Behind Closed Doors: Recruiting Female Escorts for Research Purposes

Abstract

We reflect on the different methods used to recruit cisgender female escorts for qualitative research into the social and professional context(s) they inhabit. Sex workers in general, and escorts in particular, are difficult to access as they occupy largely opaque spaces within the sex industry. This poses various challenges for researchers in the recruitment of participants. This article considers three core problems linked to the recruitment of socially excluded populations: (1) where and how to find participants; (2) how to motivate potential participants to participate; and (3) how to avoid selection bias. These can be potentially overcome by the thorough preparation of the recruitment strategy, which will be described for a qualitative research project on cis female escorts working in Belgium and the Netherlands. We start by describing the creation of online and offline recruitment materials and the role of social media as a platform to recruit participants. Further, we state that effective collaboration with gatekeepers is essential to successfully recruit within marginalized populations, as well as the ethical and practical considerations regarding financial compensation for the time and expertise of participants. This article concludes by reflecting on the added value for participants to participate in the research project, such as altruistic and therapeutic results.

Introduction

‘Hidden and hard-to-reach populations’ are sub-groups that are difficult to access for various unique economic, societal, geographical, or legal reasons (Shaghaghi et al., 2011). Many sub-groups have already been labelled as such and an extensive literature has developed around the challenges faced when researching these populations (e.g., Abrams, 2010; Barros et al., 2015; Belt et al., 2022; Kennan et al., 2012; Rockliffe et al., 2018). However, the term ‘hard-to-reach populations’ is acknowledged to be potentially problematic, as it may implicitly place responsibility on the populations themselves rather than on systemic barriers (Bradway et al., 2025). Therefore, we prefer using the term ‘socially excluded populations’, emphasizing structural and social exclusion as a cause of difficult accessibility, rather than unwillingness on the part of the group itself. Of these, sex workers ¹ are arguably one of the most striking examples, facing high levels of stigmatization and discrimination. Conceivably, this partly explains the lack of research on the sex industry ² and the underrepresentation of the experiences of sex workers in scientific research worldwide (Allen et al., 2019).

Four main characteristics of the sex industry can be discerned in order to categorize sex workers as a socially excluded population. First, sex workers experience high levels of stigmatization and discrimination by different actors, such as the media, police, and other sex workers (Benoit et al., 2005; Bowen & Bungay, 2015; Bungay et al., 2016; Daniel et al., 2023; Shaver, 2005). To avoid the risks linked to stigmatization and discrimination, such as violence or loss of reputation (Ellard-Gray et al., 2015), sex workers employ coping mechanisms including isolation, selective disclosure, and the use of personas (Daniel et al., 2023; Ham & Gerard, 2013). As a result, sex workers have a higher likelihood of stress and depression than the general population (Benoit et al., 2005). Fearing careless use of their data and further stereotyping and judgment, sex workers are highly averse to sharing their personal information with, for example, researchers, police, or medical staff (Bowen & Bungay, 2015; Bungay et al., 2016; Ellard-Gray et al., 2015; Martin, 2013; Ślęzak, 2018).

Secondly, it is impossible to estimate the size and scope of the sex industry which makes it a challenge to create a representative image (Heckathorn, 1997; Shaver, 2005). For example, persistent stigmatization and discrimination makes sex workers often distrustful of outsiders (Benoit et al., 2005) and deters them from disclosing their occupation to others. The sex industry is also characterized by high levels of mobility, both geographically and between different types of (sex) work (Bungay et al., 2016; Johnston et al., 2006), making it no easy task to obtain numerical data. In addition, contrary to the claims of many, sex workers are a heterogeneous and diverse group, which makes their needs, habits, experiences, and expectations equally diverse (Benoit & Shaver, 2006; Bungay et al., 2016).

Thirdly, sex workers are labelled as legal outcasts, as many jurisdictions criminalize (peripheral) activities in the sex industry, which enhances the stigma they experience (Benoit et al., 2005). Criminalizing the purchasing, selling, or exploitation of sex work creates a barrier to identifying as a sex worker which increases the risks of prosecution by legal and governmental authorities, such as the tax authorities. Only a few countries worldwide, including Belgium since 2022, have decriminalized the sex industry ³ with the intention of improving the working conditions, health, and safety of sex workers, decreasing their stigmatization and discrimination, and enhancing their relationship with regulators (Open Society Foundation, 2015; Struyf & Bauwens, 2021).

Finally, practical factors make sex workers particularly hard to reach, such as language barriers, unstructured lifestyles, and flexible working hours (Martin, 2013). These issues combined make it extremely difficult to reflect the diversity of the sex industry in scientific research.

Many researchers have successfully recruited sex workers by using different strategies. For example, in San Francisco and Montreal, Shaver (2005) occupied designated prostitution areas during working hours and introduced the research team to people who appeared to be working on the street. Bungay et al. (2012) initiated a respondent-driven sampling method by personally visiting massage parlors and escort agencies. ⁴ However, these strategies are not easily applicable to recruiting all types of sex workers and may be considered intrusive. Many escorts are unlikely to work in well-defined locations such as particular streets or massage parlors, they rarely interact with colleagues, and are typically unwilling to share contact details of others.

Based on the doctoral research experiences of the first author, who interviewed cisgender female escorts working in Belgium and the Netherlands, we advocate for the thorough preparation of a recruitment strategy prior to undertaking research in the sex industry. During the interviews participants were asked to reflect on the recruitment strategies used to elicit their cooperation and their thoughts on the interview process. Their answers provided us with a unique insight into how they experienced participation in a scientific study.

We explore three questions linked to the recruitment of socially excluded populations: (1) where and how to find participants; (2) how to motivate potential participants to participate; and (3) how to avoid selection bias. These methodological and ethical reflections will contribute to greater understanding in undertaking research, not only of those working in the sex industry, but also other socially excluded populations.

The Study Context: Qualitative Research With Cisgender Female Escorts

As a global phenomenon, sex work elicits high levels of work-related physical, sexual, psychological, financial, or discriminatory violence, that can occur before, during and after the performance of their job, and perpetrated by various actors (Bungay et al., 2012; Deering et al., 2014; Kinnell, 2013; Kloek & Dijkstra, 2018). Sex workers often describe the violence they encounter, as well as other risks, as “part of the job” and thus an inherent aspect of their profession (Armstrong, 2017; Bungay et al., 2012; Dalla et al., 2003; Lyons et al., 2017; Sanders et al., 2018; Ślęzak, 2017). Most international research on violence towards sex workers is focused on specific types of sex work, such as street-based sex work, massage parlors, or brothels. Little is known about escorts, who advertise their sexual services through online advertisements and relocate to an often-unknown location, such as a hotel or the client’s house, to physically meet their clients. Relocating to an unknown location can create additional risk factors and necessitates alternative security measures compared to, for example, meeting a client in a brothel. According to Ślęzak (2017), escorts ⁵ in Poland who met their clients at the latter’s home, experienced the most brutal forms of violence. In these circumstances there are no managers, security workers, or other sex workers whose presence can provide a form of social control. Escorts were found to generally distrust customers and took several preventative actions to minimize safety risks before, during, and after an appointment.

Notwithstanding the high levels of violence compared to the general public, sex workers are highly resistant to reporting victimization to the police due to four types of fear: (1) fear of punishment for illegal activities (e.g., in case of illegal residence or in countries criminalizing the sex industry); (2) fear of maltreatment by law enforcement; (3) fear of exposure towards their environment and authorities; and (4) fear of impunity of the aggressor (Struyf, 2022). Consequently, little official insight is available into the prevalence of victimization among sex workers, and the potential safety risks they face. This results in assumptions made by authorities and the public, and ill-informed polarized discourses about sex workers, the sex industry, trafficking, coercion, and violence, which fail to acknowledge the complexities of sex work (Ham & Gerard, 2013).

The doctoral research ⁶ within which this article is framed, provides the basis for developing more nuanced insights into: (1) cisgender female escort’s feelings of safety and unsafety regarding their job; (2) the safety mechanisms they apply before, during, and after an appointment with a client; and (3) the role of the police as an actor of formal social control. To answer these questions, the first author conducted in-depth interviews with 28 cisgender female escorts who work in Belgium or the Netherlands. The recruitment of these women was accompanied by various challenges, which we address below.

Where and How to Find Participants?

Designing a thorough recruitment strategy at the beginning of a research project can avoid the sense of being overwhelmed in the field, can help anticipate potential challenges during the recruitment phase, and can prevent societal and personal harm to the participants (Dewey & Zheng, 2013; Thummapol et al., 2019). This strategy should primarily focus on the materials necessary to address the targeted population, the channels through which the materials will be communicated, and the ethics associated with the direction of the project.

To ensure its feasibility and effectiveness, several stakeholder group meetings were organized at the beginning of the project, in which gatekeepers, sex workers, and other experts were asked to give feedback on the recruitment strategy. For example, the stakeholder group decided on the languages the recruitment materials should be provided in (infra) because they have a good understanding of the demographic characteristics of the population. They also helped to decide on the terminology used, and the financial compensation offered (infra). By involving stakeholders in the development of the recruitment strategy, we also gained their trust at an early stage of the research and found them willing to actively participate in recruiting participants. Additionally, this stakeholder group increased the perception of potential participants that the research was community-supported, increasing their trust towards the research team (Bonevski et al., 2014; Lobo et al., 2020).

Creating Recruitment Materials

In a communicative environment where digital media is omnipresent and essential in encouraging the interest of the target audience (Russmann & Svensson, 2017), it was necessary to create a strong and appealing visual identity for the project. An advertising agency was assigned to design a visual identity, including a logo, recruitment flyers, posters, and social media posts (see Figure 1); all produced in five different languages - Dutch, French, English, Spanish, and Romanian. For the logo, the worldwide symbol for sex work, a red umbrella, was integrated into the design, as it is not commonly recognized as such by people not involved in the sex industry. In this way, a certain degree of discretion could be guaranteed (Ellard-Gray et al., 2015). The visual materials were titled “Safety at work as an escort” and included the following information: (1) the logo of the research project and the supporting university; (2) the inclusion criteria for participants; (3) the possibility of a financial incentive (€50 per hour); (4) the guarantee of complete discretion and anonymity; (5) the contact details of the researcher ⁷ ; and (6) a QR-code that links to a secure online application form.

Figure 1.

Recruitment Poster (English)

We purposely chose the word “escort” in all communications because the term is widely understood within the Belgian and Dutch sex industries. However, using this specific term could also exclude potential participants who are technically working in this type of sex work, but who do not identify themselves as escorts, or as sex workers in general. This placed a potential, but unavoidable, limit on the diversity of the sample (Ellard-Gray et al., 2015).

Online Recruitment

In the first phase of recruitment, potential participants were targeted exclusively through online methods. The online presence of escorts has significantly increased, reflecting the necessity of online advertisements to find clients, both on websites promoting sex work and individual social media (Cunningham et al., 2018). This wider visibility creates new recruitment opportunities for researchers. According to a systematic review undertaken by Barros et al. (2015), internet recruitment appeared to have the highest response rate when recruiting many, but not all, socially excluded populations, with some groups (e.g., homeless people or the elderly) likely to have a limited online presence.

We began by posting an advertisement regarding our research on Redlights.be, the biggest website for promoting sex work in Belgium, which receives 90.000 visitors (clients) on busy days and holds contact details of 9.611 sex workers (J. Meyts, webhost Redlights.be, personal communication, October 30, 2023). In consultation with the webhost, we placed a free banner on the home page of the website with the information mentioned above, so that interested escorts could make contact and ask for more information or make an appointment for an interview. The first four interviews conducted in this research project were a direct result of the banner on Redlights.be.

In addition to the website, we shared online recruitment materials via the Instagram account especially created for this research (@SafeEscorting, see Figure 2). With the use of hashtags and reels (short videos) on Instagram, we were able to directly address the target audience. Interested escorts could contact the researcher via direct messages on Instagram, via email, or WhatsApp.

Figure 2.

Profile Page Instagram Account @SafeEscorting

The increased online presence of sex workers also translates into a greater presence on social media (Cunningham et al., 2018), which makes it important for researchers to actively use these tools to communicate with the targeted audience (Dusek et al., 2010). Furthermore, an increasing number of researchers report that they use the internet and social media as part of a broader recruitment strategy (Gioia et al., 2016; Gundur, 2017; Hashemi et al., 2024). As Gundur (2017, p. 7) states: “Using the Internet as a medium to connect with potential respondents is a strategy that should not be ignored given the rapidly increasing use of the Internet in all segments of society since the turn of the twenty-first century.” In our case, the Instagram account allowed us to reach potential participants that we would otherwise not reach using offline methods.

Over time, the Instagram account also became a place where interested parties could receive updates about the research, such as publications and presentations. Sharing study findings, and increasing community and participant engagement during the study, arguably contributed to reducing concerns harbored by a traditionally distrustful group towards research (Ellard-Gray et al., 2015; Preble, 2015). We deliberately chose Instagram as a tool for recruitment and information as the first author was familiar with the platform, and many gatekeepers were already present on it. Although it is possible to connect several social media platforms in the back office, we would recommend choosing only one. Managing several accounts at the same time is very time-consuming because of the need to respond to comments and personal messages from potential participants or interested parties in a reasonable amount of time.

Critics might argue that recruiting participants through social media constitutes a threat to their privacy, because a researcher collects the personal data of participants. Gundur (2017), however, argues that Internet recruitment is no more intrusive than traditional methods, as researchers expect the potential participants to initiate the contact. An important initiative to protect the personal data of participants is by using end-to-end encrypted programs, such as WhatsApp, to share this information (Gundur, 2017). This conforms to traditional data security protections used to secure any sensitive data for research purposes. A related, albeit ethical, issue to the privacy of participants is the question of whether to follow a participant on social media. Participants have the agency to follow researchers, however, researchers following participants is more intrusive and can potentially compromise the anonymity of the participants. Hence, we chose not to follow our participants, even if there were requests to do so.

Because the account contained keywords such as “escorts” and “sex work”, the first author received indecent proposals and pictures from men via Instagram or WhatsApp on more than one occasion. Whilst deleting the messages, blocking the accounts, and venting to her supervisors and other colleagues, it helped maintain her boundaries, these experiences are rarely discussed in ethics boards and the literature. It is striking how attentive we are when it comes to the safety and mental well-being of our participants yet often forget our own safety and mental well-being. At no point in our application to the Ethics Committee is the potential risk to the researcher probed. At no point during the preparation of the study did we as a research team reflect on these potential risks and how best to avoid them. These two observations confirm the normalization of field research in potentially high-risk contexts and the normalization of potential risks to the researcher (Bahn, 2012). After experiencing several situations of online sexual harassment, the first author never reported these incidents to the ombudsperson of the university or the police. In the Netherlands, researchers who experience unsafe situations linked to their work, can file a report with the SafeScience Project, ⁸ although this initiative is not yet widely distributed. Therefore, there is an urgent need for more awareness at both the state level and at the level of research institutions, ideally resulting in guidelines regarding ethical, legal, and social implications for researchers who conduct research in risky environments or actively use social media as a recruitment tool (Gerards, 2024; Nebeker et al., 2020). Researchers should be encouraged to reflect on potential risks at the start of their research and while preparing the recruitment strategy and data collection plan, preferably in their application for the Ethics Committee. Possible strategies could include conducting field work in pairs and sharing live location with a trustee while conducting interviews at the participants’ home (Bloor et al., 2010).

Social Media Tips and Tricks

Prior to establishing a social media account for research, we suggest thorough preparation and planning. We advise to reflect critically on what type of content is, and is not, to be shared, and the language used to communicate with followers (we used both Dutch and English). In addition, a strong visual identity makes a ‘brand’ more recognizable within the target group, and for gatekeepers. As a result, this would thus benefit the process by clearly defining how this could be developed and implemented. By planning your posts ahead of time, you can also consider important dates for your targeted audience. For example, we would post on dates like International Women’s Day and International Day to End Violence Against Sex Workers. These curated posts could ensure that your followers feel recognized and acknowledged.

To build trust with followers and attract potential participants, it is important to personalize the account by revealing your identity. When the first author made a video of herself, explaining the objectives of the research and what we were looking for, the number of followers increased, and more than thirty potential participants contacted us over a period of three days. The video was shared in Belgium and the Netherlands by multiple aid organizations that work with sex workers, and the post had the most likes up until that moment. It was notable that during the interviews, several participants referred to the video as their final push to contact us for participation and highlighted it as a source of trust in the researcher and the research. Simultaneously, the video exposed the researcher and increased the risks of harassment, which should be reflected on at the start of the recruitment process.

Social media is all about engagement, so we recommend following accounts that are linked to your topic, such as those of gatekeepers and spokespersons. By using this method, even organizations that we did not directly contact to share our content, shared our posts on their accounts and referred potential participants. It was noteworthy that whenever an organization shared one of our posts, the engagement on our account increased (new followers and new likes), and new individuals signed up for the research. This once again underscores the importance of thorough research and the gatekeeper’s wider commitment to facilitating scientific research.

Offline Recruitment

At the same time as the online recruitment, we addressed potential participants indirectly by engaging gatekeepers to support our recruitment, through both online and offline methods. Flyers and posters were sent to several aid organizations and interest groups acting for sex workers in Belgium (n = 4) and the Netherlands (n = 5). Gatekeepers were asked to present the posters and flyers at their offices and distribute them amongst possible participants. A few weeks after the distribution of the recruitment materials, several organizations contacted us to inform us that they had found interested escorts who were willing to be interviewed. Interested escorts would either contact the gatekeeper, who would then refer to the researchers if permitted to share contact details, or they would contact the researcher directly. Some interviews took place at the offices of the aid organizations; other interviews took place at the escorts’ house or a public place nearby.

How to Motivate Potential Participants?

When a potential participant contacted the first author via Instagram, WhatsApp, or email, she replied with the following information: (1) an expression of gratitude; (2) a reminder regarding the inclusion criteria for participation; (3) three possible dates to schedule the interview; (4) three possibilities regarding the location of the interview and, (5) the informed consent form as an attachment. It was important to be as flexible as possible in terms of time and location, and the choice of where the interview would take place was always left to the participant (at their home, at a public place, or at the university). Accommodating these preferences can enhance a participant’s attitude towards the interview and fosters trust towards the researcher (Elwood & Martin, 2000). The main priority was sensitivity to the participants’ comfort and safety, which resulted in many hours of traveling by car in Belgium and the Netherlands. In this context, it was equally important to ensure the physical safety of the researcher who would be meeting with an unknown participant in an often-unknown location. Therefore, the interview location was always shared with her partner and her live location via WhatsApp was always apparent until she was back in her car; a safety strategy learned from the participants.

In total, over 60 people reacted and showed interest in participating in the research, of which 28 cisgender female escorts were interviewed. There are various reasons to explain why less than half of the initial contacts resulted in an interview. The most common reasons included people who did not react to any form of communication after the initial contact or those who did not meet the inclusion criteria. Participants were selected on five inclusion criteria: (1) the participant must be of legal age to perform sex work in Belgium or the Netherlands (18 years or older) ⁹ ; (2) the participant must be cisgender female; (3) the participant works as an independent escort or works at an escort agency; (4) the participant must currently be, or have been, active as an out-call escort as her main or side occupation; and (5) the participant must work, or have worked, in Belgium or the Netherlands, independent of her nationality.

After approximately 18 in-depth interviews, no migrant and undocumented sex workers contacted the research team for participation. Due to the large international focus on human trafficking and the inherent link made to the sex industry by media and policy (Kim & Jeffreys, 2013), migrant and undocumented sex workers experience more fear of negative consequences when outing, such as deportation (Struyf, 2022). Additionally, a language barrier and cultural differences can make a trusting relationship very difficult to establish, especially for outsiders (Kim & Jeffreys, 2013). As a result, migrant and undocumented sex workers are reluctant to participate in research (Dahinden & Efionayi-Mäder, 2009). Although we initiated a more focused recruitment strategy for this missing target group, by reestablishing contact with the aid organizations and posting recruitment materials in other languages than Dutch and English on Instagram, responses were limited.

Insider led research might be the most appropriate method as their pre-understanding of the shared situation might increase trust and efficiency (Kim & Jeffreys, 2013). Such insider led research can be accomplished by peer interviews in the language of the participant, which simultaneously eliminates possible misunderstandings or misinterpretations. For this research, we did not have the funds nor the time to properly train peers, but it would be highly recommended in future research to include the lived experiences of this underserved population.

Although we did not fully apply this to the study, due to a limited budget and differing study objectives, applying a community-based approach ¹⁰ in scientific research among socially excluded populations can have a positive effect on recruiting participants (Jobe et al., 2022; Lobo et al., 2020; van der Meulen, 2011). In our research, we compiled a stakeholder group that participated in the preparations of the recruitment strategy, but also peer researchers could offer multiple advantages. According to Lobo et al. (2020, p. 2) peer researchers, for example, (former) sex workers who participate in the research “enhance the team’s understanding of sex work, bringing insights gained from shared experiences, including shared cultural identity and language, and shared demographic and socio-economic profile.” Peer researchers already have ‘their way in’, which can facilitate the recruitment and trust building process (Jobe et al., 2022). Nevertheless, collaborating with peer researchers also faces some limitations, as it does not automatically grant access to the entire population (Chavez, 2008). Peer researchers can be biased when recruiting known individuals, have less of an objective due to personal investment, and can make assumptions based on prior experiences (Chavez, 2008). These challenges should be considered when designing the research project and recruitment strategy.

What’s in it for Them?

Researchers can often ask a lot out of participants. We expect them to make time to engage with us, and share what are their sometimes traumatizing experiences. Ethics boards rightly ask researchers to thoroughly anticipate and reflect on the possible negative impact of their research on participants. As a result, research is often conducted from a protective reflex (Dyregrov, 2004). However, participating in research can also be a positive experience for participants, as reported by several international studies (Alexander, 2010; Decker et al., 2011; Dyregrov, 2004; Jaffe et al., 2015; McDonald et al., 2013; Newman et al., 1999). In other words, the potential for harm from participating in scientific research does not necessarily outweigh its benefits for participants (Alexander, 2010).

Nevertheless, researchers must reflect on the return of investment for the participant’s involvement. In our opinion, the outcome of scientific research ought to be characterized by mutually beneficial results, with researchers contributing positively to the lives of their participants. Based on Alexander’s (2010) typologies, we identified what potential gains might motivate people in marginalized contexts to participate in research and considered the following six areas: (1) therapeutic and social; (2) reflective; (3) educative; (4) empowering; (5) altruistic; and (6) financial. Throughout the interviews, most participants conversed about why they took part in the research, their experiences, and what meaning it gave derived from the interview. Their thoughts below reflect the positive elements of the interview. The participants names are pseudonymized, and some quotes have been translated to English (interviews were conducted in Dutch, English, or French).

One of the reasons why female escorts participated in this research was due to the therapeutic and social effect of the interview. By sharing their stories with a considerate, non-judgmental, and supportive listener, participants indicated that they felt heard and acknowledged. For some, the researcher was the only person who was aware of their professional activities in the sex industry, who at times served as a confidante after the interview had taken place. To foster trust before, during, and after the interview, the first author was very transparent about her personal position on sex work, voicing herself as an advocate of human rights, in favor of the decriminalization of the sex industry. Additionally, in the context of many escorts having negative experiences with men, it was advantageous that the first author was female, which helped win their trust.

The stigma surrounding sex work often makes it difficult for escorts, keeping the nature of their profession from family and friends. Such participants are often isolated, unable to speak freely to those in their social network, but in the dynamic of the interview, they were able to vent unreservedly about their experiences without shame or prejudice. Being able to tell their stories to a stranger who was interested in them, what they had to say, and the subjects they cared about, offered discernible comfort and support. In addition, pointing out shared experiences between the participants during the interviews, reassured participants that they were not alone in their encounters.

It may sound silly, but I prefer telling it to a stranger than to a family member. They usually have judgments about it, so this is easier for me – Lisanne.

It is nice to be able to talk about it with someone who, in my opinion, sees it in the same way as I do – Hannah.

A second reason mentioned by the participants was the reflective effect of the interview, whereby the questions encouraged them to reflect on their lives and offered new insights on the meanings they attached to themselves.

Interesting, because I had to think about things I never thought about before – Lucy.

It was strange to talk about those things that I never discuss and to reflect on it. Thank you – Stephanie.

Less frequently mentioned by the participants was the educational value of participation. However, some were genuinely interested in the research and others asked legal questions regarding their businesses, taxes, or national regulations during or after the interview. Some started following the Instagram account to stay up to date about the findings of the research.

It was very interesting […]. I have gone from being unaware to very knowledgeable – Steffania.

Escorts placed significant value on their agency and those we met highlighted the empowering impact of the interview. While escorts are often discussed on several societal levels, they rarely have an active voice in representing their profession.

I feel very proud to be able to contribute to this research. For me, this is empowering. I think this is something very beautiful – Cataleya.

Escorts also referred to altruistic reasons for participating, while others were open about their activism with regard to the decriminalization and destigmatization of sex work. By sharing their stories honestly, they aim to assist others in the industry by breaking the stigma that surrounds their profession (Neelakantan et al., 2023).

The taboo remains enormous; the stigma is huge. If I can contribute to removing that stigma even a little bit, that would be fantastic – Chloë.

Finally, because we gave participants the option to receive monetary compensation, the possibility of financial motivations could also be present. Although participants were informed about financial compensation before starting the interview, they perceived this as a bonus, rather than the crucial motivator (Gundur, 2017; Neelakantan et al., 2023). Some even refused payment, prioritizing other motivations. Monetary payments for participation in scientific studies have been subject to ethical contestation based on the suggested coercive nature of financial inducement (Bentley & Thacker, 2004; McNeill, 1997; Wertheimer & Miller, 2008; Wilkinson & Moore, 1997). Critics argue that financial payment undermines the voluntary consent of a participant, is harmful to the well-being of a participant, that it is exploitative, and that it can result in selection bias and socially desirable answers (Bentley & Thacker, 2004; Wertheimer & Miller, 2008). Academics in favor of financial inducements argue that it offers people extra motivation to participate in a study. As Bentley and Thacker (2004, p. 293) phrase it: “Inducements are offers, not threats, and therefore they are not usually considered to be coercive.”

For us, there were two main reasons for implementing financial payments. First, by offering the possibility of being paid, we could compensate participants for their time and expertise. During the interview, participants were not able to meet clients and therefore lost an opportunity for income (Dickert & Grady, 1999), so offering them a reimbursement was a form of respect on our part. Secondly, several aid organizations that were crucial for successful recruitment would only cooperate if participants were offered financial compensation. These aid organizations argue that sex workers are overburdened and, in a certain way, also deserve protection against exploitation by journalists and researchers.

In the planning phase of our recruitment strategy, we consulted our stakeholder group to decide on whether to offer a payment, the amount, and the form (e.g., cash, bank transfer, gift card). We chose to leave the choice of accepting financial compensation to the participant, thus guaranteeing their agency and control.

How to Avoid Selection Bias?

Although we implemented a thorough recruitment strategy, undertaking qualitative research with socially excluded populations made selection bias difficult to avoid. Selection bias is the systematic exclusion of certain types of people, places, events, or behaviors during the data collection (Bell-Martin & Marston, 2021) and has an inevitable impact on the diversity of the sample. When the most vulnerable people of a marginalized group are, for whatever reason, not included in the research, discrimination and stigmatization become reinforced in policy and aid initiatives that derive from the findings (Bell-Martin & Marston, 2021). In our research, the different recruitment strategies resulted in a sample of escorts with particular characteristics, besides the initial inclusion criteria: (1) participants who identify as sex workers/escorts; (2) participants who have the opportunity to invest time in the interview; (3) participants who do the job voluntarily and are not coerced (at that moment); and (4) participants who spoke and understood Dutch, English, or French.

Gatekeepers

Not wishing to limit our sample to escorts in receipt of help from aid organizations we worked closely with, we also recruited participants through social media and online sources (Instagram and Redlights.be). Nevertheless, aid organizations took on the crucial role as gatekeepers and thus provided most of the participants. Hence, it is likely that we excluded escorts who sought or received help from more abolitionist organizations. In addition, we only selected aid organizations which target sex workers. As a result, we missed (former) victims of human trafficking who worked as escorts in a coercive environment. To avoid this in the future, collaborations with aid organizations who target various populations is recommended.

Language

The recruitment materials were only translated in five languages due to feasibility. Consequently, escorts with other nationalities, or limited language proficiency, might not have been aware of the research. Although we specifically targeted migrants in the second phase of the recruitment strategy, the response rate remained very limited. Therefore, research on escorts, or sex work in general, would benefit from researchers who are proficient in foreign languages that are prevalent within the location, such as Spanish and Romanian in Belgium. Researchers who do not speak the language of the participant, should consider creative strategies, using an interpreter may help militate against language barriers faced by migrant sex workers who may wish to participate in the study.

Accessibility

When recruiting socially excluded populations, researchers and the means of participation must be as accessible as possible. First, financial compensation gave escorts the financial flexibility to not work for a few hours and participate in an interview. Secondly, by sharing videos of the researcher on Instagram explaining the objectives of the research and the practicalities of the interview, we created familiarity which lowered obstacles to contact the researcher. However, in this research, participants played a more active role; an approach which may have deterred potential participants as not everybody has the confidence to reach out to strangers or perceive the recruitment materials as speaking to them. It would have been interesting to have been more proactive in making the research more visible at places where sex workers are present, such as public gatherings, protests, and other sex-worker led events to expand the pool of potential participants.

Conclusion

Conducting scientific research on socially excluded populations takes a great deal of time and commitment to develop and execute. It is undoubtedly challenging for junior researchers, which makes it essential to develop a comprehensively thought out, and sometimes creative, recruitment strategy. Whilst we report on research experiences with female escorts, we have suggested how the described practices, and our reflections on the applied methodology, can easily be related to other socially excluded populations, after the necessary adjustments have been made.

We have shown that when designing a recruitment strategy, researchers must consider the online and offline places where their potential participants are present, and how they can best be reached. To map this out effectively, gatekeepers will be a helpful source of information as they often have considerable experience with contacting target groups. To help with this task, making your research more recognizable with a unique visual identity can be beneficial, both for online and offline recruiting, as it tends to confer a degree of public legitimacy.

To motivate potential participants, it is crucial to adapt to the expectations of participants in terms of time and place. Additionally, researchers must reflect on the added value a participant can experience when taking part in a scientific study. We noticed that financial compensation was never the main motivation for our participants, but it was received as a form of respect and gratitude from our side. For some gatekeepers, a financial reward is a condition for cooperation.

Finally, when studying socially excluded populations, selection bias is a major limitation. We found that gatekeepers can both limit and ensure the diversity of your sample; we suggest that researchers should not be afraid to cooperate with organizations that do not have the same mission and vision. A multilingual research group can also minimize selection bias and could attract participants that would otherwise not be motivated to collaborate.