Analysing web-based content

Thus far, we’ve considered how the internet can serve as a valuable avenue through which the criminological researcher can access relevant academic literature as well as a variety of data sources that are made available online. In this section, we move on to consider how we can approach the content of websites as an object of analysis in and of itself; the words, images and recorded sounds that comprise internet content themselves become the raw material that we study in order to learn more about crime-related issues. As with many other walks of life, the internet has rapidly become a central site or space in which public discourse about crime and related issues is produced. Online discussions and interventions concerning crime come from numerous constituencies: politicians and policy-makers; political parties and cam- paigning groups; criminal justice and law-enforcement agencies; journalists and commentators; and ‘ordinary’ members of the public. Consequently, web-based content can be analysed so as to better understand issues such as the following:

· how ‘official’ discourses construct crime problems – the kinds of crime risks we face, the characteristics of offenders and their victims, trends and patterns in crime, and so on

· how popular ‘folk’ criminology explains crime and attributes responsibility to various social, cultural, economic, psychological or other factors

· how responses to crime problems are configured and advocated – the creation of new laws, policing and law enforcement, punishment and rehabilitation.

Analysis of web-based crime discourse can focus in on a wide variety of interventions – everything from the policy pronouncements of government departments to the public expressions of interest, concern or moral outrage evident in postings on social media platforms. A variety of methodologies can be adopted, including quantitative content analysis, such as counting the incidence of various kinds of expression and mention of different issues, or measuring trends in frequency of such expressions over time (Riffe et al., 2014). This kind of analysis has been facilitated by the development of web- analytic tools that enable researchers to scour the internet for particular kinds of content and measure their presence (Herring, 2009; McMillan, 2000). However, while web searches can uncover seemingly vast amounts of content, it is important to recognize that the most commonly used search engines (such as Google) only index a small proportion of the total material hosted online. Information scientists make a distinction between the so-called ‘surface web’ (indexable by search engines)

and the ‘deep web’ (containing content that is in principle publicly accessible, but is not indexed) (Bergman, 2001). Only a small percentage of the former is covered by our usual web searches, while the latter remains invisible. As a consequence, a search for online discussions of a particular topic will not deliver all (or even a majority) of the potentially relevant content, and may also be skewed by the ways that search engines prioritize some results over others according to their ‘popularity’ (Evans, 2007). Such limitations mean that a corpus of web content will inevitably be partial and not necessarily representative of all online discourse on the topic as such. Further issues arise due to the unstable character of content hosted online – the web is a dynamic environment in which content appears and disappears with some rapidity, and undergoes constant updating and revision (Adar et al., 2009).

An alternative approach to amassing web-based content for analysis is through a targeted search (in effect, a form of purposive sampling – Palys, 2008). Through this search-based sampling strategy, you can collect data that encompasses a variety of instances of the particular phenomenon you are interested in. Data collected in this way is particularly suited to qualitative analysis, where concerns about statistical representativeness and generalizability are largely set aside. Online sampling of this kind is not restricted to websites but is also increasingly employed in order to analyse crime-related content appearing on various social media platforms, including social networking sites such as Facebook and micro-blogging platforms like Twitter. Various free online tools for searching and analysing Twitter content are now available, including the likes of Socialbearing (www. socialbearing. com). Criminological analy- sis of content derived from websites and social media is illustrated in Box 18. 1.

Analysis of online content can address not just representations or constructions of crime-related issues, as outlined above, but can also explore forms of expression that are in and of themselves criminal acts. Such offences include, first, so-called ‘content’ crimes in which users post or share material that breaches laws around obscenity, hate- ful speech or incitement to violence against particular social groups (Holt et al., 2015; Yar, 2013). Particularly topical in this regard is the use of the internet to disseminate terrorist propaganda and extremist political provocations that target minorities and vulnerable populations (Conway, 2016). Second, they include forms of harassment or abuse targeting particular individuals in online forums such as Facebook and Twitter

– so-called ‘trolling’, which may breach laws in England and Wales such as the Malicious Communications Act 1988, the Communications Act 2003 and the Protection from Harassment Act 1997 (Bishop, 2013). Third, we can identify criminal content related to online fraud, such as so-called ‘spoofing’ websites that mimic the portals of banks in order to induce victims to unwittingly share their passwords and other access-related information (Dinev, 2006). Insofar as such content is posted on publicly accessible websites and platforms (as opposed to restricted-access forums), it is available for collection, collation and analysis. Box 18. 2 offers an overview of such research related to one of the above-mentioned topics, namely the online circulation of hateful political speech.

(Continued)

online (initially in the form of websites and discussion forums, and latterly also including content on social media platforms and micro-blogging services) has been well documented (Yar, 2013: 99). Initial studies sought to identify and count the num- ber of sites containing such content, typically originating with far-right, ultra-nationalist, white supremacist and neo-Nazi groups and their supporters or sympathisers. However, attaining an accurate measure of their number is difficult, and has largely depended on users reporting their presence to anti-racism organiza- tions, internet service providers and/or the relevant authorities. As a result, such data remains indicative of the range and growth of online hate speech, but cannot provide any definitive measure. More effective have been criminological studies that use purposive sampling to map the different kinds of content appearing on such sites. For example, Gerstenfeld et al. (2003) conducted a content analysis of 157 ‘extrem- ist websites’, including variables such as ‘multimedia content’, ‘content for children’, ‘mention of economic issues’, ‘use of racist symbols’, ‘advocacy of violence’, and so on. In this way, the researchers were able to attain a clearer view of just what forms were taken by online hate speech and the prevalence of different types of content. Similar sites were used by Duffy (2003) as the basis of a qualitative analysis focused on the themes, symbols and stories that were deployed so as to generate a rhetorical effect; she found that these included appeals to ‘fairness and justice’, conspiracies about ‘Zionist’ control of the USA, and the restoration of a supposedly ‘natural order’ that dictates a ‘separation of the races’. Further studies have sought to use hateful online content in an attempt to assess the impact that exposure to such material may have on the beliefs, attitudes and outlook of those exposed to it. Lee and Leets (2002) used a range of psychological measurement techniques to assess the relative persuasiveness of different kinds of hateful conduct in terms of the attitudes reported by 108 adolescents who were exposed to such material. These studies illustrate how internet-based content can be effectively used as the basis of a variety of criminologically oriented research studies.