Dead bodies don’t lie, but statistics do
By Joel Hellewell
I’m going to use the term woman/women in prostitution even if it is a bit clunky, since it is hopefully tolerable to both sides of the debate. I have never sold sex, so I cannot speak to the experiences of women that have. I hope that what comes through is a genuine desire for the proper treatment of the data, with a view to providing good evidence for what will put an end to the violence that women in prostitution face.
Women in prostitution face far higher rates of violence and murder than the general population, these horrifying statistics are (correctly) mobilised to argue for changes in the law that might make women safer. However, in comparing the number of deaths between countries we must be careful, specifically regarding the duration over which the deaths occurred and the number of women in prostitution. An ideal comparison between countries would involve the rate of murder of women in prostitution in terms of a rate such as “women in prostitution murdered per woman in prostitution per year”.
Considered separately, accounting for duration and population makes intuitive sense. If we find that company A sells 150,000 cars over 3 years but company B sells 75,000 cars over 1 year then we would compare the sales of the two companies in “cars sold per year”, i.e. number of cars sold divided by the duration in years. In the same way, we would consider the population of a countries when comparing the number of deaths due to some disease.
Quite often the number of women murdered in prostitution is presented without considering time or population (of women in prostitution) and these numbers are used to make inferences about the effectiveness of different legal frameworks to keep women safe. Consider the map below from the Nordic model Now website:
The implication of this map (and the text surrounding it) is that less women in prostitution are murdered in Sweden compared to the other countries with more liberal prostitution laws. However, as you can see there is no consideration of the number of women in prostitution and only indirect consideration of time. The number of women in prostitution is very difficult to estimate for many reasons, women often work in secret to avoid the police or stigma from the public. If we make a rudimentary attempt to convert the number of murdered women in prostitution into a rate directly comparable between countries, can we draw the same conclusion? Below is a table showing calculations and linking to the sources, mainly wikipedia since this is mostly an exercise is making a point rather than publishing finished research.
| Country | Number of women in prostitution murdered | Number of women in prostitution | Time period | Women in prostitution murdered per woman in prostitution per year |
|---|---|---|---|---|
| Sweden | 1 (ignoring the footnote for now, see below) | Estimated at 1500 in 2003 | 17 years since implementing Nordic model | 0.000039 |
| Germany | 69 | Estimated at 400,000 | 13 years (2002-2015) | 0.000012 |
| Spain | 31 | Estimated at 300,000 | 5 years (2010-2015) | 0.000021 |
| The Netherlands | 127 | Estimated between 15,000-30,000 (used 22,500) | 30 years previous to 2013 | 0.00019 |
When considering the population of women in prostitution and time, the rate of murder of women in prostitution is lowest in Germany. There are 266 times more women in prostitution in Germany than Sweden, whereas the number of murdered women if prostitution is only 63 times higher. Thus, the rate is lower for Germany. But what if we acknowledge that the murder in Sweden was by an ex-lover and is not associated to prostitution? Does this mean that we can declare that Sweden has a rate of 0? Well there are so few women in prostitution actually in Sweden that if it’s rate of murdered women in prostitution was equivalent to Germany’s then we’d expect a woman to be murdered on average every 1/(0.000012*1500) = 55.6 years. Therefore, the job of comparing these two countries is a lot harder than it seems.
But as I’ve mentioned, these are back of an envelope calculations mostly sourced from Wikipedia and don’t take into account mediating factors such as changes in population (it would be informative to see this undertaken properly). My point though is that comparing raw numbers of murdered women in prostitution is not informative. If our goal is to reduce violence against women in prostitution, then to compare the effects of different legal frameworks we need a measure that is comparable between countries. Using a rate such as the rate of murder of women in prostitution per woman in prostitution per year is a good start.