Reconsidering the Value of Covert Research

In this article, we provide a nuanced perspective on the benefits and costs of covert research. In particular, we illustrate the value of such an approach by focusing on covert participant observation. We posit that all observational studies sit along a continuum of consent, with few research projects being either fully overt or fully covert due to practical constraints and the ambiguous nature of consent itself. With reference to illustrative examples, we demonstrate that the study of deviant behaviors, secretive organizations and socially important topics is often only possible through substantially covert participant observation. To support further consideration of this method, we discuss different ethical perspectives and explore techniques to address the practical challenges of covert participant observation, including; gaining access, collecting data surreptitiously, reducing harm to participants, leaving the site of study and addressing ethical issues.

Bernstein's research. Insights were "gathered by three embedded researchers who were simultaneously operators on the factory lines and participant observers for the study" (Bernstein, 2012, p. 185). The three researchers that were sent into the field to collect data were undergraduate students born and raised in China, who would not stand out on the production line. As Bernstein (2012, p. 185) explains, "The three students were inconspicuously placed on the factory lines as ordinary employees-only the GM, head of HR, and head of operations of the 14,000-person facility knew their true identities." Covert studies that entail some element of deception, such as Bernstein's study, have played a prominent role in the development of the social sciences. For instance, covert observations informed the creation of cognitive dissonance theory (Festinger, Riecken, & Schachter, 1965) and illuminated the poor treatment of those in asylums (E. Goffman, 1961). Deceiving participants in experiments has also revealed important aspects of human nature including our potentially lethal obedience to authority (Milgram, 1963).
Studies that employ deception have, however, sparked important debates in various fields of the social sciences. These studies have often been challenged as unethical, suggesting that participants are "manipulated" and "conned" (Erikson, 1995, p. 9) even if the deception of research participants can be unintentional (Cunliffe & Alcadipani, 2016). These criticisms hinge on the idea that deception entails conducting research without participants' fully informed consent. Informed consent describes potential participants being aware of all the relevant information regarding the risks, benefits, and implications of participating in a study, and then processing this information to make a decision whether or not to participate (S. Clarke, 1999). The ethical guidelines of the Academy of Management (AOM), for example, explicitly require the consent of participants and thus prohibits covert research. Within managerial and organizational research, in particular, the use of covert research has fallen into abeyance and the debate regarding its ethical nature has progressively died out. The American Psychological Association (APA) and the American Sociological Association, by contrast to the AOM, are more open to some forms of deception and the use of false identity, when justified by the value of the research.
In an effort to illustrate the value of covert approach in management and organization studies and rekindle the debate surrounding it, we focus on one of the most common forms of deceptive research: covert participant observation. Covert participant observation describes researchers becoming embedded in the group or organization that they are studying (Gephart, 2004), while the researchers conceal "their true identity and purports to play some other role" (Vinten, 1994, p. 33). In this article, we highlight that covert participant observation can be virtuous in many ways, providing access to otherwise unavailable data (Lauder, 2003;Leo, 1995) alongside opportunities to interpret and understand these data firsthand (Bulmer, 1982;Sullivan, Queen, & Patrick, 1958). It also reduces the risk of disturbing or inhibiting participants' natural behavior (Homan, 1980). As noted by Bernstein (2012, p. 185), employing covert observers enabled him to "avoid contaminating the environment and the behaviors [he] was attempting to observe." With reference to a range of illustrative examples, we detail how these benefits allow the method to generate novel insights that support the development of important theoretical and practical contributions.
One of our main arguments in advancing covert participant observation, when approved by an ethics board of equivalent body, is that all forms of observation rely on some degree of deception due to the practical realities of observations (Dewalt, Dewalt, & Wayland, 1998;Leo, 1995). In contrast to a simple dichotomy, we present a continuum of consent for observation, from fully overt to fully covert. We develop an associated framework to support researchers in establishing where their studies sit along this continuum, to appreciate the complexity of what constitutes covert or overt observations. While most researchers justify covert research using a consequentialist argument, claiming that the benefits outweigh the cost of deception, we consider how a number of other ethical perspectives can enlighten this debate. In particular, a situated ethics approach can help the researcher formulate guidelines for observation and potentially participation.
The article proceeds as follows. First, we broadly look at how deception relates to the question of consent in social science research and consider the perspective of different professional associations. Second, we focus on one particular form of covert research-covert participant observation-to examine the unique benefits and costs of this approach. Third, we explain the necessity for a more nuanced approach to deception in management research, illustrating the idea of a continuum of consent in terms of covert studies. Fourth, we outline key ethical perspectives to justify covert research. Fifth, we provide practical guidance to address the challenges of engaging in covert participant observation, including gaining access, collecting data, managing participant harm, leaving the site of study, and addressing ethical issues.

Forms of Deception and Consent
Deception is to "cause to accept as true what is false or to give a false impression" (Korn, 1997, p. 4). Between 1959 and, 58% of a representative sample of social psychology studies involved some form of deception (Gross & Fleming, 1982). One way to categorize the different forms of active or intentional deception is in terms of the extent to which researchers capture informed consent from their participants. We suggest that there are two broad ways that this can occur: Participants provide consent on the basis of false information, or participants provide no consent and receive no information (Sieber, 1982).
Researchers' active deceptions that rely on participants' consent, but on the basis of false information, have been used in a variety of psychological experiments. In several famous examples, studies have employed confederates (actors or stooges who participate in an experiment while really working for the researcher) without informing the real participants (e.g., Asch, 1956;Moscovici & Zavalloni, 1969). One of the most famous examples is Stanley Milgram's (1963) experiments on obedience to authority figures. Milgram recruited participants for his experiment through newspaper adverts for male participants, falsely informing them that the experiment concerned the study of memory. In the experiment, each participant was paired with another person. Though the pairing appeared to the participants to be random, the participants were always selected to play the role of teacher while one of Milgram's confederates would always act as the learner. The study measured the willingness of the participants to follow the orders of an authority figure (an experimenter, Mr. Williams) who instructed participants to punish learners' errors by administering electric shocks, with the voltage increasing in 15-volt increments for each wrong answer. Unbeknownst to the participants, the learner was not actually receiving any electric shocks. Milgram's study revealed that, despite the learners clearly communicating their distress and eventually providing no response at all, 65% of participants would eventually administer lethal electric shocks. The impact of Milgram's work was considerable, in terms of research ethics and psychological theory, and it served to illuminate the human potential for destructive obedience and ineffectual disobedience (Elms, 1994).
Studies in which participants provide no consent and receive no information tend to rely on observations and occur in more natural settings, outside the artificial confines of a laboratory. For example, Piliavin, Rodin, and Piliavin (1969) conducted a field experiment on a subway in New York, where they observed the reactions and responses of the general public to their confederates falling over. They were able to disprove the popular thesis, corroborated in laboratory experiments, that suggested that greater group sizes led to the diffusion of responsibility.
Outside of experiments or induced events, researchers have also observed everyday life in natural settings without the consent of those being observed. Common in a variety of ethnographic studies, this approach is often referred to as covert observation. For instance, Walters and Godbold (2014) observed adult behavior at children's sporting events without the consent of those attending and provided information only upon the rare request of the participants. The authors argued that "the ends justified the means" as their method provided evidence that inappropriate adult behavior toward children is not uncommon yet remains unreported (Walters & Godbold, 2014, p. 536).
One of the most common forms of deceptive research is covert participant observation, which calls for researchers to participate in the cultures, groups, or organizations they are observing while hiding their true identity. Such an approach has been employed in sociological studies across a variety of groups or settings, including factories (Bernstein, 2012), asylums (E. Goffman, 1961), Pentecostal churches (Homan, 1980), gangs of adolescent boys (Parker, 1992), football hooligans (Pearson, 2009), and young men on the run from the police (A. Goffman, 2015). Intentionally or unintentionally deceiving participants on the research motives is common to access fields of study (Cunliffe & Alcadipani, 2016). In perhaps the most notorious case, Laud Humphreys (1970aHumphreys ( , 1970b conducted a covert participative observation of public bathrooms used for secret homosexual relations (known as tea-rooming). By acting as a lookout, he gained the confidence of some of the men he observed and learned more about their experiences. Humphreys secretly recorded the license plates on the cars of these men. A year later, disguised, Humphreys visited their homes and pretended to be a health visitor and interviewed these same men about their employment, marital status, and political views. Humphreys suggested that his research, which was strictly anonymized, had shed light on a widespread but rarely studied form of human interaction that would have otherwise remained hidden because of the taboo nature of this topic (Hudson & Okhuysen, 2014). His work contributed to debunk the idea of gays as deviant, while at the same time showing that the members of this community had to remain hidden behind false identities. He offered to understand how individuals could elaborate complex lies to appear as abiding by heterosexual societal norms, while in parallel being true to themselves in parallel settings.

Diverging Views on the Need for Informed Consent
Ethical guidelines in academic associations tend to emphasize the need for fully informed consent from research participants. General guidelines of research councils and funders, such as the Tri-Agency in Canada or the National Science Foundation (NSF) in the United States, stress the "duties of honest and thoughtful inquiry" and of "being straightforward and free of . . . deception" (Tri-Agency, 2011). Without specifically referring to covert research, research councils reject deceptive methods. The NSF now requires ethical training for those involved in research, and one of the key aspects to be covered is the recruitment of subjects who are participating voluntarily in the study and building a trustworthy relationship with the researcher.
The AOM emphasizes protecting "the privacy, dignity, well-being and freedom of research participants" (Academy of Management, 2006, p. 3). It also stipulates that AOM members "obtain the informed consent of the individual or individuals, using language that is reasonably understandable to that person or persons" and requires that "written or oral consent, permission, and assent are documented appropriately" (p. 5). In this sense, covert research, which relies on the absence of informed consent (Soble, 1978), would be a breach of the AOM code of ethics.
Other fields of social science, however, provide differing views. The American Sociological Association (1999) more explicitly address the use of deception: On rare occasions, sociologists may need to conceal their identities in order to undertake research that could not practicably be carried out were they to be known as researchers. Under such circumstances, sociologists undertake the research if it involves no more than minimal risk for the research participants and if they have obtained approval to proceed in this manner from an institutional review board or, in the absence of such boards, from another authoritative body with expertise on the ethics of research. Under such circumstances, confidentiality must be maintained. (p. 12.05) The APA (2010, p. 8.07) recognizes that "the use of deceptive techniques [can be] justified by the study's significant prospective scientific, educational, or applied value" and by the fact "that effective non-deceptive alternative procedures are not feasible." Scholars must be able to "justify the value" of the knowledge acquired, and show that "nondeceptive alternative procedures are not feasible" (APA, 2010, p. 8.07).
In this way, the limited use of deception in management and organizational research relative to sociology and psychology becomes clearer when considered in light of associated ethical codes. While some management ethical guidelines conceptualize consent in black-and-white terms, other social science disciplines offer a more nuanced view and recognize the potential value of deception in research, albeit in rare situations or when alternatives are not feasible. To more clearly illustrate the value of covert research and the debate surrounding it, we focus on one of the most common deceptive approaches: covert participant observation.

The Debate Around Covert Participant Observation
Covert participant observation has yielded important theoretical contributions, and in this sense provides a good focus to illustrate the benefits of research that entails some aspect of deception. It has also been at the center of the controversy and aroused a variety of criticisms.
In the 1950s and 1960s, the School of Chicago popularized studies that utilized covert observation in the field of sociology (Bulmer, 1986). These studies entailed a participatory dimension, with students joining occupational groups and using the job as a way to finance their postgraduate studies (Roth, 1959). This research illuminated the everyday life of taxi drivers (Davis, 1959), factory workers (Roy, 1952), or jazz club employees (Becker, 1951). These studies emphasized mixing with the participants on a basis of equality and concealment, engaging in informal interaction in a situation of anonymity to explore aspects of human nature not ordinarily revealed (Bulmer, 1986).
Covert participant observation has continued to be employed across a variety of fields, such as nursing (Johnson, 1992), addiction (Power, 1989), community psychology (Farrington & Robinson, 1999), religious studies (Lauder, 2003), and medical anthropology (Oeye, Bjelland, & Skorpen, 2007). Management and organization studies also benefitted from an early tradition of covert observation in a series of monographs interested in the daily life of workers and managers. Melville Dalton (1959), who acted as an administrative assistant to collect data on the informal culture of organizational life, offered a groundbreaking study of hidden political conflicts at the executive level in three manufacturing firms and one department store. He empirically documented the limits of human rationality in the workplace and the distribution of organizational power beyond formal hierarchies. Van Maanen (1975) enrolled himself as a participant observer in a training program of the police to observe from the "bottom-up" (Van Maanen, 1981, p. 6) how trainees would socialize with each other, and experience firsthand that process. While Van Maanen had the consent from the police forces he observed, he did not have the consent of the civilians. Similarly, Laurie Graham (1995) investigated the production technique of lean management that was glorified at that time by becoming a worker on the production line of a Japanese manufacturer in Indiana. Her work was the first to provide an in-depth understanding and thorough critique of lean management by looking at the impact of employee control. She revealed the dangerous intensity of work and the health-related consequences of lean management techniques that were largely unreported. By drawing on these and other examples of covert participant observation, we go on to highlight a range of benefits of this method for developing important theoretical contributions within management and organizational research.

Benefits of Covert Participation Observation
We propose that covert observation offers three overlapping benefits. First, covert participant observation can provide access to institutions or organizations that would otherwise remain excluded from researchers. This includes secretive organizations (known to the wider public remaining discrete about their activities) and secret organizations (not known to the wider public, and without a forefront). Second, it may allow researchers to collect data as an insider, reducing the risk of those under study modifying their behaviors and can thereby reveal secretive behaviors, which may be obscured or unethical. Third, being a "pure" insider also allows the researcher to experience firsthand the phenomena under study in the same way that the participants experience it. Individually and collectively, these benefits can generate new insights and capture rich data that would often be unobtainable to support novel theoretical contributions.
Covert participant observation can provide access to behavior or organizations, or parts of organizations, that would otherwise remain inaccessible to overt research. Gaining and maintaining access is often a challenging task (Peticca-Harris, deGama, & Elias, 2016). To gain and maintain access, deception is often common practice (Cunliffe & Alcadipani, 2016). This is particularly true in secretive organizations such as the military or the police (e.g., Leo, 1995;Mann, 1957;Sullivan et al., 1958) or for organizations that have something to hide (e.g., Morales & Lambert, 2013) or that desire to remain a secret and mysterious to preserve internal values and cohesion (e.g., Festinger et al., 1965Festinger et al., /2008. In those contexts, the participants and the field are more likely to be isolated and often inaccessible to overt research projects (Lauder, 2003). As such, covert observation may be the only way to study deviance (Lauder, 2003;Leo, 1995) or minorities and underprivileged populations (Dewalt et al., 1998). Indeed, Sullivan (1959) justifies pursuing covert observations due to the "inaccessibility of institutions." One example is Leon Festinger's research into a secretive cult. In 1954, Festinger sought to understand how the members of doomsday communities who anticipated an imminent apocalypse could deal with a truth inconsistent with their belief (Festinger et al., 1965(Festinger et al., /2008). Festinger and colleagues joined the cult to observe the phenomenon from the inside and participated in group meetings. The data they collected supported the fundamental propositions of the hugely influential cognitive dissonance theory. This theory posits that individuals tend to behave in conformity with their beliefs, and discard any information that might bring them to change their beliefs and thus their behavior. Later, he designed laboratory experiments to test the propositions developed through this first qualitative exploration. The resulting cognitive dissonance theory has inspired thousands of researcher projects and generated thousands of publications, and remains a popular theory across the social sciences.
While covert participant observation can be applied to secret organizations such as cults, it can also be applied to secretive parts of more publically known organizations. For example, Leo (1995) describes that while other police activities are conducted openly and subject to public scrutiny, only in exceptional circumstances will civilians be permitted to observe interrogations. Leo waited two years to be granted access to the interrogation room. Despite "the introduction from their superiors (whom, [he] later learned, many of the detectives intensely disliked), most of the detectives initially distrusted [his] motives and the research objectives of [his] study; some bluntly told [him] so, others quietly avoided [him]" (Leo, 1995, p. 118). To avoid limiting access to potentially valuable information, he lied about his identity to be able to observe a detective interrogation room. In his own terms, he "reinvented [his] persona to fit the attributes, biases, and worldview of [his] subjects" (Leo, 1995, p. 120). Leo needed to distance himself from his identity as an academic which was 492 Organizational Research Methods 20 (3) prejudicially associated with advocacy of defendants. To achieve this identity make over, Leo adopted the same routines and culture as the police officers he was observing, and creating a rapport by faking personal and political similitude. Invoking "moral relativism," Leo (1996, p. 126) defended deceiving subjects on the basis that it is counterbalanced by the social value of the knowledge acquired, in his case, his findings on misbehaviors in the interrogation room support the idea of making police socially accountable. Similarly, Dalton (1959) explained how being undercover helped him to access the "informal" organization and the controversies of organizational life. This is important because, as Johnson (1992) stresses, in the case of occupations that play major societal roles, such as the police or the army, covert observation is one way to reveal their practices and hold them to account. In this way, such research also offers unique practical contributions. Some behaviors might not be exhibited by the participants when they know they are observed. When participants think they are interacting with one of "their own," they will behave in a more natural way (Denscombe, 2010). As such, researching covertly prevents the researchers being perceived as outsiders and therefore reduces the risk that they may modify their behaviors (Oliver & Eales, 2008;Stenger & Roulet, in press;Sullivan, 1959). Being undercover can sometimes be necessary to ensure the "naturalness" of the data (Denscombe, 2010, p. 217) and avoid the social stigma associated with being a researcher (Van Maanen, 1981). Subjects are "free from disturbance and inhibition" (Homan, 1980, p. 46). In Lauder's (2003) study of a xenophobic movement in Canada, the researcher was clearly able to compare the data collected through overt and covert observation. He later concluded the latter was the only way to obtain relevant data on the participants' behaviors and values. Similarly, Bernstein (2012) noted that the presence of a "foreigner" would perturb workers and contaminate the collection of data.
Erving Goffman (1961) relied on an almost fully covert observation in an asylum for mentally ill people. Goffman joined St. Elizabeth psychiatric hospital as an assistant to the athletic director for a year. This is a classic example of covert research: Neither the organization (i.e., the hospital) nor the subjects (staff and patients) knew that he was a researcher collecting data. Goffman does, however, mention that a couple of staff members gradually came to know that he was a researcher. Goffman sought to understand "total institutions"-environments in which participants are secluded from the world-from the inside and access a hidden reality that would not be made available to him as an overt researcher. His work focused mostly on the relationship between the inmate and the institution. His radical critique of mental hospitals as "total institutions" and the treatment of mental illnesses had a significant and long-lasting impact on practice. For instance, he singled out the treatment of mental illness at that time as ineffective and even counterproductive, aggravating the issues by isolating and dehumanizing the patients. This critique is also recognized as having helped to open mental health as a field of research for social scientists (Holmes, 2009).
Covert participation observation can sometimes require researchers to assume a "total participating role" (Sullivan, 1959, p. 399). This can enable researchers to go beyond what is expressed by the participants to experience the norms and understand the practices firsthand. As the researcher's "commitments deepen, a more refined and fixed sense of social arrangements develops" (Van Maanen, 1981, p. 40). The advocates of covert participant observation have pointed out the richness of the data obtained and that this method provides the researcher with a more profound understanding of the organizational issues at stake (Oliver & Eales, 2008;Roy, 1952).
One example of the benefit of experiencing phenomena firsthand comes from Bernstein's (2012) multimethod study in a manufacturing firm in southern China. While he did not participate directly in the collection of the qualitative data, Bernstein's team of researchers did participate directly in an initial qualitative study of factory work in China. The observers were "instruments of the research" (Bernstein, 2012, p. 187) as they "lived with those whom they studied" and went through the same process as new hires and were allocated to tasks by "floor heads and line leaders, who were unaware of their status as embeds" (Bernstein, 2012, p. 186). Had the researchers not participated in the study themselves, then it is unlikely that they could have recognized how transparency and monitoring can actually hamper the emergence of practices that support productivity.
Covert participant observation can generate new insights and knowledge that are not always possible with more overt methods. Covert participant observations are particularly well suited to studying secretive organizations, observing natural behaviors in the workplace, and gaining a firsthand and richer account of the phenomena under study. It is important to note that covert participation observation can also complement other existing methods, particularly as an initial source of propositions that can be refined or tested through different methods (as in the cases of Bernstein, 2012;Festinger et al., 1965Festinger et al., /2008. Furthermore, the data collected through covert participant observations can be integrated into established methodologies and analytical processes, such as grounded theory (Glaser & Strauss, 1967), ethnomethodology (Gephart, 1978), and phenomenology (Gill, 2014). Covert participant observation offers one way to study phenomena that often remains hidden from organizational researchers, such as emotions like anger or fear (Gill, 2015), mentalities as underlying and hidden beliefs (Clemente, Durand, & Roulet, 2017), or stigmatized organizations (Hudson & Okhuysen, 2014;Roulet, 2015). Table 1 outlines the benefits discussed in this section with reference to examples of covert participant observations from across the social sciences.

Criticisms of Covert Participant Observation
One of the first disputes around covert participant observation began in 1959, in the American Sociological Review, with an opinion piece titled "A Question of Professional Ethics" (Coser, 1959). Coser (1959) argued that deceiving organizational members to collect data was ethically unacceptable and could lead researchers to spy on behalf of senior staff. Subsequent debates emerged, such as when Erikson (1995, p. 9) criticized Leo's (1995) covert investigation of detectives' behaviors in an interrogation room because of the unethical way "he 'manipulated' and 'conned' his subjects into thinking he was someone quite other than his true self." The relatively small number of management and organizational research studies that explicitly state employing covert participant observations is likely to reflect the ethical scrutiny that the method has attracted. Covert observation is "interactionally deceitful" (Ditton, 1977, p. 10), and those who perform it have been labeled "liars" (Allen, 1997).
The first issue is that covert researchers typically mislead others about their identity or the intent of the research (Lauder, 2003). Observing participants without their informed consent is often viewed as a breach of their rights and privacy (Coser, 1959). For instance, the participants being observed may engage in behaviors they do not wish to be recorded (Dewalt et al., 1998;A. Goffman, 2015). Recognizing the rights of participants is a fundamental ethical consideration (Oliver & Eales, 2008). Covert observation therefore violates the "principle of informed consent" (Bulmer, 1982, p. 252) as it betrays the trust of participants. The information collected covertly may be used to harm those being observed, even if inadvertently. This is problematic because research is usually expected to remain harmless to the participants (Oliver & Eales, 2008), yet covert studies may pose a risk to the unknowing individuals being observed. In one example, Leo (1995) was called to court to act as a witness against the police officers he observed in an interrogation room. This meant that he was compelled by lawyers to release his field notes thereby jeopardizing the participants of his study. This might appear to be an extreme case. However, as ethics committees, funding bodies, and scientific outlets demand increasing transparency, it is likely that researchers will face increasing pressure to reveal details of their research (Dewalt et al., 1998), which makes data collected covertly increasingly problematic. Furthermore, organizations or fields that have been the object of covert observation can become more suspicious and harder to study for future generations of researchers.
A further criticism of covert approaches is that the researcher employing such a method may have to engage in morally abhorrent or risky behaviors to gain access and to avoid being discovered No. Yes, they were able to collect firsthand observations. No.
Yes, the researchers did not have to ask for the permission of the organization. Yes, the authors make the argument that the behaviors to be observed would not have been made visible to an overt observer. No. No. No.
No. Yes, policing is a field in which participants are less likely to express their feelings openly. No.
Yes, the researcher wanted to experience firsthand the socialization process.
Yes, the qualitative evidence were complemented by questionnaires.
(continued) No. (Dewalt et al., 1998). Pretending to "be someone else" can lead the researcher to act in a way that they may find questionable, as in the case of Leo's (1995) investigation of interrogation room. Dewalt et al. (1998) mention the use of illicit drugs, violence, and sexual involvement with their participants. As pointed out by Lauder (2003), the researcher might be trusted only once he or she has engaged in an action that would be questionable for outsiders but that would be interpreted as a sign of commitment to the group by insiders. Engaging in criminal behavior in the context of covert participant observation would expose the researcher to legal prosecution, arrest, and jail. The recent controversy around the work of Alice Goffman (2015), who flirted with illegality as she fully participated in the life of the gang she observed, illuminated some of the ethical issues with researchers' actions in the context of participant observation. With one of the informants, she engaged in chasing the suspected murderer of one of the individuals she had befriended during her participant observation. Existing studies have highlighted how researchers, as well as the participants that they study, can be adversely affected by performing covert research. As Punch (1986, p. 73) stated "eavesdropping, fudging over one's purpose, simulating friendship, surreptitiously reading documents, etc.-make for good data but bad consciences." Elsewhere, this has been referred to the idea of the "sociologist's original sin," when researchers experience guilt for building relationships with others for the purpose of collecting data (Davis, 1959), especially when fieldworkers may become good friends with those that they study (Thorne, 1980). This is likely to be particularly acute and significant for covert participation observers. For instance, Graham (1995, p. 15) notes that her activity was "physically demanding and emotionally draining," and Berlin (1995) notes the difficulties she experienced by disguising her identity and maintaining elaborate lies.
A further criticism concerns the validity of data collected through covert observation. Having a distanced perspective or being somewhat removed from the empirical context is often viewed as a signal of quality in data collection (Locke, Golden-Biddle, & Feldman, 2008). In a covert participant observation, the research object and the researcher are confounded: A covert observer participates in what is being observed (Oliver & Eales, 2008). When a researcher is involved as a participant observer, it is commonly suspected that he or she will therefore lack the distance required to report rigorous empirical accounts. Some scholars have viewed this issue as an epistemological question rather than a methodological flaw (Anteby, 2013). For instance, autoethnography, an approach that uses the experience of the researcher as a source of data, acknowledges the subjective aspects through which data are collected and apprehended (Gephart, 1978). The sense-making processes experienced by the researcher offer important insights that can be considered and accepted as useful subjective accounts rather than biased elements of data (Islam, 2015). Nonetheless, many scholars consider covert participant observation highly problematic as a method of data collection.

The Case for Covert Participant Research The Dimensions of Informed Consent
Declaring covert participant observation as unethical due to the method's failure to acquire consent is a simplification of a complex issue (Calvey, 2008). Observational studies are rarely fully covert or fully overt but usually situated somewhere between these two poles. To highlight this complexity, we consider observational studies in terms of two dimensions: who knows the researcher's purpose (breadth of the consent), and how much is known (depth of the consent).
In terms of the first dimension, breadth of consent, or who is aware of the study, informed consent can be collected at the organizational level, from official representatives of an organization, as well as at the individual level, from participants being observed directly or indirectly observed. In terms of the second dimension, depth of consent or how much each participant is aware of the study, different amounts of information can be provided to the participants from full disclosure of the research purpose, process, risks, and benefits to no disclosure whatsoever (Calvey, 2008). Figure 1 illustrates the two dimensions of deceit for covert participant observational studies within organizations.
To illustrate these dimensions and to highlight the nuanced nature of consent, we refer to a range of studies that can be situated at different points. At one extreme point of the continuum, researchers employ fully covert observations. For example, Postula and Postula (2011) studied liminal actions in a state-owned enterprise in Poland. They opted for covert observation because "they were concerned about employees' reaction to a strange observer" (p. 36). Also explicit in their use of an explicitly covert approach, Berlin (1995) looked at technological issues and cultural clash in the contexts of a factory and a multinational company in Venezuela. In one particular case, she "infiltrated as a spy" (p. 386) to show how local workers were less interested in acquiring technical skills. This work directly questioned the problematic role of foreign direct investment, by contrast with the positive role usually assumed, through the use of covert methods.
In the middle of the continuum, observations can be partially covert with a share of the participants knowing about the real identity of the researcher. For example, the senior members or

498
Organizational Research Methods 20 (3) gatekeepers of an organizations under study can be informed of the research process of the participant observers. This was the case for Bernstein's (2012) study of a Chinese factory. The studied facility employed "65,000 individuals in 3.1 million square feet of manufacturing space" (Bernstein, 2012, p. 184). In this context, it appears that while the organization had given its approval (the general manager, head of human resources and operations knew about the study), the employees did not provide any informed consent to participate to the study. In the same vein, Alice Goffman (2015) relied on two key gatekeepers to access the field with whom she shared her work at several stages during the observation, though many other participants were not informed. Those intermediaries vouched for her and defended her presence. This did not come without any cost: Goffman's relationship with the intermediary started after a date, and she was regularly thought of as being romantically involved with him or other subjects. After her intermediary was taken into custody, Goffman was cut from the field and could not carry out her observation anymore. Pratt (2000, p. 460) explicitly acknowledged that he employed "semiovert participant observation" in a marketing distribution company, letting his coworkers know about his inquiry, but seemingly leaving other more distant participants less informed. Presumably, Pratt did not obtain consent from the customers as he was selling products as part of his participant observation. A number of partly covert observations have been published in the management journal Administrative Science Quarterly, although they do not always explicitly mention the deceptive aspect (Bernstein, 2012;Sutton, 1991;Van Maanen, 1975). Van Maanen (1975) enrolled himself as a participant observer in a training program of the police to observe from the "bottom-up" (Van Maanen, 1981, p. 6) how trainees would socialize with each other, and experience firsthand that process. Sutton (1991) investigated the daily routine of debt collectors. While the organization and the gatekeeper knew about his research purpose, it is not clear whom among the collectors he interacted with were aware of his research purpose. However some parts of his observation were obviously covert as he for example used a "spy-and-tell" system to listen to calls between debtors and collector without being known.
The issue of informed consent is common to many forms of observation. It is unrealistic to believe that all observational studies can acquire truly complete and formal consent from all participants (Dewalt et al., 1998), especially when observing a large number of participants (Berlin, 1995). While researchers can obtain consent at the top of the hierarchical pyramid-and the consent of organizational gatekeepers-this often only assumes consent of those lower level employees. Furthermore, it is never clear what participants can consent to ex ante and how this can diverge as the observation process unfolds (Miller & Bell, 2002). Access to the field may also involve some unintentional distortion of the truth with regards to the identity of the researcher and the purpose of the research that questions the fully informed nature of the subjects' consent (Cunliffe & Alcadipani, 2016). In this way, simple conceptualizations of consent ignore the practical constraints of observation as a method of data collection.
In addition, the frontier between overt and covert research is unclear (Calvey, 2008). Although presented as an overt observation, the example of Alice Goffman's observation of Black neighborhoods in Philadelphia is particularly telling (Goffman, 2015; see her methodological note in the appendix, pp. 213-263). Her research initially focused on the life of Black women, but as she met new actors in the field, she progressively shifted her observation toward a group of men involved in illegal activities. The first group of women she observed "thought [Goffman's involvement as a tutor for adolescents in the neighborhood] was for school" (p. 221). Thus, the first group of subjects had only a vague sense that Goffman was observing them as part of a research project. As Goffman started to become interested in other groups of individuals, her identity as a researcher became less and less apparent to the individuals that ultimately became subject of the study. This reflects S. Clarke's (1999) argument that informed consent is made difficult by the struggles for subjects to fully understand the process or the nature of research, because of their training and education, but also potentially because of disinterest.
As noted in the case of Goffman's (2015) investigation, a research study is unlikely to remain in a fixed position in relation to these two dimensions throughout the period of observing and data collection. Lauder (2003) for example, initially presented himself as a researcher, but progressively feigned conversion and disguised his convictions to gain the trust of his subject. Conceiving of covert participant observations as dynamic research that moves frequently along a continuum of consent is valuable because it highlights the complexity of capturing complete consent in the field. In this way, covert participant observation is only "different in degree in the extent of deception used from 'open' research" (Bulmer, 1982, p. 253).

Ethical Justifications and Toolkit for Covert Participant Observation
In Table 2, we have contrasted the different perspectives of professional associations on covert research and deceptive approaches. We have also considered the ethical argument on which they rely, though these are not explicitly stated. While the AOM strictly requires informed consent, other associations offer greater latitude. The AOM focuses on the respect of the autonomy and dignity of participants, thus ultimately leaving them the choice to participate in studies. The APA and the American Sociological Association allow for some deception in case it is justified by the research purpose and consequences. Accordingly, deception in social sciences research, and more specifically covert participant observation, is usually justified by researchers employing a consequentialist argument. The consequentialist argument is the most common ethical justification we have found in our review of covert research across the social sciences.
The consequentialist perspective. The consequentialist argument emphasizes that the benefits of the research for society outweigh the costs to participants (e.g., Baumrind, 1979;Walters & Godbold, 2014). From a consequentialist perspective, what is morally right is the course of action that has the greatest benefits for the greatest number of people. The focus is on the consequence of the researchers' action. Covert participant observation is ethical when the social benefits outweigh the costs (Lauder, 2003), and it is the right course of action because of the positive outcome it can yield. For instance, Leo (1996, p. 126) defended deceiving subjects on the basis that it is counterbalanced by the social value of the knowledge acquired (Oeye et al., 2007). Humphreys (1970a) justified his covert observation of the double life of homosexuals to explain the social dilemma they faced and to debunk the notion that they were deviants (Wiles, 2012).
Making this consequentialist argument relies on the "justification" of the value of the covert research (APA, 2010). This would rely on some form of assessment of the social good or the advancement of science that will ultimately yield positive consequences for the society. As discussed earlier, it is possible to make the case that some covert participant observations have made significant theoretical and practical contributions across the social sciences. From a consequentialist perspective, the researcher can lie, break promises, and deceive subjects if that can be outweighed by positive consequences. One argument in favor of a consequentialist approach to covert research is the fact that only the result of the research remains, while the method and the process are only transient (Oliver & Eales, 2008). For example, it can be argued that Humphreys's (1970a) work may have put some subjects in embarrassing posture, but in a long-term perspective, his work has helped to normalize homosexuality (Wiles, 2012).
There are, however, a number of problems with adopting a consequentialist perspective in terms of covert participant research. Comparing the costs and benefits of a course of action is often practically impossible (Rawls, 1971): How to capture value? To whom? It is practically impossible to evaluate the consequences of all the actions of individuals involved for an indefinite 500 Organizational Research Methods 20(3)  (2006) Code of Ethics The research needs to respect "the privacy, dignity, wellbeing and freedom of research participants" and requires that that "written or oral consent, permission, and assent are documented appropriately." Yes. Deontological Kantian approach: Others are considered as ends and never as means and their dignity and autonomy need to be respected.

American Sociological Association (1999) Code of
Ethics "Sociologists may need to conceal their identities in order to undertake research that could not practicably be carried out were they to be known as researchers." "Sociologists do not use deceptive techniques unless they have determined that their use will not be harmful to research participants; is justified by the study's prospective scientific, educational, or applied value; and that equally effective alternative procedures that do not use deception are not feasible."

No.
Consequentialism with a focus on both the research itself and its scientific value (deception is acceptable if no other option are available) and its social value.
American Psychological Association (2010) Ethical Principles "The use of deceptive techniques [can be] justified by the study's significant prospective scientific, educational, or applied value." Scholars must be able to "justify the value" of the research.

No.
Consequentialism with a focus on the positive consequences of the research: If the social benefits of the research outweigh the cost, deception is acceptable.

Association of Social Anthropologists (2011) Ethics
Guidelines Acknowledgment that the researcher "may be able to provide only rough approximations in advance of some of the likely participants" and "Given the open-ended and often long-term nature of fieldwork, ethical decision-making has to be undertaken repeatedly throughout the research and in response to specific circumstances." "The interests and rights of those studied should come first." "Work for state or non-state organisations that is covert, and therefore breaches relations of trust and openness, is especially problematic. Overt work that is only possible because the participants are subject to coercion is also likely to breach basic ethical standards." No. Situated ethics: Covert research is acceptable in some contexts, on the conditions that the researcher constantly questions the ethicality of his or her action and research, and its consequences.
time (Israel & Hay, 2006). Some consequences of the covert research might be unintended and not anticipated by the researcher: For example, what if, in the case of Bernstein's study (2012), some workers were fired because they had been caught slacking by the observers? The researcher is unlikely to have even been aware of those potential consequences. Furthermore, Jackson (1991) explains how continued deception might hamper efficient decision processes for the deceiving party: Concretely, the covert researcher will spend more energy trying to hide his or her real identity and motive rather than conduct good research. Finally, the consequentialist perspective assumes that the researcher knows what is better for the participants than the participants themselves. S. Clarke (1999, p. 158) points out that the consequentialist view would defend the idea that individuals know the best way to maximize their own utility, and their right to informed consent should thus be inviolable. Thus the consequentialist argument can also be made against covert research.
The Kantian perspective. Similarly, a Kantian ethical perspective would require all subjects to be able to rationally make the choice of participating in an experiment, considering its emphasis on individual autonomy. From this deontological perspective, obligations do not follow from consequences but rather from the morality of actions, based in particular on the respect of human dignity (Israel & Hay, 2006). The Kantian perspective argues that "persons have unconditional worth and ought to be treated as autonomous ends and never merely as means" (S. Clarke, 1999, p. 158). Because it focuses on the moral duties we have to respect ourselves and other human beings, informed consent is a key element of deontological research ethics. The AOM Code of Ethics relies on a deontological perspective because of its focus on the inherent rights of participants (Oliver & Eales, 2008). From the perspective of deontological ethics, researchers should follow the norms and rules set up by their field. As we discussed in the first section of the article, most professional associations emphasize the need for fully informed consent. Thus, requiring informed consent has been the dominant paradigm because it recognizes Kant's point on individual autonomy and also satisfies consequentialist perspectives as it recognizes the power for participants to decide for themselves what is best. However, a deontological argument can be made in favor of covert research. A moderate Kantian view such as Parfit's objective ethical theory, blending Kantian deontology and consequentialism (Parfit, 2011), would lift the requirement for individual autonomy if the benefits for society are enormous and can be unanimously recognized by potential subjects. Parfit (2011, p. 411) indeed argues that "everyone ought to follow the principles whose universal acceptance everyone could rationally choose or will." Thus, if something can be unanimously accepted as positive, it is the right course of action. Parfit's perspective can be applied to the ethics of covert research: If ex ante it can be argued that the deceived subjects would agree that the research purpose outweighs the cost of being deceived, then covert research is justified. Such a case can however be hard to make (S. Clarke, 1999), especially when the findings and impact of research are difficult to predict or would not necessarily be recognized by the observed subjects.
Situated ethics as an alternative perspective. Some of the consequentialist arguments in favor of covert research have relied on the idea of "proportionate reason" (Angrosino & Mays de Perez, 2000;Oliver & Eales, 2008) whereby researchers can formulate a judgment on what is acceptable depending on the outcome. This assumes that researchers' "assessment of research rights and consequences, can like the research process be based on their own interpretation rather of what is right or wrong" (Oliver & Eales, 2008, p. 347). Such an approach is well anchored in the trend of situated ethics: As a philosophy of action, it can offer a less rigid lens than consequentialist or deontological approaches as it rejects the idea of universal principles or codes of actions.
In contrast to the aforementioned ethical perspectives, which proffer general principles that can be applied in a range of situations, situated ethics pay more attention to ethics as an ongoing social practice and emphasizes that ethical principles should be shaped by contextual factors rather than by 502 Organizational Research Methods 20 (3) universal codes (Nyberg, 2008). In a number of empirical contexts, it is impossible to "take a side" (i.e., determine what is right and wrong), considering the complex web of connections and motives (Calvey, 2008). This is especially true in the contexts that are harder to access for researchers. As a solution, Calvey (2008) suggests that covert observers engage in situated ethics, through constant reflection on the morality of the researcher's action in the field. Researchers need to constantly revise their assumption by questioning the ethicality of their decisions as the observation is carried out: Ethical integrity needs to be continually maintained and justified (Simons & Usher, 2000). Reflexivity has been shown to be particularly helpful for qualitative researchers to apprehend their relations to the subjects of their inquiry (Hibbert, Sillince, Diefenbach, & Cunliffe, 2014). Situated ethics requires the researcher to reflect on his or her actions and understand what sense they make in each context. Rather than offering a final answer on what is moral or not, situated ethics requires the researcher to morally question each of his or her actions to ultimately justify the choices made as an observer. Danaher and Danaher (2008) offer three steps for the researcher who chooses to engage in situated ethics. First, the researcher needs to unfreeze the ethical decision making by acknowledging the typically transitory and unpredictable nature of the field being observed. This means that the right ethical decision can be wrong at a later point in time. In the case of Humphreys (1970aHumphreys ( , 1970b, he repeatedly infiltrated some scenes that were not going to bring him any new elements of data. A situated ethics perspective would have helped him limit the scope of his covert research by questioning the research objective of some unnecessary parts of his investigation. The second suggestion of Danaher and Danaher (2008) is to unsettle taken for granted assumptions of what is right and wrong. Lauder (2003) explains how during his observation of a right-wing group, he had no other choice than lying on his identity because of the physical risks he could be facing, while at the same time not judging the behaviors of the subjects. Finally, the researcher is invited to interrogate the relationships and situations of the subjects he or she observes. Relationships and situations can be driven by power, attachment, or other invisible elements that can hamper the ethical judgment of the researchers. Some existing covert research already relies on situated ethics principles. For instance, the way Alice Goffman (2015) reflects on the way she acts (or is tempted to act) is in some way an attempt to update and refresh her decision making. She recognized the evolution of her judgment over the subjects she observed as she was carrying out her research. She caught herself being imprinted by her identity as a field participant when she considered selling drugs or stealing computers, and she immediately corrected her behavior. However, she reflected on her manhunt (the one that triggered the controversy regarding her work) only after it happened, and refused to recognize it as an ethical failure, although she did interrogate the relation she had built with the subjects.
The situated ethics perspective offers an alternative to a binary perspective of what is right or wrong (Simons & Usher, 2000). Table 3 provides a guide for the covert participant observer to carry out ethical reflexivity, by highlighting some of the questions a researcher is likely face in terms of entering the field, during the period of covert study and leaving the field. These questions reflect the ethical difficulties faced by covert researchers in the recent past (e.g., Bernstein, 2012;A. Goffman, 2015;Lauder, 2003) and can be used in the methods section but also in the presentation and framing of the findings. The situated ethics perspective is aimed at helping the researcher to make the right choices not only when carrying out the participant observation but also when interpreting and representing the realities of the participants (Simons & Usher, 2000).

Practical Considerations for the Covert Participant Observer
While covert participant observation offers many benefits, it also generates unique practical challenges. To address the paucity of guidance for researchers interested in employing covert participant observation, we consider several of these challenges and offer a variety of techniques that scholars have utilized in the field to overcome them. We address five challenges: developing a cover story, collecting data surreptitiously, reducing harm to participants, leaving the site of study, and receiving approval from ethical boards. It is important to emphasize the value of research training in developing the relevant competencies to undertake research, especially for research that entails some degree of deception. Any researcher who seeks to employ such a method should seek out training in ethnographic approaches, observational methods, and ethical issues. Furthermore, researchers must be prepared to invest considerable effort and time into the necessary planning and preparation, and this might involve learning skills that are needed for the participatory elements of the observation. We also posit that researchers usually have institutional obligation to seek approval from their ethics board or other relevant bodies prior to conducting any covert research. As we hope to indicate, conducting a covert participant observation can be laborious and requires extreme care before, during, and after a research project.

Getting In: False Identities and Real Skills
As Berlin (1995, p. 381) stated of covert participant observation, "one of the key methodological challenges in field research is gaining access to organizations" (see also Cunliffe & Alcadipani, 2016;Peticca-Harris et al., 2016). Gaining access to perform a covert participant study therefore requires consideration of how to acquire familiarity with the site of study, constructing an appropriate identity, and then developing the necessary skills to fit in.
Acquiring familiarity with the site of study. In many of the existing covert participant observation studies, researchers began by trying to learn as much as possible about the organization or field of interest in advance of conducting their study. As Jamieson (2002) points out, particularly when there is a risk of danger or violence, there is value in visiting research sites prior to conducting the actual research. This allows researchers to assess and discuss potential risks with colleagues or supervisors as well as providing more information to support gaining access. For example, Festinger et al.'s (1965Festinger et al.'s ( /2008 observations of a cult began by reading a story about a lady's (Mrs. Keech) prophecy of the end of the world in a local newspaper. Identifying this as a potential opportunity to field test research into prophecy, two of the authors "called on Mrs. Keech" to "learn whether there were other convinced persons in her orbit of influence" (Festinger et al., 1965(Festinger et al., /2008). The authors note that "the results of this first visit encouraged us to go on" (Festinger et al., 1965(Festinger et al., /2008, leading them to hire additional observers and to then join a group who were actively interested in Mrs. Keech's ideas. Constructing a new identity. Invariably, a researcher seeking to perform covert participant observation will need to be prepared to provide convincing answers to questions pertaining to their identity or, potentially, to develop and sustain a cover story. The organization or the field of interest may be secret as in Festinger et al.'s (1965Festinger et al.'s ( /2008) exploration of a cult. To ingratiate themselves with the cult required convincing the cult members that they held the same beliefs. Similarly, research into the police or the military also required researchers to lie about their identity (Leo, 1995) or the motive and focus of the research project (Huggins & Haritos-Fatouros, 1994). Because holding a minority opinion can lead to field exclusion (Clemente & Roulet, 2015), Leo (1995) explained that he disguised his political orientation to obtain the sympathy of the police officers he was observing. As Alice Goffman (2015) noted, she had to not only disguise her identity but become someone else. When asked to "account for her presence," the gatekeeper who knew about her real identity introduced her to the group as his "sister" or "godsister" (2009, p. 341; 2015, p. 228). She also mentioned to the subjects that she lived nearby (2015, p. 228). Disguising one identity while potentially developing another will require preparation. At its most straightforward, this may necessitate researchers being able to convince a senior employee in an initial interview that they would like to work for the organization they seek to study, without revealing their research goals. In Berlin's (1995) study, for example, when asked why she wanted to work there by the human resource manager, she spoke of her admiration of the organization and, it can be inferred from the article, assured the manager of her loyalty to the organization.
Developing the necessary skills. Covert observation can also call for more complex preparation, particularly for participation in professional or technical roles in which specific skills are needed. For instance, to work alongside and observe accountants and auditors requires that the researcher possess relevant competences, without which they are likely to be kept away from the organization's life (Stenger & Roulet, in press;Sullivan, 1959). It is possible that researchers could partly "fake it" (Roth, 1959), though this could harm their credibility, and their ability to observe. Nonetheless, it is highly likely that a researcher performing covert observation will have to have provide answers to questions concerning their identity from some of the individuals they are studying.
There are a number of options available to researchers to support the construction of a credible identity. The first is to draw on relevant experiences. To get hired and to work as a factory worker at Subaru, while conducting covert observation, Graham (1995) built on previous experiences as a factory worker, so as not to appear as an outsider. The second is to complete the training necessary to appear competent in a relevant role. Sullivan et al. (1958) noted that they had to complete 9 months of training before being able to enlist themselves in the military program they sought to observe. The third is to recruit fellow researchers who possess the relevant capabilities to present a credible identity. Bernstein (2012), for example, employed three of his Chinese undergraduate students to conduct his study of a Chinese production line, rather than participate himself. As he explained, "the students' personal characteristics were typical of new recruits, and the extraordinary diversity of the migrant labor pool meant that the students' small idiosyncrasies and any potential lingual accents went unnoticed. As college students, the researchers' age approximated the age of the average recruit, allowing them to blend in" (Bernstein, 2012, p. 185).

Recording Data Surreptitiously
Being a participant, and thus performing a specific role within the organization, may prevent the participant observer from recording data in more conventional ways. Furthermore, this needs to be done without raising the suspicion of the observed participants. Formal interviews, where participants are directly asked for their account of the investigated topic or collection of archival data, are not always possible without receiving unwanted attention (Burgess, 1984). For example, Thorne (1980) discusses how in his research on the illegal draft resistance movement, asking questions and taking notes led many of the participants in his study to suspect him of being a federal agent, thereby limiting his access to data and putting him at risk. As a result, covert participant observation usually generates informal data and research field notes rather than concrete quotes (Vinten, 1994). Erving Goffman (1961) also reports a number of difficulties in collecting data through a covert participant observation. For example, he stressed that he had to rely more heavily on his memory, as he could not take notes at every moment of his observation.
Approaches to collect and record data during covert participant observation. One of the most common approaches to recording data during covert observation is to employ a field diary, as E. Goffman (1961) did, to record observations at the end of the day. Dalton (1959) also relied heavily on work diaries. Diaries are particularly helpful for qualitative researchers to help them "reflexively notice their noticing" (Hibbert et al., 2014, p. 286). Researchers can create opportunities to write their notes by "slipping away" and finding places where they will be alone such as going for a walk (Festinger et al., 1965(Festinger et al., /2008 or to the bathroom (Berlin, 1995). Creative or improvised ways to collecting data may also emerge, such as Graham's (1995) use of a clipboard she used to carry out her daily job on the production line also enabling her to record observations. Stenger and Roulet (in press), for example, explain how the researcher would use Word to take field notes, making supervisors think he was working.
The process of recording notes can be managed more easily when operating in a research team. For instance, Festinger et al. (1965Festinger et al. ( /2008 noted that "when all of our observers were very fatigued and unwilling to trust their memory too much, one would go make notes while the others stayed to listen." Bernstein (2012) waited for his team to report to him in an isolated room, as the observers would take advantage of bathroom breaks to visit him. He would record their regular reports and have a debriefing with them at the end of the day on the basis of the transcripts he had 506 Organizational Research Methods 20 (3) produced. The observers were then "given a chance to confirm or challenge each other's perspectives" (Bernstein, 2012, p. 187).
Collecting data beyond the covert participant observation. Observation data may also be complemented with interviews at the time of exiting the field. L. Clarke (1996) notes that observation can provide basic insights, which can then be used to guide interviews in which the researcher can more overtly bring before the participants to discuss some aspects of their experience. After the covert participant observation, Bernstein (2012) took advantage of a debriefing and the observers revealing their true identity to deliver a survey and exit interviews. Bernstein (2012, p. 187) notes "surprisingly positive" reaction from the subjects. Covert participant observation can provide data that can be consequently triangulated and validated using other overt approaches such as interviews.

Reducing Harm to Participants
There are risks to the participants who are studied covertly. S. Clarke (1999, p. 155) highlights two harms that may occur when the infiltration of a group leads to a research output such as a publication in the public domain. First, the unwitting participants will typically suffer from feelings of betrayal if they realize that a person, whom they may have trusted, was in fact a researcher studying them. Second, whatever benefits they derived from being secretive may be lost should their activities become public. This is likely to be particularly true of marginal groups. It is therefore essential that a researcher considers the risks to their participants and, where possible, seeks to employ strategies to minimize potential harm.
Debriefings. S. Clarke (1999) claims that the most important strategy to reduce the harm to participants is debriefing, which is compulsory for researchers who adhere to the APA's standard on deception in research (for an overview of the key aspects of a debrief, see Sieber, 1983b). Lauder (2003) also mentions debriefing sessions as a way to reveal one's identity as a researcher at the end of a study and Bernstein (2012) used it as such. Nonetheless, this is typically recommended for deception experiments that occur in a laboratory or controlled setting. When debriefing, the researcher can share the original purpose of the research. In some case, there can be some direct insightful findings for the subjects. For instance, Milgram's follow-up study with his participants a year later indicated that 98.7% of his participants found the study and debriefing to be worthwhile experiences, as they were made aware of the danger of obedience and the importance of challenging authority (Sieber, 1983a). In contrast, Humphreys (1970aHumphreys ( , 1970b did not provide participants with a debriefing at any stage of his covert participant observation. Sieber (1983a, p. 5) suggests of Humphreys's decision that "[m]ost people would agree that this was the right decision, since debriefing would have horrified some of his subjects and their subsequent worries might have been life-long." Preserving anonymity. Given that debriefing may not always be appropriate or advisable in a field setting, a further way to reduce the risk of participants being adversely affected is to ensure their anonymity throughout the research process (Queen, 1959). This also means ensuring that it is not possible to identify or deduce the identity of the participants in the study in any associated research outputs. It is quite common that respondents are aware of the specific characteristics that make them identifiable by their stakeholders and thus request from the researcher a special degree of care (Shymko & Roulet, in press). Even in Humphreys's extreme form of covert participant observation, he stressed his desire to protect those he observed. He noted that "I have tried to make it impossible for any close associate to recognize the real people behind the disguised composites portrayed in this article" (Humphreys, 1970b, p. 25).

Leaving the Site of Study
Many researchers who have employed covert participant observation have struggled to identify when they should leave their study or have found leaving a challenging experience. Reflecting on his own experiences of researching mental health wards, with seemingly ambiguous consent, Taylor (1991) highlighted the need to consider the practical issue of when a study is complete as well as the personal issue of the relationships formed with a researcher's participants.
Completing the study. Covert studies seek to develop a deep understanding of a particular field and the embedded participants. Most established methodologies offer guidance on when to conclude the data collection necessary to build this understanding. For example, grounded theory (Glaser & Strauss, 1967) provides the concept of theoretical saturation. The underlying premise of this and other principles of collecting qualitative data is that the researcher should stop when no new insights are generated and the data become repetitive. This is an important point to reflect on in covert studies, and this point has been applied to Humphreys's work (1970aHumphreys's work ( , 1970b: He could have left the field of study earlier (Sieber, 1983a).
Managing the personal relationships formed with participants by easing out. As Taylor (1991, p. 238) points out, leaving the field is not simply a matter of wrapping up a study, but of "dealing with a change in how one relates to the people one has studied." Indeed, many covert studies report the development of friendships and forming close bonds with the participants being observed (e.g., Calvey, 2008;A. Goffman, 2015). To many of these researchers, their research can become allencompassing. Alice Goffman (2015), for example, explained how she struggled to go back to her normal life after her participant observation. It is thus important for researchers to consider their departure not just in the practical terms of when they have collected sufficient data but also in terms of their relationships with participants. When researchers seek to conclude their data collection, other scholars have suggested that they should "ease out" of the study (Junker, 1960), that is, gradually reduce the frequency of their interactions with the field. While this advice is typically applied to overt studies, we believe it remains valid for most covert studies. Cutting off contact too abruptly is likely to alert the participants to the absence of the researcher and, possibly, to raise concerns for the well-being of the researcher. A gradual departure therefore also necessitates the researcher providing an explanation or justification for the reduction of contact. For example, Calvey's (2008, p. 911) covert researchwhere he acted as a bouncer or nightclub doorman in the same city as his research institution-meant that he often bumped into former colleagues, to whom he would explain he was in "early retirement" or that he "couldn't stand the pace any more." Thus gradually easing out of the study, while also providing an explanation as to why, allows covert researchers to gradually "drift off" (Glaser & Strauss, 1967) from their participants without arousing suspicion.

Approval From Ethics Boards
Researchers are likely to face significant institutional constraints in seeking to conduct covert participation observation. University ethical boards, institutional review boards, professional associations, and other key stakeholders may refuse to approve covert participant observation (Becker, 2008), for example when inadequately justified. As Tope, Chamberlain, Crowley, and Hodson (2005, p ethics committee today." Such reservations stem from an often legitimate desire to minimize liabilities. Some ethical boards and professional codes of ethics may be inadvertently restricting research in ways that seem more likely to limit research than to protect subject and informant rights. In the meantime, some scientific outlets do not necessarily check for the ethical standards of the research or whether consent has been duly collected, even when the association they are affiliated with state strict guidelines. As such, ethical review boards play a vital role in protecting researchers and their host institutions, but also the participants and wider research profession. We suggest that researchers should, therefore, engage with these boards when developing their research plans. Although we have demonstrated the value and precedent set by prior covert studies, and outlined different ethical frameworks to consider this value, there are further ways to make a case and strengthen requests to employ this method. Postobservation consent. One option, in accordance with the ethics code of some professional associations (e.g., the APA), is to seek agreement from ethical boards to allow consent from participants following the collection of data. This would require detailed and thoughtful consideration of the postobservation and debriefing processes. Participant observation in this case has remained covert, but the debriefing ensures that deception was limited in time and that there is a feedback process. Such an approach, however, raises a number of risks. Participants may react negatively, and if they refuse to provide their consent, a large amount of data might be unusable. Furthermore, debriefing may not always be appropriate or an available option when it would jeopardize the researcher's safety.
Using the two-dimensional aspects of consent. A further option is to carefully consider the breadth and depth of the consent required for a proposed study. The earlier notion of a continuum of consent or dimensions of deception (see Figure 1) is important here because it underpins the idea that researchers can capture consent in a variety of different ways. It also stresses the practical challenges inherent to all observational studies, and can thus help ethical boards put covert observation into perspective with previously approved research projects for which consent was more or less pervasive. For instance, consent can be secured at the institutional level from ethical boards or universities, the organizational level through managers or other gatekeepers as well as at the individual level from the participants themselves. An observational study does not necessarily have to be simplified to as purely covert or purely overt. Ethics boards and researchers can work together to explore this idea and to identify effective ways to protect participants by blending covert and overt aspects of a study (for an example of jointly developed approach to observation, see Oeye et al., 2007).
Making a consequentialist case. As we previously discussed, the most common ethical justification is based on a consequentialist argument. Covert research is acceptable if the social benefits outweighs the cost to the participants (Lauder, 2003). Researchers can use such a justification to convince ethical boards by explaining how the harm to the recipients is minimized (e.g., stressing the preservation of anonymity), and how the positive social impact is considerable. Although covert research commonly refer to consequential arguments, they rarely list the actual benefits of the study in a rigorous and exhaustive manner. Leo (1995) claims his work unveiled police officers' misconduct and abuse of power, and can ultimately both contribute to better policing and a fairer society. In the case of the Bernstein's study (2012), the moral case for the transparency paradox is limited, although it can be beneficial for the organization's efficiency. Indeed, as this research suggests the workers on the factory line can improve processes, which can convince managers to trust their employees and give them more autonomy. Thus, when pitching their project to ethical boards, covert researchers could explain how each stakeholder can benefit from the project. This  justification can also be useful in the methods section of the article themselves so that future covert researchers can build on those justifications to make a case with their ethical boards. Table 4 summarizes each of the practical considerations of covert participant observation in terms of techniques to address each challenge, associated risks and benefits, alongside illustrative examples. It is important to stress that not all of the proposed techniques may be applicable or possible in all covert studies.

Conclusion
Due to the profound role of covert research in shaping social science both historically (e.g., Festinger et al., 2008;E. Goffman, 1961) and more recently (e.g., A. Goffman, 2009Goffman, , 2015, many professional academic associations in psychology and sociology continue to permit the use of deceptive approaches, including covert research in specific circumstances. This permission reflects the belief that sometimes covert research can be justified by the prospect of significant scientific, educational, or applied value. In an effort to support the realization of similar value within the field of management and organizational studies, where covert studies remain rare, we have focused on covert participant observation. Covert participant observation can enable researchers to gain access to communities or organizations and to collect knowledge that would otherwise remain unavailable. In some situations, covert participant observation can help create knowledge to change society for the better (Lauder, 2003) or open entirely new fields of inquiry for social scientists (E. Goffman, 1961).
Nonetheless, the justification of covert methods is context-dependent (Oliver & Eales, 2008) and brings with it many potential criticisms alongside risks to participants and researchers. As such, the aforementioned benefits of covert participant observation are available only to researchers who can carefully justify their use of the method, consider thoroughly the participant observation processes, and ensure the anonymity and protection of their participants. To support researchers in considering covert methods and their implications, we suggest that all observational studies be situated along a two-dimensional continuum of participant consent: The breadth of the observed subjects knowing about the researcher's identity matters as much as the depth of their understanding of the research's purpose. Observational studies are rarely fully covert or fully overt but, instead, usually situated somewhere between these two poles. Thus, overly simplistic categorizations of observations as either overt or covert not only preclude a potentially valuable research method but also prevent further reflection on the realities of observational research. We hope this work can bring about some reconsideration of the AOM's Code of Ethics.