What Counts As 'Evidence'?
The review focused on research published in English, mainly from 2000 onwards, and relevant to the criminalisation of the purchase of sex. Most of the empirical evidence has been drawn from countries which have introduced this legislation, with consideration given to scoping studies carried out in countries where the introduction of legislation to criminalise the purchase of sex was being explored. Quantitative and qualitative studies were included and given the narrow focus of the review, studies of various sizes and scope were included.
Attempts to maintain a clear focus for the review meant that our attention was focused on evidence relating to the criminalisation of the purchase of sex. It has been difficult to distinguish the boundaries of evidence in this context in relation to demand for prostitution and the wider issue surrounding the trafficking of people for the purpose of sexual exploitation. Ongoing debates are evident around the relationship between prostitution and human trafficking with claims that prostitution propagates trafficking via supply and demand markets. For example, Hughes (2004: 1) claims that trafficking and prostitution should be analysed as parts of 'an interlocking system'. Although outwith the remit of this review, there are claims that legal or decriminalised models of prostitution increase the market for commercial sex and that increasing demand also increases rates of human trafficking (Di Nicola et al, 2005; Cho et al, 2012  ) but as the researchers themselves acknowledge, this evidence is tentative and difficult to extrapolate from other factors. Proponents of legislation to criminalise the purchase of sex point to the 2000 Trafficking Protocol which calls upon Member States to adopt or strengthen measures that would 'discourage the demand' that fosters sexual exploitation, arguing that such legislation is a direct attempt to address the demand for prostitution and thereby sexual exploitation. The Protocol does not specifically refer to prostitution in this way however, and those who oppose criminalisation of the purchase of sex draw attention to the meaning of 'sexual exploitation' (questioning whether prostitution equals sexual exploitation per se). Where trafficking in humans for the purpose of sexual exploitation is referred to in the studies we have examined by way of contextualising legislation, we have referred to this; however the broader literature on human trafficking has been excluded.
Although much has been written on this topic, the number of empirical studies in this area is relatively small and gaps in knowledge remain (Skilbrei and Homstrom, 2013) thus it is important to acknowledge the limitations of the methodological approaches of the research reviewed. An overview of the research design, sampling and limitations of the empirical research reviewed is available in the Annex. There are a limited number of studies informed by empirical data collection and analysis specifically on the impacts of the criminalisation of the purchase of sex. This review is also limited to papers in English and therefore this puts a limitation on empirical evidence which can be drawn upon. There is likely to be a considerable amount of material we have been unable to access (from Sweden and Norway in particular).
Empirical evidence is largely focused on street-based prostitution and there is a universally acknowledged uncertainty about the numbers involved in indoor prostitution such as online and indoor sex work and the impacts of the criminalisation of purchase on these segments of prostitution. Wherever possible, we have included the evidence that has been produced in this area. There is an acknowledged need to take account of the shifting context of prostitution in relation to universal trends towards online and mobile phone technology. Much of the research available has been commissioned and/or undertaken by government bodies or by affiliated research with specific organisations. Different approaches to conducting research have also been taken by different researchers in order to highlight particular aspects of the debates (e.g. O'Neill et al, 2015).
Previous attempts within the UK to evaluate evidence in relation to demand have highlighted a number of challenges arising from the available data. For example, a review led by the Home Office, Tackling Demand for Prostitution (Home Office, 2008) aimed to assess what the government and other agencies could do to reduce the demand for prostitution in England and Wales. As part of this review, Wilcox et al. (2009) examined published research which focused on clients (predominantly men) who purchase sex. The Home Office review considered 181 studies (which met the criteria from a total of 220) from selected countries  . The review focused on the characteristics and motivations of those who procure sex, the contexts in which they procure sex, and 'what works' in tackling the demand for prostitution. The study highlighted the methodological difficulties of conducting research in this area. The authors note:
There are many gaps in the research and much of the evidence is weak or inconclusive, particularly with regard to 'what works' in reducing demand. It was also noted that prostitution is a policy domain for which the 'right' answer may not be determined solely by reference to the evidence. There are moral, political and other influences that need to be considered when tackling the demand for prostitution (Wilcox et al, 2009: Key findings).
Wilcox et al. (2009) highlighted the challenges of obtaining reliable and accurate estimates of the number of people who buy sex, due to the hidden and stigmatised nature of prostitution with estimates variable due to methodological approaches. Where measures have been implemented with the aim of reducing demand, considerable methodological challenges have been identified in terms of:
- Lack of adequate comparison group;
- Small sample size;
- Lack of representativeness;
- Inadequate 'outcome' measures (i.e. which aim to ascertain the effect of a particular intervention).
Wilcox et al. (2009) also raised several points relating to the literature on studies of demand, notably that given methodological issues, many studies are based on self-selecting or otherwise biased samples; and many studies focus on the more visible street prostitution and on male clients. Similarly, Mossman (2007) noted the challenges in assessing the impact of legislative approaches as part of an international review. She highlights four key areas of methodological difficulty:
- Limited research designed to evaluate the impact of prostitution reforms;
- Variations in legal frameworks within approaches making comparison difficult;
- Validity of reports was often limited due to lack of identifying sources of information, or methodologies used;
- Conflicting results about impacts which were frequently used to support different ideological views.
Kelly et al. (2009: 62) comment on their comparative review of prostitution 'regimes':
What emerged strongly from this review is that most approaches to prostitution lack a coherent philosophical underpinning, from which specific short and longer term aims and objectives could be drawn out and evaluated. As a consequence, much discussion and debate is founded on claims based on belief rather than a strong evidence base, or narrow assessments of selected policy goals  .
Reflecting broader contexts
While evidence on prostitution generally, and demand in particular is partial due to the very nature of this stigmatized and generally less visible activity, there is often a tendency for studies to focus on the individualised nature of the sale and purchase of sex. In many ways, this can decontextualize prostitution from the wider social contexts within which it occurs (see Scoular, 2010; Coy, 2012). Socio-economic and socio-cultural factors are crucial in considering the practices of selling and buying sex, the attitudes and environment in which it takes place (i.e. the acceptability or otherwise of prostitution) and the structural factors which impact on the extent to which prostitution is viewed as a 'choice' or not. Without considering this wider context, a simple evaluation of evidence available omits structural circumstances and the resulting discourses which emerge. For example, increased numbers involved in prostitution reflected in Sweden and Norway in 2008 is noted by Rasmussen et al, (2014) as resulting directly from the onset of the global financial crisis rather than reflective of legislative models.
Similarly, evaluations of particular regimes require some reflection upon the aims and objectives that legislators hoped to achieve via their introduction. Advantages and disadvantages have been associated with different models depending on what the aspirations for these models were at the point of legislation and implementation and ultimately, political decisions determine what policy and/or legal interventions are tasked with prioritising in terms of enforcement and subsequent resourcing. In recognition of this, some researchers have drawn attention to the particular socio-economic context of Sweden and the potential challenges which may arise for attempts to transpose a legislative framework into countries where legal, social-welfare systems and cultural attitudes are different (Gould, 2001; Bucken-Knapp and Schaffer, 2011; Scoular and Carline, 2014).
The principles and values which underpin the development of legislation may be distorted in practice during implementation processes, and the enforcement of legislation is itself part of a political decision-making process determined by broader priorities and contemporary concerns. This can mean that the principles of legislation are diluted in practice while enforcement measures and resources may not meet original aspirations. Similarly, cultural practices can affect legislative regimes which are introduced in a nation state different to that where the originating impetus for a particular model emerged. Importantly, practice and enforcement may not be closely related to the principles upon which specific policy models are based due to a range of factors such as uneven implementation and resourcing, and disparate practice through criminal justice enforcement.
These wider factors add a range of complexities to the assessment of impact and are noted by the Swedish Institute (2010) who, in their translation of selected extracts of the Swedish Government report evaluating the purchase of sexual services, note:
(...) after having read the extensive number of existing reports and studies from authorities and researchers on the subject, we realized that it would not be possible in the framework of this inquiry to produce the precise knowledge about prostitution that politicians and debaters request, but which no authorities or researchers have been able to generate in the nearly eleven years that the ban against the purchase of sexual services has been in place  .
Proposing the establishment of a national centre to support the collation and evaluation of data, the Swedish Institute (2010:12) also highlight the challenges of accessing evidence in relation to follow-up noting that "The knowledge available is difficult to grasp and, in part, difficult to assess, and is shaped by the operational focus and perspective of the agencies and organisations concerned. This makes it impossible to draw entirely reliable assessments and comparisons using the available knowledge". Further research specifically on the longer-term effects of laws on attitudes and the effects of different types of laws in different contexts has been advocated (see also Jakobsson and Kotsadam, 2010; Skilibrei and Holstrom, 2013).
Email: Justice Analytical Services