Understanding drivers of female sex workers’ experiences of external/enacted and internalised stigma: findings from a cross-sectional community-centric national study in South Africa

Abstract To deepen our understanding of sex work stigma, and to its drivers and their interrelation, we conducted an analysis using structural equation modelling of the South African National Sex Worker Survey. We enrolled 3005 women in sex work using multi-stage sampling across all South Africa’s provinces. Experience of external/enacted and internalised stigma was widespread. Non-partner rape, intimate partner violence and partner controlling behaviour (often expressions of external/enacted stigma) compounded internalised stigma. These experiences of violence, other manifestations of external/enacted stigma and food insecurity, were key drivers of internalised stigma, and often had an impact on mental health. We found that considerable protection against stigma emanated from viewing sex work positively. This resistance to stigma provided opportunities to shift the narrative. Reducing sex workers’ exposure to external/enacted stigmatising behaviour, including by enabling more to work indoors, and providing greater protection from partner violence and rape, are critical for better health and well-being. Ending the criminalisation of sex work is foundational for safer working conditions and better health outcomes for sex workers, similarly providing adequately funded mental and physical health and social care through sex work specific programmes.


Introduction
Stigma undermines sex workers' human rights and mental health and exposes them to extortion, violence, exclusion and discrimination, while reducing their access to support and care from their families, communities, health and police services (Stockton et al. 2020;Scorgie et al. 2013;Kerrigan et al. 2021;Mujugira et al. 2021;Coetzee et al. 2018).The stigma of sex work is often compounded by the intersecting stigmas of being HIV positive, having poor mental health and rape (Kerrigan et al. 2021).Although much attention is rightly directed at the harmful effects of enacted stigma on women in sex work, internalised stigma is very important and associated with enduring post-traumatic stress disorder and other harmful health outcomes (Stockton et al. 2020).Understanding stigma is critical for efforts to reduce stigma experience and build resilience among sex workers.
Stigma was classically described as having two intertwined dimensions, external stigma and internalised stigma (Goffman 1963), which were reformulated by Scambler and Hopkins as enacted and felt stigma (Scambler and Hopkins 1986).External/enacted stigma describes the actswhether physical, sexual, verbal, structural or symbolicthat express the diminished social value of the stigmatised identity and the 'disciplining' of the stigmatised individual through social and economic exclusion (Lemke 2002;Kerrigan et al. 2021).These acts are sometimes referred to as experienced stigma (Stangl et al. 2019).Internalised stigma refers to self-perceptions of spoiled identity that characteristically manifest in shame and self-blame, perceived loss of social value and social withdrawal.Recent contributions to the literature on sex work stigma have highlighted multiple forms of stigma.For example, Mujugira et al. (2021), writing about transgender male sex workers in Uganda, described four types of stigma: structural, interpersonal, internalised and anticipated.Structural stigma was defined as "cultural and societal norms, policies and laws that limit opportunities and resources available to stigmatised people".Interpersonal stigma was considered to be violence towards sex workers, such as rape.This would lead to internalised (felt) stigma, described as "negative beliefs a person holds about themselves and their identity".Anticipated stigma would be expressed in avoidance of health and related care.This approach to theorising stigma was similar in many respects to that used by Kerrigan and colleagues (Kerrigan et al. 2021) when developing a scale to measure the stigma of sex work through research conducted in Tanzania and the Dominican Republic.They identified four dimensions namely, external/enacted stigma, internalised stigma, anticipated and resisted stigma, the latter expressed through assertions of dignity.In this paper the terms external/enacted and internalised are used in discussion of the two types of stigma, recognising that internalised stigma includes shame and fear of external/enacted stigma but extends to self-blame and internalised loss of social value and withdrawal.
The dominant narrative of sex work stigma is that sex work results in a spoiled identity, however, the resistance narrative counters that sex work can be affirming (Wong, Holroyd, and Bingham 2011).To understand this, it is important to reflect that most women in sex work are poor and have experienced abuse and neglect in childhood (Footer et al. 2020;Beksinska et al. 2022).For example in Soweto, 79.5% of women entered sex work due to extreme poverty and/or homelessness (Coetzee, Jewkes, et al. 2017).As a consequence, they often have poor mental health and low self-esteem, and feel socially devalued (Jewkes et al. 2022).In this context, sex work provides an income which is higher than that earned by women living on social grants, or in very low-paid work (Richter et al. 2013) which may be a source of a little pride and fuels the stigma-resistance narrative (Wong, Holroyd, and Bingham 2011).It may also positively affirm femininity through sexual desirability, while also providing a sense of independence (Maccaro 2020).There is likely a cultural interpretation of engagement in sex work that views it through a lens of gender roles and norms.Although our dataset was a national study, and thus the population was ethically mixed, conservative gender norms emphasising women's sexual modesty and subservience predominate across most of South Africa's ethnic groups.These norms are associated with stigma in rape stigma research (Jewkes et al. 2022), but were not measured in this study.However, Black African constructions of femininity also emphasise that women are expected to be strong, to persevere in the face of hardship and provide for, and build, the home (Shai 2018).These characteristics resonate with a positive narrative of sex work enabling fortitude in the face of extreme economic adversity.
For our study, we developed a model of stigma that hypothesised that it would have the classical external/enacted and internalised dimensions and positive views of sex work.Within the structural constraints of our society, where sex work has been historically criminalised sex work and social norms are highly patriarchal, we hypothesise that sex workers' experiences of internalised stigma are influenced by the structural context of poverty and childhood trauma.
There is a larger literature on rape stigma than sex work stigma, and this shows the connections between poverty and experience of child abuse and neglect on self-esteem, self-compassion and depression, all of which are associated with internalised stigma (Boyraz et al. 2019;Jewkes et al. 2022).We hypothesised that intimate partner violence, non-partner rape and partner controlling behaviour are often manifestations of external/enacted stigma directed as discipline/punishments for engaging in sex work, and experiences of these may further fuel internalised stigma (Jewkes et al. 2022).We also hypothesised that poor mental health, particularly depression and post-traumatic stress disorder, would further drive internalised stigma (Kohli et al. 2014;Hakimi et al. 2018;N€ othling et al. 2022;Sirey et al. 2001).Research shows that internalised stigma may drive post-traumatic stress disorder, and the latter drive internalised stigma (Verelst et al. 2014).We also hypothesised that experiences of external/enacted stigma and of resistance to stigma this narrative, i.e. positive thoughts about sex work would impact internalised stigma.
To deepen our understanding of internalised stigma (shame and blame) among sex workers and to better understand its drivers and their interrelation, we conducted an analysis using structural equation modelling of data from the South African National Sex Worker Survey (Milovanovic et al. 2021).We set out from the hypothesis that internalised stigma would have structural and early childhood drivers (poverty and childhood neglect and abuse experience), be impacted by experiences of stigmatising behaviour by others including intimate partners, by experience of non-partner rape, and by poor mental health.We also hypothesised that women who felt better about themselves undertaking sex work would experience less internalised stigma.

Materials and methods
Multi-stage sampling was used to derive a self-weighting sample stratified by province.Recruitment drew on elements of respondent driven sampling methods for hard-toreach populations, which we used previously in research on sex work (Coetzee, Jewkes, et al. 2017).We stratified the sample by province and for most, one district was selected randomly.In the Northern Cape, we worked with a small outreach team based in one district.In KwaZulu-Natal and Gauteng Provinces, with larger numbers of sex workers, we spread recruitment across two and three districts, respectively.The sample size per province was based on a 2013 sex work population size estimation (Konstant et al. 2015).Initial recruits (seeds) were enrolled, interviewed and given three coupons to pass onto other sex workers in the district.We capped referral chains at 30 participants, with the average recruitment chain being 9-10 women.A full account of the methods and sample size calculation is available elsewhere (Milovanovic et al. 2021).
For inclusion in the study, participants needed to be assigned female at birth and currently identify as female, be 18 years or older, work within a selected district and have voluntarily sold or transacted in sex for financial gain (not necessarily paid in cash) in the past six months.For ethical reasons, individuals who were <18 years, or who self-reported being a victim of human trafficking were excluded from the study but referred to a partner project for appropriate assistance.
Interview guides and questionnaires were developed in collaboration with sex workers and peer educators and tested in cognitive interviews.The questionnaire and consent forms were professionally translated into ten local languages.Each partner organisation's sex work programme was asked to brief the sex work community (brothel owners and sex workers alike) in their district about the study to help ensure access and buy-in from key stakeholders.Peer educators were trained as interviewers.Data were collected from January to July 2019.Screening and informed consent were undertaken by a peer interviewer.Thereafter, participants completed a 40-minute jointly-administered survey.Participants were reimbursed about $15 for their time.
We collected demographic information and the measures shown in Table 1.The questionnaire captured demographic details and we asked 'if you had a choice to leave sex work would you leave?'.
Data analysis was conducted in Stata 17.0 and took into account the design of the sample.Categorical and continuous variables were described.Bivariable relationships were assessed using internalised stigma as a score, and dependent variable, through summing the seven items.The association between the covariates and dependent variable was calculated using multiple linear regression analysis adjusted for the district (treated as a stratum), and participants recruited by a seed were treated as a cluster and adjusted for using the xtreg command in Stata.The reliability of scale variables was assessed with the Cronbach's alpha.
We tested the latent variables for convergent and divergent validity (acceptable validity (average variance extracted AVE)>0.5]).This led to some adjustments to strengthen convergent validity.We dropped two of the five items from external/ enacted stigma with the result that the latent variable used stigma from police, a health worker and physical violence.The latent measure of positive views on sex work used four of the five itemsnot including feeling proud to be providing for the home.When we developed the questionnaire we considered that avoidance stigma would be part of internalised stigma, but we did not include the measures due to the poor convergent and divergent validity.
Structural Equation Modelling was conducted to assess the interrelationship between variables in our conceptual model and internalised stigma.We constructed a latent variable for the model outcome from seven variables indicative of internalised stigma.The correlation between each hypothesised variable and internalised stigma was then tested by building variable pairs.All associations were tested by running a full-information maximum likelihood method (FIML) to deal with missing values.This method was chosen over multiple imputations as it has been shown to yield superior results in SEM (Enders and Bandalos 2001).To assess the model fit of the observed data, we used the comparative fit index (CFI) (>0.95);Tucker-Lewis Index (TLI) (>0.95) as indicative of good fit (Tucker and Lewis 1973); and root mean square error of approximation (RMSEA) of 0.05 or less (Steiger 1990).
We fitted a path model using FIML estimation to model all available data.The final model was built based on theory and statistically meaningful modifications using backwards elimination to exclude putative variables that did not mediate any path (with significance set at the P < 0.05 level) from the exogenous variables to internalised stigma in order to ensure model parsimony.We added covariances selectively to improve model fit where they made theoretical sense.The model fit was good (RMSEA 0.044, CFI 0.965 and TLI 0.954), and discriminant validity and convergent

Results
Table 2 summarises participants' demographic characteristics.The mean age was 33.3 years (range 18-64 years).Only 22.2% of women had completed high school.Childhood abuse and neglect was very commonly reported.Most women (81.6%) were married or partnered.They sold sex in a variety of locations, but 64.6% always or sometimes sold sex on the street or in a public open space.Overall, 45.2% of women were food insecure.Overall, 52.7% of women reported having depression and 53.6%, post-traumatic stress disorder.In all, 52.7% of partners knew that she sold sex, and 81.1% of women were, or recently had been, in a relationship with a relatively or highly controlling partner, scoring in the uppermost tertile of the control scale.37.5% had experienced physical intimate partner violence and 31.9%sexual intimate partner violence.54.8% had experienced non-partner rape in the previous year, perpetrated by a policeman, client or another man.The bivariable analysis showed that past year non-partner rape, physical partner violence experience, relationship control, depression and post-traumatic stress disorder were all associated with internalised stigmatisation.37.1% of participants reported using drugs and or alcohol to help them cope with doing sex work, and this was associated with experiencing internalised stigma among those using both.
Table 2 shows that 83.5% of women would leave sex work if they had a choice, but notwithstanding this, in many respects sex work was perceived to bring positive aspects to their lives.Overall, 60.4% agreed or strongly agreed that sex work made them feel affirmed as women and a very similar percentage agreed it made them feel desirable (59.3%) (Table 3).Further, 61.4% agreed that it made them feel valued.Almost all women agreed that it made them feel independent (95.1%) and that they felt proud to be able to provide for their needs and those of their family (95.3%).
Table 3 shows that internalised stigma was quite frequently reported.23.9% agreed or strongly agreed that they could not face friends or family due to being sex workers, 40.0% were ashamed, 39.0% felt guilty, 34.2% blamed themselves, 24.5% felt they  were not worth anything, 11.1% had thought of ending their lives because they were sex workers and 19.1% felt sometimes that they deserved what happened to them because they were sex workers.Some forms of external/enacted stigma were particularly frequently experienced.80.3% had been verbally harassed or threatened in the past year because of being a sex worker, 44.2% had experienced sexual abuse because of being a sex worker and 38.0% had been beaten or threatened with a weapon in the past year because of being a sex worker.Less often participants had been treated rudely or denied help from health care workers (17.1%) or the police (19.7%) in the past year because of being sex workers.8.0% agreed or strongly agreed that they avoided going to the clinic due to fear of the treatment they would receive and 11.9% avoided going to the police.The Structural Equation Model showed that there were no direct paths to the latent variable for internalised stigma from childhood trauma, rather there were a myriad of indirect paths (Figure 1 and Table 4).Most proximally, the analysis of total effects shows that experiences of past year non-partner rape and external/enacted stigma had the largest total effects on internalised stigma, with the direction of effect showing that they increased it.The total effects of food insecurity and mental ill health were also notable, and to a lesser extent those of childhood trauma and intimate partner violence.The model also shows one strongly protective path from childhood trauma to internalised stigma that is mediated by having positive views of sex work.This was very powerful in mitigating internalised stigma.Childhood experiences of neglect and abuse negatively impacted internalised stigma through increasing the likelihood of current food insecurity, poor mental health, experience of non-partner rape and intimate partner violence, and of external/enacted stigma.

Discussion
Our findings provide further evidence of the high prevalence of having feelings of shame or blame (internalised stigma) around sex work.It also showed the very high prevalence of external/enacted stigma experienced by sex workers, as well as intimate partner violence and partner controlling behaviour, which are often themselves external/enacted stigma, and mental health problems.We have shown that the experience of these compounds internalised stigma.The greatest total effects exacerbating internalised stigma come from experience of non-partner rape and of external stigma, followed by having poor mental health and food insecurity.Having a controlling partner and experiencing intimate partner violence contribute a smaller amount.The significance of the measure of partner control can be understood by examining the locally developed items.They are framed to capture partners' actions demonstrating, very clearly, opposition to engagement in sex work, and clearly reflect stigmatisation of sex workers by their partners.We have also shown a very powerful protection that comes from viewing sex work as positively impacting on women's lives.
Our findings show that internalised stigma is strongly rooted in experiences of the behaviour of others, which finds resonance in Scrambler's argument that a key element of internalised stigma is fear of external stigma from others (Scambler 2004).We have seen the association between the stigmatising behaviour by others, likely expressed as non-partner rape (although being in sex work is not the only risk factor for rape) and followed by subsequent mental health problems.External/enacted stigma also increases the risk of experiencing intimate partner violence and controlling behaviour from partners, which undermine feelings of self-worth, impacting internalised stigma.Ultimately the impact of external/enacted stigma on internalised stigma was mediated by depression and/or post-traumatic stress disorder, or else by partner controlling behaviour.In so doing, it may have also intersected with stigma of mental illness and related feelings of inadequacy.We have previously shown in this dataset that internalised stigma leads to mental ill-health (Jewkes et al. 2021).However, the analysis we present here shows that the association is bidirectional and thus poor mental health also fuels perceptions of internalised stigma.There are many barriers for sex workers in accessing mental health care, unless it is provided through sex work programmes, however both cognitive processing therapy and cognitive behavioural therapy have been shown to reduce stigma after rape, which undoubtedly intersects with the sigma of SW and so are likely beneficial (Lomax and Meyrick 2022).They have been successfully delivered by trained and well supported lay counsellors, which is a realistic level of technology for South Africa sex work programmes.
Partner violence and rape have previously been recognised as strongly associated with sex work stigma (Kerrigan et al. 2021;Mujugira et al. 2021), but have not previously been theorised as themselves being manifestations of external stigma.Our relationship control items were developed with sex workers and on face value reflect controlling behaviour which is strongly rooted in enacted stigmatisation of sex work.Furthermore, our finding of the very large, standardised coefficients for paths from external stigma to intimate partner violence and to relationship control, strongly suggests that these forms of violence reflect the same underlying phenomenon as external stigma.Future research, particularly future stigma scale development, should take into account sex workers' intimate partners and the highly abusive nature of these relationships.Items assessing partner violence and sex work-related controlling behaviour should be included as part of stigma scales.
Many sex workers feel very positive about their work and appreciate the independence it gives them.They appreciate their ability to provide for their household as well as feeling affirmed as women, valued and sexually desirable.This runs strongly counter to the often cited view of sex work which is extremely negative.It provides an interesting perspective to help understand why women stay in sex work despite the dangers.We have shown that sex workers who are more positive about their work are much less likely to experience shame and blame.One path from childhood trauma is mediated by such positive feelings, indicating that the capacity of sex work to make women feel affirmed is rooted in their childhood experiences of abuse and neglect and this provides the context for sex work offering an affirmation that would be otherwise unavailable to some women.However, the finding that 84.5% of women would leave sex work immediately if they could, provides a reminder that even though some aspects of sex work are affirming for some women, and a minority would continue by choice, the overwhelming majority of women engaging in sex work do not have the option to leave.
Kerrigan and colleagues (Kerrigan et al. 2021), in their theoretical framing of the stigma of sex work, highlighted the observation from Foucault's work on stigma including the possibility of 'resistance'.They summarise this as a process through which individuals create new subjectivities, which allows them to some extent to free themselves from stigmatising identities (Lemke 2002).This can be seen as having parallels in the notion of positive views on sex work, to the extent that through endorsing these ideas sex workers are enabled to seize the narrative and reframe sex work as positively contributing to their lives and sense of self, rather than viewing it through the deficit lens of a lack of agency and stigma.

Limitations
We did not have a nationally representative sample of female sex workers, however, in the context of South Africa where sex work has been illegal and is highly stigmatised, sex workers are hard to access through conventional survey approaches.We believe our community-centric approach with modified respondent driven sampling recruitment represents an innovative compromise.This study is, to our knowledge, the only attempt at a national study of this population conducted in Southern Africa, although we acknowledge that we are only able to access and sample those accessible through sex work programmes.We focused on cisgender women in sex work, because of the focus of our funder.Transgender women, male and non-binary sex workers also face considerable stigma, and other challenges, and should be included in future research.This was a cross-sectional study and so we cannot state the temporality of most associations.We have measures of recent prevalent mental health problems and reports of violence and controlling behaviour but understand that many will have been longstanding.Further, the experience of external stigma was framed in the last year, but likely pertained before this.It is a limitation of this paper that we were not able to test the positive views on sex work items as a possible further dimension of the external stigma measure as has previously been done (Kerrigan et al. 2021), due to the response categories differing.Strengths of our work come from the logic underlying the associations and paths that we identified, and the fact that most were as hypothesised from what is known from research and practice-based knowledge from working with sex worker populations.

Conclusions
Conducted at a time when the buying and selling of sex were illegal in South Africa, this study represents a first attempt to understand quantitatively the experience of internalised stigma (shame and blame) in the context of external stigma and experiences of partner and non-partner violence of women in sex workers.Our findings provide quantitative support for the ubiquitous assertions that internalised and external stigma are major problems for sex workers, and closely related.We have also shown the powerful mitigating effects shown when women felt positive about the impact of sex work in their lives.This resistance to internalised and external stigma provides opportunities to shift the narrative and provides opportunities for intervention at an individual and sex work-community level.More importantly, our analysis shows that the mental health and well-being of women in sex works could be greatly enhanced by measures to protect them from externally stigmatising behaviour, including enabling more to work legally indoors, as well as enabling protection from partner violence.This requires decriminalising sex work which is foundational for enabling safer working conditions and better health outcomes for sex workers (Mccann, Crawford, and Hallett 2021), and in this respect we welcome the steps being taken in South Africa by the Government towards this.Providing adequately funded and more extensive health and social care through sex work programmes, and sensitised public health services, to meet mental health needs would also make an important contribution, as would programmes which target their paying and non-paying partners.

Figure 1 .
Figure 1.Structural equation model showing direct and mediated relationships to internalised stigma amongst female sex workers in South Africa.

Table 1 .
Key measures.Five measures asked about verbal, physical and sexual violence and stigma from health care workers and police with a 4-point response scale (never, once, a few times, often).A typical question was: how often within the past year have you experienced sexual abuse because you are a sex worker?For the SEM onlyItems were summed such that a high score indicated more stigma.Three items asking about: being physically forced to have sex when she did not want to; agreeing to sex when she did not want to because she was threatened or afraid of what he might do; or otherwise being forced to have sex.For sexual IPV these were summed and used as a score.For non-partner rape the variable was presented as a binary measure.Six items designed with the community for use among FSWs.Items: my partner worries when I dress up that I am going to sell sex; I have to hide my SW from my partner to feel safe; my partner gets angry if he thinks I am selling sex; my partner would beat me if he knew I was selling sex; my partner may take the cash I earn selling sex; my partner may rape me to punish me for selling sex.Items summed.

Table 2 .
Characteristics of the sample and associations with internalised stigma.Adjusted for whether the partner knows she sells sex.validity of the latent variables were good (AVEs: Internalised stigma 0.531, Mental health 0.626, intimate partner violence 0.696, Positive views on sex work 0.544 and External/enacted stigma 0.552).We present the direct, indirect and total standardised effects for each path.This study received approval from the University of the Witwatersrand Human Research Ethics Committee (Medical) (Ref number: 180809).

Table 3 .
Prevalence of positive assessments of the being a sex worker and of stigma.

Table 4 .
Female sex worker internalised stigma vulnerability model: direct effects, indirect effects and total effects.