Essay
New Media: Promise and Peril
In 2013 designer Zoë Quinn released an interactive online game that inspired both praise and criticism among reviewers. Soon after the game’s release, an ex-boyfriend blogged that Quinn had achieved success by cheating on him with a well-known journalist (Dewey, 2014). Hundreds of commenters took to online forums, some defending Quinn and others critiquing her perceived breach of ethics. Many took critiques a step further by “doxing”1 Quinn and her supporters by posting their personal information on blogs and social media outlets. Others Tweeted rape threats, causing Quinn and several female bloggers, journalists, and game developers to flee their homes (Dewey, 2014).
The incidents described above, labeled “Gamergate” by the media, highlight that the means engaged by perpetrators of crime, the methods of investigation used in law enforcement, and the ways in which we as a society disseminate and digest information about criminal behavior are shifting rapidly in this information age. New ways to effect social harms, such as doxing, swatting,2 phishing,3 cyber-bullying, cyber-stalking,4 trolling, revenge pornography, catfishing,5 sexting, identity theft, and hacktivism, are entering the vernacular at a concerning rate. At the same time, police agencies actively are engaging social media to identify and apprehend those engaged in wrongdoing (Lieberman, Koetzle, & Sakiyama, 2013, p. 438). More than 85% of people in the United States use the Internet, and two-thirds of these engage in social networking sites (SNS), such as Twitter or Facebook (Winkelman, Oomen Early, Walker, Chu, & Yick-Flanagan, 2015, p. 194). Unanticipated social and legal quandaries continually arise and, with these, the opportunity for new and exciting ways to study criminal justice. Although researchers across scientific, social sciences, and humanities fields have begun to explore social media as a new delta of communication, few are taking a hard look at social media as it relates to crime.
This article offers an overview of methodological approaches to studying two of the most deeply entrenched and diverse “new” media sources, Twitter and Facebook. The article first describes these SNS and identifies their particular significance among users.6 The following section discusses the ways in which Twitter and Facebook may be useful to criminal justice researchers, highlighting the different ways in which these SNS may actively be engaged: not only by persons seeking to commit or prevent crime but also by social science researchers. Lastly, the article addresses some of the challenges, both ethical and data driven, of engaging new media to study criminal justice issues and problems. This final section complicates characterizations of Internet sites as a public sphere or open source, engaging discourses about privacy concerns in studying new media and the potential for SNS to replicate or foster harmful norms regarding race, gender, and socioeconomic status. The article offers suggestions for sites beyond Facebook and Twitter of potential interest to researchers in criminology and criminal justice.
What is a “New” Medium?
Engagement with the contemporary social world rarely occurs without the influence or assistance a virtual or electronic intermediary (Altheide, 2013, p. 226). As forms of information delivery change, the ways in which we interact with one another also are altered (Altheide, 2013, p. 228). While the term “new” media arguably includes television, film, games, and other types of communication, this article focuses on “computer-mediated communication (CMC),” “human interactions occurring through the use of devices such as computers, tablets, and smartphones” (Hardacker & McGlashan, 2016, p. 81). More specifically, the article explores research regarding “social” media, “[I]nternet-based applications that … allow the creation and exchange of user-generated content” (Kaplan & Haenlein, 2009, p. 565). Ellison and boyd define SNS 2.0 as “a networked communication platform in which participants (1) have uniquely identifiable profiles that consist of user-supplied content, content provided by other users, and/or system-provided data; (2) can publicly articulate connections that can be viewed and traversed by others; and (3) can consume, produce, and/or interact with streams of user- generated content provided by their connections on the site” (2013, p. 158).
While SNS have existed since the early 2000s, it is in the past few years that use of such sites has soared, with more than 65% of online adults and 75% of online youths across ethnicities using social networking online (Lieberman et al., 2013, p. 442). As Burgess and Green (2009) observe, digital media has a vast “potential to enable active cultural participation” and “to contribute to a more inclusive cultural public sphere.” However, as described herein, with the advent of new mechanisms for participation also come significant challenges for social science researchers (pp. viii, ix).
Two of the most significant platforms for mediated social communication today are Facebook and Twitter. I have chosen to focus primarily on these two sites because, although they share the objective of virtually connecting individuals to one another, the services are distinct in their articulated goals, the format of information delivery, and the style and type of information disseminated. Thus, they offer perpetrators distinct opportunities to engage in criminal behavior, law enforcement different ways to engage the public or apprehend suspects, and scholars new and interesting opportunities to engage ideas of crime and criminal justice. As discussed below, they key to both service providers is that they “allow users to instantly create, disseminate, and consume information from any location with access to the Internet” (Wang, Gerber, & Brown, 2012, p. 231).
Facebook: “The Power to Share”
Facebook, far and away the most popular contemporary social media site, has over one billion active users worldwide (Solberg, 2010, p. 313). The site, founded to assist college students in identifying and virtually communicating with one another, has expanded to include an extremely popular instant messaging app,7 location “check in” services, and photo sharing capacities via both the parent site and its subsidiaries, Instagram and Glancee (Wilken, 2014, pp. 1088, 1090). Facebook has revolutionized social media by enabling users to connect instantly across multiple social networks using their mobile phones. The company’s mission statement describes the site’s purpose as “giv[ing] people the power to share and make the world more open and connected” (Facebook Mission Statement, 2015).
Facebook has proved a vital tool in organizing and galvanizing social protests from Ferguson, Missouri, to Delhi, India, to Cairo, Egypt (Gerbaudo, 2013; Harlow, 2013; Chai, Gordon, & Johnson, 2013; Preston, 2011). It is hard to conceive of a contemporary social movement, such as Black Lives Matter or Occupy Wall Street, that does not have a page or pages where like-minded persons can obtain information and reach out to others with common interests. The utility of Facebook in social organizing unsurprisingly makes the site an advantageous tool for engaging in criminal enterprises. Gang members, for example, have been identified using the site in order to connect with like-minded individuals, to brag about criminal activity, or to taunt those from rival enterprises in a practice called “drilling” or “Internet banging” (Patton et al., 2016, p. 312; Austen, 2013). In cultural environments where engagement in crime is connected to credibility, the integrated nature of social media provides an opportunity for participants to link a legitimate promotional campaign, such as a music video posted on YouTube, to sites on which they are engaging in actual illicit behaviors (Effron & Janik, 2012). Facebook has been implicated in recent instances of cyber-bullying, some of which have resulted in the injury or suicide of youthful targets (Associated Press, 2012; Goldman, 2010).8 Today, nearly all states have enacted anti-bullying statutes including virtual or electronic harassment among prosecutable crimes (Patchin & Hinduja, 2015, p. 69). In his article and forthcoming book, The Digital Street, Jeffrey Lane (2015) engages in ethnographic research tracing the ways in which a group of Harlem teens interweave physical and virtual relationships. Lane argues that in contemporary society, physical and virtual engagements should not be studied separately, describing “street life [as] characterized by its flow online and offline. As ethnographers, we have to keep up” (2015, p. 44).
In addition to changing the nature of organized crime, routine activities theorists posit that the infrastructures of contemporary SNS provide unique opportunities for perpetrators to locate and identify new subsets of vulnerable targets, particularly in connection with crimes such as sex trafficking, child pornography, blackmail, and fraud (Okorie, 2015; Vishwanath, 2014). Okorie (2015) describes Facebook as an “enabler” of sexual violence (p. 301). Further, “phishing,” reaching out to unsuspecting persons using a false identity in order to effect crimes, is more likely to be effective on social media than via e-mail, due to users’ being “inept at privacy settings, the amount of information provided when a friend request is accepted, and the ‘contagion’ effects of friendship on social media”—the idea that, if I trust this person, you should too (Vishwanath, 2014, p. 2). According to CNN, as of 2012, more than 83 million Facebook profiles were fake (Kelly, 2012, emphasis mine).
The visibility of posts on social media and concomitant lack of constitutional privacy protections for users have made Facebook an extremely attractive tool for criminal investigators (Boux & Daum, 2015, p. 151; Broussard, 2015). Not only does the Fourth Amendment protect information shared on social media on only a limited basis, but standing requirements often mean that information obtained from non-target users’ pages may be used without restriction against the target of the investigation. Police across the United States use various tools of network analysis in order to monitor criminal activity, profile potential criminal actors, and engage in discourse with the public about criminal justice issues (Broussard, 2015). Each novel—and often unanticipated—information sharing technology challenges societal boundaries and norms and alters the landscape of law, security studies, and policing.
Twitter: “Tell Your Stories Here”
Twitter, launched in 2006, is a “microblogging site” via which users communicate by providing brief updates of 140 characters or fewer (“Tweets”). Microblogging is best defined as “a form of personal broadcasting media where information and opinion are mixed together without an established order, usually tightly linked with current reality” (Reips & Garaizar, 2011). While Facebook’s focus is connectivity among chosen “networks” of users, Twitter self-identifies as a public forum, with the company’s privacy policy warning “you are what you Tweet” (Twitter Privacy Policy, 2015). Although Tweets may be directed toward a network of other users, or “followers,” unless certain settings are selected by the user, the default is that any person (with or without a Twitter account) will be able to access one’s Tweets. A unique feature of the service is that commenters may identify, associate, or categorize Tweets in specific groups or with other users via the addition of “hashtags.” In addition, the most popular Tweets can be widely shared via the process of re-Tweeting, whereby recipients repost popular comments, ideas, or memes on Twitter itself or on other forums, such as Facebook, Instagram, or blogs. Currently 320 million persons worldwide use the service, with 80% of commenters Tweeting from mobile phones (Guynn, 2016). Unlike Facebook’s acquisition model, however, Twitter’s growth appears to have stagnated, with critics hypothesizing that users may not fully understand how to use the service (Guynn, 2016).
In its infancy, researchers viewed Twitter primarily as a forum for inconsequential chatter, what van Dijck calls “pointless babble” (van Dijck, 2009; Rogers, 2013). As the service has grown, however, functions such as “hashtagging” have meant that the service increasingly operates similarly to Facebook in facilitating social networks. Jack Dorsey, Twitter’s CEO describes how
bird chirps sound meaningless to us, but meaning is applied by other birds. The same is true of Twitter: a lot of messages can be seen as completely useless and meaningless, but it’s entirely dependent on the recipient.
(Rogers, 2013)
Thompson (2008) poetically responds that while individual Tweets may appear insignificant or even nonsensical, “taken together over time, the little snippets coalesce into a surprisingly sophisticated portrait of your friends and family member’s lives, like thousands of dots making a pointillist painting.”
Like Facebook, Twitter has become a key source for gaining on-the-ground data regarding significant events, such as Election 2016 or the Occupy Wall Street, Arab Spring, and Black Lives Matter movements (Harris, 2015; Lotan et al., 2011; Rusbridger, 2010). As Harris (2015) writes, “documentary tools” such as Facebook and Twitter “can reach individuals through-out the nation and across the world in milliseconds, drastically slashing the time it takes to organize protests” and “influenc[ing] print and television coverage.” Rather than waiting for a nightly news report, users are able to follow incidents in “real time” and opine on government and citizens’ actions in the moment.
The immediacy of engagement offered on Twitter, coupled with its unique 140-character format, also facilitates engagement in criminal acts. Key evidence presented at the trial of two juvenile perpetrators of a 2013 gang rape in Steubenville, Ohio, for example, included Tweets, texts, SnapChats, and YouTube videos uploaded by perpetrators and observers during and after the incident, as well as texts between the victim and one of the perpetrators (Boux & Daum, 2015). Digital evidence was introduced, too, in attempts to prosecute adults with knowledge of the case for failing to report the crimes (Davidson, 2013). It was also used by a hacker group, who anonymously uploaded a video of a student-athlete at Ohio State University whom they recorded making jokes about the rape (Kimmel, 2013). Due to negative responses to the video, the student dropped out of school (Kimmel, 2013). Not only did law enforcement rely heavily on Tweets and other SNS correspondence to investigate the Steubenville rapes, in an increasingly common phenomenon, intoxicated at the time of the offense, the victim herself only became aware of events after seeing video of the attack against her (Boux & Daum, 2015).
The Steubenville prosecutions, Gamergate controversy, and a recent event where two Virginia teens were arrested in connection with the murder of a thirteen-year-old girl they allegedly met through a new SNS application called Kik, demonstrate that criminal practices are evolving at a pace with new communication technologies. At the same time, law enforcement are engaging SNS in efforts to solve and prevent crime, to assist in crowd control, to disseminate pertinent information, and to seek out and influence public opinion regarding law enforcement (Stevens, 2016; Hughes, 2016). A 2013 survey by the International Association of Chiefs of Police found that nearly 96% of police agencies surveyed used social media, over 85% of these for criminal investigations (IACP, 2013). The most common sites used were Facebook and Twitter (IACP, 2013).
Like Facebook, Twitter provides a rich site for the investigation of criminal behavior; the social media site has been referred to as the “ultimate snitch” (Knibbs, 2013). The applications of the service to law enforcement are not only responsive but also interactive and sometimes predictive. Police departments have used Twitter to track public disorder and protests, such as the riots that occurred in Vancouver during the 2011 finals of the Stanley Cup (Knibbs, 2013; Schneider, 2015). Some agencies also have launched successful campaigns encouraging citizens to use the service to report crime (Knibbs, 2013). Seattle’s “Tweets By Beat” program uses Twitter as a virtual police scanner, enabling citizens and the police to report and share experiences of criminal activities in real time (Johnson, 2012). In some areas, police are engaging in more inventive use of the service, attempting to deter drunk driving by “live tweeting” arrests for the crime (Knibbs, 2013).
The value of Twitter and Facebook in facilitating and policing criminal behavior means endless opportunities for criminal justice researchers to craft and answer criminological research questions engaging social media. However, studying virtual encounters provides unique complexities not encountered in the physical world. First and foremost is the question of methodology. The overarching—and yet and unanswered—question is the underlying nature of SNS. Are they sites for dissemination of archival data, tools for communication, or fields in which human subjects research is being conducted? The answer to this question depends upon the research methodology engaged and will dictate the ethical concerns inherent in any study. With this in mind, the following section describes some recent, creative ways in which researchers across disciplines have engaged the study of new social media. The section also offers a few suggestions as to how scholars of criminal justice and criminology might adapt these methods to answer criminological research questions.
Cross-Disciplinary Methodologies for Studying New Media in Criminal Justice
Currently, the most popular method engaged by scholars in criminal justice is survey research (Kleck, Tark, & Bellows, 2006). Kleck and colleagues recorded the methodologies used across articles published in leading peer-reviewed criminal justice and criminology publications. The study found that nearly 50% of articles in leading journals engaged some type of quantitative survey methodology (2006). In contrast, about 30% analyzed archival data, 25% discussed official statistics, and only 12% documented the results of qualitative studies, such as in-depth interviews or participant observation (Kleck et al., 2006). The majority of researchers use secondary data sets in conducting their research, most using these data to engage in statistical analysis (Kleck et al., 2006).
Mindful of the preferred methodological approaches in the field, in the subsections below I offer a few suggestions as to how criminological researchers might engage Facebook and Twitter in their existing research programs. Because so few criminal justice scholars study SNS, this section reviews research methods across disciplines that might be employed to answer criminology- and criminal-justice-oriented research questions. Across disciplines, the current trend involves a push to develop new approaches to analyzing and engaging “big data,” and “datasets that are too large for traditional data-processing systems and that therefore require new technologies” (Provost & Fawcett, 2013, p. 54). Empirically trained criminal justice and criminology researchers are ideally situated to engage in analyses of “big data,” including that mined from social media, and to apply the principles of causal analysis being used in existing research projects to critically assess and evaluate extremely large bodies of information.
Mining, Streaming, Sifting: Social Media as “Big Data”
The information relevant to criminal justice that may be obtained from Twitter and Facebook is seemingly endless. In such a large volume of postings, repostings, and status updates, a data set might comprise Tweets engaging a particular hashtag; Facebook posts in reaction to a particular event, policy, or person; or indications of actual criminal behavior, such as doxing, drilling, cyber bullying, or phishing. For researchers, the biggest challenge is first figuring out what to study and, second, the most efficient and least costly way to capture and effectively study it.
Analyzing Big Data Collected from Twitter
There are two, primary ways to access, or “mine,” public Tweets, the Firehose, which is access to all public Tweets without filtering, or streaming smaller amounts of data via an application program interface, or API (Morstatter, Pfeffer, Liu, & Carley, 2013; Puschmann & Burgess, 2013). The Firehose option is seldom used by academic researchers due to its limited availability, prohibitive cost, and the need for more sophisticated technologies, such as servers or network availability, in order to retain and analyze so much information (Morstatter et al., 2013). Therefore, the most common method engaged by researchers in mining information from Twitter is via a streaming API (Puschmann & Burgess, 2013).
In 2014 Twitter acquired social data provider Gnip, which enables researchers to purchase various “feeds” from Twitter (Twitter Blog, 2014). According to Gnip, two streaming API options are available: “sampled streams,” which deliver a “statistically valid” random sampling of the Firehose, or “filtered streams,” which deliver Tweets matching user-selected filters, such as keywords (Quist, 2011). These data may then be analyzed using the researchers’ preferred data analysis software program, such as NVivo or Dedoose (Marwick, 2013). Puschmann and Burgess (2013) describe Gnip as “one of the best tools” for researchers wanting to access Twitter’s Streaming API; however they caution that the service can be costly (p. 44). Other services offering similar access to researchers include TweetTracker, developed by researchers at Arizona State University, and Hootsuite, a for-profit service that enables users to “track, measure, and share key social media metrics” (TweetTracker; Hootsuite). Importantly, Twitter itself archives Tweets dating back to 2006 and offers a free interface, which allows the public to retrieve up to 1% of Tweets on a given topic via the “advanced search” function; however, researchers concerned with validity or generalizability of results are cautioned that the service currently selects via undisclosed algorithm what Tweets will be retrieved (Morstatter et al., 2013).
Jack Dorsey proclaims that the Twitter service excels in the context of “massively shared experiences” (Weller, Bruns, Burgess, Marht, & Puschmann, 2013). Indeed, the ability to obtain such large amounts of information on a given topic, or from a specific location or locations, has the capacity to contribute greatly to our understanding of criminal justice issues. Social scientists engaging the site for research purposes appreciate its capacity to gather and represent collective impressions about and reactions to events of large-scale importance. For example, after the gang rape of a young woman on a bus in Delhi, India, captured global attention, Purohit et al. (2016) accessed the service’s streaming API9 and analyzed 14 million Tweets to examine global opinions regarding gender-based violence. The researchers, who collaborated with the United Nations Population Fund to develop key terms relating to physical and sexual violence, were able to track Tweets based on user gender and geographical location in order to ascertain public opinion about gender-based violence. Similarly, Chaudhry (2015) in his “Racist Tweets in Canada” project utilized Hootsuite’s streaming API management service in order to identify and geotag (identify via geographic markers) instances of racial intolerance in Canada. A similar project affiliated with the University of Kentucky uses geocoding to identify homophobic, able-ist, and racist language across that country on a “Geography of Hate Map” (Chaudhry, 2015).
In addition to ascertaining global or statewide points of view on a particular topic over time, Twitter’s true strength for researchers lies in the opportunities the service presents to gain of-the-minute insights into events of worldwide significance. If one is seeking to identify trending topics in criminal justice, to measure fear of crime, or to take the temperature of the public regarding a certain incident or occurrence, Twitter is a terrific source of such information. Denef et al. (2013), for example, collected and analyzed all Tweets by the London and Manchester police forces during a series of riots in August 2011. The authors explored both police use of the social medium and public responses in order to analyze the ways in which law enforcement and the public interacted during the crisis. As Chaudhry (2015) describes, “The real-time response tweets are interesting to consider because they reveal thoughts and opinions that most people would want to remain private … This … showcases the type of rich data researchers can collect using Twitter.” A researcher might access a wealth of information about public perceptions about terrorist organization Boko Haram simply by searching “#savethegirls,” and might discover a terrorist cell by geotagging and mapping similar data.10
While one possibility on Twitter is to code the occurrence of a particular key term, phrase, or hashtag, one might also track how often an item has been re-Tweeted, or whether new concepts or “memes” are emerging (Weller et al., 2013; Thelwall, 2013). As Thelwall (2013) describes, the technology available via streaming APIs make it possible to record spikes or declines in volume of Tweets, suggesting “heated discussion,” lack of interest, or “track[ing] the activities of specific users or groups of users over time” (Thelwall, 2013). Chaudhry (2015) notes that, as Twitter has moved from a mere vehicle via which one might obtain data, questions asked by researchers have evolved from “How many followers does a given user have? … to complex queries (How does information propagate among groups of users?).” In connection with any Tweet, “metadata” are available allowing analysis of attributes or properties such as screen or user name, date and time of Tweet, geotag or location, the name associated with the screen name and, perhaps, an optional biography of the user (Hardacker & McGlashan, 2016; Driscoll & Walker, 2014). For example, Hardacker and McGlashan’s (2016) study of rape threats against a particular journalist, employing the methodology of “corpus linguistics,” concludes that one might ascertain not only if there were such threats but also whether a pattern might be established as to timing or location. Weller et al. note that Twitter increasingly functions as an “anticipatory medium,” predicting everything from criminal events to election results, natural disasters, and bull and bear markets (Wang et al., 2012; Kumar, Barbier, Ali Abbasi, & Liu, 2011). In a small study in Charlottesville, Virginia, Wang et al. (2012) modeled the likelihood of occurrence of future crimes, such as hit and run accidents, based on increased hazards reported by a local news station (pp. 231, 232).
As described above, despite the promise of the Twitter service in facilitating communication among users and in providing unique insights into social structures, opinions, and processes, use of the service until recently had experienced a deep decline (LaFrance & Meyer, 2014). Articles regarding Twitter employed terms such as “eulogy” and “twilight” (LaFrance & Meyer, 2014). Barriers to continued success—and, correspondingly, its utility to researchers—include the learning curve to using the service effectively, the lack of control of unverified content, and, as discussed below, disputes over who has control over the “public” information disseminated on the site (LaFrance & Meyer, 2014). Nonetheless, its unique 140-character format and level of accessibility to tell one’s story from anywhere at any time continue to make the service relevant for today’s criminal justice researchers. Politicians’ use of Twitter during Election 2016 may presage a renewal of the service (Heller, 2016). Further, as discussed in the following subsections, the myriad mechanisms for accessing Twitter’s data feed, as well as its overwhelmingly public perspective render the data provided in Tweets “lower hanging fruit” than those one might locate via a Facebook profile.
Mining Facebook for Big Data
Data available regarding users on Facebook are more extensive than available for Twitter, including vast amounts of personal information, such as the “user’s date of birth, political party and religious affiliations, sexual preferences, hobbies and interests, education, and current employment” (Solberg, 2010; Kosinski, 2015). Users may upload photographs or engage location services to signal their immediate geographic location to other users. This increasingly seamless integration of information, including photo sharing, instant messaging, group memberships, and email make Facebook a standout among SNS competitors (Trottier & Fuchs, 2015).
Facebook’s huge number of users means that engaging Facebook for research purposes would contribute greatly toward improving the diversity of viewpoints reflected in academic research studies (Kosinski, 2015). However, although the ability to retrieve these data, as well as to search postings, en masse would be invaluable to improving our understanding of human communication, behavior, and preferences, Facebook’s terms of service do not allow automated data collection to the extent permitted by other services, such as Twitter (Phillips, 2011). Rather than engaging an API and “streaming” data, one of the primary ways social science researchers mine data from Facebook is by recruiting participants and requesting either that they download and provide their own data or that they consent to have their profiles observed and analyzed for research purposes (Phillips, 2011; Kosinski, 2015).
Given restrictions on accessing data, the research methods engaged by social scientists studying Facebook are diverse and creative. Kosinski (2015) recommends “snowball sampling,” whereby researchers simply asks users to invite friends to participate in a research study. Facebook’s huge membership means that this method—which might often generate only a small number of participants—could result in “big data” level responses. For example, Kosinski’s myPersonality application, offering psychological personality tests to participants, “went viral” and in five years grew from 150 to 6 million Facebook participants simply via user-to-user recommendations (Kosinski, 2015).
If researchers are concerned about biases or lack of anonymity inherent in snowball sampling, they may obtain a more “random” sample of data via taking out an advertisement targeted toward a particular population or group (e.g., women between 20 and 40) (Kosinski, 2015). There are also collaborative enterprises that researchers are engaged in. For example, the Computer Forensics Research Lab at the University of Alabama-Birmingham participates in a program sponsored by the National Science Foundation Research Experience for Undergrads (REU) (Weiss & Warner, 2015). Over the summer of 2014, a group of undergraduate students at the university tracked “known criminal groups” on Facebook in order to create a database of such groups and their members to be used in nationwide law enforcement (Weiss & Warner, 2015). As part of the research project, students “created Facebook aliases that looked like cyber criminals” and joined several online groups created around criminal enterprise (Weiss & Warner, 2015). At the conclusion of the project, students identified more than 100,000 users in the criminally identified groups and worked with law enforcement to both create databases of criminal behavior and to effect arrests (Weiss & Warner, 2015).
Other larger-scale studies track the efficacy of policing with regard to criminal behaviors. For example, Lieberman et al. (2013) studied police-citizen interactions by exploring the frequency and content of Facebook postings by US police departments (p. 443). Using a web-crawler, “a computer script that automatically captures web data,” researchers “identif[ied] the 20 largest PDs in the United States that were actively using Facebook” and collected every Facebook message by those departments over a three-month period (Lieberman et al., 2013). The authors then engaged a content analysis that analyzed the type and delivery of messages over that period.
Reading Between the Lines: Social Media as Sources of Archival Data
Although the greatest benefit in researching SNS is the ability to gain insight into “real time” communications, the sheer amount of data available via Facebook and Twitter also makes the sites attractive sources for data to be used in textual analysis. For several years, Twitter has been engaged in a project to archive more than half a trillion Tweets with the Library of Congress (Scola, 2015). According to the Library of Congress, the aim of the Twitter Research Access project is to make collection of Tweets available to researchers in order to “access to a fuller picture of today’s cultural norms, dialogue, trends and events to inform scholarship, the legislative process, new works of authorship, education and other purposes” (Scola, 2015). Although the Twitter archival project is taking longer than expected, as discussed above, Twitter has launched an “advanced search” engine allowing users to gain access to Tweets dating back to 2006 (Hootsuite Blog). It is important to note, however, that rather than being random, Tweets gathered via the Search API are yielded “according to a set of heuristic algorithms that are not known to outside users … More popular Tweets are kept in memory, and less popular tweets are archived” (Driscoll & Walker, 2014). Thus, it is impossible to ensure a random sample via this method.
While the majority of criminal justice scholars may be interested in SNS as sources for “big data” about crime and criminal behavior, archival research also can be useful in diving deeper into the opinions of particular groups or in supplementing understanding on a particular research topic (Marwick, 2013, p. 109). As Marwick (2013) points out, social media such as Twitter are a “giant corpus of text,” thus “many textual analysis methods are appropriate for analyzing Twitter interaction, from qualitative coding of individual tweets to close readings of particular accounts” (p. 115; Hardacke & McGlashan, 2016). She observes that close reading and critical discourse analysis can be useful in identifying,
power relationships and links between texts and ideology. In both instances, the researcher will need to choose a relatively small sample of tweets to analyse. This may be tweets from top users; all tweets from certain users; tweets containing a particular hashtag; tweets to a particular user, and so forth.
(Marwick, 2013)
As Weller et al. (2013) describe, as Twitter commenters use the service, their practices lead to “new forms of communication and new phenomena in participatory culture” (p. xxxi). As Altheide (2013) notes, “frames are significant in defining social problems and issues” (p. 232). Schneider (2015) documents how SNS such as Facebook and YouTube function as locales in which police and citizens participate in “contests over social power,” with law enforcement increasingly and deliberately controlling how their actions are presented and discussed (p. 228). Hasinoff (2015) explores the “panic” that has welled up around the newly emergent crime category of teen “sexting” and identifies concerns with personal privacy and autonomy inherent in the development and policies regulating the sharing and dissemination of self-produced images (p. 1). Patton et al. (2016) explore how “disdain for police” was communicated via Tweets of gang members after the police shooting of a friend (p. 311). Banner (2016) employs Twitter’s Advanced Search Function to collect Tweets about accusers in recent celebrity sexual assault cases and analyze the way online comments reflect societal expectations regarding victims’ interaction with the criminal and civil justice systems.
The images, videos, and memes circulated on social media sites offer rich research opportunities for visual criminologists studying the “documentary tradition” and “politics of spectacle” in the field (Carrabine, 2014). Even Pinterest, an online scrapbooking site that in the United States is used primarily by suburban females, offers pages devoted to serial killers, true crime, and gory crime scenes (Duggan, 2015). Although potentially challenging to access, the depth of information available from SNS, particularly for qualitative researchers, makes it essential one does not disregard such sites as important loci for information about public opinion, crime, and law enforcement.
Social Media as “Fields” or “Sites” for Qualitative or Experimental Research
Although not common in criminology or criminal justice research, recent work such as that by Jeffrey Lane makes it clear that, if one wishes to engage in urban criminology, it is essential to access the SNS utilized by research participants. In his own study, Lane (2015) engaged in observation of research participants, not only in real life but also in their virtual lives on social media. This leads to the important conclusion that the assumptions about behaviors, promises, and interactions that take place in the physical world assume new (and sometimes contradictory) dynamics when continued in an online context. Marwick (2013) similarly finds value in interviewing Twitter users about how they perceive and construct imagined communities online (p. 118). Marwick argues that although she cannot ensure her research sample is representative, this is outweighed by the benefit that she gains by being able to engage extremely broad populations as to their opinions on an issue or problem in a short time. In a more abstract manner, researchers such as Boux and Daum (2015) study research questions across SNS and social and new media. After the Steubenville incident, for example, the researchers traced the prevalence of rape mythology in comments made by participants across different media (Boux & Daum, 2015).
The digital ethnography model embraced by contemporary qualitative researchers bears the promise of gleaning insights previously inaccessible to a scholarly audience. However, it is accompanied by concerns about scientific validity and generalizability, as well as ethical concerns about intruding too deeply into the private thoughts of users without explicit consent. Marwick and Boyd (2011), for example, point out that profiles on any social media may be “constructed with a hyper-aware self-consciousness, as users knew that misspellings, cultural references, and even time stamps were likely to be scrutinized by potential suitors. Similarly, social network site users select ‘markers of cool’ based on an imagined audience of friends and peers” (p. 116). It is not only qualitative but also quantitative researchers who must be especially cautious, as the theoretical framework for research in digital social media bears great influence on the type of data obtained and the potential results of each study. The article explores these issues in the following section.
Ethical Implications, Special Cautions
Although the massive amount of data available on SNS makes them extremely attractive for research purposes, it is important to consider that although posters may appreciate the visibility of comments to the general public—or even to advertisers and marketers—they may not be aware that data by and about them may be a topic of academic research (Solberg, 2010). Solberg (2010) observes that users may view the idea that their information might appear as a topic of analysis in a scholarly journal as “horrifying, or at a minimum surprising” (Solberg, 2010). These privacy concerns and expectations should be particularly carefully guarded in the context of criminal justice research, which not infrequently is undertaken in partnership with governmental organizations, such as the National Institutes of Justice, the Department of Corrections, or a local police department.
As described above, as researchers engage in research using Facebook or Twitter, a vital question to consider at the outset is whether the medium is being used as a data-collection tool or as research site where human subjects are being engaged. The Institutional Review Boards at most universities are developing procedures and policies to assist researchers in evaluating the type of research in which they are engaging and to determine whether special cautions are necessary (Solberg, 2010). Because many research projects engaging SNS involve collecting massive quantities of anonymized, streaming data, universities are likely to determine that such projects are exempt from IRB review (Solberg, 2010). However, projects that collect personal or identifying information about users, those in which researchers plan to interact with subjects via interviews or participant observation, or those who seek to study human behavior may be deserving of a higher level of institutional oversight (Solberg, 2010; Phillips, 2011). Several social media research projects have been critiqued for lack of attention to human subjects’ concerns. In 2012, for example, researchers from Cornell University collaborated with Facebook to manipulate nearly 700,000 users’ newsfeeds in order to explore how an influx of positive or negative posts would impact users’ moods (Hill, 2014). The university issued a statement that IRB approval was not needed, because, although academic researchers helped design the study, it was carried out by corporate researchers at Facebook (Hill, 2014). Even if IRB approval is deemed unnecessary by one’s university, it is important to consult the guidelines of any professional groups to which a researcher might belong. For example, the American Psychological Association prohibits solicitation of research study participants on various social media (Phillips, 2011).
Users’ expectation of privacy is not uniform across SNS. Variations in mission statements, privacy settings, and intended audience may impact the level of caution that must be employed by researchers across diverse media. Many scholars object that the Internet is a clearly public sphere, particularly in the context of sites such as Twitter that proclaim, “You are what you Tweet” (Twitter Privacy Statement, 2015). However, the intent to share personal information with “friends,” physical or digital, may not be the same as an intent to share such information for research purposes. Kosinski (2015) notes that even though Facebook users “must opt in to take a researcher’s Facebook-based personality test or to allow access to their Facebook profiles … when they hit that ‘I agree’ button on the consent form, they may not realize just how much information they’re allowing the researchers to see.” For example, “They may be fine with sharing their gender, location, and even political leanings, but they may not realize that a particular photo might reveal something about their health or sexual orientation” (Kosinski, 2015). In addition, accessing one person’s Facebook profile also may provide the researcher access to pages, comments, and photos by that person’s friends (Kosinski, 2015).
Another, significant, issue with SNS is that the sites are privately owned by corporations that purposefully do not share complete information about what information they are collecting about users or how such information is collected (Driscoll & Walker, 2014). Many of these sites and the utilities they offer users are driven by competition and advertising dollars (Wilken, 2014; Van Dijck, 2013). Increasingly, ownership of such media is concentrated in few hands; Facebook owns Instagram and popular European messaging site What’s App, Twitter owns micro-video sharing service Vine, and Google owns popular video-posting site YouTube and offers its own interactive application, Google+ (Burgess & Green, 2009; VanDijck, 2013). The fact that these platforms are owned and operated by competitive, profit-seeking corporations means that they are rapidly evolving; new functionalities, privacy settings, and changes in software come rapidly and, usually, without warning (van Dijck, 2013; Phillips, 2011). Ellison and boyd (2013) remark, “[S]cholars face a unique challenge in trying to investigate this rapidly moving phenomenon, as they struggle to understand people’s practices while the very systems through which they are enacted shift” (p. 152).
Given that collecting information from SNS is akin to reaching into a corporate-controlled “black box,” a significant concern for social science researchers is whether information retrieved from such sites ever can be valid, reliable, or generalizable to a population as a whole (Driscoll & Walker, 2014, p. 1745). The validity of information obtained from social media sites is cast into question by estimates that up to fifteen percent of Tweets may be generated by automated “bots” imitating human communication (Nesbit, 2016). Issues of both validity and generalizability are complicated by the fact that the Internet may not be inclusive and welcoming to all persons, so that the gender, age, ethnicity, or socioeconomic status of participants on a site may not reflect a society as a whole. Although Facebook and Twitter ban hate speech on their sites, incidents such as Gamergate highlight that Twitter in particular has struggled with widespread trolling and verbal abuse of female posters on the site (Hardacker & McGlashan, 2016). Although Twitter’s lack of growth may be attributed in part to corporate strategies, many commenters highlight the presence of bullying and abuse on the site as a significant contributor toward its decline (Haque, 2015). Haque (2015) writes,
We once glorified Twitter as a great global town square … But I’ve never been to a town square where people can shove, push, taunt, bully, shout, harass, threaten, stalk, creep, and mob you … for eavesdropping on a conversation that they weren’t a part of … to alleviate their own existential rage … at their shattered dreams … and you can't even call a cop.
As Burgess and Green (2009) observe of video site YouTube, as researchers, new social media give us “an opportunity to confront some of participatory culture’s most pressing problems: the unevenness of participation and voice, the apparent tensions between commercial interests and the public good; and the contestation of ethics and social norms that occurs as belief systems, interests, and cultural differences collide.”
The debates taking place in the 21st century regarding the balance between personal privacy and national security are of particular relevance to criminal justice researchers desiring to engage SNS in their projects. As Trottier and Fuchs (2015, p. 31) highlight, while social media contain great promise in marshaling power among citizens in opposition to the government—for example in the creation of Facebook groups, employment of Twitter hashtags, or dissemination of memes supporting personal data security—social media also has the potential to “amplify exposure and visibility” of evidence of criminal acts and to facilitate state surveillance and control.
The ability to geo-locate Tweets, for example, has the opportunity to provide rich information regarding the typology of persons engaging in potentially harmful or threatening behaviors, such as cyber-stalking, doxing, or trolling (Compton et al., 2014). However, in sharing information about, for example, the geographic location and timing of a Tweet, particularly with government agencies, researchers substantially may intrude on individuals’ right to privacy. Similarly, in engaging in research involving “friending” research subjects or joining various Facebook groups may expose researchers to information that suggests criminal behavior. Lane (2015) found in his study of New York City teens that many of them included people they had never met in person as “friends” on Facebook. One may collect data on criminal groups on social media with the assumption not only that such individuals truly are “networked” with one another and that social groups are static and unchanging. Databases of criminal behavior may then be collected—and perhaps shared with government agencies—based on an assumption of intent of belonging and without a system of purging and re-evaluating membership (Broussard, 2015). As Ellison and boyd (2013) caution
On one hand, SNSs offer a vibrant ‘living lab’ and access to behavioral data at a scale inconceivable to many social scientists. On the other, the data that are available present serious research ethics questions and introduces new types of biases that must be examined.
(p. 167)
As Patton et al. (2016) observe, “[M]ost youth are completely unaware that police and other governmental institutions have access to their accounts … or that certain posts can lead to time being spent in prison … It begs the question why more literature is not being produced to inform and reform the practice of police using complied social networking site evidence as an entrance into the juvenile justice system” (p. 313). For research subjects, digital data collected about them at a particular virtual moment may have lifelong, tangible implications.
Conclusion and Future Directions
To say that criminal justice researchers are at the “tip of the iceberg” with regard to new social media is an apt metaphor. Existing research methods may be outdated, unwieldy, and insufficient to capture the wealth of complex and nuanced information big data offers. The hazards in engaging in such research are not insignificant. However, what such data might reveal about criminal justice issues makes SNS a vital object of study.
While this article focuses primarily on Twitter and Facebook, these are only two among hundreds of social networking sites and applications. YouTube, with its “Broadcast yourself” message, for example, hosts more than 80 million videos and is among the top ten most visited sites (Burgess & Green, 2009). Criminal justice scholars are exploring the way that site is being used not only to protest governmental action or to expose police misconduct but also as a tool of the state to survey and assert control over citizens (Schneider, 2015; Brown, 2016). At the same time, legal scholars are debating whether activities on SNS are subject to over or under criminalization. For example, Citron and Franks (2014) advocate for increasing criminal penalties for the nonconsensual distribution of intimate photos and videos, sometimes labeled “revenge pornography,” on YouTube and other SNS, arguing that dissemination of such materials not only drives victims offline but may be tantamount to domestic violence. Hasinoff (2015), on the other hand, describes how the harms of “sexting” among teenagers may be overstated, due to societal “panic” about how teens express their sexuality using new technologies.
Almost daily, we are introduced to novel SNS being engaged in ways with potential to influence public opinion, improve law enforcement, and facilitate criminal behavior. In 2015 the application Kik, an instant messaging service popular with young teens (it permits users to “chat” not only on smartphones but on tablets, such as Kindles or iPads) came under popular scrutiny after it was discovered that two Virginia college students might have used the site to attract and murder a 13-year-old victim (Hughes, 2016). The application SnapChat also is appealing to younger users, due to the ability to select time limits (usually a few seconds) where photos or videos are available prior to being deleted, making the platform the third most popular (behind Facebook and Instagram) among millennial users (Snapchat Guidelines, 2015; Martin-Wilbourn Partners, 2015). YikYak is an application also gaining popularity, due to the ability to post short, anonymous statements to other users sharing the service in a particular location, such as a college campus or classroom (Wagstaff, 2015). Notably, although the sites promise anonymity to users, both provide detailed instructions to law enforcement as to how to access information from the services (Snapchat Guidelines, 2015; Wagstaff, 2015).
As SNS are used by perpetrators to engage in criminal behaviors, at the same time, applications are being developed to discourage such behaviors or to prevent crimes from occurring. Many police departments today offer downloadable apps that the public can use to view lists of wanted criminals or sex offenders or to track and report crime (Sterbenz, 2013). Sexual assault prevention is an area in which the field of crime prevention applications is growing exponentially. A new application called Kitestring, for example, enables users to inform the service that they are in a potentially dangerous situation; if the user does not recontact the service in a specified amount of time, the user’s emergency contacts are notified (Matthews, 2014). Similar applications, such as bSafe, engage features, such as alarms, location sharing, and even fake calls to enhance user’s personal security (“bSafe” website).
A critique of the parallel development of social science and communication or “new media” studies is that, too often, “communication technologies [are decoupled] from the contexts in which people use them” (Katz & Hampton, 2016). As Rafter (2009) observes, a key project of criminologists has been to understand the “social nature of crime.” Tapping into the rich data available via SNS provides the potential for opening up new and exciting avenues of social scientific research, as well as greatly enhancing existing criminological research projects. The failure of criminological and criminal justice researchers to access and analyze such data risks leaving a gaping hole in our understanding about contemporary social issues, phenomena, and interactions. Just this year, market researchers are predicting myriad new applications will skyrocket in popularity. Wanelo, Ello, Shots, Bebo, and Hyper may sound foreign now, but soon they may replace or augment the role of existing SNS in mediating our everyday interactions. Without a doubt, perpetrators will use these new digital technologies to facilitate crime and to identify potential victims, just as law enforcement will be using these platforms to identify suspects and to prevent and respond to criminal behavior. The challenge for researchers in criminal justice and criminology is whether and how to explore the iceberg, documenting this vital and ever-evolving area of social life.