Abstract
Sex work or prostitution has existed in all civilised countries from earliest times and seems always to have been subject to regulation by law and custom. Although there is no consensus on how the phenomenon of sex work should be regarded, historically one dominant explanatory set of ideas is that it represents problems in psychosexual development of the sex worker, arising from psychopathology set in train by early adverse experiences [1,4]. In this vein, a number of writers have suggested that sex workers may have been exposed to sexual abuse in childhood or later in their lives and that these experiences may have formulated some self-concepts which permitted their later entry into sex work [5,9]. One way in which this could occur is by use of dissociative mechanisms; the hypothesis is that the young girl learns to detach emotionally from her abusive situation and later uses the same mechanisms to allow her body to be available to clients during sex work [10,12]. Other writers suggest that the link is through early home leaving, including running away from home [13,14]. A number of these reports are severely limited by studying only street workers and by failing to provide comparison data, limiting the ability to interpret the meaning of results. Very few published reports provide data from sex workers themselves and those which do often use non-representative samples of sex workers, usually females, recruited through psychiatric clinics [15]. Most authors report only data which focus on the subjects' presumed psychopathology and downplay the role of social and economic factors motivating women to enter the sex industry.
It is now possible to study the early environment of adults retrospectively with some confidence; instruments such as the Parental Bonding Instrument have reasonable psychometric characteristics [16,17]. In this article, we use the term sex work rather than prostitution, as it is a more neutral term and it clearly is the term preferred by sex workers themselves.
We report here a preliminary survey of New Zealand female sex workers. The study was designed to overcome some of the methodological problems that have complicated this kind of research [18,20], such as: the definition of sex worker; strategies for recruiting subjects; choice of an age-matched comparison group; small samples; anecdotal evidence; and the ethical aspects of such research.
This project was seen as hypothesis-generating, aiming to generate some testable ideas about the mental health status of sex workers. We took the opportunity to use a previously collected data set of a random community sample of women as a control group.
Method
Subjects were recruited in two urban centres, Dunedin and Wellington, through the New Zealand Prostitutes Collective (NZPC). The NZPC is a support and industry lobby group set up in 1987 by a group of sex workers who considered that they needed to improve their working conditions. There is no official list maintained of those involved in sex work, although the Massage Parlours Act (1978) requires a parlour owner to keep a register of all massage workers, including their full name, date of birth and residential address (at all times). The NZPC is funded by the Ministry of Health to act as educators with the specific goal of limiting the spread of HIV/AIDS in New Zealand. A snowball sampling approach was adopted, with the first subjects being identified by the NZPC, who when interviewed were asked to suggest others whom they thought would be interested in participating. Maintaining subjects' anonymity was a prime concern at all stages of the project. For this reason, the NZPC was not prepared to support total recruitment of the sex worker population in either centre; they considered that this might allow identification of individual workers by statutory authorities.
Ethical permission was granted by the Otago Ethics Committee, who accepted our request that verbal rather than written consent be obtained and that we would not ask the women to reveal their real names but to use their ‘working’ names during the interview, if they preferred.
The women were interviewed by KP at the place of their choice. The interview contained five sections covering sociodemographic information, family background, work history, health attitudes and behaviours and abuse experiences. The women were asked to complete the Parental Bonding Instrument (PBI) for each parent [16] to measure the quality of their relationships with each parent. The socioeconomic status (SES) of each sex worker's family of origin was determined by assessing the Elley-Irving ranking for the main family wage earner, divided into high (1 and 2), medium (3 and 4) and low (5 and 6) SES levels on education and income [21]. Childhood sexual abuse was graded by levels of intrusiveness, non-genital, genital touching and intercourse, either attempted or completed. Physical abuse included being hit, punched, kicked or being hit with a weapon or object, hard enough to leave soft tissue injury.
The results for the sex worker subjects were compared with community data from the Otago Women's Child Sexual Abuse survey (OWCSA), using information from women aged 18–40 only, the age range of the sex worker sample. The OWCSA study is a random community survey of New Zealand women, selected from the electoral rolls and studied by postal questionnaire and subsequent interview. This project aimed to assess the prevalence and consequences of childhood sexual abuse on later psychological function and used similar structured instruments as the present sex workers study [22]. Physical abuse was defined in the same way as in the OWCSA, categorised into two groups: that occurring in childhood (before the age of 16), and that later (16 years or older).
The subjects' PBI scores were analysed in two different ways. First, using Student's t-test, the mean scores for the care and control scales for both parents were compared with the equivalent mean scores from the comparison group. The comparison figures used were taken from the interview sample data, weighted back to the original postal questionnaire sample. After comparison of the means, the community sample median scores were used to divide the two scales into ‘high’ and ‘low’ groupings. Second, the proportions of the two samples in each half and in the low-care/high-control quadrant were compared. This affectionless control by parents is a risk factor for later depression [23].
Statistics
The statistical analyses used were simple bivariate analyses, as befits an introductory investigation. No correction for multiple testing was included. Student's t-tests and Chi-squared analyses for significance were undertaken.
Power of the study
The subject numbers (29 sex workers and 680 controls) generate enough statistical power to detect a 25% difference in the group of interest when the base rate of the factor under consideration is 50% (power = 80% and alpha = 0.05). Where the base rate is larger or smaller than 50%, a 20% difference will be detected.
Results
The sample
Twenty-nine sex workers were interviewed, 15 in Dunedin and 14 in Wellington. Their ages ranged from 16–47 years with a mean age of 28.7 years (SD = 7.6). The 680 women from the OWCSA had an age range of 18–40 years, with a mean age of 29.1 (SD = 6.2). Two-thirds of the sex workers reported that they were currently in a ‘close or intimate’ relationship (n = 19). Ten of these women were living with their partners, four of whom were married.
Half of the sex workers (n = 15) were mothers.
Education
One-third of the sex workers reported that they had no formal qualifications and a quarter (n = 7) had secondary school level qualifications only, figures not statistically different from the control group. A greater proportion of sex workers had started tertiary education but not completed the qualification (7//29, 24%) than was reported by the OWCSA women (58/680, 9%; X 2 = 6.3, df = 1, p < 0.05).
A current socioeconomic status ranking is inappropriate, as this area of work is not listed in the commonly used schedules.
Family of origin
Ten subjects (34%) came from families classified in the high SES category (1 and 2, highly skilled work), not significantly different from the comparison group (22.0%). A smaller proportion of sex workers (8//28, 29%) came from SES 3 and 4 and a greater proportion from SES 5 and 6 (10/28, 36%) than in the control group (393/651, 60% SES 3 and 4, and 115/651, 18% for SES 5 and 6, respectively, X 2 = 11.60, df = 2, p = 0.003). Overall, the socioeconomic background of the sex workers' families did not differ from that of the comparison women.
The parents of 11 sex workers had separated (38%), a higher rate of parental separation than for the OWCSA group (8.5%) (X 2 = 11.8, df=1, p < 0.001). As a result, the sex workers were more likely to report having lived with only one parent for 6 months or more; 13//29 (45%) lived with their mother alone (cf. 93/651, 14%, of the OWCSA sample: X 2 = 18.9 df = 1, p < 0.001) and five (17%) with their fathers (cf. 20/651, 3%; X 2 = 12.8 df = 1, p < 0.001).
Sex workers were more likely to report that their mother, not their father, was the main wage earner (8//29, 28%) than were the control women (35/651, 5%, X 2 = 20.8, df = 1, p < 0.001). We did not seek information on whether other family members undertook sex work.
Age of leaving home
More sex workers left home early than the control women. One-fifth (6//29, 21%) left before their 15th birthday, and over half left by the time they were 17 (16/29, 55%). The comparable control figures were: 2% (16/672; X 2 = 30.7, df=1 p < 0.001) and 26% (174/672; X 2 = 13.4, df = 1, p < 0.001).
Pregnancy
A higher proportion of sex workers than control women had been pregnant at some time (23//29, 79%, vs 401/680, 59%; X 2 = 4.1, df = 1, p < 0.05), despite the age similarities of the two samples. There was no difference, however, in the proportion of each sample who had given birth at some stage (sex worker 15/29, 52%, OWCSA 359/680, 53%).
Sex workers were much more likely to report first being pregnant before the age of 19. Of the 23 sex workers who had been pregnant at some time, 16 (70%) became pregnant before 19 compared with only 64 of the 410 ever pregnant OWCSA women were (16%, X 2 = 20.7, df = 1 p < 0.00001). We did not study the outcome of these pregnancies.
Parental Bonding Instrument scores
Care scores
All but two sex workers (n = 27) completed a PBI questionnaire for their mother and three-quarters for their father (n = 22). The main reason for not completing a PBI on a parent was the absence of that parent from the home due to parental separation or death. One sex worker chose not to complete this questionnaire on her father, stating that her dislike of him was too intense.
Sex workers scored their mothers as significantly less caring (mean care score = 18.6, SD = 9.0, range = 2–32) than did the OWCSA women (mean = 27.2, SD = 7.7, range = 2–36, t = 5.47, df = 277, p < 0.001). The authors of the PBI have suggested a cut-off of 27 for high versus low matemal care [16]. Twice as many sex workers (79% of sample) as OWCSA women (39% of sample) reported their mothers as showing low care (t = 12.9, df = 1, p < 0.001).
Similarly, sex workers reported their fathers as less caring than those of the OWCSA control group. The mean sex worker paternal care score was 19.6 (SD = 9.8, range = 3–36) compared to a mean OWCSA score of 24.9 (SD = 8.0, range 0–36, t = 2.82, df = 272, p<0.01). Two-thirds (64%) of the sex workers scored less than 24, compared to 40% for the OWCSA group; this was not a significant difference.
The mean matemal and paternal care scores from the sex workers' PBI scales did not differ. By contrast, a significant difference was found for the OWCSA groups; mothers were reported as showing more care than fathers (t = 3.3, df = 502, p < 0//001).
Control scores
Control scores did not differ statistically between the sex worker and the OWCSA samples for either parent (maternal control: sex worker 16.5, SD = 9.1; OWCSA 13.6, SD = 5.1; paternal control: sex worker 15.4, SD = 9.3; OWCSA 11.2, SD = 7.0). There was no significant difference between the means for sex worker maternal and paternal control scores. This was in contrast to the scores from the comparison group, in which mothers were reported to have displayed a greater degree of control than fathers (t = 4.46, df = 458, p < 0.001).
Parental Bonding Instrument low care/high control categories
A greater proportion of sex workers described their mothers as in the low care/high control quadrant (15//27, 56%) than did the OWCSA sample (25%, X 2 = 9.9, df = 1, p < 0.01). The same was found for fathers, with 12 of the 22 sex workers describing their fathers as low care/high control in contrast to 24% of the OWCSA participants (X 2 = 8.4, df = 1, p < 0.01).
Childhood physical abuse
One sex worker reported that she had experienced ritual satanic abuse which involved both sexual and physical abuse. It was difficult to code the details of this experience and her data are omitted from the following analyses (n = 28). The data were collected in a somewhat different manner to the OWCSA study and direct comparisons are not available for physical abuse.
Fourteen sex workers reported being hit, kicked or punched, eight being hit with a stick or weapon and one reported that her father tried to drown her when she was 15. This provided a total of over 80% reporting at least one episode of physical violence directed against them before the age of 16.
A relative (including step-parents) had abused all but one of the 14 women. Six subjects had been hit by their mother, five by their father, two by stepfathers. There was a complex pattern of cumulative abuse, with over half of those reporting abuse being hit by more than one person and the same proportion reporting being hit more than monthly.
Childhood sexual abuse
One-third (10//28, 36%) of the sex workers reported that they had experienced genital childhood sexual abuse (fondling, without intercourse being attempted) before the age of 16. That figure is not statistically greater than the 19% of the OWCSA participants (X 2 = 3.14, df=1, p = 0.08). However, sex workers reported greater frequency of penetrative childhood sexual abuse than did the OWCSA women. One-third of the sex workers (9/28, 32%) had experienced attempted sexual intercourse before they were 16 and in seven of these nine cases the participants reported that complete penetration had occurred (7/28, 25%). Both of these figures were greater than the proportion of the random community sample of OWCSA women, of whom 9.4% had experienced attempted sexual intercourse before the age of 16 (X 2 = 8.01, df=1, p = 0.001) and 4.7% completed sexual intercourse (X 2 = 9.38, df=1, p < 0.01). Perpetrators included both non-relatives known to the girl's family (n = 10) and relatives (n = 5, father 1, brother 1, stepfather 1, grandfather 1 and uncle 1).
Discussion
Some clear findings emerge from this study, despite the small number of sex worker participants. Sex workers described their families as having more interpersonal difficulties on a number of parameters, which are known to be associated with later psychosocial problems. The sex workers' families had more parental separation with the mothers being more frequently the main wage earner. Although a relatively high rate of parental separation and divorce was found among the sex worker families, it is worth noting that almost half grew up with both parents, in a conventional nuclear family situation.
There was no overall difference in socioeconomic status between the two groups, despite women being paid less. It is interesting to speculate whether a financially independent mother was an early role model for later financial autonomy for some sex workers. This issue of adult motivation for sex work is dealt with more fully in the companion paper [Romans S, Potter K, Martin J: in preparation].
These women described their relationships with both parents as less caring and both parents as more likely to fall into the low care and high control category. Sex workers were more likely to have left home early, to have become pregnant before the age of 19 years and to have failed to complete tertiary study. They were significantly more likely than the control women to have experienced penetrative childhood sexual abuse. As the sex workers and the OWCSA subjects were asked different questions about their exposure to childhood physical abuse, it was difficult to compare the two groups on this variable. The sex workers in this study had experienced high rates of physical abuse.
Some comments about the research methods adopted are in order. A rare feature in this research area is the use of a large and randomly selected control group, with an age profile equivalent to the sex worker group. The sex workers, however, were not selected in a way which can ensure the representativeness of all New Zealand sex workers and we cannot be confident that we overcame this problem with previous research. The snowball approach enlists workers who are prepared to participate in research and such women are probably likely to be those more concerned with the health and welfare of their group. The initial subjects are likely to be well informed about health and safety issues as they were involved with the Prostitutes' Collective, a group dedicated to improving the working conditions of sex workers. Efforts were made to include the different types of sex work, massage parlours, street work, etc. It is difficult to design a method by which a truly representative sample could be recruited; a complete list from which a random sample could be chosen would be necessary. Such a list is not maintained in New Zealand, and given that sex work is an economic activity which people do intermittently and often part time, such a list would quickly become out of date. We anticipated these problems but mindful of the data-meagre, speculative nature of many of the claims previously made in the literature, we considered that this study was worth undertaking. The other methodological flaw is the small sample size: 29 sex workers in total. Some type II errors may be present in the results, particularly in PBI control figures and the total childhood sexual abuse prevalence comparisons. However, the power calculations suggest that significant differences (25% in sex worker group vs 50% in the control sample) would be found.
Bearing in mind these design limitations, we consider that this study established that these sex workers come from family backgrounds characterised by a number of significant psychosocial problems. They were more likely to have been exposed to a high level of personal abuse stress as young girls. They were more likely than control women to have travelled along a pathway of leaving home early, early school leaving, with subsequent early pregnancy, lower qualifications and reduced work opportunities. We have previously discussed the role of early pregnancy in the control data set [24]. Pregnancies before age under 19 were more likely to occur in those whose families had a non-nuclear structure, in which the parents rowed, in which physical punishment was used after the age of 12, and to girls who lacked a confidante and who were exposed to intrusive childhood sexual abuse. Several of these variables appear salient in the sex workers' early developmental experiences. The higher rate of early pregnancy was not reflected in greater rates of motherhood, suggesting more loss of pregnancies.
A cross-sectional design cannot determine causal connections but can hint at possible linkages. Hypotheses about low socioeconomic families, in which the mother as main wage earner struggles against feminised poverty, with her daughter's early pregnancy, truncated education and subsequent employment opportunities can be easily be generated but will need to be scientifically evaluated. Family factors tend to be highly correlated and it will be a challenge to disentangle pathways. To do so with confidence, future researchers will need to resolve the sampling problem in a satisfactory manner over time.
It is likely that factors motivating a woman's decision to enter the sex industry are as complex as those which contribute to a woman's decision to become involved in any other type of paid employment or occupation. They will include distal and proximal factors and will interact in a complex manner over time. The companion paper [Romans S, Potter K, Martin J: in preparation] will focus on proximal adult variables. Presented here are some suggestive finding which may be important during childhood and early adolescence. They need to be understood in the context of the later educational and occupational choices confronting women as they sort out work patterns.
Footnotes
Acknowledgements
While the opinions expressed here are those of the authors alone, we would like to record our thanks to the staff at the New Zealand Prostitutes Collective for help with the conduct of the study and comments on earlier drafts of these papers. Kathleen Potter was in receipt of a junior research fellowship from the Health Research Council of New Zealand in 1994, during the time she undertook the fieldwork reported here.