This article examines perceptions of low-income consumers receiving government assistance and the choices they make, showing that this group is viewed differently than those with more resources, even when making identical choices. A series of five experiments reveal that ethical purchases polarize moral judgments: whereas individuals receiving government assistance are perceived as less moral when choosing ethical (vs. conventional) products, income earners, particularly high-income individuals, are perceived as more moral for making the identical choice. Price is a central component of this effect because equating the cost of ethical and conventional goods provides those receiving government assistance some protection against harsh moral judgments when choosing ethically. Moreover, earning one’s income drives perceptions of deservingness, or the right to spend as one desires. Those who receive assistance via taxpayer dollars are under greater scrutiny (frequently resulting in harsher moral judgments) by others. In addition to influencing perceptions of individual consumers, the results demonstrate that such attributions extend to groups who make ethical choices on others’ behalf, and that these attributions have real monetary consequences for nonprofit organizations.
“The point is taken: You’re poor. And out of the goodness of our collective hearts we have given you a food stamp. The least you could do is … spend it on healthy, fresh, nutritious food.”
“They [welfare recipients] get wild salmon in the East Village … I’ve always said poor-people-salmon should come in a can from an outer borough!”
—Jon Stewart, The Daily Show ( Allon 2014 )
Despite being the largest economy on the planet, income and wealth inequality are at historic highs in the United States. In 2013 the median wealth of the nation’s upper-income households (i.e., those with size-adjusted incomes more than twice the median) was almost seven times larger than the median wealth of middle-income households (i.e., size-adjusted income between two-thirds and twice the median) and nearly 70 times larger than the median wealth of lower-income households (i.e., size-adjusted income less than two-thirds the median), representing the widest gaps since data collection began by the Federal Reserve 30 years ago ( Fry and Kochhar 2014 ). In addition to growing disparities across income groups, most of society’s economic gains following the Great Recession in 2008 have gone to the elite. For instance, the top 1% captured 95% of the income gains in the first three years of recovery ( Saez 2013 ). Others estimate that the top 1% hold nearly 50% of the wealth, a number at its highest since the Great Depression ( Keister 2000 ; Norton and Ariely 2011 ; Wolff 2002 ). In contrast to the wealthy or even middle class, however, life for the poor has worsened ( Shaefer and Edin 2012 ). In 2013 there were 45.3 million people living in poverty ( US Census Bureau 2014 ), suggesting that many Americans require financial support to satisfy basic needs. Indeed, nearly 50% of the population lives in a household where at least one member receives some type of direct government assistance (e.g., nutritional assistance, Social Security; Izzo 2012 ).
Given that many Americans (and indeed people in every country) are facing some degree of economic hardship, it is important to understand the unique marketplace challenges facing low-income groups, and how the choices they make are viewed by broader society and its policymakers. Specifically, we contrast perceptions of the choices made by low-income consumers receiving government assistance with those earning modest and high incomes. We develop and test a theory of consumer reactions to others’ choices, demonstrating that consumers are frequently viewed differently by others for making identical choices. Because individuals receiving government assistance are funded through taxpayer dollars, we propose that their consumption choices receive extra scrutiny because consumers view these individuals as potentially misusing “their” money.
Of course, not all marketplace choices are created equal—very few people would contest someone meeting basic needs for food, clothing, and shelter. However, the amount of money spent on food purchases, for example, has an almost infinite range (i.e., consumers can make very inexpensive choices or very costly ones). Further, these options differ in the degree to which they are environmentally friendly, sustainable, and prosocial. Here, we largely focus our investigation on making ethical choices, which involves choosing products that do not harm or exploit humans, animals, or the natural environment ( Crane 2001 ; Doane 2001 ). For example, instead of choosing conventional foods, consumers can choose organic foods grown by local farmers. Instead of choosing a regular gas-engine vehicle, individuals can reduce their carbon emissions by purchasing an environmentally friendly hybrid or electric vehicle. Across a broad spectrum of industries, there is a general upward trend in choosing ethical products like these that promote more sustainable, socially conscious lifestyles ( Dossey 2010 ; French and Rogers 2010 ).
Many ethical products benefit not only the self (e.g., they are healthier and safer), but also others both directly (e.g., fair wages, better working conditions) and indirectly (e.g., fewer pollutants in the air and water). Because it benefits others, we propose that ethical consumption will be viewed as prosocial ( Batson 1998 ), and thus inherently moral. If this is true, by extension, consumers of such products should be viewed as more moral for buying them ( Belk 1988 ). However, ethical goods possess an extra feature (in addition to being prosocial) that may limit widespread access or acceptance: a higher price . Ethical goods are generally more expensive than conventional options and are perceived as such ( De Pelsmacker, Driesen, and Rayp 2005 ; Hughner et al. 2007 ; McGoldrick and Freestone 2008 ; Padel and Foster 2005 ; Trudel and Cotte 2009 ); thus consumers must be able and willing to pay a higher price in order to engage in ethical consumption. As a result of the price premium associated with ethical products, we propose that identical choices will lead to different attributions as a function of income characteristics. Drawing on attribution theory ( Heider 1958 ; Kelley 1967 ) and equity theory ( Adams 1965 ; Huseman, Hatfield, and Miles 1987 ; Walster, Berscheid, and Walster 1973 ), our central prediction is that ethical consumption choices will polarize moral judgments: whereas individuals earning high incomes will be perceived as more moral for choosing costly, ethical (vs. more affordable, conventional) goods, those in the lowest income bracket receiving government assistance will be perceived as less moral because they are seen as “undeserving” of the right to make such choices.
The current work responds to calls for more research examining underrepresented, vulnerable groups and their experiences in the marketplace ( Mick 2006 ). In doing so, we contribute to the literature in a number of ways. Foremost, we demonstrate that moral judgments directed toward consumers making (identical) ethical choices depend on the nature of the target’s income. Consistent with our conceptualization, we identify cost as a central component of this effect because equating the cost of ethical and conventional goods differentially affects consumers depending on their income. Importantly, the ability to make ethical choices is driven by perceptions of deservingness (i.e., the freedom of choice), which are enhanced primarily by earning income. Of course, there are other possible routes to increasing perceived deservingness (e.g., greater effort), but we focus predominantly on earning income in the current research. Our final experiment reveals that negative attributions extend to groups providing ethical support for those deemed “undeserving,” and that these attributions have real monetary consequences for nonprofit organizations. Hence our research also speaks to public views on the perceived appropriateness of different expenditures by different income groups, which can inform public policies aimed at alleviating poverty.
Income and Consumption Choices
Although definitions of poverty and methodologies for measuring it vary ( Chandy and Smith 2014 ), one study estimates that “the number of households [in the United States] living on $2 or less in income per person, per day, in a given month increased from about 636,000 in 1996 to about 1.46 million households in early 2011, a percentage growth of 130 percent” ( Shaefer and Edin 2012 , 2). Living in poverty presents numerous challenges; for instance, cash-strapped individuals face uncertainty surrounding their next meal, paying basic household bills, and receiving adequate health care. Economic strain has been shown to tax cognitive resources, leaving low-income consumers less able to cope with life’s demands ( Mani et al. 2013 ). Thus a minor financial setback for more affluent consumers (e.g., car trouble or an acute illness) can be devastating for low-income consumers.
In addition to material obstacles, low-income consumers are confronted with a cultural belief that everyone, even those from the humblest of beginnings, is able to climb the socioeconomic status (SES) ladder to prosperity. Indeed, a pervasive American ideology is that hard work leads to success and that lack of success is caused by the moral failings of self-indulgence and/or a lack of self-discipline ( Katz and Hass 1988 ; Mirels and Garrett 1971 ; Shipler 2005 ; Weber 1958 ). Not working hard enough is considered an ethical lapse. This belief leads many Americans to overestimate the importance of personal or “internal” factors (e.g., laziness, low effort and motivation) relative to environmental or “external” factors (e.g., discrimination, unequal access to resources, unfavorable government policies) in their moral attributions of low-income groups ( Cozzarelli, Wilkinson, and Tagler 2001 ; Gilbert and Jones 1986 ; Lott 2002 ; Ross 1977 ), when in reality economic mobility is declining ( Athreya and Romero 2012 ). Attributions about the causes of poverty are important because they influence whether people support or oppose income redistribution via government spending on social welfare programs. For example, equity theory (i.e., the belief that people’s rewards should be proportional to their effort; Adams 1965 ) predicts that people are willing to support the poor, but only as long as the poor are perceived to be industrious ( Fong 2001 ).
We should note an important distinction between the absolute level of income (i.e., low vs. high), and the source of that income (i.e., whether it is earned or unearned) because this has also been shown to affect moral attributions. Earning low wages from a job is not the same as receiving low wages from the government; the latter is likely to be judged especially harshly. Fiske et al. (1999) found evidence that across a total of 17 stereotyped groups (e.g., rich people, businesswomen, southerners, etc.), adults receiving government assistance (i.e., “welfare” recipients) were found to be the only group that respondents both disliked and disrespected. Subsequent work has revealed that welfare recipients are perceived as low in both competence and warmth, which elicits feelings of contempt and derogation ( Cuddy, Fiske, and Glick 2007 ; Fiske et al. 2002 ). Our own work compares perceptions of low-income consumers receiving government assistance with perceptions of low-income consumers who earn their money via employment (holding the total amount of income constant).
Whether relatively rich or poor (and receiving government benefits or not), individuals of all income groups have consumption needs. In the next section, we discuss how specific consumption choices signal information to others.
Consumption Choices and Moral Attributions
Consumption choices do not exist in a vacuum; instead, they signal information (consciously or not) about the person doing the choosing ( Belk 1988 ; Berger and Heath 2007 ). For example, ordering a supersized beverage can signal status ( Dubois, Rucker, and Galinsky 2012 ), redeeming coupons can signal stinginess ( Ashworth, Darke, and Schaller 2005 ), and wearing red sneakers in the boardroom can signal competence ( Bellezza, Gino, and Keinan 2014 ). Likewise, choosing an ethical product over a conventional product should signal information about a consumer’s moral character.
We propose that a consumer’s choices lead others to draw inferences about him or her because choices are generally volitional. Attribution theory posits that people attach meaning to others’ behavior in an effort to arrive at causal explanations for events ( Calder and Burnkrant 1977 ; Heider 1958 ; Kelley 1967 ), and a key component of attribution theory is controllability ( Jones and Davis 1965 ). Similarly, equity theory predicts that one’s efforts matter, and the same person can be evaluated differently based on his or her actions (or at least intentions; Fong 2001 ). Perceptions of controllability are important to the current research because they link to inferences regarding personal responsibility, moral judgments, and moral emotions ( Weiner 1979 , 2000 ). Therefore, if product selections are perceived as freely chosen and intentional, as opposed to coerced or accidental, we propose that they convey valuable information about people, and their moral character specifically, to others. However, what we propose here, and the central contribution of this work, is that the same choice may trigger different judgments in the eyes of others, depending on the characteristics of the chooser.
Why Income Information Should Affect Moral Attributions
When individuals draw moral inferences about others based on their choices, they may value different qualities depending on the chooser’s income. For example, individuals may believe those with ample means have a duty or obligation to provide for those in less privileged social positions by being wise with their money, making donations, and behaving prosocially (i.e., “noblesse oblige”; Ostrower 1995 ); thus when wealthy consumers practice such behaviors, they are viewed favorably. Conversely, low-income individuals in general should be seen as more moral when they make thrifty choices because resisting the urge to spend money is viewed as a virtuous behavior ( Kivetz and Keinan 2006 ), and so choosing affordable options (particularly when spending “taxpayer” money) should cue positive attributions of being financially responsible.
These two qualities (prosociality and thrift) present an interesting question: How might consumers from different income groups be perceived when these qualities are at odds with each other (i.e., prosocial, yet costly)? The current research focuses on ethical products, which we propose are characterized as both prosocial and costly. Ethical choices present individuals with a choice between a conventional option and an “elevated” alternative ( Doane 2001 ). When a target consumer makes an ethical choice, we propose that these elements will be weighted differently as a function of his or her income characteristics. On one end, low-income individuals who receive government assistance may be perceived as immoral when making ethical choices because the cost element is magnified (i.e., they do not deserve to spend taxpayer dollars on so-called extras). Conversely, relatively wealthy consumers who earn their income may be perceived as more moral for the same choice because the prosocial element outshines the cost element (i.e., they deserve to spend their own money in any way they choose).
We propose that a central driver of the denigration toward low-income consumers receiving government assistance is outsiders’ views that they are less “deserving” of ethical choice, relative to those who earn their income. Of course, “deservingness” could lead to many different perceptions of what people can and cannot do with their money (e.g., choosing to invest in the stock market), beyond the current context of ethical behavior. We focus, specifically, on “deserving the right to spend money as one chooses” and its relationship with ethical choice because of the heightened cost associated with choosing ethically. We propose that spending money received from taxpayer funds may come with an expectation of how that money is spent; namely, it “should” be spent on low-cost necessities.
Deserving the freedom of choice is central to our framework because it helps explain why different income groups may be perceived differently for identical choices. According to equity theory ( Adams 1965 ; Huseman et al. 1987 ; Walster et al. 1973 ), individuals strive for a fair balance between their own inputs (e.g., hard work, effort) and outputs (e.g., salary, recognition) relative to others’ perceived inputs and outputs. Generally speaking, individuals who earn their money through paid employment are likely to be seen as receiving outputs commensurate with inputs—hence they deserve to spend their money any way they would like. Conversely, low-income individuals who receive government aid may be seen as putting forth insufficient effort to earn money; instead, they are profiting from others’ inputs. Internal attributions of laziness may dominate when their income is seen as belonging to the taxpayers who originally earned it ( Cozzarelli et al. 2001 ). As a result, aid recipients may be evaluated as less deserving of the benefits of costly, ethical products. Such an imbalance sparks feelings of injustice ( Adams 1965 ), which could lead to efforts at restoring equity via moral condemnation. We examine these questions empirically here.
OVERVIEW OF THE PRESENT RESEARCH
Across five experiments, we contrast perceptions of the choices made by low-income consumers receiving government assistance with those earning modest and high incomes. Experiment 1 provides initial evidence of an attribution reversal in the domain of organic food. While individuals earning high incomes are perceived as more moral for choosing organic (vs. conventional) food, low-income individuals receiving government assistance are perceived as less moral for the same choice. Experiment 2 equates the price of organic food and conventional food, allowing us to separate the cost component from the prosocial component of ethical consumption. Experiment 3 broadens the investigation into the domain of green vehicles and highlights the importance of perceived deservingness in driving the attribution reversal. Individuals who have not earned their money may be punished for seemingly improper resource management because they do not deserve to make costly, ethical choices. Further exploring the deservingness mechanism, experiment 4 examines the effects of other sources of unearned income (besides taxpayer dollars) on moral judgments. Lastly, experiment 5 extends our conceptualization from individuals to groups who provide ethical products to low-income consumers receiving government assistance, highlighting real monetary consequences.
EXPERIMENT 1: ORGANIC FOOD
The first experiment tests our central prediction in the domain of organic food. Prior research indicates that the production and consumption of organic food is believed to be an ethical, moral, and prosocial decision ( Harper and Makatouni 2002 ; Mazar and Zhong 2010 ; van Doorn and Verhoef 2011 ). Even though the “organic” label technically refers to a production process ( US Department of Agriculture 2012 ), consumers perceive organic food to have fewer calories than nonorganic food ( Schuldt and Schwarz 2010 ). This “health halo” may be one reason why organic food is considered to be a moral choice. In addition, organic food is associated with a higher price tag ( Hughner et al. 2007 ; Padel and Foster 2005 ), perhaps leading to the perception of organic food as a choice appropriate for higher social classes ( Harper and Makatouni 2002 ).
We predicted that for low-income consumers receiving government assistance, the perceived cost of organic food versus conventional food would overshadow the prosocial goodness, resulting in unfavorable moral evaluations. We predicted the opposite pattern for relatively wealthy consumers: the prosocial goodness would overshadow the perceived price premium, resulting in more favorable moral evaluations.
Participants and Procedure
A total of 135 students (67% female; Mage = 20.39, SDage = 1.39) from a large midwestern university completed the study in exchange for payment. The experiment followed a 2 (Annual Income: Welfare vs. $85,000) × 2 (Organic Label: Yes vs. No) between-subjects design. SES is “a composite measure that typically incorporates economic status, measured by income; social status, measured by education; and work status, measured by occupation” ( Dutton and Levine 1989 , 30). These three variables are highly interrelated but not completely overlapping. Experiment 1 focuses primarily on economic status, but we introduce work status (i.e., earning vs. not earning money) more explicitly in subsequent experiments. Because SES is often conceptualized as a continuum with poverty and high income at the extremes ( Adler et al. 1994 ), we decided to test for moral judgment effects at these opposing points. Welfare—the low point on the SES continuum—is the most commonly known form of government income assistance, so including this level enabled us to test the role of earned versus unearned income. We selected our “high” income condition because the top 25% of US households report income greater than $85,000 ( US Census Bureau 2011 ).
Participants were told that the researchers were interested in how accurately people can make judgments of other people when they only know a small amount of information about them. Utilizing a grocery list paradigm inspired by Haire (1950) , participants were presented with a list featuring eight items that belonged to a target individual. A subset of three foods (carrots, 2% milk, and cereal) was labeled organic in the organic label condition (e.g., “organic carrots”). The remaining five foods were held constant across conditions (chicken thighs, sliced bread, baking powder, ground coffee, and eggs) to mask the experiment’s premise. A single statement above the shopping list indicated that the list belonged to someone who had either been receiving welfare for the past year or earning $85,000 a year. Although we use “welfare” terminology in the set of experiments presented here, phrasing the income manipulation as “receiving government benefits” yielded similar results in other data.
Participants then evaluated the target individual along several dimensions using 7 point semantic differential scales adapted from Stein and Nemeroff (1995) . The key measure was a morality index (cruel/kindhearted, immoral/moral, uncaring/caring, and unethical/ethical; α = .85). To disguise the experiment’s purpose, the key moral qualities were embedded among filler items (e.g., practical/idealistic) that were not analyzed and were empirically distinct (i.e., amoral). In order to demonstrate that organic food has moral implications above and beyond health, we assessed the perceived healthiness of the list. This was important because consumers judge others who choose healthy food as more moral than those who choose unhealthy food ( Stein and Nemeroff 1995 ). If organic food choice is only perceived as (im)moral because of its perceived healthiness, the relationship between income and morality should disappear when we control for the healthiness of the food. As such, participants rated the overall health value of the grocery list using 7 point scales where 1 = Not at all nutritious, wholesome, fattening, good for you and 7 = Very nutritious, wholesome, fattening, good for you. A perceived health index was created by averaging responses to these four items, with the fattening item reverse-coded (α = .89). Finally, participants completed demographic items including political orientation and estimates of their own family income. These last two measures did not differ across experimental cells ( p’ s ≥ .35) or influence key results in any of the presented experiments. Subsequent experiments asked participants whether they themselves have ever received government benefits and whether their friends and immediate family members have. Perhaps because of low power, these variables did not have significant effects on our analyses of interest, so we elected not to pursue them in the present article. Our focus is on perceptions of others, but future research might explore intrapersonal qualities of the perceiver.
Results and Discussion
A 2 (Income) × 2 (Organic Label) analysis of covariance (ANCOVA) was conducted with the morality index as the dependent variable and the perceived health index as the covariate. Perceived health was significantly correlated with morality ( r (133) = .23, p < .01), so controlling for health allowed us to test the purity of the morality effect, above and beyond positive health perceptions. The results revealed only a significant interaction between Income and Organic Label, suggesting that moral judgments of a consumer purchasing organic food versus conventional food depend on the nature of the target’s income ( F (1, 130) = 8.78, p < .01, = .06; figure 1 ). Note that the interaction remains significant without the covariate ( F (1, 131) = 7.38, p < .01, = .05).
Follow-up tests on the adjusted means showed that welfare recipients were perceived as marginally less moral when they purchased organic food versus nonorganic food ( M = 4.39, standard error [SE] = .14 vs. M = 4.78, SE = .14; p = .06, d = .48). Conversely, targets earning $85,000 a year were perceived as significantly more moral when they purchased organic food versus nonorganic food ( M = 5.03, SE = .14 vs. M = 4.60, SE = .14; p < .05, d = .53). Thus the very same action is judged as either moral or immoral depending on who made the choice. Within the organic condition, targets earning $85,000 a year were viewed as significantly more moral than welfare recipients ( p < .01, d = .79)—a difference in moral perceptions that does not hold when both groups purchase nonorganic food ( p = .36). Other attributions may exist for the conventional list (e.g., frugality), which is one reason why we do not see a difference between the income groups. The design of this experiment also provides a conservative test of our predictions because the conventional grocery list did not highlight the alternative choice of buying organic food.
In sum, while relatively poor individuals receiving government assistance are viewed as less moral when purchasing costly, ethical food products, relatively wealthy consumers are viewed as more moral. Our results hold when controlling for health perceptions, providing evidence that mere differences in the perceived nutritional content are not driving the results, and showing instead that organic food also results in other (moral) attributions.
EXPERIMENT 2: ORGANIC FOOD AT A DISCOUNTED PRICE
We have proposed that two key dimensions of ethical consumption are cost and prosociality. In experiment 2, we test our conceptualization further by creating equivalence in the cost of the ethical (vs. conventional) products. We elected to hold price constant for theoretical reasons, more so than practical reasons. Ethical products do generally cost more than conventional options ( De Pelsmacker et al. 2005 ; Hughner et al. 2007 ; McGoldrick and Freestone 2008 ; Padel and Foster 2005 ; Trudel and Cotte 2009 ). Nevertheless, removing cost from the equation allows us to examine if, as expected, the derogation that those receiving government assistance experience can be mitigated if they purchase ethical goods but do not spend more money in doing so. In fact, such individuals may even receive the same boost as their wealthier counterparts experience in such a case because they can still benefit from the prosocial element of such goods.
An additional goal of experiment 2 was enhancing external validity in two ways. First, we used a single item at a lower price point (vs. an entire grocery list) to represent a more modest expenditure. Second, we added two new income conditions. Because the welfare recipients and $85,000 earners in experiment 1 differed in both income level (low vs. high) and source (unearned vs. earned), we included a “low-income earner” condition. Note that low-income earners are distinct in that they may share favorable attributions with both low-income consumers receiving government assistance (i.e., “saving money on a tight budget is honorable”) and high-income earners (i.e., “spending hard-earned money on ethical consumption is honorable”). The second new condition was based on the fact that many families receiving government support (e.g., via the Supplemental Nutrition Assistance Program, commonly known as SNAP) do so in addition to earning money through paid employment. In the United States, individuals can earn close to $12,000 annually and still be eligible for federal aid programs ( US Department of Health and Human Services 2015 ). Including a group of “partial” income earners in our experiment allows us to test whether earning even a small amount of income can overcome the stigma associated with receiving government benefits, or if the acceptance of any government benefits overshadows earned income.
Participants and Procedure
A total of 608 adults (44% female; Mage = 33.84, SDage = 11.44) from Amazon Mechanical Turk (MTurk; Paolacci, Chandler, and Ipeirotis 2010) completed the experiment in exchange for a $.50 payment. Participants were randomly assigned to one of 12 conditions in a 4 (Annual Income: $12,000 in welfare benefits vs. $4,000 in welfare benefits + $8,000 earned vs. $12,000 earned vs. $85,000 earned) × 3 (Price: Lower-Priced Nonorganic, Higher-Priced Organic, Same Price Organic–Discounted) between-subjects design. For simplicity, we refer to the “$4,000 in welfare benefits + $8,000 earned” as “majority earners.” We chose $12,000 to represent a relatively low income level because the poverty guideline in the United States for one individual is $11,770 ( US Department of Health and Human Services 2015 ). In other words, this is the level of income used to determine financial eligibility for federal aid programs.
All participants read about a person named Anna (in her 30s) and her recent trip to the grocery store. After receiving information about her income (e.g., Anna “receives $12,000 a year in welfare benefits,” “earns $12,000 a year from her job”), participants read the following scenario:
Anna is going grocery shopping and wants to buy a container of precut pineapple. While looking at the produce display, she is presented with two options: organic pineapple or regular (nonorganic) pineapple. The organic pineapple is priced at $5.99, while the regular pineapple is priced at $3.49.
Everyone was exposed to these same two price points, which were calibrated based on advertisements from national retailers. Participants randomly assigned to the Lower-Priced Nonorganic condition were then told, “After some thought, Anna decides to purchase the regular pineapple for $3.49.” Those in the Higher-Priced Organic condition were told, “After some thought, Anna decides to purchase the organic pineapple for $5.99.” Those in the Same Price Organic–Discounted condition read, “Before making a decision, Anna notices that the organic pineapple is actually on sale for $3.49 this week. Anna decides to purchase the organic pineapple for $3.49.” After reading about Anna’s decision, participants completed the morality (α = .93) and perceived health (α = .83) indices from experiment 1.
Results and Discussion
A 4 (Income) × 3 (Price) ANCOVA was conducted with the morality index as the dependent variable and the perceived health index as the covariate. As in experiment 1, perceived health was significantly correlated with morality ( r (606) = .25, p < .001), so controlling for health allowed us to test for differences in moral judgments above and beyond positive health perceptions. A significant main effect for price ( F (2, 595) = 9.69, p < .001, = .03) was qualified by a significant interaction ( F (6, 595) = 4.81, p < .001, = .05; figure 2 ). Note that the interaction remains significant and is slightly stronger without the covariate ( F (6, 596) = 5.22, p < .001, = .05).
The first set of analyses focused on differences within each income condition. Follow-up tests on the adjusted means showed that $12,000 welfare recipients were perceived as significantly less moral when they purchased higher-priced organic food versus lower-priced nonorganic food ( M = 4.45, SE = .16 vs. M = 5.59, SE = .15; p < .001, d = 1.06). When the price of organic and nonorganic food was equivalent, however, there was no difference in moral judgments directed toward welfare recipients. Specifically, purchasing lower-priced nonorganic food was evaluated similarly to purchasing organic food for the same price ( M = 5.31, SE = .16; p = .20), with the latter choice seen as significantly more moral than purchasing higher-priced organic food ( p < .001; d = .80). This “protective effect” of purchasing discounted organic food is also present among majority earners, who show an identical pattern to welfare recipients. Compared to majority earners who purchased higher-priced organic food ( M = 4.77, SE = .15), majority earners who chose lower-priced nonorganic food ( M = 5.40, SE = .15; p < .01, d = .58) or discounted organic food ( M = 5.45, SE = .16; p < .01, d = .63) were judged as significantly more moral.
Different patterns emerge for the two income earning groups. Among $12,000 earners, moral judgments did not differ as a function of purchase decision (lower-priced nonorganic food: M = 5.20, SE = .16; higher-priced organic food: M = 4.89, SE = .15; same price organic food: M = 5.19, SE = .15; p’ s ≥ .15). Unlike the other three income groups, $85,000 earners were perceived as significantly more moral when they purchased higher-priced organic food versus lower-priced nonorganic food ( M = 5.31, SE = .16 vs. M = 4.86, SE = .15; p < .05, d = .41). Choosing discounted organic food ( M = 5.16, SE = .15) over nonorganic food yielded a directionally favorable, but nonsignificant difference ( p = .17).
The next analysis compared judgments within each of the three price conditions. For targets purchasing nonorganic food, both $12,000 welfare recipients and majority earners were perceived as significantly more moral than $85,000 earners ( p’ s ≤ .01, both d’ s ≥ .49). The significant difference between welfare recipients and $85,000 earners for relatively neutral behavior is likely because the scenario highlighted the alternative of buying organic food (vs. in experiment 1 where the alternative was not made salient for the conventional grocery list). The two groups receiving government assistance did not differ from each other ( p = .37); nor did the two 100% earning conditions ( p = .12). The linear trend across income conditions (welfare – partial earners – $12,000 earned – $85,000 earned) was also significant ( t (595) = 3.49, p < .001). For targets purchasing premium-priced organic, we see the opposite pattern such that both $12,000 and $85,000 earners were perceived as significantly more moral than $12,000 welfare recipients ( p’ s < .05, both d’ s ≥ .41). The two groups receiving government assistance did not differ from each other for this choice ( p = .13). The results also revealed that $85,000 earners were marginally more moral than the $12,000 earners ( p = .06, d = .38), suggesting an additive effect for earning a higher level of income. Again, the linear trend across income levels was significant ( t (595) = 3.85, p < .001). Lastly, the four income groups were viewed similarly when they purchased discounted organic food ( p’ s ≥ .18).
In sum, the results from experiment 2 replicate and extend the results from experiment 1. First, we observe the same attribution reversal for $12,000 welfare recipients and $85,000 earners such that the former is morally derogated and the latter is acclaimed when choosing expensive ethical goods over more affordable conventional goods. Moreover, including two additional income groups provided new insight. The results suggest that partially earning one’s income is insufficient to distinguish oneself from someone who receives all of his or her income via government assistance. Despite earning twice as much income than the size of their benefit check (i.e., $8,000 vs. $4,000), majority earners experienced virtually identical judgments as those receiving all of their income from the government for each purchase decision. Our findings also suggest that $12,000 earners are a unique group, distinct from both $12,000 welfare recipients and $85,000 earners. We examine this pattern more in depth in the following experiments because it suggests the likely presence of multiple attributions. Finally, and most importantly, offering ethical goods at a discounted price protects low-income consumers receiving government assistance from harsh moral judgments. Consistent with our framework, eliminating the cost variable from the moral attribution equation allows this group to benefit from the prosocial element of ethical goods.
EXPERIMENT 3: ENVIRONMENTALLY FRIENDLY VEHICLE RENTALS
Experiments 1 and 2 examined our predictions in the domain of organic food. However, our framework suggests that morality effects may extend to other consumption domains characterized by both prosociality and cost. As such, we selected environmentally friendly products for experiment 3. In addition to a perceived “green” price premium ( Chang 2011 ), prior research indicates that consumers who purchase green versus conventional goods are perceived as more cooperative, altruistic, and ethical ( Mazar and Zhong 2010 ). Thus we predicted that consumers would be perceived as more moral when they choose green versus conventional goods, but only if they are relatively wealthy. Low-income consumers receiving government assistance, in contrast, should be perceived as less moral for the same action because using green products is costly (in addition to being prosocial).
The results from experiment 2 also revealed that low-income earners are judged differently than income groups at either end of the SES spectrum. We speculate that they share favorable attributions with both high-income earners (e.g., being lauded for choosing ethically) and those receiving government assistance (e.g., being lauded for economical behavior). Both attributions may operate simultaneously among low-income earners, producing the distinct pattern observed in experiment 2. As such, we included low-income earners again to examine more closely the independent effects of income level and source on moral attributions.
Experiment 3 was also designed to test for process more directly. We have proposed that earning income is associated with the right to make costly, ethical choices. If this is true, then targets who earn their own money (whether relatively low or high absolute amounts) will be perceived as more deserving of spending freedom than targets who do not earn their income. Deservingness should then drive moral judgments when ethical goods are chosen, but not when conventional goods are chosen. The former case conveys direct information about the person’s morality, whereas the latter case is simply the baseline option; “deserving” the right to an expensive choice is only relevant if the choice is indeed costly.
Participants and Procedure
A total of 181 adults (54% female; Mage = 35.80, SDage = 12.69) from MTurk completed the experiment in exchange for a $.50 payment. We developed a rental car scenario in which the target consumer chooses to rent an environmentally friendly (hybrid) vehicle or a more basic vehicle. The experiment followed a 3 (Annual Income: $12,000 in welfare benefits vs. $12,000 earned vs. $85,000 earned) × 2 (Rental: Prius vs. Corolla) between-subjects design. We chose the Prius because it has remained the top-selling hybrid electric vehicle (of over 40 models available for purchase) since its introduction in 2000 ( Alternative Fuels Data Center 2014 ).
To examine whether morality perceptions differed across the three income conditions before any consumption choices, participants rated the target’s morality before he made a purchase decision. The only information participants received about “John” was his age (in his 30s) and income level. Participants rated his morality with the same four item index used previously (“pre-morality”; α = .94). Everyone was then presented with the following scenario:
John currently lives in the city and needs to rent a car for an event that weekend. While he is looking on a rental car website, he is presented with two options: a Toyota Corolla and a Toyota Prius. Renting the Prius (a hybrid) would be more expensive than renting a more basic car. Specifically, it will cost John 35% more to rent the Prius.
We selected a 35% premium to be on par with that offered by several rental car companies. Images of the cars were presented (in a counterbalanced order), with John’s selection circled in red. Participants were randomly assigned to the Prius (“After some thought, John ultimately decides to pay more and rent the Toyota Prius”) or Corolla (“After some thought, John ultimately decides to pay less and rent the Toyota Corolla”) condition. After reading about John’s decision, participants completed the morality index again (“post-morality”; α = .95).
Our proposed mediator, deservingness, was assessed by asking participants to rate their agreement with three statements ( John deserves to spend his money as he chooses, John has the right to make his own spending choices, and How John spends his money is up to him ) on 7 point scales where 1 = Strongly disagree and 7 = Strongly agree. A deservingness index was created by averaging responses to these items (α = .92).
Results and Discussion
To examine baseline impressions of the target as a function of income before choice (“pre-morality”), we conducted a one-way analysis of variance (ANOVA) with income as the independent variable. The results revealed a significant main effect ( F (1, 178) = 5.03, p < .01, = .05), suggesting that individuals judge morality differently based on income information alone. Planned contrasts revealed that both $12,000 earners ( M = 4.69, standard deviation [SD] = 1.17) and $85,000 earners ( M = 4.68, SD = .87) were perceived as significantly more moral than $12,000 welfare recipients ( M = 4.12, SD = 1.31; both t’ s ≥ 2.73, both p’ s ≤ .01, both d’ s ≥ .46). Moreover, the two income earning conditions did not differ ( t (178) < 1), indicating that earning one’s income may be more influential in shaping moral judgments than absolute income level.
The primary analysis focused on morality ratings following the target’s rental choice. Due to statistical limitations of using differences scores (e.g., they are less reliable than either the pre- or posttest scores and are subject to ceiling and floor effects; Tabachnick and Fidell 2006 ), we conducted a 3 (Income) × 2 (Rental) ANCOVA with post-morality ratings as the dependent variable and pre-morality ratings as the covariate. A significant main effect of income ( F (2, 174) = 4.74, p = .01, = .05) was qualified by a significant interaction ( F (2, 174) = 16.03, p < .001, = .16; figure 3 ). Note that the interaction remains significant and the pattern holds without the covariate ( F (2, 175) = 7.86, p = .001, = .08) and when using difference scores as the dependent variable ( F (2, 175) = 15.52, p < .001, = .15).
Follow-up tests on the adjusted means showed that targets earning $85,000 a year were perceived as significantly more moral when they rented an environmentally friendly Prius than when they rented a conventional Corolla ( M = 4.91, SE = .16 vs. M = 4.20, SE = .17; p < .01, d = .78). Conversely, $12,000 welfare recipients were perceived as significantly less moral when they rented a Prius versus a Corolla ( M = 4.13, SE = .17 vs. M = 5.30, SE = .17; p < .001, d = 1.29). The $12,000 earners were viewed similarly whether they chose to rent a Prius or a Corolla ( M = 4.99, SE = .17 vs. M = 5.14, SE = .17; p = .54). These results fully replicate experiment 2.
To better understand the results of the $12,000 earners, we conducted planned contrasts within the two rental conditions. Specifically, it is possible that two different judgments are being made for low-income earners. On the one hand, they may be rewarded for saving money when they forgo unnecessary car upgrades (like welfare recipients). Yet they may also be rewarded for seeking a more ethical, although costly, rental car over the conventional alternative (like their relatively wealthier counterparts). Consistent with this dual attribution prediction, the results revealed that both $12,000 welfare recipients and $12,000 earners were perceived as significantly more moral than $85,000 earners when they saved money by choosing the conventional vehicle (both p’ s < .001, both d’ s ≥ 1.03). The two low-income conditions did not differ ( p = .50). Moreover, both $12,000 earners and $85,000 earners were perceived as significantly more moral than $12,000 welfare recipients when they spent extra money by choosing the green vehicle (both p’ s < .001, both d’ s ≥ .86). The two earning conditions did not differ ( p = .74). As in experiment 2, these results emphasize the importance of both income level and income source in shaping moral judgments.
An ANCOVA was also conducted on the deservingness index (with pre-morality evaluations as a covariate), which revealed only the predicted income main effect ( F (2, 174) = 19.55, p < .001, = .18). Relative to low-income individuals receiving government assistance (adjusted M = 5.00, SE = .16), both $12,000 earners (adjusted M = 6.15, SE = .16) and $85,000 earners (adjusted M = 6.27, SD = .16) were perceived as more deserving of choice (both p’ s < .001, both d’ s ≥ .95).
Recall that our conceptualization predicts that income earners will be seen as more deserving than nonearners, but deservingness should drive perceptions of morality only when ethical choices are made, not when conventional choices are made. To assess the role of deservingness in driving the effect of income on moral judgments, we ran a moderated mediation model with pre-morality evaluations as a covariate (model 15; Hayes 2013 ). This model specifies a process where X (income) predicts M (deservingness), which predicts Y (moral judgments), but the effect of M on Y depends on V (rental car choice). Because income consisted of three levels, we used dummy codes so that the welfare recipient served as the reference category. Model 15 was then run twice to obtain all of the parameter estimates: once with the $12,000 earner/welfare recipient comparison as X (and the $85,000 earner/welfare recipient comparison as an additional covariate) and once with the $85,000 earner/welfare recipient comparison as X (and the $12,000 earner/welfare recipient comparison as an additional covariate; Hayes and Preacher 2014 ).
The results reveal that $12,000 earners and $85,000 earners were both viewed as significantly more deserving of spending freedom than welfare recipients (β’s ≥ 1.15, both t’ s ≥ 5.10, both p’ s < .001). In turn, greater deservingness was associated with greater morality (β’s ≥ .13, both t’ s ≥ 2.28, both p’ s < .05). However, the positive effect of perceived deservingness on moral judgments was moderated by rental car choice in both models (β’s ≥ .29, both t’ s ≥ 2.92, both p’ s < .01). Examination of the relative indirect effects revealed that deservingness predicts morality when both $12,000 earners and $85,000 earners behave ethically by choosing to rent the Prius ($12,000: β = .38; 95% confidence interval [CI], .18–.65; $85,000: β = .35; 95% CI, .17–.57), but not when they rent the Corolla ($12,000: β = −.07; 95% CI, −.35 to .16; $85,000: β = −.02; 95% CI, −.32 to .24). Thus consistent with our predictions, deservingness is unrelated to morality when the target chooses the conventional option, but deservingness mediates the effect of earning income on morality when the target chooses ethically.
In sum, these findings demonstrate that the attribution reversal observed in experiments 1 and 2 is not unique to organic food, and that moral judgments depend on both income level (low vs. high) and income source (earned vs. unearned). The results from experiment 3 also provide initial support for the underlying psychological mechanism—individuals who earn versus do not earn their income are seen as relatively more deserving of choice and, as a result, they are viewed as more moral when choosing ethically. These results highlight the scrutiny low-income individuals receiving government assistance face regarding their monetary choices. They are viewed as less deserving than income earners, which subsequently prevents them from receiving moral credit when choosing ethically. In our next experiment, we examine the scrutiny that comes with misusing taxpayer dollars more directly. If, as we have proposed, low-income consumers receiving government assistance are seen as undeserving due to improper money management ( Cozzarelli et al. 2001 ), then spending taxpayer dollars on ethical goods could lead to perceptions among individuals that the recipients are abusing “their” money. To examine this possibility, the next experiment tests whether negative attributions of low-income consumers receiving aid can be mitigated by spending income from sources other than taxpayer dollars.
EXPERIMENT 4: ADDITIONAL SOURCES OF UNEARNED INCOME
Our experiments thus far have demonstrated that perceptions of deservingness drive moral judgments. To the degree that the source of assistance also changes perceptions of deservingness, we would also predict an impact on moral judgments. In particular, we propose that when financial assistance is given to an individual by a family member or a charity, instead of being given by the government from taxpayer dollars, other consumers will perceive the individual as more deserving of having the right to choose how the money is spent.
We base this prediction on two lines of reasoning. First, going back to the critical role that volition plays in attributions ( Jones and Davis 1965 ), because financial assistance given by family or charity is given freely and directly by the respective donor, perceptions of deservingness should be higher for the recipients because someone (or some group) deemed them deserving of the money. In contrast, because taxpayers do not have any direct say in how and to whom the government gives financial assistance, feelings of deservingness should be lower for those receiving the assistance. Second, it is these very taxpayers who are making the moral judgments of the recipients; it is, albeit indirectly, “their” money being given to these individuals (i.e., recipients are using “my inputs” to unfairly receive positive outputs; Adams 1965 ). When a charity gives money, unless the person making judgments gave money to that particular charity, he or she has not earned a say in how the charity spends its money. If moral derogation is indeed specific to the perceptions of deservingness as we propose, then spending unearned money received from family and/or a charity should be seen as similarly moral to spending earned money. In experiment 4, we examine this possibility by manipulating the source of unearned income and again test the mediational role of deservingness.
Participants and Procedure
A sample of 121 adults (38% female; Mage = 30.89, SDage = 9.38) recruited through MTurk completed the survey in exchange for a $.40 payment. Participants were presented with a scenario about “Daniel” or “Danielle” (matched with their gender), a 35-year-old who was asked for a donation while shopping for groceries. The male version of the scenario was as follows:
When Daniel approached the checkout line, the cashier asked him if he would be interested in donating money to a local charity. The charity, called “Better Streets for Us” is described as a neighborhood beautification project that plants flowers along city streets. Daniel agrees and gives the cashier $100 for the cause.
We wanted to ensure that participants would notice the size of the donation, so we elected to use a relatively high amount. We acknowledge that the larger amount decreases the external validity of the scenario, but we were willing to make the tradeoff in order to effectively test the theoretical underpinnings of our conceptualization.
We manipulated the source of income used to donate to the cause. Specifically, participants received information about the target’s income and were randomly assigned to one of four conditions: Welfare (“receives $12,000 a year in welfare benefits”), Charity (“receives $12,000 a year from a local charity”), Family (“receives $12,000 a year from his or her immediate family”), or Earned (“earns $12,000 a year”). Note that in all conditions both the donation size ($100) and income amount ($12,000) were held constant. As in previous experiments, the key dependent variable was the morality index (α = .90). Responses to two items ( Daniel has earned the right to spend his money as he chooses and Daniel deserves to spend his money as he chooses ) were averaged to form a deservingness index ( r = .87) for mediational analyses.
Results and Discussion
A one-way ANOVA conducted on the morality index revealed a significant effect of income source ( F (3, 117) = 3.82, p = .01, = .09). Replicating our previous experiments, the $12,000 welfare recipient was evaluated as significantly less moral than the $12,000 earner for donating money to a charitable cause ( M = 5.37, SD = 1.34 vs. M = 6.15, SD = .99; t (117) = 2.83, p < .01, d = .68). The results also revealed that donating money received from welfare ( M = 5.37) was viewed as significantly less moral than donating money received from charity ( M = 6.10, SD = .98; t (117) = 2.65, p < .01, d = .64) or family ( M = 6.17, SD = .94; t (117) = 2.85, p < .01, d = .70). Targets in the three “non-taxpayer” (i.e., charity, family, and earned money) conditions were perceived as similarly moral ( t’ s < 1). Thus consumers are most dissatisfied when low-income groups donate money received through government assistance—likely because consumers viewed the recipients as undeserving of the right to donate their money as they wish because they themselves contributed to this money via their own paycheck.
To assess the role of deservingness in driving the effect of income source on moral judgments, we ran a mediation model (model 4, Hayes 2013 ). Relative to individuals receiving government assistance ( M = 4.16, SD = 1.81), those who received money from charity ( M = 5.13, SD = 1.84; β = .97, t (117) = 2.47, p = .01, d = .54), their family ( M = 4.90, SD = 1.36; β = .74, t (117) = 1.88, p = .06, d = .47), or from employment ( M = 6.29, SD = .91; β = 2.14, t (117) = 5.42, p < .001, d = 1.53) were perceived as more deserving when making a monetary donation. Also note that individuals who earned their income were seen as more deserving than the other three, nonearning groups ( t’ s ≥ 3.00, p’ s < .01, d’ s ≥ .81). When we regressed morality on deservingness, we found a significant positive effect (β = .37, t (116) = 6.72, p < .001). Further, the indirect effects for both charity (β = .36; 95% CI, .04–.78) and income earners (β = .79; 95% CI, .47–1.26) were significant, and family marginally so (β = .28; 95% CI, .00–.67). The overall pattern indicates that spending money received from the government is judged more harshly than spending other sources of income, even if such income is not earned.
In sum, the results of experiment 4 highlight that spending tax dollars, specifically, negatively impacts moral judgments. However, a “tax dollars” account is not sufficient to fully explain all of the effects because income earners were seen as significantly more deserving than the other three groups. Thus consistent with the previous three experiments, earning is a central component of attributions of deservingness, but not the only one, because individuals who receive money from family and/or charity are seen as more worthy than individuals receiving government-issued benefits. This result follows from our volitional account: someone made a conscious choice to give to them. In contrast, when tax dollars are given to low-income recipients, individuals themselves are not directly involved in that choice (the way the charity or family members would have been), and consequently feelings of deservingness are lower.
To explore the role of volition more precisely, in the final experiment, we provided participants themselves with a conscious choice—the choice of whether or not to donate money to a charity dedicated to helping low-income consumers receiving government assistance.
EXPERIMENT 5: REAL DONATIONS FOR AN ORGANIC CHARITY
Up until this point we have demonstrated that attributions regarding others’ morality are made from the choices they make individually for themselves. According to attribution theory ( Jones and Davis 1965 ), the personal control inherent in making one’s own choices is what should drive outsiders’ moral judgments. Thus given the critical role that volition has in moral judgments, if we remove control (of choice) from the consumers and allow another entity to act as an agent making decisions on the consumers’ behalf, we would predict moral judgments to shift to the decision-making agent instead.
Many organizations (e.g., food banks, homeless shelters, churches) provide food to individuals, such that the recipient has little control over his or her own food choices. In this context, the attributions should shift away from the individual (who receives the food) and toward the organization (that makes the food choices), since the latter possesses volitional control over the choices made. Thus perceptions of resource management (and potential for mismanagement) would shift from the target consumer to the charitable organization itself.
In experiment 5, we return to the context of organic foods. Despite nonprofit organizations having a prosocial mission, we propose that because organic foods are generally costly, outsiders should view a charity more negatively if they choose to provide organic (vs. conventional) food to individuals seen as undeserving. However, since the individuals receiving the meals did not choose their food themselves (i.e., choice was exogenously imposed on them), our framework would predict that their morality should not be affected as a function of whether they receive organic food or not. A secondary purpose of this study was to capture moral judgments as reflected in real-world decision making to ascertain its practical significance. Interestingly, some nonprofit organizations have already begun hosting organic Thanksgivings at homeless shelters ( Organic Soup Kitchen 2011 ). Negative moral attributions may translate to less donor support for events like these, and a reduced willingness to help would also support a negative association between the poor and ethical goods.
Participants and Procedure
A total of 153 participants (69% female; Mage = 21.37, SDage = 2.78) from a large midwestern university completed the study as part of an hour-long session in exchange for payment. The experiment featured a one-factor design where a fictitious charity named “Helping Hands” was described as either providing “meals” or “organic meals” to area families who have been receiving welfare for the past year. At the beginning of the session, participants received a $12 payment in small denominations (seven $1 bills and one $5 bill) to encourage a later donation ( Small, Loewenstein, and Slovic 2007 ). They proceeded to complete a series of unrelated measures spanning approximately 45 minutes.
Once the online studies were complete, the computer screen instructed participants to open the manila envelope that had been placed at their individual cubicles. Inside the envelope, they found a charity request letter along with a small white envelope. Students read that the business school was assisting the local chapter of Helping Hands in their donation efforts because food pantries “typically have difficulty securing donations in the first six months of the year after the holidays.” The letter informed participants that each dollar they donated would be used to purchase items such as canned soups and stews, cereal and oatmeal, granola bars, pasta, and canned fruits and vegetables. In the organic condition, the word organic appeared in front of the foods (e.g., organic pasta); it was omitted in the nonorganic condition, which was otherwise identical. Once participants made anonymous decisions to donate or not and placed money ($0 to $12) into the white envelope, they were instructed to place all materials inside the unmarked manila envelope ( Kogut and Ritov 2005 ; Small et al. 2007 ). Following study completion, all donations were sent to an actual local food bank.
Once the participants made their anonymous donations and raised their hand to signal that they were done, participants were informed of one more short survey that the experimenter “forgot to include in the manila envelope.” Participants then completed a one page questionnaire including attitudes toward the charity (bad/good, negative/positive, undesirable/desirable, unfavorable/favorable, and dislike/like; α = .97) and moral perceptions of the families (immoral/moral and unethical/ethical; r = .79). The morality measure was different here for two reasons: parsimony and because this experiment was run earlier in the research program. Despite differences in measurement, the results from all previous experiments remain the same when we conducted the key analyses using the two item version of the morality index.
Lastly, we assessed frequency of monetary donations, food pantry donations, and demographics, but these factors did not affect the analyses. Completed surveys were placed in the unmarked manila envelope with the other materials, sealed, and left at their workstation. A lab assistant collected the envelopes once all participants had left and put them in a single box for the experimenter to pick up at the end of the day.
Results and Discussion
As predicted, participants donated significantly less money to Helping Hands when it was described as providing organic food to area families compared to nonorganic food ( M = $.85, SD = $1.14 vs. M = $1.35, SD = $1.85; t (151) = 1.98, p = .05, d = .33). Attitudes toward Helping Hands were significantly less favorable when they were described as donating organic food compared to nonorganic food ( M = 5.37, SD = 1.16 vs. M = 5.87, SD = 1.37; t (145) = 2.37, p < .05, d = .39). There was no difference between conditions in moral perceptions of the families served by the charity ( Morganic = 5.47, SDorganic = 1.06 vs. Mnonrganic = 5.57, SDnonrganic = 1.12; t (127) < 1). As expected, being recipients of organic (vs. nonorganic) food from organizations does not have an effect on moral attributions, consistent with our attribution framework.
We next tested whether overall attitudes toward the charity mediate the effect of reduced monetary donations (model 4; Hayes 2013 ). The results revealed that attitudes were significantly associated with donation amount (β = .32, t (144) = 3.33, p = .001), and the indirect effect of food type on donations was significant (β = −.16; 95% CI, −.38 to −.03). Thus donations were lower because the organization making the organic choice was perceived less favorably, not because the recipients were denigrated.
To add credence to our volitional account, we sought to determine whether perceptions would differ if it were the families themselves (vs. the charity) who requested organic food. According to our theorizing, harsh moral judgments should transfer back to the families if they exhibited agency over the choice. A separate sample of 110 individuals (27% female; Mage = 25.50, SDage = 4.39) from MTurk was presented a description of Helping Hands as a charity dedicated to providing meals to families who have been on welfare for the past year. Participants were randomly assigned to one of four “food request” conditions: recipient families had ostensibly made a special request for “organic food only,” “kosher food only,” or “lactose- or dairy-free food only.” The latter two conditions were included because they represent requests that would be need based (religious or allergy driven), whereas organic could be seen as a discretionary choice, but all are costly choices. The remaining participants were in the “no request” control condition in which no additional information was provided. Participants evaluated the families on the family morality index used in experiment 5 ( r = .95).
As predicted, perceptions of the families were significantly less favorable when they had requested organic food compared to when there was no such request ( M = 5.20, SD = 1.49 vs. M = 6.02, SD = 1.16; t (106) = 2.49, p = .01 , d = .63). Perceptions of the families did not differ from the control, however, when they requested either kosher ( M = 5.70, SD = 1.20) or lactose-free food ( M = 5.79, SD = 1.08; both p’ s > .35). Importantly, this pattern suggests that changes in moral judgments are not simply a matter of making any request; morality is implicated only when needy families request organic food or make a costly ethical choice specifically. Taken together, experiment 5 and this follow-up study demonstrate that the chooser is punished, even if the chooser is an organization, and the mere receipt of costly ethical goods alone does not necessarily harm one’s morality.
Despite recent interest in the lives of less fortunate consumers ( Adkins and Ozanne 2005 ; Ordabayeva and Chandon 2011 ; Saatcioglu and Ozanne 2013 ), very little research has examined the stigma low-income individuals receiving government assistance face in the marketplace. The current research responds to previous calls for additional studies ( Mick 2006 ) by shedding light on a historically underrepresented group in the marketing literature. This is an important area of inquiry because the present results indicate that consumers are not equal when it comes to the consumption of certain products.
Across five experiments, we tested a framework suggesting that identical choices lead to different moral judgments depending on income characteristics. In the case of ethical goods, relatively wealthy consumers can afford the price premium of these goods and therefore are perceived as more deserving of being able to choose them, resulting in their being viewed as more virtuous when they do so (i.e., the prosocial attribution is weighted more heavily while the cost attribution is diminished; experiments 1–3). For low-income consumers receiving government assistance who are viewed as less deserving, however, the expense of ethical products becomes magnified and overpowers the moral goodness (i.e., the prosocial attribution is diminished while the cost attribution is enhanced). In support of our proposed mechanism, low-income consumers who earned their income were somewhat protected from harsh judgments (experiments 2 and 3). Consumers who earn their money (whether small or large amounts) are seen as more deserving of choice, and as a result, they are viewed as more moral when choosing ethically.
As a further test of our framework, we find that equating the cost of ethical and conventional goods provides low-income consumers receiving government assistance some protection against harsh moral judgments when choosing ethically (experiment 2). Buying discounted organic food results in judgments comparable to those directed toward relatively wealthy earners choosing premium-priced organic food. This is an important boundary condition because it suggests that minimizing the cost component (and perhaps indirectly highlighting the prosocial component) of ethical choices may lead to more favorable perceptions of low-income consumers receiving government assistance. Although consumers are less punitive when unearned money does not come from government sources, those who earn money are still considered the most deserving and moral (experiment 4).
Experiment 5 also presents an important implication for charitable and nonprofit organizations. Participants donated less money to a charity when they believed their dollars would be buying organic versus conventional food for area families in need. Interestingly, the nature of the food did not impact perceptions of the families receiving support, but it did impact attitudes toward the charity itself. The key factor appears to be perceived responsibility for the organic choice. Paradoxically, organizations that expend time, corporate social responsibility effort, or employee resources toward causes designed to “go above and beyond” for those in need (e.g., donating organic vs. conventional food, new vs. used clothing, providing high-quality subsidized housing), may actually experience pushback or even backfire effects for trying to give low-income consumers receiving government assistance more than people feel they deserve.
Limitations and Future Directions
The current research suggests several potential avenues for future work. Gaining a better understanding of the marketplace challenges faced by low-income consumers receiving government assistance is important both theoretically as well as from a policy perspective. If this segment is seen as undeserving of ethical products, then government programs may garner less support among US voters if legislators initiate efforts to subsidize more of these items. For instance, most states explicitly prohibit Women, Infants and Children (a supplemental nutrition program) beneficiaries from spending benefits on organic items in most product categories ( US Department of Agriculture 2015 ). Our research speaks to one reason why this may be the case. Interestingly, while some states have formalized policies against the widespread adoption of ethical consumption choices, other states are becoming more supportive of these choices for all income groups (likely because ethical products have prosocial benefits for society). For instance, California hopes to make electric cars more affordable by offering vouchers to low-income residents ( Medina 2014 ). If consumers receiving government assistance take advantage of these sorts of programs, we may actually observe more favorable attributions because the higher cost element has been minimized, provided that others are aware of such programs. On that note, because prosocial behavior is often expensive, our results highlight the importance of taking both the price of the goods, as well as the consumer’s purchasing power, into account.
Our results reveal a consistent pattern of moral attribution effects on the ends of the SES continuum (with some movement in the middle). Although we did include low-income wage earners and partial earners in several experiments, future studies might test for moral judgment effects at additional points along the spectrum. For instance, the moral attributions of those earning extremely high incomes may actually be less favorable than those earning more moderate levels of income. Very wealthy consumers (i.e., the top 1%) may be seen as having large amounts of money at their disposal (eliminating the cost variable completely from the moral attribution equation). When they do purchase energy-efficient light bulbs or fair trade coffee, they may experience criticism for not doing more . In this way, it may be that another asymmetry exists in how the wealthy are perceived when making smaller, less expensive ethical purchases. Taking things further, it might also be the case that the wealthy are judged even more harshly for inaction; that is, high-income consumers who buy nothing when faced with an ethical purchasing decision may actually be viewed very negatively (e.g., as “stingy” or “cheap”), whereas low-income consumers receiving government assistance may be judged as more moral for not making a purchase and saving money. That said, the present results also indicate that when it comes to moral judgments, the absolute level of one’s income may not be as important as perceptions of having earned that income.
Although we investigated some boundary conditions of the moral judgment effect, several other contextual factors remain. Importantly, qualities of the perceivers themselves are likely to have a strong impact on their moral judgments of others (e.g., people who themselves choose ethical options may be more appreciative of others who do the same). Exploring individuals’ own habits and attitudes (vs. judgments of others) is an important area for future research. Second, examining whether and how our results extend to items characterized by just prosociality (e.g., living-wage clothing from discount retailers; Thomasson 2014 ) or cost (e.g., used furniture/clothing) would also provide greater insight.
Additionally, no information was provided regarding the circumstances surrounding the target’s income, which may influence perceptions of deservingness ( Petersen et al. 2011 ). For example, was the individual recently laid off from a job he or she had held for 20 years, or someone who is able-bodied but free-riding off the system? Although we examined the source of unearned income among low-income groups, it would be interesting to examine the source of income among high-income groups. The $85,000 target was always described as having earned his or her income in our experiments. Additional studies could examine whether results hold for targets who receive a similarly sized inheritance or earn their money working as a special needs teacher versus a used car salesperson.
Understanding the circumstances surrounding the target’s income might also facilitate deliberate, conscious judgment making, which might attenuate negative moral judgments of low-income consumers receiving government assistance. Indeed, research suggests that using situational information when drawing dispositional inferences is an effortful process ( Gilbert and Hixon 1991 ; Gilbert, Pelham, and Krull 1988 ). Once activated, stereotypes influence judgments unless one exerts cognitive effort to be less stereotypic. Although not tested explicitly in the current research, it is likely that participants engage in heuristic processing when evaluating the target, drawing on income stereotypes to form moral attributions. Prior work finds that low prejudiced individuals tend to make stereotypic judgments of others automatically, just as highly prejudiced people do, but this former group attempts to correct initial judgments ( Devine 1989 ). If we can encourage consumers to engage in deliberate processing, it seems possible to minimize harsh moral judgment directed toward aid recipients who purchase ethical goods by, for example, focusing their attention on external circumstances (e.g., recently being laid off).
Some large retailers are beginning to offer ethical products at reduced price points (e.g., Wal-Mart; Warner 2006 ). Less expensive ethical goods will allow low-income consumers receiving assistance to have the best of both worlds in terms of purchasing ethical products at affordable prices. Yet offering the same product at different prices to different segments may not be wise because such dynamic pricing can compromise fairness perceptions ( Haws and Bearden 2006 ). Will increased accessibility change the cost and prosocial associations of ethical consumption that made it “good” in the first place? Will lifting economic restrictions be sufficient to promote ethical product adoption ( Botti et al. 2008 ), or do social judgments need to change? These are provocative questions, and ones we hope will be explored in future research.
Not only do low-income consumers receiving government assistance face fiscal challenges, they face cultural challenges. Our research demonstrates that this group is morally derogated while those earning modest to high incomes are morally acclaimed for making identical choices. Across five experiments, we identify an attribution reversal: while low-income individuals receiving government assistance are perceived as less moral when choosing ethical (vs. conventional) products, income earners, particularly high-income individuals, are perceived as more moral for the identical choice (which usually costs more). Earning one’s income drives perceptions of deservingness, or the right to spend as one desires. Even charities that try to donate ethical goods to the less fortunate may be judged more harshly and receive less support. We believe these asymmetric moral outcomes are both theoretically compelling and practically relevant, and we hope they will provide a strong foundation for future work on attributions and marketplace stereotyping.
DATA COLLECTION INFORMATION
The first author collected data for experiments 1 and 5 from a paid subject pool at the University of Michigan during May and June 2011 and April 2012, respectively. Data for experiments 2 (October 2015), 3 (August 2014), 4 (March 2014), and the follow-up to experiment 5 (December 2012) were collected by the first author using Amazon MTurk. Statistical analyses for all experiments were conducted by the first author under the supervision of the other authors.