A rapid evidence assessment gathers, analyses and reports on as much literature as possible within a specified period, generally to inform policy development (Government Social Research Service, nd). The search is systematic and follows agreed methods. Our search adopted a SPIDER design (Cooke et al, 2012), which offers greater relevance to social science research questions than models directed at assessing interventions. This framework was applied to the research question: What is known about the characteristics, vulnerabilities and on-and offline behaviour of victims of online-facilitated child sexual abuse and exploitation? (see Table A.1).

Search methods

Two information specialists conducted a pilot to test the search strategy to ensure that relevant results would be retrieved. Our search strategy was made up of a number of search strings, each of which are key terms that were searched for individually and together to build up to a sensitive and specific overall search (see Tables A.2 and A.3). The pilot results illustrated the power of key words in the right search order to influence outcome. For example, the number of results retrieved from combining the Sample, Phenomena of Interest and Evaluation strings returned 1,588 research reports, but when combined with the ‘Design’ string this reduced to 78 results. The design category was removed at this stage as it was limiting the results far too drastically. After some testing we also added solicitation/blackmail/extortion to the sample as this found new and relevant results.

For some databases the search strings were too long, for example JSTOR and ATM digital. Other databases place a limit on the number of wildcards (a way of truncating a search term so that all the possible variations of that word are searched for, such as child*, sext*) and Boolean operators (AND/OR) that can be used. Such restrictions led to the development of bespoke search strings for these databases. One platform would only allow a single study to be extracted at a time; we therefore limited results to the first 50 searches on the database.

A range of databases was included spanning social science, humanities and technology research (see Table A.4).