Go to top of page

Survey methodology

Survey development

The Sex Worker Migration survey used in this study was an expanded and amended version of the Hong Kong sex worker organisation Zi Teng’s 2006–07 Chinese sex worker survey that aimed to identify the needs of Chinese sex workers. This survey was administered in six other countries but was conducted in Australia by Scarlet Alliance, the Australian Sex Workers Association. The survey was redeveloped by the AIC in collaboration with Scarlet Alliance and its committee of sex workers from Thai, Chinese and Korean backgrounds (the steering committee). The project was approved by the AIC’s Human Research Ethics Committee; research approval was also sought and received from the AIDS Council of New South Wales (ACON).

The steering committee provided essential input into cross-cultural and sex work-specific issues during the development, administration and analysis of the survey. Scarlet Alliance and the steering committee also provided advice on the sensitivity of particular questions, particularly those with the potential to cause discomfort to, and/or discourage respondents from, completing the survey. These questions were marked in the survey as optional.

The survey comprised 51 multiple-choice questions and 11 open-ended questions categorised into four major areas: demographics, migration experience, working conditions and access to services (see Appendix C). The survey was translated into Thai, Korean and Chinese, with at least two of these language groups (Thai and Chinese) significantly represented in migrant sex worker populations according to the literature and advice from sex worker organisations. Translations were undertaken by accredited independent translators, with cross-checking by members of Scarlet Alliance and the steering committee.

All multiple-choice questions had an ‘other’ option that participants could choose if none of the listed options matched their experience. The number of open-ended questions was kept to a minimum to allow for easier analysis, and referred to matters where responses could not be reliably anticipated (eg the reason(s) for migrants not wanting to return to Australia, their reasons for not being satisfied with their income, and the advice they would give to other migrant sex workers).

A draft survey was piloted with two sex workers in the Australian Capital Territory by AIC researchers and Scarlet Alliance staff before the administration of the survey.

Survey administration and collection

The AIC partnered with Scarlet Alliance, which then entered into arrangements with the Sex Industry Network (SIN), ACON (on behalf of SWOP NSW) and the AIDS Action Council (on behalf of SWOP ACT), to administer the survey using their collective networks and outreach capacities and services. The survey used a snowballing and convenience-sampling method, drawing on the knowledge of the sex worker organisations involved in the research to ensure that the survey was administered to a cross-section of workplace types in each location.

Surveys were collected face-to-face between February and November 2010 in:

  • Sydney and Newcastle;
  • Melbourne;
  • Brisbane, Townsville and Toowoomba;
  • Adelaide;
  • Canberra; and
  • Perth and Kalgoorlie.

As the collection of information on migrant sex workers’ workplace experiences, migration experiences and access to services was the overarching aim of this research project, migrant sex workers were targeted specifically for survey collection. This involved focusing on gaining access to workers from the language groups and cultural backgrounds identified in the literature and industry knowledge as constituting the majority of migrant sex workers (ie Thai, Chinese and Korean), and the sectors they work in (off-street premises). However, face-to-face collection took place in a range of workplaces, including brothels, massage parlours, strip clubs, escort agencies, private sex workers’ homes and other workplaces, and also at non-work venues such as sexual health clinics and street-based worker drop-in centres. Where possible, workplaces were contacted in advance of data collection and informed that sex workers involved in a research project on migrant sex workers were seeking to collect surveys from workers in their establishment. Collection was also conducted by sex worker organisations as part of their regular outreach activities, using previously established relationships with workplaces. Australian-born sex workers were included in the survey population as a comparative group.

Additional collection sites included Chinese, Korean and Thai-language sexual health clinics at Melbourne and Sydney Sexual Health Centres, and at a sex worker organisation (Magenta) in Perth. Where relevant, medical staff of these health clinics were briefed on the project and gave potential participants a summary of the project and the option to participate.

All respondents were given a small gift for participating in the survey. See Appendix A for more detail on the survey collection at each site.

To ensure that collection included sex workers who worked in exploitative and/or tightly controlled environments, survey collectors employed the following strategies:

  • conducting multiple visits to workplaces at various times and on various days to engage with different members of management;
  • working with staff at sexual health clinics to approach sex workers to participate in the study, outside the workplace setting;
  • ensuring that sex workers could complete the survey privately without interference or observation by any management or others; and
  • targeting survey collection at workplaces where there was anecdotal evidence of bad or exploitative work conditions, and workplaces that had recently experienced police or immigration visits.

All collectors involved in the administration of the survey were currently working or had at one time worked as a sex worker. Multilingual sex workers whose first language was Thai, Chinese or Korean were funded to participate in the survey collection, while sex workers whose first language was English participated as volunteers. Survey collection training took place in Sydney, Melbourne, Adelaide and Canberra. All survey collectors were trained in two areas—outreach skills and survey administration.

Administration guidelines were developed that set out the sequenced steps for administering the survey and a comprehensive explanation of each survey question and definitions. The administration guidelines also included alternative administration steps to be followed if the respondent was unable to read the language in which the survey was written and the collector was required to assist in completing the survey. Aside from the few instances of assisted surveys (n=11), all surveys were completed in private, without assistance or input from the collector. Completed surveys were sealed in envelopes by the participants and remained sealed until they were received by the AIC for data processing. Respondents were encouraged to answer the open-ended questions in their own language. Responses in languages other than English were later translated by independent accredited translators, and cross-checked by Scarlet Alliance’s steering committee.

All collectors were required to have the administration guidelines with them at collection and to refer to the question explanations when asked by the respondent about definitions or interpretations of survey questions. In addition to the administration guidelines, collectors were required to have a project information pamphlet ready to supply to all survey respondents. This pamphlet was available in English, Korean, Chinese and Thai and included information on the project and contact information for relevant services (as well as contact information in case they wanted more information on the research).

The survey was also made available online (in English, Thai, Korean and Chinese) and distributed exclusively to Scarlet Alliance members to ensure that only sex workers received it. The online collection ran for the last three months of the collection period, from September 2010 until November 2010. The majority of survey responses (98%, n=582) were collected face-to-face, with two percent (n=10) of the 592 responses collected using the online survey.

The risk of the survey being completed by the same person more than once was managed as follows. For the online survey, the following paragraph was added to the consent and confidentiality information:

This survey has also been collected face-to-face by peer collectors at various sites in Adelaide, Melbourne, Sydney, Newcastle, Canberra, Townsville, Perth, Kalgoorlie and Brisbane. If you think you may have already completed the survey then, please do not complete the survey online again. If you have already completed this online survey, please do not complete the survey online again.

For the face-to-face survey collection, the same survey sites were visited by the same collectors, who indicated they would recognise the respondent if they had already completed the survey. However, it is acknowledged that this method is not infallible and ran the risk of respondents being surveyed more than once. In addition, there was also a risk of cross-state and territory double counting if a respondent had moved interstate during the collection period. For future survey administration, using the same text as used in the online consent and confidentiality information in the face-to-face collection would be considered. To ensure the integrity of survey responses, the collectors were not permitted to participate in the survey themselves or respond to questions about their own experiences and work conditions.

The risk of a non-sex worker completing the survey was mitigated by restricting the survey collection sites to sex industry workplaces or to individuals who had disclosed to medical personnel at sexual health centres that they were sex workers. Other non-workplace sites such as drop-in centres for street-based workers were also considered appropriate.

Fifty surveys were excluded from the analysis for not meeting the exclusion criteria developed to ensure surveys were completed to a substantial level. These criteria included two phases.

Phase 1

All cases that did not meet the following requirements were dropped from the dataset:

  • answered at least four demographic questions; and
  • answered at least four workplace condition questions.

And if classified as a migrant:

  • answered at least two workplace satisfaction questions; and
  • answered at least three migration experience questions.

Phase 2

The remaining cases were screened to ensure that questions vital for analysis had been responded to. Cases that did not meet the following requirements were dropped from the dataset:

  • answered at least two of the questions concerning age (Q1), gender (Q2) and birth country (Q3);
  • answered at least one of the questions pertaining to hours worked per day (Q27), days worked per week (Q28), clients seen per week (Q29) and workplace type (Q31); and
  • answered at least one of the questions pertaining to regularity of payment (Q32), proportion of wage received (Q33) and contract conditions (Q35).

The statistical testing employed in the analysis of the survey responses for the purpose of this report is outlined in Appendix C.

Survey limitations

Sampling bias

Migrant and non-migrant respondents were not distributed evenly across the states and territories. Indeed, there were greater proportions of migrants in New South Wales (52%) and South Australia (19%) than in other states and territories (Table 2).

It is important to note that these distributions are not representative of the sex worker population; rather they are likely to be indicative of a sampling bias resulting from the snowball and convenience-sampling method used. For instance, the bias to respondents in New South Wales is largely a result of the main survey collection coordinator, Scarlet Alliance, being based in Sydney. The bias to South Australia for migrant respondents is most likely due to the dedicated survey collection of SIN based in Adelaide and their targeting of migrant workers for survey administration.

Table 2 State and territory collection site by migrant status (%)
State and Territory Migranta Non-migrant Totala,b
NSW 52 40 49
Vic 9 13 10
Qld 5 21 9
WA 10 14 11
SA 19 3 15
NT 0 1 0c
Tas 0 0 0c
ACT 5 8 6
Total (n) 411 151 591

a: Excludes one survey completed by a migrant respondent that had no recorded state or territory

b: Includes respondents with an unclassified migrant status

c: n=1 but rounded to 0 percent

Note: Percentages may not add to 100 due to rounding

Source: AIC, Sex Worker Migration and Vulnerabilities to Trafficking 2010 [computer file]

Survey content

It was determined during the collection phase that several survey questions had been affected by wording or printing errors, which had not or could not have been picked up in piloting and which created problems in interpreting some responses.

For the question on the proportion of wages received from their employer (Q33), it was intended that the respondent would indicate on a line (by drawing a mark) the proportion of wages they received relative to the categories ‘none of it’, ‘half of it’ and ‘all of it’. However, most respondents circled one of the three categories rather than making a mark on the line. This was rectified by training survey collectors to advise respondents on how to properly respond to the question. As this training occurred during the collection phase, it is possible that a number of surveys collected prior to the training were thus affected.

The question on contracts (Q35) asked respondents if they had ever been on a contract; however, the multiple-choice responses include a comparison of their current workplace conditions with their contract terms. Therefore, it would have been difficult for respondents who were previously but not currently on a contract to answer the question.

Respondents were also asked whether they had experienced at work a range of situations that were positive or negative in nature (eg incidents of abuse or the provision of support and services), and whether these situations involved their boss, receptionist, co-workers, sex worker organisations, police, staff from the then Department of Immigration and Citizenship (DIAC; now Department of Immigration and Border Protection), or other government agencies (Q39). However, this question did not have an option for respondents to indicate that they had not experienced the listed situations. Therefore, it was not possible to distinguish between respondents who did not select any experiences because they were not relevant or because they skipped the question.

The survey question on respondents’ help-seeking behaviour (ie from whom would they seek assistance for a range of scenarios—involving criminal, financial, work, health or immigration issues; Q48) was problematic because, similar to Question 39 described above, a list of options was provided including ‘I don’t know’ and ‘I wouldn’t seek help for this’, but not ‘none of those listed’. The options that were provided were relevant mainly to scenarios relating to crime and justice rather than health, work conditions and financial issues. Feedback from the steering committee suggested that the formatting of the question and responses into a table might have discouraged or confused respondents. This same formatting issue was raised for Q39. Further, although the wording of the question implied that only one option should be selected for each scenario, unlike all other questions in the survey there was no direct instruction to select one response only. The subsequent high non-response rate for this question is likely an indication that the issues listed above may have prevented or discouraged some respondents from answering this question (see Appendix B).

Missing responses

Twenty-nine of the 60 survey questions had a non-response rate of 10 percent or more of the entire sample (see Appendix A). This was due to incomplete surveys and respondents skipping questions. This limited the extent to which the responses to some questions could be generalised to the entire sample. It also created potential validity issues in comparing migrant and non-migrant responses, particularly where the non-response rate differed between these two groups for certain questions.

Migrant sex workers in New Zealand

A survey tool nearly identical to the one used in this research project was administered among 124 migrant sex workers in New Zealand in 2012. This survey was conducted by the New Zealand Prostitutes’ Collective and was separate to this project. Given the similarities in the methodology applied in the research projects, findings from the New Zealand survey will be compared with the AIC/Scarlet Alliance study regarding the work and migration situations of both sample groups. A summary of the findings from the New Zealand survey is provided in Appendix D.

Last updated
3 November 2017