58.90% 3.009 5.431 (2.303) $5.00 three-quarters sour absent 8 16.20% 1.793 3.798 (2.198) $5.00 three-quarters fine present Note: In the between-subject design, respondents were randomized to evaluate one of the eight scenarios. The number of observations per scenario ranged from 111-113. For the within-subject design, 109 respondents ranked all eight scenarios based on the likelihood of saving/wasting. Superscript a indicates based on dichotomous choice question where 1=“Pour out the milk” and 0=“Drink the milk”; b indicates based on 5-point scale response where 1=“Definitely drink” and 5=“Definitely pour out”; c indicates that vignettes were ranked such that 1=Most likely to drink; 8=Most likely to pour out (waste). To further examine which factors are likely to lead consumers to pouring out the milk (i.e., food waste), we estimated an ordered logit regression for the 5-point likelihood of waste scale. Equation (3) presents the results, with significant coefficients bolded. Likelihood of Waste= Intercept – 0.045*Price – 0.159*Three-quarters Full– 1.575*Smells Fine + 0.007*Replacement Present. (3) As in study 1, the intercept term is replaced with four threshold parameters in the ordered logit model. Further, Three-quarters Full, Smells Fine, and Replacement Present are indicator variables; their effects are relative to One-quarter Full, Smells Slightly Sour, and Replacement Absent, respectively. Looking at equation (3), it is clear that the smell variable drives the waste decision in the case of milk. When milk smells fine (as opposed to slightly sour), consumers were significantly less likely to pour out the milk. The price, fullness, and replacement variables had no statistically significant impact on the waste decision. Table 6 extends equation (3) to explore the potential for heterogeneity in preferences for the vignette attributes. We again observed that the smell attribute dominated the waste decision, where consumers were generally less likely to pour out milk that smells fine. This finding did not hold across all consumers, though. Younger participants (aged 18–44) and Democrats were significantly more likely to throw out milk even when it smelled fine compared to those who were 65 years or older and non-Democrats, respectively. In addition, Democrats were less likely to pour out the milk when the carton is fuller relative to non-Democrats. Table 6 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Between-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 6 Ordered Logit Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Between-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a a −0.098 (0.209)† 0.04 (0.528) −2.55* (0.532) −0.319 (0.524) Female vs. Male 0.117 (0.532) −0.001 (0.114) 0.366 (0.287) −0.547 (0.288) 0.392 (0.287) Age 18–24 vs. 65 and older −0.425 (1.01) 0.229 (0.228) −0.499 (0.565) 1.793* (0.586) −0.455 (0.564) Age 25–34 vs. 65 and older −0.548 (0.966) 0.073 (0.208) −0.412 (0.522) 1.685* (0.530) 0.051 (0.522) Age 35–44 vs. 65 and older −1.089 (0.985) 0.244 (0.211) −0.253 (0.529) 1.762* (0.538) −0.441 (0.528) Age 45–54 vs. 65 and older −0.158 (0.956) 0.072 (0.206) −0.190 (0.517) 0.664 (0.530) −0.150 (0.516) Age 55–64 vs. 65 and older −0.690 (0.953) 0.231 (0.203) −0.088 (0.505) −0.197 (0.512) −0.401 (0.507) SNAP vs. Non-SNAP 0.409 (0.719) −0.032 (0.160) −0.237 (0.402) 0.413 (0.401) 0.426 (0.402) College degree vs. No degree −0.373 (0.580) 0.066 (0.126) 0.163 (0.314) 0.220 (0.317) −0.216 (0.314) Democrat vs. Other parties 0.825 (0.499) −0.179 (0.109) −0.587* (0.272) 0.556* (0.278) 0.229 (0.271) Obese vs. Non-obese 0.863 (0.617) −0.243 (0.130) 0.367 (0.323) −0.046 (0.323) −0.111 (0.322) Kids in household vs. No kids 0.954 (0.625) −0.091 (0.135) −0.366 (0.338) −0.236 (0.340) 0.004 (0.338) Medium vs. Low income −0.587 (0.642) 0.145 (0.138) 0.386 (0.346) −0.198 (0.347) 0.360 (0.345) High vs. Low income 0.348 (0.812) 0.007 (0.172) 0.013 (0.430) −0.845 (0.431) 0.538 (0.429) Number of Observations 894 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Superscript a indicates that the model has four threshold parameters not reported here; asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Results: Within-Subject Design Table 5 presents the summary statistics for the within-subject design. Similar to the between-subject design, we observe two clusters of means. The cluster that is more likely to drink the milk (rankings closer to 1) all share the characteristic that the milk smells fine, while the other cluster that is more likely to pour out the milk (rankings closer to 8) all share milk that smells slightly sour. We confirmed the impact of the smell variable on the decision to waste by running a regression on the waste rankings as a function of the vignette variables. Equation (4) shows that consumers are significantly less likely to throw out milk that smells fine relative to milk that smells slightly sour Waste Ranking =5.431+ 0.009*Price – 0.050*Three-quarters Full– 1.798*Smells Fine – 0.083*Replacement Present. (4) Using subject-specific regression estimates, we can examine heterogeneity in preferences by modeling the coefficient for each decision factor as a function of socio-demographic variables (see table 7). The table shows there was significant heterogeneity in the smell attribute. Here, males and younger participants (with the exception of the 45–54 year olds) were more likely to pour out milk when it smells fine relative to females and older participants (ages 65 and up), respectively. Within the replacement category, our results revealed that females and higher-income consumers were more likely to pour out the milk when a replacement was readily available. Though there was less variation in preferences based on price and fullness, we found that individuals who were 55–64 years old were less likely to waste when prices were high (relative to those 65 years and older) and that SNAP recipients were less likely to pour out the milk when the carton was fuller (relative to non-recipients). Table 7 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Within-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Table 7 Subject-Specific Regression Results with Socio-Demographic*Vignette Attribute Interactions (Study 2, Within-Subject Design) Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Interaction with … Intercept Price ¾ full vs. ¼ full Smells fine vs. Slightly sour Replacement present vs. absent n/a 4.739* (1.105) 0.293 (0.231) −0.238 (0.505) −2.167* (0.813) −0.270 (0.549) Female vs. Male 0.878 (0.605) −0.079 (0.126) −0.019 (0.276) −1.893* (0.445) 0.745* (0.301) Age 18–24 vs. 65 and older −0.484 (1.252) −0.093 (0.261) 0.024 (0.572) 1.845* (0.921) −0.206 (0.623) Age 25–34 vs. 65 and older −1.774 (1.033) 0.166 (0.216) 0.572 (0.472) 2.091* (0.760) −0.361 (0.514) Age 35–44 vs. 65 and older 0.341 (1.119) −0.173 (0.233) −0.200 (0.511) 1.792* (0.823) −0.978 (0.556) Age 45–54 vs. 65 and older 0.736 (1.078) −0.173 (0.225) −0.055 (0.493) 0.875 (0.793) −0.996 (0.536) Age 55–64 vs. 65 and older 1.185 (1.064) −0.525* (0.222) −0.148 (0.487) 2.327* (0.783) −0.609 (0.529) SNAP vs. Non-SNAP 1.418 (0.828) −0.324 (0.173) −0.896* (0.378) −0.019 (0.609) 0.512 (0.412) College degree vs. No degree 0.558 (0.618) −0.039 (0.129) −0.249 (0.282) −0.568 (0.454) −0.009 (0.307) Democrat vs. Other parties 0.134 (0.581) −0.047 (0.121) 0.162 (0.266) 0.061 (0.428) −0.139 (0.289) Obese vs. Non-obese 0.749 (0.791) −0.180 (0.165) 0.194 (0.362) −0.070 (0.582) −0.273 (0.393) Kids in household vs. No kids 0.146 (0.764) 0.002 (0.159) 0.326 (0.349) −0.254 (0.562) −0.376 (0.380) Medium vs. Low income −0.291 (0.765) −0.049 (0.160) 0.173 (0.350) 0.057 (0.562) 0.717 (0.380) High vs. Low income −0.655 (0.831) −0.011 (0.173) 0.190 (0.380) 0.373 (0.611) 0.825* (0.413) Number of Observations 109 109 109 109 109 R-Squared 0.16 0.14 0.13 0.27 0.15 Note: The term “vs.” is used for indicator variables to denote the reference category that was dropped from the regression. In the case of Smells fine vs. Slightly sour, for example, the variable is coded such that 1=Milk Smells Fine and 0=Milk Smells Slightly Sour. Asterisk * denotes significance at the 5% level; † indicates that standard errors are in parentheses. Discussion This study revealed that the keep/waste decision is not always a straightforward one. It is an economic decision, with both costs and benefits; the outcome depends on several contextual factors as well as individual-level characteristics. For waste reduction efforts to be effective, it is critical to understand the household decision process as well as the potential heterogeneities that exist across households. In the case of meal leftovers, respondents were generally less likely to waste the leftovers when the meal cost was high, when there were leftovers for a whole meal, when there were no future meal plans, and when the meal was prepared at home. Many of these relationships have a very obvious time component. Leftovers can save individuals time when there is enough for a whole meal and there are no future meal plans; further, when a meal is prepared at home, there is already a time cost for that meal (albeit a sunk cost) that people do not want to discount by throwing the leftovers out. With milk, the decision to waste was heavily impacted by food safety considerations as reflected in the smell of the product. Not surprisingly, milk that smelled slightly sour was more likely to be thrown out than milk that smelled fine—signaling individuals’ aversion to consuming a product they believe could make them or their family members ill (Graham-Rowe, Jessop, and Sparks 2014; Neff, Spiker, and Truant 2015). However, a subset of consumers opted to throw out the milk even when it smelled fine. This may be due to the expiration date information given in the vignette. In all scenarios, the milk was one day past its expiration, which may have caused some consumers to throw it out regardless of sensory properties. Our results have important implications for policymakers and the food manufacturing and retailing industries. For example, the milk results provide further evidence that more consumer education is needed on date labeling—particularly among those consumers who are likely to throw out milk even when it smells fine. Sensory properties are touted as an important signal for discerning product quality and safety, yet some consumers strictly rely on (potentially misleading) date labels. A new voluntary labeling initiative encourages food manufacturers and retailers to streamline the labeling terms used down to these two: “BEST If Used By” and “USE by” to signal product quality (e.g., the product may not taste as expected, but is safe to consume) and safety (e.g., product should not be consumed after this date), respectively (GMA 2017). Secondly, our findings suggest that there is heterogeneity in how consumers approach waste decisions. Consistent with past research (Quested, Ingle, and Parry 2013; Thyberg and Tonjes 2016), we found that younger individuals (18–44 years) were more likely to waste food than older consumers. Interestingly, this group was more likely to waste in contexts where a decision to waste was less intuitive—when the meal was higher-priced and when the milk smelled fine. One possible explanation for this may be that individuals with lower marginal productivities in meal preparation are likely to waste more (Lusk and Ellison 2017). It is likely that older individuals have acquired more skill in food preparation, and that retired individuals have more time for such activities. It may also be the case that younger consumers purchase more convenience-oriented items (frozen, microwavable, etc.) that are not well-suited for leftovers. We also found that political affiliation, sex, income level, SNAP participation, and children in the home may impact how a household approaches a waste decision, but the effects are not always straightforward or intuitive. For example, we cannot simply conclude that households with children waste less because there are more mouths to feed, or that Democrats waste less because they tend to be more environmentally-conscious. Rather, we find that consumer segments respond differently to different decision attributes. As educational campaigns are identified as a critical solution in reducing food waste (Rethink Food Waste through Economics and Data 2016), understanding the heterogeneity in waste behaviors can enable policymakers or other advocacy groups to better target educational efforts to the households most susceptible for high levels of food waste. Limitations and Areas for Future Research While this study is one of the first to examine specific food waste decisions, more work is needed to fully understand food waste at the household level. One limitation of the current study is that waste behaviors are self-reported, which may underestimate food waste (Food Loss and Waste Protocol 2016). However, as long as the self-report bias is constant across treatments, we can still identify the marginal effects of our decision attributes, even if our overall estimated level of waste is underestimated. Future studies should work to replicate these findings where behaviors are non-hypothetical and do not rely on self-reported waste data. Regarding the vignette design, this study was limited in that the outcomes for food were narrowly defined. Particularly with the milk vignette in study 2, the only outcomes proposed were for the individual to drink the milk or pour it down the drain. There was no option to use the milk for cooking, for example, which may have been more acceptable to individuals in some cases (e.g., when the milk smelled slightly sour) than drinking the milk. Footnotes 1 One notable exception is the WRAP program in the United Kingdom. This research group has undertaken rigorous household food waste audits (to learn more about the methodology used, see Quested, Ingle, and Parry 2013) and has developed an extensive consumer-facing education campaign, “Love Food Hate Waste”. For an evaluation of the impact of this campaign on household food waste behavior, see Quested and Ingle (2013). 2 We focus our discussion on attitudes toward and motivations for food waste. For more information on consumers’ knowledge and awareness of food waste, refer to Neff, Spiker, and Truant (2015); Stefan et al. (2013); and Parizeau, von Massow, and Martin (2015). For a discussion on waste-promoting and waste-reducing behaviors, see Neff, Spiker, and Truant (2015). 3 We present results for the waste scale outcome variable, rather than the dichotomous keep/waste variable because the scale allows for more variation across respondents. Logistic regression results for the dichotomous outcome variable are available upon request. 4 In total, five regressions are estimated: one for the intercept and one for each of the four vignette attributes. References Affognon H. , Mutungi C. , Sanginga P. , Borgemeister C . 2015 . Unpacking Postharvest Losses in Sub-Saharan Africa: A Meta-Analysis . World Development 66 : 49 – 68 . Google Scholar CrossRef Search ADS Aguinis H. , Bradley K.J . 2014 . Best Practice Recommendations for Designing and Implementing Experimental Vignette Methodology Studies . Organizational Research Methods 17 ( 4 ): 351 – 71 . http://dx.doi.org/10.1177/1094428114547952 Google Scholar CrossRef Search ADS Ajzen I. 1991 . The Theory of Planned Behavior . Organizational Behavior and Human Decision Processes 50 ( 2 ): 179 – 211 . http://dx.doi.org/10.1016/0749-5978(91)90020-T Google Scholar CrossRef Search ADS Alexander C.S. , Becker H.J . 1978 . The Use of Vignettes in Survey Research . Public Opinion Quarterly 42 ( 1 ): 93 – 104 . http://dx.doi.org/10.1086/268432 Google Scholar CrossRef Search ADS Becker G.S. 1965 . A Theory on the Allocation of Time . The Economic Journal 75 ( 299 ): 493 – 517 . http://dx.doi.org/10.2307/2228949 Google Scholar CrossRef Search ADS Bellemare M.F. , Çakir M. , Peterson H.H. , Novak L. , Rudi J . 2017 . On the Measurement of Food Waste . American Journal of Agricultural Economics . 99 ( 5 ): 1148 – 1158 . Google Scholar CrossRef Search ADS Buzby J.C. , Wells H.F. , Hyman J . 2014 . The Estimated Amount, Value, and Calories of Postharvest Food Losses at the Retail and Consumer Levels in the United States . Washington DC : U.S. Department of Agriculture, Economic Research Service . Daniel C. 2016 . Economic Constraints on Taste Formation and the True Cost of Healthy Eating . Social Science & Medicine 148 : 34 – 41 . Google Scholar CrossRef Search ADS Epstein D. , Mason A. , Manca A . 2008 . The Hospital Costs of Care for Stroke in Nine European Countries . Health Economics 17 ( S1 ): S21 – S31 . Google Scholar CrossRef Search ADS PubMed Food and Agriculture Organization of the United Nations . 2016 . FAO and Messe Düsseldorf Lead the SAVE FOOD – Global Initiative on Food Loss and Waste Reduction. Available at: http://www.fao.org/save-food/background/en/. Food Loss and Waste Protocol . 2016 . Food Loss and Waste Accounting and Reporting Standard: Version 1.0. Available at: http://www.wri.org/sites/default/files/FLW_Standard_final_2016.pdf. Food Waste Reduction Alliance . 2013 . About the Food Waste Reduction Alliance. Available at: http://www.foodwastealliance.org/about-us-page-2/. Graham-Rowe E. , Jessop D.C. , Sparks P . 2014 . Identifying Motivations and Barriers to Minimising Household Food Waste . Resources, Conservation and Recycling 84 : 15 – 23 . Google Scholar CrossRef Search ADS Graham-Rowe E. , Jessop D.C. , Sparks P. 2015 . Predicting Household Food Waste Reduction Using an Extended Theory of Planned Behavior . Resources, Conservation and Recycling 101 : 194 – 202 . Google Scholar CrossRef Search ADS Grocery Manufacturers Association . 2017 . Grocery Industry Launches New Initiative to Reduce Consumer Confusion on Product Date Labels. Available at: http://www.gmaonline.org/news-events/newsroom/grocery-industry-launches-new-initiative-to-reduce-consumer-confusion-on-pr/. Gunders D. 2012 . Wasted: How America is Losing Up to 40 Percent of Its Food from Farm to Fork to Landfill. Natural Resources Defense Council (NRDC) Issue Paper: 12-06-B. Available at: http://www.indianasna.org/content/indianasna/documents/NRDC_Wasted_Food_Report.pdf. Hainmueller J. , Hangartner D. , Yamamoto T . 2015 . Validating Vignette and Conjoint Survey Experiments against Real-World Behavior . Proceedings of the National Academies of Science 112 ( 8 ): 2395 – 400 . http://dx.doi.org/10.1073/pnas.1416587112 Google Scholar CrossRef Search ADS Hodges R.J. , Buzby J.C. , Bennett B . 2011 . Postharvest Losses and Waste in Developed and Less Developed Countries: Opportunities to Improve Resource Use . Journal of Agricultural Science 149 : 37 – 45 . Google Scholar CrossRef Search ADS Kapteyn A. , Smith J.P. , van Soest A . 2007 . Vignettes and Self-Reports of Work Disability in the United States and the Netherlands . The American Economic Review 97 ( 1 ): 461 – 73 . http://dx.doi.org/10.1257/aer.97.1.461 Google Scholar CrossRef Search ADS Kristensen N. , Johansson E . 2008 . New Evidence on Cross-Country Differences in Job Satisfaction Using Anchoring Vignettes . Labour Economics 15 ( 1 ): 96 – 117 . http://dx.doi.org/10.1016/j.labeco.2006.11.001 Google Scholar CrossRef Search ADS Lazell J. 2016 . Consumer Food Waste Behaviour in Universities: Sharing as a Means of Prevention . Journal of Consumer Behavior 15 ( 5 ): 430 – 39 . http://dx.doi.org/10.1002/cb.1581 Google Scholar CrossRef Search ADS Lusk J.L. , Ellison B . 2017 . A Note on Modelling Household Food Waste Behaviour . Applied Economics Letters 24 ( 16 ): 1199 – 202 . http://dx.doi.org/10.1080/13504851.2016.1265070 Google Scholar CrossRef Search ADS Lusk J.L. , McCluskey J . Forthcoming. Understanding the Impacts of Food Consumer Choice and Food Policy Outcomes . Applied Economic Perspectives and Policy . Mondéjar-Jiménez J.-A. , Ferrari G. , Secondi L. , Principato L . 2016 . From the Table to Waste: An Exploratory Study on Behavior Towards Food Waste of Spanish and Italian Youths . Journal of Cleaner Production 138 ( 1 ): 8 – 18 . Google Scholar CrossRef Search ADS Neff R.A. , Spiker M.L. , Truant P.L . 2015 . Wasted Food: U.S. Consumers’ Reported Awareness, Attitudes, and Behaviors . PLoS ONE 10 ( 6 ): e0127881 . Google Scholar CrossRef Search ADS PubMed Newsome R. , Balestrini C.G. , Baum M.D. , Corby J. , Fisher W. , Goodburn K. , Labuza T.P. , et al. 2014 . Applications and Perceptions of Date Labeling of Food . Comprehensive Reviews in Food Science and Food Safety 13 ( 4 ): 745 – 69 . Google Scholar CrossRef Search ADS Parizeau K. , von Massow M. , Martin R . 2015 . Household-level Dynamics of Food Waste Production and Related Beliefs, Attitudes, and Behaviors in Guelph, Ontario . Waste Management 35 : 207 – 17 . Google Scholar CrossRef Search ADS PubMed Rethink Food Waste through Economics and Data . 2016 . A Roadmap to Reduce U.S. Food Waste by 20 Percent. Available at: http://www.refed.com/downloads/ReFED_Report_2016.pdf. Qi D. , Roe B.E . 2016 . Household Food Waste: Multivariate Regression and Principal Components Analyses of Awareness and Attitudes among U.S. Consumers . PLoS ONE 11 ( 7 ): e0159250 . Google Scholar CrossRef Search ADS PubMed Quested T. , Ingle R . 2013 . West London Food Waste Prevention Campaign Evaluation Report. WRAP Research Report, Project Code: CFP104-030. Available at: http://www.wrap.org.uk/sites/files/wrap/West%20London%20Food%20Waste%20Campaign%20Evaluation%20Report_1.pdf. Quested T. , Ingle R. , Parry A . 2013 . Household Food and Drink Waste in the United Kingdom 2012. WRAP Research Report, Project Code: CFP102. Available at: http://www.wrap.org.uk/sites/files/wrap/hhfdw-2012-main.pdf.pdf. Quested T.E. , Marsh E. , Stunell D. , Parry A.D . 2013 . Spaghetti Soup: The Complex World of Food Waste Behaviors . Resources, Conservation and Recycling 79 : 43 – 51 . Google Scholar CrossRef Search ADS Refsgaard K. , Magnussen K . 2009 . Household Behaviour and Attitudes with Respect to Recycling Food Waste – Experiences from Focus Groups . Journal of Environmental Management 90 : 760 – 71 . Google Scholar CrossRef Search ADS PubMed Stancu V. , Haugaard P. , Lähteenmäki L . 2016 . Determinants of Consumer Food Waste Behavior: Two Routes to Food Waste . Appetite 96 : 7 – 17 . Google Scholar CrossRef Search ADS PubMed Stefan V. , van Herpen E. , Tudoran A.A. , Lähteenmäki L . 2013 . Avoiding Food Waste by Romanian Consumers: The Importance of Planning and Shopping Routines . Food Quality and Preference 28 : 375 – 81 . Google Scholar CrossRef Search ADS Thyberg K.L. , Tonjes D.J . 2016 . Drivers of Food Waste and their Implications for Sustainable Policy Development . Resources, Conservation and Recycling 106 : 110 – 23 . Google Scholar CrossRef Search ADS U.S. Department of Agriculture . 2013 . USDA and EPA Launch U.S. Food Waste Challenge. Washington DC: USDA Press Release Number 0112.13. Available at: http://www.usda.gov/wps/portal/usda/usdahome?contentidonly=true&contentid=2013/06/0112.xml. Waste and Resources Action Programme . 2017 . Food Waste Reduction. Available at: http://www.wrap.org.uk/food-waste-reduction. Williams H. , Wikström F. , Otterbring T. , Löfgren M. , Gustafsson A . 2012 . Reasons for Household Food Waste with Special Attention to Packaging . Journal of Cleaner Production 24 : 141 – 8 . Google Scholar CrossRef Search ADS Wilson N.L.W. , Rickard B.J. , Saputo R. , Ho S.-T . 2017 . Food Waste: The Role of Date Labels, Package Size, and Product Category . Food Quality and Preference 55 : 35 – 44 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Journal

Applied Economic Perspectives and PolicyOxford University Press

Published: Feb 20, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off