ASRXXX10.1177/0003122418806284American Sociological ReviewSchilke and Rossman 8062842018

American Sociological Review 29–­1 It’s Only Wrong If It’s © American Sociological Association 2018 https://doi.org/10.1177/0003122418806284DOI: 10.1177/0003122418806284 Transactional: Moral journals.sagepub.com/home/asr Perceptions of Obfuscated Exchange

Oliver Schilkea and Gabriel Rossmanb

Abstract A wide class of economic exchanges, such as bribery and compensated adoption, are considered morally disreputable precisely because they are seen as economic exchanges. However, parties to these exchanges can structurally obfuscate them by arranging the transfers so as to obscure that a disreputable exchange is occurring at all. In this article, we propose that four obfuscation structures—bundling, brokerage, gift exchange, and pawning—will decrease the moral opprobrium of external audiences by (1) masking intentionality, (2) reducing the explicitness of the reciprocity, and (3) making the exchange appear to be a type of common practice. We report the results from four experiments assessing participants’ moral reactions to scenarios that describe either an appropriate exchange, a quid pro quo disreputable exchange, or various forms of obfuscated exchange. In support of our hypotheses, results show that structural obfuscation effectively mitigates audiences’ moral offense at disreputable exchanges and that the effects are substantially mediated by perceived attributional opacity, transactionalism, and collective validity.

Keywords obfuscation, economic sociology, bribery, legitimacy, vignette experiments

In the early twentieth century, the social cus- through websites such as SeekingArrange- tom of “treating” developed as an alternative ments.com, which connect “sugar babies” to to older notions of courtship and prostitution. “sugar daddies.” Just as a century ago, trans- Women who participated in treating were actional sex remains a liminal category that called “charity girls,” because they received can pass as legitimate sociability but remains entertainment or goods as a gift and gave sex discreditable because it violates proscriptions as a gift but never sold sex as a commodity. The anti-vice society that investigated this custom begrudgingly placed charity girls in aThe University of Arizona the intermediate category of “near whores” bUniversity of California-Los Angeles (Clement 2006:1). This label refers to the fact that these women had not taken money from Corresponding Author: men, and this made the offense of their behav- Gabriel Rossman, UCLA, Department of Sociology, 264 Haines Hall, Los Angeles, CA ior less shameful than prostitution. Similar 90095-1551 practices have been reinvigorated in this century Email: [email protected] 2 American Sociological Review 00(0) on exchange. In other words, it can be per- questions of the boundaries and distinctions ceived as either basically dating or basically between market exchange and other types of commercial sex work. For example, one exchange. Boundaries exist in terms of what “sugar baby” seemed surprised that a jealous goods are usually understood to be legitimate friend told her parents that she was a “prosti- market goods (versus items excluded from tute” (Sales 2016). The case of transactional trade) and how to distinguish when a set of sex is just one example of obfuscated transfers qualifies as market exchange (versus exchange. Similar practices can be used to other forms of exchange, such as gifts). How- obfuscate political bribery, commercial brib- ever, recent investigations have challenged ery, and human tissue markets. In all these the categorical approach to markets and mor- cases, observers may have a less unfavorable als. These studies emphasize that the line moral reaction to the exchange to the extent between market transfers and non-market that they understand it as structurally differ- transfers is often blurry, and this ambiguity ent from a quid pro quo. can be the object of obfuscation by partici- In this article, we argue that an economic pants and moral evaluation by audiences exchange, such as bribery, that would offend (Åkerström 2014; Fridman and Luscombe audiences if it were an open quid pro quo can 2017; Hoang 2018; Lainer-Vos 2013; Mears be made less offensive when it takes on a dif- 2015; Rossman 2014). ferent structure. In particular, actors can miti- The growing literature on obfuscated gate audience disapproval of disreputable exchange shows that moral perceptions are exchange by practicing structural obfusca- not only about whether a set of goods can tion—defined as arranging a set of transfers legitimately be exchanged, but whether a set so as to obscure the fact that a disreputable of transfers is understood as an exchange at exchange is occurring. Expanding upon Ross- all. This implies that participants who seek to man (2014), we articulate and experimentally retain their honor will reshape sets of trans- test the effects of four such obfuscatory fers to make them look less like exchanges. exchange structures. First, bundling occurs Consistent with this literature, our investiga- when two seemingly unrelated exchanges or tion is concerned with obfuscated disreputa- relationships cross-subsidize one another. ble exchange; however, our work differs Second, brokerage occurs when ego pays a significantly from prior investigations in its broker, who in turn pays alter, who then gives scope and methods. The extant literature a favor to ego. Third, gift exchange involves focuses on identifying actual instances of asynchronous transfers that, in principle, are obfuscation or accusations of obfuscation. voluntary and not strictly reciprocal. Finally, However, because it relies on qualitative real- pawning involves using favors to redeem a world cases, this literature cannot systemati- debt.1 Although these four structures involve cally measure the moral evaluations of different relational work, we propose that audiences nor establish the causal effect of they all can render an illicit commensuration obfuscation on these evaluations. Indeed, less explicit. In so doing, they can sublimate because obfuscation is only observable when transactionalism across a wide range of sub- it is partially or wholly unsuccessful, research stantive contexts. based on real-world cases necessarily sam- This article draws on and contributes to ples on the dependent variable. Through ran- several research streams, most notably work domized experiments in which we vary the in sociology, anthropology, economics, law, structure of exchange but keep the basic and philosophy on exchange and contested topic, we are able to complement extant qual- commodities (Bourdieu 2000; Fiske 1992; itative findings. In addition to showing that Krawiec 2009; Molm 2003; Roth 2007; San- obfuscation does occur, we also show how del 2000; Walzer 1983; Zelizer 2005). This and why it is effective through mediation cross-disciplinary literature has focused on analysis. The mechanisms highlighted in this Schilke and Rossman 3 analysis are key to understanding how obfus- differences in moral evaluations through cation works to limit the moral fallout from these three theoretical mechanisms, which exchanges that are substantively equivalent to also distinguish between structurally obfus- quid pro quos. cated and directly negotiated (i.e., quid- This article also contributes to research on pro-quo) forms of disreputable exchange. By the social psychology of morality. Sociologi- elucidating these mechanisms, we aim to cal social psychologists traditionally empha- contribute to integrating the streams of socio- size economic or belongingness needs, and logical research on cultural phenomena and moral concerns have only recently taken social psychological processes (Collett and center stage in their research (cf. Hegtvedt and Lizardo 2014). Scheuerman 2010; Lawler and Hipp 2010; Finally, this article also speaks to institu- Stets and Carter 2011). In contrast, within tional theory, which is fundamentally con- psychology, the study of morality has blos- cerned with legitimacy (Meyer and Rowan somed, uncovering individual dispositions 1977; Suchman 1995). Recently, institutional and cognitive processes in moral evaluations scholars have advocated applying this con- (Haidt and Kesebir 2010)—but ignoring social cept to the context of market exchanges and structural considerations. Research on the role studying its moral components (Anteby 2010; of exchange structure, which is central to Quinn 2008). Largely unexamined so far, sociological social psychology (Molm, however, are the micro-foundations or drivers Collett, and Schaefer 2007; Willer, Flynn, and of legitimacy (see also Bitektine and Haack Zak 2012), has yet to find its due place in the 2015; Haack et al. 2016; Lucas and Lovaglia morality literature. In this study, departing 2006). Our findings help fill this gap. We from earlier research on the social psychology show not only that the type of actor or traded of exchange structures, we follow the lead of resource matters but also that perceptions of moral psychology (Haidt and Kesebir 2010) in legitimacy are a function of the structure adopting the viewpoint of the external observer through which the resource is traded. We pro- of the exchange (rather than of a participant) vide evidence for the notion that how an to shed light on audiences’ moral perceptions. exchange is achieved can shape evaluation This perspective reflects that individuals’ above and beyond who is involved and what actions are not only shaped by their own feel- is being exchanged. Thus, we shift the focus ings and preferences but also by the moral from the actor and the resource to the approval or disapproval of audiences. exchange structure and, in so doing, uncover In this article, we seek to answer the ques- a neglected way in which actors can achieve tion of whether distinct exchange structures and maintain legitimacy. will cause differences in the extent to which The rest of this article is structured as fol- audiences find an otherwise illicit exchange lows. We first derive hypotheses from a morally acceptable. Specifically, we investi- review of the nature of exchange taboos and gate four types of exchange structures and obfuscation through transfer structure. Spe- show that these structures appreciably shape cifically, we propose both direct effects and observers’ moral responses to otherwise com- mediated pathways for how structural obfus- parable disreputable transactions. Adding cation practiced by actors shapes the moral process evidence, we also identify three evaluations of audiences. In our theorizing, causal mechanisms that help explain the path- we treat the four exchange structures as a set, ways through which exchange structures emphasizing their common consequence of influence morality judgments. These mecha- mitigating moral offense. We then test our nisms are perceived attributional opacity, hypotheses through a series of experimental transactionalism, and collective validity (as studies. These studies address reputable trans- we will discuss later). Specifically, we pro- fers, open disreputable exchange, and obfus- pose that structural obfuscation affects cated exchange in the contexts of political 4 American Sociological Review 00(0) bribery, commercial bribery, and compen- list of disreputable exchanges across cultures, sated adoption. We close with a discussion of explaining what counts as disreputable our empirical findings and their contributions exchange is an inquiry in its own right (Nel- to different literatures. son 1969; Sandel 2000; Satz 2010). Disrepu- table exchange encompasses a wide variety of phenomena, going well beyond the obvious Morally Disreputable examples of vice and corruption, and in some Exchange contexts it can be pervasive, although the The fundamental starting point for this article exact magnitude is difficult to calculate.3 We is that people sometimes object morally to draw from a variety of substantive contexts to economic transactions precisely because they illustrate and test our arguments, with the aim are seen as putting prices on things belonging of developing a general understanding of the in non-market contexts (Krawiec 2009; San- role of structural obfuscation in moral del 2000; Walzer 1983).2 We define disrepu- perceptions. table exchange as provision of goods or services that is culturally prohibited, where the moral objection is contingent on an under- Structural Obfuscation standing of this provision as economic. Notably, If exchange taboos rely on the extent to this definition excludes cultural prohibitions which an interaction can be interpreted as in which morality focuses on action regard- transactional, this leads us to ask: Can people less of compensation (Krawiec 2009). For effectively mitigate moral disapproval by instance, murder is wrong, and not just if it is structurally obfuscating such exchanges? To a paid hit, but even if it is provided for free. address this question, we investigate the In contrast, most contemporary Americans degree to which specific types of structural tolerate political advocacy and casual sex but obfuscation can alleviate the offense of not bribery and prostitution, even though the exchange taboos. Based on Rossman (2014), only difference between the former and the Rossman and colleagues (2016), and an latter is whether money changes hands. In extensive reading of relevant historical, eth- these cases, moral revulsion occurs primarily nographic, and journalistic accounts, we because someone is perceived as taking an identify a set of four different structures action out of venal motives rather than through which transactions can be obfus- because of the action itself. A hit job for free cated: bundling, brokerage, gift exchange, is still murder, whereas bribery for free is an and pawning. These structures effectively oxymoron. Paying for political favors or sex allow one party to pay the other and to is disreputable specifically because it breaches receive non-market goods in return, but in the moral limits to fungibility of market and ways that make it less conspicuous that an non-market goods (Espeland and Stevens economic transaction has occurred. By exten- 1998; Sandel 2000; Walzer 1983). sion, we expect that obfuscation can mitigate This class of disreputable economic trans- the moral offense of what would otherwise actions qua economic transactions includes be considered disreputable exchange. The political bribery, commercial bribery, human subtlety and ambiguity of obfuscated tissue markets, compensated adoption, and exchanges allows them to be characterized prostitution. Taking a comparative-historical by a “two-fold truth”: they are meaningfully perspective, the category also encompasses characterized by whatever social practice exchange taboos less salient to contemporary they appear to be on the surface, and they are U.S. culture, such as riba (usury) in Islam, ultimately morally disreputable commensu- simony (bribing clergy) in medieval Europe, rations (Bourdieu 2000).4 and senators’ involvement in commerce in Figure 1 depicts different forms of trans- ancient Rome. As suggested by the diverse fers (adapted from Rossman 2014). In each Schilke and Rossman 5

Figure 1. Quid-Pro-Quo versus Obfuscated Exchange Structures Note: Each panel in the figure represents a set of transfers. Ego and alter represent parties to a set of transfers. “A” represents money or a market good. “B” represents a non-market good, such as a political favor. Capital letters denote valuable goods; lowercase letters denote goods of relatively trivial value. part of the figure, one actor (ego) gives a disapproval—defined as the extent to which market good (A) to another actor (alter), who audiences to an interaction characterize the in turn gives a non-market good (B) to ego. In actors or the interaction as a whole as offen- quid-pro-quo exchange, the two goods are sive, demonstrating bad behavior, or deserv- explicitly exchanged for one another. The ing punishment. For each structure, we provide subsequent panels all describe forms of illustrative examples from real-world cases. obfuscation. We depict generic obfuscation as Note that this creates something of a paradox. ego receiving a non-market good and alter Obfuscation is only observable when it does receiving a market good, but in such a way not work; therefore, all examples we give are that the linkage between the two goods, or only partially or temporarily efficacious. We even their source, is no longer readily appar- revisit this issue of temporarily or partially ent. All the specific obfuscation structures effective obfuscation in the conclusion. introduce some form of buffer that obscures reciprocity, although the specific form of Forms of Structural Obfuscation buffer varies. Specifically, with gifts and pawning, time buffers reciprocity; with bro- Bundling. From 1994 to 2011, flamboyant kerage, a third party serves this role; and with media magnate Silvio Berlusconi served bundling, a second set of transfers occurring intermittently as prime minister of Italy. Ber- in parallel obscures reciprocity of market for lusconi retained ownership of his Mediaset non-market goods. corporation throughout his time in office. The In the next section, we first outline each of pronounced pattern of companies in highly the four obfuscation structures. We then regulated industries buying advertising time hypothesize about how they affect moral when Berlusconi was in office strongly 6 American Sociological Review 00(0) suggests that these companies were seeking Brokerage. In the 1980s, record labels to buy political influence (DellaVigna et al. relied on outside consultants called “indies” 2016). Similarly, decades earlier in the United to get their songs played on the radio (Dannen States, “everybody knew that a good way to 1990; Rossman 2012). In theory, radio pro- get Lyndon [Johnson] to help you with gov- motion consists of sending albums and press ernment contracts was to advertise over his kits to radio stations as well as visiting the radio station” (Caro 1990:103). Indeed, some stations to promote a single and explain the commercial interests even purchased ads label’s plan for the artist. In reality, during despite not having business in KTBC’s mar- this period, radio promotion was character- ket of Austin.5 Likewise, since the 2016 U.S. ized by “payola” (commercial bribery). Indies presidential election, a major line of criticism would pay radio station staff in cash or of President Trump has been that parties seek- cocaine, and the record labels would in turn ing to influence him could purchase services pay indies for ensuring that their songs from or otherwise favor his business interests received airplay. Record labels would seldom (usually referred to as “emoluments,” in ref- pay radio stations directly; rather, the pay- erence to a clause of the Constitution). Exam- ments went through the indies, and the labels ples of this can be seen when diplomats stay were deliberately oblivious to the details at Trump’s DC hotel or when Trump’s devel- (Dannen 1990). opment deals overseas gain (difficult and Brokerage occurs when an exchange somewhat discretionary) regulatory approval between two parties is mediated by a third (Paddock et al. 2016). party. In our particular usage, we understand Bundling occurs when two morally reputa- brokerage as structurally resembling a chain- ble exchanges cross-subsidize one another to or network-generalized exchange (the most synthesize the equivalent of a morally dis- famous example of which is a “kula ring”). In reputable exchange. In Figure 1, ego and alter such an exchange, ego gives to broker, broker trade a valuable market good (A) for a low- gives to alter, and alter gives to ego, so that value market good (a) and a low-value non- reciprocity is cyclical rather than direct market good (b) for a valuable non-market (Yamagishi and Cook 1993). Critically, our good (B). This combination makes it seem use of the term brokerage encompasses the that market goods are traded for market goods idea that some of the directed ties involve the and that non-market goods are traded for non- transfer of market goods and others of non- market goods. Therefore, bundling obfuscates market goods. As shown in Figure 1, market the fact that, in some meaningful sense, a good (A) flows from ego to broker to alter, market good (A) is being traded for a non- but then alter sends a non-market good (B) to market good (B). For instance, it is licit for a ego. This structure allows the commensura- company to purchase advertising time from a tion of market and non-market goods without broadcaster, and it is licit for a company to direct reciprocity, since at no point does any- petition a policymaker for favorable regula- one trade A for B directly. Rather, value goes tion. However, when the broadcaster is owned in a cycle, and over the course of that cycle by the policymaker, the company could be the nature of the circulating value changes paying above-market rates for these services from market to non-market. Whereas prior and thus implicitly buying influence with the research emphasizes the fact that brokerage policymaker. Likewise, intimacy between can facilitate negotiations between low-vol- professionals and clients can be interpreted as ume traders (Stovel and Shaw 2012), we are bundling, or “differentiated ties” in Zelizer’s more interested in the ways brokerage serves (2005:95) terminology, such as when a law- observers’ moral concerns (Stovel, Golub, yer who has had sex with his client accepts and Milgrom 2011). Although brokerage is payment for legal services discounted by often associated with the network centrality $200. measure of “betweenness” (Marsden 1982), Schilke and Rossman 7 our approach brackets any issues of whole arduously long evenings at nightclubs build network structure and highlights how a focal relationships with the women and woo them interaction is structured. In our usage, we with gifts rather than pay them wages (Mears emphasize instances in which the broker 2015). In Fiske’s (1992) relational models serves to obfuscate the fact that a disreputable theory, disreputable exchange is defined as transaction is occurring. using an inappropriate relational model for a particular good, for instance charging money Gift exchange. The Roman Catholic for a good that ought to be given as a gift Church does not recognize divorce. However, (Fiske and Tetlock 1997). Similarly, social it does allow for annulment, which is a ruling psychologists refer to gift exchange as a that the original marriage was never valid due reciprocal exchange, in contrast to direct to violating one or another aspect of canon negotiated exchange (i.e., quid quo pro) law. Many Catholic churches charge a fee of (Emerson 1981; Molm et al. 2007). a few hundred dollars for annulment applica- tions, which is understood to cover the Pawning. When the gangster Henry Hill expenses of processing the request. However, was arrested, he became a cooperating witness this fee can create the perception that annul- for the U.S. Attorney and recounted to the ments can be “bought.” For the explicit pur- journalist Nick Pileggi (1985) how the Luc- pose of avoiding such a perception, in August chese crime family controlled JFK airport. 2014 Bishop Gerald Barnes declared that his The key to accessing the airport’s cargo was diocese of San Bernardino, California, would recruiting inside men among the teamsters, no longer accept fees for annulments, because baggage handlers, and security guards. As a “annulments—like God’s mercy—are with- general rule, airport workers did not sell tips out price” (Barnes 2014). Rather, he instructed on cargo to the mob in aspiration for profit; priests to encourage (but not require) that rather, they offered them in desperation to petitioners tithe to their parishes, not only clear their debts. Many workers at JFK airport while the annulment was being processed, but were gambling addicts who were deep in debt indefinitely, as a sign of acknowledging the to the mob, and “a tip on good cargo loads “mutual rights and duties” of the church com- could make up for a lot of paychecks and buy munity. In other words, wishing to avoid the back a lot of IOUs” (Pileggi 1985:88). Indeed, perception of annulments as a commodity, gangsters owned bars and bookies near the Bishop Barnes treats them as a gift. However, airport and would deliberately carry gambling this gift implies a relationship that carries its debts in order to recruit collaborators. own reciprocal obligations. Pawning is an obfuscation structure in Gift exchange occurs when two parties which people take on and then repay debts give one another gifts within the context of a using goods or services they would be reluc- relationship (Bourdieu 1990, 2000; Fiske tant to sell directly. As seen in Figure 1, in 1992; Lawler and Hipp 2010; Mauss [1923] pawning, ego loans alter a market good (A) 1967; Molm and Cook 1995). One party gives and alter explicitly acknowledges a market the other a gift, and then, usually with a time debt (IOU). When alter is unable to repay the delay, the second party gives a gift to the first. IOU with market goods, alter instead pro- In Figure 1, ego and alter are engaged in gifts vides a non-market good (B) to explicitly as obfuscation. Ego gives alter a market good cancel the debt. The practice resembles gift (A), and later on alter gives ego a non-market exchange in that it is asynchronous, but it dif- good (B). The gifts are framed as expressions fers in that the debt is explicit and usually of the relationship, and therefore the fact that denominated in money, rather than an infor- a market good was traded for a non-market mal obligation with a fuzzy value as in gift good is sublimated. For instance, party pro- exchange. Likewise, reciprocity explicitly moters who arrange for models to spend cancels the debt and may be explicitly 8 American Sociological Review 00(0)

Figure 2. Serial-Mediation Model Note: The expectation underlying Hypothesis 2D is that the serial-mediation path (i.e., the indirect effect illustrated through consecutive solid arrows in Figure 2) will be significant. Dashed arrows indicate direct effects net of the main serial-mediation path shown as solid arrows. The model will estimate these effects, but our theoretical focus is on the serial-mediation path. demanded, whereas in gift exchange reciproc- structural obfuscation and audiences’ moral ity is more sublimated. That is, gift exchange reactions. We theorize a mediated model is relational and involves favors, whereas (shown in Figure 2), in which we anticipate pawning is contractual and involves debts. In that obfuscation will operate to a significant Fiske’s (1992) terminology, pawning is an extent through three intervening processes: aspect of market pricing, whereas gift perceived attributional opacity, transactional- exchange is equality matching or authority ism, and collective validity. We suggest that ranking. The felt need to repay debts can these three causal mechanisms help explain override other moral concerns, and under the how structural obfuscation reduces moral pressure of debt we see people commit such disapproval. In our model, we consider audi- misdeeds as enslaving populations, sacking ences’ own view of the interaction (i.e., their cities, and stripping churches of relics (Grae- perception of attributional opacity and trans- ber 2011). actionalism) and their judgment of how oth- ers will view it (i.e., their perception of Summary. We outlined four forms collective validity). This is consistent with the through which disreputable exchanges can be sequential decision-making model outlined and are obfuscated. Although these forms by Tolbert and Zucker (1996), whereby indi- may differ in structure, the following hypoth- viduals will assess both their own subjective esis emphasizes how they serve similar pur- interpretation of the situation and likely inter- poses of mitigating the moral disapproval of pretations by external sources before forming observers: their overall evaluation. We start by discuss- ing each mediator separately, and then we Hypothesis 1: Structural obfuscation of disrep- present a hypothesis regarding their causal utable exchange (versus taboo quid pro quo) ordering. decreases moral disapproval. First, we propose that structural obfusca- tion may increase perceived attributional Mechanisms Underlying Structural opacity—defined as a lack of clarity about Obfuscation actors’ intentions, especially pertaining to their exchange motivation. We propose that, A theory of obfuscated exchange should not compared to quid pro quo, audiences will find only show that obfuscation affects moral per- it more difficult to clearly determine actors’ ceptions, but also how and why. To identify exchange motivation. Whereas quid-pro-quo relevant mechanisms, we investigate possible exchanges typically suggest parties’ explicit mediators of the relationship between understanding of the purpose of the exchange, Schilke and Rossman 9 the more complicated obfuscatory exchange one stay clear of moral limits to fungibility structures make it more difficult to determine (Espeland and Stevens 1998; Sandel 2000; parties’ motives, as these structures afford Walzer 1983). An exchange that explicitly “the denial (Verneinung) of interest” commensurates the incommensurable attaches (Bourdieu 2000:192). Just as taboo sugges- a concrete market value to a non-market good. tions framed in indirect speech can allow for Pricing the priceless will therefore provoke plausible deniability (Pinker, Nowak, and Lee moral outrage and cries for punishment (Fiske 2008), structurally obfuscated exchanges can and Tetlock 1997). However, as perceived make it hard to attribute actors’ intentions. transactionalism decreases, the commensura- Attributional opacity, in turn, can help miti- tion of incommensurables is masked. As such, gate moral disapproval. Opacity is often the exchange feels less culturally inappropriate viewed as a source of inefficiency in eco- and thus will summon comparatively less moral nomic exchange, but it can help facilitate the disapproval. moral acceptability of otherwise problematic Third, we consider that moral judgments transactions (Lainer-Vos 2013). Audiences of obfuscated exchange can benefit from per- frequently rely on their inferences about ceived collective validity—broadly defined as intentions to make moral judgments (Kelley the extent to which an action is thought to be 1971). As a result, violations that are clearly “right and proper” by most people (Dorn- intentional will be judged more harshly than busch and Scott 1975; Granovetter 2007; comparable behaviors that appear to occur Weber [1922] 1978). This notion refers to the more incidentally (Malle and Knobe 1997). observer’s perception that others in society Conversely, obscuring ulterior motives can will view an action as appropriate for its produce more favorable audience reactions social context (Bitektine and Haack 2015; (Hahl, Zuckerman, and Kim 2017). Taken Suchman 1995). Objects high in collective together, these arguments indicate that attri- validity have a strong taken-for-granted char- butional opacity will mediate the effect of acter, to an extent that alternatives are “liter- structural obfuscation on moral disapproval. ally unthinkable” (Zucker 1983:5). Two Second, we argue that structural obfuscation dimensions of validity are endorsement (i.e., may reduce moral disapproval because it can peer support) and authorization (i.e., support lessen perceived transactionalism—defined as from a superordinate entity) (Johnson, Dowd, the perception of a direct causal relationship and Ridgeway 2006). Endorsement occurs between the goods exchanged between parties. when comparable actions are prevalent, mak- Overt quid pro quos make the contingency ing the validity of a given action seem taken involved in the exchange explicit (I give you x for granted. In other words, the more com- if, and only if, you give me y). In contrast, monly observed an action is in everyday life, obfuscation allows individual transfers (e.g., the more collectively valid it is perceived to the transfer of money and the transfer of a be (i.e., “everybody does it”). In authoriza- favor) to be seen as existing in separation rather tion, an action is tolerated by key institutions than being made in direct exchange for each of society, most notably the judicial system. other (Emerson 1976). Obfuscation creates Both of these validity clues—commonness temporal or spatial, and thus ultimately cogni- and legality—apply to many types of exchanges tive, distance between individual transfers and organized through obfuscation structures. thus weakens the audience’s perception that Indeed, many types of obfuscated taboo they are directly causally linked. In this way, exchange are quite common even among obfuscation structures reduce the explicitness respectable people. Political campaign contribu- with which one thing is exchanged for another, tions are typically not provided in a quid- and the transfers are no longer viewed as neces- pro-quo exchange for policy favors, but a smart sarily reciprocal. Sublimated transactionalism, lobbyist will recommend that a legislative sup- in turn, is often more palatable because it helps plicant support campaigns. Moreover, structural 10 American Sociological Review 00(0) obfuscation is frequently not explicitly prohib- causal chain. Specifically, we expect attribu- ited by the law. Many types of obfuscation are tional opacity to have the most proximal not blatant enough to be considered illegal effect, with audiences immediately scrutiniz- beyond a reasonable doubt (Noonan 1984; ing actors’ intentions, given that attributions Pinker et al. 2008). Through their perceived col- are generally believed to occur quickly (Har- lective validity, obfuscation structures gain a vey et al. 2014). These attributions, in turn, taken-for-granted status in the eyes of the likely affect perceived transactionalism, such observer. As Walker and colleagues argue, peo- that exchanges in which parties appear to act ple rely on their understanding of consensual with intent are viewed as transactional viewpoints when forming their own assessment (Weiner 1986). Next, perceptions of transac- of what is right and wrong (Walker, Rogers, and tionalism will inform validity beliefs, with Zelditch 1988; Zelditch and Walker 1985). If highly transactional exchange being viewed “everybody does it,” it cannot be improper as lacking collective approval. Finally, collec- (Noonan 1984). As such, when obfuscation tive validity functions as a direct driver of structures help enhance collective validity individuals’ ultimate judgment (Walker et al. beliefs, this perceived validity will decrease the 1988; Zelditch and Walker 1985). These argu- audience’s moral disapproval. ments suggest a serial model, as outlined in In summary, we propose three key mecha- Figure 2 and summarized in Hypothesis 2D: nisms that help explain how structural obfus- cation is perceived and ultimately judged by Hypothesis 2D: The negative effect of struc- third parties. We predict that structural obfus- tural obfuscation (versus taboo quid pro cation reduces moral disapproval indirectly quo) on moral disapproval is mediated by by affecting perceived attributional opacity, perceived (1) attributional opacity, followed by (2) transactionalism, followed by (3) col- transactionalism, and collective validity. Thus, lective validity. we hypothesize as follows:6

Hypothesis 2A: Perceived attributional opac- ity mediates the negative effect of structural Empirical Analyses obfuscation (versus taboo quid pro quo) on Study Overview moral disapproval, such that structural ob- fuscation increases attributional opacity, and To test our hypotheses, we conducted four attributional opacity in turn decreases moral vignette experiments in which study partici- disapproval. pants were randomly assigned to read about Hypothesis 2B: Perceived transactionalism me- either an explicit disreputable quid pro quo, diates the negative effect of structural obfus- an obfuscated transaction, or an otherwise cation (versus taboo quid pro quo) on moral similar but reputable transfer (between-sub- disapproval, such that structural obfuscation jects design). The experimental method decreases transactionalism, and transaction- allows us to tightly control the study design to alism in turn increases moral disapproval. reflect our theoretical interests. The vignettes Hypothesis 2C: Perceived collective validity we used closely reflect every obfuscation mediates the negative effect of structural obfuscation on moral disapproval, such that structure as well as quid-pro-quo and cultur- structural obfuscation (versus taboo quid ally appropriate conditions. Notwithstanding pro quo) increases collective validity, and their limitations, vignette experiments have collective validity in turn decreases moral the ability to establish causal inference and disapproval. isolate underlying mechanisms while using realistic contexts that alleviate potential con- Finally, in considering how the three medi- cerns about external validity (Aguinis and ators work in concert, we expect an underly- Bradley 2014; Finch 1987; Schilke 2018; ing pattern of serial mediation, whereby Sniderman and Grob 1996). Particularly in individual mediators function in a sequential our context of studying audience perceptions, Schilke and Rossman 11

Table 1. Experimental Manipulations – What Does the Supplicant Provide to the Person from Whom They Seek a Favor?

Condition

Quid pro Gift ex- Study Appropriate quo Bundling Brokerage Pawning change

1 & 4: Political Loans Provides Discounts Sponsors Forgives Previously bribery a pen immediate campaign voter unpaid made campaign office rent registration campaign campaign contribution through bills contribut- NGO ions 2: Commercial Loans Provides cash Overpays for Pays Forgives Previously bribery a pen bribe used car honorarium damages to gifted a used via dealers’ car car association 3: Baby- Brings Provides cash Overpays for Pays via a Pays off Previously selling coffee and payment used car lawyer medical gifted a used muffins bills car Trivial value Value of $10,000 the mundane realism of vignette experiments Mechanical Turk (total n = 1,517). The exper- is comparatively high (Mutz 2011), because iments were run between April 2015 and June the experimental setting of reading about an 2017. Please see our data archive for the exchange resembles many relevant real-world original data and data processing scripts experiences (e.g., reading the news). (https://osf.io/yxafu/). We crafted vignettes about bribery and baby-selling in order to cover moral objec- Study 1: Political Bribery tions to exchanges associated with concerns about fairness and disgust (Munger 2011; This study is the first of four in which we Sandel 2000). Study 1 presents a political used a completely randomized single-factor bribery vignette in which a defense contractor design with six conditions. Participants read a requests a favor from a congressman. Study vignette involving either a taboo quid pro 2’s vignette describes a commercial bribery quo, a culturally appropriate situation, or one scenario in which a car dealer requests that of the four types of obfuscated exchange. In corporate ship more high-demand cars to his Study 1, we used a vignette about political dealership. Study 3’s vignette describes a bribery between a defense contractor and a couple seeking to adopt a baby from a new congressman. mother. Study 4 returns to the congressman scenario from Study 1 and introduces three Participants. As in the studies that fol- mediators underlying the effect of structural low, Study 1’s participants were recruited via obfuscation. Table 1 summarizes the manipu- an advertisement posted on Amazon’s lations in all four studies, which differ only in Mechanical Turk (AMT) and offered US$1 in terms of what the supplicant (the defense compensation. AMT is an online crowdsourc- contractor, the Honda dealer, and the adopting ing service that offers large volumes of small parents) offers to the person from whom they web-based tasks to online workers for mone- are seeking a favor (the congressman, the tary compensation. AMT allows experiments Honda executive, and the birth mother). All to be run comparatively inexpensively, pro- four experiments were embedded in online vides access to a broad cross-section of the surveys, with subjects recruited from Amazon population, and has repeatedly been shown to 12 American Sociological Review 00(0) have the capacity to produce highly reliable median respondent had an annual household data (Buhrmester, Kwang, and Gosling 2011; income between $50,000 and $59,999, and Weinberg, Freese, and McElhattan 2014). 57.5 percent of respondents had at least an Research using AMT has become increas- undergraduate college degree. ingly common, not only in sociology but also Study 1, like our other three studies, in psychology, economics, and political sci- included two attention screeners. One item ence. Although AMT respondents are not per- hidden within a number of other items read, “If fectly representative of the U.S. population you read this, please select strongly agree” (e.g., they are more politically liberal on aver- (adapted from Berinsky, Margolis, and Sances age), they do feature substantial demographic 2014). Another item asked what issue the sce- diversity and are thus often preferable to col- nario was mainly about (with four answer lege student samples, especially for research choices provided). Between 92.7 and 96.7 per- concerning political issues (Willer, Feinberg, cent of respondents in each study answered and Wetts 2016).7 both questions correctly. We experimented Recently, researchers have flagged the with dropping inattentive participants and possibility of fraud in AMT responses (Den- found that doing so did not materially affect nis, Goodson, and Pearson 2018; Ryan 2018); the results of the experiments. We follow the this is most apparent in repeated geotags that recommendations of Berinsky and colleagues have low-quality responses and seem to (2014) and report results without excluding reflect overseas workers using U.S.-based participants based on attention screeners. The VPNs. We checked all four of our studies and alternative results are available on request; in found far fewer repeated geotags than in these addition, they can be replicated by changing reports, with almost all repeated geotags only line 7 of the replication file posted on OSF. being used twice. Excluding all respondents with repeated geotags yielded no substantive Procedure and materials. In all our changes to the results. We suspect our data do experiments, participants were recruited to an not suffer from these problems because we academic research study that involved read- completed our data collection by June 2017 ing a brief scenario and responding to several and restricted our recruiting to U.S.-based survey questions. In Study 1, participants users with excellent reputations (i.e., at least were randomly assigned to read one version 100 completed tasks and at least a 98 percent of a story about an interaction between a approval rating). While we are thus confident defense contractor and a congressman (Part A of the appropriateness of our AMT sample, of the online supplement). We changed the we encourage future research to validate our protagonists’ names, but our vignette is results using other types of samples, such as loosely based on the prosecution of Repre- online panels recruited through probability- sentative William Jefferson, a congressman based sampling (Freese and Visser 2010). best known for being caught by the FBI in Online panels have made experimental data 2005 with $90,000 in cash hidden in his collection affordable, fast, and efficient, but freezer (McCann and Redlawsk 2006; Peo- they come with a certain loss of experimental ples 2013). In all six versions of our vignette, control (Bitektine, Lucas, and Schilke 2018), the defense contractor asks for the congress- which is why future research conducted in a man’s help in endorsing his company’s equip- laboratory setting would be helpful. ment for testing by the Army, and the Study 1 had a total of 306 participants (137 congressman ultimately grants this request. women, 169 men), 252 (82.4 percent) of The specific versions of the vignette dif- whom identified as white (including multira- fered as follows. In the quid-pro-quo condi- cial people who are part white). Participants tion, the congressman asks for a $10,000 were U.S. residents ranging in age from 19 to campaign contribution, which the contractor 79 years (M = 35.33 years, SD = 12.04). The provides. In the culturally appropriate Schilke and Rossman 13

Figure 3. Mean Disapproval by Condition, Study 1 Note: Cohen’s d (relative to quid pro quo): culturally appropriate (–2.02), bundling (–.45), brokerage (–.46), pawning (–.70), gift (–1.15). Cohen’s d—the difference in two groups’ means divided by their pooled standard deviation—is a common measure of effect size for mean comparisons. Cohen (1988) suggests that d = |0.2| can be interpreted as a “small” effect size, |0.5| reflects a “medium” effect size, and |0.8| indicates a “large” effect size. condition, the congressman asks to borrow a moral disapproval, calculated as the simple pen, which he returns after using.8 In each of average of the two protagonists’ badness and the other four conditions, the contractor pro- punishability items (Cronbach α’s for this vides $10,000 to support the congressman’s measure ranged from .79 to .88 across Studies campaign but does so in a fashion following 1 to 4).9 Figure 3 summarizes the condition one of the four obfuscation structures out- means. Results of a t test revealed that disap- lined earlier: bundling, brokerage, gift proval was significantly lower in the culturally exchange, or pawning. After reading the appropriate condition (Mappropriate = 2.48) than vignette, participants were given a short sur- in the quid-pro-quo condition (Mquid pro quo = vey about their perceptions of the contractor 4.94). This substantial gap validates the fact and the congressman. To capture our depend- that we based the vignette on a culturally ent variable of moral disapproval, this survey inappropriate exchange, which is also true in included measures for badness and punisha- the subsequent studies. Moreover, the four bility that we adopted from Huebner, Lee, and obfuscation conditions were significantly Hauser (2010), anchored on seven-point lower in disapproval compared to the quid- answer scales (ranging from “very bad” to pro-quo condition (all p’s < .05): Mbundling = “very good” and “should be severely pun- 4.31, Mbrokerage = 4.37, Mpawning = 4.06, and ished” to “should not be punished at all”). Mgift exchange = 3.37. Moral disapproval of all Here and in all subsequent studies, partici- obfuscation structures fell in between that of pants then provided basic sociodemographic the appropriate exchange and quid-pro-quo information. Finally, participants were asked conditions (in most cases, it was closer to if they had any comments about the study. quid pro quo than to appropriate).10 These They were then given a code word to be empirical results provide support for Hypoth- entered in AMT in order to receive US$1 in esis 1.11 compensation. Discussion. Study 1 offers initial experi- Results. For the purpose of hypotheses mental support for our account that structural testing, we created a composite measure for obfuscation effectively mitigates audience 14 American Sociological Review 00(0) disapproval. Political bribery is a prime exam- Results. Figure 4 displays the condition ple of a disreputable exchange qua economic means. Consistent with Study 1, disapproval transaction, but it is certainly not the only one was significantly lower in the culturally

(Noonan 1984). To provide convergent evi- appropriate condition (Mappropriate = 2.45) than dence for Study 1’s finding (Lykken 1968) in the quid-pro-quo condition (Mquid pro quo = and to investigate whether it could be general- 4.72). Moreover, respondents showed signifi- ized to other types of exchange settings, we cantly less disapproval in three of the four designed two follow-up vignettes, the first of obfuscation conditions compared to the quid- which is about commercial bribery. pro-quo condition (all p’s < .05): Mbrokerage = 4.16, Mpawning = 3.57, and Mgift exchange = 3.81. Only bundling had a non-significant effect Study 2: Commercial Bribery (Mbundling = 4.45, t(100) = −1.09, p > .10, Participants. In this study, 304 U.S. resi- Cohen’s d = −.22). dents (143 women, 161 men) participated and completed all measures used in the analyses. Discussion. Study 2 offers further evi- Of these respondents, 244 (80.3 percent) dence that structural obfuscation can decrease identified as white. Participants ranged in age disapproval of a disreputable exchange. Note from 19 to 68 years (M = 34.68 years, SD = that the effect of bundling was not significant 11.62). The median respondent’s household in Study 2. In the context of this scenario, the earned between $30,000 and $39,999, and above-fair-value price the car dealer paid for 47.0 percent of participants had at least a col- the regional manager’s private used car may lege degree. have been too salient to hide its relationship to the favor the car dealer received in return. Materials. Participants were randomly To broaden our investigation from taboos on assigned to read one of six versions of a bribery, our next study relies on vignettes vignette (Part B of the online supplement) about a much more intimate issue: adoption that we adapted from an actual commercial with compensation for the birth mother. bribery RICO case. In that case, the federal government found that Honda dealers rou- Study 3: Baby-Selling tinely bribed executives at Honda USA head- quarters for access to popular models in the Participants. This study included 303 par- 1980s and early 1990s (Dunfee, Smith, and ticipants (138 women, 165 men), 250 (82.5 Ross 1999). These models sold at a premium percent) of whom identified as white. Partici- to the manufacturer’s sticker price, which is pants were U.S. residents ranging in age from why dealers’ demand for them exceeded the 19 to 74 years (M = 34.60 years, SD = 10.88). number of cars corporate was able to supply. The median annual household income was In all versions of our vignette, a Honda between $40,000 and $49,999, and 55.1 dealer has a meeting with a regional manager percent of participants had at least a college from Honda corporate, the dealer requests to degree. be allowed to buy scarce car models, and the regional manager ultimately approves his Materials. In Study 3, participants read a request. As in the political bribery vignettes, scenario describing a couple asking a new we used one vignette involving the explicit mother for permission to adopt her baby. transfer of $10,000 (quid-pro-quo condition), Despite the considerable industry in adoption one vignette in which one character briefly and assisted reproduction, substantial aspects loans the other a pen but no money changes of the process are obfuscated in real life, in hands (culturally appropriate condition), and part because buying babies or human tissues four vignettes involving an obfuscated trans- is illegal in most jurisdictions (Spar 2006). fer of $10,000. However, while the laws are clear and the Schilke and Rossman 15

Figure 4. Mean Disapproval by Condition, Study 2 Note: Cohen’s d (relative to quid pro quo): appropriate (−1.70), bundling (−.22), brokerage (−.41), pawning (−.92), gift (−.65).

Figure 5. Mean Disapproval by Condition, Study 3 Note: Cohen’s d (relative to quid pro quo): appropriate (−1.30), bundling (−.32), brokerage (−.25), pawning (−.79), gift (−.54). vast majority of people in modern societies quid-pro-quo condition, a culturally appropri- strongly believe that parenthood should not ate condition involving no transfers (aside be bought or sold (Ertman 2003), in practice from a light breakfast), and four conditions there is often only a fine line between baby- involving obfuscated transfers of $10,000.12 selling and legitimate adoption (Kapstein 2003). In all versions of the vignette (Part C of the Results. As shown in Figure 5, respond- online supplement), a couple has been trying ents assigned to the culturally appropriate unsuccessfully to have a baby when they (Mappropriate = 2.60), pawning (Mpawning = learn that a co-worker is seeking to give her 3.27), and gift exchange (Mgift exchange = 3.48) baby up for adoption, and they successfully conditions reported significantly less disap- convince the birth mother to let them adopt proval than those assigned to the quid-pro- her baby. As in the two previous studies, there quo condition (Mquid pro quo = 4.18), all are six versions of this vignette: a $10,000 p’s < .01. However, we did not find a 16 American Sociological Review 00(0)

significant effect for either bundling (Mbundling object of greater disapproval and that obfus- = 3.81) or brokerage (Mbrokerage = 3.89), cation is more important for justifying her p’s > .10. behavior than that of the adopting couple. To explore the specifics of these results This finding suggests that what is considered further, we conducted separate analyses for objectionable about an exchange may not the two parties involved, the birth mother and always be the exchange as a whole but rather the adopting parents (Part C, Figure S1 and a particular role within the exchange. Specifi- Table S1, in the online supplement). Interest- cally, the party who could be understood to ingly, results for the birth mother are largely have alienated her honor or intimate connec- consistent with those in the previous two tions may be perceived as more objectionable studies: not only pawning and gift exchange than the party who induces her to do so. This but also bundling and brokerage have signifi- interpretation is consistent with Noonan’s cant effects that are consistent with Hypothe- (1984) observation that the sender of mone- sis 1. However, when focusing on the adopting tary funds in disreputable exchanges may parents only, the results differ drastically: sometimes experience comparatively less none of the four obfuscation structures has a social disapproval than the receiver. Like- significant effect. This may be because even wise, people seem to be reluctant to blame in the quid-pro-quo condition, subjects show only those who seek a morally desirable end. In moderate disapproval of the adopting parents this case, the end goal of raising a child seems to foster empathy for the adopting parents, to (=M adoptingparents 3.59), quidpro quo and thus there is a the extent that even an explicit quid pro quo floor effect for obfuscation to have a signifi- is understood as essentially benign. cant legitimating effect within this limited That one party sometimes bears most of the range. Indeed, when the data are treated as stigma or risk of sanction from a taboo “long,” with vignette characters nested within exchange is consistent with the qualitative lit- respondents, both the main effect and all erature, which shows that, in many substan- interaction effects for vignette character are tive areas, one party seeks obfuscation to significant. As such, it appears that partici- avoid dishonor. For example, Clement (2006) pants judged the birth mother more harshly found that charity girls drew a strong moral and were more sensitive to obfuscation struc- boundary between transactional sex and pros- ture in evaluating her (Part C, Table A2, in the titution, but the men who treated them under- online supplement). Studies 1 and 2 similarly stood treating to be a cheap form of prostitution, had greater disapproval for the person receiv- with no real distinction of morality or honor. ing a bribe than the person giving a bribe, but This is consistent with a general double stand- the gap was only a third that seen in Study 3 ard in most cultures suggesting that women’s and the interaction effects were also much sexual honor is more fragile than men’s. Other smaller. examples of asymmetric honor and obfusca- tion transcend sex. For instance, record labels Discussion. Consistent with prior results, tended to be more obfuscatory about payola in Study 3 shows that structural obfuscation the 1980s than radio stations were, although diminishes disapproval of disreputable this is no longer the case (Dannen 1990; Ross- exchange—but only for pawning and gift man 2012). Likewise, among the Tiv people in exchange when considering both parties to Nigeria, the person who trades subsistence the exchange jointly. When we conducted goods gains honor from the person who trades separate analyses for the “seller” and “buyer” prestige goods, and the parties are proportion- sides of the exchange, we saw greater effec- ately furtive or boastful about the transaction tiveness of structural obfuscation for the birth (Bohannan 1955). In summary, who gets mother than for the adopting parents. It is blamed for an illicit transaction is not inherent noteworthy both that the birth mother is the in the nature of the good itself but rather in Schilke and Rossman 17

Figure 6. Mean Disapproval by Condition, Study 4 Note: Cohen’s d (relative to quid pro quo): appropriate (−1.79), bundling (−.61), brokerage (−.31), pawning (−.81), gift (−1.17). other contingencies; this can be explored in mediators between the presentation of the future experimental and historical research. scenario and the disapproval items for the Nonetheless, the results of Study 3 are dependent variable. We measured perceived generally in line with those of the previous attributional opacity by asking participants to studies. Despite notable exceptions, the gen- indicate their agreement with the statement, eral pattern of results across Studies 1, 2, and “The motives of the characters in the scenario 3 largely supports Hypothesis 1. Thus, our were unclear to me.” Perceived transactional- findings are consistent with our claim that ism was measured with the item, “Representa- structurally obfuscated exchange will miti- tive Baker receiving something of value gate audience disapproval (as compared to shaped his decision to write a letter to the quid-pro-quo exchange). One limitation in army’s lab.”13 Finally, we gauged perceived these studies is that they did not address the collective validity by including two items, one causal mechanisms underlying the effect of for Representative Baker and one for Mr. obfuscatory exchange. We therefore designed Campbell, asking participants to indicate their Study 4 to test our second hypothesis regard- agreement with the statement, “Most people ing mediating factors. in our society would find [Representative Baker’s/Mr. Campbell’s] behavior appropri- ate” ( = .88). These mediators were anchored Study 4: Mediating Mechanisms α on a five-point answer scale (ranging from Participants. In our final study, 604 U.S. “strongly disagree” to “strongly agree”) and residents (302 women, 302 men) participated, presented in counterbalanced order. 491 (81.3 percent) of whom identified as white. These participants ranged in age from Results. Consistent with our prior studies’ 19 to 79 years (M = 35.95 years, SD = results, participants assigned to the culturally 11.58). The median respondent had an annual appropriate condition reported significantly household income between $50,000 and less disapproval (Mappropriate = 2.59) than did $59,999, and 56.3 percent of participants had those assigned to the quid-pro-quo condition at least a college degree. (Mquid pro quo = 4.83). Moreover, as shown in Figure 6, disapproval was also significantly Materials. Study 4 used the experimental lower (all p’s < .05) in the four obfuscation procedures of Study 1 but added measures for conditions compared to the quid-pro-quo 18 American Sociological Review 00(0)

condition: Mbundling = 3.95, Mbrokerage = 4.43, noteworthy findings emerge from these Mpawning = 3.78, and Mgift exchange = 3.28. results. First, and most importantly, the spe- These results largely replicate those of Study cific indirect effect #4 (sequentially involving 1 and provide additional support for Hypoth- all three proposed mediators) is statistically esis 1. significant, lending empirical support to the We began testing Hypotheses 2A, 2B, and serial model structure proposed in Hypothesis 2C by running three simple mediation boot- 2D. That is, structural obfuscation is associ- strapping tests (Preacher and Hayes 2004; the ated with perceived attributional opacity, mediation tests were run with 5,000 bootstrap which in turn is associated with perceived samples using the PROCESS macro in SPSS; transactionalism, which then affects per- Hayes 2017), in which each of the three ceived collective validity, ultimately causing potential mediators was analyzed indepen- variation in moral disapproval. Moreover, the dently (Hayes 2017: Model 4).14 Using boot- mediating effect of perceived attributional strap mediation analysis to calculate the opacity on disapproval is not direct (i.e., the indirect effect has the advantage of greater specific indirect effect #1 in Table 3 is not statistical power without assuming multivari- significant). Rather, the effect occurs indi- ate normality in the sampling distribution rectly through perceived transactionalism and (Preacher and Hayes 2008). If the 95 percent collective validity (indirect effects 2, 3, and 4 bias-corrected confidence interval (95 percent in Table 3). Attributional opacity can thus be CI) for the parameter estimate did not contain considered a proximal mediator in the model. zero, then the respective indirect effect was Perceived transactionalism, the intermediate statistically significant, and mediation was mediator, is significant for three of the obfus- established. As shown in Figure 7, the indi- cation structures; its indirect effects 5 and 6 rect effect of obfuscation (versus quid pro (in Table 3) fall just short of statistical signifi- quo) on disapproval through perceived attri- cance at the 5 percent level in the case of butional opacity was significant in all four brokerage. Finally, collective validity can be obfuscation conditions, such that obfuscation considered a distal mediator, as it has a par- increased perceived attributional opacity and ticularly strong effect on disapproval (column perceived attributional opacity decreased dis- 5 in Table 2), consistent with earlier research approval. Likewise, perceived transactional- (Walker et al. 1988; Zelditch and Walker ism emerged as a significant mediator in all 1985). Collective validity is substantially four obfuscation conditions, with obfuscation involved in several of the indirect effects decreasing perceived transactionalism and (indirect effects 3, 4, 6, and 7 in Table 3), perceived transactionalism increasing disap- playing a key role in explaining disapproval proval. The third mediator, perceived collec- variations across conditions. tive validity, was significant in three out of It is worth noting that the statistical signifi- four obfuscation conditions; the indirect cance of the focal indirect effect #4, covering effect of obfuscation on disapproval through the series of all proposed mediators, suggests perceived collective validity was significant that none of the three mediators alone can be for bundling, pawning, and gift exchange, but considered “sufficient” in predicting moral not for brokerage. These results lend empiri- disapproval. Rather, a combination of the cal support to Hypotheses 2A and 2B and three mediators appears to be required to partial support to 2C. comprehensively explain how structural We then examined all three mediators obfuscation influences moral evaluations. simultaneously in a single serial-mediation Notably, the majority of the effect of struc- model (see schematic in Figure 2) and using tural obfuscation on moral disapproval is Hayes (2017: Model 6), as summarized in indirect and runs through the proposed medi- Tables 2 and 3 and discussed in greater detail ators. Depending on the condition, the ratio of in Part D of the online supplement. Several total indirect effect (i.e., the sum of all Schilke and Rossman 19

Figure 7. Simple Mediation Models, Study 4 Note: n = 501. Quid pro quo is the baseline condition. The statistics directly above the dashed arrow indicate the direct effect of structural obfuscation on disapproval, not accounting for the mediator. The statistics below the dashed arrow indicate the direct effect of structural obfuscation on disapproval, with the mediator included in the regression model. Asterisks denote significant coefficients p(† < .10; *p < .05; **p < .01; ***p < .001; two-tailed tests). 20 American Sociological Review 00(0)

Table 2. Serial-Mediation Regression Results, Study 4

3. 2. Perceived 4. 1. Attributional Transaction- Collective 5. Disapproval Opacity alism Validity Disapproval

Exchange Structure (Omitted: Quid Pro Quo) Bundling −.883*** .543*** −.219† .397** −.320* (.193) (.156) (.113) (.144) (.146) Brokerage −.404* .419** −.165 .017 −.193 (.192) (.156) (.112) (.142) (.144) Pawning −1.056*** .542*** −.440*** .401** −.372* (.191) (.155) (.112) (.144) (.147) Gift Exchange −1.554*** .698*** −.431*** .768*** −.554*** (.193) (.157) (.114) (.147) (.152) Mediators Perceived Attributional Opacity −.330*** .096* −.017 (.032) (.045) (.045) Perceived Transactionalism −.391*** .231*** (.057) (.060) Perceived Collective Validity −.765*** (.045) Constant 4.833*** 1.720*** 5.157*** 3.945*** 5.572*** (.136) (.110) (.096) (.318) (.369) R2 .135 .044 .235 .236 .527

Note: Standard errors in parentheses. n = 501. †p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests). specific indirect effects) to total effect ranges manipulated obfuscation structures. Further between 52.2 and 64.7 percent. analyses revealed a serial-mediation pattern, Overall, the mediation results support the whereby the effects of obfuscation unfold in a model depicted in Figure 2 and predicted in sequence beginning with perceived attribu- Hypothesis 2D. They may also help explain tional opacity, continuing with transactional- why brokerage represents the weakest of the ism, and finally affecting collective validity obfuscation structures across all the studies. before informing moral disapproval ratings. When controlling for perceived attributional Our mediation analyses are strongly support- opacity, the effects of brokerage on perceived ive of the proposed causal order. Nonetheless, transactionalism and on collective validity as discussed in Part D of the online supple- were both non-significant (columns 3 and 4 in ment, future research would be helpful to cor- Table 2). This suggests that the (compara- roborate the mediation results reported here. tively weak but significant) main effect of brokerage on disapproval can be largely explained only by perceived attributional Conclusions opacity and its downstream effects (as well This article has shown that respondents evince as, of course, factors that remained unob- moral disapproval of disreputable exchange, served in this study). that moral disapproval declines when a dis- reputable exchange is obfuscated, and that this Discussion. This experiment showed that effect is mediated by three cognitive pro- three mediators—perceived attributional opac- cesses: perceptions of attributional opacity, ity, transactionalism, and collective validity— transactionalism, and collective validity. Sup- explain a significant share of the effects of the porting Hypothesis 1, in each of the Schilke and Rossman 21

Table 3. Serial-Mediation Indirect Effects of Obfuscation on Disapproval, Study 4

95% CI Completely Ratio of Lower Upper Standardized Indirect to Indirect Effect Limit Limit Indirect Effect Total Effect

1: Exchange Structure→M1→Disapproval Bundling −.068 .038 −.002 .011 Brokerage −.053 .031 −.002 .018 Pawning −.067 .040 −.002 .009 Gift Exchange −.080 .054 −.003 .008

2: Exchange Structure→M1→M2→Disapproval Bundling −.087 −.015 −.009 .047 Brokerage −.070 −.009 −.007 .079 Pawning −.086 −.015 −.009 .039 Gift Exchange −.109 −.021 −.012 .034

3: Exchange Structure→M1→M3→Disapproval Bundling −.099 −.006 −.009 .045 Brokerage −.081 −.003 −.007 .076 Pawning −.095 −.007 −.009 .038 Gift Exchange −.118 −.007 −.011 .033

4: Exchange Structure→M1→M2→M3→Disapproval Bundling −.097 −.026 −.012 .061 Brokerage −.078 −.014 −.010 .102 Pawning −.093 −.026 −.012 .051 Gift Exchange −.120 −.036 −.015 .044

5: Exchange Structure→M2→ Disapproval Bundling −.116 −.010 −.012 .057 Brokerage −.108 .000 −.009 .094 Pawning −.205 −.040 −.023 .096 Gift Exchange −.191 −.040 −.022 .064

6: Exchange Structure→M2→M3→Disapproval Bundling −.137 −.012 −.015 .074 Brokerage −.111 .004 −.011 .122 Pawning −.222 −.068 −.030 .125 Gift Exchange −.216 −.066 −.028 .083

7: Exchange Structure→M3→ Disapproval Bundling −.548 −.066 −.069 .343 Brokerage −.215 .191 −.003 .031 Pawning −.542 −.085 −.070 .290 Gift Exchange −.831 −.361 −.128 .378

Note: n = 501. M1: perceived attributional opacity, M2: perceived transactionalism, M3: perceived collective validity. experiments, obfuscation structures tended to across several substantive areas: political brib- be more morally palatable than quid pro quos. ery, commercial bribery, and compensated Our theory identified four distinct forms of adoption. As such, our four experiments pro- obfuscation, which—despite their differences vide strong convergent evidence in support of in terms of structure—we show reduce audi- the proposed main effect, which is largely ences’ moral disapproval of disreputable robust across different obfuscation structures, exchanges. Notably, our hypothesis held operationalizations, and exchange contexts. 22 American Sociological Review 00(0)

Another noteworthy aspect of this research brokerage less effective than they are in pertains to the causal process model. In this nature? model, we proposed three intervening causal One possibility is that bundling and broker- mechanisms to produce the abovementioned age are simply easier to achieve. Obfuscation effect: perceived attributional opacity, transac- not only conceals the intent to accomplish a tionalism, and collective validity. Structural transaction through a set of transfers but also obfuscation makes it more difficult for audi- transforms the set of transfers, especially in ences to clearly ascribe actors’ motivations the case of gift exchange and pawning. This is and intentionality in the exchange; it calls into something we learned about obfuscation just question the causal link between the quid and by writing the vignettes used in our studies, the quo; it also creates a sense that the observed even before showing them to the study partici- behavior is more common, and likely more pants. We hit several dead ends in drafting legal, than a comparable quid-pro-quo scenarios, as not all empirical contexts are exchange. Overall, our article significantly compatible with all four obfuscation struc- expands the literature on obfuscation by pro- tures. In particular, writing vignettes involving viding empirical insight into the effects on gift exchange and pawning required us to give moral evaluations, as well as by adding knowl- the characters a preexisting relationship. In edge of causal mechanisms—a fundamental contrast, nothing in the nature of bundling and component for theory building and enrich- brokerage prevents them from operating well ment (Merton 1949; Stinchcombe 1991). in a spot market. The fact that gift exchange Revisiting our empirical findings in greater and pawning involve delay requires not only detail, it is noteworthy that Hypothesis 1 was managing relationships and forgoing instant more strongly supported for some obfuscation gratification, but also carrying a formal or structures than for others. In particular, gift informal debt, along with whatever implicit exchange and pawning were always the most financing costs and risk this may involve effective obfuscation structures in mitigating (Bourdieu 1990). Carrying an implicit or moral disapproval. Bundling and brokerage explicit debt requires liquidity until reciproc- were relatively less effective, being significant ity is achieved and comes with the risk that in some cases and falling below significance this may never occur. In this sense, bundling in others. However, regardless of whether and brokerage are less effective than gift their effects were statistically distinguishable exchange and pawning, but they are also from quid pro quo, they were always weaker cheaper, in terms of being easier to accom- than the effects of gift exchange and pawning. plish or carrying lower risk or tacit interest. If This was unexpected, as an inductive reading this is the case, then someone might be willing of the qualitative data shows that all obfusca- to obfuscate through bundling or brokerage tion structures are common, not that one or even with the knowledge that the more costly two are systematically more effective than the options of gifts and pawning would be more others (Rossman 2014). effective. Notably, Hoang (2018) found that in Given these notable differences, we face Vietnam, foreign real estate investors lack the the puzzle of why all four obfuscation struc- patience or tact to bribe public officials with tures are ubiquitous in qualitative cases, gifts and instead rely on bundling and broker- despite the fact that our experimental work age, which puts them at a disadvantage versus found some of them to be relatively ineffec- gift-giving local developers. However, Hoang tive at mitigating moral disapproval. In other (2018) also found that brokerage (and to a words, has this article identified the practical lesser extent, bundling) was extremely effec- upshot that people who use bundling or bro- tive in maintaining the reputations of foreign kerage are committing economic transgres- real estate investors, who were seen as more sions ineffectively? Or, alternatively, did our ethical and less cynical than local gift-based experiments artifactually render bundling and investors. Schilke and Rossman 23

As suggested by Hoang’s (2018) study, even though all the facts were clearly laid out bundling and brokerage may be comparably in the course of a few hundred words. This effective to gift exchange and pawning in effect through perceived attributional opacity real-world situations. If so, something about should be much more efficacious in real life, our use of relatively terse vignette experi- where facts are generally not presented in as ments may have suppressed this fact. Most terse a fashion as they are in vignettes and notably, obfuscation can work in the real may not even be publicly available at all. For world by rendering the interaction difficult to this reason, our vignette methodology sug- observe. Our mediation analyses in Study 4 gests a conservative bias for the efficacy of testing Hypotheses 2A, 2B, and 2C shed obfuscation. For instance, in the bundling some light on the differential effects of the condition of Studies 1 and 4, one reads that four obfuscation structures and how they are the lobbyist not only leased office space to the shaped by observation. Specifically, these congressman but had also given him a bargain analyses revealed that only two of the three rent. In real life, the perceived attributional cognitive-process mediators were significant opacity—and by extension, obfuscation—in for brokerage. Brokerage affected moral dis- this case would be almost total, because these approval through attributional opacity (i.e., facts would not naturally occur in the same difficulty in clearly deciphering transactants’ brief document. Rather, an accuser would exchange motivation) and transactionalism have to somehow notice in the sea of paper (i.e., the perception of a direct causal relation produced by a congressional office that the between the quid and the quo). However, its office rent listed in campaign finance filings indirect effect through collective validity was is below market and that the landlord is part not statistically significant. We found similar of the same firm recommended in a letter to results for bundling: although all three indi- the Army. Among the most merciful things in rect effects were significant, the mediating the world is the inability of audiences to our roles of perceived transactionalism and col- actions to correlate their contents. Our actions lective validity were relatively weak. These might be scandalous if they were understood findings provide initial insight into the dis- in context, but the lack of joint salience often tinct consequences of different modes of prevents this comprehension, and by exten- obfuscation. Some obfuscation structures sion, disapproval. (i.e., pawning and gift exchange) substan- This is consistent with a common pattern tially activate all three cognitive processes, in qualitative data: bundling and brokerage whereas others (i.e., bundling and brokerage) are extremely effective at obfuscation until work by increasing perceived attributional attention is drawn to them, at which point opacity while not appreciably invoking per- they become scandalous (Faulkner 2011). For ceived transactionalism or collective validity. instance, Lyndon Johnson’s obfuscation of These results suggest a possible reason taking bribes bundled with radio advertising why bundling and brokerage are common in was extremely effective until Caro accused qualitative data, even though they appeared to him of having obfuscated corruption, at which be relatively ineffective at mitigating moral point Johnson’s historical memory took a disapproval in our experiments. Specifically, substantial hit. This suggests a possible ave- bundling and brokerage may have obfusca- nue for future work to further disentangle the tory power simply because they make the scope conditions for obfuscation. For exam- complete circuit of exchange harder to ple, obfuscation should be comparatively less observe. However, they may be relatively effective given an accusation denouncing it as ineffective at making the set of transfers tantamount to quid pro quo. Conversely, appear less transactional or collectively valid decreasing the temporal coincidence of infor- once they are observed. In Study 4, bundling mation about individual transfers may further and brokerage reduced attributional opacity mitigate disapproval by complicating the 24 American Sociological Review 00(0) identification of how those transfers come 2012). This research program has drawn atten- together to create the overall obfuscated tion to contested valuation of goods (e.g., exchange structure. Almeling 2011; Healy 2006; Zelizer 1985, Our empirical work in this article was 2012). For the most part, this literature either designed to test and extend a theory of highlights the question of which commodities exchange obfuscation and to contribute to the can (and cannot) be exchanged through mar- social psychology of morality as well as to kets or emphasizes the details of negotiating institutional theory as it pertains to legitimacy commensuration. However, the central impulse judgments; however, its findings also speak to of this literature has largely been a critique of other literatures. For instance, although we reductionist accounts rather than a positive focus on the abstract patterns and commonali- identification of relevant schemata. Settling for ties across different areas of exchange, we an end goal of showing how rich and compli- expect this study can provide conceptual tools cated exchange can be, extant relational work and research questions to those interested spe- sociology has stopped short of revealing sys- cifically in substantive areas of morally con- tematic patterns (cf. Healy 2013). In this arti- tested exchange, especially accusation and cle, we go beyond this limited ambition of contestation of gray areas. This article pro- critiquing reductionism and providing nuance vides a theoretical tool to apply to such areas to show specific ways that structuring disrepu- as political bribery, commercial bribery, and table exchange effectively mitigates its moral reproductive markets (Åkerström 2014; offense. Our experiments identify a set of sys- Faulkner 2011; Fridman and Luscombe 2017; tematic patterns to obfuscating exchange that Hoang 2018; Rossman 2012; Spar 2006). help relieve the audience’s discomfort, render- Beyond the settings already covered in our ing what would be an offensive quid pro quo vignettes, we expect similar patterns can be into merely an innocuous set of transfers. found in contexts such as pharmaceutical detailing, transactional sex/flirtation, Islamic Acknowledgments finance, and investor-citizenship (Hoang 2015; The authors gratefully acknowledge the comments and King and Bearman 2017; Kuran 2004; Mears contributions provided by Aleksander Ellis, Alan Fiske, 2015; Surak 2016; Swidler and Watkins 2007; Sarah Harkness, Linda Molm, Martin Reimann, Reihan Zelizer 2005). As the literature suggests, there Salam, three anonymous reviewers, the ASR editors, and is not a clear line between reputable and dis- participants of the Innovation and Creativity Workshop at reputable exchange. Rather, audiences to UCLA Anderson School of Management, the 2016 Southern California Symposium on Network Economics ambiguous transfers must actively develop an and Game Theory at UCLA, the Organizational Behavior understanding of whether to tolerate or to Seminar at Stanford GSB, the Duke University Depart- accuse. Our work offers a systematic and gen- ment of Sociology Colloquium, and the 2017 American eralized typology to this largely qualitative Sociological Association Regular Session on Economic literature. In turn, further qualitative work can Sociology. reveal the real-world details and contingencies for actors to enact these structures and for Funding audiences to evaluate them. Preparation of this article was supported in part by a In the early twentieth century, social sci- Research Small Grant from the Center for Leadership ence took the distinction between market and Ethics, Eller College of Management to the first author non-market exchange as a central research and a UCLA Senate Faculty Research Grant to the sec- ond author. question (Mauss [1923] 1967; Weber [1922] 1978). However, this issue lay largely dormant in American sociology for several decades Notes until reinvigorated with a new wave of inter- 1. Rossman (2014) only describes “pawning” in a est—specifically, the fertile new paradigm of footnote as a variant of gift exchange, but in this relational work (e.g., Bandelj 2012; Block article we treat it as a distinctive form. As we will Schilke and Rossman 25

discuss, these two structures differ primarily in Obfuscating bribes was insufficient to keep John- that pawning involves explicit debts, whereas gift son’s historical memory free of the knowledge of exchange involves sublimated obligations. corruption, but it was effective long enough to let 2. By definition, non-market goods cannot legitimately Johnson advance to the Senate and the presidency. be exchanged for money. However, it may or may Even today, corruption is not the major focus of the not be licit to commensurate non-markets goods with historical memory of Johnson. We can also consider each other. For instance, it is culturally prohibited to cases where obfuscation is only partially effective, buy sex or political favors and to trade sex for politi- even contemporaneously. The fact that obfuscation cal favors, but sex can be exchanged for sex and is practiced may give plausible deniability to sup- political favors for political favors. See the discus- porters, who can choose to ignore transgressions sion between Rossman and Tabarrok in Rossman and more readily than they could in the case of an open colleagues (2016). Furthermore, in some cultures quid quo pro. Indeed, this fact pattern of contempo- special monies are designated for transfers involving rary public accusations that are rejected by support- human life: slavery, betrothals, and wrongful death ers in the face of ambiguity describes how Johnson suits (Bohannan 1955; Graeber 2011). got away with stealing the 1948 Texas primary and 3. Estimating the exact magnitude of these markets is how President Trump has largely sloughed off accu- unfeasible, both because they are furtive by nature sations of conflicts of interest. and because the category of disreputable exchange 6. Note that all the mediated pathways involve a com- is not yet the basis for aggregating systematic data. bination of a positive and a negative partial effect. Most analyses of illicit or shadow economies do not Because the product of a positive figure and a nega- distinguish between markets that are illegal because tive figure is a negative figure, hypotheses 2A, 2B, they are understood as markets (e.g., prostitution and 2C all imply negative indirect effects of obfus- and bribery) versus markets in contraband (e.g., cation on moral disapproval. drugs and unlicensed firearms). Moreover, discrep- 7. We experimented with supplementary analyses, ancy models find it difficult to distinguish black including additive effects and interactions with markets from the informal economy (e.g., babysit- exchange structure, for political party preference, ting) (Dank et al. 2014). Nonetheless, we can get a gender, college education, and age. About the same vague sense of scale for some illicit markets that number of these additional effects are statistically are only illicit by virtue of being understood as significant as one would expect to see by chance. markets. A recent analysis estimated that in seven 8. We had one character borrow a pen from the other in U.S. cities, prostitution accounted for between the culturally appropriate condition in order to have .03 percent (Denver) and .11 percent (Atlanta) of filler text where the other conditions describe $10,000 local GDP, or slightly larger than the market for transfers; this kept the vignettes at a similar length. illegal drugs (Dank et al. 2017). It is even more 9. In addition to our disapproval metric combining the difficult to estimate the volume of public sector “behave” and “punish” responses regarding the char- bribery, but a lower-bound estimate is suggested acters in the story, we also collected several questions by the approximately 570 official corruption con- reflecting moral outrage (Kennedy and Kray 2014). victions per year throughout the United States in These questions asked to what extent the respondent the past three decades (Cordis and Milyo 2016). felt the scenario was disgusting, objectionable, upset- Public sector corruption is an even more important ting, offensive, shameful, contemptible, and morally issue in emerging markets like Vietnam, where it acceptable (reverse-coded). The disapproval and completely structures such industries as real estate moral outrage indices were strongly correlated (r’s development (Hoang 2018). ranged from .70 to .81 across the studies). Regres- 4. If estimating the volume of disreputable exchange sion results using the two measures had highly simi- is difficult, estimating the volume of obfuscated lar model fit statistics, effect sizes, and coefficient exchange is impossible, as it is not only furtive but significance levels. We emphasize reporting of the inherently contentious. Nonetheless, we can make “behave” and “punish” disapproval metric, as it some estimates. The total volume of payola around allows us to disaggregate reactions to the interacting both 1980 (when it was obfuscated via brokerage) parties specifically rather than to the interaction as and 2004 (when it was obfuscated via brokerage a whole, which proves important in Study 3. More- and gifts) was about $150 to $250 million per year over, choosing the disapproval over the outrage met- (in 2018 real dollars) (Rossman 2012). ric as our primary dependent variable is consistent 5. Caro’s portrait of Johnson is a good illustration of with the cognitive (rather than emotional) emphasis how obfuscation is only available to the analyst of our theoretical argument. Nonetheless, we include if it fails. However, it is worth noting that John- both metrics in the data archive. son’s obfuscations were very successful for a very 10. As will be seen, in Studies 2, 3, and 4 obfuscation long time. Caro’s accusation came 50 years after also falls closer to quid pro quo than to appropriate Johnson’s underlying transgressions. This means transfers. In the conclusion, we discuss how this may Johnson escaped scrutiny for his entire lifetime. reflect the conservative bias of our methodology. 26 American Sociological Review 00(0)

11. To test robustness, we collected an index for Berinsky, Adam J., Michele F. Margolis, and Michael perceived realism (the average of “the scenario W Sances. 2014. “Separating the Shirkers from the sounded phony” [reverse-coded], “the scenario Workers? Making Sure Respondents Pay Attention sounded realistic,” and “things actually happen in on Self-Administered Surveys.” American Journal of the real world that are similar to the scenario”). Post Political Science 58(3):739–53. hoc analyses using regressions showed that results Bitektine, Alex, and Patrick Haack. 2015. “The ‘Macro’ in all four vignette studies are robust to including and the ‘Micro’ of Legitimacy: Toward a Multilevel this index in the models. Theory of the Legitimacy Process.” Academy of Man- 12. The pawning condition in Study 3 involves medical agement Review 40(1):49–75. debt incurred in the delivery. To many audiences, Bitektine, Alex, Jeffrey W. Lucas, and Oliver Schilke. this medical debt may feel more legitimately tied to 2018. “Institutions under a Microscope: Experimen- custody of the baby than would debt the birth mother tal Methods in Institutional Theory.” Pp. 147–67 in had incurred for other reasons. This suggests cau- Unconventional Methodology in Organization and tion in interpreting the pawning effect in Study 3. Management Research, edited by A. Bryman and D. We thank an anonymous reviewer for flagging this A Buchanan. Oxford, UK: Oxford University Press. issue after our data collection was complete. Block, Fred. 2012. “Relational Work in Market Econo- 13. Studies 1, 2, and 3 also collected perceived transac- mies.” Politics & Society 40(2):135–44. tionalism, but not our other mediators. The effects Bohannan, Paul. 1955. “Some Principles of Exchange of perceived transactionalism in Studies 1, 2, and 3 and Investment among the Tiv.” American Anthro- are similar to those in Study 4. We do not present pologist 57(1):60–70. them here in order to maintain thematic focus on Bourdieu, Pierre. 1990. The Logic of Practice. Cam- mediators in Study 4 and to avoid the impression bridge, UK: Polity. that perceived transactionalism is more important Bourdieu, Pierre. 2000. Pascalian Meditations. Stanford, than the other mediators. CA: Stanford University Press. 14. Consistent with Hypotheses 2A, 2B, 2C, and 2D, Buhrmester, Michael, Tracy Kwang, and Samuel D. the mediation analyses reported here focused on Gosling. 2011. “Amazon’s Mechanical Turk: A New comparing the quid pro quo to the four obfuscation Source of Inexpensive, yet High-Quality, Data?” Per- conditions; thus, they did not incorporate data on spectives on Psychological Science 6(1):3–5. the appropriate condition, which was not central to Caro, Robert A. 1990. Means of Ascent: The Years of Lyn- the causal logic of our theory. However, we also ran don Johnson. New York: Alfred A. Knopf. all mediation analyses while including the appropri- Clement, Elizabeth Alice. 2006. Love for Sale: Court- ate condition, and the results pertaining to Hypoth- ing, Treating, and Prostitution in New York City, eses 2A, 2B, 2C, and 2D are robust. 1900–1945. Chapel Hill: University of North Caro- lina Press. Cohen, Jacob. 1988. Statistical Power Analysis for the ORCID iDs Behavioral Sciences. Hillsdale, NJ: Lawrence Erl- Oliver Schilke https://orcid.org/0000-0001-6832-1677 baum Associates. Collett, Jessica L., and Omar Lizardo. 2014. “Localizing Gabriel Rossman https://orcid.org/0000-0002-3556-0601 Cultural Phenomena by Specifying Social Psycho- logical Mechanisms.” Social Psychology Quarterly References 77(2):95–99. Aguinis, Herman, and Kyle J. Bradley. 2014. “Best Prac- Cordis, Adriana S., and Jeffrey Milyo. 2016. “Measur- tice Recommendations for Designing and Implement- ing Public Corruption in the United States: Evidence ing Experimental Vignette Methodology Studies.” from Administrative Records of Federal Prosecu- Organizational Research Methods 17(4):351–71. tions.” Public Integrity 18(2):127–48. Åkerström, Malin. 2014. Suspicious Gifts: Bribery, Dank, Meredith, Bilal Khan, P. Mitchell Downey, Cybele Morality, and Professional Ethics. New Brunswick, Kotonias, Deborah Mayer, Colleen Owens, Laura NJ: Transaction Publishers. Pacifici, and Lilly Yu. 2014. Estimating the Size and Almeling, Rene. 2011. Sex Cells: The Medical Market for Structure of the Underground Commercial Sex Econ- Eggs and Sperm. Berkeley: University of California omy in Eight Major US Cities. Washington, DC: Urban Press. Institute. Retrieved May 25, 2018 (https://www.urban. Anteby, Michel. 2010. “Markets, Morals, and Practices of org/sites/default/files/publication/22376/413047- Trade: Jurisdictional Disputes in the U.S. Commerce estimating-the-size-and-structure-of-the-underground- in Cadavers.” Administrative Science Quarterly commercial-sex-economy-in-eight-major-us-cities. 55(4):606–638. pdf). Bandelj, Nina. 2012. “Relational Work and Economic Dannen, Fredric. 1990. Hit Men: Power Brokers and Fast Sociology.” Politics & Society 40(2):175–201. Money inside the Music Business. New York: Times Books. Barnes, Gerald. 2014. “Untitled Pastoral Letter on DellaVigna, Stefano, Ruben Durante, Brian Knight, and Annulment.” Retrieved July 6, 2015 (https://goo.gl/ Eliana La Ferrara. 2016. “Market-Based Lobby- eK85tF). ing: Evidence from Advertising Spending in Italy.” Schilke and Rossman 27

American Economic Journal: Applied Economics Haidt, Jonathan, and Selin Kesebir. 2010. “Morality.” 8(1):224–56. Pp. 797–832 in Handbook of Social Psychology, Dennis, Sean A., Brian M. Goodson, and Chris Pearson. edited by S. T. Fiske and D. T. Gilbert. Hobeken, NJ: 2018. “Mturk Workers’ Use of Low-Cost ‘Virtual Wiley. Private Servers’ to Circumvent Screening Meth- Harvey, Paul, Kristen Madison, Mark Martinko, T. Rus- ods: A Research Note.” Retrieved August 25, 2018 sell Crook, and Tamara A. Crook. 2014. “Attribution (https://papers.ssrn.com/sol3/papers.cfm?abstract_ Theory in the Organizational Sciences: The Road id=3233954). Traveled and the Path Ahead.” Academy of Manage- Dornbusch, Sanford M., and W. Richard Scott. 1975. ment Perspectives 28(2):128–46. Evaluation and the Exercise of Authority. San Fran- Hayes, Andrew F. 2017. Introduction to Mediation, Mod- cisco, CA: Jossey-Bass. eration, and Conditional Process Analysis: A Regres- Dunfee, Thomas W., N. Craig Smith, and William T. sion-Based Approach. New York: Guilford Press. Ross. 1999. “Social Contracts and Marketing Ethics.” Healy, Kieran. 2006. Last Best Gifts: Altruism and the Journal of Marketing 63(3):14–32. Market for Human Blood and Organs. Chicago: Uni- Emerson, Richard M. 1976. “Social Exchange Theory.” versity of Chicago Press. Annual Review of Sociology 2(1):335–62. Healy, Kieran. 2013. “Dealing with Awkward Relations.” Emerson, Richard M. 1981. “Social Exchange Theory.” Retrieved August 17, 2017 (https://kieranhealy.org/ Pp. 30–65 in Social Psychology: Sociological Per- files/papers/awkward-relations.pdf). spectives, edited by M. Rosenberg and R. H Turner. Hegtvedt, Karen A., and Heather L. Scheuerman. 2010. New York: Basic Books. “The Justice/Morality Link.” Pp. 331–60 in Hand- Ertman, Martha M. 2003. “What’s Wrong with a Parent- book of the Sociology of Morality, edited by S. Hitlin hood Market? A New and Improved Theory of Com- and S. Vaisey. New York: Springer. modification.” North Carolina Law Review 82(1). Hoang, Kimberly Kay. 2015. Dealing in Desire: Asian Espeland, Wendy Nelson, and Mitchell L. Stevens. 1998. Ascendancy, Western Decline, and the Hidden Cur- “Commensuration as a Social Process.” Annual rencies of Global Sex Work. Oakland: University of Review of Sociology 24(1):313–43. California Press. Faulkner, Robert R. 2011. Corporate Wrongdoing and the Hoang, Kimberly Kay. 2018. “Risky Investments: How Art of the Accusation. London, UK: Anthem Press. Local and Foreign Investors Finesse Corruption-Rife Finch, Janet. 1987. “The Vignette Technique in Survey Emerging Markets.” American Sociological Review Research.” Sociology 21(1):105–114. 83(4):657–85. Fiske, Alan P. 1992. “The Four Elementary Forms of Huebner, Bryce, James J. Lee, and Marc D. Hauser. 2010. Sociality: Framework for a Unified Theory of Social “The Moral-Conventional Distinction in Mature Relations.” Psychological Review 99(4):689–723. Moral Competence.” Journal of Cognition and Cul- Fiske, Alan P., and Philip E. Tetlock. 1997. “Taboo Trade- ture 10(1):1–26. Offs: Reactions to Transactions That Transgress the Johnson, Cathryn, Timothy J. Dowd, and Cecilia L. Spheres of Justice.” Political Psychology 18(2):255– Ridgeway. 2006. “Legitimacy as a Social Process.” 97. Annual Review of Sociology 32:53–78. Freese, Jeremy, and Penny Visser. 2010. “Data Collected Kapstein, Ethan B. 2003. “The Baby Trade.” Foreign by Time-Sharing Experiments for the Social Sci- Affairs 82(6):115–25. ences.” NSF Grant 0818839. Kelley, Harold H. 1971. “Moral Evaluation.” American Fridman, Daniel G., and Alexander Luscombe. 2017. Psychologist 26(3):293–300. “Gift-Giving, Disreputable Exchange, and the Man- Kennedy, Jessica A., and Laura J. Kray. 2014. “Who Is agement of Donations in a Police Department.” Willing to Sacrifice Ethical Values for Money and Social Forces 96(2):507–528. Social Status? Gender Differences in Reactions to Graeber, David. 2011. Debt: The First 5,000 Years. New Ethical Compromises.” Social Psychological and York: Melville House. Personality Science 5(1):52–59. Granovetter, Mark. 2007. “The Social Construction of King, Marissa, and Peter S. Bearman. 2017. “Gifts and Corruption.” Pp. 152–72 in On Capitalism, edited Influence: Conflict of Interest Policies and Pre- by V. Nee and R. Swedberg. Stanford, CA: Stanford scribing of Psychotropic Medications in the United University Press. States.” Social Science & Medicine 172:153–62. Haack, Patrick, William McKinley, Oliver Schilke, and Krawiec, Kimberly D. 2009. “Show Me the Money: Lynne G. Zucker. 2016. “The Theory-Method Gap Making Markets in Forbidden Exchange.” Law and in Organizational Legitimacy Research: A Critical Contemporary Problems 72(3):i–xiv. Review, Synthesis, and Directions for Future Research.” Kuran, Timur. 2004. Islam and Mammon: The Economic Presented at the American Sociological Association Predicaments of Islamism. Princeton, NJ: Princeton Annual Meeting, August 20–23, Seattle, WA. University Press. Hahl, Oliver, Ezra W. Zuckerman, and Minjae Kim. Lainer-Vos, Dan. 2013. “The Practical Organization of 2017. “Why Elites Love Authentic Lowbrow Culture: Moral Transactions: Gift Giving, Market Exchange, Overcoming High-Status Denigration with Outsider Credit, and the Making of Diaspora Bonds.” Socio- Art.” American Sociological Review 82(4):828–56. logical Theory 31(2):145–67. 28 American Sociological Review 00(0)

Lawler, Edward J., and Lena Hipp. 2010. “Corruption Paddock, Richard C., Eric Lipton, Ellen Barry, Rod as Social Exchange.” Pp. 269–96 in Advances in Nordland, Danny Hakim, and Simon Romero. 2016. Group Processes, Vol. 27, edited by S. R. Thye and “Potential Conflicts around the Globe for Trump, E. J. Lawler. Bingley, UK: Emerald Group Publish- the Businessman President.” The New York Times. ing Limited. Retrieved September 29, 2017 (https://www.nytimes. Lucas, Jeffrey W., and Michael J. Lovaglia. 2006. “Legit- com/2016/11/26/us/politics/donald-trump-international- imation and Institutionalization as Trust-Building: business.html). Reducing Resistance to Power and Influence in Orga- Peoples, Clayton D. 2013. “Campaign Finance and nizations.” Pp. 229–52 in Advances in Group Pro- Policymaking: PACs, Campaign Contributions, and cesses, Vol. 23, edited by S. R. Thye and E. J. Lawler. Interest Group Influence in Congress.” Sociology Bingley, UK: Emerald Group Publishing Limited. Compass 7(11):900–913. Lykken, David T. 1968. “Statistical Significance in Pileggi, Nicholas. 1985. Wiseguy: Life in a Mafia Family. Psychological Research.” Psychological Bulletin New York: Simon and Schuster. 70(3):151–59. Pinker, Steven, Martin A. Nowak, and James J. Lee. Malle, Bertram F., and Joshua Knobe. 1997. “The Folk 2008. “The Logic of Indirect Speech.” Proceedings Concept of Intentionality.” Journal of Experimental of the National Academy of Sciences 105(3):833–38. Social Psychology 33(2):101–121. Preacher, Kristopher J., and Andrew F. Hayes. 2004. Marsden, Peter V. 1982. “Brokerage Behavior in “SPSS and SAS Procedures for Estimating Indi- Restricted Exchange Networks.” Pp. 201–218 in rect Effects in Simple Mediation Models.” Behav- Social Structure and Network Analysis, edited by P. ior Research Methods, Instruments, & Computers V. Marsden and N. Lin. Beverly Hills, CA: Sage. 36(4):717–31. Mauss, Marcel. [1923] 1967. The Gift: Forms and Func- Preacher, Kristopher J., and Andrew F. Hayes. 2008. tions of Exchange in Archaic Societies. Translated by “Asymptotic and Resampling Strategies for Assessing I. Cunnison. New York: Norton. and Comparing Indirect Effects in Multiple Mediator McCann, James A., and David P. Redlawsk. 2006. “As Models.” Behavior Research Methods 40(3):879–91. Voters Head to the Polls, Will They Perceive a ‘Cul- Quinn, Sarah. 2008. “The Transformation of Morals in ture of Corruption?’” PS: Political Science and Poli- Markets: Death, Benefits, and the Exchange of Life tics 39(4):797–802. Insurance Policies.” American Journal of Sociology Mears, Ashley. 2015. “Working for Free in the VIP: Rela- 114(3):738–80. tional Work and the Production of Consent.” Ameri- Rossman, Gabriel. 2012. Climbing the Charts: What can Sociological Review 80(6):1099–1122. Radio Airplay Tells Us About the Diffusion of Innova- Merton, Robert K. 1949. “On Sociological Theories of tion. Princeton, NJ: Princeton University Press. the Middle Range.” Pp. 39–53 in Social Theory and Rossman, Gabriel. 2014. “Obfuscatory Relational Work Social Structure, edited by R. K. Merton. Glencoe, and Disreputable Exchange.” Sociological Theory IL: Free Press. 32(1):43–63. Meyer, John W., and Brian Rowan. 1977. “Institution- Rossman, Gabriel, Michael C. Munger, Alan Page Fiske, alized Organizations: Formal Structure as Myth and Alex Tabarrok. 2016. “The Exchanges We Hide.” and Ceremony.” American Journal of Sociology Cato Unbound. Retrieved May 10, 2018 (https:// 83(2):340–63. www.cato-unbound.org/print-issue/2080). Molm, Linda D. 2003. “Theoretical Comparisons of Roth, Alvin E. 2007. “Repugnance as a Constraint Forms of Exchange.” Sociological Theory 21(1):1–17. on Markets.” Journal of Economic Perspectives Molm, Linda D., Jessica L. Collett, and David R. Schae- 21(3):37–58. fer. 2007. “Building Solidarity through Generalized Ryan, Timothy J. 2018. “Data Contamination on MTurk.” Exchange: A Theory of Reciprocity.” American Jour- Retrieved August 25, 2018 (http://timryan.web.unc nal of Sociology 113(1):205–242. .edu/2018/08/12/data-contamination-on-mturk/). Molm, Linda D., and Karen S. Cook. 1995. “Social Sales, Nancy Jo. 2016. “Daddies, ‘Dates,’ and the Girl- Exchange and Social Networks.” Pp. 209–235 in friend Experience: Welcome to the New Prostitu- Sociological Perspectives on Social Psychology, tion Economy.” Vanity Fair. Retrieved December edited by K. S. Cook, G. A. Fine, and J. S. House. 12, 2016 (http://www.vanityfair.com/style/2016/07/ Boston, MA: Allyn & Bacon. welcome-to-the-new-prostitution-economy). Munger, Michael C. 2011. “Euvoluntary or Not, Sandel, Michael J. 2000. “What Money Can’t Buy: Exchange Is Just.” Social Philosophy and Policy The Moral Limits of Markets.” Tanner Lectures on 28(2):192–211. Human Values 21:87–122. Mutz, Diana C. 2011. Population-Based Survey Experi- Satz, Debra. 2010. Why Some Things Should Not Be ments. Princeton, NJ: Princeton University Press. for Sale: The Moral Limits of Markets. New York: Nelson, Benjamin. 1969. The Idea of Usury, from Tribal Oxford University Press. Brotherhood to Universal Otherhood. Chicago: Uni- Schilke, Oliver. 2018. “A Micro-Institutional Inquiry into versity of Chicago Press. Resistance to Environmental Pressures.” Academy of Noonan, John T. 1984. Bribes. New York: Macmillan. Management Journal 61(4):1431–66. Schilke and Rossman 29

Sniderman, Paul M., and Douglas B. Grob. 1996. “Inno- of an Online Factorial Survey between a Population- vations in Experimental Design in Attitude Surveys.” Based and a Crowdsource-Recruited Sample.” Socio- Annual Review of Sociology 22:377–99. logical Science 1:292–310. Spar, Debora L. 2006. The Baby Business: How Money, Weiner, Bernard. 1986. An Attribution Theory of Motiva- Science, and Politics Drive the Commerce of Concep- tion and Emotion. New York: Springer. tion. Boston: Harvard Business School Press. Willer, Robb, Matthew Feinberg, and Rachel Wetts. Stets, Jan E., and Michael J. Carter. 2011. “The Moral 2016. “Threats to Racial Status Promote Tea Party Self: Applying Identity Theory.” Social Psychology Support among White Americans.” Retrieved August Quarterly 74(2):192–215. 26, 2016 (http://papers.ssrn.com/abstract=2770186). Stinchcombe, Arthur L. 1991. “The Conditions of Willer, Robb, Francis J. Flynn, and Sonya Zak. 2012. Fruitfulness of Theorizing About Mechanisms in “Structure, Identity, and Solidarity: A Comparative Social Science.” Philosophy of the Social Sciences Field Study of Generalized and Direct Exchange.” 21(3):367–88. Administrative Science Quarterly 57(1):119–55. Stovel, Katherine, Benjamin Golub, and Eva M. Mey- Yamagishi, Toshio, and Karen S. Cook. 1993. “General- ersson Milgrom. 2011. “Stabilizing Brokerage.” ized Exchange and Social Dilemmas.” Social Psy- Proceedings of the National Academy of Sciences chology Quarterly 56(4):235–48. 108(Supplement 4):21326–32. Zelditch, Morris, and Henry A. Walker. 1985. “Legiti- Stovel, Katherine, and Lynette Shaw. 2012. “Brokerage.” macy and the Stability of Authority.” Pp. 1–25 in Annual Review of Sociology 38(1):139–58. Advances in Group Processes, Vol. 1, edited by E. J. Suchman, Mark C. 1995. “Managing Legitimacy: Strate- Lawler. Greenwich, CT: JAI Press. gic and Institutional Approaches.” Academy of Man- Zelizer, Viviana A. 1985. Pricing the Priceless Child: agement Review 20(3):571–610. The Changing Social Value of Children. New York: Surak, Kristin. 2016. “Global Citizenship 2.0: The Basic Books. Growth of Citizenship by Investment Programs.” Zelizer, Viviana A. 2005. The Purchase of Intimacy. IMC Working Papers. Retrieved August 16, 2017 Princeton, NJ: Princeton University Press. (http://investmentmigration.org/download/global- Zelizer, Viviana A. 2012. “How I Became a Relational citizenship-2-0-growth-citizenship-investment- Economic Sociologist and What Does That Mean?” programs/). Politics & Society 40(2):145–74. Swidler, Ann, and Susan Cotts Watkins. 2007. “Ties of Zucker, Lynne G. 1983. “Organizations as Institutions.” Dependence: AIDS and Transactional Sex in Rural Research in the Sociology of Organizations 2:1–47. Malawi.” Studies in Family Planning 38(3):147–62. Tolbert, Pamela S., and Lynne G. Zucker. 1996. “The Institutionalization of Institutional Theory.” Pp. 175– Oliver Schilke is an Assistant Professor of Management 90 in Handbook of Organization Studies, edited by S. and Organizations (tenure track) and an Assistant Profes- Clegg, C. Hardy, and W. R. Nord. London, UK: Sage. sor of Sociology (by courtesy) at the University of Ari- Walker, Henry A., Larry Rogers, and Morris Zelditch. zona. Oliver is primarily interested in organizational 1988. “Legitimacy and Collective Action: A Research theory, social psychology, and economic sociology. Note.” Social Forces 67(1):216–28. Much of his current research contributes to elaborating Walzer, Michael. 1983. Spheres of Justice: A Defense of the microfoundations of institutional theory. Pluralism and Equality. New York: Basic Books. Weber, Max. [1922] 1978. Economy and Society: An Gabriel Rossman is an Associate Professor of Sociol- Outline of Interpretive Sociology. Translated by S. ogy at the University of California-Los Angeles. His Edward and M. Rheinstein. Berkeley: University of research interests include the production of culture and California Press. economic sociology. He is currently working on a com- Weinberg, Jill D., Jeremy Freese, and David McElhattan. puter simulation of diffusion and a book on obfuscated 2014. “Comparing Data Characteristics and Results exchange.