No Data on Prostitution? Just Make It Up!

The way people make up data about sex work is not new. There are people who want to exaggerate the number of sex workers whilst treating us as a scourge, to shock people with the scale of the supposed problem, while others want to present the need for sex workers’ rights as a pressing concern due to a large number of individuals being impacted.

How many people are there who sell sex in a given country or city? To answer that question we first need to create parameters for who we’re counting, then we need a way of measuring that population. Given that sex workers are often criminalized and stigmatized, in most places we are actively and intentionally avoiding detection. If a population does not want to be counted, they are a lot harder to measure.

Without having concrete numbers, politicians and activists alike can have a hard time getting people on board with their proposed policy changes. To argue that something is worth the time and effort of changing laws, it is generally expected that the person proposing the change can demonstrate the size of the impact. When it comes to groups that are highly stigmatized, sometimes that expectation can be waived due to the public support for criminalization of some sort making it a popular move for a government to make, but to make the topic seem like a big enough problem for the public to work themselves up into a frenzy you still need some sort of number to give them.

We’ve seen this issue of making up numbers over and over. Parent-Duchatelet complains about it in his book “De la prostitution dans la ville de Paris: considérée sous le rapport de l’hygiène publique, de la morale et de l’administration” – he spent 8 years researching and writing it, finally published in 1836 after he died, and analyzed data on sex work in Paris going back to the late 1700s. He complained openly about police commissioners telling him that there were 60,000 or 50,000 prostitutes in London, or 30,000 in Paris, with absolutely no source for those claims. Then, by assessing the records of registered prostitutes in Paris, he gives his own figures for each year up to 1834 regarding the number of prostitutes. He offers these numbers as a counter to the estimates of 30,000 or so and claims there were no more than just over 4,000 in the 1830s. Of course, he acknowledge that not all prostitutes will be registered, but then uses these figures as if unregistered prostitutes are a negligible number.

People in the modern day make these same errors, though they have differently flawed ways of measuring the populations. Prostitution is currently legal in the UK, with both the buying and selling of sex being permitted but brothels and various forms of solicitation or travel or management of sex workers being illegal. Unlike Parent-Duchatelet, the Office for National Statistics couldn’t use any form of registry of prostitutes, so they relied on commissioning a report – “Calculating the Number of Sex Workers and Contribution to Non-Observed Economy in the UK“. This report estimated a total of 72,816 sex workers based on the number of us who access support services. They took the average number of clients per sex worker support service, from those they contacted, and multiplied this by the total number of services. This estimate is deeply flawed and doesn’t get us anywhere near close to an accurate number. Only a fraction of active sex workers will use support services.

These figures are absurd. in 1834 Parent-Duchatelet thought he was reasonable with suggesting there were only a few thousand prostitutes in Paris in the 1800s; now some sex worker organisations in the modern day seem willing to accept figures like 72,000 across the UK as being accurate. I’d posit to you than both then and now those figures were significant underestimates. I won’t lie to you by making up my own figures. I’ll merely point out that since the data collection methods being used only account for a fraction of sex worker, like only those who access support services, the real figure must be much higher.

Estimates themselves are not always bad. In our current situation where sex workers are not a population that can easily be counted, if we want to have any chance of working out what amount of resources sex workers need then we should have some sort of idea of the population size even if the method is flawed. The way to get a more accurate estimate would be to take a representative sample of the population and ask them to anonymously submit information about whether they had ever sold sex among a series of other questions. Some people would not be forthcoming, but this would at least give us a closer estimate than only looking at those who identify themselves as sex workers publicly or to support services.

From the report that the University of Bristol published, “Nature of prostitution and sex work in England and Wales”, which was commissioned by the Home Office and is offered on the gov.uk website, we see many of these flawed calculation methods.

Something I must praise about this research by the University of Bristol is that it openly admits these figures are serious under-estimates and the researchers seem to understand the reasons why. Under the above tables, they include the note:

“There are some caveats that apply to all the sources reviewed. The generation of prevalence estimates depends in the first place on the identification of the population of interest. In this study, sex work and prostitution have been defined using a broad approach that includes all sexual services. None of the studies reviewed contain data on all of these groups or follows the methodology recommended by the World Health Organisation (WHO)/ United Nations Programme on HIV/AIDS (UNAIDS). Existing data is likely severely to under-estimate figures for groups who are less likely to be in contact with prostitution and sex work support services, such as escorts, victims of trafficking and exploitation, migrant and or highly mobile populations. Additionally, our own qualitative research suggests that certain groups such as those involved in sugaring or BDSM, as well as people who engage in opportunistic sex for payment, do not necessarily view themselves as part of the sex industry: thus they are likely to be excluded from any counts.”

Unfortunately, this nuance is lost when people reference this exact work, or specifically cite figures from any of the sources which are criticized in their report. People do not want to admit they don’t know the totals, want to have a figure, and often providing a source at all is enough to make people assume a claim is true and legitimate even if the content of that source admits the figure is guesswork.

It is not just assessing the total number of sex workers which causes this problem, but also understanding the demographic make-up of sex workers as a group. We don’t have good information about the gender breakdown of sex worker populations, the rate of sexual assault, or the average age at which people began selling sex.

The small populations of sex workers we can find and study won’t necessarily be representative. Someone could assess street sex workers in a specific area and conclude that almost all sex workers are cis women, but just the same they could assess people engaging in transactional sex within men’s prisons and conclude that almost all sex workers are cis men. Where researchers tend to look will be influenced by their biases, so places like prisons are often ignored and informal settings for selling sex like Grindr or some sugar daddy dating websites will equally be overlooked.

Even when we get statistics from government institutions, who people assume will be good sources, a lot of their data comes from arrests or welfare checks or brothel raids or support services. The subset of sex workers who police are more likely to come across and collect data on, or who are going to access support services most frequently, will be those who are vulnerable and that results in a skew to the data that is collected.

One way we can get valuable data on increases and decreases in risk for sex workers is to look at one method of information collection and compare it to information collected in the same way at a later date. For example, the 2020 Médecins du Monde report on the April 2016 law change in France which criminalized the buying of sex was able to do this very well. It offers us the information that from April to September the year before the law change, 41 sex workers reported suffering violent assaults. From April to September the next year (shortly after the law passed), 71 were reported. We should be careful not to assume these numbers are the total violent assaults, as they most likely make up a very small fraction because sex workers notoriously do not trust the police or report crimes against themselves, but what we can do is compare these numbers to each other to get an idea of the level of increase.

However, if changes happen to the reporting system being used then we have an issue with comparability. If the reports are being sent to a third party rather than the police, how well-know that third party is will change the number of reports. If a service that accepts reports from sex workers becomes twice as well-known in the space of a year and reports double, did the number of instances of violence actually go up or is that a result of people being more aware that they can report? If we’re looking at police reports and they go down the same year the police increase their deportations of migrant sex workers, did they actually go down or are migrant sex workers avoiding making reports because they’re scared of being removed from the country?

I’m not going to suggest that we shouldn’t use any of the estimates that are made about the number of sex workers or our demographics, when campaigning for sex workers’ rights. There are plenty of situations where government officials will want a figure or where members of the public want evidence that a certain legal approach lowers or increases violence before they decide if they’re for or against it. What is vital is that we openly acknowledge the limitations of these statistics, rather than parroting them as if they’re objective facts.

The next time you see someone saying there are 100,000 sex workers in the UK and think that number is shockingly large, remember that is a huge underestimate by the Home Office’s own admission. When you hear a number for the amount of violent assaults against sex workers in the last year or month, remember that sex workers are a population that is very unlikely to report abuse and that those numbers can only be compared to see increases or decreases rather than to give us a representative total.

We don’t have good data about sex work and continuing to make it up until we do is not a useful way to encourage better research methods.