“Fact-Checking Africa”

The prominence of “fake news” today has sparked an open challenge to the legitimacy of traditional news media. As a result, a series of independent data-driven organisations are emerging to fact-check legacy news media as well as other news sources. This study examines how these actors advocate and adopt journalistic practice and the perceived impact they have on news journalism. We draw our data from in-depth interviews with 14 practitioners working in three organisations—Code for Africa, Open Up and Africa Check—that are currently leading major data and fact-checking operations in sub-Saharan Africa. Our findings show that while these non-journalistic actors are at the periphery of news media as institutions, their operations, activities and goals are at the heart of journalistic discourse. In their data strategies, they emerge as data advocates and activists seeking to reformulate fact-checking processes within news media.


Introduction
The practice of fact-checking has recently gained prominence as an attempt to mitigate the presumed effects of "fake news" (understood here as news with intent to disseminate deceiving content), especially outside legacy media organisations (Graves and Cherubini 2016;Tandoc, Lim, and Ling 2018). At the same time, there has been a global rise of independent fact-checking organisations (e.g. Full Fact and PolitiFact) and data-driven non-profits (e.g. Code for America and Hack/Hackers) that advocate for data-driven practices. By February 2018, there were 149 active fact-checking organisations around the globe, according to Duke University Reporter's Lab (Stencel 2018).
In their strategies and activities, these organisations have incorporated technologic, data (referring the aggregation of facts and information in digital data sets or databases) and journalism practices with varying goals, from holding media and politicians accountable to engaging citizens in news production (Fridkin, Kenney, and Wintersieck 2015;Wintersieck 2017).
Research into fact-checking as a data-driven practice is growing, but there is still scarcity of literature addressing the impact on journalistic discourse by non-traditional journalistic actors. Further, existing studies focus on fact-checking and data journalism as practised in the West (Graves 2016;Graves and Cherubini 2016). In proposing the need to diversify sites of the study of fact-checking, Graves (2018) acknowledges the data-driven practice represents a "rare example of a genuinely transnational movement in journalism" that is yet to be explored especially outside the United States (p. 2). Indeed, there are currently few studies of this journalism reform project that tackle the growing adoption of data-driven practices in sub-Saharan Africa-the reason we are interested in this region. While contextualisation is inevitable in interrogating how data-driven organisations in sub-Saharan Africa generate journalistic discourse, this study intentionally abstains from focusing on the particularities of "Africa" (Mbembe 2001) data-driven practices that are often exoticised or are in conflict with Western journalism(s).
We argue that in the understudied region of sub-Saharan Africa, there are developments making the study of data-driven practices interesting for journalism studies in their own accord. There is an emergence of international non-governmental organisations (INGOs) staking out a role in news production (Wright 2018). Three non-profits with activities that span several countries in sub-Saharan Africa-Code for Africa, Open Up and Africa Check-are specific examples of this growing trend. Their activities represent an "entanglement" of practices with traditional news organisations (Baack 2018) as they advocate for the inclusion of data-driven practices within mainstream news organisations, which has the potential to change journalistic discourses and epistemologies. While the non-profits lay claim to the "transnational movement" (Graves 2018), in their vision, goals and activities, they also profess to leading an Africa-wide data-driven campaign, even though their activities are currently confined to a few countries.
The aim of this study is therefore threefold: first, to study how fact-checking is being practised in leading data-driven organisations in sub-Saharan Africa; second, to assess how these organisations at the periphery of news media make sense of factchecking, which lies at the heart of journalistic discourse; and third, to analyse the goals of these practitioners and their perceived impact on the discourse.
Qualitative interviews with practitioners of Africa Check, Open Up and Code for Africa show how journalism discourse can be adopted by actors outside of traditional news organisations and expand that discourse by introducing new perspectives. We argue that these external organisations challenge traditional journalism practice by adopting and advocating for a renewed journalistic discourse that incorporates datadriven practices. They then attempt to transfer back the expanded discourse to news organisations with the potential to affect practice.

Literature Review
Even though fact-checking is a practice often found in organisations outside traditional news media, it has become a defining trait within the heart of journalistic discourse. To be a journalist is associated with verification. The capacity journalism has to generate consensual knowledge is contingent on their authority to convince the public that they adhere to journalistic norms and values. However, in an environment where misinformation threatens news organisations, the process of fact-checking in itself has become an epistemological tool that several actors beyond traditional journalism "FACT-CHECKING AFRICA" propose. Therefore, we turn to fact-checking as a device for shaping news dynamics and how it incorporates external actors that adopt the epistemological qualities often attributed to news media.

Epistemologies of Journalism, Data as Factual Evidence
The aspiration to uphold the standards of journalistic professionalism not only helped shape news organisations' norms and values during the twentieth century, but also legitimised journalists in their role to ascertain what events could be considered news. In this sense, € Ornebring (2017, 75) argues that journalists are "epistemic workers" whose professional role involves "assessing the knowledge claims made by others, and then making knowledge claims of their own". Such power to explicate the world is what Ekstr€ om (2002) calls the epistemology of journalism, in which forms of knowledge and production of knowledge are tightly connected to the public acceptance of those knowledge claims (Andrejevic 2013). However, these epistemological changes are spurred by the adoption of quantification and data-driven processes in journalism practice (Coddington 2015;Ferrer-Conill 2017).
Data, as a source of power, "opens new paths for imagining how journalistic investigations develop epistemologically relevant revelations" (Lewis and Westlund 2014, 454). Thus, it is no surprise that an epistemological shift ignited by processes of datafication has stronger ramifications in a profession that prides itself in producing knowledge that is anchored in facticity and verification. Following the logics of factchecking, data have the ability to become a journalistic source that, veiled in a discourse of objectivity, signifies a qualitative measure of journalistic success.
The reconciliation of ideal journalism with an actual working practice based on reproduction and retransmission of knowledge claims made by other sources strengthens the "epistemological paradox of journalism" ( € Ornebring 2017, 85). The increasing adoption of data-powered processes is amplifying the epistemological shift (Parasie 2014) in which the power to explicate knowledge is placed on data rather than on the journalist.
If data are conceptualised as an aggregation of facts, their value in journalism lies in telling the story the best way possible (Rogers 2013). Furthermore, data and algorithmic practices allow newsworkers to cope with informational flow through computational fact-checking as a form of automated evaluation of veracity (Ciampaglia et al. 2015). Moreover, the capacity to streamline fact-checking through data practices further reveals the complexity of who dictates factual inaccuracies. Regardless of its practitioner, the epistemic nature of factual verification places it at the heart of journalistic discourse.

Fact-Checking as Journalistic Discourse
In a profession as porous as that of the journalist, the strict adherence to journalism standards such as objectivity, autonomy and verifying sources helped demarcate the boundaries in which journalists operate (Mindich 1998). Journalism self-representation cements itself around facts, truth and reality (Zelizer 2009). Factchecking, as a part of the verification process, slowly became synonymous with journalistic practice, separating newsworkers from other opinion commentators (Phillips 2010).
With the rapid expansion of digital media, news production started to diverge visibly from the professional standards established in the social imaginary (Scott 2005). Online newsrooms focused their resources on "adapting stories to the Web and turning them around quickly, not fact-checking" (Singer 2003, 152). A similar narrative is offered by Nyhan and Reifler (2015) who venture that the infrequent resource of fact-checking in contemporary news media might be the result of commercial urges, dwindling resources and demands for faster content iteration.
An increasingly polarized and opaque media landscape, spurred by low barriers to create content online, resulted in two major developments. On the one hand, the increase in fake news through blogs, social media and alternative media has tainted legacy media, which is now being questioned (Bounegru et al. 2017). On the other hand, the explosion of fact-checking in the United States and Europe (Graves and Cherubini 2016;Graves and Glaisyer 2012) derived in a landscape that is both diverse in the amount of fact-checking and fact-checkers. As news media are being questioned and scrutinized, fact-checking has been adopted by a multitude of independent organisations as a way to scrutinize political actors, alternative media outlets and legacy news organisations.
In this sense, the epistemological properties of journalism, the discourse building upon objectivity and facticity as a mode to establish its understanding of reality (Kunelius 2006) is embedded in the practice of fact-checking. Thus, as Graves (2016, 11) proposes, "fact-checkers practice a kind of journalism that is undeniably of this moment, but also claims to revitalise the truth-seeking ideas so vital to the field's self-understanding". Interestingly, this places fact-checking at the heart of journalistic discourse, yet, as news organisations are being challenged, the practice of fact-checking seems to be moving towards the periphery of journalism, performed by external actors.
These external actors, such as INGOs, civic and fact-checking organisations "differ in terms of their organisational aim and funding" (Brandtzaeg and Følstad 2017;Spyksma 2017;Wright 2018) from legacy news media. Even if they operate as actors at the periphery of news media organisations, they merge data-driven practices and factchecking services, promoting "entanglements" of practices (Baack 2018) and a nascent journalistic discourse around issues of quality journalism effectively becoming "unintentional journalists" (Spyksma 2017). However, at stake is the contestation of "journalistic authority" (Carlson 2017) which is all the more reason why a study of the intervention of non-journalistic actors in the production and consumption of knowledge is important. To further explore the external actors' input to journalistic discourse, through data-driven practices geared towards fact-checking in the African context, we pose the following questions: RQ1: What are the goals and activities of data and fact-checking organisations in relation to news journalism?
RQ2: How do data and fact-checking organisations operate regarding traditional journalistic discourse(s)?

Method
This study takes a qualitative approach, based on in-depth interviews with practitioners from three data and fact-checking civic organisations in sub-Saharan Africa. To find a range of interviewees whose insights would give a valuable overview of the operations of these organisations, our sampling was strategic-identifying organisations and individuals with lead positions as stated on the organisation's websites. Our sample comprised 14 respondents-seven men and seven women-involved with the following non-profit organisations: Code for Africa (5), Africa Check (4) and Open Up (5). These organisations were selected owing to the prominence of their work in social media and public forums, mostly English-speaking Africa. To contextualise our cases, we briefly give an overview of the three organisations studied.
Code for Africa is a civic organisation which has since 2015 been at the forefront of various data-driven practices and projects within and outside news organisations. It has branches and partnerships with 15 civic organisations (as of May 2018) that include Cameroon, Ethiopia, Ghana, Kenya, Morocco, Nigeria, Sierra Leone, South Africa, Tanzania and Uganda. Its operations involve working with newsrooms to train data journalists or aggregation of data in digital platforms for investigative journalism and fact-checking purposes. A successful example is a joint partnership with Kenya's Star newspaper to create a doctors' database and a public platform-dubbed Dodgy Doctors-to weed out quacks in the medical profession (see: https://health.the-star.co.ke/).
Open Up focuses on data and civic mobilisation as well as training of traditional journalists. It is based in Cape Town, South Africa. An example of Open Up's project is a searchable repository of government documents in South Africa available to the public (opengazettes.org.za) through which claims made by state organs/officials can be monitored and verified.
Africa Check is a fact-checking organisation founded in 2012 and based at the University of the Witwatersrand in Johannesburg. It conducts fact-checking and training projects for news organisations in South Africa, Kenya, Nigeria and Senegal. Africa Check investigates claims made by public officials or news organisations, using a methodology based on verified sources of evidence. It disseminates its reports mostly through social media. The organisation has in the past few years offered timely factchecks of "State of the Nation" addresses by South African and Kenyan presidentstheir reports becoming source of news for mainstream news organisations.
These organisations receive funding from international bodies such as Bill and Melinda Gates Foundation and the International Center for Journalists (ICFJ), but also generate revenue from small-scale commercial data projects.
Six interviews with practitioners of Africa Check, Code for Africa and Open Up were conducted in person in Cape Town, Johannesburg and Nairobi while eight (as well as the follow-ups) were done through Skype between October 2016 and June 2017. The respondents were based in Kenya, Nigeria and South Africa.

Findings and Discussion
The following themes emerged after the analysis of data: peripheral actors adopting journalistic discourses; data-powered fact-checking strategies; and expanding the meta-journalistic discourse. The respondents here are identified using codes: CA (Code for Africa), AC (Africa Check) and OU (Open Up).

Peripheral Actors Adopting Journalistic Discourses
Our respondents comprised editors, digital storytellers, directors, developers, data evangelists, data journalists, data wranglers and program managers. Their roles were fluid and overlapping and often evolved from one to another. Most participants took part of several training camps, such as the School of Data Fellowship Programme in South Africa, and advocated for use data in journalism. As one respondent explained: Our major focus at the moment is in digital journalism, how to engage the media in utilizing data and building skills to do more investigative journalism using data and more detailed reporting. (CA2) While they were aware they operated at the periphery of news media, they adhered to journalistic norms and values, both in terms of service to the citizen and holding the powerful to account. An Open Up practitioner described their practice as "service journalism", which he argued "is very much around how you give people information which is relevant, personal, timely that they use in their lives" (OU1). At the same time, a respondent from Africa Check reflected on their perceived role in keeping journalists accountable: We occupy a strange place. We are fact checkers … (and) we are also a media development agency and we have that watchdog role which we're not too comfortable with because to have a watchdog role means someone's put you in that position. (AC3) Occupying journalistic spaces is a result of dissatisfaction with the current local journalism practices. Questioning the processes of truth production legitimises their aims for questioning the media, as explained by one of the respondents: A problem with media all over the world is that when they get press releases, they re-write it a little bit, they don't check any of the information themselves and that is presented to the public as news or as a report, but it is not scrutinized. (AC2) Among the activities and efforts, data-driven processes, such as data liberation, data analysis and storytelling techniques, are predominant. Furthermore, as a way to renew journalistic practice, these organisations collaborate with news media by teaching and training journalists. This is a major part of their operations. As a Code for Africa respondent explained: We build individual capacity as much as just teaching them (journalists) the different data pipeline of finding the data, going to search for the right data, looking at the various positive sides, understanding what statistics means really, teaching them how to adapt this to their stories in as much as they are talking to the editor and the publisher how to embed charts in those stories. (CA1) The practitioners, while taking non-traditional journalistic roles, generally define and link their work to legacy news media practices of sourcing and fact-checking. Their contribution to data as a source of news journalism as well as the training work in and outside the newsrooms seeks to reinforce journalism practice but also challenges "FACT-CHECKING AFRICA" traditional practices of legacy news media. Therefore, through their aims, goals and activities, these peripheral actors outside of the news media spheres could be said to be adopting normative values typical of journalistic operations.

Data-Powered Fact-Checking Strategies
Data and its diverse applications in journalism are prominent in these emerging peripheral journalistic actors. However, each organisation shows different patterns of specialisation. Open Up and Code for Africa specialise in data practices and data journalism, while Africa Check focuses mainly on fact-checking strategies.
The respondents make clear that "regarding media development in the different countries, it's not like one size fits all" (CA3). However, data processes were usually scaled and oriented to applicability within newsrooms or audiences. A respondent from Open Up explained the ways in which data were presented to journalists: We created that demand for data and went through the process of, first of all, getting them aware of open data and data journalism and then scaling them up and finding the data, collecting, cleaning, interrogating the data and presenting the data and using it in the normal news cycle. (OU5) Fact-checking was taken as a data-powered strategic tool in the campaign against public misinformation through the spread of fake news on social media. Data were used to fact-check a story or "give a story more meat or basis" (CA3). However, when asked about the quality of the data, the respondents showed a reflexive process of considering data quality is uneven depending on the source of data. A trustworthy source usually went unchallenged, simply because they did not have the resources to cross-check all data. A reflexive approach was presented when discussing what fact-checking strategies entail. Here, some respondents were reluctant to equate data to truth: So we also try to teach that data is not the ultimate source of truth, it's just one more source for a story which you then need to treat and the other sources and also look and verify it and if you can't verify it then you need to make a judgment call as to whether you are going to use in your story, or not. (OU1) Others seem to understand data as a source that carries objective qualities: If you're transparent about your information if you don't use it in a misleading way, that creates a proxy for objectivity that is important. There, you're an activist for the truth but you're not an activist in any other sense. (AC4) Although with varying views of data as a source of fact-checking, the data strategies of these organisations were geared towards verifying information disseminated on news media or countering the spread of "fake" news. The respondents saw their activities as filling a gap in news journalism today since legacy news media remain weak in an information ecology that demands knowledge and skill in using diverse applications and data tools.

Expanding the Meta-Journalistic Discourse
The perceived impact these organisations have is anchored in their organisations' goals. A respondent from Africa Check stated: We have two main goals. One is to help members of the public to make informed decisions or to have accurate data based on factual evidence and then the second is where we try to help raise the standard of journalism or to help journalists be better at fact checking and doing their jobs. (AC2) They make use of their technological capacities to bypass news media and address the public by creating tools that are built to empower citizens to inform themselves. This expands the action and performativity of news organisations to external actors and eventually to the audiences themselves. A respondent explained: We have to stop pretending that media are the gatekeepers of information and that we are the only ones that can make sense of it and we need to start sharing our sources and our resources and our methodology with our audiences, to educate them to become fact checkers themselves. (AC4) Furthermore, their attempt to expand the journalistic practice is evident through their collaboration with news media. All three organisations train journalists to develop data skills and data-driven fact-checking practices. Their perceived impact on news media reflects a newfound form of truth-seeking and transparency that is passed onto journalists within news organisations. Establishing new methodologies within the newsroom has the potential to affect practice. As a respondent stated: I sometimes say we're like the missionaries of fact-checking … (Our job) is to take not only the messaging out there, not only to explain why fact checking is important but to also teach people or to share the methods that we use. (AC4) However, they were aware their work was akin to that of the advocate or the activist and that sometimes their efforts met resistance in news media. The friction with journalists they train is apparent among all respondents. Regardless of the eventual willingness to collaborate, the reluctance towards data practices is apparent: Trying to get them to think of it against what they are used to for years in the style that they are used to, so we put up with words like, 'we don't want to do this … ' In fact, the prevalent word is, 'we can't' … they really don't see the possibilities of some things. (CA1) There is therefore a constant awareness that their approach is foreign to the current understanding journalists have of their own practice, but that news organisations feel is necessary for the current media landscape. Overcoming the resistance may lead to expanding the meta-journalistic discourse.

Concluding Remarks
In this article, we argue that fact-checking carries both journalistic and epistemic discourses that go beyond the figure of the journalist, and that when adopted by extra-journalistic actors, expand data-driven practices by shaping fact-checking within news organisations, but also by bypassing news media and reaching citizens directly.
Returning to the research questions motivating the study, we can see that while there are diverging views, a predominant set of answers emerge.
RQ1: How do data journalism and fact-checking organisations define their goals and activities? How do they perform the tasks of fact-checking? "FACT-CHECKING AFRICA" The aims of the non-profit organisations are a combination of traditional normative journalistic goals, such as informing the public, truth-seeking, storytelling and factchecking; and attempts to create and establish data-powered tools and strategies that change the ways in which news organisations operate and how users access data. The findings point to a divergent understanding of the organisations' activities ranging from journalism, activism and advocacy. These activities comprise news media training, data-related practices and the creation of tools directed at the public.
Fact-checking activities are done at two levels. First, to corroborate the data sets, the organisations gather and compile. The strategies behind the corroboration processes are similar to typical processes of traditional verification. However, due to resource scarcity, the degree of verification is contingent to the trust of the source of data. Second, to hold accountable governmental institutions and media organisation, the strategies are akin to those of watchdog journalism.
RQ2: How do data journalism and fact-checking organisations operate regarding traditional journalistic discourse(s)?
These actors operate at the periphery of journalism. However, their practices are supported and guided by journalistic discourse(s). Our data show that these organisations adhere current journalistic discourses and well-established properties of journalism, such as fact-checking, verification and accountability. Their discontent with the current state of journalism is the basis to legitimise strategies of appropriation, extraction, expansion and re-introduction of journalistic discourse. In this sense, they change and expand the existing discourse. Moreover, they also incorporate new discourse that collides with current journalistic discourse. We could identify two major approaches. First, there is an advocacy to being open about sources. While traditional journalism is anchored in the notion of transparency, sources are protected by journalistic integrity. In the case of data journalism, the intent is to make source as available to the public as possible. Second (and related to the first), a push for establishing data-driven strategies within news media advocates for an open and detailed account of practice. They advocate for a process of verification that is not exclusive to news media but should be replicable by citizens. Once more, the notion of transparency is taken to practice in a way that tries to empower citizens to also fact-check media organisations. This could be understood as a threat to journalistic authority.
This study shows that data and fact-checking non-profits operate under journalistic discourse, but that they expand it to their own organisational goals and re-introduce it by training journalists. We find that in the African context, there is heightened role of peripheral actors in the form of non-profits (see also Wright 2018). We also note that, in reference to recent studies in other contexts (Graves and Cherubini 2016;Graves 2016.), civic organisational approaches are predominant in the data and journalistic practices of the non-profits in Africa.
It is beyond the scope to study if the non-profits' attempts to accomplish the truth-telling mission affect journalistic practice, but considering the processes in which meta-journalistic discourses emerge, the potential to slowly redefine how news is produced seems apparent. This study contributes to two strands of journalism studies: first, to fact-checking research, by focusing on African cases, usually absent in the contemporary literature; and second, to data journalism, by capturing the way by which data advocates attempt to promote data-driven practices by expanding the current journalistic discourse(s). Future studies may explore data-driven practices in African journalistic cultures and their implication to legacy news media as well as comparative approaches that contextualise practices in different journalistic contexts.